Active repression of cell fate plasticity by PROX1 safeguards hepatocyte identity and prevents liver tumourigenesis

Abstract

The present invention relates to a method for identifying transcription factors that act as suppressors of unwanted cell fate plasticity. Transcription factors that are identified by the method of the present invention can be used, e.g., in the prediction and treatment of cancer.

Description

[0001] Deutsches Krebsforschungszentrum 1 DK17279PC Stiftung des offentlichen Rechts et al.

[0002] Active repression of cell fate plasticity by PROXI safeguards hepatocyte identity and prevents liver tumourigenesis

[0003] Field of the invention

[0004] The present invention relates to a method for identifying transcription factors that act as suppressors of unwanted cell fate plasticity. Transcription factors that are identified by the method of the present invention can be used, e.g., in the prediction and treatment of cancer.

[0005] Background

[0006] Cell fate plasticity enables stem and progenitor cells to generate all cell types of the body (Sanchez Alvarado and Yamanaka 2014). During differentiation, this plasticity typically is restricted to reach a terminally-differentiated state with a defined phenotype. Conversely, unlocked cellular plasticity emerged as a cancer hallmark (Hanahan 2022). In this context, blocked differentiation or even de- and trans-differentiation of mature cells to a malignant state contributes to neoplasia. Precise mechanisms that govern phenotypic plasticity often remain elusive, but transcription factors are important regulators of this process.

[0007] Individual lineage-specific selectors or master regulators can activate gene regulatory networks to induce specific tissues or cell types (Balsalobre and Drouin 2022; Garcia-Bellido 1975; Lewis 1978; Ohno 1979). Loss of master regulators in mature cells, such as deletion of Pax5 in B cells or Ptfla in acinar cells (Cobaleda, Jochum, and Busslinger 2007; Krah et al. 2015), can cause cancer. Repressors that silence unwanted genes are also important to define cell fate in development (Gray and Levine 1996; Lim et al. 2024). A prominent example is the developmental repressor REST (Schoenherr and Anderson 1995), which represses neuronal genes in non-neuronal cells. However, one repressor silencing one cell identity raises a logistical problem, since hundreds of repressors would be needed to silence all alternate identities in each cell type. Recently, a new kind of cell type-specific “safeguard repressor” that might resolve this conundrum was discovered. Unlike REST, the transcription factor MYT1L is almost exclusively expressed in neurons and directly binds and represses several non-neuronal genes to promote neuronal cell identity (Q. Y. Lee et al. 2020; Mall et al. 2017). MYT1L is expressed lifelong, and loss of MYT1L is associated with mental disorders and brain cancer (Hu et al. 2013; Weigel et al. 2023). Whether this type of repression exists in other lineages, and how it contributes to cellular plasticity in cancer, is unclear. Deutsches Krebsforschungszentrum 2 DK17279PC Stiftung des offentlichen Rechts et al.

[0008] Here, an in silico screen to identify cell type-specific safeguard transcription factors across 18 cell types that promote and maintain cell identity by actively suppressing cell fate plasticity and cancer was developed. Hepatocyte-specific candidates, revealing that PR0X1 enhanced hepatocyte identity by repressing alternate cell fate master regulators during reprogramming was experimentally validated. In mice, Proxl was required for efficient hepatocyte regeneration after injury and sufficient to block liver tumour initiation and progression. Interestingly, manipulating PR0X1 levels can toggle transformed hepatocytes between cholangiocarcinoma and hepatocellular carcinoma fates. Together, these findings support a model in which cell typespecific safeguard repressors actively suppress unwanted cell fate plasticity to induce and maintain cell identity and block tumourigenesis.

[0009] Overview on the present invention / Definitions

[0010] The present invention relates to a method for identifying one or more candidate transcription factors which suppress plasticity in a cell type of interest, comprising: a) assessing the expression levels of a plurality of transcription factors in i) the cell type of interest and ii) control cell types which differ from the cell type of interest, b) selecting one or more transcription factor which has a high expression level in the cell type of interest, but a low expression level in the control cell types, c) counting the number of binding motifs for each transcription factor selected in step b) in the promoter regions of i) signature genes of the cell type of interest and ii) signature genes of the control cell types, and d) identifying a candidate transcription factor which suppresses plasticity in the cell type of interest based on the result of step c).

[0011] In accordance with the present invention, transcription factors which suppress plasticity in a cell type of interest shall be identified. Such transcription factors preferably, also maintain the cell identity of the cell type of interest. They could be used as markers in the prognosis of cancer. Thus, the present invention also relates to the identification of markers for the prognosis of cancer. Moreover, the studies carried out in the context of the present invention indicate that the overexpression of such transcription factors could prevent tumour formation. Therefore, such transcription factors, or vectors expressing such transcription factors could be used in the treatment or prevention of cancer. Accordingly, the method of the present invention allows for identifying transcription factors which can be in the treatment of prevention of cancer.

[0012] The individual steps of the method of the present invention are preferably carried out as described herein below. Moreover they can be carried out as in the Examples section. In general, Deutsches Krebsforschungszentrum 3 DK17279PC Stiftung des offentlichen Rechts et al. terms used herein are to be given their ordinary and customary meaning to a person of ordinary skill in the art and, unless indicated otherwise, are not to be limited to a special or customized meaning.

[0013] In step a) of the above method the expression levels of a plurality of transcription factors shall be assessed. The term “plurality of transcription factors” as used herein, preferably, refers to at least 10, more preferably, at least 100, even more preferably at least 500, and most preferably at least 1000 transcription factors.

[0014] The term “expression level” preferably refers to the expression level of the gene encoding the transcription factor, e.g. the level of mRNA encoding the transcription factor. In an alternative embodiment, the expression level is the level of the transcription factor protein.

[0015] As used herein, the term “level” includes any and all measure of quantity deemed suitable by the skilled person, and in particular includes an absolute amount of a transcription factor as such (or the mRNA of encoding the transcription factor), a relative level, or a concentration of the a transcription factor as such (or the mRNA of encoding the transcription factor), as well as any value or parameter which correlates thereto or can be derived therefrom, in an embodiment by standard mathematical operations. Such values or parameters comprise intensity signal values from all specific physical or chemical properties obtained from the said compounds by direct measurements, e.g., intensity values in mass spectra or NMR spectra. Moreover, encompassed are all values or parameters which are obtained by indirect measurements specified elsewhere in this description.

[0016] The assessment of the expression levels is preferably the determination of the expression level. The term “determining” as used herein refers to semi quantitative or quantitative determination of the expression of transcription factor referred to herein. Preferably, the expression level of the plurality of transcription factors is determined (or has been determined) in a sample of cells of the individual cell types. Thus, the determination of the expression level is done in vitro. In an embodiment, the expression levels are (or have been) determined at a single cell level, for example by RNA sequencing.

[0017] Information on the expression level of transcription factors can be also obtained from databases. Such databases (such as single-cell expression atlas described in the examples) are well known in the art.

[0018] In a preferred embodiment of the above method, step a) thus comprises a) providing information on the expression levels of a plurality of transcription factors in i) the cell type of interest and ii) control cell types which differ from the cell type of interest, Deutsches Krebsforschungszentrum 4 DK17279PC Stiftung des offentlichen Rechts et al.

[0019] The “cell type of interest” is the cell type for which transcription factor which suppress plasticity should be identified. It can be any cell type deemed appropriate. In an embodiment, the cell type of interest are astrocytes, B cells, bladder cell, cardiac muscle cells, endothelial cell, epidermal cells, epithelial cells, fibroblasts, granulocytes, hepatocytes, keratinocytes, microglial cells, monocytes, neurons, oligodendrocytes, skeletal muscle satellite cells, T cells, or type B pancreatic cells.

[0020] The control cell types are cell types which differ from the cell type of interest. Preferably, at least 5, more preferable at least 10 and more preferably at least 17 different control cell types are used. Preferably, the expression level is determined (has been determined) for each transcription factor in each cell line.

[0021] It is to be understood that the cell type of interest and the control cell types are from the same species. Preferably, the cell types are mammalian cell types such as mouse, bovine, avian, canine, equine, feline, ovine, porcine, or primate (including humans and non-human primates) cell types. In a preferred embodiment, the method is carried out for human cell types. Moreover, it is to be understood that the transcription factors are from the same species as the analyzed cells. Typically, the transcription factors are naturally occurring transcription factors, such as human transcription factors.

[0022] Step b) of the above method comprises a selection step, i.e. selecting one or more transcription factors which have a high expression level in the cell type of interest, but a low expression level in the control cell types. The selection can be based on the comparison of the levels in the cell type of interest. Preferably, transcription factors are selected which have a high expression level in the cell type of interest versus a low expression level in the control cell types. Accordingly, transcription factors are selected which have an increased expression in the cell type of interest as compared to the expression in the control cell types.

[0023] The selected transcription factors are then further analyzed in step c). In step c) the number of binding motifs (i.e. DNA binding motifs) is counted in the in the promoter regions of i) signature genes of the cell type of interest and ii) signature genes of the control cell types. As promotor region, a region of 2 kb upstream the transcription start site could be used.

[0024] Preferably, the promoter regions of at least 100, more preferably of at least 300, even more preferably of least 500 and most preferably at least 1000 signature genes are analyzed for each cell type. Preferably, the signature genes cell type-enriched genes, i.e. genes that are most highly expressed in the cell type. More preferably, they genes that are most highly expressed in Deutsches Krebsforschungszentrum 5 DK17279PC Stiftung des offentlichen Rechts et al. the cell type, as compared to the other cell types. They can be identified, e.g. as in the Examples by using Seurat FindMarkers.

[0025] In the promoters of each of these signature genes, the number of DNA-binding motifs for each selected transcription factor, is counted for the cell type of interest and for each control cell type. Afterwards the number of binding motifs in the cell type of interest is compared to the number in the control cell types. A low number of binding motifs in the promoter regions of signature genes of the cell type of interest versus a high number of binding motifs in the promoter regions of signature genes of the control cell types is indicative for a transcription factor which suppresses plasticity in a cell type of interest. Thus, the DNA-binding motifs are depleted in promoter regions of signature genes of the cell type of interest, but enriched in promoter regions of signature genes of the control cell types.

[0026] Moreover, a repression score can be calculated based on the expression levels and the DNA- binding motif depletion at signature genes for each transcription factor in each cell type as in the Examples.

[0027] Preferably, the method of the present invention further comprises the step of assessing whether the selected transcription factor exhibits a life-long expression in the cell type of interest. A life-long expression, i.e. throughout life in the respective cell type, further indicates that the factors act as suppressors of unwanted cell fate plasticity.

[0028] The method of the present invention can comprise further steps such as the experimental validated that the identified transcription factor suppresses plasticity in the cell type of interest. For example, the transcription factor can be overexpressed, e.g. in the cell type of interest, to assess whether the overexpression would reduce plasticity. Moreover, it could be overexpressed in embryonic fibroblasts to assess whether the overexpression induces reprogramming in the cell type of interest.

[0029] The definitions and explanations also apply to the following.

[0030] The present invention relates to a method for assessing a whether a transcription factor suppresses plasticity in a cell type of interest, comprising a) providing the expression level of the transcription factor in i) the cell type of interest and ii) control cell types which differ from the cell type of interest, b) comparing the expression level of said transcription factor in the cell type of interest to the expression level in the control cell types, Deutsches Krebsforschungszentrum 6 DK17279PC Stiftung des offentlichen Rechts et al. c) counting the number of binding motifs of said transcription factor in the promoter regions of i) signature genes of the cell type of interest and ii) signature genes of the control cell types, wherein a high expression level in the cell type of interest versus a low expression level in the control cell types, in combination with a low number of binding motifs in the promoter regions of signature genes of the cell type of interest versus a high number of binding motifs in the promoter regions of signature genes of the control cell types is indicative for a transcription factor which suppresses plasticity in a cell type of interest.

[0031] The identified transcription factor, or an agent that increases the level of the transcription factor (preferably in the derived cell type) may be used in the prevention or treatment of cancer. The type of cancer to be prevented or treated typically depends on the cell type for which the transcription factor acts as suppressor of cell fate plasticity. An overview of the cell type and the respective cancer to be treated or prevented is provided in the following table. For example, a transcription factor which acts as suppressor of unwanted cell fate plasticity in hepatocytes can be used in the treatment of hepatocellular carcinoma. For example, a transcription factor which acts as suppressor of unwanted cell fate plasticity in bladder cells can be used in the treatment of bladder cancer. Deutsches Krebsforschungszentrum 7 DK17279PC Stiftung des offentlichen Rechts et al.

[0032] Accordingly, the present invention relates to the transcription factor identified by the method of the present invention, or an agent that increases the level of the transcription factor (preferably in the derived cell type) for use in preventing or treating cancer in a patient.

[0033] The term “transcription factor” typically also encompasses variants having at least 90% sequence identity to the respective transcription factor (preferably on the amino acid level). Preferably, the transcription factor is PROXI, which is as such known to the skilled person. More preferably, the transcription factor is human PROXI, in a preferred embodiment comprising the amino acid sequence of Genbank Acc No. NP_001257545.1.

[0034] The present invention further relates to a method for assessing cancer, comprising the steps of a) determining the level at least one transcription factor identified by the method of the present invention in a sample from the subject, and b) assessing cancer based on the level determined in step a).

[0035] In a preferred embodiment the method comprises comparing the level of said at least one transcription factor to a suitable reference level.

[0036] In a preferred embodiment, the assessment of cancer is the prediction of the risk of developing cancer. Preferably a decreased level of the at least one transcription factor as compared to the reference level is indicative of a subject who is at risk of developing cancer. Preferably an increased level of the at least one transcription factor as compared to the reference level is indicative of a subject who is not at risk of developing cancer.

[0037] In a preferred embodiment, the assessment of cancer is the prediction of the risk of mortality in a patient suffering from cancer. Preferably a decreased level of the at least one transcription factor as compared to the reference level is indicative of a subject who is at increased risk of mortality. Preferably an increased level of the at least one transcription factor as compared to the reference level is indicative of a subject who is at decreased risk of mortality (due to said cancer). Deutsches Krebsforschungszentrum 8 DK17279PC Stiftung des offentlichen Rechts et al.

[0038] In a preferred embodiment, the term “assessing”, as used herein, refers to establishing information about the status of the indicated disease or condition, in particular its severity, prognosis, treatment options, and / or other relevant information. As the skilled person understands in view of the description herein, the method of assessing described herein is an aid in determining the status of a subject with regard to the disease. Thus, the method of assessing preferably is an aid in determining whether a subject suffers from said disease, in particular cancer, and / or on the severity of said disease. The determination may be based on the aforesaid assessment, however, preferably is based on the aforesaid assessment and further diagnostic information, such as anamnesis data, general physical and / or mental examination findings, and / or additional metabolic and / or histological data. Moreover, assessing disease may further include assessing whether a subject is at risk of suffering from the disease, exhibits a medical condition which deteriorates with respect to the disease, and / or establishing a prognosis with respect to the disease. Thus, assessing preferably comprises prognosticating the disease, monitoring disease therapy, predicting outcome of a pre-determined disease therapy, predicting disease relapse, assessing whether a subject is susceptible to a pre-determined disease therapy, and / or establishing a recommendation for disease therapy of a subject. Preferably, the method of assessing as such does not lead to a diagnosis of disease.

[0039] As referred to herein, assessing disease, in a preferred embodiment, may also refer to establishing information about differentiating between subtypes of disease as specified herein below, i.e. said assessing preferably is an aid in differentially diagnosing the indicated disease subtypes; as the skilled person will understand, establishing a differential diagnosis may be based on the aforesaid assessment, however, preferably is based on the aforesaid assessment in combination with further diagnostic information, preferably as specified herein above. Thus, assessing disease may relate to assessing whether a subject suffers from a first subtype of a disease or a second subtype of said disease. Accordingly, assessing as used herein includes classifying subtypes of disease. Preferably, in particular in case of differentiating of subtypes of disease, the subject was diagnosed to show at least one symptom shared by said subtypes. Thus, assessing preferably comprises assisting in triaging said subject, in particular assisting in determining whether said subject is susceptible to a pre-determined therapy, preferably as specified herein below, or is not susceptible to said pre-determined therapy. Thus, assessing preferably comprises assisting in identifying a subject benefiting from a pre-determined therapy and / or assisting in identifying subject not benefiting from said pre-determined therapy.

[0040] In a preferred embodiment, assessing as referred to herein may relate to a rule-in assessment, i.e. to identifying a subject as belonging to a group of subjects sharing a common feature, e.g. as suffering from a severe form of disease. Assessing, however, may also relate to a rule-out assessment, i.e. to identifying a subject as not belonging to a group of subjects sharing a common feature, e.g. as not suffering from a severe form of disease. Thus, the assessment may aid Deutsches Krebsforschungszentrum 9 DK17279PC Stiftung des offentlichen Rechts et al. in establishing a differential diagnosis and / or in triage. As the skilled person is aware of, e.g. excluding a severe form of disease in a subject showing symptoms thereof may be of help to decide on the further course of treatment.

[0041] As will be understood by those skilled in the art, the assessment made in accordance with the present invention, although usually preferred to be, may not be correct for 100% of the investigated subjects. However, the term typically requires that a statistically significant portion of subjects can be correctly assessed. Whether a portion is statistically significant can be determined without further ado by the person skilled in the art using various well known statistic evaluation tools, e.g., as specified herein above.

[0042] In a preferred embodiment, the term "cancer", as used herein, relates to a disease of an animal, including man, characterized by uncontrolled growth by a group of body cells (“cancer cells”). This uncontrolled growth may be accompanied by intrusion into and destruction of surrounding tissue (invasion) and possibly spread of cancer cells to other locations in the body (metastasis). Preferably, also included by the term cancer is a recurrence of cancer after treatment (relapse). Thus, preferably, the cancer is a solid cancer, a metastasis, or a relapse thereof. Cancer may be induced by an infectious agent, preferably a virus, more preferably an oncogenic virus, more preferably Epstein-Barr virus, a hepatitis virus, Human T-lymphotropic virus 1, a papillomavirus, or Human herpesvirus 8. Cancer may, however, also be induced by chemical compounds, e.g. a carcinogen or alcohol abuse, by physical factors, such as ionizing radiation, or endogenously, e.g. caused by spontaneous mutation. Preferably, the cancer is liver cancer, more preferably hepatocellular carcinoma, cholangiocarcinoma, or a mixed form thereof (hepatocellular- cholangiocarcinoma). In the case of cancer, preferably, a "severe form of disease" is a form of cancer with a bad prognosis, preferably with a survival rate of less than 50%, preferably less than 25%, and / or a median life expectancy of less than 2 years, preferably less than 1 year, and / or a TMN staging of at least 3. Correspondingly, a "mild form of disease" preferably is a form of cancer with a good prognosis, preferably with a survival rate of at least 50%, preferably at least 75%, and / or a median life expectancy of at least 2 years, preferably at least three years, and / or a TMN staging of less than 3.

[0043] In a preferred embodiment, the term “sample”, as used herein, refers to a biological sample from a body fluid, preferably blood, plasma, serum, saliva or urine, or a sample derived from cells, tissues or organs, e.g., by biopsy, wherein said sample preferably is known or suspected to comprise T cells. Preferably, the sample is a cell-comprising sample, preferably a sample comprising cancer cells, more preferably tumor sample. Biological samples can be derived from a subject by techniques known in the art. For example, blood samples may be obtained by blood taking, bone marrow sample can be obtained e.g. by aspiration, while tissue or organ samples are to be obtained, e.g., by biopsy. The aforementioned samples may be pre-treated before they Deutsches Krebsforschungszentrum 10 DK17279PC Stiftung des offentlichen Rechts et al. are used according to the present invention. Said pre-treatment may include treatments required to release or separate the biomarker(s) and / or the analyte(s) or to remove excessive material or waste. Suitable techniques comprise centrifugation, extraction, fractioning, ultrafiltration, protein precipitation followed by filtration and purification and / or enrichment of compounds. Moreover, other pre-treatments may be carried out in order to provide the biomarker and / or analyte in a form or concentration suitable for the intended determination. Suitable and necessary pre-treatments depend on the means used for carrying out the methods of the invention and are well known to the person skilled in the art. Pre-treated samples as described before preferably are also comprised by the term “sample” as used in accordance with the present invention.

[0044] In a preferred embodiment, the term “subject”, as used herein, relates to a vertebrate animal, preferably to a mammal. Preferably, the subject is a primate and, more preferably, a human. The subject may be an apparently healthy subject, however, preferably, is a subject known or suspected to suffer from disease. Preferably, the subject known or suspected to suffer from disease is a subject showing symptoms of disease, in particular cancer, such as liver cancer. Preferably, the subject is a subject in need of differential diagnosis of disease. Thus, the subject preferably is a subject showing at least one symptom of subtypes of disease, preferably as specified herein above. In a preferred embodiment, a subject under medical treatment or diagnosis, preferably a human, may also be referred to as a "patient".

[0045] Preferably, the sample to be tested comprises cells of the type for which the transcription factor acts as suppressor of cell fate plasticity. For example, if the transcription factor acts as suppressor of cell fate plasticity in hepatocytes, the sample comprises hepatocytes. For example, if the transcription factor acts as suppressor of cell fate plasticity in bladder cells, the sample comprises bladder cells. Preferably, the determination of the level is done in in vitro. The cancer to be assessed is typically a cancer which is associated with the cell type for which the transcription factor acts as suppressor of cell fate plasticity.

[0046] In a preferred embodiment, the identified transcription factor is a transcription factor that has been identified in the Examples below cell type specific safeguard repressor, or the human counterpart transcription factor. In a particularly preferred embodiment, the transcription factor is PROXI. The transcription factor can be used in the assessment, treatment or prevention of liver cancer, preferably hepatocellular carcinoma and / or cholangiocarcinoma.

[0047] The term "transcription factor" is understood by the skilled person; in a preferred embodiment, the term relates to a polypeptide controlling the rate of transcription of genetic information from DNA to messenger RNA, preferably by binding to a specific DNA sequence. As the skilled person understands in view of the description herein, the transcription factors referred to herein, e.g. PROXI, are at the same time factors determining cell status (cell identity) and biomarkers Deutsches Krebsforschungszentrum 11 DK17279PC Stiftung des offentlichen Rechts et al. of said cell status. Thus, the transcription factors referred to herein may also be referred to as "biomarkers" which serve as indicators for disease or a physiological state as referred to herein. In accordance, the transcription factor as biomarker may be determined as the polypeptide as such, but may also be determined as an other gene product of its gene, e.g. as an mRNA. The transcription factor preferably is detectable in a sample of a subject, in particular is a metabolite of the subject's metabolism. Moreover, the transcription factor as a biomarker may also be a molecular species which is derived from said metabolite. In such a case, the actual metabolite will be chemically modified in the sample or during the determination process and, as a result of said modification, a chemically different molecular species, i.e. the analyte, will be the determined molecular species. Preferred modes of determination and analytes for the biomarkers of the present description are described in the context of the respective biomarkers herein below. Moreover, as is understood by the skilled person, a biomarker according to the present invention need not necessarily correspond to one molecular species. Rather, the biomarker may comprise variant molecular species, e.g. translated from splice variants, glycosylation variants, peptidase processing variants, and the like. In an embodiment, the variants share at least one determinable feature, e.g. an epitope or an activity.

[0048] In a preferred embodiment, the term “determining”, as used herein, refers to semi quantitative or quantitative determination of a transcription factor referred to herein; thus, in a preferred embodiment, "determining a level" of a transcription factor relates to determining any measure of a quantity, e.g. an amount or concentration, of a transcription factor in a sample. Determining a level of a transcription factor may be carried out by any technique which allows for establishing a measure of quantity of a biomarker in a semi quantitative or quantitative manner. In a preferred embodiment, the level of a transcription factor can be determined by determining a complex of an analyte with a detection compound, in particular an antibody or fragment thereof, i.e. in an immunoassay. Said determining of a complex of the analyte may be performed in any format deemed appropriate by the skilled person, in particular a sandwich, competition, or other assay format. Said assays will develop a signal which is indicative for the level of a biomarker. In a preferred embodiment, the transcription factor is determined based on its encoding mRNA or a cDNA derived therefrom, e.g. via qPCR, a quantitative sequencing method, hybridization, or any other method deemed appropriate by the skilled person. The transcription factors to be determined in accordance with the present invention are as such known in the art. Moreover, methods for the determination of the amounts of transcription factors are known to the skilled person as well. For example, the transcription factor can be determined as described in the Examples section. Also, as described in more detail herein below, in addition or as an alternative to a transcription factor as specified herein, a multitude of target genes of said transcription factor may be determined. Deutsches Krebsforschungszentrum 12 DK17279PC Stiftung des offentlichen Rechts et al.

[0049] The term “reference”, as used herein, relates to a value, e.g. an amount or any value derived therefrom, e.g. a score, which can be correlated to a medical condition and, preferably, which allows for the assessment as referred to herein to be made, more preferably enables allocation of a subject into either a group of subjects suffering from a disease or condition or being at risk for developing it, or a group of subjects which do not suffer from said disease or condition or which are not at risk for developing it. Such a reference can be a threshold value, e.g. a threshold amount, which separates these groups from each other. Accordingly, the reference may be a value which allows for allocation of a subject into a group of subjects suffering from a disease or condition or being at risk for developing it, or not. For example, the reference may be a value which allows for allocation of a subject into a group of subjects suffering from cancer, or being at risk of developing cancer. The reference may, however, also be a reference range, e.g., in an embodiment, a range of values for which cancer can be excluded. Furthermore, the reference may be a value calculated from the aforesaid values, e.g. from the levels of one or more transcription factors, optionally in combination with further diagnostic results, in an embodiment to provide a score, e.g. a predictor score, a prognostic score, or any other score deemed appropriate by the skilled person. A suitable reference separating two or more groups of subjects can be provided without further ado e.g. by the statistical tests referred to herein elsewhere based on levels of transcription factors from suitable reference groups as specified herein below. As the skilled person understands, it may not always be possible, although preferred, to provide a reference unambiguously allocating each and every possible level of a transcription factor to one of the aforesaid groups; thus, there may be a range of values for which a clear assessment cannot be provided. Preferably, however, as indicated above, a reference enables the assessment to be made for each and every level of transcription factor which may be measured. As the skilled person understands, the specific value of a reference may depend on the assessment intended and on parameters thereof; thus, the reference value for diagnosing cancer may typically be different from the reference value for predicting cancer. Relevant parameters having an influence on the reference may in particular be sensitivity and specificity of assessment.

[0050] A reference may in particular be derived from at least one reference group, the term "reference group" relating to a group of subjects with known status with regard to the assessment. Thus the reference group may e.g. be a group of subjects for which it is known whether they suffer from cancer. The population of subjects in a reference group preferably comprises a plurality of subjects, e.g. at least 5, 10, 50, 100, 1,000, or 10,000 subjects. Typically, the subject to be diagnosed and the subjects of the said reference group are of the same species. The reference applicable for an individual subject may vary depending on various physiological parameters such as age, gender, or subpopulation. As is understood by the skilled person, in case prevalence of a disease in the population is low, e.g. at most 5%, preferably at most 1%, a reference may also be derived from the average population. Assuming that contribution of actually afflicted subjects is low, such an average population reference group may be treated as a reference group Deutsches Krebsforschungszentrum 13 DK17279PC Stiftung des offentlichen Rechts et al. known not to suffer from cancer; preferably, in such case, the size of the reference group is sufficiently high, e.g. at least 100, more preferably at least 1000, even more preferably at least 10000 subjects. In view of the description herein, the skilled person understands that a reference group may, in principle, also be a mixed population of subjects with regard to cancer, provided that the status of each member of said mixed population with regards to cancer is or becomes known before deriving a reference from such group.

[0051] In a preferred embodiment, reference amounts can, in principle, be calculated for a cohort of subjects based on the average or mean values for a given parameter such as biomarker amount by applying standard statistically methods. In particular, accuracy of a test such as a method aiming to diagnose an event, or not, is best described by its receiver-operating characteristics (ROC) (see especially Zweig 1993, Clin. Chem. 39:561-577). The ROC graph is a plot of all of the sensitivity / specificity pairs resulting from continuously varying the decision threshold over the entire range of data observed. The clinical performance of a diagnostic method depends on its accuracy, i.e. its ability to correctly allocate subjects to a certain prognosis or diagnosis. The ROC plot indicates the overlap between the two distributions by plotting the sensitivity versus 1 -specificity for the complete range of thresholds suitable for making a distinction. On the y- axis is sensitivity, or the true-positive fraction, which is defined as the ratio of number of truepositive test results to the product of number of true-positive and number of false-negative test results. This has also been referred to as positivity in the presence of a disease or condition. It is calculated solely from the affected subgroup. On the x-axis is the false-positive fraction, or 1 -specificity, which is defined as the ratio of number of false-positive results to the product of number of true-negative and number of false-positive results. It is an index of specificity and is calculated entirely from the unaffected subgroup. Because the true- and false-positive fractions are calculated entirely separately, by using the test results from two different subgroups, the ROC plot is independent of the prevalence of the event in the cohort. Each point on the ROC plot represents a sensitivity / -specificity pair corresponding to a particular decision threshold (i.e. reference). A test with perfect discrimination (no overlap in the two distributions of results) has an ROC plot that passes through the upper left corner, where the true-positive fraction is 1.0, or 100% (perfect sensitivity), and the false-positive fraction is 0 (perfect specificity). The theoretical plot for a test with no discrimination (identical distributions of results for the two groups) is a 45° diagonal line from the lower left corner to the upper right corner. Most plots fall in between these two extremes. If the ROC plot falls completely below the 45° diagonal, this is easily remedied by reversing the criterion for "positivity" from "greater than" to "less than" or vice versa. Qualitatively, the closer the plot is to the upper left corner, the higher the overall accuracy of the test. Dependent on a desired confidence interval, a threshold can be derived from the ROC curve allowing for the diagnosis or prediction for a given event with a proper balance of sensitivity and specificity, respectively. Accordingly, the reference to be used Deutsches Krebsforschungszentrum 14 DK17279PC Stiftung des offentlichen Rechts et al. for the aforementioned method of the present invention, i.e. a threshold which allows to discriminate between subjects being at risk and not being at risk can be generated, usually, by establishing a ROC for said cohort as described above and deriving a threshold amount therefrom. Dependent on a desired sensitivity and specificity for a diagnostic method, the ROC plot allows deriving suitable thresholds. It will be understood that an optimal sensitivity may be desired for excluding a subject for being at increased risk (i.e. a rule-out), whereas an optimal specificity may be envisaged for a subject to be assessed as being at an increased risk (i.e. a rule-in).

[0052] In a preferred embodiment, the term “comparing” as used herein encompasses comparing the determined level for a transcription factor as referred to herein to a reference. It is to be understood that comparing as used herein refers to any kind of comparison made between the value for the amount with the reference. However, it is to be understood that preferably identical types of values are compared with each other, e.g., if an absolute amount is determined, the reference shall also be an absolute amount, if a relative amount is determined, the reference shall also be a relative amount, etc. The aforesaid comparison of identical types of values is also referred to a comparing to a "corresponding" value, e.g. a corresponding reference, herein. The term comparing also encompasses comparing a calculated score with a suitable reference score. Thus, preferably, the corresponding reference is a value of the quantitative parameter allowing the assessment to be made.

[0053] In a preferred embodiment, the comparison may be carried out manually or computer assisted. The value of the level and the reference can be, e.g., compared to each other and the said comparison can be automatically carried out by a computer program executing an algorithm for the comparison. The computer program carrying out the said evaluation preferably provides the desired assessment in a suitable output format. As set forth above, it is also envisaged to calculate a score, e.g. a "predictor score", in particular a single score, and to compare this score to a reference score. The calculated score in an embodiment combines information on the amounts of one or more biomarker(s), optionally with additional diagnostic information, such as a Child- Pugh score, a Barcelona Clinic Liver Cancer (BCLC) score, and / or a TNM stage. Moreover, in the score, contributing parameters may be weighted in accordance with their contribution to the establishment of the differentiation, wherein the weighting factor of the individual parameters may be different. The score can be regarded as a classifier parameter for the assessing as set forth herein. In particular, it enables providing the assessment based on a single score. Thus, the skilled person does not have to interpret the entire information on the amounts of the individual parameters. Using a scoring system as described herein, values of different dimensions or units for the biomarkers may be used since the values will be mathematically transformed into the score. The reference score to be applied may be elected based on the desired sensitivity and / or the desired specificity. Deutsches Krebsforschungszentrum 15 DK17279PC Stiftung des offentlichen Rechts et al.

[0054] In a preferred embodiment, as the skilled person understands in view of the description herein, the deviation indicative of a pre-determined assessment depends on the type of reference selected. I.e., in case a healthy reference is used, a deviation, in particular a decrease, of the level of the transcription factor preferably is indicative of disease or of a severe form of disease, while in case a reference from subjects known to suffer from cancer is used, an essential lack of a deviation preferably is indicative of disease, while an increase compared to said reference preferably is indicative of absence of disease.

[0055] In a preferred embodiment, as the skilled person understands in view of the description herein, a multitude of target genes of a transcription factor referred to herein can be used as a surrogate marker of the activity of the transcription factor. Thus, in case PR0X1 is the transcription factor, the means and methods described herein for PR0X1 as a biomarker may also be achieved by using at least 5, preferably at least 10, more preferably at least 25, still more preferably at least 50, most preferably all, PROXI target genes, preferably selected from the list provided in Table 2. Also preferably, at most 119, more preferably at most 100, even more preferably at most 75, most preferably at most 50 of said target genes are determined.

[0056] Table 2: PROX 1 target genes: gene symbols, and human and mouse reference sequence Gen- jank Acc Nos. Deutsches Krebsforschungszentrum 16 DK17279PC Stiftung des offentlichen Rechts et al. Deutsches Krebsforschungszentrum 17 DK17279PC Stiftung des offentlichen Rechts et al. Deutsches Krebsforschungszentrum 18 DK17279PC Stiftung des offentlichen Rechts et al.

[0057] Also, the present invention relates to the transcription factor PR0X1 (preferably human PR0X1), or an agent that increases the level of PR0X1 for use in preventing or treating hepatocellular carcinoma in a patient. In a preferred embodiment, the present invention also relates to the transcription factor PR0X1 (preferably human PR0X1), or an agent that increases the level of PR0X1 for use in treating liver cancer and / or liver damage in a subject, preferably for improving liver regeneration in a subject. Deutsches Krebsforschungszentrum 19 DK17279PC Stiftung des offentlichen Rechts et al.

[0058] In a preferred embodiment, the terms "treating" and “treatment” refer to an amelioration of the diseases or disorders referred to herein or the symptoms accompanied therewith to a significant extent. Said treating as used herein also includes an entire restoration of health with respect to said diseases or disorders. It is to be understood that treating, as the term is used herein, may not be effective in all subjects to be treated. However, the term shall require that, preferably, a statistically significant portion of subjects suffering from a disease or disorder referred to herein, or an identifiable subgroup thereof, can be successfully treated. Whether a portion is statistically significant can be determined without further ado by the person skilled in the art using various well-known statistic evaluation tools, e.g., determination of confidence intervals, p-value determination, Student's t-test, Mann-Whitney-U test etc. Details are found in Dowdy and Wearden, Statistics for Research, John Wiley & Sons, New York 1983. Preferred confidence intervals are at least 95%, at least 97%, at least 98% or at least 99 %. The p-values are, preferably, 0.05, 0.01, 0.005, 0.001, or 0.0001.

[0059] In a preferred embodiment, the term “preventing”, as used herein, refers to significantly reducing the likelihood with which the disease or condition develops in a subject, preferably within a defined window (prevention window) starting from the administration of the PROXI or the agent increasing the level of PROXI. Typically, the prevention window is up to 1 month, preferably up to 2 months, more preferably up to 6 months, even more preferably up to one year, most preferably more than one year. However, it will be understood that the preventive window may, dependent on the kind of agent administered, also be several years up to the entire life time. It will be understood that prevention may not occur in 100% of the subjects to which the PROXI or agent increasing the level of PROXI has been administered. The term, however, requires that the prevention occurs in a statistically significant portion of subjects (e.g. a cohort in a cohort study). Whether a portion is statistically significant can be determined without further ado by a person skilled in the art using well-known statistic evaluation tools, e.g., those described herein above.

[0060] The agent that increases the level of the transcription factor, preferably in the cell type of interest, can be any compound that increases the expression in the cell type of interest. In an embodiment, said agent is a polynucleotide comprising an expressible nucleic acid sequence encoding the identified transcription factor, i.e. under control of a suitable promoter. Said polynucleotide may be present on a vector comprising, for example a viral vector such as a retroviral vector, lentiviral vector, or adenoviral vector, or adeno-associated viral vector (AAV) can be administered to a subject. Preferably, the transcription factor or agent that increases the level of the transcription factor, is administered to the subject such that is delivered to the tissue or organ of interest (in case of hepatocellular carcinoma to the liver). Deutsches Krebsforschungszentrum 20 DK17279PC Stiftung des offentlichen Rechts et al.

[0061] In a preferred embodiment, the agent that increases the level of PROXI, which is also referred to as "agent increasing the level of PROXI" herein, may be any composition of matter causing a PROXI polypeptide to become present, in particular after its application to a cell, preferably to a subject. Thus, the agent increasing the level of PROXI may preferably be any peptide or polypeptide comprising PROXI. From such an agent increasing the level of PROXI, PROXI may be liberated, e.g. by proteolysis (e.g. by a proteasome), by hydrolysis, e.g. of an amido or ester bond to a carrier molecule, and / or by fusion of a lipid vesicle, e.g. of a nanoemulsion, with a cell membrane. Corresponding compositions and methods are known in the art. Also preferably, the agent increasing the level of PROXI is a nanoemulsion comprising at least PROXI; corresponding compositions are known in the art. Also preferably, the agent increasing the level ofPROXl may be the PROXI as such. The agent increasing the level ofPROXl may, however, also be an agent causing a host cell to synthesize PROX 1 or a polypeptide comprising the same; for the peptides and polypeptides which may be produced from such an agent, reference is made to the description herein above. A corresponding agent increasing the level ofPROXl may in particular be a polynucleotide encoding at least PROXI or a polypeptide comprising the same, preferably a polynucleotide encoding at least PROXI. The polynucleotide may be any polynucleotide deemed appropriate by the skilled person for the intended use, taking into account e.g. mode of administration, target cell, required dose and duration of expression, and the like. Thus, the agent increasing the level ofPROXl may be an mRNA, an expression construct, optionally comprised in a vector, and the like. Thus, the agent increasing the level of PROXl preferably is an mRNA or a DNA, preferably double-stranded DNA. In a preferred embodiment, increasing the level of PROX 1 is PROXI overexpression in a cell, preferably compared to cells of the same type in the absence of the agent increasing the level of PROXl. Thus, PROXI overexpression in a liver cancer cell preferably is overexpression compared to a cancer cell in the absence of the agent increasing the level of PROXl. More preferably, said overexpression is constitutive overexpression.

[0062] The terms "liver damage" and "liver regeneration" are understood by the skilled person. In a preferred embodiment, liver damage is a damage caused by a chemical agent, in particular an alcohol, such as ethanol or methanol, or caused by an infectious agent, such as a hepatitis virus.

[0063] Accordingly, the present invention relates to a method for assessing hepatocellular carcinoma comprising the steps of a) determining the level ofPROXl in a sample from the subject, and b) assessing cancer based on the level determined in step a).

[0064] In a preferred embodiment, the present invention relates to a method of assessing liver cancer comprising at least aforesaid steps a) and b), wherein assessing liver cancer may in particular be differentiating between hepatocellular carcinoma and cholangiocarcinoma. Deutsches Krebsforschungszentrum 21 DK17279PC Stiftung des offentlichen Rechts et al.

[0065] Finally, the present invention relates to the use of antibody which binds the identified transcription factor (such as PR0X1), or an antigen binding fragment thereof, for assessing cancer, preferably in a sample from the subject.

[0066] It is to be understood that in the specification and in the claims, “a” or “an” can mean one or more of the items referred to in the following depending upon the context in which it is used. Thus, for example, reference to “an” item can mean that one item or more than one of those items can be utilized.

[0067] As used in the following, the terms “have”, “comprise” or “include” are meant to have a nonlimiting meaning or a limiting meaning. Thus, having a limiting meaning these terms may refer to a situation in which, besides the feature introduced by these terms, no other features are present in an embodiment described, i.e. the terms have a limiting meaning in the sense of “consisting of’ or “essentially consisting of’. Having a non-limiting meaning, the terms refer to a situation where besides the feature introduced by these terms, one or more other features are present in an embodiment described.

[0068] Further, as used in the following, the terms “preferably”, “more preferably”, “most preferably”, "particularly", "more particularly", “typically”, and “more typically” are used in conjunction with features in order to indicate that these features are preferred features, i.e. the terms shall indicate that alternative features may also be envisaged in accordance with the invention.

[0069] Further, it will be understood that the term “at least one” as used herein means that one or more of the items referred to following the term may be used in accordance with the invention. For example, if the term indicates that at least one item shall be used this may be understood as one item or more than one item, i.e. two, three, four, five or any other number. Depending on the item the term refers to the skilled person understands as to what upper limit the term may refer, if any.

[0070] The term "about" in the context of the present invention means + / - 20%, + / - 10%, + / - 5%, + / - 2 % or + / - 1% from the indicated parameters or values. This also takes into account usual deviations caused by measurement techniques and the like.

[0071] In view of the above, the following embodiments are particularly envisaged:

[0072] Embodiment 1 : A method for identifying a candidate transcription factor which suppresses plasticity in a cell type of interest, comprising: Deutsches Krebsforschungszentrum 22 DK17279PC Stiftung des offentlichen Rechts et al. a) assessing the expression levels of a plurality of transcription factors in i) the cell type of interest and ii) control cell types which differ from the cell type of interest, b) selecting one or more transcription factors which has a high expression level in the cell type of interest, but a low expression level in the control cell types, c) counting the number of binding motifs for each transcription factor selected in step b) in the promoter regions of i) signature genes of the cell type of interest and ii) signature genes of the control cell types, and d) identifying a candidate transcription factor which suppresses plasticity in the cell type of interest based on the result of step c).

[0073] Embodiment 2: A method for assessing a whether a transcription factor suppresses plasticity in a cell type of interest, comprising a) providing the expression level of the transcription factor in i) the cell type of interest and ii) control cell types which differ from the cell type of interest, b) comparing the expression level of said transcription factor in the cell type of interest to the expression level in the control cell types, c) counting the number of binding motifs of said transcription factor in the promoter regions of i) signature genes of the cell type of interest and ii) signature genes of the control cell types, wherein a high expression level in the cell type of interest versus a low expression level in the control cell types, in combination with a low number of binding motifs in the promoter regions of signature genes of the cell type of interest versus a high number of binding motifs in the promoter regions of signature genes of the control cell types is indicative for a transcription factor which suppresses plasticity in a cell type of interest.

[0074] Embodiment 3 : A method for assessing cancer, comprising the steps of a) determining the level at least one transcription factor identified by the method of claim 1 in a sample from the subject, and b) assessing cancer based on the level determined in step a).

[0075] Embodiment 4: The method of claim 3, wherein step b) comprises comparing the level of said at least one transcription factor to a suitable reference level.

[0076] Embodiment 5: The method of claim 3 or 4, wherein the assessment of cancer is the prediction of the risk of developing cancer.

[0077] Embodiment 6: The method of any one of claims 3 to 5, wherein the transcription factor is PR0X1 and the cancer in a preferred embodiment is liver cancer, preferably is hepatocellular carcinoma and / or cholangiocarcinoma.

[0078] Embodiment 7: A transcription factor identified by the method of claim 1, or an agent that increases the level of the transcription factor, for use in the prevention or treatment of cancer.

[0079] Embodiment 8: The transcription factor for use of claim 7, wherein the transcription factor is PR0X1 and the cancer is hepatocellular carcinoma. Deutsches Krebsforschungszentrum 23 DK17279PC Stiftung des offentlichen Rechts et al.

[0080] Embodiment 9: The transcription factor for use of claim 7 or 8, wherein the agent is a polynucleotide comprising an expressible nucleic acid sequence encoding the identified transcription factor.

[0081] Embodiment 10: A method for assessing liver cancer in a sample of a subject, comprising the steps of

[0082] (a) determining a level at least one transcription factor being PROXI in a sample from the subject, and

[0083] (b) assessing liver cancer based on the level determined in step a).

[0084] Embodiment 11 : The method of embodiment 10, wherein step b) comprises comparing the level of said at least one transcription factor to a suitable reference level.

[0085] Embodiment 12: The method of embodiment 10 or 11, wherein said assessing cancer is predicting a risk of developing cancer.

[0086] Embodiment 13: The method of any one of embodiments 10 to 12, wherein said assessing cancer is predicting a mortality risk of a subject suffering from cancer.

[0087] Embodiment 14: The method of any one of embodiments 10 to 13, wherein said sample is a sample comprising cancer cells.

[0088] Embodiment 15: The method of any one of embodiments 10 to 14, wherein the cancer is hepatocellular carcinoma.

[0089] Embodiment 16: PROXI or an agent increasing the level of PROXI in a cell for use in the prevention or treatment of cancer.

[0090] Embodiment 17: The subject matter of embodiment 16, wherein said agent increasing the level of PROXI is a polynucleotide comprising an expressible nucleic acid sequence encoding PROXI.

[0091] Embodiment 18: The subject matter of embodiment 16 or 17, wherein said polynucleotide is comprised in a vector, preferably a viral vector, more preferably a retroviral vector, a lenti- viral vector, an adenoviral vector, or an adeno-associated virus (AAV) vector.

[0092] Embodiment 19: The subject matter of any one of embodiments 16 to 18, wherein said cancer is liver cancer, preferably hepatocellular carcinoma.

[0093] Embodiment 20: Use of PROXI for assessing liver cancer.

[0094] Embodiment 21 : Use of PROXI or an agent increasing the level of PROXI in a cell in the manufacture of a medicament for treating and / or preventing cancer.

[0095] Embodiment 22: A method for treating and / or preventing liver cancer in a subject, said method comprising

[0096] (A) administering PROXI or an agent increasing the level of PROXI in a cell to said subject, and

[0097] (B) thereby treating and / or preventing cancer in said subject.

[0098] All references cited throughout this specification are herewith incorporated by reference with respect to the specifically mentioned disclosure content as well as in their entireties. Deutsches Krebsforschungszentrum 24 DK17279PC Stiftung des offentlichen Rechts et al.

[0099] FIGURES

[0100] Figure 1. Safeguard repressors revealed by a computational and experimental screen.

[0101] (a) Transcription factors with high expression in the target cell type, and motif depletion in target cell type-specific promoters, have high safeguard repressor scores (see also Fig. 6c).

[0102] (b) Top six safeguard repressor candidates across eighteen cell types based on safeguard repressor score > 0. Shown are lifelong expression based on Tabula Muris Senis, repressor / acti- vator activity, and tumour suppressor role from published reports. Asterisks: factors reported to promote indicated cell fate (data not shown) Bold: candidates in this study.

[0103] (c) Expression and motif presence analysis of 1,296 transcription factors highlights six hepatocyte safeguard repressor candidates.

[0104] (d) Left, Proxl expression in 18 cell types from Tabula Muris. Right, number of PROXI motifs in promoters of cell type marker genes.

[0105] (e) Predicted survival benefit or deficit of liver candidates. Log rank test based on Kaplan-Meier curves from hepatocellular carcinoma patients in TCGA, segregated by high or low expression of each candidate (see Fig. 6h).

[0106] (f) Developmental expression of the top six hepatocyte repressor candidates in mouse liver (Cardoso-Moreira et al. 2019).

[0107] (g) Experimental validation of safeguard repressor candidates by lentiviral overexpression during 4inl -induced hepatocyte (iHep) reprogramming from mouse embryonic fibroblasts (MEFs).

[0108] (h) Representative TJP1 immunofluorescence images of induced hepatocytes following overexpression of top three hepatocyte repressor candidates or GFP control.

[0109] (i) Number of TJP1+ cells based on immunofluorescence quantification (top) and Albumin secretion from ELISA measurements (bottom) at day 14 of iHep reprogramming with indicated candidates.

[0110] Bar graphs show mean values from three biological replicates, error bars = SD, Dunnett’s test, * p-adj < 0.05, ** p-adj < 0.01.

[0111] Figure 2. PROXI suppresses hepatocyte transformation and liver cancer formation and progression.

[0112] (a) PROXI gene expression in tumour samples from HCC patients (n=62) and paired normal tissue (n=59) from the TIGER-LC dataset (Chaisaingmongkol et al. 2017).

[0113] (b) PROXI protein levels in HCC patient liver sections within tumours and adjacent non-tu- mour tissues.

[0114] (c) Overall survival of 364 HCC patients segregated by PROXI expression levels (40% cutoff for high-expression cohort) (Menyhart, Nagy, and Gyorffy 2018). Deutsches Krebsforschungszentrum 25 DK17279PC Stiftung des offentlichen Rechts et al.

[0115] (d) Overall survival of 1,173 HCC patients segregated by PR0X1 mutation status (chromosomal amplification including PROXI or unaltered) (Ahn et al. 2014; Cerami et al. 2012; Gao et al. 2013; Harding et al. 2019; Ng et al. 2022; Weinstein et al. 2013; R. Xue et al. 2019).

[0116] (e) Confluence percentage of Hep3B cells after 7 days of culture upon induction of PROXI shRNA-knockdown (KD) or overexpression (OE) normalised to uninduced controls.

[0117] (f) Differentially closed and opened regions two days upon PROXI overexpression in Hep3B cells compared to controls determined by ATAC-seq.

[0118] (g) Proxl and hepatocyte marker gene expression in healthy (day 0) and MYC-induced mouse HCC model (day 28) based on single-cell gene expression analysis (Li et al. 2023). Mean expression of Proxl and hepatocyte signature genes across all cells is shown.

[0119] (h) Representative mouse livers following hydrodynamic tail vein injection (HDTVI) to induce Myc overexpression and Trp53 knockout together with Constitutive Proxl overexpression (OE) (n=5).

[0120] (i) Percentage of GFP+ tumours in mice treated as in (h), indicating no Proxl -IRES-GFP positive tumours, at endpoint.

[0121] (j) Overall survival of mice treated as in (g) following Constitutive Proxl overexpression (OE) (n=5) or doxycycline-inducible Late Proxl OE (n=4) at day 14 compared to control (n=8).

[0122] (k) HDTVI-induced livers tumours following KrasG12D overexpression and Trp53 knockout together with Constitutive Proxl overexpression (OE) (n=4).

[0123] (i) Percentage of GFP+ tumours in mice treated as in (k), indicating lack of Proxl -IRES-GFP positive tumours.

[0124] (m) Overall survival of mice treated as in (k) following Constitutive Proxl or control overexpression (n=4).

[0125] Bar graphs and scatter plots show mean values from specified biological replicates, error bars = SD, Mann- Whitney test (b), Log rank test (c,d,j, and m), one-sample t-test assuming 0 as a theoretical mean (e) or unpaired t-test (a, i and 1), * p-adj < 0.05, ** p-adj < 0.01, *** p-adj < 0.001, **** p-adj < 0.0001.

[0126] Figure 3. Multilineage repression by PROXI is sufficient and necessary to promote hepatocyte cell fate.

[0127] (a) Schematic of DDC diet-induced liver injury and regeneration in mice.

[0128] (b) Proxl and hepatocyte signature levels of mice treated as in (a) based on single-cell gene expression pseudotime analysis(Li et al. 2023).

[0129] (c) Liver injury and recovery as in (a) following cre-mediated deletion in conditional Proxlfl / fl knockout mice.

[0130] (d) Number of HNF4+ cells per area in the livers from mice treated as in (c) (n=3).

[0131] (e) Serum levels of Alkaline phosphatase from mice treated as in (c) (n=3). Deutsches Krebsforschungszentrum 26 DK17279PC Stiftung des offentlichen Rechts et al.

[0132] (f) Hepatocyte reprogramming time course with or without Proxl analysed by single-cell RNA- seq, with 22,761 cells from two biological replicates following clustering and UMAP projection.

[0133] (g) Annotation of cells in (f) based on experimental treatment and time point.

[0134] (h) Projection of hepatocyte (top) and fibroblast (bottom) identity scores onto all cells in (f).

[0135] (i) Quantification of hepatocyte (top) and fibroblast (bottom) identity scores in hepatocyte cluster from (f), shown as boxplots with p-values (two-tailed t-test) for each time point and treatment.

[0136] (j) Correlation of various cell identity scores with 4inl or PR0X1 activity.

[0137] (k) Reprogramming of MEFs to induced hepatocytes (top), neurons (middle), or myocytes (bottom) with 4inl, Ascii, or Myodl overexpression, respectively. Representative immunofluorescence of TJP1 (hepatocyte), TUBB3 (neuronal), or Desmin (myocyte) marker proteins at day 14 of respective reprogramming protocols with or without Proxl overexpression.

[0138] (l) Immunofluorescence quantification of cells in (k) (n=3).

[0139] (m) Knockout of Proxl during hepatocyte reprogramming via cre-mediated deletion of exon 2 in Proxlfl / fl MEFs.

[0140] (n) TJP1 immunofluorescence at day 14 of hepatocyte reprogramming in Proxl- / - or Proxlfl / fl cells.

[0141] (o) Analysis of cells in (n) quantifying the number of TJP1+ cells (n=3), and amount of Albumin secretion upon Proxl deletion (n=5). p, Proportion of reprogrammed cells in (n) positive for Desmin or TJP1.

[0142] Bar graphs show mean values from specified biological replicates, error bars = SD, unpaired t- test in (d, e, 1, and o), * p-adj < 0.05, *** p-adj < 0.001.

[0143] Figure 4. Alternative cell fate inducers are directly repressed by PROXI.

[0144] (a) PROXI interaction partners identified by mass-spectrometry upon immunoprecipitation from mouse liver (n=4).

[0145] (b) Fusion proteins containing the PROXI DNA-binding domain (DBD) and the VP64 activator or EnR repressor domains (not to scale).

[0146] (c) Representative TJP1 immunofluorescence upon 4inl -induced hepatocyte reprogramming with indicated PROXI fusion constructs at day 14.

[0147] (d) Quantification of TJP1+ induced hepatocytes generated in (c) (n=3).

[0148] (e) Normalised Albumin secretion of cells in (c) (n=5).

[0149] (f) Proportion of reprogrammed cells in (e) positive for indicated cell type-specific markers and morphology.

[0150] (g) Normalised counts of differentially-expressed genes from RNA-seq of cells in (c), compared to DBD as a control, at day 7 (n=2).

[0151] (h) Percent overlap of up- and down-regulated genes in (g). Fisher test, * p < le-03, ** p < le- 06. Deutsches Krebsforschungszentrum 27 DK17279PC Stiftung des offentlichen Rechts et al.

[0152] (i) Chromatin accessibility (n=3) and gene expression analysis (n=2) comparing hepatocyte reprogramming with or without Proxl at indicated time points. Genes were clustered based on gene (red) up- and (bue) down-regulation and (yellow) in- and (teal) decreased accessibility displayed as scaled logFC compared to control.

[0153] (j) Overlap of cell identity marker genes with genes present in each cluster. Overlaps with p- adj < 0.01 are shown.

[0154] (k) Enrichment or depletion of transcription factor binding from CUT&RUN (PROXI) or motif presence (all others) in the promoters of genes within each cluster. Log2 odds ratio with p-adj < 0.05 shown.

[0155] (l) Computational prediction of the transcription factors most important for PROXI -enhanced hepatocyte reprogramming over time. m, Expression correlation of indicated transcription factors and their target genes predicts activator vs repressor function.

[0156] (n) Representative TJP1 immunofluorescence upon 4inl -induced hepatocyte reprogramming with overexpression or shRNA-mediated knockdown of Prrxl or Pparg at day 14.

[0157] (o) Albumin protein quantification of cells in (n) by Western blot. Data is normalised to respective controls in boxplots (n=5).

[0158] (p) Proposed PROXI gene regulatory network, by which repression of direct downstream transcription factors represses alternate cell identities.

[0159] Bar graphs show mean values, error bars = SD, Dunnett’s test in (d and e), Fisher’s LSD test in (o), * p-adj < 0.05, ** p-adj < 0.01.

[0160] Figure 5. PROXI regulates HCC to CCA fate trajectories in liver cancer.

[0161] (a) PROXI gene expression in tumour samples from cancer patients with HCC (n=62) and CCA (n=91) subtypes from the TIGER-LC dataset (Chaisaingmongkol et al. 2017).

[0162] (b) Correlation of indicated HCC and CCA markers and fate regulators with PROXI expression in patients from (a).

[0163] (c) Immunohistology of HDTVI-tumour models from HCC (Myc / Trp53) and CCA (Akt / Notch) mice at the endpoint generated using hematoxylin and eosin (HE) staining as well as KRT19 and HNF4 antibodies following Proxl -knockdown or overexpression compared to control (each n=5).

[0164] (d) Quantification of KRT19 (CCA-marker) and HNF4 (HCC-marker) positive cells in GFP+ tumours from (c) shown as percentage.

[0165] (e) Transcriptome analysis of tumour nodules from mice in (c) using RNA-seq following knockdown of Proxl (n=2) or Late Proxl overexpression (n=2-3). Heatmap of selected differentially expressed cholangiocyte- and hepatocyte-related genes is shown.

[0166] Bar graphs and scatter plots show mean values from specified biological replicates, error bars = SD, unpaired t-test, ** p-adj < 0.01, *** p-adj < 0.001, **** p-adj < 0.0001. Deutsches Krebsforschungszentrum 28 DK17279PC Stiftung des offentlichen Rechts et al.

[0167] Figure 6. In silico and reprogramming screen identifies safeguard repressors.

[0168] (a) Single cell UMAP of 18 cell types annotated by the Tabula Muris consortium.

[0169] (b) Cell type-specific gene signatures used in this study are displayed across all cells in (a).

[0170] (c) Equations used to calculate a safeguard repressor score for each transcription factor (see Methods for details).

[0171] (d) Expression and motif presence analysis of 1,296 transcription factors highlights three safeguard repressor candidates in neurons, including MYT1L.

[0172] (e) Mytll expression in 18 cell types from Tabula Muris and number of MYT1L motifs in promoters of cell type marker genes.

[0173] (f) Developmental expression levels of the top three neuronal safeguard repressor candidates in the mouse brain (Cardoso-Moreira et al. 2019).

[0174] (g) Odds ratio of PROXI CUT&RUN peaks from mouse liver at promoters of cell type marker genes.

[0175] (h) Kaplan-Meier survival curves for HCC patients in TCGA, segmented by high (blue) or low (orange) expression of each of the candidate hepatocyte safeguard repressors. Log rank p-values are shown.

[0176] (i) Detection of FL AG-tagged hepatocyte safeguard repressor candidates upon overexpression in MEFs at day two of hepatocyte reprogramming using anti-FLAG Western blot.

[0177] (j) Western blot analysis of Albumin and E-cadherin protein levels at day 7 of hepatocyte reprogramming with indicated hepatocyte safeguard repressor candidates.

[0178] (k) Quantification of Albumin and E-cadherin protein expression in (j), normalised to total protein expression.

[0179] (l) Gene expression analysis of indicated hepatocyte markers in cells treated as in (j) using qRT- PCR.

[0180] Bar graphs show mean values from three biological replicates, error bars = SD, Dunnett’s test, * p-adj < 0.05, ** p-adj < 0.01, *** p-adj < 0.001, **** p-adj < 0.0001.

[0181] Figure 7. PROXI is low in HCC tumours and overexpression induces chromatin closure and dose-dependent growth delays in mouse HCC cell lines.

[0182] (a) Representative PROXI histological micrographs of livers from HCC patients with areas depicting tumour and non-tumour tissue, showing no PROXI expression in the tumours.

[0183] (b) PROXI overexpression following stable integration into Hep3B cell lines and doxycycline treatment for two days (PROXI OE) compared to control (Ctr) determined by qRT-PCR (n=6).

[0184] (c) Inducible shRNA-mediated PROXI knockdown upon doxycycline treatment for two days in Hep3B cell lines (shPROXl) compared to control (shCtr) determined by qRT-PCR (n=6).

[0185] (d) MA-plot of differential accessible regions (DAR) following two days of PROXI overexpression in Hep3B cells, based on ATAC-seq.

[0186] (e) PC A of DARs in (d) labelled by condition and replicate. Deutsches Krebsforschungszentrum 29 DK17279PC Stiftung des offentlichen Rechts et al.

[0187] (f) IGV tracks of PR0X1 binding by CUT&RUN and chromatin accessibility at the MYC locus based on ATAC-seq in Hep3B cells two days after PROXI overexpression displayed as accessibility change compared to control.

[0188] (g) GSEA normalised enrichment scores (NES) for MYC targets and Apoptosis following 7 days of PR0X1 overexpression in Hep3B cells, or two days after inducible-Proxl overexpression in HCC mouse model (day 16 harvest).

[0189] (h) Doxycycline dose-dependent Proxl overexpression in primary mouse tumour-derived cell lines transformed with Trp53 knockout together with overexpression of Myc (Myc / Trp53) (n=2) or Kras(G12D) (Kras / Trp53) (n=3) determined by Western blot following three days of doxycycline treatment.

[0190] (i) Quantification of PR0X1 protein levels of cells treated as in (h).

[0191] (j) Confluency of cells treated as in (h) normalised to uninduced controls.

[0192] Bar and boxplot graphs show mean values from indicated biological replicates, error bars = SD, two-sided t-test (b,c and i) or one-sided t-test assuming 0 as mean for (j), * p-adj < 0.05, ** p- adj < 0.01, *** p-adj < 0.001, **** p-adj < 0.0001.

[0193] Figure 8. PROXI prevents liver cancer induction and progression in mice.

[0194] (a) Strategy to induce HCC-like tumours by stable Myc overexpression and Trp53 knockout using HDTVI together with constitutive Proxl overexpression compared to GFP controls.

[0195] (b) Representative hematoxylin and eosin (HE), HNF4, and KRT19 histological micrographs of livers treated as in (a) and Fig. 2h at both 0.5x and 5x magnification, showing high HNF4 and low KRT19 expression.

[0196] (c) Number of tumours upon treatment as in (a) following overexpression (OE) of Proxl -IRES- GFP (n=3) vs GFP control (n=4).

[0197] (d) Size of tumours (cross-sectional area) in (c) with or without GFP expression indicating transgene expression.

[0198] (e) Schematic of doxycycline-inducible Late Proxl OE at day 14 following HDTVI-tumour induction as in (a).

[0199] (f) Representative mouse livers treated as in (e) following histological staining for HNF4 and CASP3 at day 16.

[0200] (g) Quantification of CASP3 protein levels in GFP+ tumours in (f) shown as percentage, following Late Proxl overexpression (OE) (n=3) vs GFP control (n=4). h, Representative GFP staining in livers treated as in (e) at the endpoint.

[0201] (i) Quantification of GFP+ tumour nodule numbers shown in (h).

[0202] (j) Western blot of PROXI protein levels following PGK- and Efla-promoter driven overexpression in vivo following HDTVI and in vitro following transfection in respective Myc / Trp53 models compared to controls. Deutsches Krebsforschungszentrum 30 DK17279PC Stiftung des offentlichen Rechts et al.

[0203] (k) Representative mouse livers following HDTVI to induce Myc overexpression and Trp53 knockout together with Proxl overexpression (OE) using PGK- or Efla-promoters (n=4) compared to PGK-GFP controls (n=5).

[0204] (l) Overall survival of mice treated as in (k). Log rank test.

[0205] (m) Strategy to induce liver tumours with complex identity following HDTVI-mediated KrasG12D overexpression and Trp53 knockout together with constitutive Proxl overexpression compared to controls.

[0206] (n) Representative histology sections of livers treated as in (m) and Fig. 2k stained with HE, HNF4, and KRT19 at both 0.5x and 5x magnification, showing high KRT19 and absence of nuclear HNF4 signal in this tumour model.

[0207] (o) Number of tumours upon treatment as in (m) following overexpression (OE) of Proxl- IRES-GFP (n=4) vs GFP control (n=4).

[0208] (p) Size of tumours (cross-sectional area) in (o) with or without GFP expression indicating transgene expression.

[0209] Bar graphs show mean values from indicated biological replicates, error bars = SD, unpaired t- test, * p-adj < 0.05, ** p-adj < 0.01, *** p-adj < 0.001, **** p-adj < 0.0001.

[0210] Figure 9. Impaired regeneration upon DDC-liver injury in Proxl deleted mice.

[0211] (a) Representative histological staining for PROXI and GFP of mouse livers two weeks following AAV GFP-cre-mediated Proxl deletion in conditional Proxl knockout mice Proxl?1compared GFP-A-cre control mice at day 14 just before injury induction.

[0212] (b) Hematoxylin and eosin (HE), HNF4, SOX9, and KRT19 stainings of liver sections treated as in (a) and following two weeks of DDC-diet and two weeks of recovery with normal diet.

[0213] (c) Percentage of HNF4+ and KRT19+ cells in liver sections from mice treated as in (b) (n=3).

[0214] (d) Serum levels of aspartate transaminase and alanine aminotransferase from mice treated as in (b) (n=3).

[0215] Bar graphs show mean values from three biological replicates, error bars = SD, unpaired two- sided t-test, * p-adj < 0.05.

[0216] Figure 10. Single cell transcriptome quality control and marker gene expression during hepatocyte reprogramming timecourse with or without Proxl.

[0217] (a) PCA of cells based on cell cycle gene expression before and after adjustment for cell cycle effects using Seurat, b-c, UMAP of cells labelled by expression of Proxl (b) and 4inl (c), inferred from PROXI and 4inl activity.

[0218] (d) LTMAP of all cells labelled by condition and biological replicate following demultiplexing using barcode oligos.

[0219] (e) Hepatocyte identity score in each cell on UMAP.

[0220] (f) Expression of selected individual hepatocyte marker genes in all cells.

[0221] (g) Fibroblast, adipocyte, myocyte, and neuron identity scores projected onto the UMAP. Deutsches Krebsforschungszentrum 31 DK17279PC Stiftung des offentlichen Rechts et al.

[0222] (h) Expression of selected fibroblast, adipocyte, myocyte, and neuronal marker genes in all cells.

[0223] (i) Quantification of fibroblast, adipocyte, myocyte, and neuron identity scores shown as boxplots with p-values (two-tailed t-test) for each time point and treatment.

[0224] Fig. 11. PROXI chromatin binding sites are enriched at gene promoters.

[0225] (a) PR0X1 chromatin binding was determined by CUT&RUN using FLAG antibodies upon overexpression of FLAG-tagged PR0X1 in MEFs and compared to GFP overexpression as control identifying 25,519 high-confidence PR0X1 binding sites. CUT&RLTN using IgG antibodies served as an antibody control and showed no signal at these peaks.

[0226] (b) Proportion of PR0X1 binding sites located in indicated genomic regions.

[0227] (c) Log2 ratio of observed PR0X1 binding sites to expected binding events in indicated genomic regions highlight enrichment at promoters.

[0228] (d) Negative correlation of fibroblast identity score with PR0X1 activity across all single cells during hepatocyte reprogramming time course, p-value < 2.2e-16.

[0229] Figure 12. PROXI inhibits reprogramming to myocyte and neuronal cell fates.

[0230] (a-c) Reprogramming of MEFs to induced hepatocytes, neurons, or myocytes using 4inl, Ascii, or Myodl overexpression, respectively. Protein quantification of E-cadherin (hepatocyte), TUBB3 (neuronal), or Desmin (myocyte) marker proteins at day 14 of respective reprogramming protocols with or without Proxl overexpression using Western blot analysis.

[0231] (d) Representative immunofluorescence microscopy of cells reprogrammed as in (a-c) with or without Proxl or Mytll overexpression in respective reprogramming protocols using indicated antibodies.

[0232] (e) Quantification of successfully reprogrammed cells in (d), based on displayed immunofluorescence markers and morphology.

[0233] (f) Percentage of reprogrammed cells in (d) positive for indicated markers and morphological criteria.

[0234] Bar graphs show mean values from seven (a) or three (b, c, e) biological replicates, error bars = SD, Dunnett’s test, * p-adj < 0.05. ** p < 0.01, *** p < 0.001.

[0235] Figure 13. Depletion of Proxl reduces hepatocyte reprogramming efficiency and fidelity.

[0236] (a) Expression of Proxl and indicated hepatocyte marker genes at day 14 of 4inl-induced hepatocyte reprogramming upon treatment with Proxl or control shRNAs based on qRT-PCR.

[0237] (b) Representative TJP1 immunofluorescence of cells in (a) at day 14.

[0238] (c) Quantification of normalised Albumin secretion in cells from (b).

[0239] (d) Expression of Proxl and indicated hepatocyte marker genes at day 14 day of hepatocyte reprogramming using Proxlfl / fl MEFs upon treatment with ere (Proxl- / -) or Acre (Proxlfl / fl) as isogenic control based on qRT-PCR. Deutsches Krebsforschungszentrum 32 DK17279PC Stiftung des offentlichen Rechts et al.

[0240] (e) E-cadherin (hepatocyte) and Desmin (muscle) protein quantification of cells in (d) based on Western blot analysis.

[0241] (f) Differential gene expression of cells in (d) based on RNA-seq (n=2) highlights downregulation of hepatocyte-specific genes, such as Krtl8 and Trf, and upregulation of alternate fatespecific genes, such as Myh9 (muscle) and Map2 (neuron), upon Proxl deletion.

[0242] Bar graphs show mean values from four (a) or five (c-e) biological replicates, error bars = SD, two-tailed t-test, * p-adj < 0.05, ** p-adj < 0.01, *** p-adj < 0.001, **** p-adj < 0.0001.

[0243] Figure 14. PROXI interacts with the repressive NuRD complex in liver but not in hippocampus.

[0244] (a) Immunoprecipitation (IP) of PROXI from primary mouse liver and hippocampus compared to IgG control followed by Western blot using indicated antibodies.

[0245] (b) Mass spectrometric identification and analysis of differential PROXI interaction partners between hippocampus (red, n=4) and liver (blue, n=4) (see Methods).

[0246] (c) Cartoon of the repressive NuRD complex highlighting liver-specific PROXI interaction partners.

[0247] Figure 15. PROXI target gene repression mimics PROXI function while target activation has the inverse effect.

[0248] (a) Reprogramming of MEFs to induced hepatocytes, neurons, or myocytes using MyoDl, Ascii, or 4inl overexpression, respectively. PROXI target gene activation was induced by fusion of PROXI DNA-binding domain (DBD) to the VP64 activator, while silencing was triggered by EnR repressor fusion. Full-length PROXI and DBD served as controls. Protein quantification of E-cadherin (hepatocyte, n=7), Tubb3 (neuronal, n=3), or MYH and Desmin (muscle, n=3) marker proteins at day 7 of respective reprogramming experiments using Western blot analysis.

[0249] (b) Representative immunofluorescence microscopy images of cells in (a) at day 14 of reprogramming using indicated antibodies.

[0250] (c) Quantification of successfully reprogrammed cells in (b) based on displayed immunofluorescence markers and morphology.

[0251] (d) Percentage of reprogrammed cells in (b) positive for indicated markers and morphological criteria.

[0252] (e) Expression of indicated hepatocyte marker genes at day 7 of 4inl -induced hepatocyte reprogramming upon overexpression of indicated PROXI fusion constructs based on qRT-PCR (n=6).

[0253] Bar graphs show mean values from specified number of biological replicates, error bars = SD, Dunnett’s test, * p-adj < 0.05, ** p-adj < 0.01, *** p-adj < 0.001, **** p-adj < 0.0001.

[0254] Figure 16. PROXI predominantly closes bound chromatin and silences associated genes. Deutsches Krebsforschungszentrum 33 DK17279PC Stiftung des offentlichen Rechts et al.

[0255] (a) MA-plot of differentially accessible regions (DAR) at day two of 4inl -induced hepatocyte reprogramming with or without Proxl overexpression based on ATAC-seq (n=3).

[0256] (b) Percentage of closed and opened regions upon Proxl overexpression in (a).

[0257] (c) Correlation of DARs between indicated conditions and replicates in (a) and upon overexpression of GFP or Proxl in MEFs.

[0258] (d) PCA of DARs in (c) labelled by condition and replicate.

[0259] (e) Top, mean Tn5 transposon adapter insertions at indicated conditions from (a) centred at PR0X1 binding sites from CUT&RUN experiments. Bottom, changes in chromatin accessibility at PR0X1 binding sites between 4inl + GFP and 4inl + Proxl indicate decreased accessibility upon Proxl overexpression.

[0260] (f) Differential gene expression of cells in (a) based on RNA-seq (n=2), for genes with differ- entially-accessible regions (p-adj < 0.05) within 2 kb of their TSS and an overlapping CUT&RUN peak with a PROXI binding motif, showing global downregulation of target gene transcription upon Proxl overexpression.

[0261] Figure 17. Induction of hepatocyte gene cluster 1 and repression of alternate fate cluster 2 by PROXI in bulk and single cell gene expression datasets over time.

[0262] (a) Bulk gene expression changes during 4inl -induced hepatocyte reprogramming at indicated time points clustered by differential expression and chromatin accessibility. Median gene expression of differential expressed genes between Proxl and GFP control are presented as boxplots for both clusters, two-tailed t-test (p-values shown).

[0263] (b) Aggregate expression of genes from bulk gene expression-derived cluster 1 and 2 projected onto UMAP of 4inl -induced reprogramming single-cell transcriptomics data from Fig. 2b, highlight overlap of cluster 1 with hepatocyte cluster and cluster 2 with alternate fate cluster.

[0264] Figure 18. Prrxl and Pparg are PROXI target genes and their repression in part explains PROXI activity during reprogramming.

[0265] (a) Representative IGV tracks of PROXI chromatin binding based on CUT&RUN (n=3) and chromatin accessibility based on ATAC-seq (n=3) displayed as accessibility change at day two day of 4inl -induced hepatocyte reprogramming with or without Proxl at the promoters of Prrxl (top), Pparg (middle), Ebf2 (bottom). PROXI motifs and CUT&RUN peaks are displayed.

[0266] (b) Representative TJP1 immunofluorescence staining at day 14 of hepatocyte reprogramming with indicated combinations of transcription factor overexpression and / or shRNA-mediated knockdown treatments.

[0267] (c) Quantification of Albumin protein levels by Western blot at day 14 of hepatocyte reprogramming with overexpression of Prrxl or Pparg and co-overexpression of GFP control (left) o Proxl (right). Deutsches Krebsforschungszentrum 34 DK17279PC Stiftung des offentlichen Rechts et al.

[0268] (d) Expression of Pparg and Prrxl upon shRNA-knockdown at day 14 of hepatocyte reprogramming determined by qRT-PCR.

[0269] (e) Albumin protein levels determined by quantitative Western blot at day 14 of hepatocyte reprogramming upon knockdown of Prrxl or Pparg with or without Proxl overexpression.

[0270] Bar graphs show mean values normalised to total protein levels in (c and e), or GAPDH expression in (d), from four biological replicates, error bars = SD, Dunnett’s test (c and e) or two- tailed t-test (d), * p-adj < 0.05, ** p-adj < 0.01, *** p-adj < 0.001, **** p-adj < 0.0001.

[0271] Figure 19. PROXI loss can enhance HCC liver tumour formation in mice.

[0272] (a) Overall survival of 135 HCC patients with high PROXI expression levels (40% cutoff for high-expression cohort) compared to CCA patients (Menyhart, Nagy, and Gyorffy 2018).

[0273] (b) Representative photographs of mouse livers at endpoint of HCC tumour modelling induced by Myc overexpression and Trp53 knockout using HDTVI together with PX330-sgRNA-Proxl knockout compared to respective negative control.

[0274] (c) Number of tumours upon treatment as in (a) following Proxl knockout (KO) vs control (Ctr) (n=3).

[0275] (d) Survival curve of HCC mice treated as in (a) comparing Proxl KO (n=4) with control (n=3). Log rank test.

[0276] (e) Representative mouse livers at endpoint of HCC tumour modelling as in (a) with EFla- GFP-shProxl knockdown compared to respective negative control. Histology sections of corresponding livers are shown in Fig. 5c

[0277] (f) Total number of tumour nodules with diameter > 0.2 mm at day 14 of tumour induction as in (d) upon shRNA-mediated control or Proxl knockdown (n=5).

[0278] (g) Tumour size (cross-sectional area) following treatments as in (d) displaying tumours with or without GFP expression (n=5).

[0279] (h) Number of tumours upon treatment as in (a) following Proxl knockdown (shProxl) vs control (shCtr) (n=5).

[0280] (i) Survival curve of HCC mice treated as in (d) comparing knockdown of Proxl with control (n=5). Log rank test.

[0281] Bar graphs show mean values from specified biological replicates, error bars = SD, two-way ANOVA and Sidak correction (f) or two-sided t-test (g), * p-adj < 0.05, ** p-adj < 0.01, *** p- adj < 0.001, **** p-adj < 0.0001.

[0282] Figure 20. PROXI can shift liver tumour fate from CCA to HCC in mice.

[0283] (a) Strategy to induce liver tumours with CCA identity following HDTVI-mediated Akt and Notch 1 receptor intracellular domain (NICD) (Akt / Notch) overexpression with constitutive Proxl overexpression compared to controls.

[0284] (b) Representative mouse livers at endpoint of CCA tumour modelling induced as in (a). Histology sections of corresponding livers are shown in Fig. 5c. Deutsches Krebsforschungszentrum 35 DK17279PC Stiftung des offentlichen Rechts et al.

[0285] (c) Quantification of GFP + tumours following constitutive overexpression (OE) of Proxl - IRES-GFP (n=4) vs GFP control as in (a), shown as percentage (n=5).

[0286] (d) Size of tumours (cross-sectional area) in (c) with or without GFP expression indicating transgene expression.

[0287] (e) Number of tumours upon treatment as in (c).

[0288] (f) Survival curve of CCA mice treated as in (c). Log rank test.

[0289] (g) Schematic of doxycycline-inducible Late Proxl OE at day 20 following HDTVLCCA tumour induction as in (a).

[0290] (h) Representative mouse livers 1-2 weeks following Late Proxl OE induced in CCA model as in (g).

[0291] (i) Histology sections of livers treated as in (g) stained with HE, KRT19, HNF4, and SOX9.

[0292] (j) Quantification of KRT19 and SOX9 (CCA-marker) and HNF4 (HCC-marker) positive cells in GFP+ tumours following Late Proxl overexpression (OE) (n=4) vs GFP control (n=4) from (g) shown as percentage.

[0293] Bar graphs show mean values from specified biological replicates, error bars = SD, two-sided t-test (b) or two-way ANOVA and Sidak correction (c), * p-adj < 0.05, ** p-adj < 0.01, *** p- adj < 0.001, **** p-adj < 0.0001.

[0294] EXAMPLES

[0295] Materials and Methods

[0296] The Examples merely illustrate the invention. They shall by no means be interpreted as limiting the scope of the invention.

[0297] No statistical methods were used to predetermine the sample size for the experiments. Animals for primary cultures and in vivo experiments were selected randomly before indicated treatments. The investigators were blinded to the microscopy analysis and quantification. Otherwise, no blinding and randomisation were performed.

[0298] Human material

[0299] Formalin-fixed, paraffin-embedded human liver tissue samples were retrieved from the Medical Faculty Mannheim, Heidelberg University, for immunohistological analyses. The study was approved by the local ethics committee (permit number 2012-293N-MA).

[0300] Primary mouse cell lines

[0301] Mouse embryonic fibroblasts (MEFs) were harvested from El 3.5 embryos of C57BL / 6N (wildtype) or C57BL / 6J (Proxlfll) mice (Martinez-Corral et al. 2015) as described before (Adrian- Deutsches Krebsforschungszentrum 36 DK17279PC Stiftung des offentlichen Rechts et al.

[0302] Segarra, Weigel, and Mall 2021; Mall et al. 2017). The distal portions of all limbs from 3-4 embryos were dissected, placed in 100 pL trypsin, cut thoroughly, and incubated in a total of 1 mL trypsin (37°C, 15 min). Trypsin was inactivated by the addition of cell suspension to 25 mL MEF media (DMEM; Invitrogen) containing 10% cosmic calf serum (CCS; Hy clone), betamercaptoethanol (Sigma), non-essential amino acids, sodium pyruvate, L-glutamine, and peni- cillin / streptomycin (all from Invitrogen). MEFs were then cultured in MEF media and either cryopreserved or passaged twice using trypsin before reprogramming experiments.

[0303] Hepatocellular carcinoma cell lines

[0304] Mouse primary liver cancer cell lines were derived from C57BL / 6N female mice following HDTVI with either Myc OE / Trp53 KO or A7z / .s(G I 2D) OE / Trp53 KO, and human Hep3B cells were transduced with lentivirus prepared from indicated plasmids (data not shown) in DMEM (Invitrogen) containing 10% foetal bovine serum (Sigma), non-essential amino acids, sodium pyruvate, L-glutamine, and penicillin / streptomycin (all from Invitrogen). Puromycin selection was performed (2 pg / ml) for 2 days to generate stable cell lines. Cellular proliferation rates were determined beginning 1 day after seeding with media containing 2 pg / ml or the indicated amount of doxycycline (Sigma) or without doxycycline (for controls) for a total of 7-10 days (depending on the proliferation rate of the line) using the IncuCyte S3 live-cell imaging system (Essen BioScience, Hertfordshire, UK). 4 brightfield images per well were acquired every 4 h at a magnification of 10x and analysed using the IncuCyte S3 2019B software. Experiments were performed in 2 to 3 biological replicates, with 5 to 6 technical replicates each.

[0305] Animal experiments

[0306] For hydrodynamic tail vein injection (HDTVI), 2 mL of sterile 0.9% NaCl solution, corresponding to 10% of body weight, containing the plasmids of interest was injected into the tail vein of 8-week-old female C57BL / 6N mice within 5 to 7 s (Largaespada and Collier 2008; Moon et al. 2019; Revia et al. 2022; Tschaharganeh et al. 2014; W. Xue et al. 2014). Depending on the vector used, this technique allowed for liver-specific gene knockouts and / or overexpression by in vivo transfection of hepatocytes. Each mouse was injected with 20 pg of pX330-based plasmid for sgRNA-mediated gene knockout of Tp53 or Proxl, and 10 pg of pT3-EF la-based or pT3-PGK-based plasmid for transposon-mediated stable overexpression of Myc, Kras, Akt or NICD. For Proxl knockdown, 2 pg of CMV-Sleeping Beauty transposase and 20 pg of miR- E-based plasmid with / Vo / -targeting shRNA and a GFP reporter were co-injected. For constitutive Proxl overexpression, 4 pg of CMV-Sleeping Beauty transposase and 10 pg of pT3- PGK-based or pT3-EF la-based plasmid with 7 ox7-cDNA and a GFP reporter was co-injected (data not shown). For late Proxl overexpression, 4 pg of CMV-Sleeping Beauty transposase and 10 pg of pT3-Tre-based plasmid with 7 ox7-cDNA and a GFP reporter were co-injected. Each experimental group involving HDTVI contained at least 5 mice, with all mice monitored daily. For mice that received Tre-driven transgenes, a doxycycline-containing diet (6.25% Deutsches Krebsforschungszentrum 37 DK17279PC Stiftung des offentlichen Rechts et al. doxycycline hyclate, Envigo Teklad, Indianapolis) was given beginning at day 14 or 20 post- HDTVI, depending on the model. Upon euthanasia of mice (at indicated time points or at humane endpoint), relevant organs were harvested and photographed. Survival data were analysed based on the time between HDTVI and euthanasia at a humane endpoint. After euthanasia, tumour samples were taken for RNA and protein analysis. The remaining tissue was incubated in 4% paraformaldehyde for a minimum of 24 h for subsequent histological analysis. For Di- ethyl-l,4-dihydro-2,4,6-trimethyl-3,5-pyridindicarboxylat (DDC, 137030 Sigma-Aldrich)-in- duced liver injury, 4-month-old Proxl^ mice (Martinez-Corral et al. 2015) were injected into the tail vein with 150 pl of PBS with 5 x 1011genomic particles of Adeno-associated virus 8 (AAV8) carrying ere or Acre-recombinase (data not shown). After 14 days, liver injury was induced by providing 0.1% DDC mixed with a standard diet (3437, KLIBA NAFAG) for 2 weeks followed by normal diet for another 2 weeks. After euthanasia, blood and tissue were harvested for subsequent analysis. All animal experiments were performed in compliance with ethical regulations and approved by the regional ethics board in Karlsruhe, Germany.

[0307] Blood biochemical analysis

[0308] Blood was collected and serum was freshly isolated by centrifugation (12,000 g, 10 min, 4°C). Serum was stored at -20°C until analysis. Aspartate transaminase (AST), alkaline phosphatase (ALP), and alanine aminotransferase (ALT) levels were detected as per manufacturer instructions (Fujifilm DRI-CHEM SLIDE).

[0309] Histology

[0310] Paraformaldehyde-incubated livers were embedded in paraffin and 2 pm slices were processed as previously described for immunohistochemistry (H4C) staining (Gallage et al. 2022). BONDMAX (Leica Biosystems) was used for automated staining. BondTM citrate solution (AR9961, Leica), BondTM EDTA solution (AR9640, Leica), or BondTM proteolytic enzyme kit (AR9551, Leica) were used for antigen retrieval (data not shown). Sections were incubated in antibodies diluted in BondTM primary antibody diluent (AR9352, Leica Biosystems) followed by secondary antibody (Leica Biosystems) incubation and staining with Bond Polymer Refine Detection Kit (DS9800, Leica Biosystems). Slides were scanned with an Aperio AT2 slide scanner (Leica Biosystems) at 20x, then annotated and analysed with Aperio ImageScope (vl2.4.0.5043, Leica) for determining the size of tumour nodules. Marker staining quantifications were analyzed in QuPath (vO.4.3) (Bankhead et al. 2017) using the positive cell detection option with the same settings for each marker across all sections. Certified pathologists (H.W. and D.T.) performed the histopathological analysis of paraffin-embedded liver tumour sections.

[0311] PROXI immunoprecipitation and mass spectrometry

[0312] For each immunoprecipitation, one hippocampus or liver of 2 to 3 -month-old mice was used per biological replicate. The fresh tissue was lysed in 1 mL lysis buffer containing (in mM): Deutsches Krebsforschungszentrum 38 DK17279PC Stiftung des offentlichen Rechts et al.

[0313] 0.5% Tween-20, 50 Tris pH 7.5, 2 EDTA, 1 DTT, 1 PMSF, 5 NaF (all from Sigma), and complete protease inhibitor (Roche) for 15 min at 4 °C and processed for immunoprecipitation as described previously (Mall et al. 2017; Weigel et al. 2023) using 2 pg PROXI or control IgG (Sigma) antibody per reaction (data not shown). Bound proteins were enzymatically digested with trypsin using an AssayMAP Bravo liquid handling system (Agilent technologies) running the autoSP3 protocol as described here (Muller et al. 2020). A LC-MS / MS analysis was carried out using a Vanquish Neo UPLC system (Thermo Fisher Scientific) directly connected to an Orbitrap Exploris 480 mass spectrometer for a total of 60 min per sample. Peptides were online desalted on a trapping cartridge (Acclaim PepMap300 Cl 8, 5pm, 300 A wide pore; Thermo Fisher Scientific) with a loading volume of 60 ul using 30 ul / min flow of 0.05% TFA in water. The analytical multistep gradient (300 nl / min) was performed with a nanoEase MZ Peptide analytical column (300 , 1.7 pm, 75 pm x 200 mm, Waters) using solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in acetonitrile). For 45 min the concentration of B was linearly ramped from 5% to 30%, followed by a quick ramp to 80%, after four min the concentration of B was lowered to 2% and a 3 column volumes equilibration appended. Eluting peptides were analyzed in the mass spectrometer using data-dependent acquisition (DDA) mode. A full scan at 60k resolution (380-1400 m / z, 500% AGC target, 100 ms maxIT) was followed by up to 1.5 s of MS / MS scans. Peptide features were isolated with a window of 1.2 m / z, fragmented using 26% NCE. Fragment spectra were recorded at 15k resolution (100% AGC target, 150 ms maxIT). Dynamic exclusion was set to 10 s. Data analysis was carried out by MaxQuant (version 2.1.4.0) (Tyanova, Temu, and Cox 2016) using an organism-specific database extracted from Uniprot.org (mouse reference database with 1 protein sequence per gene, containing 21,957 unique entries from May 3rd, 2023). Settings were set to default with the following adaptions. Separate parameter groups were assigned for liver and hippocampus samples. Separate Label free quantification (LFQ) per parameter group was enabled. Besides the LFQ approach based on the MaxLFQ algorithm (Cox et al. 2014), quantification was also done based on iBAQ-values (Schwanhausser et al. 2011). The statistical analysis of proteins has been conducted as follows: Adapted from the Perseus recommendations (Tyanova and Cox 2018), protein groups with valid values in 70% of the samples of at least 1 condition were used for statistics. In addition, missing values, being completely absent in 1 condition, were imputed with random values drawn from a downshifted (2.2 standard deviation) and narrowed (0.3 standard deviation) intensity distribution of the individual samples. For missing values with no complete absence in one condition, the R package missForest (Stekhoven and Buhlmann 2012) was used for imputation. No additional normalization was applied to the iBAQ values that were used in the statistical analysis. The statistical analysis was performed with the R-package limma (Ritchie et al. 2015) with an adapted contrast setup from chapter 9.5 Interaction Models. Within the eBayes function the options robust and trend were set to TRUE. The p-values were adjusted with the Benjamini-Hochberg method for the multiple testing. PROXI interactors were considered significant with an absolute LogFC > 1, p-value < 0.05, and quality score > 0.5. In Deutsches Krebsforschungszentrum 39 DK17279PC Stiftung des offentlichen Rechts et al. addition, interactors were filtered according to nuclear location (based on https: / / www.protein- atlas . org / ab out / download) .

[0314] Recombinant virus production

[0315] Lentivirus was produced through transfection of lentiviral backbones containing indicated transgenes along with third-generation packaging plasmids into HEK293T cells according to the Trono laboratory protocol (data not shown) (Dull et al. 1998). Lentivirus was concentrated from HEK293T culture supernatant through ultra-centrifugation (23,000 rpm, 2 h, 4°C) and stored at -80°C or used immediately. AAV8 vectors were produced by polyethyleneimine (PEI) triple-transfection of HEK293T cells using indicated plasmids as described before (data not shown) (Becker et al. 2022). AAV vectors were purified using iodixanol gradient density centrifugation followed by buffer exchange to PBS. AAV vector quantification was conducted by droplet digital (dd)PCR using the BioRad ddPCR system. Each 20 pL PCR contained 5 pL diluted virus template, 10 pL of the ddPCR Supermix for Probes (no dUTP; BioRad), 4 pL nuclease-free H2O and 1 pL ITR-primer / probe mix (final concentration: 900 nM for primers, 250 nM for probes; ITR f: GGAACCCCTAGTGATGGAGTT (SEQ ID NO:1), ITR r: CGGCCTCAGTGAGCGA (SEQ ID NO:2), ITR_probe: HEX-CAC- TCCCTCTCTGCGCGCTCG-BHQ1 (SEQ ID NO:3). The measured copy number of vector templates per reaction was corrected by the input volume and dilution factor to calculate vector genomes per pL vector stock.

[0316] Direct reprogramming from MEFs

[0317] Wild-type or Proxlfl / fl MEFs were transduced by incubation with lentivirus prepared from indicated plasmids (data not shown) in MEF medium with 8 pg / ml polybrene (Sigma) for 16 to 20 h. Medium was exchanged to MEF medium containing 2 pg / ml doxycycline (Sigma) to induce transgene expression. For myocyte and neuronal reprogramming (Adri an- Segarra, Weigel, and Mall 2021b), MEFs were transduced with lentivirus containing Myodl or Ascii, respectively, along with rtTA-containing lentivirus. After 48 h, all medium was exchanged with N3 medium (DMEM / F12) containing N2 supplement, B27, 20 pg / ml insulin, penicillin / strep- tomycin (all from Invitrogen), and doxycycline to continue transgene expression. Medium was changed every 2 days for the remainder of the reprogramming. For hepatocyte reprogramming (Song et al. 2016), MEFs were seeded onto collagen-coated plates and transduced 1 day later with 4inl -containing lentivirus. 1 day following doxycycline induction the medium was supplemented with 0.5* volume Hepatocyte Culture Medium (HCM, Lonza) containing 5% FBS (Life Technologies) and 2 pg / ml doxycycline. On day 2, all medium was exchanged to a mixture of % MEF medium and % HCM + 5% FBS with 2 pg / ml doxycycline. On day 3, all medium was exchanged to HCM + 5% FBS with 2 pg / ml doxycycline. Medium was then changed every 2 days for the remainder of reprogramming. For domain fusion experiments published protocols Deutsches Krebsforschungszentrum 40 DK17279PC Stiftung des offentlichen Rechts et al. were followed (Adrian- Segarra, Weigel, and Mall 2021b, 2021a), and for shRNA-based knockdown cells were treated with lentivirus targeting the indicated gene with 2 independent-shRNA constructs (data not shown).

[0318] Immunofluorescence quantification

[0319] To calculate the efficiency of neuronal induction, the total number of TUBB3 -expressing cells with complex neurite outgrowth (cells with a round cell body and at least one thin process with a length at least double the diameter of the cell body) was counted manually (Mall et al. 2017). Any TUBB3 -positive and Desmin-negative cells that did not meet the morphological criterion were considered TUBB3 -positive, non-neuronal cells (Fig. I lf). To calculate the efficiency of myocyte cell induction, the total number of Desmin-expressing cells was counted manually. Dual TUBB3- and Desmin-positive cells were considered as a separate category of mixed identity cells. To calculate the efficiency of hepatocyte reprogramming, the total number of cells for which TJP1 staining formed a complete border around the nucleus (stained by DAPI) was counted manually. All quantifications were performed 14 days after transgene induction by immunofluorescence microscopy. Fluorescence micrographs were captured automatically using a Nikon Ti2 microscope with a Ti-HCS system, the Nikon S Plan Fluor ELWD 20* NA 0.45 objective, Nikon DS-Qi2 CMOS camera (2404x2404), and Lumencor Sola SE II light source. Quantifications were based on the mean number of positive cells across 10 randomly selected 20 x magnification fields of view per biological replicate, with at least 3 biological replicates. The number of reprogrammed cells in each treatment condition was then normalised to the number of reprogrammed cells in the control condition. To calculate the fraction of total reprogrammed cells positive for indicated markers and morphological criteria, the number of cells in each of the 5 categories was divided by the total number of cells present in any of the 5 categories, pooled across all replicates.

[0320] Computational safeguard repressor screen

[0321] Using single-cell gene expression and cell type annotations from Tabula Muris (The Tabula Muris Consortium et al. 2018b), the expression of 1,296 transcription factors (Lambert et al. 2018) in 18 selected cell types using median normalised CPM units was analysed. In addition, the top 1,000 cell type-enriched genes were retrieved using Seurat FindMarkers (data not shown). In the promoters of each of these signature genes (±2 kb around the TSS), the number of DNA-binding motifs for each transcription factor was determined (derived from CIS-BP 1.94d (Lambert et al. 2018; Weirauch et al. 2014) mapped to mmlO). Pairwise comparisons were performed between cell-type-specific gene signatures to remove overlapping genes and calculate mean transcription factor motif density. For each cell type, transcription factor expression specificity and transcription factor binding motif enrichment in signature gene promoters were calculated by Z-score scaling to obtain expression and Zmotif respectively. The higher / expression, the more specifically the transcription factor is expressed in the corresponding cell Deutsches Krebsforschungszentrum 41 DK17279PC Stiftung des offentlichen Rechts et al. type. Transcription factors with positive Zmotif, and high expression might function as activators of cell type-specific genes. Conversely, a negative Zmotif, indicates depletion of DNA binding motifs of this transcription factor at genes specific to the analysed cell type, indicating that it could act as a safeguard repressor by silencing cell type-unspecific genes. The PR0X1 CUT&RUN peaks were harnessed, filtered using Hocomoco vl2 motifs (Kulakovskiy et al. 2018), to assess the gene signature binding enrichment using a one-tailed hypergeometric test. A searchable database of this bioinformatics analysis was deployed here: apps.embl.de / safe- guard. In addition, a safeguard repressor score was calculated for each transcription factor defined as the sum of / expression and -1 * Zmotif, following Zmax-normalisation to ensure equal weighting of expression and motif bias (Fig. lb; Fig. 6c). Transcription factors were further annotated as cell type-specific and life-long expressed using data from Tabula Muris Senis (The Tabula Muris Consortium et al. 2020) based on the following two criteria: (i) the mean expression in Tabula Muris Senis must be at least 50% of that in Tabula Muris; (ii) the mean expression must be higher in a specific cell type compared to the mean expression across all 18 cell types in Tabula Muris. In addition, TFs were categorised as life-long expressed using a mouse developmental gene expression atlas (Cardoso-Moreira et al. 2019) for the brain, heart, and liver if they exhibited high expression (in the highest quantile) in a continuous manner (in more than 70% of developmental time points and replicates). Safeguard candidates for microglia, neurons, oligodendrocytes, and astrocytes from the brain data; cardiomyocytes from the heart data; and hepatocytes from the liver data were analysed. 77% of the safeguard repressor candidates in these cell types (17 / 22), including Proxl and Mytll, were found to be life-long expressed in an organ-specific manner, which compared to all TFs expressed in these tissues reached statistically significant enrichment using a Fisher's exact test. Top safeguard repressor candidates were also categorised as activators, repressors, or dual activator / repressors as well as having a known tumour suppressor role or not, in the cell type of interest, based on literature review (data not shown).

[0322] Bulk RNA-seq library generation

[0323] Primary mouse liver tumour samples were dissected and placed into TRIzol (Invitrogen). Samples were crushed and then homogenised using QIAshredder (Qiagen). Reprogrammed hepatocytes or Hep3B cells were harvested from culture plates by the addition of TRIzol to the cultures at indicated time points. RNA harvested in TRIzol was isolated using the RNA Miniprep kit (Zymo Research). For bulk RNA sequencing, libraries were prepared according to the dUTP protocol (Levin et al. 2010) and paired-end sequencing (2xl00bp) was performed on the No- vaSeq 6000 or NovaSeq X Plus platforms (Illumina).

[0324] RNA-seq data processing

[0325] Raw reads were mapped to the reference genome mmlO or hg38 using STAR (Dobin et al. 2013). Differential gene expression was determined using DESeq2 (R package version 1.28.1) Deutsches Krebsforschungszentrum 42 DK17279PC Stiftung des offentlichen Rechts et al.

[0326] (Love, Huber, and Anders 2014) with size factor normalisation and Wald significance tests. For bulk MEF reprogramming data, the primary MEF line as a covariate was used. Com- plexHeatmap (version 2.12.1) (Gu, Eils, and Schlesner 2016) was used to generate heatmaps. In Fig. 4i, genes were included in the analysis if they had an abs (log2FC) > 0.75 at any of the time points.

[0327] CUT&RUN library preparation

[0328] Cells were harvested with Accutase and strained through a 70 pm strainer followed by CUT&RUN processing (Skene and Henikoff 2017). 250,000 cells were washed twice with 1 mL wash buffer (20 mM HEPES-KOH pH 7.5, 150 mM NaCl, 0.5 mM spermidine, and IX Roche Complete Protease Inhibitor), then resuspended in 200 pL of ice-cold cell lysis buffer (10 mM Tris-HCl pH 7.5, 10 mM NaCl, 3 mM MgCh, 0.1% Tween-20, 0.1% NP-40, and 1% BSA in ddH2O) for 3-5 min. 1 mL of ice-cold wash buffer was then added to stop cell lysis, and nuclei were centrifuged (500g, 10 min, 4°C). Nuclei were resuspended in 200 pL wash buffer, and 100,000-150,000 nuclei were taken to another tube. Concanavalin-A beads (Polysciences) were pre-activated in cold binding buffer (20 mM HEPES-KOH pH 7.5, 10 mM KC1, 1 mM CaC12, and 1 mM MnC12). Nuclei were centrifuged and buffer was removed. Activated beads were then added to the pellet. The bead-cell suspension was rotated (RT, 10 min). The supernatant was removed on a magnet and the beads were resuspended in antibody buffer (0.2 mM EDTA, 0.05% w / v digitonin in wash buffer). 1 pg primary antibody (rabbit anti- FLAG, in vitro, or rabbit anti-Proxl, in vivo) or control (mouse IgG, Sigma) was added (data not shown), and cells were incubated on a nutator (overnight, 4°C). Beads were washed twice in digitonin-wash buffer (0.05% w / v digitonin in wash buffer), resuspended in 700 ng / mL pAG- MNase (Protein Expression and Purification Core Facility, EMBL, Heidelberg) in digitonin- wash buffer, and rotated (4°C, 1 h). pAG-MNase-loaded beads were then washed twice in digitonin-wash buffer, resuspended in di gitonin- wash buffer, and placed on ice. 1 uL of 100 mM CaC12 was added to induce chromatin digestion, and the mixture was incubated on ice (30 min). 50 uL of 2x stop buffer (340 mM NaCl, 20 mM EDTA, 4 mM EGTA, 0.05% w / v digitonin, 50 ug / mL RNase A, 50 ug / mL glycogen, 0.5 ng / mL spike-in E. coli DNA) was added, and the suspension incubated at 37°C for 10 min to release chromatin fragments from cells. The supernatant was subjected to phenol-chloroform extraction, and purified DNA fragments were used for library preparation with NEBNext DNA Library Prep Kit for Illumina (NEB E7645). Libraries were then sequenced (paired-end, 2x40 bp) on the NextSeq 550 and 2000 platform (Illumina). Livers from 2 to 3-month-old C57BL / 6J mice were harvested and directly processed for nuclei isolation using liver swelling buffer (10 mM Tris pH 7.5, 2 mM MgC12, 3 mM CaC12) with the help of a douncer. Homogenized tissue was passed through a 70 pm strainer and centrifuged (400g, 5 min, 4°C). Tissue pellets were resuspended in liver lysis buffer (10 mM Tris pH 7.5, 1% NP-40, 2 mM MgC12, 10% Glycerol, 3 mM CaC12) and centrifuged again. Pellets containing the nuclei were washed twice in PBS and then the same protocol as for cells was Deutsches Krebsforschungszentrum 43 DK17279PC Stiftung des offentlichen Rechts et al. followed. Hippocampi from 2 to 3-month-old C57BL / 6J mice were harvested and directly processed for nuclei isolation. Tissue was lysed in brain lysis buffer (0.5% Tween-20, 50mM Tris pH 7.5, 2mM EDTA, lx DTT, lx NaF, lx Protease inhibitor cocktail) by mechanical homogenization in a 1.5 mL Eppendorf tube. After 15 min of incubation on ice, nuclei were pelleted by centrifugation (1 min, 3200 rpm, 4°C). Then, the samples were processed as described for cells. For mouse livers and hippocampi, 500,000 nuclei were used per sample instead of 100,000- 150,000 for cells. In addition, 5 pg of target antibody was used per sample.

[0329] CUT&RUN data processing

[0330] CUT&RUN data were analysed using the nf-core / cutandrun pipeline vl.0.0 with Nextflow version 21.05.0 (Cheshire et al. 2022; P. A. Ewels et al. 2020). Reads were aligned to mmlO or hg38. Software versions: bedtools (v2.30.0) (Quinlan and Hall 2010), bowtie2 (2.4.2) (Langmead and Salzberg 2012), deeptools (v3.5.0) (Ramirez et al. 2016), DESeq2 (vl.28.0) (Love, Huber, and Anders 2014), fastqc (vO.11.9), multiqc (vl. l l) (P. Ewels et al. 2016), picard (v2.23.9) (Broad Institute 2019), python (v3.8.3), samtools (vl.10) (Danecek et al. 2021), Genrich (vO.6.1) (https: / / github.com / jsh58 / Genrich), TrimGalore (vO.6.6) (Krueger et al. 2021), ucsc (v377). Consensus peaks were defined by running Genrich -m 30 -e chrM -r -1 5 -q 0.3. To determine peaks containing motifs, the PROXI binding motif (CIS-BP ID: M03445 2.00) was mapped onto mm 10 or hg38, extended each resulting motif peak to a total width of 50 bp, and ran bedtools intersect -wa on the CUT&RUN consensus peaks and motif peaks, respectively. Genes defined as direct target genes based upon CUT&RUN were determined by running Homer annotatePeaks.pl on the final peak set and filtering for genes with a peak within ±lkb of their TSS.

[0331] ATAC-seq library preparation

[0332] Nuclei from liver and hippocampus tissues were isolated as described for CUT&RUN. Cells on culture plates were washed twice with PBS and detached by Accutase digestion (4-6 min, RT) followed by AT AC processing. Cell suspensions were placed into an equal volume of MEF medium followed by centrifugation (500g, 5 min, 4°C) before resuspension in ice-cold PBS + 1% BSA. Cells were then strained through a 70 pm filter, and centrifuged (500g, 5 min, 4°C). 50,000 cells or nuclei were resuspended in 50 pL of ice-cold cell lysis buffer containing 0.1% NP-40 and 0.01% digitonin in wash buffer (10 mM Tris-HCl pH 7.5, 10 mM NaCl, 3 mM MgCE, 0.1% Tween-20, and 1% BSA in ddH2O) for 3-5 min. 1 mL of ice-cold wash buffer was then added to stop cell lysis, and nuclei were centrifuged (500g, 10 min, 4°C). 4.5 pL of each of the tagmentation oligos, Tn5_ME and Tn5-R1N (data not shown), were annealed in oligo annealing buffer (10 mM Tris-HCl pH 7.5, 50 mM NaCl, and 10 mM EDTA final concentration in ddH2O) by heating at 95°C (3 min) followed by a ramp down by 1°C to 25°C. The same procedure was performed for Tn5_ME and Tn5-R2N (data not shown). Tn5 was assembled with annealed oligos by combining 50 pL Tn5 (1 mg / mL stock) with 25 pL annealed Deutsches Krebsforschungszentrum 44 DK17279PC Stiftung des offentlichen Rechts et al.

[0333] Tn5_ME and Tn5-R1N and 25 pL Tn5_ME and Tn5-R2N. Nuclei were resuspended in 40 pL tagmentation buffer (38.8 mM Tris-acetate, 77.6 mM K-acetate, 11.8 Mg-acetate, 18.8% dimethylformamide, and 0.12% NP-40 in ddH2O), to which 5 pL ice-cold PBS + 1% BSA and 5 pL pre-assembled Tn5 was added. Samples were incubated on a Thermomixer (37°C, 30 min, 500 rpm) before being subjected to MinElute (Qiagen) cleanup to extract tagmented DNA. Eluted DNA was pre-amplified with P5 and P7 primers (data not shown) using the NEBNext HF 2x PCR Master Mix (New England Biolabs) in a thermocycler set to 72°C (5 min), 98°C (30 sec), and 5 cycles of 98°C (10 sec), 63°C (30 sec), and 72°C (1 min). A qPCR side reaction was performed with the resulting pre-amplified libraries to determine the necessary additional cycles (5 cycles fewer than the number of cycles corresponding to % of max fluorescence) for complete amplification. After finishing amplification, 50 pL of each library was subjected to 2- sided size selection by addition of 27.5 pL (0.55x) AMPure XP beads, incubation (5 min), transfer of supernatant to new tubes, addition of 42.5 pL (1.4x) AMPure XP beads to the supernatant, incubation (5 min), three washes with 80% ethanol, and elution of DNA. Resulting libraries were then sequenced on the NextSeq 2000 platform (Illumina).

[0334] ATAC-seq data processing

[0335] ATAC-seq data from in vitro samples were analysed with a custom Snakemake pipeline. Raw reads were quality-checked with fastqc (v0.11.8), trimmed with trimmomatic (v0.38), and aligned to UCSC mmlO or hg38 with bowtie2 (v2.3.4.3) () (Langmead and Salzberg 2012)). Aligned reads were cleaned and base-recalibrated (to take account of Tn5 insertion biases) with samtools (vl.10) (Danecek et al. 2021) and picard (v2.18.16) (Broad Institute 2019). Reads were filtered with bedtools (v2.27.1) (Quinlan and Hall 2010), samtools, and picard. Peaks were called using macs2 (2.1.2) and coverage was calculated with deeptools (v3.1.3) (Ramirez et al. 2016). Final quality checks were performed with multiqc (vl.6) (P. Ewels et al. 2016). Differential peak analysis was performed with DiffBind (v3.4.11) (Stark and Brown 2011) as described in the authors’ vignette (same version). ATAC-seq from in vivo samples were analyzed using the nf-core / atacseq pipeline v2.1.2 with Nextfl ow version 23.10.1. Reads were aligned to mmlO. Consensus peaks were defined by running Genrich -j -m 30 -e chrM -r -1 15 -q 0.01 -a 500 -g 15 -1 15 -d 50.

[0336] Footprint analysis

[0337] To determine the PROXI footprint, 11 bp-wide motif peaks (CIS-BP ID: M03445 2.00, mapped on mmlO or hg38) were intersected with CUT&RUN consensus peaks, respectively, using bedtools intersect -wa. This generated peaks present in the CUT&RUN consensus peak set that contain motifs and are centred on the motif. DiffTF (vl.8) (Berest et al. 2019) was then run using ATAC-seq bam files and the motif-centered peak set (Berest et al. 2019).

[0338] Single-cell RNA-seq multiplexing and library generation Deutsches Krebsforschungszentrum 45 DK17279PC Stiftung des offentlichen Rechts et al.

[0339] For single-cell RNA-sequencing, reprogrammed cells were dissociated into single cells after 2 PBS washes by Accutase digestion. Cells were lifted using cell scrapers, then suspensions were then passed through 70 pm filters into 9 mL of pre-warmed MEF medium. Cells were pelleted (800 rpm, 5 min) and resuspended in 1 mL HBSS + 0.04% BSA to > 3.5 million cells / mL, then fixed through the addition of 4* volume of ice-cold methanol to a final concentration of 80% methanol. Cells were then stored at -20° C. Barcoding oligonucleotides for multiplexing were designed based upon the ClickTag scheme (Gehring et al. 2020) and ordered with a 5’ amine group label (data not shown). A methyltetrazine group was conjugated to the oligos via the 5’- amine group. Membrane proteins on the fixed cells were conjugated to an amine-Zraw -cy- clooctene group. Subsequent incubation of oligos and cells according to the ClickTag protocol (Gehring et al. 2020) allowed chemical labelling of cells. -10,000 cells were multiplexed and loaded per GEM well in the Chromium Controller (lOx Genomics), and the Chromium Single Cell 3' v2 reagent kit was used according to the manufacturer’s instructions for gene expression library generation. Modifications were performed at the cDNA and library preparation steps as suggested in the ClickTag protocol to generate barcoded oligonucleotide libraries in parallel. Gene expression and barcode libraries were then diluted to equimolar amounts, pooled at a 9 to 1 ratio, and 26 + 98 bp paired-end sequencing was performed using NovaSeq 6000 platform (Illumina).

[0340] Single-cell RNA-seq data processing

[0341] Reprogramming single-cell RNA-seq data generated in this study was analysed using lOx Genomics Cell Ranger (v4.0.0) and Seurat (v4.3) (Hao et al. 2021). Cells containing fewer than 1,000 features, containing fewer than 2,000 reads, or with mitochondrial genes comprising over 20% of genes, were discarded. Cell doublets were removed using Scrublet (Wolock, Lopez, and Klein 2019) with a threshold of 0.35. Cells were demultiplexed based on their hashtag oligos by running Seurat HT0Demux() recursively. Expression values of cell cycle genes were regressed out using vars. to. regress in ScaleData() to reduce heterogeneity caused by cycling cells. Cells were projected into 2-dimensional space using the UMAP algorithm. Cells were clustered using the Leiden clustering algorithm with 40 dimensions and a resolution parameter of 0.35.

[0342] Regulon analysis

[0343] The PROXI regulon was defined as genes that exhibited downregulation in PROXI repressor fusion or upregulation in activator fusion, and contained a PROXI CUT&RUN peak and motif within 1 kb of the TSS. For all other transcription factors Dorothea (vl.7.2, all confidence levels) (Garcia- Alonso et al. 2019) was used to build their target gene regulon. A sub-gene-regu- latory network containing the PROXI regulon and the regulons of transcription factors that are Deutsches Krebsforschungszentrum 46 DK17279PC Stiftung des offentlichen Rechts et al. directly regulated by PR0X1 was constructed. GRaNPA was used to determine the most important transcription factors based on this sub-gene-regulatory network and differential expression analysis between Proxl overexpression and control (Kamal et al. 2023).

[0344] Activity score quantification for 4inl TFs and Proxl

[0345] 4inl transcription factors activity was inferred based on the aggregate expression of all genes within the 4inl regulon from Dorothea, calculated using Seurat addModuleScore(). PROXI activity was calculated by taking the inverse of the aggregate expression, calculated with add- ModuleScore(), of a subset of the PROXI regulon that consists of 79 high-confidence PROXI repressed target genes (data not shown). These contained a PROXI CUT&RUN peak and motif within 1 kb of the TSS and their promoters were differentially closed at day 2 following PROXI overexpression based on ATAC-seq log2FC threshold (GFP vs Proxl of > 1.

[0346] Filtering cells without transduction

[0347] PROXI activity was used as a basis for the removal of cells from the Proxl condition which were not successfully transduced with roxf overexpression lentivirus. For each Proxl -labelled cell, the proportion of GFP-labelled cells with a lower PROXI activity than that cell (GFPpro- portion), and the proportion of Proxl -labelled cells with a higher PROXI activity than that cell (Proxl proportion) were determined. Cells in which GFPproportion was larger than Proxl proportion were then excluded. A similar process was employed using 4inl- and MEF-labelled cells to remove cells from the dataset that were not transduced with 4inl overexpression lentivirus.

[0348] Signature gene analysis

[0349] For analysis of cell type gene signatures, marker genes from the Panglao database (Franzen, Gan, and Bjbrkegren 2019) were used as input with Seurat addModuleScore(). The Pearson correlation coefficient between cell type geneset scores and 4inl or PROXI activity across all cells was used as the correlation score between cell identity and transcription factor activity.

[0350] Reanalysis of Myc-driven liver tumour and DDC-liver injury scRNA-seq data

[0351] Published Myc-driven liver tumour single-cell RNA-seq (Chen et al. 2021) data was re-ana- lysed using Seurat along with the provided metadata. After data normalisation with SCTrans- form, dimensional reduction using PC A and UMAP, using the top 30 dimensions were performed. Clustering was then carried out using the Leiden clustering algorithm with a resolution of 0.2. To ascertain cell identity, scores were calculated based on the Panglao dataset using Seurat module scores. For our specific study objectives, the focus was narrowed to wild type samples classified as “healthy” and those from day 28 upon Myc-overexpression. Published DDC-liver injury single-cell RNA-seq data (Li et al. 2023) was re-analyzed using Seurat in conjunction with provided metadata, focusing on cells generated using the lOx platform and samples from DDC-injected subjects. After data normalisation with SCTransform, dimensional Deutsches Krebsforschungszentrum 47 DK17279PC Stiftung des offentlichen Rechts et al. reduction was carried out using UMAP and PCA based on 50 dimensions. Predefined cell type annotations were used by the main authors and calculated cell identity using Seurat module scores with reference to the Panglao dataset. Pseudotime trajectory was defined using Monocles. Subsequently, Proxl expression and hepatocyte identity scores across this trajectory were analysed.

[0352] Statistics

[0353] Data are presented as mean ± SD. No statistical methods were used to predetermine the sample size. The exact number of technical and biological replicates as well as the applied statistical tests are indicated in the figure or in the figure legend.

[0354] Survival analysis

[0355] For all survival analyses, including patients and mouse models, survival durations were plotted as Kaplan-Meier curves, and a log rank test used to analyse the statistical significance of differences in survival outcomes. For HCC patient survival impact analysis in Fig. le, -log Revalues) from the log rank test are shown, with values positive if survival was improved with higher candidate expression, and negative if survival was poorer.

[0356] Plasmid constructs

[0357] DNA constructs were generated by DNA synthesis (Sigma) or PCR amplification of cDNA with Q5 polymerase followed by ligation into restriction-digested vectors using indicated enzymes and T4 DNA ligase (NEB). qPCR primers

[0358] DNA oligonucleotide primers for quantitative PCR were ordered from Sigma.

[0359] Antibodies

[0360] All primary antibodies used in this study can be found in Table 1. Secondary Alexa-conjugated antibodies for immunofluorescence were used at 1 :2,000 (Invitrogen) and secondary IRDye- conjugated antibodies for Western blot were used at 1 : 10,000 (LI-COR). Deutsches Krebsforschungszentrum 48 DK17279PC Stiftung des offentlichen Rechts et al.

[0361] Table 1. Antibodies

[0362] Example 1: PROXI expression positively correlates with survival in hepatocellular carcinoma patients.

[0363] The role of PR0X1 in liver cancer is contradictory, as some studies indicate a tumour-promoting function, while others propose a tumour suppressor role (Yanfeng Liu et al. 2013; Shimoda et al. 2006). Thus, the role of PROXI as a putative hepatocyte cell fate safeguard in liver cancer Deutsches Krebsforschungszentrum 49 DK17279PC Stiftung des offentlichen Rechts et al. was examined. It was first assessed whether PR0X1 expression is dysregulated in HCC patients (Chaisaingmongkol et al. 2017). It was found that expression in tumour samples (n=62) was lower than in paired normal tissue controls (n=59) (Fig. 2a). Immunohistological analysis confirmed that PROXI protein levels are significantly reduced in HCC patient tumours compared to the adjacent non-tumour liver tissue (Fig. 2b; Fig. 12a). Then, it was analysed whether PROXI expression levels in HCC patients correlated with disease prognosis. In a cohort of 364 patients with both transcriptome and survival data (Menyhart, Nagy, and Gyorffy 2018), it was found that patients with high PROXI expression had a median survival of 81.9 months, while patients with low PROXI levels had a median survival of only 47.4 months (Fig. 2c). Strikingly, it was found that chromosomal amplifications that encompass PROXI are associated with increased survival in HCC patients (Ahn et al. 2014; Cerami et al. 2012; Gao et al. 2013; Harding et al. 2019; Ng et al. 2022; Weinstein et al. 2013; R. Xue et al. 2019) (Fig. 2d). These genetic alterations and expression changes suggest that PROXI has a tumour-suppressive role in HCC patients.

[0364] Example 2: Dose-dependent suppression of cancer cell proliferation by PROXI.

[0365] To investigate the effect of PROXI levels on the physiology of human liver cancer cells, the expression in an HCC cell line (Hep3B) in vitro was manipulated. Specifically, stable Hep3B cell lines with inducible PROXI overexpression or knockdown constructs were generated, respectively (Figs. 7b, c; Methods). While PROXI overexpression decreased proliferation by 60%, shRNA-mediated depletion significantly enhanced proliferation in vitro (Fig. 2e). To investigate the molecular mechanism behind the antiproliferative activity of PROXI, the effects of PROXI on chromatin organisation were characterised using an assay for transposase-acces- sible chromatin followed by sequencing (ATAC-seq). On day two of Proxl overexpression, 4,353 differentially-accessible peaks (padj < 0.05) of which 4,074 (93.6%) were found to be closed compared to GFP control (Fig. 2f; Figs. 7d,e). To identify PROXI target genes, CUT&RUN was conducted at the same time point with antibodies against PROXI or IgG as control. 16,183 peaks harbouring PROXI motifs and found that PROXI binding was enriched at sites that decreased in accessibility upon overexpression were identified (Fisher test, p < 2.2e- 16, odds ratio 1.6). The PROXl-bound regions included the MYC locus (Fig. 7f). RNA-se- quencing (RNA-seq) at day two followed by Gene Set Enrichment Analysis (GSEA) and transcription factor importance analysis (GRaNPA; Methods) confirmed that MYC and MYC targets were downregulated upon PROXI overexpression, which coincided with increased expression of an apoptosis gene signature (Figs. 7f,g). Next, the dose-dependent effect of PROXI levels on primary mouse liver cancer cells was investigated. To that end, lentivirus was used to introduce doxycycline-inducible Proxl in two primary cell lines derived from in vivo with distinct drivers, combining Trp53 knockout with either Myc or Kras (G12D) overexpression (Revia et al. 2022). In both models, a significant reduction in proliferation in a PROXI dose- Deutsches Krebsforschungszentrum 50 DK17279PC Stiftung des offentlichen Rechts et al. dependent manner was observed (Figs. 7h-j). This shows that PROXI primarily closes chromatin in liver cancer cells and reduces their proliferation in vitro in a dose-dependent manner by regulating gene expression.

[0366] Example 3: PROXI blocks hepatocyte transformation and liver cancer progression in mice.

[0367] Next, it was tested whether PROXI can suppress liver tumourigenesis by preventing cell fate plasticity and tumour initiation in mice. First, the expression of Proxl using available singlecell data from an HCC mouse model was assessed (Chen et al. 2021) and it was found that Proxl expression is lower in transformed cells compared to healthy hepatocytes (Fig. 2g). In addition, these tumour cells also exhibit decreased overall hepatocyte identity based on cell type-specific gene expression patterns (Fig. 2g). To investigate whether manipulating PROXI can prevent liver tumour formation, an established HCC mouse model was used. To that end, stable Myc overexpression and Trp53 knockout (MyclTrp53) was introduced via transposable elements into mouse livers via hydrodynamic tail-vein injection (HDTVI) (Fig. 2h; Fig. 8a; Methods). After two weeks, mice developed multifocal liver carcinomas resembling HCC nodules with solid and trabecular growth patterns (Fig. 8b). Tumours exhibited strong expression of the hepatocyte-specific transcription factor, hepatocyte nuclear factor-4a (HNF4a), did not express the biliary epithelial cell marker, keratin 19 (KRT19), and lacked glandular structures typical for cholangiocarcinomas (CCAs) (Fig. 8b). Constitutive overexpression of Proxl in this model led to fewer tumour nodules at the endpoint (6.7 vs 39.5 nodules, p < 0.001) and, most strikingly, all tumours in the rox7-IRES-GFP animals were GFP-negative, indicating that tumour cells strongly selected against Proxl overexpression (Fig. 2i; Figs. 8c-d). Indeed, Proxl overexpression significantly increased median survival from 29 days to 64 days and caused a significant increase in overall survival duration with several mice surviving the 100-day experiment (Fig. 2j; Figs. 81). To test dose-dependent effects during in vivo tumourigenesis, low with high PROXI overexpression was compared using PGK- and Efla-promoters, respectively, and it was found that only high PROXI levels significantly increased survival (Figs. 8j-l). This increase is biologically significant compared to other treatments such as the kinase inhibitor drug sorafenib, approved for the treatment of advanced primary liver cancer, which only increases survival in comparable mouse models by around 8 days (Rudalska et al. 2014).

[0368] Given the striking benefits of Proxl overexpression on tumour formation, and because of the beneficial correlation between survival in HCC patients with PROXI amplification, which presumably occurs later in tumour development, it was decided to also test the effect of PROXI during tumour progression. To that end, the same HCC mouse model was used and combined Deutsches Krebsforschungszentrum 51 DK17279PC Stiftung des offentlichen Rechts et al. it with doxycycline-inducible Proxl overexpression. Proxl expression was induced after tumour nodules had formed for 14 days post-HDTVI (Fig. 8e). Two days following doxycycline treatment, a >4-fold increase in apoptosis in tumours was observed as judged by CASP3 histology compared to control (Figs. 8f-g). These results were substantiated by gene expression analysis based on RNA-seq two days following Proxl overexpression, which showed upregulation of apoptosis and downregulation of pro-proliferative MYC gene signatures, mirroring the findings in Hep3B cells in vitro described above (Fig. 2g). Importantly, late Proxl overexpression resulted in fewer GFP+ tumour nodules at the endpoint (219.5 vs 13.75 nodules, p < 0.001) and significantly increased median survival from 17 days to 36.5 days in a hepatocellular-cholan- giocarcinoma model (Fig. 2j; Fig. 8h-i), suggesting that PROXI also blocks tumour progression. To determine whether PROXI broadly suppresses liver cancer, a second murine model induced by HDTVI-mediated oncogenic Kras (G12D) overexpression and Trp53 knockout (KraslTrp53) was employed (Fig. 8m). This model presents features of both HCC and CCA, characterized by morphologically complex tumours with loss of HNF4a and gain of KRT19 staining (Fig. 8n). Proxl overexpression in this context also resulted in a significant reduction in tumour nodule count at the endpoint (3.25 vs. 19.5 nodules, p < 0.05), with all nodules lacking Proxl -IRES-GFP, and extended median survival from 26.5 days to 47.5 days (Fig. 2k-m; Figs. 8o-p). Collectively, these findings demonstrate that PROXI functions as a tumour suppressor by impeding tumour initiation and progression in distinct liver cancer models.

[0369] Example 4: PROXI is required for efficient liver regeneration upon injury.

[0370] Beyond development and cancer, cell fate plasticity also plays a key role in other fate transitions, such as during regeneration following injury or direct cell reprogramming. It was therefore investigated whether PROXI can also regulate cell fate plasticity in these processes. Upon liver injury, mature hepatocytes can undergo a dedifferentiation process associated with the reactivation of progenitor-like programs followed by proliferation and differentiation to regenerate functional hepatocytes (Li et al. 2023). Analysing available single-cell data along inferred pseudotime spanning pre-injury, injury and, post-injury cells from a 3, 5-Diethoxy carbonyl- 1,4- Dihydrocollidine (DDC)-induced mouse liver injury model (Li et al. 2023), the levels of Proxl, were assessed and it was found that its expression sharply decreased following injury and gradually recovered during hepatocyte regeneration (Fig. 3a-b). Concordantly, hepatocyte identity decreased during the injury-induced dedifferentiation and was reactivated quickly during recovery. To test whether PROXI is required during DDC-induced liver injury and regeneration conditional Proxl knockout mice were used (Martinez-Corral et al. 2015), in which exon 2 of Proxl is flanked by loxP sites (Proxl" ") (Fig. 3c). Compared to AAV Acre-transduced controls, cre-transduced homozygous Proxlzdeletion reduced the density of HNF4+ hepatocytes by -33% and increased the number of KRT19+ cholangiocytes following regeneration (Fig. Deutsches Krebsforschungszentrum 52 DK17279PC Stiftung des offentlichen Rechts et al.

[0371] 3d; Figs. 9a-c). In addition, serum levels of the liver injury marker alkaline phosphatase (ALP) were elevated two-fold in 7 ox7-deleted mice after DDC withdrawal, while alanine aminotransferase (ALT) and aspartate aminotransferase (AST) showed the same trend but did not reach significance (Fig. 3e; Fig. d). Together, this suggests that PR0X1 is required for efficient hepatocyte regeneration after liver injury.

[0372] Example 5: Liver cell fate is enhanced by active suppression of alternate cell identities.

[0373] To experimentally investigate how PR0X1 could enhance liver fate acquisition, the effects of Proxl overexpression during hepatocyte reprogramming was assessed. Specifically, single-cell RNA-seq (scRNA-seq) of untreated MEFs and on days 2, 7, and 14 of 4inl-induced hepatocyte reprogramming together with overexpression of Proxl or GFP as control was performed (Fig. 31). Following data processing and quality control filtering 22,761 cells from two reprogramming experiments were grouped into seven major clusters (Fig. 3f; Figs. lOa-d). Cells with Proxl overexpression were clearly distinct from control cells and largely grouped into three clusters that closely corresponded to the three harvest time points (Figs. 3f,g). Two alternative fate clusters populated with GFP control cells were identified (Figs. 3f,g). Each cell was scored for expression of marker genes of different cell types from the curated Panglao database (Methods), and one cluster strongly expressing MEF genes was found, reflecting the starting cell identity, and one cluster displaying a clear hepatocyte signature, indicating successful reprogramming (Fig. 3h). Overall, Proxl overexpression significantly increased the number of successfully-reprogrammed hepatocytes >7-fold (2,527 Proxl cells vs 347 control, p < 2.2e-16). Importantly, Proxl overexpression significantly increased hepatocyte cell identity in single cells at each time point, as measured by the expression of hepatocyte marker genes (Fig. 3i; Figs. 10e,f). Strikingly, 7 ox7-overexpressing cells downregulated the initial MEF cell fate more efficiently and to a greater degree, and displayed lower expression of markers of alternate cell identities, such as fibroblasts, adipocytes, myocytes, and neurons (Fig. 3i; Figs. lOg-i). These data indicated that PROXl strongly enhances hepatocyte cell fate and suppresses gene expression programs of fibroblast and alternate cell types.

[0374] Next, it was determined whether PR0X1 promotes hepatocyte identity by regulating cell typespecific gene regulatory networks. To identify PR0X1 target genes during reprogramming, FLAG-tagged Proxl or GFP was overexpressed and CUT&RUN was conducted two days later with antibodies against FLAG or IgG as control (Fig. Ila). 25,519 peaks were identified harbouring PR0X1 motifs and found that PR0X1 binding was enriched at gene promoters (Figs. llb,c). Genes within 1 kb from a PR0X1 binding peak as targets were defined and their expression was analysed to determine PR0X1 activity in single cells during reprogramming (Methods). PR0X1 activity exhibited a significant negative correlation with the fibroblast Deutsches Krebsforschungszentrum 53 DK17279PC Stiftung des offentlichen Rechts et al. identity score, suggesting that PR0X1 directly represses genes of the donor cell fate (Fig. 3j; Fig. lid). Expanding this analysis to other cell fates, it was found that PR0X1 activity negatively correlated with all tested cell identities, except for hepatocyte and oligodendrocyte signatures (Fig. 3j). Interestingly, the activity of the liver inducers - F0XA3, GATA4, HNF1A, and HNF4A (4inl) - correlated positively with hepatocyte identity genes but also with several alternative cell identities, such as astrocytes, epithelial cells, and cholangiocytes (Fig. 3j). These signatures negatively correlated with PR0X1 activity, suggesting that PR0X1 can directly repress non-hepatic gene signatures, potentially activated by 4inl, to specifically promote the desired hepatocyte fate.

[0375] Example 6: PROXI can block alternate neuronal and muscle cell reprogramming.

[0376] To determine whether PR0X1 can actively repress alternative cell fates, the effect of its overexpression on neuronal and myocyte reprogramming was tested. Therefore, Proxl or GFP were overexpressed together with the pro-neuronal transcription factor Ascii, which can convert fibroblasts into functional neurons (Chanda et al. 2014). Unlike reprogramming to hepatocytes, which was significantly enhanced by PROXI, neuronal reprogramming was almost entirely abolished upon Proxl overexpression, as determined by TUBB3 protein expression (Figs. 3k, 1; Fig. 12). Myocyte fate was among the top six signatures repressed by PROXI, and expression of the muscle marker protein Desmin was decreased by PROXI during hepatocyte reprogramming (Figs. 3j; Fig. 121) It was therefore also tested whether PROXI could actively suppress muscle cell reprogramming induced by Myodl overexpression (Davis, Weintraub, and Lassar 1987). PROXI nearly fully abolished induced myocyte reprogramming determined by Desmin protein expression and quantification of Desmin positive cells (Figs. 3k, 1; Fig. 12). Interestingly, while MYODI alone did not induce any liver-like cells, 18% of reprogrammed cells expressed TJP1 and displayed hepatic morphology upon Proxl co-expression (Fig. 121). Of note, the neuronal safeguard repressor MYT1L had an analogous effect, inhibiting muscle and liver cell reprogramming while promoting only neuronal identity during reprogramming (Figs. 12d-f). This is in line with the previous finding that MYODI exhibits promiscuous activity and suggests that safeguard repressors, such as PROXI or MYT1L, can redirect this activity to promote a specific, non-repressed cell fate (Q. Y. Lee et al. 2020). In summary, PROXI not only suppressed neuronal and muscle genes during hepatocyte reprogramming, but also strongly reduced, and even redirected, the cell fates induced by neuronal and muscle master regulators.

[0377] Example 7: PROXI is necessary to prevent alternate fates during hepatocyte reprogramming. Deutsches Krebsforschungszentrum 54 DK17279PC Stiftung des offentlichen Rechts et al.

[0378] Since exogenous Proxl overexpression was sufficient to suppress alternate fates, it was determined whether endogenous Proxl expression is necessary for efficient hepatocyte reprogramming. First, shRNA-mediated Proxl knockdown was performed during hepatocyte reprogramming. Proxl downregulation by qRT-PCR was confirmed, and it was found that several hepatocyte marker genes, such as Alb, Krtl8, and Cdhl, were decreased upon Proxl depletion (Fig. 13a). Monitoring Albumin secretion and TJP1 immunofluorescence confirmed significantly impaired liver reprogramming upon Proxl knockdown (Figs. 13b, c). To substantiate these results, mouse embryonic fibroblasts from the conditional Proxl knockout mice (Proxl^A MEF) were prepared (Fig. 3m; Methods). As expected, cre-mediated homozygous Proxlzdeletion significantly reduced the overall expression of Proxl and other liver markers compared to Acre- transduced isogenic controls (Fig. 13d). Importantly, conditional Proxl deletion decreased the number of TJP1 -positive hepatocyte-like cells >7-fold and significantly lowered Albumin secretion per cell (Figs. 3n,o). Conversely, the fraction of cells that expressed the muscle marker Desmin significantly increased upon Proxl deletion during liver reprogramming, while expression of E-cadherin decreased (Fig. 3p; Fig. 13e). Gene expression analysis by RNA-seq confirmed that genetic Proxlzdeletion during hepatocyte reprogramming decreased the expression of many liver markers, and increased the expression of alternate fate markers such as neuronal Map2 or muscle Myh9 (Fig. 131). Overall, these experiments showed that PR0X1 is necessary and sufficient for efficient liver cell fate induction by repressing alternate fates.

[0379] Example 8: Direct repression of PROXI target genes enhances liver fate while activation induces fate plasticity.

[0380] In mice, Proxl is also expressed in some neural stem cells in the hippocampus and cerebellum, in which it has been shown to promote neurogenesis (Karalay et al. 2011; Lavado et al. 2010), and is also a key regulator of lymphatic endothelial cell differentiation and maintenance (Petrova 2002; Wigle 2002). In these contexts, PROXI mainly activates gene expression in combination with coactivators, such as NR2F2 (a.k.a COUP-TFII) (Aranguren et al. 2013; Iwano et al. 2012). Conversely, in the liver, PROXI was found to interact with corepressors, such as histone deacetylases (Armour et al. 2017). To systematically expand on these findings, immunoprecipitation was performed followed by mass spectrometry to identify PROXI interaction partners in primary hepatocytes and neurons from mouse liver and hippocampus, respectively. Intriguingly, it was found that PROXI interacted with 9 out of 14 members of the repressive Nucleosome Remodeling and Deacetylase (NuRD) complex only in the liver. The NuRD complex contains histone deacetylases (HDACs) that catalyse the removal of acetyl groups from histones to mediate repression. This supports the notion that cell type-specific cofactor interactions enable PROXI to switch between gene activation and repression (Fig. 4a; Fig. 14) Hence, cofactor-dependent effects were uncoupled to study which gene regulatory networks can be targeted directly by PROXI to regulate cell identity. To directly activate or Deutsches Krebsforschungszentrum 55 DK17279PC Stiftung des offentlichen Rechts et al. repress PR0X1 target genes, the DNA-binding domain (DBD) of PROXI was fused to either a transcriptional activator (VP64) or the Engrailed repressor (EnR) (Fig. 4b). As a control, the DBD without an effector domain was expressed. While the DBD alone did not affect hepatocyte reprogramming, the repressor fusion improved hepatocyte fate induction similarly to full-length PR0X1, as determined by the number of TJP1 -positive hepatocyte-like cells and amount of Albumin secretion per cell (Figs. 4c-e; Figs. 15a-c). Conversely, the activator fusion had a dominant negative effect and significantly impaired hepatocyte reprogramming. Furthermore, the fraction of reprogrammed cells that expressed alternate neuronal or myocyte markers was decreased by the repressor fusion and increased by the activator fusion (Fig. 4f; Fig. 15d).

[0381] Since PR0X1 inhibited neuronal and myocyte reprogramming, the effects of the activator and repressor fusions in these contexts were also tested. As expected, the repressor fusion reduced myocyte and neuronal induction, but a similar effect was observed when the DBD alone was expressed (Figs. 15a-c). However, the fraction of TJP1 -positive hepatocyte-like cells was increased upon repressor fusion expression during neuronal and myocyte reprogramming (Fig. 15d). Conversely, the activator fusion reduced neuronal cell induction and enhanced myocyte induction, and increased the fraction of Desmin-positive cells in both settings relative to DBD (Figs. 15a-d). Transcriptome analysis verified that, during hepatocyte reprogramming, the EnR fusion triggered a similar response in gene expression as full-length PR0X1, causing induction of hepatocyte genes such as Hnf4a and Krtl8, and concomitant repression of non-hepatocyte genes such as the cholangiocyte marker Krtl9 and the fibroblast transcription factor Prrxl (Fig. 4g; Fig. 20e). Indeed, a significant overlap was found between genes downregulated by PR0X1 and the repressor fusion (Figs. 4g, h). Strikingly, the activator fusion had the opposite effect, decreasing hepatocyte gene expression and increasing alternate fate gene expression. Taken together, the experiments suggest that PR0X1 directly binds and represses genes that drive non-hepatocyte fates, and thereby actively suppresses alternative fate trajectories to promote hepatocyte identity.

[0382] Example 9: PROXI decreases chromatin accessibility and expression of alternate fate signature genes.

[0383] To investigate the effects of PR0X1 on chromatin organisation during hepatocyte reprogramming, ATAC-seq was performed. On day two of 4inl -induced hepatocyte reprogramming, 111,411 differentially-accessible peaks (padj < 0.05) were found upon overexpression of Proxl compared to GFP control, of which 85,140 (76.4%) were closed (Fig. 4i; Figs. 16a, b). Proxl or GFP control were also overexpressed without 4inl and obtained a similar number of differentially-accessible peaks (109,917), of which 84,800 (77.1%) were closed upon Proxl overexpression. Hence, as in cancer cells (Fig. 21), PROXI closed many chromatin regions, explaining -80% of the variation between conditions (Figs. 16c, d). Next, chromatin remodelling at Deutsches Krebsforschungszentrum 56 DK17279PC Stiftung des offentlichen Rechts et al.

[0384] PR0X1 -bound target genes using the PR0X1 CUT&RUN DNA-binding data were characterised (Fig. 11). A repressive signature and decreased accessibility at PROXI -bound sites upon Proxl overexpression were found, corroborating its repressive role (Fig. 16e; Methods). Furthermore, 74% of genes with a PR0X1 -bound and differentially-accessible region in their promoters were downregulated upon Proxl overexpression, and none of the 71 upregulated genes were hepatocyte-specific markers (Fig. 161). This shows that PR0X1 primarily closes chromatin at directly-bound target genes, reducing their expression in fibroblasts and the early stages of hepatocyte reprogramming.

[0385] To explore how PR0X1 promotes hepatocyte fate at later stages of reprogramming, and to identify repressed PR0X1 target genes that might mediate these effects, a time course transcriptome analysis using bulk RNA-seq was performed. Specifically, differential gene expression during hepatocyte reprogramming with GFP (control) or Proxl overexpression at days 7, 14, and 28 was assessed (Fig. 4i). Globally, 8,036 protein-coding genes deregulated between Proxl and control at least at one of the three-time points were found (Methods). Of these, 3,629 were up- and 3,264 were down-regulated consistently across all three-time points upon Proxl overexpression compared to control. Based on promoter accessibility at day two and differential gene expression over time, these genes were grouped into 2 clusters (Fig. 4i). Genes in cluster 1 were upregulated throughout the 4 weeks and displayed increased promoter accessibility at day two upon Proxl overexpression (t-test, p < 1.487e-05) (Fig. 17a). Cluster 1 was enriched for hepatocyte identity genes, but also contained some non-hepatic genes such as cholangiocyte and oligodendrocyte markers (Fig. 4j). On the other hand, cluster 2 was down-regulated upon Proxl overexpression throughout the 28-day reprogramming experiment and had reduced promoter accessibility at day two (t-test, p < 6.637e-09) (Fig. 4i; Fig. 17a). Genes in these clusters were enriched in many non-hepatocyte identity markers, including neuron and muscle genes, as well as fibroblast and adipocyte genes (Fig. 4j). During the 28-day reprogramming period, the expression levels of genes in cluster 2 increased in control cells but decreased upon Proxl overexpression (Fig. 17a). Notably, genes from cluster 2 were expressed mostly in the PROX1- repressed alternative fate clusters of the single-cell dataset (Fig. 17b). By contrast, genes from clusters 1 were highly expressed in reprogrammed hepatocytes in the single-cell transcriptomics dataset.

[0386] Next, it was determined whether any of these clusters were directly regulated by the overexpressed transcription factors. To determine the direct effects of the liver reprogramming factors FOXA3, GATA4, HNF1A, and HNF4A (4inl), the target genes of these four transcription factors and combined them into a 4inl regulon containing 145 target genes were retrieved (Methods). In addition, a PROXI regulon comprising 1,411 genes, which were bound by PROXI based on CUT&RUN was defined and direct regulation as determined through activator or re- Deutsches Krebsforschungszentrum 57 DK17279PC Stiftung des offentlichen Rechts et al. pressor fusion transcriptome analysis was shown (Fig. 4g; Fig. 11). The hepatocyte gene-enriched cluster 1 was significantly enriched for 4inl target genes and was strongly induced during reprogramming (Figs. 4i-k). Conversely, PR0X1 target genes were significantly depleted in cluster 1, suggesting that 4inl directly enhances hepatocyte maturation while PR0X1 indirectly promoted this effect (Figs. 4j,k). Indeed, cluster 2, which contained 8 alternate fate signatures, was significantly enriched for direct PR0X1 targets (Figs. 4j,k). Since this cluster was downregulated upon Proxl overexpression (Fig. 4i), this further supports that PR0X1 mediates active repression of unwanted fates to promote liver cell induction and maturation.

[0387] The list of PR0X1 target genes in Table 2 herein above was generated by first defining Proxl lose regulon from Proxl cut&run and ATAC data without expression-based filtering (n = 1411) in MEF to hepatocyte reprogramming. From this, we derived Proxl targets (n = 448) by retaining only genes with significant expression changes upon Proxl fusion construct overexpression in MEFs. This set was then intersected with genes downregulated in Proxl overexpression in Akt / Notch mouse cholangiocarcinoma models (Proxl targets neg akt OE, n = 289) and with genes upregulated in Proxl knockout in Hepatocellular Carcinoma mouse models (Proxl_targets_pos_HCC_KO, n = 208). The final list of Table 2 contains genes present in both of these latter sets.

[0388] Example 10: Prrxl and Pparg are two alternate fate inducers repressed by PROXI.

[0389] To understand how PROXI silences many non-hepatocyte cell fates, and to identify key PROXI target genes, GRaNPA was employed, a computational method to predict the importance of specific transcription factors based on differential gene expression (Kamal et al. 2023). Gene regulatory networks for PROXI were constructed and direct PROXI targets were annotated to be transcription factors. As expected, differential expression at day two is almost exclusively explained by PROXI (Fig. 41). At day 7, additional transcription factors were predicted, namely the direct PROXI targets PRRX1 and EBF2, to regulate differential gene expression (Fig. 41). Importantly, based on network analysis, PROXI remained important at days 14 and 28, as did its direct targets PRRX1 and EBF2. The observation that PROXI remains important throughout the 4-week experiment supports its function in hepatocyte fate maintenance (Fig. 41). Starting from day 28, additional transcription factors predicted to be regulated by PROXI were found, such as the cardiac regulator HAND2 (Fernandez-Perez et al. 2019) (Fig. 41). Strikingly, the regulons of all transcription factors targeted by PROXI were enriched in the repressed non-hepatocyte cluster 2 (Fig. 4j,k). Unlike PROXI, all downstream transcription factors are predicted to be activators based on expression correlation with their target genes (Fig. 4m) Therefore, the gene regulatory network analysis suggests that PROXI safeguards hepatocyte identity by silencing alternate cell fate genes both directly and by repression of transcription factors that activate non-hepatocyte gene programs. Deutsches Krebsforschungszentrum 58 DK17279PC Stiftung des offentlichen Rechts et al.

[0390] The donor fibroblast identity and alternate adipocyte identity were among the gene signatures most strongly repressed by PR0X1, based upon bulk and single-cell transcriptomic inference during hepatocyte reprogramming (Figs. 3j and 4j). In line with this, the analyses identified master regulators of both cell types as direct PR0X1 targets. EBF2 is a co-activator of PPARG, and together they can drive adipogenesis (Jimenez et al. 2007; Rosen et al. 1999). PRRX1 is a master transcription factor of stromal fibroblasts, regulating the differentiation of mesodermal cell types and driving fibrosis during wound healing (Leavitt et al. 2020; K.-W. Lee et al. 2022). It was verified that Prrxl and Pparg were bound by PROXI at their promoters and displayed reduced chromatin accessibility upon Proxl overexpression, confirming that they are direct PROXI targets (Fig. 18a). Prrxl or Pparg were overexpressed during 4inl-mediated hepatocyte reprogramming and observed, similar to Proxl deletion, reduced Albumin expression per cell by -50% and impaired liver fate induction based on TJP 1 -immunofluorescence (Figs. 4n,o). In line, overexpression of Prrxl or Pparg together with Proxl cancelled the positive effects of PROXI during liver reprogramming (Figs. 18b, c). Conversely, shRNA-mediated depletion of Prrxl or Pparg increased Albumin expression per cell -1-2 fold, and the number of TJPl-positive hepatocyte-like cells upon 4inl expression (Figs. 4n,o; Figs. 18b, d,e), similar to Proxl overexpression. Depletion of Prrxl or Pparg alongside Proxl overexpression did not further enhance hepatocyte reprogramming based on Albumin protein levels and TJP 1 -immunofluorescence, further indicating that both act downstream of PROXI (Fig. 4p; Fig. 18e). In summary, it was found that PROXI suppresses increased plasticity by repressing non-hepatic cell identities via direct transcriptional silencing of master regulators of alternate lineages, including fibroblast-specific Prrxl and the adipocyte regulator Pparg.

[0391] Example 11: PROXI regulates plasticity between cholangiocarcinoma vs hepatocellular carcinoma fates.

[0392] Cellular plasticity and transdifferentiation also play important roles in liver cancer. The two dominant forms of primary liver cancer, cholangiocarcinoma (CCA) and hepatocellular carcinoma (HCC) (Fan et al. 2012; Seehawer et al. 2018), which differ markedly in cellular composition and morphology (Farazi and DePinho 2006; Rizvi and Gores 2013), can both arise from hepatocytes. However, the transcriptional mechanisms that regulate the transformation of hepatocytes to HCC vs CCA are largely unknown. To investigate whether PROXI could play a role in this process, it was assessed whether PROXI expression differs between CCA and HCC samples from 153 liver cancer patients (Chaisaingmongkol et al. 2017). It was found that the median PROXI expression was 1.4-fold higher in HCC (n=62) patient samples than in CCA (n=91) (Fig. 5a). Interestingly, survival in HCC patients with high PROXI expression is significantly increased compared to CCA patients (Figs. 19a). This was mirrored by the positive correlation of PROXI expression and the HCC marker HNF4A (Fig. 5b). Intriguingly, not only Deutsches Krebsforschungszentrum 59 DK17279PC Stiftung des offentlichen Rechts et al. the CCA marker KRT19 but also the PR0X1 targets PRRX1 and PPARG exhibited a negative correlation with PR0X1 expression in tumour tissue. This suggests that PR0X1 could also prevent plasticity in primary liver cancer by suppression of alternate fate regulators and may regulate the transformation trajectory of hepatocytes towards HCC instead of CCA.

[0393] To determine whether PR0X1 loss influences liver cancer identity in vivo, CRISPR-mediated Proxl knockout during HDTVI-mediated HCC formation in the MydTrp53 model was performed. Consistent with a tumour suppressor role, Proxl knockout in this model increased the number of tumours more than two-fold and decreased the median survival by 10 days (Figs. 19b-d) To substantiate the results and directly follow perturbed cells, knockdown of Proxl was performed using shRNA constructs coupled to a GFP-reporter in the same model. Proxl knockdown led to a 10-fold increase in the number of microscopically detectable tumour nodules two weeks following HDTVI, but did not affect tumour number and survival at the endpoint (Figs. 19e-i). Strikingly, Proxl knockdown induced a morphological shift from HCC towards CCA, with the formation of glandular tumour structures accompanied by significantly decreased HNF4 expression and concomitant increase in KRT19 expression (Figs. 5c, d).

[0394] Next, it was investigated whether PROXI gain could drive liver cancer identity from CCA towards HCC. To that end, an established CCA mouse model using HDTVI-mediated overexpression of Akt and the Notchl receptor intracellular domain (NICD) (Akt / Notcli) was employed (Fan et al. 2012). After two to three weeks this model induced multifocal liver carcinomas with glandular structures and high KRT19 levels, which lacked HNF4a expression and mimicked CCA (Fig. 5c; Figs. 20a, b). Constitutive overexpression of Proxl in this model led to fewer tumour nodules at the endpoint (59.8 vs 245.2 nodules, p < 0.05) and significantly increased median survival from 54 days to 88 days, indicating that Proxl overexpression also impaired hepatocyte formation also in this model (Figs. 20c-f). Importantly, Proxl overexpression, when combined with Akt / Notch, induced a morphological shift from CCA towards HCC, with a reduction of glandular structures, significantly decreased KRT19 expression, and a concomitant increase in HNF4 expression (Figs. 5c, d). Strikingly, performing doxycycline-inducible late Proxl overexpression after tumour nodules were allowed to form for 20 days following HDTVI also resulted in decreased gland-like morphologies and reduced expression of the chol- angiocyte markers KRT19 and SOX9 (Figs. 20g-j).

[0395] Transcriptome analysis of tumour nodules from the CCA (AktlNotdi) and the HCC (MydTrp53) mice corroborated these shifts, with Proxl knockdown in HCC causing downregulation of several hepatocyte markers such as Alb, Apoal, Cdhl, Trf, and Ttr, and upregulation of cholangiocyte markers, such as Krtl9 and Sox9 (Fig. 5e), while Proxl overexpression in CCA resulted in the opposite effect. In line with expression in human liver cancer patient samples, it was observed that manipulating PROXI in mouse liver cancer models could switch the Deutsches Krebsforschungszentrum 60 DK17279PC Stiftung des offentlichen Rechts et al. transformation trajectory of hepatocytes between CCA and HCC. Overall, this suggests that PROXI can act as a safeguard repressor to maintain hepatocyte cell identity and prevent liver disease in vivo, with lower levels permitting increased plasticity and higher levels reducing the potential for transformation and transdifferentiation.

[0396] Example 12: RESULTS

[0397] Identifying safeguard repressor candidates across eighteen cell types.

[0398] To globally identify safeguard repressors that can suppress cell fate plasticity and maintain cell identity by actively silencing alternative fates, three cardinal features were defined and they should: i) exhibit lifelong and cell type-specific expression; ii) bind and repress genes expressed in other cell types; and iii) promote and maintain a particular cell identity (Fig. la). To identify candidates, focus was placed on 18 well-characterised cell types spanning all three germ layers and used expression data from Tabula Muris, a single-cell gene expression atlas derived from multiple tissues of approximately three-month-old adult mice (The Tabula Muris Consortium et al. 2018a) (Fig. 6a). First, cell type-specific expression for all 1,296 transcription factors detected across the 18 cell types were calculated. Suitable candidates exhibit high expression levels in the desired cell type while displaying low expression in others. Next, cell type-specific gene expression signatures for all 18 cell types were created (Fig. 6b). For each transcription factor the number of binding motifs in the promoters of these signature genes were counted. For safeguard repressor candidates, it was expected that their DNA-binding motifs are depleted at signature genes of the desired cell type and enriched at alternative fate genes. Finally, transcription factor expression levels was integrated and DNA-binding motif depletion at signature genes to derive a safeguard repressor score for each transcription factor in each cell type was determined (Fig. 6c; Methods). A searchable database of this analysis is accessible at apps.embl.de / safeguard (username: reviewer; password: safeguardl2345).

[0399] Following this approach, 59 candidates across 18 cell types were shortlisted (Fig. lb). 50 of these transcription factors have reported repressor or dual activator / repressor function based on literature evidence (data not shown). Further 33 candidates were classified as lifelong-expressed based on continued expression in approximately two-year-old mice from Tabula Muris Senis (The Tabula Muris Consortium et al. 2020). Indeed, using a developmental gene expression atlas for heart, brain, and liver tissues it was found that 77% (17 / 22) of the candidates were expressed at a high level throughout life in the respective cell type (Cardoso-Moreira et al. 2019), which is a significant enrichment compared to all transcription factors expressed in these organs (data not shown). Across all analysed germ layers and cell types, 27 candidates satisfied the criteria for lifelong safeguard repressors. Based on previous work, 14 of these candidates Deutsches Krebsforschungszentrum 61 DK17279PC Stiftung des offentlichen Rechts et al. promote the predicted cell fate during development or reprogramming (data not shown), and some act, at least in part, as repressors, including TBX5 in cardiomyocytes and 0LIG2 in oligodendrocytes (Waldron et al. 2016; Zhou, Choi, and Anderson 2001). The screen also identified MYT1L as a safeguard repressor, confirming the previous experimental evidence that MYT1L induced and maintained neuronal identity by actively repressing non-neuronal genes (Fig. lb; Figs. 16d-f) (Q. Y. Lee et al. 2020; Mall et al. 2017). For the top hepatocyte candidate PROXI, chromatin binding data was generated from primary mouse liver using CUT&RUN that confirmed binding enrichment at many non-hepatocyte specific genes corroborating the results from the motif-based screen (Figs. Ic-d; Fig. 6g). Interestingly, 59% (16 / 27) of the candidates are reported to have tumour-suppressive roles in their respective cell types, this enrichment was especially prominent for endodermal candidates including in hepatocytes (Fig. lb)

[0400] PROXI is the most biologically-relevant hepatocyte-specific safeguard repressor.

[0401] Cell fate loss plays a crucial role in liver disease, including dedifferentiation and transdifferentiation in the development of primary liver cancer (Seehawer et al. 2018; Tschaharganeh et al. 2014). Thus, the expression of the hepatocyte-specific candidates in hepatocellular carcinoma (HCC) patients from The Cancer Genome Atlas (TCGA) were analysed. Of the top six predicted safeguard repressors, high expression of PROXI, KLF15, ONECUT2, and ZNF771 was associated with better prognosis, suggesting they could function as putative liver tumour suppressors (Fig. le; Fig. 6h; Methods). However, only Proxl. Onecul2. and Klfl5 showed high lifelong expression throughout development and ageing in the liver (Fig. lb, f).

[0402] To test whether the top three candidates promote hepatocyte identity, their effects in the context of cell fate reprogramming were studied. Previous work showed that overexpression of four liver transcription factors - Foxa3. Gata4, Hnfla, &n Hnf4a - from a single polycistronic vector (4inl) reprogrammed mouse embryonic fibroblasts (MEFs) towards induced hepatocytes (Song et al. 2016). Thus, the effects o Proxl, Onecut2, or Klfl5 overexpression during hepatocyte reprogramming were assessed (Fig. 1g). While all candidate factors were expressed at a similar level, only PROXI increased expression levels of several hepatocyte-specific proteins and genes, such as Cdhl and Krtl8 (Figs. 16i-l). Indeed, compared to baseline reprogramming, PROXI overexpression generated >10-fold more hepatocyte-like cells based on morphology and TJP1 expression, and increased Albumin secretion per cell 17-fold (Figs. lh,i). PROXI is a Prospero homeobox transcription factor that plays a role in the development of several tissues, including the liver (Seth et al. 2014). The striking promotion of hepatocyte cell fate by PROXI during reprogramming and the association of high expression with better survival in HCC patients encouraged us to investigate the role of PROXI as a potential liver tumour suppressor. Deutsches Krebsforschungszentrum 62 DK17279PC Stiftung des offentlichen Rechts et al.

[0403] Example 13: DISCUSSION

[0404] Several dozen transcription factors might be expressed in a given cell, but only a handful of so- called selectors or master regulators are sufficient to induce specific lineage identities by activating gene expression during development. Interestingly, ablating lineage regulators can cause expression of alternative lineage genes. For instance, the deletion of Pax5 or EBF1 in B cells results in the upregulation of myeloid gene expression (Cobaleda et al. 2007; Nechanitzky et al. 2013). Similarly, genetic removal of Bell lb in T cells leads to the derepression of NK cellspecific genes, facilitating their transdifferentiation into NK-T cells (Li et al. 2010). In addition, loss of some of these genes, such as Pax5, contributes to cancer formation (Cobaleda, Jochum, and Busslinger 2007; Hanahan 2022; Krah et al. 2015). In addition, mounting evidence suggests that repressive factors play critical roles in safeguarding cell fate induction and maintenance by preventing unwanted gene expression (Lim et al. 2024). For instance, loss of repressive polycomb group proteins in mouse embryonic stem cells (mESCs) causes activation of developmental transcription factors (Boyer et al. 2006), and deletion of the BAF chromatin-remodelling complex member Brm during directed cardiogenesis from mESCs induces neuronal transcription factor expression, which shifts the identity of precardiac mesoderm to neural precursors (Hota et al. 2022). Besides these ubiquitously-expressed repressive chromatin regulators, recent examples of cell type-specific transcription repressors have been reported to safeguard cell fate (Lim et al. 2024). For example, the male germ cell-specific repressor Kmg inhibits the expression of somatic lineage genes in D. melanogaster (Kim et al. 2017). Similarly, the neuronspecific repressor MYT1L prevents the expression of non-neuronal genes in neurons to induce and maintain mouse neuronal cell fate and function (Q. Y. Lee et al. 2020; Mall et al. 2017). Here, evidence is provided that PROXI represses non-hepatocyte genes to promote and cement liver cell identity.

[0405] The findings allow to outline a speculative model to explain the mechanism of action of such cell type-specific safeguard repressors. It is proposed that safeguard repressors regulate target genes that are inappropriately accessible (Pujadas and Feinberg 2012), and that they do so (i) based on their affinity to specific DNA motifs, while exhibiting (ii) cell type-specific, and (iii) continuous expression. Analysis of 1,296 transcription factors identified 27 potential safeguard repressor candidates across 18 cell types, with over 50%, including MYT1L, previously linked to promoting their predicted cell fates. PROXI, essential for hepatocyte commitment in mice and human cells (Du et al. 2014; Huang et al. 2014; Sosa-Pineda, Wigle, and Oliver 2000), meets the criteria of a hepatocyte safeguard repressor. First, PROXI exhibits high lifelong expression in hepatocytes, enhanced reprogramming efficiency 10-fold, and suppressed most tested alternate cell identity gene signatures. Second, deletion of Proxl reduced hepatocyte reprogramming by 87% and impaired liver regeneration following injury. Third, PROXI binding Deutsches Krebsforschungszentrum 63 DK17279PC Stiftung des offentlichen Rechts et al. was associated with decreased chromatin accessibility and gene repression of alternate fate regulators.

[0406] This contrasts with reports studying the role of PR0X1 in other lineages. Notably, in mice, Proxl is expressed in some cells in the hippocampus and cerebellum, in which it has been shown to promote neurogenesis (Karalay et al. 2011; Lavado et al. 2010), and is also a master regulator of lymphatic endothelial cell differentiation (Petrova 2002; Wigle 2002). In these contexts, PROXI is reported to mainly induce gene expression in combination with coactivators, while in hepatocytes corepressor interactions are reported (Aranguren et al. 2013; Armour et al. 2017; Iwano et al. 2012). In line with this, it was found that PROXI was interacting with the repressive NurD complex in the liver but not in the hippocampus. Decoupling PROXI target gene regulation from the effects of recruited cofactors by fusion of the DNA binding domain of PROXI to activator and repressor effector domains confirmed that PROXI target gene repression promoted hepatic fate, while target gene activation induced alternate fates. Hence, depending on cofactor interaction a transcription factor can switch from a terminal selector into a safeguard repressor. Further studies are needed to characterise whether additional repressors guard cell fate canalisation in this manner, and if their life-long expression might help them to act as “terminal repressors” to prevent cell fate plasticity or even dedifferentiation and disease in mature cell types.

[0407] In line with this, the role of PROXI in maintaining hepatocyte cell fate and preventing liver cancer in vivo was investigated. PROXI expression is decreased in liver cancer, and low levels are associated with poor survival in HCC patients, suggesting that PROXI prevents hepatocyte plasticity and transformation. In mice, it was found that Proxl overexpression strongly reduces neoplastic transformation and progression in a dose-dependent manner across three HDTVI- based liver cancer models. This contrasts with previous studies, in which Proxl overexpression in established tumour cells induced greater migratory and metastatic potential upon transplantation (Y Liu et al. 2015; Yanfeng Liu et al. 2013). In line with a role in suppressing tumour initiation, Proxl knockdown or knockout accelerated tumour formation in the in vivo HCC model. Strikingly, resulting tumours displayed an identity switch away from HCC and towards CCA. Conversely, in CCA patients PROXI levels are low, and its overexpression in mouse models was able to shift CCA to HCC-like tumours. Indeed, hepatocytes are capable of giving rise to both HCC and CCA (Seehawer et al. 2018), and the data indicate that PROXI may be an important regulator of this decision, supporting the notion that safeguard repressors can prevent cell fate plasticity and block cancer development, progression and transdifferentiation in vivo.

[0408] In conclusion, the study shows that continuous cell type-specific repression of alternate fates is essential for cell fate induction and maintenance. Identifying and mechanistically characterising Deutsches Krebsforschungszentrum 64 DK17279PC Stiftung des offentlichen Rechts et al. similar factors in other cell types, guided by computational tools such as the one presented here, could help generate cells for biomedical applications and reveal targets that prevent cell fate plasticity and disease.

[0409] Literature

[0410] - Sanchez Alvarado, A. & Yamanaka, S. Rethinking differentiation: stem cells, regeneration, and plasticity. Cell 157, 110-119 (2014).

[0411] - Hanahan, D. Hallmarks of cancer: new dimensions. Cancer Discov. 12, 31-46 (2022).

[0412] - Balsalobre, A. & Drouin, J. Pioneer factors as master regulators of the epigenome and cell fate. Nat. Rev. Mol. Cell Biol. 23, 449-464 (2022).

[0413] - Garcia-Bellido, A. in Ciba Foundation Symposium 29 — Cell Patterning (eds Porter, R. & Rivers, J.) Ch. 8 (Ciba Foundation, 1975).

[0414] - Lewis, E. B. A gene complex controlling segmentation in Drosophila. Nature 276, 565-570 (1978).

[0415] - Cobaleda, C., Jochum, W. & Busslinger, M. Conversion of mature B cells into T cells by dedifferentiation to uncommitted progenitors. Nature 449, 473-477 (2007).

[0416] - Krah, N. M. et al. The acinar differentiation determinant PTF1A inhibits initiation of pancreatic ductal adenocarcinoma. eLife 4, e07125 (2015).

[0417] - Gray, S. & Levine, M. Transcriptional repression in development. Curr. Opin. Cell Biol. 8, 358-364 (1996).

[0418] - Lim, B., Domsch, K., Mall, M. & Lohmann, I. Canalizing cell fate by transcriptional repression. Mol. Syst. Biol. 20, 144-161 (2024).

[0419] - Schoenherr, C. J. & Anderson, D. J. The neuron-restrictive silencer factor (NRSF): a coordinate repressor of multiple neuron-specific genes. Science 267, 1360-1363 (1995).

[0420] - Lee, Q. Y. et al. Pro-neuronal activity of Myodl due to promiscuous binding to neuronal genes. Nat. Cell Biol. 22, 401-411 (2020).

[0421] - Mall, M. et al. Mytll safeguards neuronal identity by actively repressing many non-neuronal fates. Nature 544, 245-249 (2017).

[0422] - Hu, J. et al. Neutralization of terminal differentiation in gliomagenesis. Proc. Natl Acad. Sci. USA 110, 14520-14527 (2013).

[0423] - Weigel, B. et al. MYT1L haploinsufficiency in human neurons and mice causes autism-associated phenotypes that can be reversed by genetic and pharmacologic intervention. Mol. Psychiatry 28, 2122-2135 (2023).

[0424] - The Tabula Muris Consortium, et al. Single-cell transcriptomics of 20 mouse organs creates a Tabula Muris. Nature 562, 367-372 (2018).

[0425] - The Tabula Muris Consortium, et al. A single-cell transcriptomic atlas characterizes ageing tissues in the mouse. Nature 583, 590-595 (2020). Deutsches Krebsforschungszentrum 65 DK17279PC Stiftung des offentlichen Rechts et al.

[0426] - Cardoso-Moreira, M. et al. Gene expression across mammalian organ development. Na- ture571, 505-509 (2019).

[0427] - Seehawer, M. et al. Necroptosis microenvironment directs lineage commitment in liver cancer. Nature 562, 69-75 (2018).

[0428] - Tschaharganeh, D. F. et al. p53-dependent Nestin regulation links tumor suppression to cellular plasticity in liver cancer. Cell 158, 579-592 (2014).

[0429] - Song, G. et al. Direct reprogramming of hepatic myofibroblasts into hepatocytes in vivo attenuates liver fibrosis. Cell Stem Cell 18, 797-808 (2016).

[0430] - Seth, A. et al. Proxl ablation in hepatic progenitors causes defective hepatocyte specification and increases biliary cell commitment. Development 141, 538-547 (2014).

[0431] - Liu, Y. et al. PROXI promotes hepatocellular carcinoma metastasis by way of up-regulating hypoxia-inducible factor la expression and protein stability. Hepatology 58, 692-705 (2013).

[0432] - Shimoda, M. et al. A homeobox protein, Proxl, is involved in the differentiation, proliferation, and prognosis in hepatocellular carcinoma. Clin. Cancer Res. 12, 6005-6011 (2006).

[0433] - Chaisaingmongkol, J. et al. Common molecular subtypes among asian hepatocellular carcinoma and cholangiocarcinoma. Cancer Cell32, 57-70 (2017).

[0434] - Ahn, S.-M. et al. Genomic portrait of resectable hepatocellular carcinomas: implications of RBI and FGF19 aberrations for patient stratification. Hepatology 60, 1972-1982 (2014).

[0435] - Cerami, E. et al. The eBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov 2, 401-404 (2012).

[0436] - Gao, J. et al. Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci. Signal. 6, pll (2013).

[0437] - Harding, J. J. et al. Prospective genotyping of hepatocellular carcinoma: clinical implications of next-generation sequencing for matching patients to targeted and immune therapies. Clin. Cancer Res. 25, 2116-2126 (2019).

[0438] - Ng, C. K. Y. et al. Integrative proteogenomic characterization of hepatocellular carcinoma across etiologies and stages. Nat. Commun. 13, 2436 (2022).

[0439] - Weinstein, J. N. et al. The Cancer Genome Atlas Pan-Cancer analysis project. Nat. Genet. 45, 1113-1120 (2013).

[0440] - Xue, R. et al. Genomic and transcriptomic profiling of combined hepatocellular and intrahe- patic cholangiocarcinoma reveals distinct molecular subtypes. Cancer Cell 35, 932-947 (2019).

[0441] - Menyhart, O., Nagy, A. & Gyorffy, B. Determining consistent prognostic biomarkers of overall survival and vascular invasion in hepatocellular carcinoma. R. Soc. Open Sci. 5, 181006 (2018).

[0442] - Revia, S. et al. Histone H3K27 demethylase KDM6A is an epigenetic gatekeeper of mTORCl signalling in cancer. Gut 71, 1613-1628 (2022). Deutsches Krebsforschungszentrum 66 DK17279PC Stiftung des offentlichen Rechts et al.

[0443] - Chen, W. S. et al. Single-cell transcriptomics reveals opposing roles of Shp2 in Myc-driven liver tumor cells and microenvironment. Cell Rep. 37, 109974 (2021).

[0444] - Li, L. et al. Kupffer-cell-derived IL-6 is repurposed for hepatocyte dedifferentiation via activating progenitor genes from injury-specific enhancers. Cell Stem Cell 30, 283-299 (2023).

[0445] - Martinez-Corral, I. et al. Nonvenous origin of dermal lymphatic vasculature. Circ. Res. 116, 1649-1654 (2015).

[0446] - Chanda, S. et al. Generation of induced neuronal cells by the single reprogramming factor ASCL1. Stem Cell Rep. 3, 282-296 (2014).

[0447] - Davis, R. L., Weintraub, H. & Lassar, A. B. Expression of a single transfected cDNA converts fibroblasts to myoblasts. Cell 51, 987-1000 (1987).

[0448] - Karalay, O. et al. Prospero-related homeobox 1 gene (Proxl) is regulated by canonical Wnt signaling and has a stage-specific role in adult hippocampal neurogenesis. Proc. Natl Acad. Sci. USA 108, 5807-5812 (2011).

[0449] - Lavado, A., Lagutin, O. V., Chow, L. M. L., Baker, S. J. & Oliver, G. Proxl is required for granule cell maturation and intermediate progenitor maintenance during brain neurogenesis. PLoS Biol. 8, el000460 (2010).

[0450] - Petrova, T. V. Lymphatic endothelial reprogramming of vascular endothelial cells by the Prox-1 homeobox transcription factor. EMBO J. 21, 4593-4599 (2002).

[0451] - Wigle, J. T. An essential role for Proxl in the induction of the lymphatic endothelial cell phenotype. EMBO J. 21, 1505-1513 (2002).

[0452] - Aranguren, X. L. et al. COUP-TFII orchestrates venous and lymphatic endothelial identity by homo- or hetero-dimerisation with PROXL J. Cell Sci. 126, 1164-1175 (2013).

[0453] - Iwano, T., Masuda, A., Kiyonari, H., Enomoto, H. & Matsuzaki, F. Proxl postmitotically defines dentate gyrus cells by specifying granule cell identity over CA3 pyramidal cell fate in the hippocampus. Development 139, 3051-3062 (2012).

[0454] - Armour, S. M. et al. An HDAC3-PROX1 corepressor module acts on HNF4a to control hepatic triglycerides. Nat. Commun. 8, 549 (2017).

[0455] - Kamal, A. et al. GRaNIE and GRaNPA: inference and evaluation of enhancer-mediated gene regulatory networks. Mol. Syst. Biol. 19, el 1627 (2023).

[0456] - Fernandez-Perez, A. et al. Hand2 selectively reorganizes chromatin accessibility to induce pacemaker-like transcriptional reprogramming. Cell Rep. 27, 2354-2369 (2019).

[0457] - Jimenez, M. A., Akerblad, P., Sigvardsson, M. & Rosen, E. D. Critical role for Ebfl and Ebf2 in the adipogenic transcriptional cascade. Mol. Cell. Biol.27, 743-757 (2007).

[0458] - Rosen, E. D. et al. PPARy is required for the differentiation of adipose tissue in vivo and in vitro. Mol. Cell 4, 611-617 (1999).

[0459] - Leavitt, T. et al. Prrxl fibroblasts represent a pro-fibrotic lineage in the mouse ventral dermis. Cell Rep. 33, 108356 (2020). Deutsches Krebsforschungszentrum 67 DK17279PC Stiftung des offentlichen Rechts et al.

[0460] - Lee, K.-W. et al. PRRX1 is a master transcription factor of stromal fibroblasts for myofibro- blastic lineage progression. Nat. Commun. 13, 2793 (2022).

[0461] - Fan, B. et al. Cholangiocarcinomas can originate from hepatocytes in mice. J. Clin. In- vest.122, 2911-2915 (2012).

[0462] - Farazi, P. A. & DePinho, R. A. Hepatocellular carcinoma pathogenesis: from genes to environment. Nat. Rev. Cancer 6, 674-687 (2006).

[0463] - Rizvi, S. & Gores, G. J. Pathogenesis, diagnosis, and management of cholangiocarcinoma. Gastroenterology 145, 1215-1229 (2013).

[0464] - Li, P. et al. Reprogramming of T cells to natural killer-like cells upon Bell lb deletion. Science 329, 85-89 (2010).

[0465] - Nechanitzky, R. et al. Transcription factor EBF1 is essential for the maintenance of B cell identity and prevention of alternative fates in committed cells. Nat. Immunol. 14, 867-875 (2013).

[0466] - Kim, J. et al. Blocking promiscuous activation at cryptic promoters directs cell type-specific gene expression. Science 356, 717-721 (2017).

[0467] - Pujadas, E. & Feinberg, A. P. Regulated noise in the epigenetic landscape of development and disease. Cell 148, 1123-1131 (2012).

[0468] - Du, Y. et al. Human hepatocytes with drug metabolic function induced from fibroblasts by lineage reprogramming. Cell Stem Cell 14, 394-403 (2014).

[0469] - Huang, P. et al. Direct reprogramming of human fibroblasts to functional and expandable hepatocytes. Cell Stem Cell 14, 370-384 (2014).

[0470] - Sosa-Pineda, B., Wigle, J. T. & Oliver, G. Hepatocyte migration during liver development requires Proxl. Nat. Genet. 25, 254-255 (2000).

[0471] - Rudalska, R. et al. In vivo RNAi screening identifies a mechanism of sorafenib resistance in liver cancer. Nat. Med. 20, 1138-1146 (2014).

[0472] - Liu, Y. et al. PROXI promotes hepatocellular carcinoma proliferation and sorafenib resistance by enhancing P-catenin expression and nuclear translocation. Oncogene 34, 5524- 5535 (2015).

[0473] - Moorman, A. R. et al. Progressive plasticity during colorectal cancer metastasis. Nature 637, 947-954 (2025).

[0474] - Adri an- Segarra, J. M., Weigel, B. & Mall, M. in Neural Reprogramming: Methods in Molecular Biology Vol. 2352 (ed. Ahlenius, H.) 1-12 (Springer, 2021).

[0475] - Moon, S.-H. et al. p53 represses the mevalonate pathway to mediate tumor suppression. Cell 176, 564-580 (2019).

[0476] - Xue, W. et al. CRISPR-mediated direct mutation of cancer genes in the mouse liver. Nature 514, 380-384 (2014).

[0477] - Davis, G. D. & Kayser, K. J. Chromosomal Mutagenesis Vol. 435, pp. 95-108 (Humana Press, 2008). Deutsches Krebsforschungszentrum 68 DK17279PC Stiftung des offentlichen Rechts et al.

[0478] - Bankhead, P. et al. QuPath: open source software for digital pathology image analysis. Sci. Rep. 7, 16878 (2017).

[0479] - Muller, T. et al. Automated sample preparation with SP3 for low-input clinical proteomics. Mol. Syst. Biol. 16, e9111 (2020).

[0480] - Tyanova, S., Temu, T. & Cox, J. The MaxQuant computational platform for mass spectrometry-based shotgun proteomics. Nat. Protoc. 11, 2301-2319 (2016).

[0481] - Cox, J. et al. Accurate proteome-wide label-free quantification by delayed normalization and maximal peptide ratio extraction, termed MaxLFQ. Mol. Cell. Proteomics 13, 2513-2526 (2014).

[0482] - Schwanhausser, B. et al. Global quantification of mammalian gene expression control. Na- ture473, 337-342 (2011).

[0483] - Tyanova, S. & Cox, J. Perseus: a bioinformatics platform for integrative analysis of proteomics data in cancer research. Methods Mol. Biol. 1711, 133-148 (2018).

[0484] - Stekhoven, D. J. & Biihlmann, P. MissForest — non-parametric missing value imputation for mixed-type data. Bioinformatics 28, 112-118 (2012).

[0485] - Ritchie, M. E. et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 43, e47 (2015).

[0486] - Dull, T. et al. A third-generation lentivirus vector with a conditional packaging system. J. Virol. 72, 8463-8471 (1998).

[0487] - Becker, J. et al. Ex vivo and in vivo suppression of SARS-CoV-2 with combinatorial AAV / RNAi expression vectors. Mol. Ther. 30, 2005-2023 (2022).

[0488] - Adri an- Segarra, J. M., Weigel, B. & Mall, M. in Neural Reprogramming: Methods in Molecular Biology Vol. 2352 (ed. Ahlenius, H.) 227-236 (Springer, 2021).

[0489] - Lambert, S. A. et al. The human transcription factors. Cell 172, 650-665 (2018).

[0490] - Weirauch, M. T. et al. Determination and inference of eukaryotic transcription factor sequence specificity. Cell 158, 1431-1443 (2014).

[0491] - Kulakovskiy, I. V. et al. HOCOMOCO: towards a complete collection of transcription factor binding models for human and mouse via large-scale ChlP-seq analysis. Nucleic Acids Res. 46, D252-D259 (2018).

[0492] - Levin, J. Z. et al. Comprehensive comparative analysis of strand-specific RNA sequencing methods. Nat. Methods 7, 709-715 (2010).

[0493] - Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15-21 (2013).

[0494] - Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).

[0495] - Gu, Z., Eils, R. & Schlesner, M. Complex heatmaps reveal patterns and correlations in multidimensional genomic data. Bioinformatics32, 2847-2849 (2016). Deutsches Krebsforschungszentrum 69 DK17279PC Stiftung des offentlichen Rechts et al.

[0496] - Skene, P. J. & Henikoff, S. An efficient targeted nuclease strategy for high-resolution mapping of DNA binding sites. eLife 6, e21856 (2017).

[0497] - Cheshire, C. et al. nf-core / cutandrun: nf-core / cutandrun v2.0 Copper Cobra. Zenodo; doi.org / 10.5281 / zenodo.10606804 (2022).

[0498] - Ewels, P. A. et al. The nf-core framework for community-curated bioinformatics pipelines. Nat. Biotechnol. 38, 276-278 (2020).

[0499] - Quinlan, A. R. & Hall, I. M. BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics 26, 841-842 (2010).

[0500] - Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. Nat. Methods 9, 357-359 (2012).

[0501] - Ramirez, F. et al. deepTools2: a next generation web server for deep-sequencing data analysis. Nucleic Acids Res. 44, W160-W165 (2016).

[0502] - Ewels, P., Magnusson, M., Lundin, S. & Kaller, M. MultiQC: summarize analysis results for multiple tools and samples in a single report. Bioinformatics 32, 3047-3048 (2016).

[0503] - Broad Institute. Picard Toolkit, GitHub; github.com / broadinstitute / picard (2019).

[0504] - Danecek, P. et al. Twelve years of SAMtools and BCFtools. GigaScience 10, giab008 (2021).

[0505] - Krueger, F., James, F., Ewels, P., Afyounian, E. & Schuster-Boeckler, B. FelixKrue- ger / TrimGalore: vO.6.7 — DOI via Zenodo. Zenodoh; doi.org / 10.5281 / zenodo.5127899 (2021).

[0506] - Stark, R. & Brown, G. D. DiffBind: differential binding analysis of ChlP-seq PEAK data. bioconductor.org / packages / devel / bioc / vignettes / DiffBind / inst / doc / DiffBind.pdf (2011).

[0507] - McLean, C. Y. et al. GREAT improves functional interpretation of cis-regulatory regions. Nat. Biotechnol. 28, 495-501 (2010).

[0508] - Berest, I. et al. Quantification of differential transcription factor activity and multiomics- based classification into activators and repressors: diffTF. Cell Rep. 29, 3147-3159 (2019).

[0509] - Gehring, J., Hwee Park, J., Chen, S., Thomson, M. & Pachter, L. Highly multiplexed singlecell RNA-seq by DNA oligonucleotide tagging of cellular proteins. Nat. Biotechnol. 38, 35- 38 (2020).

[0510] - Hao, Y. et al. Integrated analysis of multimodal single-cell data. Cell 184, 3573-3587 (2021).

[0511] - Wolock, S. L., Lopez, R. & Klein, A. M. Scrublet: computational identification of cell doublets in single-cell transcriptomic data. Cell Syst. 8, 281-291 (2019).

[0512] - Garcia-Alonso, L., Holland, C. H., Ibrahim, M. M., Turei, D. & Saez-Rodriguez, J. Benchmark and integration of resources for the estimation of human transcription factor activities. Genome Res. 29, 1363-1375 (2019).

[0513] - Franzen, O., Gan, L.-M. & Bjorkegren, J. L. M. PanglaoDB: a web server for exploration of mouse and human single-cell RNA sequencing data. Database 2019, baz046 (2019). Deutsches Krebsforschungszentrum 70 DK17279PC Stiftung des offentlichen Rechts et al.

[0514] - Kamal, A., Lim, B., Mall, M. & Zaugg, J. B. Safeguard Proxl code repository. Zenodo; doi.org / 10.5281 / zenodo.14771231 (2024).

Claims

Deutsches Krebsforschungszentrum 71 DK17279PC Stiftung des offentlichen Rechts et al.CLAIMS1. A method for assessing liver cancer in a sample of a subject, comprising the steps of(a) determining a level at least one transcription factor being PR0X1 in a sample from the subject, and(b) assessing liver cancer based on the level determined in step a).

2. The method of claim 1, wherein step (b) comprises comparing the level of said at least one transcription factor to a reference level.

3. The method of claim 1 or 2, wherein said assessing cancer is predicting a risk of developing cancer.

4. The method of any one of claims 1 to 3, wherein said assessing cancer is predicting a mortality risk of a subject suffering from cancer.

5. The method of any one of claims 1 to 4, wherein said sample is a sample comprising cancer cells.

6. The method of any one of claims 3 to 5, wherein the liver cancer is hepatocellular carcinoma and / or cholangiocarcinoma.

7. PR0X1 or an agent increasing the level of PR0X1 in a cell for use in the prevention or treatment of liver cancer.

8. PR0X1 or an agent increasing the level of PR0X1 in a cell for use of claim 7, wherein said agent increasing the level of PR0X1 is a polynucleotide comprising an expressible nucleic acid sequence encoding PR0X1.

9. PR0X1 or an agent increasing the level of PR0X1 in a cell for use of claim 7 or 8, wherein said polynucleotide is comprised in a vector, preferably a viral vector, more preferably a retroviral vector, a lentiviral vector, an adenoviral vector, or an adeno-as- sociated virus (AAV) vector.

10. PROXI or an agent increasing the level of PROXI in a cell for use of any one of claims 7 to 9, wherein said liver cancer is hepatocellular carcinoma and / or cholangiocarci- noma.Deutsches Krebsforschungszentrum 72 DK17279PC Stiftung des offentlichen Rechts et al.

11. Use of PR0X1 for assessing liver cancer.

12. The use of claim 11 comprising at least one, preferably all steps of the method according to any one of claims 1 to 6.

13. A method for identifying a candidate transcription factor which suppresses plasticity in a cell type of interest, comprising: a) assessing the expression levels of a plurality of transcription factors in i) the cell type of interest and ii) control cell types which differ from the cell type of interest, b) selecting one or more transcription factors which has a high expression level in the cell type of interest, but a low expression level in the control cell types, c) counting the number of binding motifs for each transcription factor selected in step b) in the promoter regions of i) signature genes of the cell type of interest and ii) signature genes of the control cell types, and d) identifying a candidate transcription factor which suppresses plasticity in the cell type of interest based on the result of step c).

14. A transcription factor identified by the method of claim 13, or an agent that increases the level of the transcription factor, for use in the prevention or treatment of cancer.

15. A method for assessing a whether a transcription factor suppresses plasticity in a cell type of interest, comprising a) providing the expression level of the transcription factor in i) the cell type of interest and ii) control cell types which differ from the cell type of interest, b) comparing the expression level of said transcription factor in the cell type of interest to the expression level in the control cell types, c) counting the number of binding motifs of said transcription factor in the promoter regions of i) signature genes of the cell type of interest and ii) signature genes of the control cell types, wherein a high expression level in the cell type of interest versus a low expression level in the control cell types, in combination with a low number of binding motifs in the promoter regions of signature genes of the cell type of interest versus a high number of binding motifs in the promoter regions of signature genes of the control cell types is indicative for a transcription factor which suppresses plasticity in a cell type of interest.

Images