Minor spliceosome targeting compositions and their use in cancer treatment
A nucleic acid agent targeting the minor spliceosome inhibits U6atac snRNA to address resistance in CRPC and NEPC, altering the transcriptome and proteome, providing a novel therapeutic approach for lethal prostate cancer and potentially other cancer types.
Patent Information
- Application Number
- JP2024531654
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2021-08-17
- Filing Date
- 2022-08-17
- Publication Date
- 2024-11-19
AI Technical Summary
Current cancer treatments, particularly for castration-resistant prostate cancer (CRPC) and neuroendocrine prostate cancer (NEPC), face challenges due to intrinsic or acquired resistance mechanisms, including AR axis reactivation and neuroendocrine differentiation, with limited understanding of minor spliceosome (MiS) function in cancer progression and resistance.
A pharmaceutical nucleic acid agent targeting the minor spliceosome (MiS), specifically inhibiting U6atac snRNA, is developed to downregulate MiS activity, combined with platinum-containing drugs, to block cancer cell proliferation and prevent recurrence.
The MiS inhibition effectively reduces cancer cell proliferation and recurrence by altering the transcriptome and proteome, offering a novel therapeutic strategy for lethal prostate cancer types and potentially extending to other cancer types.
Smart Images

Figure 2024543181000009 
Figure 2024543181000010 
Figure 2024543181000011
Abstract
Description
[Technical field]
[0001] This application claims the benefit of European Patent Application (EP) No. 21191594.7, filed August 17, 2021, which is incorporated herein by reference.
[0002] The present invention relates to pharmaceutical nucleic acid agents capable of downregulating or inhibiting the activity of minor spliceosome (MiS) for use in the treatment or prevention of recurrence of cancer. The present invention also relates to a method for the diagnosis of cancer. [Background technology]
[0003] Cancers constantly evolve by modifying their transcriptome while evading therapeutic intervention. Constantly exploited splicing leads to changes in the isoforms of key cancer genes involved in disease progression. Currently, little is known about MiS in cancer progression. Herein, we investigate PCa as an example cancer to understand the role of MiS in disease progression.
[0004] Androgen deprivation therapy (ADT) is used to treat advanced PCa. However, for PCa resistant to ADT, second-generation potent androgen receptor signaling inhibitors (ARSi), such as enzalutamide and abiraterone, are used. Although initially effective, patients eventually develop this ultimately fatal disease due to intrinsic or acquired resistance to ARSi in the form of castration resistant prostate cancer (CRPC). In CRPC, resistance to ARSi is mediated by intratumor heterogeneity driven by AR axis reactivation, including AR amplification, AR mutations, AR coactivators, and ligand-independent AR activation. Another mechanism that has recently gained increased recognition in the long-term treatment of ARSi (AR inhibition) is the transdifferentiation of adenocarcinoma (CRPC-adeno) to neuroendocrine PCa (CRPC-NE), a highly lethal and AR-independent PCa variant. In general, CRPC-adeno and CRPC-NE are defined by the expression / non-expression of characteristic markers such as AR and KLK3 (CRPC-adeno) or the absence of SYP, CHGA and AR (CRPC-NE). The expression of these markers can be quantified by the so-called AR score or NEPC score.
[0005] CRPC-adeno and CRPC-NE share similar genomic landscapes but dramatically different transcriptomes, suggesting non-coding RNA events and RNA splicing as potential mechanisms of PCa transdifferentiation and progression. Indeed, multiple studies have suggested that alternative splicing of AR transcripts plays a key role in the therapeutic resistance of CRPC-adeno. At the same time, the splicing factor SRRM4 has been identified as a key promoter of CRPC-NE transdifferentiation. In general, non-canonical splicing is widely associated with prostate tumorigenesis, and many PCa-related genes exhibit isoform switching during cancer initiation and progression. However, little is known about the pathways that control it, and a unified hypothesis explaining the molecular origin of those isoforms is still lacking. Therefore, understanding the mechanisms underlying splicing aberrations in PCa is essential both to predict tumor progression and to discover key regulators. Although the role of canonical splicing in cancer has been widely studied, our current understanding of the interplay between minor intron splicing and cancer is lacking.
[0006] Minor introns (<0.5%) that require the minor spliceosome are found in most major intron-bearing genes that are spliced by the major spliceosome. These minor intron-containing genes (MIGs) perform diverse functions in different molecular pathways. Despite their diverse functions, MIGs are highly enriched in the essentialome (a list of genes essential for survival). The essentiality of MIGs is reflected in early embryonic lethality upon inhibition of MiS in mice, zebrafish, and Drosophila. Furthermore, loss of U11 snRNA in the developing cerebral cortex results in aberrant splicing of MIGs, which results in cell cycle defects and defects in rapidly dividing neural stem cells. With regard to cancer, dysregulation of minor intron splicing has been implicated in Peutz-Jegger syndrome and myelodysplastic syndromes, which often transition to gastrointestinal cancers and acute myeloid leukemia, respectively. There is further evidence that MiS components show robust associations with increased risk of scleroderma (U11 / U12-65K protein), AML (U11-59K protein), and familial PCa (U11 snRNA). Indeed, a microRNA (miRNA) profiling study of Finnish high-risk PCa families identified altered U11 snRNA expression as a risk factor for PCa development. Summary of the Invention [Problem to be solved by the invention]
[0007] Based on the above-mentioned state of the art, the object of the present invention is to provide means and methods for treating or preventing recurrence of cancer. [Means for solving the problem]
[0008] This object is achieved by the subject matter of the independent claims herein, together with further advantageous embodiments described in the dependent claims, the examples, the figures and the general description. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0009] Summary of the Invention Herein, we show that MiS function, controlled by the AR axis, increases with the stage and degree of differentiation of (prostate) cancer. Indeed, the MiS component U6atac snRNA may serve as an additional marker for cancer diagnosis. We show that siU6atac-mediated MiS inhibition is more effective in blocking PCa cell proliferation than current state-of-the-art combination therapy such as EZH2 inhibitor / enzalutamide. Finally, we show that other MiS components can also be targeted and that MiS inhibition also blocks proliferation of other cancer cell types. Overall, this study uncovers a novel pathway, the minor spliceosome, as a therapeutic entry point for lethal PCa and shows that this strategy extends to other cancer types.
[0010] A first aspect of the present invention relates to a pharmaceutical nucleic acid agent capable of downregulating or inhibiting the activity of the minor spliceosome (MiS) for use in the treatment or prevention of recurrence of cancer.
[0011] A second aspect of the present invention relates to a method for assigning to a patient a likelihood of having or developing cancer, wherein when the expression level of snRNA U6atac is 2 to 3, a high likelihood of having or developing cancer is assigned.
[0012] A third aspect of the invention relates to a pharmaceutical nucleic acid agent for use according to the first aspect, wherein a patient is assigned to have or be likely to develop cancer, or to have or be likely to develop cancer of an aggressive (advanced) treatment-resistant phenotype, according to the method of the second aspect.
[0013] Terms and Definitions For the purposes of interpreting this specification, the following definitions shall apply, and where appropriate, terms used in the singular shall include the plural and vice versa. In the event that a definition set forth below conflicts with any document incorporated herein by reference, the definition set forth herein shall control.
[0014] As used herein, the terms "comprising," "having," "containing," "including," and other similar forms and their grammatical equivalents are intended to be equivalent in meaning and that the item or items following any one of these words are intended to be open-ended in that they are not an exhaustive list of such item or items, or are limited to only the item or items listed. For example, an item "comprising" components A, B, and C can consist of components A, B, and C (i.e., contain only components A, B, and C), or can include not only components A, B, and C, but also one or more other components. Thus, "comprising" and similar forms and their grammatical equivalents are intended and understood to include disclosure of embodiments that "consist essentially of" or "consist of."
[0015] Where a range of values is provided, unless the context clearly dictates otherwise, it is understood that each intervening value, to the tenth of the unit of the lower limit, between the upper and lower limit of that range, and any other stated or intervening value within that stated range, is encompassed within the disclosure, subject to any specifically excluded limit in the stated range. Where one or both of the limits are included in the stated range, ranges excluding either or both of those included limits are also included in the disclosure.
[0016] As used herein, reference to "about" a value or parameter includes (and describes) the variation directed to the value or parameter itself. For example, a statement referring to "about X" also includes the statement "X."
[0017] As used in this specification, including the appended claims, the singular forms "a," "or," and "the" include plural references unless the context clearly dictates otherwise.
[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art (e.g., cell culture, molecular genetics, nucleic acid chemistry, hybridization techniques, and biochemistry). Standard techniques are used for molecular, genetic, and biochemical techniques (see generally Sambrook et al., Molecular Cloning: A Laboratory Manual, 4th ed. (2012) Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, and Ausubel et al., Short Protocols in Molecular Biology (2002) 5th ed., John Wiley & Sons, Inc.) and chemical techniques.
[0019] The term snRNA U6atac as used herein relates to: [ka]
[0020] General molecular biology: nucleic acid sequence, expression The term "gene" refers to a polynucleotide that contains at least one open reading frame (ORF) that can encode a particular polypeptide or protein after being transcribed and translated. A polynucleotide sequence can be used to identify larger fragments or full-length coding sequences of the gene with which it is associated. Methods for isolating larger fragment sequences are known to those of skill in the art.
[0021] The term "gene expression" or "expression", or the term "gene product", may refer to either or both of the process of production of nucleic acids (RNA) or production of peptides or polypeptides - and their products, also called transcription and translation, respectively, or any of the intermediate processes that regulate the processing of genetic information to result in a polypeptide product. The term "gene expression" also applies to the transcription and processing of RNA gene products, such as regulatory RNA or structural (e.g., ribosomal) RNA. Where the expressed polynucleotide is derived from genomic DNA, expression may include splicing of mRNA in eukaryotic cells. Expression can be assessed at both the level of transcription and translation, in other words, the mRNA and / or protein product.
[0022] The term "downregulation" or "suppression" of expression herein relates to the ability to reduce the number of RNA molecules in a cell.
[0023] The term "nucleotide" in the present context refers to a building block of a nucleic acid or a nucleic acid analog, which oligomer is capable of forming selective hybrids with an RNA or DNA oligomer based on base pairing. The term "nucleotide" in this context includes the building blocks of classical ribonucleotides adenosine, guanosine, uridine (and ribosylthymine), cytidine, and the building blocks of classical deoxyribonucleotides deoxyadenosine, deoxyguanosine, thymidine, deoxyuridine, and deoxycytidine. It also includes analogs of nucleic acids, such as phosphothioates, 2'O-methylphosphothioates, peptide nucleic acids (PNA; N-(2-aminoethyl)-glycine units linked by peptide bonds with the nucleic acid base attached to the alpha carbon of glycine) or locked nucleic acids (LNA; 2'O,4'C methylene bridged RNA building blocks). When referring to a "hybridization sequence" in the present specification, such a hybridization sequence may be composed of any of the above nucleotides, or a mixture thereof.
[0024] In the present context, the term "capable of forming hybrid or hybridizing sequences" refers to sequences that are capable of selectively binding to their target sequence under conditions present in the cytosol of a mammalian cell. Such hybridizing sequences may be continuously reverse-complementary to the target sequence or may contain gaps, mismatches or additional unmatched nucleotides. The minimum length of a hybridizable sequence depends on its composition and backbone chemistry in terms of the inclusion of C or G nucleotides, which contribute more to the binding energy than A or T / U nucleotides.
[0025] In the present context, the term "hybridizing sequence" encompasses polynucleotide sequences that comprise or consist essentially of RNA (ribonucleotides), DNA (deoxyribonucleotides), phosphothioate deoxyribonucleotides, 2'-O-methyl modified phosphothioate ribonucleotides, LNA and / or PNA nucleotide analogs. In certain embodiments, a hybridizing sequence according to the invention comprises 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29 or 30 nucleotides. In certain embodiments, the hybridizing sequence comprises deoxynucleotides, phosphothioate deoxynucleotides, LNA and / or PNA nucleotides, or mixtures thereof.
[0026] The term "antisense oligonucleotide" in the context of this specification relates to an oligonucleotide that has a sequence substantially complementary to an RNA and is capable of hybridizing with the RNA. Such an antisense effect on an RNA results in the modulation, in particular the inhibition or suppression, of the biological effect of the RNA. If this RNA is an mRNA, the expression of the resulting gene product is inhibited or suppressed. Antisense oligonucleotides can consist of DNA, RNA, nucleotide analogs and / or mixtures thereof. The skilled artisan knows various commercial and non-commercial sources for the calculation of the theoretically optimal antisense sequence for a given target. Optimization can be performed both in terms of the nucleic acid base sequence and the backbone (ribo, deoxyribo, analog) composition. There are numerous sources of actual physical oligonucleotides, which are generally synthesized by solid-state synthesis.
[0027] The term "siRNA" (small / short interfering RNA) in the context of this specification relates to an RNA molecule capable of interfering with (in other words inhibiting or preventing) the expression of a gene that contains a nucleic acid sequence that is complementary to or hybridizes to the sequence of the siRNA in a process called RNA interference. The term "siRNA" is meant to encompass both single-stranded and double-stranded siRNAs. siRNAs are typically characterized by a length of 17-24 nucleotides. Double-stranded siRNAs can be derived from longer double-stranded RNA molecules (dsRNA). The prevailing theory is that this longer double-stranded RNA is cleaved by an endoribonuclease (called Dicer) to form double-stranded siRNAs. In a nucleoprotein complex (called RISC), the double-stranded siRNAs are unwound to form single-stranded siRNAs. Most RNA interference works by binding of siRNA molecules to mRNA molecules with complementary sequences, leading to the degradation of the mRNA. RNA interference can also occur when siRNA molecules bind to intronic sequences in pre-mRNA (immature, unspliced mRNA) in the nucleus of a cell, causing the degradation of the pre-mRNA.
[0028] The term "shRNA (small hairpin RNA)" in the context of this specification relates to an artificial RNA molecule with a tight hairpin turn that can be used to silence the expression of a target gene via RNA interference (RNAi).
[0029] "sgRNA" (single guide RNA) in the context of this specification relates to an RNA molecule capable of sequence-specific inhibition of gene expression via the CRISPR (clustered regularly interspaced short palindromic repeats) mechanism.
[0030] The term "miRNA (microRNA)" in the present context relates to small non-coding RNAs (containing about 22 nucleotides) that function in RNA silencing and post-transcriptional regulation of gene expression.
[0031] Join: The term "specific binding" in the context of the present invention refers to the property of a ligand that binds to its target with a certain affinity and target specificity. The affinity of such a ligand is indicated by the dissociation constant of the ligand. A specifically reactive ligand has a dissociation constant ≦10 upon binding to the target. -7 mol / L, but has a dissociation constant at least three orders of magnitude higher when interacting with a molecule that has nearly the same chemical composition as the target but a different three-dimensional structure.
[0032] As used herein, the term "pharmaceutical composition" refers to a compound of the present invention or a pharma- ceutically acceptable salt thereof, together with at least one pharma- ceutically acceptable carrier. In certain embodiments, the pharmaceutical composition of the present invention is provided in a form suitable for topical, parenteral, or injectable administration.
[0033] As used herein, the term "pharmaceutically acceptable carrier" includes any solvents, dispersion media, coatings, surfactants, antioxidants, preservatives (e.g., antibacterial agents, antifungal agents), isotonic agents, absorption delaying agents, salts, preservatives, drugs, drug stabilizers, binders, excipients, disintegrants, lubricants, sweeteners, flavorings, dyes, and the like, and combinations thereof, as known to those skilled in the art (see, e.g., Remington: the Science and Practice of Pharmacy, ISBN0857110624).
[0034] As used herein, the term "treating" or "treatment" of any disease or disorder (e.g., cancer) refers, in one embodiment, to alleviating the disease or disorder (e.g., delaying, preventing, or reducing the onset of the disease or at least one of its clinical symptoms). In another embodiment, "treating" or "treatment" refers to alleviating or improving at least one physical parameter, including those that may not be discernible to the patient. In yet another embodiment, "treating" or "treatment" refers to modulating the disease or disorder, either physically (e.g., stabilization of a discernible symptom), physiologically (e.g., stabilization of a physical parameter), or both. Methods for assessing the treatment and / or prevention of a disease are generally known in the art, unless otherwise described herein below.
[0035] Detailed Description of the Invention A first aspect of the present invention relates to a pharmaceutical nucleic acid agent capable of downregulating or inhibiting the activity of the minor spliceosome (MiS) for use in the treatment or prevention of recurrence of cancer.
[0036] In a particular embodiment, the agent is capable of downregulating or inhibiting the expression of snRNA U6atac, in particular by hybridizing to SEQ ID NO: 001 and leading to its degradation or inhibition.
[0037] In certain embodiments, the agent is, or encodes, an antisense oligonucleotide. In certain embodiments, the agent is, or encodes, an siRNA.
[0038] In a particular embodiment, the cancer is a fatal cancer of advanced, treatment-resistant phenotype (neuroendocrine carcinomas and adenocarcinomas, in particular neuroendocrine carcinomas).
[0039] In certain embodiments, the cancer is selected from glioblastoma, breast cancer, chronic myeloid leukemia (CML), bladder cancer, colon cancer, renal cancer, and prostate cancer.
[0040] In certain embodiments, the cancer is prostate cancer.
[0041] In certain embodiments, the cancer is selected from castration-resistant prostate cancer (CRPC), neuroendocrine prostate cancer (NEPC), castration-resistant neuroendocrine prostate cancer (CRPC-NE), and small cell prostate cancer. In certain embodiments, the cancer is androgen receptor negative NE type and / or androgen receptor negative and NE negative type fatal cancer. Advanced resistant prostate cancer does not respond to ADT (androgene deprivation therapy) and ARSi (androgen receptor inhibitor).
[0042] In certain embodiments, the agent is administered in combination with a platinum-containing complex, hi certain embodiments, the agent is administered in combination with a platinum-containing agent selected from carboplatin, satraplatin, cisplatin, dicycloplatin, nedaplatin, oxaliplatin, picoplatin, and / or triplatin.
[0043] A second aspect of the present invention relates to a method for assigning a patient a likelihood of having or developing cancer. If the expression level of snRNA U6atac is 2-3, a high likelihood of having or developing cancer is assigned. In situ testing, there can be four levels: 0, 1, 2, and 3. 2-3 corresponds to moderate / strong expression.
[0044] In certain embodiments, an expression level of snRNA U6atac of 2 or 3 is assigned a high likelihood of having or developing cancer.
[0045] In certain embodiments, a high likelihood of having or developing an advanced, treatment-resistant phenotype of cancer is assigned if the expression level of snRNA U6atac is 2 or 3. In certain embodiments, a high likelihood of having or developing an advanced, treatment-resistant phenotype of cancer is assigned if the expression level of snRNA U6atac is 2 or 3.
[0046] We observed strong differences in U6atac expression between 1) benign tissues and cancers, and 2) primary cancers and advanced or metastatic cancers. We concluded that this relationship is valid for all cancers, with a higher U6atac score indicating a more advanced stage of cancer.
[0047] In certain embodiments of the second aspect, the cancer is selected from glioblastoma, breast cancer, chronic myeloid leukemia (CML), bladder cancer, colon cancer, renal cancer, and prostate cancer. In certain embodiments of the second aspect, the cancer is prostate cancer.
[0048] A third aspect of the invention relates to a pharmaceutical nucleic acid agent for use according to the first aspect, wherein a patient is assigned to have or be likely to develop cancer, or to have or be likely to develop cancer of an advanced, treatment-resistant phenotype, according to the method of the second aspect.
[0049] A further aspect of the invention relates to a method for treating or preventing the recurrence of cancer in a patient, said method comprising administering to the patient a pharmaceutical nucleic acid agent according to the first aspect.
[0050] A further aspect of the invention relates to a method for treating or preventing recurrence of cancer in a patient, said method comprising the steps of: a. obtaining a tumor sample from said patient; b. determining the expression level of snRNA U6atac in the sample; c. identifying the patient as being at high risk of cancer recurrence if the expression level of the snRNA U6atac is 2 or 3; d. Treating said patient at high risk of cancer recurrence with an anti-neoplastic agent.
[0051] In certain embodiments, the patient is treated by administration of a pharmaceutical nucleic acid agent according to the first aspect.
[0052] A further aspect of the invention relates to a system for carrying out the method according to the second aspect.
[0053] A further aspect of the invention relates to the use of an agent capable of determining the expression level of snRNA U6atac in the manufacture of a kit for the detection of cancer.
[0054] Medical Treatments, Dosage Forms, and Salts Similarly, within the scope of the present invention is a method or treatment of cancer in a patient in need thereof comprising administering to the patient a nucleic acid sequence vector according to the above description.
[0055] Similarly, there is provided a dosage form for the prevention or treatment of cancer comprising a non-agonist ligand or an antisense molecule according to any of the above aspects or embodiments of the invention.
[0056] Those skilled in the art will recognize that the drug compounds specifically mentioned herein may exist as pharma- ceutically acceptable salts of the drug, which contain an ionized drug and an oppositely charged counterion. Non-limiting examples of pharma- ceutically acceptable anionic salt forms include acetate, benzoate, besylate, bitatrate, bromide, carbonate, chloride, citrate, edetate, edisylate, ebonate, estolate, fumarate, gluceptate, gluconate, hydrobromide, hydrochloride, iodide, lactate, lactobionate, malate, maleate, mandelate, mesylate, methyl bromide, methyl sulfate, mucate, napsylate, nitrate, pamoate, phosphate, diphosphate, salicylate, disalicylate, stearate, succinate, sulfate, tartrate, tosylate, triethiodide, and valerate salts. Non-limiting examples of pharma-ceutically acceptable cationic salt forms include aluminum, benzathine, calcium, ethylenediamine, lysine, magnesium, meglumine, potassium, procaine, sodium, tromethamine, and zinc.
[0057] The dosage form may be for parenteral administration, such as in the form of subcutaneous, intravenous, intrahepatic, intramuscular injection, etc. Optionally, pharma- ceutically acceptable carriers and / or excipients may be present.
[0058] Pharmaceutical Compositions and Administration Another aspect of the present invention relates to a pharmaceutical composition comprising a compound of the present invention, or a pharma- ceutically acceptable salt thereof, and a pharma- ceutically acceptable carrier. In a further embodiment, the composition comprises at least two pharma- ceutically acceptable carriers as described herein.
[0059] In certain embodiments of the present invention, the compounds of the present invention are generally formulated into pharmaceutical dosage forms in order to provide an easily controllable dosage of the drug and to present the patient with a clear and easy to administer product.
[0060] The pharmaceutical composition can be formulated for enteral administration, particularly oral or rectal administration. Furthermore, the pharmaceutical composition of the present invention can be in a solid form (including, but not limited to, capsules, tablets, pills, granules, powders, or suppositories) or in a liquid form (including, but not limited to, solutions, suspensions, or emulsions).
[0061] The pharmaceutical composition can be formulated for parenteral administration, eg, by intravenous (iv) injection, intradermal, subcutaneous, or intramuscular administration.
[0062] The dosing regimen of the compounds of the present invention will vary depending on known factors such as the pharmacodynamic properties of the particular agent and its mode and route of administration; the recipient's species, age, sex, health, medical condition, and weight; the nature and extent of the condition; type of concurrent treatment; frequency of treatment; route of administration, the patient's renal and hepatic function, and the desired effect. In certain embodiments, the compounds of the present invention may be administered once daily or the total daily dosage may be administered in divided doses two, three, or four times daily.
[0063] In certain embodiments, the pharmaceutical composition or combination of the present invention may be a unit dosage of about 1-1000 mg of active ingredient(s) for a subject of about 50-70 kg. The therapeutically effective dosage of the compounds, pharmaceutical compositions, or combinations thereof depends on the subject's species, weight, age, individual condition, and the disorder or disease being treated or its severity. A physician, clinician, or veterinarian of ordinary skill can readily determine the effective amount of each active ingredient required to prevent, treat, or inhibit the progression of the disorder or disease.
[0064] The pharmaceutical compositions of the present invention can be subjected to conventional pharmaceutical operations such as sterilization, and / or can contain conventional inert diluents, lubricants, or buffers, as well as adjuvants such as preservatives, stabilizers, wetting agents, emulsifiers, and buffers. They can be produced by standard processes, such as conventional mixing, granulation, dissolving, or lyophilization processes. Many such procedures and methods for preparing pharmaceutical compositions are known in the art, see, for example, L. Lachman et al., Theory and Practice of Industrial Pharmacy, 4th ed., 2013 (ISBN8123922892).
[0065] Methods of manufacture and treatment according to the present invention The present invention further encompasses, as an additional aspect, the use of a nucleic acid agent as identified herein, or a pharma- ceutically acceptable salt thereof, as defined in detail above, for use in a method for the manufacture of a medicament for the treatment or prevention of cancer.
[0066] Similarly, the invention encompasses a method for treating a patient diagnosed with a cancer-related disease, which method involves administering to the patient an effective amount of a nucleic acid agent identified herein, or a pharma- ceutically acceptable salt thereof, as defined in detail herein.
[0067] The present invention further encompasses the use of a nucleic acid agent capable of detecting the expression level of the snRNA U6atac identified herein for use in the manufacture of a kit for detecting cancer.
[0068] For example, whenever alternatives for a single separable feature, such as a target gene, a type of nucleic acid drug, or a medical indication, are presented herein as "embodiments," it is understood that such alternatives may be freely combined to form separate embodiments of the invention disclosed herein.
[0069] The present invention is further illustrated by the following examples and figures, from which further embodiments and advantages can be derived, which are intended to illustrate the invention without limiting its scope. [Brief description of the drawings]
[0070] [Figure 1-1] FIG. 1 shows that MIG expression patterns and MiS component expression correlate with PCa progression. A) Enrichment of 542 MIGs among proteins interacting with proteins encoded by 403 cancer genes and 26 prostate cancer genes at distances (d)=1, d=2, d=3, d=4, d=5 in human protein-protein interaction networks. Enrichment values indicate the log2 fold change in the presence of MIGs in distance bins compared to a randomly sampled set of genes of the same size. Error bars indicate standard deviation. Asterisks indicate P<10E-5. B) The largest connected sub-network of all interactions between prostate cancer genes and MIGs, capturing 87 genes (inset). The curve shows the distribution of network size when we randomly sample an equally sized set of non-MIGs (542). C) Sub-network of interactions between 74 MIGs and 19 prostate cancer genes (enlarged inset of 1B). [Figure 1-2]D) Relative performance of MIGs and non-MIGs for clustering gene expression data from PCAWG for nine different cancer types (biliary, breast, colorectal, lung, ovarian, pancreatic, gastric, thymic, and prostate). Each box corresponds to a 1,000-length simulation in which non-MIGs were randomly sampled among all non-MIGs in the genome. The plot shows a negative relationship between the relative abundance of non-MIGs and the quality of clustering, quantified using the silhouette coefficient. P-values are based on two-tailed t-tests, with each sample being assigned a value that is the difference between the silhouette coefficient of one out of 1000 100% non-MIG samples and the corresponding silhouette coefficient when only MIGs are used for clustering (i.e., the silhouette coefficient corresponding to 0% non-MIGs). E) Relative performance of MIGs and non-MIGs for clustering gene expression data of different stages of prostate cancer progression. Gene expression data were obtained from GTEx, TCGA, and SU2C, respectively. These datasets represent increasing stages of tumor progression (from healthy tissue in GTEx to advanced prostate tumors in SU2C). Clustering was performed with an n_cluster value of K=3. Each box in this plot corresponds to a simulation of length 1000 in which non-MIGs were randomly sampled among all non-MIGs in the genome. The plot shows a negative relationship between the relative abundance of non-MIGs and the quality of clustering, quantified using the Silhouette Coefficient. P values are based on two-tailed t-tests, with each sample value taken as the difference between the Silhouette Coefficient of one of 1000 100% non-MIG samples and the corresponding Silhouette Coefficient when only MIGs were used for clustering (i.e., the Silhouette Coefficient corresponding to 0% non-MIGs). [Figure 1-3]F) Box plot showing the distribution of RNU11 expression values (counts per million (CPM)) across primary tumor samples in which RNU11 was detected. Samples (n=23) were grouped by low (6+7) or high (8+9) Gleason score (n=23, Wilcoxon test, p-values shown). [Figure 1-4] G) Pearson correlation analysis between RNPC3 and SOX2. Organoids are color-coded by transcriptome NEPC score, where a score >0.4 indicates a CRPC-NE phenotype and a score <0.4 indicates a CRPC-Adeno phenotype (Beltran et al., PMID26855148). Organoids are sized by AR score. [Figure 1-5] H) Heatmap showing RNAseq expression (FPKM) of prostate cancer cell lines ordered by increasing NEPC score and decreasing AR score. NEPC score and AR score were calculated based on FPKM values of 70 and 27 gene sets to estimate the likelihood of a test sample being CRPC-NE or CRPC-adeno, respectively. I) U6atac snRNA expression as x-fold siScrambled normalized to the mean values of GAPDH and ACTB gene transcripts in different PCa cell lines (RWPE n=5, LNCaP n=7, DU145 n=7, PC3 n=3, PCc8 n=9, L-AR n=10, C4-2 n=7, 22Rv1 n=5, PM154 n=5, H660=5). Data are presented using box plots. Box plots show minimum, 1st quartile, median, 3rd quartile and maximum (2-tailed Mann-Whitney test, ns; not significant, p>0.05, *p<0.05, **p<0.01, ***p<0.001). Each data point represents x-fold of LNCaP cells obtained from a single experiment. Experiments were performed in triplicate. [Figure 1-6] J) U6atac score from TMA analysis (Wilcoxon test, p=0.0038). [Figure 2-1]Figure 2 shows that MiS activity is regulated by AR and elevated in prostate cancer. A) Representative images of validation of the RNU6atac BaseScope probe in human primary, primary with potential metastasis, and metastatic PCa tissues. Scale bar represents 50 μm. [Figure 2-2] Quantification of U6atac scores for all TMA analyses is shown on the right. Between-group comparisons were performed using Wilcox tests (primary vs primary (metastatic: p = 0.004, primary vs metastatic: p < 2.22e-16, primary (metastatic) vs metastatic: p = 0.013). [Figure 2-3] B) Normalized luminescence values of minor / major spliceosome luc reporter plasmids in different PCa cell lines of different PCa subtypes (RWPE n=9, LNCaP n=30, DU145 n=7, PC3 n=10, C4-2 n=22, 22Rv1 n=13, VCAP n=7, PM1262 n=13, PM154 n=20, MSK16 n=7, H660=12, mean ± SEM, One-way Anova; ns p>0.05, *p<0.05, **p<0.01, ***p<0.001). Each data point represents a single experiment, experiments were performed in triplicate. Abbreviations: Pri-PCa=primary prostate cancer, CRPC-Ad=castration-resistant adeno prostate cancer, CRPC-NE=castration-resistant neuroendocrine prostate cancer. [Figure 2-4] C) Normalized luminescence values of minor / major spliceosome luc-reporter plasmids in LNCaP and C4-2 PCa cells chronically treated with abiraterone, enzalutamide, and charcoal stripped (c / s) medium (LNCaP: abiraterone n=19, c / sn=17, enzalutamide n=7; C4-2: abiraterone n=24, c / sn=29, enzalutamide n=12, mean ± SEM, One-way Anova; ns p>0.05, *p<0.05, **p<0.01, ***p<0.001). Each data point represents a single experiment, performed in triplicate. [Figure 2-5]D) Normalized luminescence values of minor / major spliceosome luc-reporter plasmids in LNCaP and L-AR cells treated with siAR, c / s medium, and c / s medium + DHT (10 uM) (LNCaP n=34, L-AR n=19, L-AR+siAR n=7, L-AR c / s n=8, L-AR c / s+DHT n=13, mean ± SEM, normal one-way Anova; ns p>0.05, *p<0.05, **p<0.01, ***p<0.001). Each data point represents a single experiment performed in triplicate. Abbreviations: c / s=charcoal stripped medium, DHT=dihydrotestosterone. [Figure 2-6] E) Representative 40x confocal micrographs of PLA signals from heteromerization (red puncta) of AR and PDCD7 or HSP90 (positive controls) and the associated negative IgG rabbit controls. PLA was performed using primary antibodies anti-PDCD7 (rabbit monoclonal) and androgen receptor (mouse monoclonal), and oligonucleotide-conjugated PLA secondary probes [mouse, M(+) and rabbit, R(-)]. Scale bar indicates 25 μm. This experiment was repeated three times. [Figure 3-1]Figure 3 shows that inhibition of MiS by U6atac siRNA effectively alters the PCa transcriptome. A) U6atac snRNA expression as x-fold of siScrambled (control) normalized to the mean values of GAPDH and ACTB gene transcription in C4-2 cells (U6atac KD n=14, U6atac over. n=7). Data are presented using box plots displaying minimum, first quartile, median, third quartile and maximum. Statistical analysis was evaluated using unpaired two-tailed t-test, ns p>0.05, *p<0.05, **p<0.01, ***p<0.001. Each data point represents a single experiment, performed in triplicate. B) Expression of CoA3 spliced and unspliced transcripts as x-fold of siScrambled normalized to the mean GAPDH and ACTB gene transcripts in C4-2 cells treated with siU6atac RNA for 48, 72, and 96 hours (n=3, two-way anova; ns p>0.05, *p<0.05, **p<0.01, ***p<0.001). Experiments were performed in triplicate. [Figure 3-2]C) Box plots showing the 10th to 90th percentile missplicing index for all 431 minor introns that showed retention in LNCaP, C4-2, and 22Rv1 cell lines, and PM154 organoids after 96 h of treatment with siScrambled (scr) or siU6atac. Significance was determined by Kruskal-Wallis test followed by post-hoc Dunn test: asterisks (*) indicate significance compared to the appropriate siScrambled control for each siU6atac sample, plus signs (+) indicate significance compared to LNCaP siU6atac, and hashtags (#) indicate significance compared to PM154 siU6atac. *p<0.05, **p<0.01, ***p<0.001. D) Bar graph showing the distribution of alternative splicing (AS) events around all minor introns (from 2 exons upstream to 2 exons downstream) in LNCaP, C4-2, and 22Rv1 cell lines and PM154 organoids after 96 h of treatment with siScrambled (scr) or siU6atac. Different colors indicate different alternative splicing categories. E) Venn diagram (not to scale) depicting the overlap of MIGs with significantly increased minor intron retention after 96 h of siU6atac treatment in LNCaP, C4-2, and 22RV1 cell lines (left) and PM154 organoids (right) compared to the appropriate 96 h siScrambled control. F) Venn diagram (not to scale) depicting the overlap of MIGs with significantly increased alternative splicing following 96 h of siU6atac treatment compared to appropriate 96 h siScrambled controls in LNCaP, C4-2, and 22RV1 cell lines (left) and PM154 organoids (right). [Figure 3-3]G) STRING network showing the association of shared (i) prostate cancer-associated MIG with elevated minor intron retention (red) and (ii) downregulated protein-coding genes (grey; downregulated MIG in blue) in three cell lines LNCaP, C4-2, and 22RV1 (left) and PM154 organoid (right). H) Venn diagram (not to scale) showing overlapping biological pathways generated by Ingenuity Pathway Analysis (IPA) of the LNCaP, C4-2, 22RV1 (LCR) gene list (G; left) and PM154 (P) gene list (G; right). I) Venn diagram (not to scale) showing overlapping biological networks generated by Ingenuity Pathway Analysis (IPA) of the LNCaP, C4-2, 22RV1 (LCR) gene list (H; left) and PM154 (P) gene list (G; right). [Figure 4-1] Figure 4 shows that U6atac-mediated MiS inhibition effectively alters the PCa proteome in a cell-type-specific manner. A) Volcano plot showing the most significantly increased (top right) and decreased (top left) proteins in C4-2 cells treated with siU6atac (96 h) compared to siScrambled control (pooled data from three co-IP replicates). The x-axis represents log2 fold change (FC) values, and the y-axis represents -log10 of adjusted p-values. Grey dots represent non-differentially expressed proteins (non-DE), orange dots represent differentially expressed proteins (DE), black dots represent differentially expressed MIG-encoded proteins (DE MIG), and red dots represent differentially expressed MIG-encoded proteins with statistically significant increase in minor intron retention in transcription. B) Venn diagram (not to scale) showing overlap of proteins upregulated after U6atac KD in LNCaP, C4-2, 22Rv1, and PM154 cells as assessed by mass spectrometry analysis. C) Venn diagram (not to scale) showing overlap of proteins downregulated after U6atac KD in LNCaP, C4-2, 22Rv1, and PM154 cells as assessed by mass spectrometry analysis. [Figure 4-2] D) Immunoblot showing expression levels of selected proteins down- or upregulated by mass spectrometry in C4-2 cell line treated with siScrambled or siU6atac RNA for 96 hours. GAPDH was used as a loading control. E) Table summarizing differential expression of selected proteins after siU6atac treatment in LNCaP (L), C4-2 (C), 22Rv1 (R), and PM154 (P) cells. F) Venn diagram (not to scale) showing overlap of genes showing up- or downregulation (left) or downregulation (right) by mass spec (prot.) and transcriptionally (trans.) in LNCaP, C4-2, 22Rv1, and PM154 cells. [Figure 4-3] G) Quantification of flow cytometric analysis of C4-2 and PM154 cells treated with siU6atac for 72 and 96 hours and then stained with EdU and Hoechst (n=3, mean ± SEM, ordinary two-way anova; ns p>0.05, *p<0.05, **p<0.01, ***p<0.001). H) Pseudocolor plot showing G1 / 0, S and G2 selection gates in C4-2 cells treated with siU6atac or siScrambled for 72 hours. Quantification of flow cytometric analysis of PM154 organoids treated with siU6atac for 72 and 96 hours and then stained with EdU and Hoechst (n=3, mean ± SEM, ordinary two-way anova; **p=0.001, ****p<0.001). I) Pseudocolor plot showing the gates for G1 / 0, S and G2 selection in PM154 organoids treated with siU6atac or siScrambled for 72 h. Quantification of flow cytometry analysis of PM154 organoids treated with siU6atac for 72 and 96 h and subsequently stained with EdU and Hoechst (n=3, mean ± SEM, ordinary two-way anova; **p=0.001, ****p<0.001). [Figure 5-1]Figure 5 shows that single cell RNAseq corroborates siU6atac-mediated cell cycle defects and reveals PCa lineage dependency on MiS function. A) UMAP view of LNCaP cell lines showing clusters at optimal resolution (0.2). B) Histograms of LNCaP cell lines showing the contribution of each sample in each cluster. C) UMAP view of LNCaP cell lines showing cell cycle phase of LNCaP cell lines for each cell. D) Histograms of LNCaP cell lines showing the percentage of cell cycle phase in each cluster. [Figure 5-2] E) UMAP display of LNCaP cell line showing contribution of siScrambled and siU6atac samples (left); cell cycle phase of siScrambled and siU6atac samples (center); and cell phase of each cell (right). F) UMAP display of PM154 cell line showing contribution of siScrambled and siU6atac samples (left); cell cycle phase of siScrambled and siU6atac samples (center); and cell phase of each cell. [Figure 5-3] G) Histograms of LNCaP cell line showing percentage of cell cycle phases of siScrambled and siU6atac cells. Table shows number of cells in each cell cycle phase. p-values were calculated using Fisher's exact test. H) Histograms of PM154 cell line showing percentage of cell cycle phases of siScrambled and siU6atac cells. Table shows number of cells in each cell cycle phase. p-values were calculated using Fisher's exact test. [Figure 5-4] I) UMAP representation of LNCaP cell line showing AR scores calculated for siScrambled and siU6atac samples. Violin plot shows AR scores calculated for each cell cycle phase. p-values were calculated using Wilcoxon test. J) UMAP representation of LNCaP cell line showing EMT scores calculated for siScrambled and siU6atac samples. Violin plot shows EMT scores calculated for each cell cycle phase. p-values were calculated using Wilcoxon test. [Figure 5-5]J) UMAP of LNCaP cell line showing EMT scores calculated for siScrambled and siU6atac samples. Violin plots show EMT scores calculated at each cell cycle phase. p-values were calculated using Wilcoxon test. [Figure 6-1]Figure 6 shows that U6atac and MiS are possible therapeutic targets in cancer. A) Cell viability (mean ± SEM, ordinary two-way anova; ns p>0.05, *p<0.05, **p<0.01, ***p<0.001) in human PCa cell lines (LNCaP n=5, L-AR n=4, C4-2 n=6, 22Rv1 n=6, Pm154 n=6), human fibroblasts (HS27 n=6) and primary mouse prostate cells (MS2514 n=3, MS2513 n=3). Experiments were performed in triplicate. B) Growth curves of C4-2 cells stably overexpressing U6atac normalized to C4-2 cells treated at multiple time points with siU6atac normalized to C4-2 cells treated with siScrambled or normalized to C4-2 cells expressing EV plasmid. Data are pooled from 16 biologically independent experiments (C4-2 siU6atac) and 4 biologically independent experiments (C4-2 U6atac) (mean + / - SD, ordinary two-way anova; ns p>0.05, *p<0.05, **p<0.01, ***p<0.001). C) Growth curves of LNCaP and L-AR cells treated with siU6atac at several time points. Data are normalized to siScrambled control and represent pooled results from 4 biologically independent experiments (mean + / - SEM, ordinary two-way anova; ns p>0.05, *p<0.05, **p<0.01, ***p<0.001). D) Cell viability of different human cancer cell lines treated with siU6atac at several time points. Data are normalized to siScrambled control and represent pooled results from four biologically independent experiments (mean + / - SEM, ordinary two-way anova; ns p>0.05, *p<0.05, **p<0.01, ***p<0.001). E) Cell viability of human bladder cancer UMOC-9 cells treated with siRNA against U6atac, RNPC3, and Scrambled, or cisplatin (1uM). Data represent pooled results from four biologically independent experiments (mean + / - SD, ordinary one-way anova; ns p>0.05, *p<0.05, **p<0.01, ***p<0.001).F) Growth curves of LNCaP (n=4), L-ENZ (n=6), and C4-2 (n=4) cells treated with siScrambled, siU6atac, siEZH2+ / -Enzalutamide (10uM). Data represent pooled results from biologically independent experiments (mean+ / -SEM, ordinary two-way anova; ns p>0.05, *p<0.05, **p<0.01, ***p<0.001). [Figure 6-2] G) Scatter plot of RNAseq results of patient-derived PCa organoids (large dots) overlaid with clinical data (small dots) from the SU2C dataset. Y-axis represents AR score, X-axis represents NEPC score. H) Brightfield micrographs of MSK8 cells 8 days after siRNA treatment against U6atac. Cells treated with scrambled siRNA are shown as control. Scale bar (red): 50 μm. Bar blots summarize cell counts at day 8 (n=3, paired t-test, *p=0.0307). I) Cell viability of patient-derived PCa organoids. Data are normalized to scrambled control and represent pooled results from three biologically independent experiments (mean + / - SEM, two-way anova; ns p>0.05, *p<0.05, **p<0.01, ***p<0.001). [Figure 7-1] Figure 7 shows that U6atac expression is increased in cancer. A) Representative images of the expression of the RNU6atac BaseScope probe in human ovarian, small cell lung cancer (SCLC), (ADC), skin, uterus and prostate benign tissues (n=15) and tumor tissues (n=33). Scale bar represents 20 μm. Bottom panel shows U6atac score from TMA analysis (Wilcoxon test, p=0.0038). [Figure 7-2] B) U6atac intensity in different samples (Wilcoxon test, stomach: p=8.78E-9, follicular thyroid carcinoma: p=0.63, poorly differentiated thyroid carcinoma: p=6.26E-6). [Figure 8]Figure 8 shows that U6atac expression is increased in metastatic PCa. U6atac score in all samples tested (Wilcoxon test, primary vs primary with potential metastasis: p=0.0004, primary vs metastatic: p<2.2E-16, primary with potential metastasis vs metastatic: p=0.013). [Figure 9-1] Figure 9 shows that expression of MiS components and MiS activity increase with PCa progression. A) Cell viability of human PCa cell lines treated with 20uM enzalutamide for 48 hours as measured using CellTiter-GLO. Data normalized to untreated WT control (ctr.). (LNCaP n=6, L-AR n=3, L-rENZ n=6, C4-2 n=3, 22Rv1 n=4, PM154 n=6, PC3 n=3, PC3Pro4 n=3, mean±SEM, ordinary one-way anova; ns p>0.05, *p<0.05). Experiments were performed in triplicate. B) Normalized luminescence values of minor / major spliceosome luc reporter plasmids in PC3 cells (n=10) and its subline PC3PPro4 (n=4) (top) and LNCaP cells (n=37), its AR-overexpressing subline L-AR (n=19) and its enzalutamide-resistant subline L-rENZ (n=7, mean ± SD, ordinary one-way anova; ns p>0.05, ***p, ****p<0.0001). Each data point represents a single experiment, performed in triplicate. [Figure 9-2] C) Splice index of hSCN4A and p120 reporter plasmids in PC3, PC3Pro4, LNCaP, and L-rENZ cells (n=3, unpaired two-tailed t-test, ****p<0.0001). Splice index was calculated as the ratio of spliced to unspliced expression values of hSCN4A and p120 in cells transfected with hSCN4A and p120 minigene reporter plasmids for 72 h. Expression values were normalized to the average GAPDH and ACTB gene transcripts and calculated as x-fold of siScrambled (siScr.). [Figure 10-1]Figure 10 shows that siU6atac RNA reduces MiS activity. A) Expression of U6atac and p120 spliced and unspliced as x-fold of siScrambled (siScr.) normalized to the average GAPDH and ACTB gene transcripts in 22Rv1, PM154, L-AR, LNCaP, L-rENZ, and H660 cells treated with siU6atac RNA for 96 hours. Data are presented using box plots. Box plots show minimum, 1st quartile, median, 3rd quartile and maximum (n=4, ordinary one-way ANOVA, 22Rv1 / U6atac: *p=0.0193, 22Rv1 / CoA3 spliced: *p=0.0315, LNCaP: **p=0.0017, H660: ***p=0.0003, ****p<0.0001). Each data point represents a single experiment, performed in triplicate. B) U6atac as x-fold of siScrambled (siScr.), and spliced and unspliced expression of hSCN4A, hSCN8A, p120 were normalized to the average values of GAPDH and ACTB gene transcription in C4-2 cells transfected with hSCN4A, hSCN8a, and p129 minigene reporter plasmids and treated with siU6atac RNA for 96 hours (hSCN4A n=3, hSCN8A n=4, p120 n=5). Data are presented using box plots. Box plots show minimum, 1st quartile, median, 3rd quartile and maximum (unpaired two-tailed t-test, ns p>0.05, ***p<0.001, ****p<0.0001. Each data point represents a single experiment, performed in triplicate. C) Normalized luminescence values of minor spliceosome luc-reporter plasmid in LNCaP (siU6atac n=5, anisomycin n=8) and PM154 (siU6atac n=5, anisomycin n=4) cells treated with siU6atac (96 hr) or anisomycin (1ug / ml, 4 hr) (mean ± SEM, normal one-way anova; ns p>0.05, ***p<0.001, ****p<0.0001). Each data point represents a single experiment, performed in triplicate. [Figure 10-2]D) U6atac snRNA expression as x-fold of siScrambled (siScr.) normalized to the mean values of GAPDH and ACTB gene transcription in LNCaP, C4-2, 22Rv1 and PM154 cells treated with siU6atac RNA for 48 and 96 hours (n=5). Data are presented using box plots. Box plots show minimum, 1st quartile, median, 3rd quartile and maximum (One-way Anova; ****p<0.0001). Experiments were performed in triplicate. [Figure 11] Figure 11 shows that siU6atac induces intron retention and AS in PCa. A-D) Overlap (left) and total (right) numbers of (A) up-regulation of protein-coding genes, (B) down-regulation of protein-coding genes, (C) up-regulation of MIG, and (D) down-regulation of MIG in LNCaP, C4-2, and 22RV1 cell lines and PM154 organoids, respectively. [Figure 12] Figure 12 shows that siU6atac reduces the expression of U6atac and spliced CoA3 transcripts. Expression of U6atac snRNA, CoA3 spliced and unspliced as x-fold of siScrambled (siScr.) normalized to the average GAPDH and ACTB gene transcripts (n=3) in LNCaP, C4-2, 22Rv1 cells and PM154 organoids treated with siU6atac RNA for 96 hours. Box plots show minimum, 1st quartile, median, 3rd quartile and maximum (normal one-way anova; 6Uatac: *p=0.0294, ***p=0.004(22Rv1), 0.0002(LNCaP, ****p<0.0001; CoA3 spliced: ***p=0.0025(PM154), 0.0035(22Rv1), ***p=0.0003) [Figure 13-1]Figure 13 shows that siU6atac affects protein expression in PCa. A) Immunoblots of BRCA1, Brd9, EED, AR, NKX3.1, Rb1, AURKA, BRAF, CDN2A, CDN1A, and JNK1 / 2 in LNCaP, C4-2, 22Rv1 cells and PM154 organoids treated with siU6atac or siScrambled (scr.) for 96 hours. [Figure 13-2] B) Quantification of immunoblots. Values were normalized to GAPDH (ordinary one-way anova, C4-2: *p=0.0485 (Rb1), 0.0196 (CDN1A); 22Rv1: *p=0.0230). [Figure 14-1] Figure 14 shows scRNAseq in PM154 organoids. A) UMAP view of PM154 cell lines showing clusters at optimal resolution (0.2). B) Histogram of PM154 cell lines showing the contribution of each sample in each cluster. [Figure 14-2] C) UMAP view of the LNCaP cell line showing the cell cycle phase of each cell. D) Histogram of the LNCaP cell line showing the percentage of cell cycle phase in each cluster. [Figure 14-3] E) UMAP representation of PM154 cell line showing AR scores calculated for siScrambled and siU6atac samples. Violin representation of PM154 cell line showing AR scores calculated for each cell cycle phase. p-values were calculated using Wilcoxon test. [Figure 14-4] F) UMAP representation of PM154 cell line showing EMT scores calculated for siScrambled and siU6atac samples. Violin representations show EMT scores calculated for each cell cycle phase. p-values were calculated using Wilcoxon test. G) AR scores of LNCaP and PM154 cell lines calculated for siScrambled and siU6atac samples using bulk RNAseq. [Figure 15] Figure 15 shows scRNAseq analysis of MIG expression in LNCaP cells. Violin plots showing expression levels of selected MIGs indicating significant PCa nodes (Figure 1B, C) in LNCaP cells. [Figure 16] Figure 16 shows scRNAseq analysis of MIG expression in PM54 organoids. Violin plot showing expression levels indicating significant PCa nodes (Figure 1B, C) in PM154 cells. [Figure 17] Figure 17 shows that siU6atac and siRNPC3 affect PCa cell proliferation. A) U6atac snRNA and RNPC3 expression as x-fold of siScrambled (siScr.) normalized to the mean value of GAPDH and ACTB gene transcription in LNCaP, L-AR, C4-2, 22Rv1, PM154, HS27, and MS2514 cells and organoids treated with siU6atac (n=5) and siRNPC3 (n=3) RNA for 96 hours. Data are presented using box plots. Box plots show minimum, 1st quartile, median, 3rd quartile and maximum (2-way anova; LNCaP: **p=0.0076, ****p<0.00'1; L-AR: ***p=0.0006, ****p<0.0001; C4-2: ns p>0.05, ****p<0.0001; 22Rv1: **p=0.0048, ****p<0.0001; PM154: **p=0.0024, ****p<0.0001; HS27: *p=0.0190, ***p=0.0005; Ms2514: **p=0.0025, ***p=0.0007). B) Growth curves of LNCaP, C4-2, L-AR, 22Rv1 cells and PM154 organoids treated with siU6atac for 0, 48, 72, and 96 hours. Data are normalized to siScrambled control and pooled from 8 (siU6atac) and 6 (siRNPC3) biologically independent experiments (mean + / - SEM, ordinary two-way anova; ns p>0.05, **p<0.01, ****p<0.001). [Figure 18-1]FIG. 18 shows that MiS inhibition blocks cancer cell viability and proliferation. A) U6atac snRNA expression as x-fold of siScrambled (siScr.) normalized to the mean values of GAPDH and ACTB gene transcription in LNCaP (n=4) and L-AR (n=8) cells treated with siU6atac RNA for 96 hours. Data are presented using box plots. Box plots show minimum, 1st quartile, median, 3rd quartile, maximum (ordinary one-way anova, ****p<0.0001). B) Proliferation curves of LNCaP and L-AR cells (n=3, mean + / - SEM, ordinary two-way anova; ****p<0.00001). C) Expression of U6atac and U12 snRNA, and RNPC3 and PDCD7 as x-fold of siScrambled (siScr.) normalized to the mean gene transcription of GAPDH and ACTB in C4-2, DLD1, HEK293, K-562, LN18, MDA-MB-231, and UMOC-9 cells treated with siU6atac, siPDCD7, siRNPC3, and siU12 RNA for 96 hours (n=3). Data are presented using box plots. Minimum, first quartile, median, third quartile, and maximum are displayed (unpaired two-tailed t-test, p-values are indicated in the figure). [Figure 18-2]D) Cell viability assay measured using CellTiter-GLO in C4-2, DLD1, HEK293, K-562, LN18, and MDA-MB-231 treated cells treated with siPDCD7, siRNPC3, and siU12 RNA (48, 72, 96, and 120 hours). Data were normalized to siScrambled (siScr.) control and represent pooled results from four biologically independent experiments (mean + / - SEM, two-way anova; p-values shown in figures). E) Expression of U6atac snRNA, CoA3 spliced and unspliced after siU6atac (96 h) treatment, and EZH2 expression after siEZH2 (96 h) treatment as x-fold higher than siScrambled (siScr.) and normalized to the mean values of GAPDH and ACTB gene transcripts in LNCaP (n=4), C4-2 (n=4) and L-rENZ (n=6) cells. Data were presented using box plots. Minimum, first quartile, median, third quartile and maximum values are displayed (unpaired two-tailed t-test, p-values shown in figures). E) Expression of U6atac snRNA, CoA3 spliced and unspliced as x-fold of siScrambled (siScr.) normalized to the mean gene transcripts of GAPDH and ACTB in MSK10 (n=3), MSK14 (n=3), MSK16 (n=5), MSK8 (n=3), and PM154 (n=4) patient-derived PCa organoids treated with siU6atacRNA for 96 hours. Data are presented using box plots. Minimum, 1st quartile, median, 3rd quartile, and maximum are displayed (unpaired two-tailed t-test, p-values are indicated in the figure). [Figure 19] Figure 19 shows that MiS inhibition has a stronger effect on cancer than normal cells. Brightfield micrographs of C4-2, HS27, MS2514, and MS2513 cells 96 hours after treatment with siU6atac and siRNPC3. Cells treated with scrambled siRNA are shown as controls. Scale bar (red): 100 μm. [Figure 20]Figure 20 shows that minor splicing correlates with PCa progression. Spliced CoA3 as x-fold expression of unspliced normalized to the mean value of GAPDH and ACTB gene transcripts in RWPE (n=6), LNCaP (n=5), Du145 (n=6), PC3 (n=3), PCc8 (n=4), L-AR (n=3), C4-2 (n=8), 22Rv1 (n=5), PM154 (n=5), MSK16 (n=1), and H660 (n=4) cells and organoids. Each data point represents a single experiment, performed in triplicate. [Figure 21] Figure 21 shows MIG expression in LNCaP, C4-2, 22Rv1, and PM154 cells. Heatmap shows log10 transformed TPM values. Colors represent percentiles (10th-90th), with blue indicating low expression and red indicating high expression. UpSig list shows genes significantly more highly expressed in one cell line compared to all other cell lines by ANOVA (P<0.05) and post-hoc Tukey test (P<0.05) with BH P-value adjustment. [Figure 22-1]FIG. 22 shows that p38MAPK regulates MiS in PCa-NE. A) Normalized luminescence values of minor / major spliceosome luc reporter plasmid in LNCaP n=8 / 3, C4-2 n=6 / 4, Pm154 n=5 / 3 (mean±SEM, two-way Anova; ns p>0.05, *p<0.05, ****p<0.0001). Each data point represents a single experiment, performed in triplicate. B and C) Normalized luminescence values of minor / major spliceosome luc reporter plasmid in LNCaP, C4-2, 22Rv1, and Pm154 cells (n=3) treated with siScrambled or siMAPK13 / 14 for 96 hours (mean±SEM, one-way Anova; ns p>0.05, ****p<0.0001). Each data point represents a single experiment, performed in triplicate. The bottom panel shows MAPK13 / 14 snRNA expression normalized to the mean values of KD control: GAPDH and ACTB gene transcripts as x-fold increase over siScrambled. Data are presented using box plots. Box plots show minimum, 1st quartile, median, 3rd quartile and maximum (One-way Anova, ns; not significant, p>0.05, *p<0.05, **p<0.01, ***p<0.001). Experiments were performed in triplicate. [Figure 22-2] D) Schematic diagram of U6atc turnover regulation. AR blocks U6atac turnover in CRPC, where MiS activity is increased, whereas p38 is increased in NEPC. [Figure 23-1]FIG. 23 shows the relative performance of MIGs and non-MIGs for clustering gene expression data from A) 9 cancer types (biliary tract, breast, colorectal, lung, ovarian, pancreatic, gastric, thymic, and prostate) and B) 23 cancer types from the PCAWG consortium. Each box corresponds to a simulation of length 1000 randomly sampling non-MIGs among all non-MIGs in the genome. The plot shows a positive relationship between the relative abundance of MIGs and the quality of clustering, quantified using the Silhouette Coefficient. P values are based on a two-tailed t-test, with each sample's value taken as the difference between the Silhouette Coefficient of one of 1000 100% non-MIG samples and the corresponding Silhouette Coefficient when only MIGs are used for clustering (i.e., the Silhouette Coefficient corresponding to 0% non-MIGs). [Figure 23-2] FIG. 23 shows the relative performance of MIGs and non-MIGs for clustering gene expression data from A) 9 cancer types (biliary tract, breast, colorectal, lung, ovarian, pancreatic, gastric, thymic, and prostate) and B) 23 cancer types from the PCAWG consortium. Each box corresponds to a simulation of length 1000 randomly sampling non-MIGs among all non-MIGs in the genome. The plot shows a positive relationship between the relative abundance of MIGs and the quality of clustering, quantified using the Silhouette Coefficient. P values are based on a two-tailed t-test, with each sample's value taken as the difference between the Silhouette Coefficient of one of 1000 100% non-MIG samples and the corresponding Silhouette Coefficient when only MIGs are used for clustering (i.e., the Silhouette Coefficient corresponding to 0% non-MIGs). [Figure 24]Figure 24 shows the expression of U11, U12, U4atac and U6atac as x-fold of their respective LNCaP and PC3 expression values, normalized to the mean values of GAPDH and ACTB gene transcripts in L-AR (n=5), L-rENZ (n=4) and PC3Pro4 (n=5) cells. Data are presented using box plots. Box plots show minimum, 1st quartile, median, 3rd quartile and maximum (ordinary one-way Anova, ns p>0.05, *p<0.05). Each data point represents a single experiment, performed in triplicate. [Figure 25-1] Figure 25 shows that MiS affects REST splicing: A) Expression of canonical REST and REST4 as x-fold in LNCaP cells normalized to the average gene transcription of GAPDH and ACTB in LNCaP, C4-2, 22Rv1, and PM154 cells (n=4, One-way Anova, ns p>0.05, *p<0.05, **p<0.01). Each data point represents a single experiment, performed in triplicate. [Figure 25-2] B) Scatter plot showing the relationship between the expression levels of REST / REST4 genes and the AR / NEPC score (n=249, Pearson correlation test). Box plot showing the expression levels of REST / REST4 genes in samples AR vs. NEPC positive (score >=0.4, Wilcoxon test). [Figure 25-3]C) Canonical REST expression after siU6atac treatment as x-fold of siScrambled ctr cells normalized to the mean values of GAPDH and ACTB gene transcripts in LNCaP (n=4), C4-2 (n=5), 22Rv (n=6), and PM154 (n=3) cells (One-way Anova, ns p>0.05, *p<0.05, ***p<0.001). Each data point represents a single experiment, performed in triplicate. D) Bulk RNAseq TPM values of REST4 and REST? isoforms in LNCaP, C4-2, 22Rv1, and PM154 cells treated with siU6atac and siScrambled for 48 hours. Red font indicates cases with a fold change threshold above 1.5. E) REST4 expression after siU6atac (48-72 h) treatment as x-fold of siScrambled ctr cells normalized to the average gene transcription of GAPDH and ACTB in LNCaP, C4-2, 22Rv, and PM154 cells (n=5, one-way anova, ns p>0.05, *p<0.05). Each data point represents a single experiment, performed in triplicate. F) qRT-PCR products using primers binding to exon 3 (fw) and exon 4 (rv) of REST loaded onto a 3% agarose gel. Experiments were performed in LNCaP, C4-2, 22Rv1, and PM154 cells treated with siScrambled or siU6atac (48 h). The size of the canonical REST is expected to be 150 bp, and the size of the REST4 samples is expected to be 200 bp. [Figure 26-1]Figure 26 shows AR and minor splicing. A) U6atac expression as x-fold of LNCaP cells normalized to the mean of GAPDH and ACTB gene transcription in LNCaP and L-rENZ cells, and L-AR cells treated with siSCrambled, siAR (n=2), c / s medium, and / or DHT (n=4) (unpaired t-test, *p<0.05). Each data point represents a single experiment, performed in triplicate. B) Immunoblots showing expression levels of PARP (MIG control, n=5), EZH2 (n=8), Rb1 (n=3), and AR (n=8) in LNCaP and C4-2 cell lines treated with siScrambled RNA or siU6atac RNA for 96 hours. GAPDH was used as a loading control (unpaired multiple t-test, ****p<0.001). [Figure 26-2] B) Immunoblots showing expression levels of PARP (MIG control, n=5), EZH2 (n=8), Rb1 (n=3), and AR (n=8) in LNCaP and C4-2 cell lines treated with siScrambled RNA or siU6atac RNA for 96 hours. GAPDH was used as a loading control (unpaired multiple t-test, ****p<0.001). [Figure 27-1] Figure 27 shows A) confluence (y-axis) and B) viability (y-axis) of different cell lines treated with siU6atac / siScrambled for 120 hours (n=3). Cells are ranked according to doubling time (DT). Normal cells are depicted in green. C) U6atac expression as x-fold of siScrambled after siU6atac treatment of all cells used in that experiment. Each data point represents a single experiment, performed in triplicate. Data are presented using box plots. Box plots display minimum, 1st quartile, median, 3rd quartile and maximum. [Figure 27-2]Figure 27 shows A) confluence (y-axis) and B) viability (y-axis) of different cell lines treated with siU6atac / siScrambled for 120 hours (n=3). Cells are ranked according to doubling time (DT). Normal cells are depicted in green. C) U6atac expression as x-fold of siScrambled after siU6atac treatment of all cells used in that experiment. Each data point represents a single experiment, performed in triplicate. Data are presented using box plots. Box plots display minimum, 1st quartile, median, 3rd quartile and maximum. [Figure 28] FIG. 28 shows A) growth curves of mCherry-positive HS27 cells and GFP-positive C4-2 cells grown in co-culture treated with siScrambled or siU6atac. Data are expressed as x-fold of siScrambled. Data represent pooled results from biologically independent experiments. (n=3, mean + / - SEM, ordinary two-way anova; ns p>0.05, *p<0.05) B) U6atac expression as x-fold of siScrambled after siU6atac treatment of all cells used in that experiment. Each data point represents a single experiment, and experiments were performed in triplicate. Data are presented using box plots. Box plots show minimum, 1st quartile, median, 3rd quartile, and maximum (n=3, mean ± SEM, ordinary one-way anova; ****p<0.0001). [Figure 29] FIG. 29 shows rescue experiments with hairpins against RNU6atac that demonstrated that significant MIG expression was dependent on U6atac. [Figure 30-1]FIG. 30: A) Brightfield and fluorescent micrographs of C42'U6atac cells (scale=50uM). B) U6atac expression in C4-2 EV and C42'U6atac cells (n=3). Each data point represents a single experiment, performed in triplicate. Data are presented using box plots. Box plots show minimum, first quartile, median, third quartile and maximum (n=3, mean + / - SEM, ordinary one-way anova; *p<0.05). C) Growth curves of C4-2'U6atac cells (blue) and C4-2 cells treated with siU6atac (pink). Data are presented as x-fold increase of C4-2'EV or siScrambled, respectively. Data represent pooled results from biologically independent experiments (n=3, mean + / - SEM, ordinary two-way anova; ****p<0.0001). D) Experimental setup for rescue experiments. [Figure 30-2] E) 1% agarose gel showing endogenous U6atac (1427bp) and CRISPERylated U6atac (707bp) amplified in untreated (NTC) C4-2 and C4-2'U6atac cells transfected with empty backbone vector Px458 or gRNA targeting U6atac (CRISPR), F) U6atac expression in C4-2 and C42'U6atac cells treated with gRNA against U6atac (n=3). Data are presented as x-fold of cells treated with empty vector Px458. Each data point represents one experiment; experiments were performed in triplicate. Data were presented using box plots. Box plots display minimum, first quartile, median, third quartile, and maximum. (n=3, mean + / - SEM, ordinary one-way anova; ***p<0.001), G) Brightfield micrographs (scale=50uM) of C4-2 and C4-2'U6atac cells treated with either empty vector Px458 or gRNA against U6atac, H) Viability of C4-2 and C4-2'U6atac cells treated with gRNA against U6atac. Data are expressed as x-fold of cells treated with empty vector Px458 (n=2, experiments performed in triplicate). [Figure 31-1]Figure 31 shows the Pearson correlation analysis between SNRNP48, SNRNP25, SNRNP35, UBE2K, ZMAT5, PDCD7 (MiS protein), ZRSR2, CENPA (MiS and major spliceosome protein), and EZH2. Organoids are color-coded by transcriptome NEPC score, with a score >0.4 indicating a CRPC-NE phenotype and a score <0.4 indicating a CRPC-adeno phenotype (Beltran et al., PMID26855148). Organoids are sized by AR score. [Figure 31-2] Figure 31 shows the Pearson correlation analysis between SNRNP48, SNRNP25, SNRNP35, UBE2K, ZMAT5, PDCD7 (MiS protein), ZRSR2, CENPA (MiS and major spliceosome protein), and EZH2. Organoids are color-coded by transcriptome NEPC score, with a score >0.4 indicating a CRPC-NE phenotype and a score <0.4 indicating a CRPC-adeno phenotype (Beltran et al., PMID26855148). Organoids are sized by AR score. [Figure 31-3] Figure 31 shows the Pearson correlation analysis between SNRNP48, SNRNP25, SNRNP35, UBE2K, ZMAT5, PDCD7 (MiS protein), ZRSR2, CENPA (MiS and major spliceosome protein), and EZH2. Organoids are color-coded by transcriptome NEPC score, with a score >0.4 indicating a CRPC-NE phenotype and a score <0.4 indicating a CRPC-adeno phenotype (Beltran et al., PMID26855148). Organoids are sized by AR score. EXAMPLES
[0071] Example 1: MIG plays an important role in cancer initiation and progression Since MIG is highly enriched in cell cycle regulation and survival, we investigated whether MIG is enriched in biological pathways utilized by cancer-causing genes. To this end, we investigated whether MIG-encoded proteins are enriched among proteins that interact with proteins encoded by cancer-causing genes in 160,881 protein-protein interaction (PPI) networks between 15,366 human proteins currently in the HINT database (Das, J. et al., BMC Syst Biol 2012, 6, 92). We hypothesized that through these protein-protein interactions, MIG is the primary recipient of molecular information cascaded from cancer-causing genes. We considered 403 cancer-causing genes in the current set of 15,366 interacting proteins in the Cancer Genome Interpreter database (Bigas, N et al., Genome Medicine 2018, 10(1), 25) and calculated the shortest distance d of all proteins encoded by the cancer-causing genes (where d = 1 indicates a direct interaction). In each distance bin, we measured the enrichment of 542 MIG-encoded proteins (inset, Figure 1A) and found that MIGs were significantly enriched in the immediate vicinity of the network of cancer-causing genes (permutation test: d = 1, 2; p < 10E-5). However, MIGs were significantly diluted at greater distances (d > 3, p < 10E-5) (inset, Figure 1A). We then focused on the immediate interactors of proteins encoded by the 26 PCa-causing genes (inset, Figure 1A) (Bigas, N et al., supra). We found that MIG-encoded proteins were significantly enriched as immediate interactors of these 26 PCa proteins. Because MIG was significantly enriched from cancer-causing genes at d=1, we next identified three PPI sub-networks of interactions between MIG-encoded proteins and the 26 PCa proteins.Indeed, we found that MIG-encoded proteins formed a significantly large (permutation test, p<6.8x10E-4) tight interaction web between 17 PCa proteins and 72 MIG-encoded proteins, capturing more than 95% of the proteins (Fig. 1B, C). Similarly, a significant (permutation test, p<3x10E-5) large tight sub-network was obtained between MIG-encoded proteins and proteins encoded by all 403 cancer-causing genes (Fig. 1A). Such findings suggest that proteins encoded by cancer-causing genes in general, and PCa-causing genes in particular, tend to significantly interact with MIG-encoded proteins in the underlying human interaction network.
[0072] The direct interaction of MIG with cancer-causing proteins led us to explore to what extent MIG (compared to non-MIG) shows higher differential expression between different cancer types. To address this question, we first performed hierarchical sample clustering (HSC) of nine different solid cancer types using normalized MIG expression data, and then assessed how well this MIG expression could cluster samples by cancer type. We first assessed the quality of the resulting clustering by visualizing the clustered data in a dendrogram and associated heatmap, where the structure could be easily seen in cases where only MIG was used to cluster samples (Figure 1D). To assess whether MIG clusters data more successfully versus non-MIG, we performed the same type of visual analysis on datasets that differ in the relative abundance of MIG and non-MIG. In particular, this controlled approach involved gradually contaminating the pool of MIGs with increasing numbers of non-MIG samples such that 0%, 10%, 20%, 30%, ... 100% of all genes in a given gene set were non-MIG. In taking this approach, we discovered that a gradual increase in the relative abundance of non-MIGs (within the gene set used for clustering) progressively worsens the quality of the clustering of the data. This can be seen visually by observing the deteriorating structure of the heatmap associated with various relative abundances of non-MIGs.
[0073] We therefore measured this degradation of clustering (i.e., visual structure in the heatmap) with a more objective and quantitative approach. To quantitatively assess how well a given expression dataset is clustered, we used the Silhouette Coefficient, where higher coefficient values point to a more clearly defined correct clustering. We used simulations (a scheme based on resampling of the data) to generate 1000 gene sets for each fraction of non-MIG genes, and then calculated the Silhouette Coefficient associated with each resampled gene set. This quantitative approach reproduced what we observed visually in the heatmap: the Silhouette Coefficient associated with the clustering of expression data on gene sets from the non-MIG-rich fraction resulted in less effective clustering in these nine different cancer types (Figure 1E; two-tailed t-test p<2.2E-16). That MIGs enable better clustering performance than non-MIGs suggests that MIG expression is more informative in defining the differences between these nine cancer types.
[0074] We next assessed whether MIG expression might also show greater gene expression differences (compared to non-MIG) across different PCa progression stages. This was performed on transdifferentiation analysis across different transcriptomes as indicated by principal component analysis. In particular, to obtain data representative of different stages of PCa progression, we used prostate samples from GTEx (normal tissues), TCGA (primary PCa samples), and SU2C (CRPC-adeno and CRPC-NE) datasets. Similar to our clustering analysis of nine different cancer types detailed above, we observed that MIG expression results in better clustering performance in these different stages of prostate cancer progression, although the visual disparity between heatmaps associated with non-MIG 0% and non-MIG 100% is not as pronounced as we observed in our pan-cancer analysis. The same simulation-based scheme in which the silhouette coefficient was measured in our pan-cancer analysis was similarly applied across these broad cancer progression stages. Here again, we observed that MIG expression resulted in higher clustering performance than non-MIG expression (Figure 1F and G; two-tailed t-test p<2.2E-16). Together, these findings indicate that MIG transcription is closely related to cancer development and progression.
[0075] Example 2: Expression of MiS components correlates with PCa progression The upregulation of MIG with PCa progression and transdifferentiation requires increased MiS activity, which would therefore be controlled by the levels of MiS components. We therefore queried the expression dynamics of U11 snRNA (RNU11), a key MiS component, in the TCGA data of PCa (PRAD). We observed a significant association (Wilcoxon test, p=0.023) between high-grade (Gleason scores 8 and 9) PCa patients and high RNU11 expression (Figure 1). Another unique MiS component is the protein RNPC3, which is associated with cancer-associated scleroderma. We therefore queried the RNA-seq data of 18 PCa organoids (CRPC-adeno and CRPC-NE), and found that the expression of RNPC3 tended to be positively correlated with SOX2, a pluripotent stem cell and early differentiation marker, and EZH2, an epigenetic regulator, respectively (Figure 1). We observed that expression of RNPC3 parallels PCa disease progression with the lowest expression in benign and hormone-sensitive PCa cells (LNCaP), intermediate expression in malignant CRPC-adeno cells (22Rv1), and highest expression in CRPC-NE cells (H660) and patient-derived organoid lines (WCM154 and Msk16) (Figure 1). The same was observed for U6atac snRNA. U6atac snRNA is regulated with a high turnover rate (and is therefore a key control node for MiS activity). High expression of RNU6atac and RNPC3 further correlates with high NEPC and low AR scores, which are indicators of poor PCa prognosis due to the plasticity of the CRPC-NE lineage (Figure 1). Moreover, its increased expression is particularly informative, since low levels of U6atac essentially reduce the activity of MiS. Indeed, quantitative RT-PCR data confirmed that expression of MiS components correlated with PCa progression: expression of four MiS snRNAs - U6atac, U4atac, U11, and U12 - was higher in CRPC-adeno and CRPC-NE cell lines compared with hormone-responsive LNCaP and RWPE cells, or AR-low Du145 and PC3 cells (Figure 1).
[0076] To test the prediction that U6atac expression increases with cancer development and progression, we developed an in-situ probe specific for U6atac. We used this probe to investigate various cancer types and found that U6atac expression was significantly higher in cancer tissues compared to matched benign samples (Wilcox test, p=0.0038). This finding indicates the strong specificity of U6atac expression to highly proliferative tissues (pan-cancer tissue microarray (TMA)) (Figures 1 and 7). Consistently, thyroid and colon cancers showed higher U6atac intensity scores in cancers compared to corresponding benign tissues (Figure 7). Taken together, our data indicate that expression of both MiS components and MIG is robustly associated with cancer development and progression.
[0077] Example 3: MiS activity correlates with PCa progression Consistent with the increased expression of U6atac in more aggressive PCa cell lines (Figure 1), we also observed significantly increased expression of U6atac in primary and metastatic PCa patient samples compared to benign prostate (Wilcox test, p=0.0004). More importantly, we found that metastatic PCa had significantly elevated U6atac expression compared to primary PCa (Wilcox test, p=0.013) (Figures 2A and 8). Therefore, we hypothesized that MiS activity would increase with PCa progression. To test this, we used minor and major splicing reporter plasmids (kindly provided by the Dreyfuss laboratory) in a variety of PCa cell lines that reflect the extent of PCa progression and transdifferentiation. We found a trend for minor intron splicing to be more efficient in therapy-resistant C4-2 and 22Rv1 cells compared to hormone-sensitive LNCaP cells (Figure 2B) or AR-low DU145 and PC3 cells. We also observed a six-fold increase in minor intron splicing in CRPC-NE cell lines and patient-derived organoid lines (PM-154, PM1262, MSK16, and NCI-H660) (Figure 2B). In contrast, major intron splicing was unaffected in all cell lines and organoid lines tested (Figure 2B). This indicates that, unlike the major spliceosome, the activity of the minor spliceosome dynamically increases with PCa progression. This prompted us to explore the molecular mechanisms underlying the regulation of MiS activity in PCa.
[0078] AR signaling plays a key role in PCa progression, which is often the apex of the oncogenic pathway. Therefore, we hypothesized that the activity of MiS may be related to AR signaling throughout PCa progression. To simulate the stress response and reactivation of AR signaling in PCa, we mimicked the therapeutic resistance mechanism and subjected PCa cells to charcoal-stripped (C / S) medium, ARSi with abiraterone and enzalutamide, and long-term androgen deprivation therapy (ADT). We observed a significant increase in MiS activity in cells exposed to ADT / ARSi, whereas this treatment had limited effect on major splicing (One-way Anova; * p=0.0117, *** p and **** p<0.001) (Fig. 2C). Similar associations were observed in PC3 cells and its well-differentiated subline, PC3Pro4, as well as in hormone-sensitive LNCaP cells and their sublines overexpressing AR (L-AR and / or ENZ-resistant (L-rENZ)) (Fig. 9A and B). PC3Pro4, L-AR, and L-rENZ showed a significant increase in MiS activity, whereas major spliceosome activity was unchanged (Fig. 9B) (One-way Anova; *** p and **** p<0.001). This increase in MiS activity in these lines was further confirmed by minor intron containing p120 and hSCN4A minigene constructs. We observed that L-AR, L-rENZ, and PC3Pro4 cells displayed higher splicing efficiency in both reporter assays compared to their native LNCaP and PC3 lines (Figure 9C). Finally, both L-rENZ and PC3Pro4 cells expressed higher MiS snRNA levels compared to their respective wild type (WT) lines (Figure 24).
[0079] We next investigated the relationship between AR signaling and minor intron splicing. Here, we used a luciferase reporter as a readout of minor intron splicing in therapy-sensitive LNCaP cells. Overexpression of AR led to a significant increase in MiS activity (Figure 2D II) (One-way Anova; **** p<0.001), which was not observed with the major spliceosome reporter (Fig. 2D II). MiS activity was reduced by siRNA against AR (Fig. 2D III). Knockdown (KD) of AR was confirmed by qRT-PCR. To further explore the control of MiS activity by AR signaling, we treated L-AR cells with hormone-deprived medium to block AR activation. This was confirmed by a decrease in KLK3 expression. Again, we found that blocking AR activation led to a decrease in MiS activity (Fig. 2D IV). Finally, by adding dihydrotestosterone (DHT) to hormone-deprived medium, MiS activity was rescued to the level observed with L-AR alone (Fig. 2D V). Similar findings were observed with U6atac snRNA expression and p120 minor intron splicing minigene reporter. In contrast, major spliceosome activity was not affected by AR modulation (Fig. 2D).
[0080] The association between AR activation and MiS activity suggests the existence of an AR-MiS regulatory axis. To address this, we determined whether AR interacts with MiS proteins. Co-immunoprecipitation (Co-IP) with an antibody targeting AR revealed interaction with PDCD7 (and vice versa), one of the seven proteins unique to the MiS complex, suggesting that AR directly regulates MiS through PDCD7 (Figure 2E). AR45, an AR splice variant known to heterodimerize with the full-length version, and ZRSR2, known to interact with MiS, were used as positive controls. Further support was provided by demonstrating in situ interaction of AR with PDCD7 and the AR interactor HSP90 by proximity ligation assay (Figure 2F). Taken together, this data suggests that MiS activity is regulated by AR signaling, can be stimulated by ADT, and increases as PCa disease progresses to CRPC-adeno and CRPC-NE.
[0081] Example 4: MiS inhibition leads to minor intron splicing abnormalities in prostate cancer Based on our findings that MiS plays an important role in PCa progression, we next inhibited MiS in PCa by targeting U6atac snRNA, which usually shows a higher turnover rate but is detected at higher levels in advanced cancer stages. We used siRNA against siU6atac in four PCa cell lines: LNCaP (primary PCa, treatment sensitive), C4-2 (CRPC-adeno, treatment resistant), and 22RV1 (CRPC-adeno, treatment resistant) cell lines and patient-derived organoid PM154 (CRPC-NE, treatment resistant). First, we chose C4-2 cells to establish the kinetics of U6atac downregulation (Figure 3A). Consistent with the increase in U6atac expression in C4-2 cells overexpressing U6atac (~6-fold, t-test: p<0.0001), successful downregulation of U6atac snRNA in C4-2 cells upon treatment with siU6atac (~7-fold, unpaired t-test: p<0.0001) was confirmed by qRT-PCR (Figure 3A). We next investigated the kinetics of MiS inhibition in minor intron splicing by analyzing COA3, a single intron carrying minor intron. The resulting inhibition of the minor spliceosome was reflected in a progressive increase in unspliced (unspliced) COA3 native minor intron reporter transcript over time, with the highest change observed at 96 hours post-transfection (the latest time point examined) (Figure 3B). Similar results were obtained in other cell lines (Figure 10A). This finding was further confirmed by reduced splicing of three minor intron splicing minigene reporters, p120, hSCN8A, and hSCN4A, in siU6atac-treated cells compared to siScrambled controls (Figure S10B).Furthermore, the absence of minor intron splicing was observed by a decrease in the luminescence signal from the luc-minor splicing reporter (Figure S10C).
[0082] To capture the global effect of MiS inhibition, we performed ribo-deleted total RNAseq on the four cell types treated with siU6atac and siScrambled for 96 h. Comparable levels of KD of U6atac were observed in all four cell lines (Figure 10D). We first determined the overall level of minor intron retention by quantifying the missplicing index for each cell line. Comparing the median missplicing index for each sample revealed significantly increased minor intron retention in siU6atac-treated LNCaP, C4-2, 22RV1, and PM154 compared to their respective scrambled siRNA controls (Figure 3C) (Kruskal-Wallis test followed by post-hoc Dunn's test, respectively). * p<0.05, ** p<0.01, ***p<0.001). Consistent with our previous reports on the cell type-specific effects of MiS activity (Figures 1 and 2), we found that the retention level of minor introns in siU6atac-treated C4-2 cells was significantly higher than that in both siU6atac-treated PM154 and LNCaP cells (Kruskal-Wallis test and post-hoc Dunn test: C4-2 vs. LNCaP p<0.0001; C4-2 vs. PM154 p=0.0296). Similarly, the retention level of minor introns in siU6atac-treated 22Rv1 cells was also significantly higher than that in siU6atac-treated LNCaP cells (Kruskal-Wallis test and post-hoc Dunn test: p=0.0001). Because MiS inhibition is also predicted to induce alternative splicing (AS) across minor introns (in addition to increasing minor intron retention), we next quantified the number of AS events detected in all eight samples. Indeed, we found a robust increase in the number of AS events occurring around minor introns in all four samples treated with siU6atac, with the most abundant changes occurring in cryptic splice site usage. With regard to both minor intron retention and AS around minor introns, the greatest effect of siU6atac treatment was observed in C4-2 cells, followed by 22RV1 cells, then PM154, and finally LNCaP cells (Figure 3D).
[0083] Next, we investigated whether there were cell type-specific effects of MiS inhibition by performing an intersection analysis of MIGs with significantly elevated minor intron retention or AS. Here, we separated three cell lines from organoids due to their different origins, culture conditions, genomic structures, and PCa phenotypes (CRPC-NE), which were also reflected in the principal component analysis. Among the 380, 419, and 390 MIGs with significantly elevated minor intron retention in LNCaP, C4-2, and 22RV1 cells, respectively, we found that 337 MIGs were common to all (Figure 3E). PM154 organoids were found to have 379 MIGs with elevated minor intron retention, of which 303 MIGs were shared with the three cell lines. GO term enrichment analysis of each set of MIGs with elevated minor intron retention showed enrichment in fundamental molecular pathways, such as RNA and DNA processing, vesicular trafficking, and mRNA splicing. Similarly, among 159, 255, and 225 MIGs with significantly elevated AS events in LNCaP, C4-2, and 22RV1 cells, respectively, 137 MIGs were common to all (Figure 3F). In PM154 cells, we detected 196 MIGs with significantly elevated AS events (two-tailed Student's T-test, p<0.05), of which 111 were shared with the three cell lines.
[0084] Aberrant minor intron splicing in siU6atac-treated samples should affect the global transcriptome of PCa, which we captured by analyzing differential gene expression. We set the gene expression threshold at 1 transcript per million (TPM) and found 68 genes significantly upregulated (log2FC≧1, p≦0.01) and 691 genes significantly downregulated (log2FC≦-1, p≦0.01) in siU6atac-treated LNCaP cells compared to siScrambled using isoDE2 ( FIG. 11 ). Similarly, we found that 154, 228, and 121 genes were upregulated, and 787, 638, and 189 genes were downregulated, respectively, in C4-2, 22Rv1, and PM154 organoids treated with siU6atac for 96 hours (Figure 11). Cross analysis of LNCaP, C4-2, and 22Rv1 cells revealed 11 upregulated protein-coding genes common to the three cell lines, with only one (CNST) being MIG. In contrast, we found 268 downregulated protein-coding genes common to the three cell lines, with 18 being MIG. Using DAVID, we found that the shared downregulated genes were enriched in many gene ontology (GO) terms related to the cell cycle. Alternatively, in PM154 cells treated with siU6atac for 96 h, 189 downregulated genes were enriched in the GO term cell differentiation, which is specific to this cell line.Gene Set Enrichment Analysis (GSEA) based on the distinct gene sets of up- and down-regulated genes confirmed that proliferation-promoting pathways (E2F targets, G2M checkpoint, mitotic spindle) and DNA repair pathways were decreased by siU6atac-mediated MiS inhibition, and further revealed a decrease in prostate-specific pathways such as androgen response or spermatogenesis.
[0085] Next, we attempted to elucidate the molecular defects downstream of the aberrant minor intron splicing in conjunction with the transcriptome changes captured by RNAseq. To this end, we obtained a list of MIGs with elevated minor intron retention common to all of the LNCaP, C4-2, and 22Rv1 cell lines (Figure 3G), and selected MIGs that we identified as direct interactors of prostate cancer causative genes (Figures 1B and C). Here, we focused only on minor intron retention, rather than MIGs with AS. This is because it is a single molecular event (minor introns are either retained or not, whereas multiple AS events can occur simultaneously), thus simplifying downstream molecular predictions. This resulted in 55 MIGs for the LNCaP, C4-2, and 22RV1 cell lines. We hypothesized that the retention of minor introns in these MIGs may affect the transcriptome, which would be reflected in the downregulated non-MIGs. We then investigated whether MIGs and non-MIGs formed PPI networks by performing STRING network analysis on the combined list of 55 MIGs that retained minor introns and 268 downregulated genes in all three cell lines. Most of the downregulated genes formed a tight central network surrounded by MIGs with high minor intron retention (Figure 3H). We performed a similar analysis on PM154 organoids and found that although the network was somewhat diffuse, MIGs with high retention were clearly associated with the abundance of downregulated genes (Figure 3H). We then used Ingenuity pathway analysis (IPA) to i) uncover the biological pathways underlying these networks, and ii) determine whether the same pathways affected in cell lines were also affected in organoids (Figure 3). Of the 71 and 212 significantly enriched biological pathways for cell lines and organoids, respectively, we found that 52 were shared.The overlapping pathways included terms such as cyclin and cell cycle control, PCa signaling, and MAPK signaling (Figure 3). Of note, both cell lines and organoids were enriched in numerous unique terms related to apoptosis and MAPK signaling. Finally, we attempted to decompose the large STRING network into functionally related subnetworks and then determine whether similar subnetworks affect cell lines and organoids. We found that only a single subnetwork was shared in both conditions, where the three main disease associations were cancer, hematological disease, and immune disease (Figure 3). All remaining subnetworks in cell lines and organoids did not overlap. Importantly, the subnetworks found to be enriched in the three cell lines were often related to cell cycle (e.g., G2 / M DNA damage checkpoint control and G1 / S checkpoint control) and DDR (e.g., the role of BRCA1 in DNA damage response and DNA double-strand break repair by homologous recombination). In contrast, organoid subnetworks were frequently associated with oncogenic signaling pathways (e.g., PI3K / Akt signaling and PTEN signaling); actions performed by the cytoskeleton (e.g., morphological changes, EMT, and actin cytoskeleton) that control cancer hallmarks such as invasion, metastatic spread, migratory capacity, and cell movement. MIGs with high minor intron retention were found to be involved in nearly all subnetworks in both cell lines and organoids. Overall, siRNA-mediated U6atac downregulation led to robust minor intron splicing defects in many subsets of MIGs involved in cancer-relevant pathways.
[0086] Example 5: Effect of MiS inhibition on the prostate cancer proteome To investigate whether the transcriptional changes captured in each cell line extended to differences in protein production, we performed LC-MS / MS on the same cell lines treated with siRNA (Figures 4 and 12). Upon U6atac knockdown (log2FC>±1, p≦0.05), we found 314 upregulated and 353 downregulated proteins in LNCaP cells; 390 upregulated and 753 downregulated proteins in C4-2 cells; 341 upregulated and 689 downregulated proteins in 22RV1 cells; and 197 upregulated and 455 downregulated proteins in PM154 cells (Figures 4A-E). For all cells, we found that most differentially expressed MIG-encoded proteins were downregulated (Figure 4A). Importantly, we found that almost all MIGs with significantly higher minor intron retention were downregulated by LC-MS / MS (Figure 4A). For three cell lines (LNCaP, C4-2, and 22Rv1), we found shared upregulation of only eight proteins (Figure 4B), whereas 45 proteins were downregulated in all three (Figure 4C). None of the common upregulated proteins were encoded by MIGs, whereas nine of the common downregulated proteins were found to be MIGs, including SLC12A7, CEP170, TCEA3, POLA2, XPO4, WDFY1, ORC3, C17orf75, and SRPK1. Except for CEP170 and TCEA3 proteins, the remaining proteins were significantly elevated in minor intron retention of their respective transcripts by RNAseq in all cell lines. More importantly, POLA2 and SRPK1 are MIG-encoded proteins that directly interact with proteins encoded by the PCa causative genes identified in Figure 1C.
[0087] Comparison of the proteomic data of the four analyzed cell lines representing PCa disease progression revealed cell type, and therefore likely PCa subtype, and context-dependent MiS-dependent proteomes, very similar to the RNAseq analysis. Each cell line expressed a unique set of up- and down-regulated proteins, including MIGs and non-MIGs (Figure 4B and E), each important for cancer biology in its own way. Among them, AR (non-MIG), Rb1 (non-MIG), the epigenetic regulator EED (MIG), a common target during PCa treatment, and the proliferation marker Ki67 (non-MIG) were strongly reduced by siU6atac treatment in the CRPC-adeno line C4-2 (Figure 4D). Similarly, DNA damage response (DDR) proteins BRCA1 and AURKA (non-MIG), and JNK2 were decreased in treatment-resistant C4-2, 22Rv1, and PM154 lines, but increased in treatment-responsive LNCaP cells (Figure 4E). AURKA is associated with lineage plasticity and neuroendocrine differentiation in PCa. Consistent with previous results, we observed an increase in tumor suppressors such as cell cycle regulators CDN1A (p21) (non-MIG) and CDN2A (non-MIG) in all three cell lines tested. In contrast, we found that these proteins were decreased in neuroendocrine organoids PM154 treated with siU6atac (Figure 4E). Finally, we also observed that MiS inhibition increased proapoptotic proteins such as BCL10 (non-MIG). GSEA based on signature gene sets of up- and down-regulated proteins confirmed that siU6atac-mediated MiS inhibition decreased pro-growth pathways (E2F target, G2M checkpoint, mitotic spindle) and DNA repair pathways, whereas increased apoptotic and stress-sensing pathways (p53, unfolded protein response, UV response, or hypoxia-mediated oxidative stress). Differential protein expression was verified by WB analysis (Figure 4D and Figure S13).
[0088] Next, we investigated the agreement between the transcriptome and proteome data, which were generally discordant in each cell line (Figure 4F). MIGs with high retention upon MiS inhibition have been reported to escape NMD (nonsense-mediated decay). For example, we found that minor intron retention and AS were elevated upon siU6atac treatment, although proteins such as EED, JNK1 / 2, BRAF, and RAF1 were decreased, although their expression by RNAseq remained unchanged (Figure 4A-E). We also found that proteins such as AR and Rb1 were only decreased in the proteome, suggesting that an indirect control mechanism was being implemented via MiS. Nevertheless, we identified 57, 127, 117, and 8 genes whose mRNA transcripts and encoded proteins were significantly downregulated in 96 h siU6atac-treated LNCaP, C4-2, 22Rv1 cells, and PM154 organoids, respectively (Figure 4F). For example, we found that the proliferation marker Ki67 was downregulated in the transcriptome as well as the proteome of LNCaP and C4-2 cells. 22Rv1 cells showed a decrease in the DDR proteins BRCA1 and AURKA / B. In contrast, CDN1A(p21) was upregulated in the transcriptome and proteome in all three cell lines. When these four gene sets were independently submitted to DAVID, we found a high degree of overlap in GO terms, mainly centered around the cell cycle. For example, genes with transcripts and encoded proteins that were similarly downregulated in LNCaP, C4-2, and 22RV1 cells were enriched in cell division, mitotic nuclear division, spindle poles, sister chromatid cohesion, the G2 / M transition of the mitotic cell cycle, cell proliferation, and DNA strand elongation involved in DNA replication.Similarly, in siU6atac-treated PM154 organoids, eight genes that were both transcriptionally and proteomically downregulated were enriched in a single GO term (metaphase plate formation), which was also enriched in siU6atac-treated 22RV1 cells. In summary, siU6atac-mediated MiS inhibition altered the transcriptome and proteome, both of which were enriched in cell cycle regulators. Based on the molecular signature, we next investigated whether MiS inhibition actually affects cell cycle progression. Therefore, we performed FACS analysis at 72 and 96 hours post-transfection in siU6atac-treated therapy-resistant C4-2 cells, and found that G1 / G0 cells were significantly increased (two-way ANOVA, 72 hours: ** p=0.0027, 96 hours: ** p=0.0018), and was significantly decreased in S phase cells (two-way ANOVA, 72 h ** p=0.0080, 96 hours: ** p=0.0085) (Figure 4H). PM154 organoids also showed similar results under siU6atac conditions (Figure 4I). These cell cycle defects observed here revealed that MiS inhibition causes G1 / S cell cycle arrest in PCa.
[0089] Example 6: Single-cell RNAseq reveals cell cycle defects caused by MiS inhibition Next, we wanted to clarify the cellular heterogeneity of response to siU6atac-mediated MiS inhibition. Therefore, we also performed single-cell RNA sequencing (scRNAseq) on LNCaP cells and PM154 organoids after siU6atac. After standard data processing and quality control procedures (Methods), we obtained transcriptome profiles for 8206 siScrambled cells and 6730 LNCaP cells, as well as 11181 siScrambled cells and 11475 PM154 organoids. We employed unsupervised clustering to identify the heterogeneity of response to siU6atac. K-means clustering and UMAP projections integrating data across both genotypes revealed nine major cellular clusters (Figure 5A). We found that clusters 0, 3, 4, 6, 7, and 9 contained equal numbers of siScrambled and siU6atac treated cells (Figure 5B). Clusters 1 (Fisher's exact test, p=3.16934e-69), 2 (Fisher's exact test, p=1.702664e-63), and 5 (Fisher's exact test, p=3.977987e-05) showed a higher proportion of siScrambled cells compared to siU6atac cells. Finally, cluster 8 (Fisher's exact test, p=1.533612e-14) contained a predominance of siU6atac treated cells compared to siScrambled (Figure 5B). GO term enrichment analysis of the top 5 upregulated genes in cells belonging to clusters 1, 2, and 5 revealed cell division, G1 / S transition, and mitotic nuclear division. In contrast, no Go term enrichment was found in the top 5 genes in cluster 8. The enrichment of cell cycle terms, combined with previous findings that MiS inhibition affects cell cycle control in cancer (Figures 3, 4), led us to employ a list of cell cycle regulators as a means to sort single cells based on cell cycle phase. We therefore overlaid cell cycle phases on the UMAP representation of the unsupervised data clustering (Figure 5D).We found that clusters 0, 3, 4, 6, 7, and 9 had the highest percentage of G1 cells (Figure 5E), while clusters 1 and 2 had the highest percentage of S and G2 / M cells, respectively (Figure 5E). We performed a similar analysis on PM154 organoids, resulting in seven clusters (Figure 14A). The majority of cells present in cluster 5 (Fisher's exact test, p=3.287745e-40) were from siScrambled, whereas cluster 6 (Fisher's exact test, 2.738288e-57) was predominantly siU6atac-treated cells (Figure 14B). No Go term enrichment was found in the upregulated genes in these clusters, although cluster 5 showed a high percentage of S and G2 / M cells, similar to clusters 1, 2, and 5 in LNCaP cells, respectively (Figure 14C,D).
[0090] Comparison of the distribution of cells from each condition in the nine clusters and the distribution of cells in different phases of the cell cycle led us to investigate cell cycle defects through scRNAseq analysis (Figure 5). We then overlaid cell identity (i.e., siScrambled or siU6atac) onto the UMAP of clusters along with cell cycle phase (Figure 5). For LNCaP cells, we found a significant enrichment of siU6atac-treated cells in G1 phase (Fisher's exact test, p=3.18e-308, OR=3.89, 95% CI 3.60-4.19) and a reduction in these cells in S phase (Fisher's exact test, p=9.80e-112, OR=3.22, 95% CI 2.88-3.60) (Figure 5). Similarly, for PM154 organoids, we observed a significant enrichment of siU6atac cells in G1 phase (Fisher's exact test, p=1.56e-145, OR=2.05, 95% CI 1.93-2.16) and a reduction in S phase (Fisher's exact test, p=7.07e-55, OR=1.57, 95% CI 1.48-1.66) (Figure 5). These findings are consistent with the FACS analysis (Figure 4).
[0091] Since siU6atac successfully blocked cell cycle progression in PCa and AR signaling is a key driver of PCa, we next investigated the AR score (a key metric of CRPC-adeno progression) in LNCaP cells and PM154 organoids treated with siU6atac (Figure 5 and Figure 14E). We found that siU6atac could indeed significantly reduce the AR score of both models in G1 (Wilcoxon test, p<2.22e-16), G2M (Wilcoxon test, LNCaP: p=5.8e-16, PM154: p<2.22e-16) and S phases (Wilcoxon test, LNCaP: p=3.4e-07, PM154: p<2.22e-16). We did not observe any shift in AR score after siU6atac treatment in bulk RNAseq (Figure 14G), which is not surprising since we lost single-cell resolution and the expression of some of the key AR score markers was diluted. Another hallmark of cancer progression is EMT, which we investigated using the EMT score (Dong, B et al., Communications Biology 2020, 3(1), 778). We found that EMT scores were also significantly decreased in siU6atac-treated cells and organoids in G1 (Wilcoxon test, p<2.22e-16), G2M (Wilcoxon test, LNCaP: p=2.9e-10, PM154: p=2.9e-06) and S (Wilcoxon test, LNCaP: p=3.4e-07, PM154: p=0.0089) phases (Figure 5 and Figure 14F). Next, we investigated the expression of MIGs, identified as the most important nodes from our PPI analysis, in the scRNAseq data (Figures 1B and C). Indeed, we found that the number of siU6atac-treated cells and organoids expressing these MIGs, such as AURKA, EED, and PARP1, was significantly lower (Figures 15 and 16).
[0092] Taken together, scRNAseq revealed siU6atac-mediated transcriptome remodeling that contributes to PCa progression and lineage plasticity by reducing cell cycle, AR signaling, and EMT.
[0093] Example 7: The minor spliceosome is essential for PCa proliferation and survival Administration of siU6atac significantly reduced the proliferation of hormone-responsive LNCaP cells and therapy-resistant L-AR, C4-2, 22RV1, and PM154 cells and organoids (Figure 6A, and Figure 17A and B). Previous reports have shown tissue-specific minor intron splicing as reflected by the difference in the ratio of minor intron retention across various tissues. We therefore investigated whether siU6atac treatment would result in a similar context-dependent inviability. Indeed, we observed that upon siU6atac treatment, the confluence of L-AR cells was significantly reduced compared to LNCaP cells (Figure 6B, Figure 18A). L-AR cells showed a higher proliferation defect than LNCaP cells, likely due to the higher doubling rate of L-AR cells (Figure 18B). It was shown that L-AR cells (and therapy-resistant cells) have higher MiS activity than natural hormone-responsive cell lines. These data suggest that the higher MiS basal activity (Figure 2) indicates an increased dependency on this non-canonical splicing mechanism. Of note, the viability of non-cancer cells such as fibroblasts (HS27) and primary mouse prostate cells (M2514) was not affected by siU6atac-mediated MiS inhibition (Figures 6A, 17A, and 19). In accordance with the KD data, overexpression of U6atac in C4-2 cells again promoted cell proliferation (Figure 6C). This is consistent with previous findings showing that MiS is essential for cell proliferation and plays a major role in cell cycle progression.
[0094] Although we targeted U6atac to inhibit MiS, the minor spliceosome complex is composed of other unique components, including snRNAs and snRNPs. Therefore, we chose to inhibit MiS by targeting another component of MiS, RNPC3. Indeed, KD of RNPC3 caused a similar response to siU6atac in PCa cells, indicating that the observed decrease in cell proliferation and viability may be due to a decrease in MiS activity in general and is not a specific observation for U6atac (Figure 6A, Figure 17A and B). Importantly, we found that not only does MiS inhibition affect PCa, but also the viability of other cancer cell lines, such as colon, kidney, breast, glioblastoma, and CML, is decreased by KD of several MiS components (Figure 6D and Figure 18C and D). This finding reveals that MiS is essential for other cancer types and that dependency on its integrity may represent a broad phenomenon not limited to PCa. Notably, among all siRNAs tested (RNPC3, PDCD7, U12, U6atac), KD of U6atac was found to be the most efficient. Surprisingly, even KD of U6atac was not sufficient to reduce the viability of bladder cancer lines UMOC9 and SW680. However, siU6atac (in combination with cisplatin) led to a significant decrease in viability (Figure 6E and F). This suggests that in some cases, MiS inhibition may work synergistically with current state-of-the-art treatments. Indeed, combination therapy such as EZH2 inhibition, which sensitizes to ARSi (enzalutamide) treatment, is required for CRPC (Ku, SY et al., Science 2017, 355(6320), 78-83; Bai, Y. et al., J Biol Chem 2019, 294(25), 9911-9923). Therefore, we tested whether U6atac inhibition + enzalutamide could be as effective as or better than EZH2 inhibition + enzalutamide. We observed that U6atac KD + enzalutamide was more effective than EZH2 + enzalutamide in CRPC C4-2 cells and enzalutamide-resistant LNCaP (L-rENZ) cells (Figure 6G and Figure 18E).Surprisingly, U6atac KD alone was significantly better than EZH2 plus enzalutamide.
[0095] Finally, we wanted to further explore the effects of MiS inhibition in a model system that captures the heterogeneity of PCa disease better than the cell culture studies used so far. We therefore applied siRNA against U6atac in 3D cultured PCa patient-derived organoids. By overlaying the RNAseq results of these organoids with clinical data from the SU2C study (Figure 6H), we confirmed that these model systems are clinically relevant. Treatment of organoids MSK8 (PCa8, CRPC-Ad), MSK10, 16, 14 (PCa10, 16, 14, CRPC-NE) and PM154 (CRPC-NE) with siU6atac significantly (two-way ANOVA, p=0.0246) reduced organoid proliferation and viability (Msk16: P=0.019) (Figures 6I and J, Figure 18F). Taken together, these data indicate that knocking down MiS components, particularly U6atac, is sufficient to reduce cancer proliferation and viability.
[0096] In summary, we performed the first-in-the-field evaluation of MiS in PCa using a multi-pronged approach combining RNAseq, mass spectrometry, FACS, and scRNAseq, demonstrating that inhibition of MiS leads to profound cell cycle defects through aberrant splicing of MIG, with fundamental implications for PCa progression, survival, and lineage plasticity.
[0097] Example 8: Minor splicing correlates with PCa progression Our rationale for using reporter assays as splicing readouts is that evaluation of splicing dynamics of endogenous genes is often complicated by multiple factors (chromatin structure, transcription rate, trans-acting factors, and activity of the nonsense mediated decay (NMD) pathway). Thus, the field of splicing, especially minor intron splicing, has a long history of using splicing reporters such as the ones we used in our study. The rationale, especially when comparing splicing in different cell types or cancer stages, is that expression of the transfected splicing reporter is independent of endogenous modes of control. Thus, by expressing the same amount of reporter in different cell lines, we can truly examine the efficiency of MiS activity.
[0098] Evaluating whether changes are seen in the reporter assay reflects changes in endogenous MIG within each cell line. This question concerns the splicing kinetics analysis of CoA3 that we performed to confirm that siU6atac does indeed affect the minor splicing of MIG (Figures 10, 12, 18). CoA3 is one of the rare genes that has only one intron, a minor intron, and therefore represents an excellent endogenous MIG reporter of minor intron splicing. To address this, we now measure the expression of spliced CoA3 as x-fold (splice index) over unspliced CoA3 in most of the cell lines used in this paper (Figures 9E, 24). Although the values of endogenous CoA3 splicing were highly variable for the reasons explained above, we also observed a trend for more advanced PCa lines to show higher CoA3 splice index.
[0099] Example 9: MIG expression in LNCaP, C4-2, 22Rv1, and PM154 cells The inventors further analyzed the RNAseq datasets of four cell lines (LNCaP, C4-2, 22Rv1, and PM154) representing different progression stages of PCa. In particular, the inventors analyzed the RNAseq data of samples treated with scrambled siRNA representing the de novo control transcriptome (Figures 21, 24). The inventors found that the number of MIGs expressed, and thus the required amount of MiS, increases with the progression of PCa and androgen receptor signaling inhibition (ARSi) therapy resistance (LNCaP (3 MIGs) < C4-2 (5 MIGs) < 22Rv1 (16 MIGs) < PM154 (41 MIGs)). The levels of endogenous expressed MIGs within each cell line are similar to the results of the inventors' reporter assay in this analysis.
[0100] Interestingly, the inventors observed that while the treatment-responsive LNCaPs expressed MIGs that play a role in the antiviral response, the MIGs of C4-2 cells contained proteins involved in DNA damage and autophagy. The MIGs of 22Rv1 included MAPK11, as well as the tumor antigen (CTAG2) or the splicing factor (SRPK3), and the MIG repertoire of the PM154 organoids included, among others, MAPK10; chromatin remodeling factors such as ACTL6B; and several proteins involved in cytoskeletal signaling (CEP170, EML4, PDE6D), and also a protein involved in DNA damage repair (MSH3). Therefore, it is possible to claim that as PCa progresses, the intracellular MIG pool acts favorably for cancer survival. The inventors also compared the endogenous AS events in the RNAseq data of samples treated with scrambled siRNA, and found that there was only one AS event with a significant difference in all four cell lines, which occurred more frequently in PM154. This event occurs within TSPYL2, which has recently been shown to contribute to abiraterone resistance, an ARSi in CRPC, through transcriptional regulation of CYP genes.
[0101] Example 10: p38MAPK regulates MiS in PCa-NE To answer the question of why CRPC-NE cells / organoids usually exhibit high MiS activity despite low AR activity, and whether there are multiple mechanisms that control MiS activity, we explored AR-independent mechanisms of MiS regulation in NEPC. pP38MAPK was recently found to induce neuroendocrine differentiation in LNCaP cells. Furthermore, p38MAPK has been suggested to promote invasion, metastasis, and immune evasion in treatment-resistant PCa cells. Interestingly, p38MAPK is partially regulated by the rapid turnover of U6atac snRNA. The Dreyfuss laboratory showed that anisomycin-mediated activation of p38MAPK leads to increased stability of U6atac, which in turn activates MiS. Based on these evidences, we hypothesized that p38MAPK controls MiS activity in NE organoids.
[0102] To address this, we tested the effect of anisomycin, which activates p38MAPK, in our five organoids (MSK8, MSK10, MSK16, PM154, and PM1262) transfected with luc-splicing reporter assay constructs, as well as LNCaP and C4-2 PCa-adeno cells, and H660 PCa-NE cells.In a recently published study on patient-derived organoids from MSKCC (named MSK) and The inventorsill Cornell (named PM), Tang et al. classify MSK8 as PCa-adeno-like stem cell organoids, MSK16 and PM1262 as organoids that enrich for WNT signals, and MSK10 and Pm154 as organoids that enrich for genomic NEPC signatures. Surprisingly, the effect on minor intron splicing was observed only in PCa-NE cells and organoids (H660, PM154, and MSK10), suggesting that MiS activity in neuroendocrine PCa may be regulated via the p38 MAPK axis (Figures 22A and D).
[0103] To directly test this, we investigated the expression and minor intron splicing of all four p38 MAPKs (MAPK11, 12, 13, and 14) in our omics dataset. Integrating both siU6atac RNAseq and proteomics, our results indicate that MAPK14 and 13 may be key components available to control MiS activity. We found that all four p38 MAPK family members showed intron retention or AS when U6atac was knocked down (KD) in LNCaP, C4-2, 22Rv1, PM154 cells and organoids. Comparing RNAseq and proteomic data, we found that MAPK14 protein expression, however, was only reduced in 22Rv1 cells, which are considered intermediate between the adeno and neuroendocrine stages, and MAPK13, which promotes neurotoxicity and is required for prostate epithelial differentiation, was only reduced in PM154 organoids.
[0104] We therefore performed siRNA-mediated KD of MAPK14 or 13 in LNCaP, C4-2, 22Rv1, and H660 cells, and MSK8, 10, 16, PM154, and PM1262 organoids, along with luc-splicing reporter assays (Figure 22B). We found that MAPK13 downregulation did not show significant minor splicing changes, as reflected by the luciferase activity of the minor intron reporter. siMAPK14 significantly reduced minor splicing only in H660 cells and PM154 organoids. Together, these findings reveal a novel mechanism for regulating MiS in neuroendocrine PCa in the absence of AR. Importantly, the major splicing reporter construct showed no significant changes. In summary, we demonstrate that MiS is regulated in a bimodal manner in prostate adenocarcinoma, where AR activity informs MiS, while MAPK13 and 14 are potential regulators of MiS activity in neuroendocrine PCa (Figure 22D).
[0105] Example 11: Minor intron splicing pathways associated with prostate cancer progression Our goal is to answer the question of whether MIG can affect mRNA levels in the transcriptome and whether U6atac affects gene expression through its function in MiS. Currently, U6atac is only known to control minor intron splicing through its interaction with U4atac as part of MiS. Knockdown of U6atac is predicted to result in increased retention of minor introns, which was captured by our RNAseq analysis (Figure 3).
[0106] In this case, it is surprising that no significant downregulation of MIG expression levels with intron retention is observed. However, there are multiple explanations for this result. The first major NMD (nonsense-mediated decay) player, including UPF1 and NCBP2, is MIG, which itself exhibits aberrant splicing. Thus, because the NMD pathway itself is impaired, these aberrantly spliced transcripts are able to escape NMD, which is reflected in the lack of downregulation of MIG. Despite this lack of downregulation, aberrant splicing of MIG would affect the function of the encoded proteins, since the majority of the transcripts cannot code for full-length proteins. Indeed, we have previously shown that aberrant splicing of MIG by MiS inhibition results in transcripts that escape mRNA surveillance mechanisms. Moreover, we have previously shown that aberrantly spliced MIG transcripts bind to ribosomes, which may contribute to the observed cell cycle defects, whereby the production of novel, potentially toxic proteins. Finally, there is emerging literature supporting intron retention as a distinct form of regulation that is not exclusive to NMD.
[0107] To analyze MIGs that affect the transcriptome, MIGs perform various functions whose functions are disrupted upon aberrant splicing of these MIGs. For example, MIGs are splicing factors (SFs), subunits of polymerase II and III (Pol II, III), transcription factors (TFs), RNA-binding proteins (RBPs), chromatin remodeling factors (CRFs), and ribonucleoprotein particles (RNPs). Overall, KD of U6atac causes aberrant splicing of MIGs, partially escaping NMD and no downregulation is observed, but ultimately impacting directly on transcription and / or indirectly on the cellular transcriptome.
[0108] As mentioned above, siU6atac-mediated aberrant splicing of MIG has global transcriptomic effects. Since our goal was to understand how the primary defect of MiS activity inhibition manifests in cancer cells and ultimately leads to cell cycle defects and cell death, we sought to capture the global transcriptomic changes that underpin cell cycle defects and cell death. Thus, we first focused on either MIG or non-MIG genes that are affected by siU6atac.
[0109] We then added further PCa-associated MIGs to the list, such as JNK2 and JNK3, E2F1 and E2F2, RNPC3, Brd9, BRAF, ACTL6A, SMARCA1, and ACTL6B (also known as BAF53b), which we have recently described as essential NEPC biomarkers. Below, we show a table that clarifies the dynamic and context-dependent nature of minor splicing in different PCa cell lines (Table 1). In that sense, we found that siU6atac caused a context-dependent decrease in the expression of many MIGs and non-MIGs with known functions in NEPC transdifferentiation, such as the previously mentioned p38MAPK or BAF53B, and BAF53A, EED, EZH2, AURKA, likely based on intron retention and AS events (Table 1). [Table 1]
[0110] Example 12: Minor splicing is associated with attenuation of cell cycle genes The major cell cycle defects are mediated by MIG, which is actively involved in cell cycle control. Importantly, we showed that highly retained MIG, which is downregulated at the protein level, is highly enriched in cell cycle control (GO term enrichment analysis). Thus, the main molecular defect causing the observed cell cycle defects is splicing abnormality of MIG. Moreover, it is now well documented that minor splicing disruption is associated with cell cycle defects, affecting S-phase, mitosis, and cytokinesis during brain development. Since we demonstrated that siU6atac results in minor intron splicing defects similar to those observed in other model systems, it is not unreasonable to link the cell cycle defects observed in PCa to splicing abnormalities. Indeed, this is reflected in our molecular transcriptome / proteome cross-analysis (Figure 4). Collectively, our data in PCa fit well with the literature linking aberrant splicing of MIG to the cell cycle in the developing cerebral cortex and limb.
[0111] Example 13: Differential clustering of cancer types Next, we found that differential clustering improved significantly when more cancer types (and therefore samples) were included. In particular, the difference in clustering based on MIG and non-MIG widened as a result of including more cancer types (i.e., the difference in silhouette scores between MIG and non-MIG widened after including more cancer types in our analysis). The results of this revised analysis are shown in Figure 25.
[0112] Figure 4B suggests that a strictly linear relationship may exist between the proportion of MIGs in a sample and the resulting silhouette score (as opposed to the possibility of a more complex, non-linear relationship, e.g., one that may indicate saturation). To assess the extent to which the proportion of MIGs affects the silhouette score in a linear fashion, we used a simple linear regression to regress the median silhouette score on the proportion of MIGs. In particular, we take the median silhouette score for a given portion of MIGs to be the median score from those associated with the same 1000 resampled gene sets used to generate Figure 4B. The results of this regression strongly indicate a linear relationship, and we provide the relevant summary statistics below: t-statistics and p-values related to the slope: t-statistic: 64.97 p-value: 2.45e-13 R Squared: 0.9979 F-statistic: 4221 (at 1 and 9 degrees of freedom) Residual standard error: 0.0005889, at 9 degrees of freedom
[0113] In particular, we chose to include only cancer types for which a sufficient number of samples with expression data were available. Thus, we only enforced that each cancer type was represented by a sufficiently large number of samples (we chose 18 as the threshold). This allowed us to retain 23 cancer types across the sample size of N=1224.
[0114] Example 14: MiS affects REST splicing To answer the central question in the field of understanding PCa drug resistance to ARSi, whether MiS is indeed involved in PCa CRPC / NE transition, we previously reported that 10-15% of CRPC-adeno undergo lineage plasticity to an AR-negative state, some expressing neuroendocrine surface markers, and others being double negative. REST is known to be a critical factor for neuroendocrine transdifferentiation.
[0115] RE1-silencing factor (REST) is well defined as a repressor of neural differentiation. Loss of REST expression was associated with upregulation of genes used to define CRPC-NE (e.g. SYP and CHGA). Furthermore, loss of REST expression precedes neuroendocrine differentiation in PCa. Therefore, loss of REST has been considered as a potential promoter of NEPC PCa. Relevant to this study, REST is controlled through dynamic alternative splicing such that when intron 3 contains a mini-exon, the mini-exon results in a premature stop codon that truncates the open reading frame. A truncated form of this REST protein, called REST4, cannot bind to the RE1 silencing element but is able to block REST FL / DNA contacts. Thus, REST4 acts as a dominant negative, such that increased levels of REST4 result in inhibition of REST function allowing NE differentiation. We observed by qPCR that the expression of REST is dynamic among four different cell lines (LNCaP>C4-2>22Rv1>PM154) representing PCa progression (Figure 25A). Importantly, when we compared the expression of REST4, the upregulation of this isoform in 22Rv1 and PM154 was highly significant (Figure 25A). Analysis of SU2C data is consistent with these results (Figure 25B). We found that canonical REST correlated positively with AR score, while it correlated negatively with NEPC score. In line with this, the expression of REST was significantly decreased in NEPC samples (NEPC score >0.4) compared to AR samples (AR score >0.4). In contrast, when we queried SU2C for the expression of REST isoforms, we found that REST4 correlated negatively with AR score and positively with NEPC score. Here, expression of REST4 was significantly increased in NEPC samples (NEPC score>0.4) when compared to AR samples (AR score>0.4).
[0116] When we relate this information to minor splicing, we find that KD of U6atac increases canonical REST after 96 h and decreases REST4 expression after 48-72 h in NE-like cells (22Rv1) and NE organoids (PM154) (Figure 25C and D). Surprisingly, our RNAseq data of scrambled and siU6atac cells showed no change in total gene expression of REST. However, we observed compensatory changes in two alternatively spliced REST isoforms (REST4 and REST'3c+) after siU6atac (48 h) treatment (Figure 25D-F); MiS inhibition by siU6atac led to downregulation of REST4 in 22Rv1 and PM154 cells and organoids (Figure 25D-F). As mentioned above, upregulation of REST4 acts as a dominant negative and blocks the function of REST. This finding suggests that MiS is not only correlated with PCa progression, but may also be involved in the transition from CRPC to NEPC through the regulation of alternative REST splicing. Thus, we are now able to demonstrate that MiS is indeed involved in the CRPC / NE transition through the regulation of alternative REST splicing.
[0117] Example 15: AR and minor splicing First, we show that MIG is a key player that executes oncogenic programs downstream of PCa causative genes. This arrangement of MiG positions MiS as an important target for PCa therapeutics. In particular, we show that MiS is not only directly downstream of the AR axis that blocks U6atac turnover (Fig. 2, 26A), but also that inhibition of MiS leads to downregulation of AR protein (Fig. 4, 26B). Taken together, these findings indicate that MiS is closely involved in the progression of PCa by AR. Thus, we suggest that MiS is a therapeutic target for CRPC-adeno. Furthermore, it is known that the function of REST is important for the progression to AR-independent RPC-NE. In this regard, please refer to the previous section. We show that p38MAPK plays an important role in the control of MiS when CRPC-NE is established (Fig. 21). In summary, we hypothesized that MiS is regulated by various oncogenes (e.g., AR, p38MAPK) and is linked to tumorigenic pathways. Thus, the therapeutic window in the context of PCa is multifaceted and can be exploited for the treatment of both CRPC-adeno and CRPC-NE.
[0118] Example 16: MiS inhibition affects PCa cells more than normal cells Next, we revisited cell doubling (Figure 27) and performed Incucyte experiments with all cell lines used in Figure 3. Briefly, cells treated with siU6atac or siScramble for 120 hours were measured every 6 hours. At the end of each Incucyte experiment, the viability of all samples was examined with a cell titer glow (CTG) assay. Figure 26 plots the Incucyte and CTG results (y-axis) against the doubling time (DT) of the cell line (x-axis) as x-fold of the scrambled control. Cell lines were ranked by doubling time. We did not observe a correlation between cell proliferation (doubling time) and siU6atac response. Thus, we conclude that while minor splicing certainly impacts cell cycle and proliferation, MiS dependency does not simply track cell proliferation rate. Instead, MiS dependency is cancer type and stage specific. Indeed, we also show that PCa-specific pro-tumorigenic factors, such as AR, REST, and p38MAPK, are all coupled to MiS activity and may play a role in defining MiS dependency. Taken together, we conclude that inhibition of MiS will disproportionately affect PCa cells compared to normal cells.
[0119] Example 17: siU6atac-mediated MiS inhibition affects PCa Consistent with our hypothesis that MiS inhibition would disproportionately affect PCa cells over normal cells, we show that MS2514 (normal mouse prostate cells) with a low doubling time (DT) (Figure 26) and HS27 (normal human fibroblasts) with a high doubling time (Figure 26) do not respond significantly to siU6atac (Figure 26 and Figure 3A). Importantly, human fibroblasts HS27 have a similar doubling time as LNCaP or 22Rv1 cells, but are significantly less sensitive to KD of U6atac.
[0120] To further address tumor selectivity, we performed co-culture experiments with C4-2-GFP cancer cells and HS27-mCherry fibroblasts (Figure 27). Briefly, C4-2 cells were fluorescently tagged with green fluorescent AIEDot nanoparticles, while HS27 cells were tagged with red fluorescent AIEDot nanoparticles. Both cell lines were then co-cultured and then treated with scrambled or U6atac siRNA. Cell proliferation was measured every 6 hours for a total of 5 days. After the experiment was completed, the cell mix was FACS sorted to separate C4-2 and HS27 cells, and the KD of U6atac was confirmed by FACS sorting followed by qRT-PCR. As expected, siU6atac significantly reduced the green fluorescent signal (C4-2) but did not affect the red fluorescent signal, even though the KD of U6atac was stronger in HS27 cells compared to C4-2 cells (Figure 27).
[0121] Taken together, we show that siU6atac-mediated MiS inhibition is not only related to proliferation rate, but rather specifically affects PCa, which uses many MIGs to execute its oncogenic program. In that sense, we now lay the foundation to address the challenge of the therapeutic window of titrating MiS inhibition in order to translate our findings into the clinic.
[0122] Example 18: Molecular splicing defects due to MiS inhibition In PCa, we found that downstream signaling of MiS encompasses key MIGs such as BRAF, PARP1, EED, ACTL6A and B, MAPK8, 9, 13 and 14. However, the exact MIG repertoire is cell type dependent. In this context, we found, for example, that KD of U6atac caused a decrease in ACTL6A transcripts in LNCaP and C4-2 cells, but not in 22Rv1 and PM154 cells (Figures 3 and 4). However, ACTL6B, a paralogue of ACTL6A, showed increased intron retention after siU6atac treatment in 22Rv1 and PM154 cells, but not in LNCaP and C4-2 cells. Similarly, siU6atac caused an increase in MAPK9 in hormone-responsive LNCaP cells but a decrease in therapy-resistant C4-2, 22Rv1, and PM154 cells ( Figures 4A–E ), whereas PARP1 showed intron retention in all four cell lines.
[0123] Here, we identified PCa-associated MIGs by RNAseq, qRT-PCR (Figure 3), mass spectrometry (masspec) and immunoblot analysis in U6atac KD cells (Figure 4, Figure 26 (PARP)). In all, we relied on disrupting the repertoire of MIGs, which we show to be a novel therapeutic strategy for PCa.
[0124] To show that PCa-associated MIG destruction is primarily mediated by U6atac, we performed rescue experiments (Figure 8A). We generated C4-2 cells stably expressing a doxycycline (dox)-inducible hairpin for U6atac. We treated these cells, as well as sh-scrambled control cells, with 2ug / ml doxycycline for 48, 72, and 96 hours, froze half of the cells as U6atac KD, and washed the other half three times with PBS to remove doxycycline and recultured for another 72 hours (U6atac KD rescue cells). Finally, we performed qPCR for U6atac and multiple MIGs (CoA3, PARP, EED, ACTL6B, PTEN). In line with our hypothesis that siU6atac reduces MIGs important for PCa, depletion of doxycycline from the medium rescued U6atac and shU6atac, which caused a reduction in the tested MIGs. However, we observed a reduction only after 48 hours. At later time points, we observed a gradual increase in all tested genes, which may be due to potential positive feedback. We found that, unlike siRNA, shU6atac, which reproduces the increase in minor intron retention, does not result in significant cell death. Indeed, the difference in efficacy between shRNA and siRNA is inherent due to the underlying mechanisms by which they act. However, this experiment allowed us to demonstrate a molecular splicing defect by inhibiting MiS.
[0125] Example 19: U6atac KD is responsible for observed cancer deaths To test whether we could rescue the cell death phenotype observed in the siU6atac experiment, we used CRISPR-mediated KO of endogenous RNU6atac in cells overexpressing U6atac snRNA from a plasmid transfected into C4-2 cells (C4-2'U6atac cells) (Figure 9A and B). We designed two gRNAs that target adjacent sites in the RNU and 6ATAC genes (Figure 9). The rationale here is that only endogenous U6atac will be reduced, while cells expressing the U6atac plasmid will still express functional U6atac (Figure 24).
[0126] As expected, overexpression of U6atac plasmid significantly increased the proliferation of C4-2 cells (Figure 9C). In line with that hypothesis, C4-2'U6atac cells showed no decrease in U6atac expression even though the gene was successfully CRISPER ablated (Figure 9E and F). In line with our siRNA-based U6atac KD approach, KO of U6atac caused cell death in C4-2 cells but not in C4-2'U6atac cells (Figure 9G and H). This demonstrates that U6atac KD is responsible for the cancer death we observed.
[0127] Example 20: Pull-down experiment using anti-PDCD7 antibody (AB) We performed pull-down experiments with anti-PDCD7 AB for two reasons: 1) so far, anti-PDCD7 is the only reliable AB among the seven unique MiS protein ABs. 2) Unlike the major intron 5' splice site, the minor intron 5' splice site is first recognized and bound by a protein, PDCD7. Since we wanted to pull-down pre-mRNA, we chose to perform PD with this direct interacting protein.
[0128] For 1H and 1l, we performed a correlation analysis of all seven proteins unique to MiS with ZRSR2 and CENPA, proteins common to the minor and major spliceosomes (Figure 30). Thus far, studies of ZRSR2 and CENPA KD have indeed reported a predominant defect in splicing of the minor intron.
[0129] Of course, loss of function of ZRSR proteins in the testis has also been reported as a defect in the major intron. Interestingly, we observed that the majority of MiS proteins showed a positive correlation trend with EZH2, while ZMAT5 and PDCD7 showed a negative correlation trend with EZH2. Taken together, these findings indicate that MiS and its components are dynamically regulated with PCa progression, and combinatorial changes in the expression of MiS components require further study. Furthermore, because these proteins have not been extensively studied, we do not understand other potential functions they may perform. For example, PDCD7 was identified as the Programmed Cell Death Protein 7 gene, as its name suggests, and reported as a potential transcription factor that activates the transcription e-cadherin. Overall, this analysis is one of the first to begin to position MiS as a whole in PCa progression. Of course, one caveat is that MiS proteins may have multiple functions, but currently snRNAs are thought to be involved only in the execution of minor splicing.
[0130] Example 21: Number of MIGs expressed in cell lines Here, we identified the number of MIGs expressed in all three cell lines (LCR-515 MIGs) or PM154 organoids (555 MIGs) and submitted these lists to DAVID, which yielded 81 and 99 significant GO terms, respectively. We then intersected these GO terms reflecting functional enrichment of MIGs at baseline with GO terms generated by MIGs with elevated minor intron retention in the cell lines (39 GO terms) and the organoids (42 GO terms). For cell lines, this intersection approach identified 54 GO terms specific to MIGs expressed at baseline, 12 GO terms specific to MIGs with retention, and 27 shared GO terms. For organoids, 70 GO terms were specific to MIGs expressed at baseline, 13 were specific to MIGs with retention, and 29 were shared. For both cell lines and organoids, we found that: GO terms specific to MIG expressed at baseline included: cell cycle, spliceosome complex assembly, RNA metabolic processes, DNA repair, small GTPase binding, etc.; shared GO terms included MAP kinase activity, nucleotide excision repair, snRNA transcription from RNA polymerase II promoters, ATP binding, nuclear pores, etc.; and GO terms specific to MIG with retention included RAN GTPase binding, protein serine / threonine kinase activity, polA RNA binding, and protein transport. Thus, pathway enrichment by misspliced MIG reflects only a portion of pathways regulated by MIG at baseline, and also includes new terms that are not present when only focusing on MIG expressed at baseline.
[0131] Example 22: Conclusion In the field of PCa, there is an active debate about PCa staging. Therefore, to clarify our analysis, we present a classification scheme of different PCa cell lines based on Bluemn et al. (Figure 29). 22Rv1 represent intermediate PCa stage, are AR positive but express some NEPC markers. In other words, they have some transcriptomic similarity with neuroendocrine PCa cell lines (Figure 31), and therefore we placed them at the transition point from CRPC to NEPC on the graph. In Figure 2B, we swapped the labels of C4-2 and 22Rv1 to be consistent with the previous order.
[0132] Here, we report for the first time that siU6atac and MiS inhibition are potential therapeutic targets. It will take considerable effort and time to translate this finding into the clinic to specifically treat resistant and lethal PCa. We therefore hoped to show that this novel strategy is indeed a treatable approach by comparing it with the most exciting treatments in the field, such as the combination of enzalutamide and EZH2 inhibition. Although this combination approach is not approved by the FDA / EDA, there are several phase 1 and phase 2 trials investigating this combination in CRPC. This combination approach is based on the demonstration by several groups (Mu, P. et al., Science 2017, 355(6320), 84-88; Ku, SY et al., supra) that EZH2 inhibitors resensitize resistant PCa to enzalutamide treatment. Thus, the significance of the findings in Figure 6G is that it clearly shows that siU6atac and MiS inhibition have high therapeutic value and potential. The inventors make this clear in the text.
[0133] Example 23: Discussion "MIG expression distinguishes cancer from benign tissues". MIG expression and the pathways they control are closely associated with oncogenes (Figure 1A). Thus, MIG is a key bottleneck for various molecular pathways downstream of oncogenes and, in this way, is part of the aberrant regulatory logic of cancer-associated phenotypes. MiS therefore likely represents another master regulator of cancer whose inhibition triggers tumor checkpoint disruption. Indeed, MIG expression discriminated between different cancer types, while minor intron retention levels were informative for PCa progression (Figure 1). Thus, MIG, with its diverse functions, is an ideal effector of various pathways driven by oncogenes. Moreover, minor intron splicing may be a critical control node for cancer progression, consistent with the idea that cancer progression requires increased cell cycle speed. Consistent with these findings, we found that MiS snRNA, a key regulator of MiS activity, is upregulated in cancer and throughout cancer progression (Figures 1 and 2A). These findings suggest that MiS activity is enhanced and actively regulated with cancer progression. Furthermore, increased expression of U11, PDCD7, and RNPC3 is associated with increased PCa, AML, and anti-RNPC3 cancer-associated scleroderma, respectively. An even more dramatic upregulation was observed for the MiS catalytic core snRNA U6atac, which is controlled by turnover rather than transcriptional upregulation. According to the "molecular switch theory," U6atac is rapidly turned over. It is therefore the limiting snRNA that determines MiS activity. Cellular stress signals, especially activation of p38 MAPK, rapidly increase the expression of U6atac, with enhanced splicing of MIG, which is otherwise inefficiently spliced or degraded. Thus, stabilization of elevated levels of U6atac can enhance the activity of the minor spliceosome. In other words, the level of U6atac can control the expression of MIG "on demand" when needed.Consistent with this model, upregulation of U6atac was observed in tumors of various tissues compared to their respective benign tumors (Figure 1), a finding that underscores the notion that U6atac snRNA levels are a key regulator of minor spliceosome activity during cancer progression.
[0134] "MiS activity increases with cancer progression". Expression of MIG is uniquely dependent on minor intron splicing efficiency. Thus, it is not surprising that we found differences in minor intron splicing efficiency when comparing different PCa subtypes. Indeed, when we examined U6atac levels during prostate cancer progression, we showed that U6atac expression was closely correlated with Pca progression and positively associated with Pca metastasis (Figure 2A). Results supporting this idea were observed with minor intron splicing reporters. The minor intron splicing reporters were more efficiently spliced in cell lines representing more aggressive cancer stages (Figure 2B). For example, splicing of minor introns, unlike major introns, was more efficient in various therapy-resistant CRPC-adeno cells and various CRPC-NE organoids (Figure 2B). Considering the "molecular switch theory", the regulation of MIG may represent a dynamic adaptation mechanism by cancer cells in response to changing environmental conditions, as seen in therapy resistance. In this sense, we found that MIG splicing increases with prolonged ARSi treatment of prostate cancer. Resistance to ARSi is almost always based on reactivation of the AR axis in prostate cancer. This is consistent with our data showing that MIG splicing is controlled by AR in PCa (Figure 2D-F). These findings, together with our previous findings of U6atac as a potent regulator of MiS activity, strongly suggest that control of U6atac levels will be a point of therapeutic intervention. Together, these findings reveal a unique role for MiS activity in cancer development and progression, positioning MiS as a potential therapeutic target.
[0135] "siU6atac blocks the cell cycle in cancer cells". Considering that U6atac snRNA had no other known function besides MiS, siU6atac specifically inhibited the function of MiS. This was reflected as an increase in the retention of minor introns (Figure 3). Indeed, siU6atac led to the highest number of MIGs with high retention of minor introns compared to published reports of MiS inhibition by other components. Unlike MiS inhibition via other components, the largest number of AS events observed in siU6atac samples were cryptic splicing events carried out by the major spliceosome (Figure 3D). MiS inhibition by siU6atac is expected to disrupt the U4atac-U6atac-U5 tri-snRNP, which is recruited after recognition of the 5' splice site by U11 / U12 di-snRNP. Thus, the exon definition requirements are mostly met, and exon skipping by the major spliceosome is unlikely to occur (Figure 3D, red). This finding reinforces the idea that inhibition of the minor spliceosome induces specific types of AS events depending on the specific MiS component targeted and the cell type targeted. This finding opens a new approach to target snRNAs that can directly inhibit the minor spliceosome.
[0136] The cell type-specific response to MiS inhibition is reflected as a different number of MIGs with minor intron splicing changes per cell line (Figure 3C). Indeed, principal component analysis by either minor intron retention or AS showed that C4-2 responded closer to LNCaP, while 22Rv1 was closer to PM154, in response to MiS inhibition (Figure 11). This finding is consistent with our understanding that LNCaP, C4-2, 22Rv1, and PM154 can be ordered in this order in PCa progression. The separation of PM154 organoids from the three cell lines was also evident in the overlap analysis of minor intron retention and AS events (Figure 3). STRING network analysis showed how the simultaneous aberrant splicing of MIGs converged to a central node composed of downregulated genes that are mainly enriched for cell cycle control (Figure 3). Interestingly, similar analysis was performed on PM154 organoids, but did not yield similar results. This suggests that MiS inhibition is cell type-specific, tumor stage- and context-dependent, which is consistent with the hierarchical clustering of the nine cancer types using MIG expression (Figure 1D). However, IPA analysis with MIG and downregulated genes interacting with PCa in the three cell lines and organoids showed distinct cancer-related biological pathways (Figure 3). This underlines the idea that MiS inhibition can perturb different MIGs in different cell lines and still converge on the same biologically relevant endpoints of cancer (i.e., cell cycle and survival). For example, biological network analysis revealed a MAPK-centric network in adeno cell lines, whereas PARP1-, ACTL6A-, and SRPK1-driven networks were observed in organoids. The latter finding is consistent with the idea that CRPC-NE organoids are molecularly distinct from the three CRPC-adeno cell lines.
[0137] Perturbations in biological pathways predicted based on transcriptome changes are ultimately implemented by the proteins produced. Mass spectrometry analysis showed downregulation of many aberrantly spliced MIGs. However, we could not observe a one-to-one correlation (Figure 4). This is consistent with published reports of discordance between the transcriptome and the proteome. Nevertheless, proteins (MIGs and non-MIGs) that were up- or downregulated under siU6atac conditions were associated with PCa progression. For example, AURKA, which has been shown to be important for CRPC differentiation and aggressiveness, was increased in treatment-sensitive LNCaP cells but decreased in CRPC-adeno cells 22Rv1 and CRPC-NE organoid PM154. Similarly, the polycomb group protein EED, known to control the expression levels of AR and a potential target in CRPC, was decreased in C4-2 and 22Rv1 cells. In contrast, we observed increased protein levels of the G1 cell cycle arrest mediator CDN1A in all CRPC-adeno cells (Figure 4A-E). Although there is an essential discrepancy between the transcriptome and the proteome, we found that genes with both transcript and protein downregulated were highly enriched in the GO terms cell cycle control and DDR. Thus, despite dynamic fluctuations in mRNA and protein levels, the core molecular defect is one that inhibits proliferation. Indeed, FACS analysis confirmed G1 / G0 arrest in C4-2 cells and PM154 organoids (Figure 4).
[0138] "ScRNAseq reveals decreased AR and EMT scores in siU6atac-treated cells and organoids". The cell cycle defects observed by FACS analysis (Figure 4) were further confirmed by scRNAseq analysis, which revealed that siU6atac-treated cells were enriched in G1 phase (Figure 5). We also observed an increase in the number of siU6atac-treated cells in S phase, consistent with enrichment of the DDR pathway. Again, these findings clearly demonstrate a cell type-specific effect of MiS inhibition that ultimately results in blocking proliferation, a key feature for any therapeutic strategy. Consistent with this, we observed a decrease in EMT activity after MiS inhibition. This suggests that targeting MiS blocks not only cancer progression but also metastasis (Figures 5 and 13). However, the main strategy for PCa treatment is to block the activity of AR. Interestingly, we found that inhibition of MiS leads to a decrease in AR activity scores in PCa cells and organoids (Figures 5 and 14). This reveals inhibition of AR transcriptional activity as a novel mechanism of action of MiS that may be cancer type or even cancer stage dependent. Considering that MiS activity, and thus MiS-dependent signaling, is tissue and cancer type specific, this finding indicates that MiS may represent a dynamic adaptive mechanism of cancer cells responding to changing environmental conditions. Thus, MiS may be a common denominator of prominent cancer-promoting axes, such as the AR axis in PCa, and may be utilized as an all-in-one target against many cancer types.
[0139] "The minor spliceosome is essential for PCa proliferation and survival". siU6atac-mediated MiS inhibition substantially affected cancer cells but not human fibroblasts or benign mouse prostate cells (Figure 6A and B). This finding has therapeutic relevance since MiS inhibition appears to be specifically detrimental to highly proliferative cells. KD of U6atac led to a decrease in LNCaP cell viability, whereas the proliferation index was not significantly affected by MiS inhibition (Figure 6C). This difference in response by LNCaP cells can be explained by their slow cell division rate, which was shown to be dependent on AR activation. Consistent with previous results naming AR as a regulator of MiS activity, we found that overexpression of AR increased the proliferation of LNCaP cells, thereby making them sensitive to siU6atac-mediated MiS inhibition (Figure 6C). Thus, U6atac is a promising target for CRPC-adeno PCa with persistent (re)activation of AR. Our study clearly indicates the importance of inhibiting MiS, which consists of important components other than U6atac. Indeed, we found that using siRNA against a unique MiS protein, RNPC3, resulted in similar effects in PCa as seen with siU6atac (Figure 6A). However, we found that MiS inhibition can have differential effects based on both the cell type inhibited and the MiS component targeted for inhibition. For example, we found that siU6atac, while fairly broadly effective, was best in PCa and glioblastoma cells (Figure 6D). Similarly, targeting other MiS components, such as RNPC3 and PDCD7, showed significant results in PCa and glioblastoma cells (Figure 18). In contrast, KD of U12 had no significant effect on viability in all cell lines tested (Figure 18).
[0140] Interestingly, KD of U6atac alone had no effect on bladder cancer cells, but was highly effective when combined with cisplatin (Figure 6E and F). This finding suggests that U6atac inhibition may be an effective complement to combination therapy for bladder cancer. Considering that many patients with invasive bladder cancer are cisplatin-ineligible, the combination of cisplatin with siU6atac may represent a promising option for treating those patients. Similarly, enzalutamide-resistant CRPC may benefit from combination therapy including siU6atac. Enzalutamide alone increased CRPC growth in resistant cells, highlighting the need for new CRPC treatments (Figure 6G). For example, EZH2 inhibition has been shown to overcome enzalutamide resistance in cultured PCa cells and xenograft models. Surprisingly, we found that KD of U6atac itself was sufficient to block the proliferation of CRPC cells, which was superior to enzalutamide, siEZH2, or siEZH2 / enzalutamide (Figure 6G). Importantly, we showed that MiS inhibition also works in organoids derived from PCa patients (CRPC-adeno and CRPC-NE), whose transcriptome signatures correlate with those of patients tested in the SU2C study (Figure 6H).
[0141] Taken together, we show that MiS activity plays a key role in PCa progression and that MiS inhibition is a viable therapeutic target. We show that inhibition of various MiS components can block cancer cell proliferation and survival, but U6atac was the most effective in our study. We show that MiS activity corresponds to AR signaling across the stages of PCa progression, reflected by increased U6atac expression. Indeed, this finding suggests that U6atac may also be used as a diagnostic marker for lethal PCa and other cancers. Regardless of the stage of PCa, siU6atac can successfully inhibit proliferation and survival by disrupting pathways such as MAPK, cell cycle, and DNA repair. Although the current study used PCa as a model to examine the efficacy of MiS inhibition as a therapeutic strategy, these observations may be extended to other cancer types.
[0142] Example 24: Materials and Methods Cell and organoid lines LNCaP (male, ATCC, RRID: CVCL_1379), C4-2 (male, ATCC, RRID: CRL-3314), 22Rv1 (male, ATCC, RRID: CRL-2505), PC3 (male, ATCC, RRID: CRL-3470), DLD-1 (2 (male, ATCC, RRID: CCL-221), L-rENZ, and L-AR cells were cultured on poly-L-lysine-coated plates in RPM supplemented with 10% FBS (Gibco, 10270106) and 1% penicillin-streptomycin (Gibco, 11548876). I medium (Gibco, A1049101). RWPE cells (male, ATCC, RRID: CVCL_3791) were maintained in Keratinocyte Serum-Free Medium (Gibco, 17005075) supplemented with bovine pituitary extract and human recombinant EGF, 1% penicillin-streptomycin (Gibco, 11548876). HEK293T cells (female, ATCC, RRID: CVCL_0063), VCaP (male, ATCC, RRID: CRL-2876), MDA-MB-231 (female, ATCC, RRID: HTB-26) K-562 (female, ATCC, RRID: CCL-243), LN-18 (male, ATCC, RRID: CRL-2610), PC-3M-Pro4 and DU145 cells (male, ATCC, RRID: CVCL_0105) were maintained in DMEM (Gibco, 31966021) supplemented with 10% FBS and 1% penicillin-streptomycin. NCI-H660 cells (male, ATCC, RRID: CRL-5813) were maintained in DMEM supplemented with 5% FBS, 1% penicillin-streptomycin (Gibco, 11548876), 0.005 mg / ml insulin ... Cells were maintained in RPMI medium (Gibco, A1049101) supplemented with 0.01 mg / ml apo-transferrin (Sigma-Aldrich, I9278), 0.01 mg / ml apo-transferrin (Sigma-Aldrich, T1147), 30 nM sodium selenite (Sigma-Aldrich, S9133), 10 nM hydrocortisone (Sigma-Aldrich, H6909), 10 nM beta-estradiol (Sigma-Aldrich, E2257) and L-glutamine (final concentration 4 mM) (Sigma-Aldrich, G7513).PC-3M-Pro4 cells were kindly provided by Dr. Kruithof-De Julio. LNCaP-AR cells were kindly provided by Dr. Sawyers and Dr. Mu (Memorial Sloan Kettering Cancer Center). L-ENZ cells were established by constant exposure to enzalutamide. Briefly, low passage LNCaP cells were treated overnight with 20uM enzalutamide in C / S medium. The next day, the medium was replaced with regular RPMI (10%FBRS, 1%P / S) and surviving LNCaP cells (~10%) were maintained until they reached ~80% confluence. This procedure was repeated twice. The concentration of enzalutamide was then increased to 40uM for the 3rd treatment and to 80uM for the 25th treatment. From then on, cells were treated once a week with 80uM enzalutamide.
[0143] All cell lines were cultured at 37 °C and 5% CO2. All cell lines were authenticated by STR analysis and regularly (every 3 months) tested for mycoplasma.
[0144] CRPC-Adeno patient-derived organoids from MSKCC-PCa8, 10, 14, and 16 were kindly provided by Dr. Chen (Memorial Sloan Kettering Cancer Center). All organoids, including WCM154, were maintained in 3D according to previously described procedures (Puca, L., et al., Nat Commun 2018, 9(1), 2404; Gao, D., et al., Cell 2014, 159(1), 176-187). Briefly, GlutaMAX 1x (Thermo Fisher Scientific, 35050061), HEPES 1mM (Thermo Fisher Scientific, 15630056), AA 1x (Life Technologies, 15240-062), 1% penicillin-streptomycin, B27 (Thermo Fisher Scientific, 17504001), N-acetylcysteine 1.25mM (Sigma-Aldrich, A9165), recombinant mouse EGF 50ng / ml (PeproTech, 315-09), human recombinant FGF-10 20ng / ml (Peprotech, 100-26), recombinant human FGF-basic 1ng / ml (Peprotech, 100-18B), A-83-01 500nM (Tocris, 29-391-0), SB202190 Advanced DMEM (Thermo Fisher Scientific, 31966047) supplemented with 10 μM (Sigma-Aldrich, S7076), nicotinamine 10 mM (Sigma-Aldrich, N0636), (DiHydro)testosterone 1 nM (Fluka, 10300), PGE2 1 μM (Tocris, 2296), Noggin conditioned medium (5%) (PeproTech, 120-10C) and R-spondin conditioned medium (5%) (PeproTech, 315-32). The final resuspended pellet was mixed with growth factor reduced matrigel (VWR, BDAA356239) in a 1:2 volume ratio. A drop of 40 μl of the cell suspension / matrigel mixture was pipetted into each well of a 6-well cell suspension culture plate (Huberlab, 7.657185). The plate was placed in a cell culture incubator at 37°C and 5% CO2 for 30 min to allow the droplets to solidify.Then, 3 ml of human organoid medium was added to each well. During the growth of the organoids, 50% of the medium was replaced every 3-4 days. The organoids were passaged as soon as they reached a size of 200-500 um. For this purpose, the organoid droplets were mixed with TrypLE Express (Gibco) and placed in a 37 °C water bath for up to 5 min. The resulting cell clusters and single cells were washed and re-cultured following the procedure described above.
[0145] In situ validation collection Tissue microarrays were kindly provided by the Translational Research Unit (TRU) Platform, University of Bern (www.ngtma.com). For PCa, we used TMAs from the Bern PCBM cohort (Briganti, A. et al., Eur Urol 2013, 63(4), 693-701) (28 patients) and tissue microarrays from 210 primary prostate tissues from the European Multicenter Study of High-Risk Prostate Cancer Clinical and Translational Research Group (EMPaCT) (Tosco, L. et al., Eur Urol Focus 2018, 4(3), 369-375; Chys, B. et al., Front Oncol 2020, 10, 246; Dawson, H. et al., Histopathology 2020, 76(4), 572-580).
[0146] Mass spectrometry analysis LNCaP, C4-2, 22Rv1, and PM154 cells (400,000 cells) were seeded in 6-well plates and treated with siScrambled or siU6atac RNA (16 pmol) for 96 h. 96 h after transfection, cells were harvested and 50% of the cell pellet was used for KD confirmation of U6atac by qRT-PCR. The remaining pellet was washed twice with PBS and subjected to mass spectrometry (MS) analysis:
[0147] Cells were lysed in 8M urea / 100mM Tris pH8 / protease inhibitors by sonication for 1 min with 10 s intervals on ice. The supernatant was reduced, alkylated and precipitated overnight. The pellet was resuspended in 8M urea / 50mM Tris pH8 and protein concentration was measured using the Qubit Protein assay (Invitrogen). 10μg of protein was digested with LysC for 2 h at 37°C followed by trypsin overnight at room temperature. 800ng of digest was loaded in random order onto a precolumn (C18 PepMap100, 5μm, 100A, 300μm ID x 5mm length) with solvent C (0.05% TFA in water / acetonitrile 98:2) at a flow rate of 50μL / min.
[0148] After loading, peptides were eluted in a Homepack analytical nanocolumn (Reprosil Pur C18-AQ, 1.9 μm, 120A, 0.075 mm i.d. × 500 mm length) in backflush mode using an acetonitrile gradient from 5% to 40% solvent B (0.1% formic acid in water / acetonitrile 4, 9:95) at a flow rate of 250 nL / min for 180 min. The column effluent was directly coupled to a Fusion LUMOS mass spectrometer (Thermo Fischer, Bremen; Germany) via a nanospray ESI source.
[0149] Data acquisition is performed in data-dependent mode, with precursor ion scans recorded in the Orbitrap at a resolution of 120,000 (m / z = 250) comparable to the top-speed fragment spectra of the most intense precursor ions in the Linear Trap with cycle times of up to 3 seconds.
[0150] Generation of U6atac-overexpressing cell lines LV290591-RNU6ATAC lentiviral vector (human) (CMV) (pLenti-GIII-CMV-GFP-2A-Puro) and the corresponding empty vector control were purchased from ABM. DNA was amplified by chemical transformation of One Shot Mach1 T1 phage-resistant chemically competent E. coli cells (Invitrogen, C862003). Lentivirus was produced in HEK293T cells by transfection of the construct, and then transduced into C4-2 cells using the medium containing the virus. Three days after transduction, cells were subjected to puromycin selection (1 μg / mL). After the selected cells reached 80% confluence, FACS sorting was performed for GFP positivity. This was repeated three times.
[0151] Drug treatment For DHT stimulation experiments, cells were hormone-starved for 48 h in phenol red-free RPMI medium (Gibco, 11-835-030) and 10% charcoal-stripped FBS (Gibco, A3382101) and then treated with 10 nM dihydrotestosterone (Fluka, 10300) for 24 h.
[0152] For long-term ADT treatment, cells were exposed weekly to 20 μM enzalutamide (Selleck Chemicals, S1250) or 10 uM abiraterone.
[0153] For proliferation experiments, cells were treated with siRNA and then 2 hours later with 20 μM enzalutamide, which was re-administered 3 days later.
[0154] For splicing reporter assays, cells were exposed to 1ug / ml anisomycin (Sigma Aldrich, A9789) for 4 hours.
[0155] Proximity ligation assay L-AR cells (50,000 cells) were seeded on inventorsll μ-Slide8 (ibidi, 80826). The next day, cells were washed once with ice-cold PBS and fixed with 4% PFA for 10 min. Cells were then permeabilized with PBS + 0.2% Triton for 10 min. Proximity ligation assays using Duolink® In Situ Red Starter Kit Mouse / Rabbit (Sigma-Aldrich, DUO92101-1KT) were performed according to the manufacturer's instructions. Briefly, primary monoclonal antibodies against mouse AR (Thermo Fisher Scientific, MA5-13426), rabbit HSP90 (Abcam, ab203085), rabbit PDCD7 (Abcam, 121258), and rabbit IgG isotype control antibody (Thermo Fisher Scientific, 026102) were diluted in Duolink antibody diluent (1:50, 1:200, 1:100, and 1:1000). Cells were incubated in AB solution overnight at 4°C. The next day, cells were washed twice and incubated with PLUS and MINUS PLA probes for 1 h at 37°C in a humidity chamber. Afterwards, cells were washed twice and incubated in ligation mix for 30 min and amplification solution for 100 min at 37°C (humidity chamber). After two final washes for 10 min, slides were mounted with DAPI-containing medium and monitored under a fluorescent microscope (LEICA, DMI4000B).
[0156] Cell transfection and siRNA-mediated knock-down cell ON-TARGET plus siRNA SMART pool siRNAs for U6atac, AR, EZH2, PDCD7, mouse RNPC3 and non-targeting (siScrambled) siRNAs were purchased from Dharmacon. siRNAs for RNU6atac and RNU12 and Silencer Select negative control were purchased from Thermo Fisher Scientific, and siRNA for mouse U6atac was purchased from Ambion. Adherent cells were transfected at each time point with 16 pmol of 20 μM siRNA per well using Lipofectamine™ RNAiMAX transfection reagent (Thermo Fisher Scientific, 13778150).
[0157] Organoids Prior to transfection, organoids were cultured in human organoid growth medium for 2-3 weeks. The medium was removed and these organoids were first mechanically dissociated. To obtain single cells, organoids were trypsinized in 1 ml of TriplE (Thermo Fisher Scientific, 12605036) for 15-18 min at 37°C. The reaction was stopped with 1 ml of growth medium and cells were spun at 300g for 5 min. Cells were then filtered and counted. One million cells per condition were plated in 6 wells. Lipofectamine™ RNAiMAX complexes were prepared according to the standard Lipofectamine™ RNAiMAX procedure: 5ul of RNAiMAX reagent and 40nM siRNA+10%FBS were each diluted in 125ul of Opti-MEMH medium. Both mixtures were pooled and incubated for 10 min before adding the siRNA reagent complex to the cells. The cell / siRNA mixture was centrifuged at 600g for 60 min at 32°C and then incubated overnight at 37°C. The next day, the cells were resuspended and collected by centrifugation (300g, 5 min, room temperature). The pellet was resuspended in 280ul of Matrigel and the mixture was separated into 7 drops, which were added to 6 wells. Organoids were cultured in human organoid medium for 96 hours (CTG assay) or 7 days (cell counting assay).
[0158] RNA extraction from cells and qRT-PCR Cells were harvested for RNA isolation using the ReliaPrep™ miRNA Cell and Tissue Miniprep System (Promega, Z6212). Synthesis of complementary DNA (cDNA) using the FIREScript RT cDNA Synthesis Kit (Solis BioDyne, 06-15-00200) and real-time reverse transcription PCR (RT-PCR) assays using HOT FIREPol EvaGreen qPCR Mix Plus (Solis BioDyne, 08-24-00020) were performed using the manufacturer's protocol. Quantitative real-time PCR was performed on a ViiA 7 system (Applied Biosystems). All quantitative real-time PCR assays were performed with three technical replicates. For relative quantification of quantitative real-time PCR data, GAPDH, ACTB were used as housekeeping genes.
[0159] Single cell sequencing Cell counting and viability assessment were performed using a ViCell XR cell counter and viability analyzer (Beckman Coulter, BA30273). GEM generation and barcoding, reverse transcription, cDNA amplification, and 3' gene expression library generation steps were then all performed using the specified 10x Genomics reagents according to the Chromium Next GEM Single Cell 3' Reagent Kit v3.1 User Guide (10x Genomics CG000204 Rev D). In general, 11.8–27.5 μL of each cell suspension (600–1,400 cells / μL) and 15.7–31.4 μL of nuclease-free water were used for the recovery of 10,000 target cells. GEM generation was followed by GEM reverse transcription incubation, clean-up steps, and 10–12 cycles of cDNA amplification. The resulting cDNA was quantitatively and qualitatively assessed using a Thermo Fisher Scientific Qubit 4.0 fluorometer with the Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific, Q32854) and an Advanced Analytical Fragment Analyzer System with the Fragment Analyzer NGS Fragment Kit (Agilent, DNF-473), respectively. 3' gene expression libraries were then constructed using 11-12 cycles of sample indexing PCR steps. The generated cDNA libraries were quantitatively and qualitatively tested using fluorimetry and capillary electrophoresis, as described above. The cDNA libraries were pooled and sequenced on an Illumina NovaSeq6000 sequencer using a NovaSeq 6000 S2 Reagent Kit v1.5 (100 cycles; Illumina 20028316) and two NovaSeq 6000 S4 Reagent Kits v1.5 (200 cycles; Illumina 20028313) with a loading concentration of 300 pM (150 pM for runs using the XP workflow), paired-end and single index. Read settings were as follows: read 1: 28 cycles, i7 index: 8 cycles, i5: 0 cycles, and read 2: 91 cycles.The quality of the sequencing runs was assessed using the Illumina Sequencing Analysis Viewer (Illumina version 2.4.7), and all base call files were demultiplexed and converted to FASTQ files using Illumina bcl2fastq conversion software v2.20. All steps were performed on the next-generation sequencing platform at the University of Bern.
[0160] Bulk RNA sequencing Total RNA was extracted from LNCaP, C4-2, 22Rv1, and PM154 cells treated with siU6atac or siScrambled for 96 h. DNase treatment was included as recommended. The quantity and quality of extracted RNA was assessed using a Thermo Fisher Scientific Qubit 4.0 fluorometer with the Qubit RNA BR Assay Kit (Thermo Fisher Scientific, Q10211) and an Advanced Analytical Fragment Analyzer System with the Fragment Analyzer RNA Kit (Agilent, DNF-471), respectively. cDNA libraries were then generated using a combination of Illumina Stranded Total RNA Prep, Ligation with Ribo-Zero Plus (Illumina, 20040529) and IDT for Illumina RNA UD Index Sets A and B (Illumina, 20040553 and 20040554, respectively). Illumina protocols were followed with the recommended input of 100ng of total RNA. The quantity and quality of the generated NGS libraries were assessed using a Thermo Fisher Scientific Qubit 4.0 fluorometer with the Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific, Q32854) and an Advanced Analytical Fragment Analyzer System with the Fragment Analyzer NGS Fragment Kit (Agilent, DNF-473), respectively. As a further quality control step, prior to NovaSeq 6000 sequencing, the pooled cDNA library pool was subjected to paired-end sequencing on an iSeq 100 sequencer using iSeq 100 i1 Reagent v2, 300 cycles (Illumina, 20040760).Library pools were repooled to ensure equal numbers of reads / library and then paired-end sequenced on an Illumina NovaSeq 6000 instrument using the NovaSeq 6000 S4 Reagent Kit v1.5, 300 cycles (Illumina, 20028312). The quality of the sequencing run was assessed using Illumina's Sequencing Analysis Viewer (Illumina version 2.4.7), and all base call files were demultiplexed and converted to FASTQ files using Illumina's bcl2fastq conversion software v2.20. The average number of reads / library was 82 million. RNA quality control assessment, library creation, and sequencing runs were performed on the next-generation sequencing platform at the University of Bern, Switzerland.
[0161] U6atac in situ hybridization mRNA ISH was performed with automated staining using Bond RX (Leica Biosystems) and Basescope® technology (Advanced Cell Diagnostics, Hayward, CA, USA). All slides were dewaxed with Bond dewaxing solution (product code AR9222, Leica Biosystems), heat-induced epitope retrieval was performed at 95°C for 15 min at pH 9 in Tris buffer base (code AR9640, Leica Biosystems) and protease treatment was performed for 5 min. The following probes from RNAscope 2.5 LS (Advanced Cell Diagnostics) were used: BaseScope™ LS probe-BA-Hs-RNU6ATAC-1zz-st-C1 product number 1039918, PPIB-1zz product number 710178, DapB-1zz product number 701028 were used as positive and negative controls, respectively. The efficiency of the probe was tested using 50% U6atac-overexpressing C4-2 cells (5 million cells) treated with siU6atac RNA.
[0162] All probes were incubated for 120 min at 37°C. The Basescope™ 2.5 LS assay (product no. 323600, Advanced Cell Diagnostics) was used as a preamplification system. The reaction was then visualized using Fast red (Bond polymer Refine Red detection, Leica Biosystems, product no. DS9390) as a red chromogen for 20 min. Finally, the samples were counterstained with hematoxylin, air-dried and mounted with Aquatex (Merck). Slides were scanned and photographed using a Pannoramic 250 (3DHistech). U6atac intensity was manually scored by a pathologist (Mark Rubin) blinded to the clinical data, using the digital online TMA scoring tool Scorenado (University of Bern, Switzerland), which was specifically developed for TMA scoring of de-arrayed spots.
[0163] The sample annotated as "center" was used for the analysis of TMA data. The U6ATAC score was calculated by multiplying the percentage of positive cells by the intensity. If multiple values were recorded in one block, the sample with the highest score was used. Comparisons between groups were performed using the Wilcox test.
[0164] Comparison of U6ATAC expression in PCa and PCBM included 325 primary, 25 potentially metastatic and 32 metastatic samples from 24 patients.
[0165] Flow cytometry C4-2 and PM154 cells were seeded in 6-well plates (500,000 cells / well) and transfected with siU6atac or siScrambled RNA for 72 and 96 h as described above. Flow cytometric cell cycle analysis was performed using the Click-iT™ EdU Alexa Fluor™ 488 Flow Cytometry Assay Kit (Thermo Fisher Scientific, C10420). Briefly, EdU (10 uM) was added to the medium and cells were incubated for 1 h at 37°C. Cells were washed with 1% BSA in PBS and fixed with 100 ul of Click-iT fixative for 15 min. After three further washing steps, cells were permeabilized with 100 ul of 1xClick-iT saponin-based reagent for 15 min. Click-iT reaction cocktail was prepared according to the manufacturer's instructions and 500 ul of reaction mix / condition was incubated with cells for 30 min at room temperature. Cells were washed and resuspended in 500ul of saponin-based permeabilization buffer. Hoechst (1ug / ml) was added to the reaction mixture 20 minutes before analysis. Cells were analyzed on a BD LSR II Flow Cytometer (BD Biosciences) using FACSDiva software at the FACSlab Core facility of the University of Bern. Data were further quantified with FlowJo 10.7.1. Values were calculated as fold change compared to siScrambled-treated controls.
[0166] Immunoblotting Cells were lysed in GST-Fish buffer (10% (v / v) glycerol, 50 mM Tris-HCl pH 7,4, 100 mM NaCl, 1% (v / v) Nonidet P-40, 2 mM MgCl2, 1 mM PMSF) with freshly added protease and phosphatase inhibitors. Total protein concentration was measured using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). 50ug protein samples were resolved by SDS-PAGE on 4-15% Mini-Protean TGX gels (BioRad, 456-1084) and transferred to nitrocellulose membranes using the iBlot2 system (Thermo Fisher, IB23001). Blots were blocked with 5% milk / TBST or BSA / TBST for 1 hour at room temperature and incubated overnight at 4°C with primary antibodies dissolved in 5% BSA / TBST buffer. The membrane was washed three times and then incubated with horseradish peroxidase-conjugated secondary antibodies for 1 h at room temperature. After three washes, the signals were visualized by chemiluminescence using Luminata Forte substrate (Thermo Fisher Scientific, WBLUF0100) for strong antibodies and The inventorssternBright Sirius-HRP substrate (Witec AG, K-12043-D10) for weak antibodies. Images were captured with a FUSION FX7 EDGE imaging system (Witec AG).
[0167] Luciferase reporter assay CMV-luc2CP / ARE (major intron splicing reporter), CMV-luc2CP (empty vector backbone control) and luc1CFH4 (minor intron splicing reporter) were kindly provided by Dr. Gideon Dreyfuss (University of Pennsylvania). DNA was amplified via chemical conversion of One Shot Mach1 T1 phage-resistant chemically competent E. coli cells (Invitrogen, C862003) and subjected to Sanger sequencing.
[0168] To measure the splicing rate of minor / major introns, cells were seeded (8000 / well) in white 96-well plates (Huberlab, 7.655 098) and treated according to the assay conditions. 24 hours before analysis, cells in each condition were co-transfected with each reporter plasmid and empty vector backbone plasmid. Briefly, 1.5ul of P3000 reagent and 0.5ug of DNA, and 1.5ul of Lipofectamine P300 were each diluted in 25ul of Opti-MEMH medium. Both mixtures were pooled and incubated for 20 minutes before the solution was added to the cells. Luciferase expression was measured using the Dual-Glo® Luciferase Assay System (Promega, E2940): medium was removed, 100ul of PLB was added, and cells were frozen at -20°C for 2 hours. After a 1 hour shaking step, 100ul of LAR substrate was added and firefly luciferase expression was measured on a Tecan Infinite M200PRO reader, values were calculated as x-fold of CMV-luc2CP expression and then x-fold of the respective reference control.
[0169] Minigene reporter assay pRSV2-p120-AmpR XL10 (p120), pCMV7.1 SCN4A fragment (hSCN4A) and pLUX hSCN8A-minigene AmpR STbl3 (hSCN8a) were kindly provided by Dr. Mark-David Ruepp (King's College London). DNA was amplified by chemical transformation of One Shot Mach1 T1 phage-resistant chemically competent E. coli cells (Invitrogen, C862003).
[0170] To determine the splice index of minigenes, cells were seeded in 6 wells (350,000 / well) and treated according to the assay conditions. SiRNA was added 96 hours before measurement. Cells were transfected with each minigene 72 hours before measurement. Briefly, 10 μl of P3000 reagent + 1.5ug DNA and 7.5 μl of Lipofectamine P3000 were each diluted in 125 μl of Opti-MEMH medium. Both mixtures were pooled and incubated for 20 minutes before the solution was added to the cells. 48 hours before measurement, the medium was replaced with medium supplemented with 10% charcoal stripped FBS (Gibco, A3382101), and 100 nM DHT was added to each condition 24 hours before measurement. qRT-PCR was performed to confirm the KD and determine the splice index of each minigene. The minigene splice index was calculated by forming a ratio between the normalized mRNA levels of minigene-transfected cells and the mRNA levels of WT cells, taking into account the transfection efficiency, and then dividing the value corresponding to the spliced minigene by the value corresponding to the unspliced minigene.
[0171] Cell proliferation experiments survival rate Cells were seeded (400,000 cells) in 6-well plates and treated overnight according to assay conditions. Cells were then seeded (8000 cells / well, n=3 / condition) in Poly-L-Lysine-coated 96-well plates, and WCM154 cells were seeded (5000 cells / well, n=3 per condition) in collagen-coated 96-well plates. The remaining cells were used for U6atac KD control by qRT-PCR. Cell viability was measured using the CellTiter-Glo® Luminescent Cell Viability Assay on a Tecan Infinite M200PRO reader after 24, 48, 72, and 96 hours according to the manufacturer's instructions (Promega, G9243). Viability values were calculated as x-fold of cells transfected with siRNA for 0 hours. Cell confluence (n=4 per condition) was determined using an IncuCyte S3 instrument and IncuCyte S3 2018B software (Essen Bioscience, Germany). Values were calculated as x-fold of timepoint 0, then calculated as fold-change of confluence compared to siScrambled-treated control.
[0172] Organoids Organoids were transfected with siRNA as previously described. The next day, 160000 cells were suspended in 320ul of Matrigel and 40ul droplets (1 drop / well, time point 4, n=2) were plated in suspension in 48-well plates (Huberlab, 7.677 102). The remaining cells were plated in 6 wells for U6atac KD control in qRT-PCR. The 24-well plates were incubated at 37°C for 3 min and then incubated upside down at 37°C for 20 min. Afterwards, 500ul of organoid medium was added and viability was measured using the CellTiter-Glo 3D Cell Viability Assay (Promega, G9683).
[0173] Co-immunoprecipitation For co-immunoprecipitation (co-IP), the cytoplasmic fraction of LNCaP-AR cells was isolated using the Universal CoIP kit (Active Motif, 54002). Chromatin in the cytoplasmic fraction was mechanically sheared using a Dounce homogenizer (Fisher Scientific, 11898502). Plasma membranes and debris were pelleted by centrifugation, and protein concentrations of the clarified lysates were measured using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, 23227). 1 μg of rabbit anti-PDCD7 antibody (ab131258, Abcam) and rabbit anti-AR antibody (ab133273, Abcam), and 1 μg of rabbit IgG isotype control antibody (Thermo Fisher Scientific, 026102) were incubated with 1 mg of protein supernatant overnight at 4°C with gentle rotation. The following morning, 30 μl of protein G magnetic beads (Active Motif) were washed twice with 500 μl of CoIP buffer and incubated with the antibody-containing lysate for 2 h at 4 °C with gentle rotation. The bead-bound PDCD7 or AR complexes were washed twice with CoIP buffer, followed by two washes with a buffer containing 150 mM NaCl, 50 mM Tris-HCL (pH 8), and protease and phosphatase inhibitors. The washing steps were performed at 4 °C with gentle rotation. The bead-bound proteins, supernatants, and input controls were reduced and denatured in 40 μl of Laemmli buffer containing DTT by boiling at 95 °C for 5 min. The magnetic beads were removed from the solution using a magnetized pipette rack (Thermo Fisher Scientific, 11757325) and 20 μl of reduced proteins were loaded onto an SDS-PAGE gel, followed by immunoblotting using iBlot (Life Technologies).The membranes were blocked with 5% BSA solution and then incubated overnight with the respective antibodies against the targets of interest: AR, PDCD7, AR45 (androgen receptor antibody (carboxy-terminal antigen), Cell Signaling Technologies, 54653S), and ZRSR2 (Abcam, ab223062). Protein signals were detected using HRP-conjugated native anti-rabbit IgG antibody (CST, #5127) and Luminata Forte substrate (Thermo Fisher Scientific, WBLUF0100) using a FUSION FX7 EDGE imaging system (Witec AG).
[0174] Protein-protein interaction network analysis Assuming that cancer genes perturb large molecular networks through interactions with other genes, we determined the distances of MIGs in the protein-protein interaction network of 160,881 interactions between 15,366 human proteins currently in the HINT database (Das, J., et al., supra). In particular, MIGs that directly interact with cancer-causing genes are separated by a distance of d=1, while proteins separated by intermediate proteins are separated by a distance of d=2 from the cancer gene in question. To determine these interaction distances, we considered a list of 403 cancer-causing genes from 186 different cancer types currently in the Cancer Genome Interpreter database (https: / / www.cancergenomeinterpreter.org) in the 160,881 interactions of the underlying protein-protein interaction network between 15,366 human proteins currently in the HINT database. The profile of the number of MIGs that are separated by a given distance d from each cancer gene,
number
number
number
[0175] To find the tight web of MIGs and cancer genes, we scatter these direct interactions between cancer genes and MIGs in the underlying human protein-protein interaction network and determine the size of the largest connected subnetwork S. We randomly sample MIGs N = 100,000 times. The size of the largest subnetwork is determined by the number of these random sets, S r Calculate using
number
[0176] Principal component analysis We integrated MIG expression data from three sources: prostate samples from GTEx (healthy prostate tissue), prostate cancer samples from TCGA (primary prostate cancer samples), and prostate cancer samples from SU2C (advanced prostate cancer). Gene expression values were normalized according to the procedure employed by GTEx (and detailed in our pan-cancer analysis). PCA analysis was performed on this normalized gene expression matrix, allowing the resulting data to be visualized as a projection onto the space spanned by the first two PCs. Remarkably, this visualization showed that it captured the progression from these three broad stages of prostate cancer progression, from healthy tissue in GTEx (lower end of the first PC) to advanced stages in SU2C (upper end of the first PC).
[0177] Quantitative comparison between MIG-based and non-MIG-based gene expression clustering The Silhouette Coefficient was used to characterize the relative performance of gene expression clustering for gene sets containing different relative abundances of MIGs and non-MIGs. The Silhouette Coefficient provides an objective measurement metric that measures the visual discernment of the presence (or lack thereof) of structure in a given heatmap. This coefficient constitutes an unsupervised approach that provides a score ranging from -1 to 1, with scores closer to 1 indicating well-defined and denser clustering (i.e., more meaningful structure in a given heatmap). This coefficient quantifies how similar a given data point (i.e., sample) is to its own cluster versus different clusters. In our study, the data points consist of vectors of length N, where N is the number of distinct samples in the expression matrix, and the number of distinct clusters is predefined as n_cluster=9 in our pan-cancer analysis, since this analysis was performed on datasets of nine different cancer types. The silhouette coefficient has been employed for similar purposes in previous studies (Belorkar, A. et al., BMC Bioinformatics 2016, 17(Suppl 17), 540; Van Laere, SJ et al., Clin Cancer Res 2013, 19(17), 4685-96; Zhao, S. et al., Biol Proced Online 2018, 20, 5; Lovmar, L. et al., BMC Genomics 2005, 6, 35).
[0178] Before calculating the silhouette coefficient, agglomerative clustering was performed on the normalized gene expression matrix. Gene expression values from 9 cancer types (biliary tract cancer, breast cancer, colorectal cancer, lung cancer, ovarian cancer, pancreatic cancer, prostate cancer, gastric cancer, thyroid cancer) with a total of N=444 samples were obtained from PCAWG, and gene expression normalization was performed using the same approach as that adopted in GTEx (PMID:29022597). Briefly, the entire gene expression matrix was normalized using quantile normalization. Inverse quantile normalization was then applied to this quantile-normalized matrix to map it to a standard normal distribution (which also allowed the inventors to remove outliers). For PCa progression analysis, gene expression values from GTEx, TCGA, and SU2C were integrated into one expression matrix. Briefly, the pre-normalized expression matrix was normalized using quantile normalization and subsequently mapped to a standard normal distribution by inverse quantile normalization; this step also removed outlier genes. This normalization scheme is the same as that employed in the previous GTEx study (PMID: 29022597). Using this normalized gene expression matrix, clustering was performed using hierarchical clustering by employing the Euclidean metric and Ward Linkage.
[0179] This entire analysis was performed using different expression matrices with different percentages of MIG genes, and the results are plotted in Figure 1F and G. Each box in this plot (e.g., the right-most box representing the 100% non-MIG set) corresponds to a 1000-length simulation in which non-MIGs are randomly sampled from among all non-MIGs in the genome. Thus, for the right-most box, 1000 random sets were sampled, each with a composition of 100% non-MIGs. P values are based on two-tailed t-tests, with each sample value taken as the difference between the Silhouette coefficient of one of the 1000 100% non-MIG samples and the corresponding Silhouette coefficient when only MIGs are used for clustering (i.e., the Silhouette coefficient corresponding to 0% non-MIGs).
[0180] MSI analysis Primary tumor RNA-seq patient samples from The Cancer Genome Atlas (TCGA) were randomly queried using the GenomicDataCommons package (http: / / github.com / Bioconductor / GenomicDataCommons) in R. The minor intron retention pipeline developed by Olthof et al. (https: / / github.com / amolthof / minor-intron-retention) was run on queried TCGA samples from the following cohorts: breast invasive carcinoma (BRCA; N=20), cholangiocarcinoma (CHOL; N=20), colon adenocarcinoma (COAD; N=20), lung adenocarcinoma (LUAD; N=20), ovarian serous cystadenocarcinoma (OV; N=20), pancreatic adenocarcinoma (PAAD; N=20), prostate adenocarcinoma (PRAD; N=16), and thyroid carcinoma (THCA; N=20). For prostate cancer analysis, additional samples were analyzed from the Stand Up to Cancer dataset: androgen receptor (AR; N=20) and neuroendocrine (NE; N=22). NE samples were those with an NEPC score >0.4, and AR samples were those with an NEPC score ≤0.4. Prostate cancer samples from the Genome-Tissue Expression Portal were analyzed similarly (GTEX; N=20). Kruskal-Wallis and post-hoc Dunn tests were performed between the TCGA and prostate cohorts. Heatmaps were generated using the default clustering method for the prostate across MIG and pan-cancer TCGA cohorts using the gplots package in R (https: / / CRAN.R-project.org / package=gplots). GO enrichment analysis was performed on the clustered genes for each cancer cohort, and genes with average MSI values in the ranges of 0, 0–0.04, 0.04–0.75, and 0.75–1 were grouped together.
[0181] Quantification of RNU11 according to Gleason score Gene expression data for primary prostate cancer samples were retrieved from The Cancer Genome Atlas (TCGA) in the form of raw counts. Sequencing reads were aligned to the human reference genome (hg38) using STAR (Lovmar, L. et al., BMC Genomics 2005, 6, 35). Gene expression was quantified at the gene level using Gencode annotations (v29) (Harrow, J. et al., Genome Res 2012, 22(9), 1760-74). Subsequent analysis and library size normalization were performed using the edgeR pipeline (Chen, Y. et al., F1000Res 2016, 5, 1438). RNU11 mapping reads were identified in 23 of 497 samples (5%), indicating that this gene product is difficult to capture with canonical PolyA+ sequencing technology. Nevertheless, a clear association was identified between Gleason score and RNU11 mRNA expression in these 23 samples. Significance was assessed using the nonparametric Wilcoxon test.
[0182] Single-cell RNAseq Single cell pretreatment and quality control: Reads were aligned to the human genome reference sequence (GRCh38) using Cell ranger analysis pipeline v6.0.1, and gene-cell matrices were generated from these data. Gene expression matrices were analyzed using Seurat 4.0.3 (https: / / github.com / satijalab / seurat). We removed low-quality cells and multiplets by excluding genes detected in less than 5 cells and discarding cells with ≥ 10000 but < 1000 genes detected. Cells with > 25% mitochondrial gene counts were also removed.
[0183] UMI counts were normalized with the NormalizeData Seurat function using the LogNormalize normalization method with default parameters (10000 scale.factor).
[0184] Cell cycle phase classification and cell score: Prediction of cell cycle phase for each cell was performed using the Seurat CellCycleScoring function. Scores were calculated and cell phases (G2 / M, S, and G1) were assigned as previously described (Tirosh, I. et al., Science 2016, 352(6282), 189-196). Fisher's exact test was performed to determine whether sIU6 cells had significantly different cell numbers compared to Scr cells in G1 or S phase using the fisher.test function in R.
[0185] We used the Seurat AddModuleScore function to assess the degree to which individual cells express a particular predefined gene set. We defined scores to estimate the activity of the prostate AR pathway and EMT status as previously described (Dong, B. et al., Communications Biology 2020, 3(1), 778). The AR pathway gene set included AR, KLK3, KLK2, FKBP5, TMPRSS2, FOXA1, GATA2, SLC45A3, and CDH2, CDH11, FN1, VIM, TWIST1, SNAI1, ZEB1, ZEB2, and DCN in EMT status. Violin plots were drawn using Seurat, and p-values were calculated using the Wilcoxon test (Hadley Wickham, ggplot2: Elegant Graphics for Data Analysis 2016).
[0186] Integration and batch modification of multiple datasets: To combine multiple datasets and minimize batch effects between them, we combined our six samples for each cell line (three replicates for SCR and three replicates for siU6) following the procedure in Seurat v4.0.3 (Stuart, T. et al., Cell 2019, 177(7), 1888-1902.e21).
[0187] Briefly, we used the FindVariableFeatures function (selection.method=“vst”) to select the most variable genes for each dataset and ranked them according to the number of datasets in which they were independently identified as highly variable. The 2000 most variable genes were then combined by merging pairs of datasets according to a given distance.
[0188] Integration anchors represent two cells predicted to originate from a common biological state in both datasets using Canonical Correlation Analysis (CCA), which was performed using the FindIntegrationAnchors function. For each pair of anchor cells, the expression difference between the two expression vectors was used to correct the expression of the target dataset. This step was performed using the IntegrateData function. This process resulted in an expression matrix containing the batch effect-corrected expression of 2000 selected genes for all cells of the six samples for each cell line.
[0189] Dimensionality reduction and clustering: PCA was performed on the scaled data using the RunPCA Seurat function (npcs=30). Uniform Manifold Approximation and Projection (UMAP), a nonlinear dimensionality reduction method, was performed using the RunUMAP Seurat package function to embed cells in 2D space. A K-nearest neighbor graph (KNN) based on Euclidean distance in PCA space was constructed and cells were clustered with the Louvain algorithm (resolution=0.2) using the FindNeighbors and FindClusters Seurat functions. We selected the optimal clustering resolution using the clustree R package (v0.4.3) (Zappia, L., et al., GigaScience 2018, 7(7)). Bar plots were performed using the dittoSeq R package (Bunis, DG et al., Bioinformatics 2020, 36(22-23), 5535-5536).
[0190] Differential gene expression analysis: Differentially expressed genes (DEGs) were identified between the different clusters using the FindAllMarkers function from the Seurat package (one-sided Wilcoxon rank sum test, p-values were adjusted for multiple testing using Bonferroni correction). To calculate DEGs, all genes were tested for their expression in at least 25% of cells in either of the two populations being compared and had an expression difference of at least 0.25 on the natural log scale. Heatmaps were created by selecting the top 5 genes in each cluster using the DoHeatmap Seurat function.
[0191] RNAseq Gene Expression: Paired-end total RNA reads from each replicate (N=4) were mapped to the mm10 genome via Hisat2 (Kim et al., 2015). Reads mapping to multiple alignments were excluded. Gene expression values were calculated using IsoEM2 (Mandric et al., 2017). Differential gene expression was calculated by IsoDE2 (Mandric et al., 2017), which uses a 200-replicate bootstrapping method to generate 95% confidence intervals for the expression of each gene and then statistically compares these values between experimental conditions. We adopted thresholds of log2FC≥1, P≤0.01 for upregulation and log2FC≤-1, P≤0.01 for downregulation.
[0192] Minor intron retention: We report minor intron retention as the misspling index here, by the methodology described in Olthof et al., 2019. Briefly, we extracted uniquely mapped reads from the region of interest (from 2 exons upstream to 2 exons downstream) around the minor intron. We then calculated the misspling index by dividing the sum of reads mapping to the 5' splice site and 3' splice site of the minor intron by the sum of reads mapping to the 5' splice site, 3' splice site, and 2x canonically spliced reads. We considered only introns that met our filtering criteria (requiring more than 4 exon-intron boundary reads, ≥1 read mapping to both 5' splice site and 3' splice site, and intron coverage >95%) as retained in all replicates of a condition. Statistically significant overall minor intron retention was determined using the Kruskal-Wallis test and post-hoc Dunn test (P ≤ 0.05). Determination of individual MIGs with significantly elevated minor intron retention was calculated using a two-tailed Student's T test (P ≤ 0.05).
[0193] Alternative splicing: We adopted the methodology reported by Olthof et al., 2017 for our alternative splicing analysis. Briefly, we used BEDTool to classify the difference in 5' and 3' splice site usage around the region of interest for all minor introns, binning them into one of eight categories (Figure 3D). We then calculated the missplicing index by quantifying the number of AS reads per category per sample divided by the sum of AS reads and canonically spliced reads. We used a filtering criterion that considered an intron to have AS only if the average missplicing index across all replicates for each condition was greater than 10%. Furthermore, we normalized the number of reads supporting an AS event by the total sequencing depth. Therefore, we included only AS events with more than one read per 3 million uniquely mapped reads in our analysis. Determination of individual MIGs with significantly elevated AS was calculated using a two-tailed Student's T test ( P ≤ 0.05).
[0194] DAVID's Analysis: Gene lists were submitted to DAVID for gene ontology (GO) enrichment. We considered only GO terms with a Benjamini-Hochberg adjusted P value ≦0.05 as significant.
[0195] STRING and Ingenuity Pathway Analysis: In the case of LNCaP, C4-2, and 22Rv1, overlapping MIGs with significantly elevated minor intron retention, which were also found to be associated with causative genes in prostate cancer, were grouped with overlapping protein-coding genes with significant downregulation. This list was submitted to STRING under default parameters to obtain a gene interaction network. The same list was then submitted as a core analysis to IPA with default parameters. All biological networks and pathways reported by IPA were significant using a cutoff of Benjamini-Hochberg adjusted P-value ≤ 0.05. The same analysis was also performed on PM154 alone.
[0196] Principal component analysis: Principal component analysis was performed using clustvis (Metsalu and Vilo, 2015) with default parameters. Ellipses indicate 95% confidence intervals.
[0197] GSEA Analysis: To perform gene set enrichment analysis, RNAseq data were pre-ranked using the metric: log10(P-value) / sign(logFC). GSEA was performed using GSEA v.4.0.3 software. Hallmark gene sets were obtained from the GSEA website (www.broadinstitute.org / gsea / ) and used for enrichment of siU6atac-related pathway genes. Dot plots were used to visualize the most significantly enriched terms. Normalised enrichment score (NES) and false discovery rate (FDR) were applied to screen for siU6atac pathway enrichment after 1000 permutations of gene sets for analysis.
[0198] Mass spectrometry (MassSpec) MS data were interpreted using MaxQuant (version 1.6.1.0) against the SwissProt human database (release 2019_07) using default MaxQuant settings, which allowed for a precursor ion mass deviation of 10 ppm for the initial search, and a maximum peptide mass of 5500 Da; matching between runs with a matching time window of 0.7 min was effective, but prevented between different groups of replicates by the use of non-consecutive fractions. Additionally, the four cell lines were treated as different parameter sets and normalized independently. Settings that differed from the default also included: strict trypsin cleavage rules allowing three missed cleavages, fixed carbamidomethylation of cysteines, variable oxidation of methionines, and acetylation of protein N-termini.
[0199] Protein intensities are reported as MaxQuant Label Free Quantification (LFQ) values, and Top3 values (sum of intensities of the three most intense peptides); for the latter, variance stabilization was used for peptide normalization. Missing peptide intensities were input in the following way, provided there was at least one identification within a group: two missing values in a group of replicates were replaced by a draw from a Gaussian distribution of width 0.3 × sample standard deviation centered at the mean of the sample distribution minus 1.8 × sample standard deviation, while one missing value per group would be replaced according to the Maximum Likelihood Estimation (MLE) method. Imputation at the protein level was performed for both LFQ and Top3 values if at least one group of replicates had at least two measured intensities; missing values in this case were drawn from a Gaussian distribution of width 0.3 × sample standard deviation centered at the mean of the sample distribution minus 2.5 × sample standard deviation. Differential expression tests were performed using a moderated t-test empirical Bayes (R function EBayes from the limma package version 3.40.6) on the imputed LFQ and Top3 protein intensities. Additionally, the Benjamini and Hochberg method was applied to correct for multiple testing. The criterion for statistically significant differential expression was a maximum adjusted p-value of 0.05 for large fold changes, which asymptotically decreased to 0 as the log2 fold change approached 1 (curve parameter 1x the overall standard deviation). The protein imputation step was repeated 20 times to allow flagging proteins that showed persistently significant differential expression throughout the cycles.
[0200] Relative protein abundance was calculated using imputed iTop3. Differential expression was calculated using empirical Bayes test. Protein up- and down-regulation was determined by setting thresholds of Benjamini-Hochberg adjusted P-value ≦0.05, log2 FC (fold change) ≧1 or ≦-1, respectively.
[0201] JPEG2024543181000007.jpg254153JPEG2024543181000008.jpg254153
Claims
1. A pharmaceutical nucleic acid agent capable of downregulating or inhibiting the activity of the minor spliceosome (MiS) for use in the treatment or prevention of recurrence of cancer.
2. The pharmaceutical nucleic acid agent for use according to claim 1, wherein the agent is capable of downregulating or inhibiting the expression of snRNA U6atac.
3. 3. A pharmaceutical nucleic acid agent for use according to claim 1 or 2, wherein the agent is or encodes an antisense oligonucleotide.
4. A pharmaceutical nucleic acid agent for use according to any one of claims 1 to 3, wherein the agent is or encodes an siRNA.
5. The pharmaceutical nucleic acid agent for use according to any one of claims 1 to 4, wherein the cancer is an advanced, treatment-resistant phenotype cancer.
6. The pharmaceutical nucleic acid agent for use according to any one of claims 1 to 5, wherein the cancer is selected from glioblastoma, breast cancer, chronic myeloid leukemia (CML), bladder cancer, colon cancer, renal cancer, and prostate cancer.
7. The pharmaceutical nucleic acid drug for use according to claim 6, wherein the cancer is prostate cancer.
8. The pharmaceutical nucleic acid drug for use according to claim 7, wherein the cancer is selected from castration-resistant prostate cancer (CRPC), neuroendocrine prostate cancer (NEPC), castration-resistant neuroendocrine prostate cancer (CRPC-NE), and small cell prostate cancer.
9. 9. The pharmaceutical nucleic acid agent for use according to any one of claims 1 to 8, wherein the agent is administered in combination with a platinum-containing complex, in particular in combination with a platinum-containing drug selected from carboplatin, satraplatin, cisplatin, dicycloplatin, nedaplatin, oxaliplatin, picoplatin, and / or triplatin.
10. 1. A method for assigning a likelihood of having or developing cancer to a patient, comprising: - When the expression level of snRNA U6tac is 2 or 3 The method further comprising assigning a likelihood of having or developing cancer.
11. - When the expression level of snRNA U6tac is 2 or 3 11. The method of claim 10, wherein the patient is assigned a high likelihood of having or developing cancer of an advanced, treatment-resistant phenotype.
12. 12. The method of claim 10 or 11, wherein the cancer is selected from glioblastoma, breast cancer, chronic myeloid leukemia (CML), bladder cancer, colon cancer, renal cancer, and prostate cancer.
13. The method according to claims 10 to 12, wherein the cancer is prostate cancer.
14. The method according to claims 10 to 13, wherein said expression level is determined in a tumor sample isolated from said patient.
15. A pharmaceutical nucleic acid agent for use according to any one of claims 1 to 9, wherein the patient is assigned to have or be likely to develop cancer, or to have or be likely to develop cancer of an advanced, treatment-resistant phenotype, by the method of claims 10 to 14.
16. A method for treating or preventing the recurrence of cancer in a patient, the method comprising administering to the patient a pharmaceutical nucleic acid agent according to any one of claims 1 to 9.
17. 1. A method for treating or preventing recurrence of cancer in a patient, the method comprising the steps of: a. obtaining a tumor sample from the patient; b. determining the expression level of snRNA U6tac in said sample; c. identifying the patient as at high risk of cancer recurrence if the expression level of the snRNA U6tac is 2 or 3; d. Treating said patient at high risk of cancer recurrence with an anti-cancer drug. The method comprising:
18. The method of claim 17, wherein the patient is treated via administration of a pharmaceutical nucleic acid agent according to any one of claims 1 to 9.
19. A system for carrying out the method according to claims 10 to 14.
20. Use of an agent capable of determining the expression level of snRNA U6atac in the manufacture of a kit for the detection of cancer.
Citation Information
Patent Citations
minor spliceosome test system for modulating cell division
DE102007031574A1
