Improved method for quantifying the cellular composition of a biological sample

Abstract

The invention relates to a method for determining at least part of the cellular composition of at least one biological sample comprising leukocytes, wherein the proportion of at least one type of leukocyte is determined based on the identified methylation levels, in order to predict the relative distribution of that type of leukocyte in the sample. The invention further relates to the use of at least one artificial nucleic acid molecule for determining at least part of the cellular composition of a biological sample comprising leukocytes according to the above-mentioned method. In the context of the invention, the following CpG sites according to the invention with cell-type-specific methylation could be identified: cg22381196 (DHODH), cg12483340 (TMEM87A), cg06270401 (DYRK4), cg23054181 (FAM169BP), cg26076724 (RP11-146I2.2), cg12249234 (KSR1), cg04468741 (MICAL2), cg05074138 (CTLA4), cg05705140 (LINC01237), cg15564619 (SKI), cg11531557 (HMBOX1), cg02212339 (TRPV1), cg22488278 (ZFYVE28), cg02240030 (MAD1L1) and cg24408769 (JARID2).

Description

[0001] Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen 23521WO Improved method for quantifying the cellular composition of a biological sample The invention relates to a method for determining at least part of the cellular composition of at least one biological sample comprising leukocytes, wherein the proportion of at least one leukocyte type is determined based on the identified methylation levels in order to predict the relative distribution of this leukocyte type in the sample. The invention further relates to the use of at least one artificial nucleic acid molecule for determining at least part of the cellular composition of a biological sample comprising leukocytes according to the above-mentioned method. The analysis of the cellular composition of biological samples, in particular blood, is a fundamental diagnostic method that allows conclusions to be drawn about many diseases.To date, differential blood counts, for example, have been generated primarily using flow cytometry, immunophenotypic analyses, and cell morphological characteristics. The examination of white blood cells is an important component of hematological diagnostics. Nevertheless, many challenges remain even today. First, the analysis is difficult to standardize, which leads to the same blood samples producing completely different results in different laboratories. Second, blood cannot be stored for too long (a maximum of a few hours). For venous blood collection, patients usually have to see a specialist. The coronavirus crisis has clearly shown that this is not always possible and that self-collection might be useful. Depending on which cell types are to be analyzed, several methods are currently available. A stained blood smear can be examined under the microscope.There are also automatic cell counters that display information on the number of granulocytes, lymphocytes, and monocytes. If the determination of the cellular immune status, which also distinguishes between the various lymphocyte subtypes, is required, this can be achieved using antibody staining followed by flow cytometry. Flow cytometry measurements are relatively labor-intensive and cost-intensive and can only be applied to fresh blood samples. Fresh blood, which must be less than 24 hours old, is required. In addition, the required blood volume is usually larger than for a small capillary blood sample from a fingertip. Epigenetic methods such as cell-type-specific DNA methylation (DNAm) can also be used to determine the number of leukocyte subsets in the blood.However, most known methods are based on larger methylation profiles (>100 CpG dinucleotides) rather than on individual CpGs. However, the selection of a few or individual CpGs is particularly advantageous for clinical applications. In contrast to conventional methods based on cell morphology or the expression of surface markers, cellular deconvolution based on DNAm values ​​is also suitable for frozen or dried blood. For example, EP 3561074 A1 discloses a method for epigenetic blood cell counting, which identifies a quantitative picture of the cell composition in a blood sample using a normalization standard.The normalization standard is a nucleic acid molecule comprising at least one marker region specific for each of the blood cells to be detected and at least one cell-nonspecific control region, these regions being present in the same number of copies on this molecule and / or a natural blood cell sample of known composition. EP 3382033 A1 discloses an alternative epigenetic method for determining the cellular composition of peripheral blood. In this method, DNA methylation (DNAm) is measured at specific genomic regions by pyrosequencing and used for the relative quantification of granulocytes, lymphocytes, monocytes, and CD4. + -T cells, CD8 +T cells, B cells, and NK cells. The object of the invention is to improve the accuracy of determining the cellular composition of biological samples by expanding the selection of cell-type-specific CpG dinucleotides. The object is achieved according to the invention by a method for determining at least part of the cellular composition of at least one biological sample comprising leukocytes, which method comprises the following steps: - identifying DNA methylation levels of at least one region of the genomic DNA in the sample, wherein the specific region comprises at least one CpG dinucleotide, and wherein the specific region comprises at least one nucleotide sequence selected from the group consisting of cg22381196 (DHODH), cg12483340 (TMEM87A), cg06270401 (DYRK4), cg23054181 (FAM169BP), cg26076724 (RP11-146I2.2), cg12249234 (KSR1), cg04468741 (MICAL2), cg05074138 (CTLA4), cg05705140 (LINC01237), cg15564619 (SKI), cg11531557 (HMBOX1), cg02212339 (TRPV1), cg22488278 (ZFYVE28), cg02240030 (MAD1L1) and cg24408769 (JARID2); and - determining the proportion of at least one leukocyte type based on the identified methylation levels in order to predict the relative distribution of that leukocyte type in the sample, wherein - the methylation level of at least one CpG dinucleotide selected from the group consisting of CpG dinucleotides cg22381196 (DHODH), cg12483340 (TMEM87A), cg06270401 (DYRK4) and at least one CpG dinucleotide located within a region of 1,000 nucleotides upstream and / or downstream of each of said CpG dinucleotides indicates the proportion of granulocytes, - the methylation level of the CpG dinucleotide cg04468741 (MICAL2) or at least one CpG dinucleotide located within a region of 1,000 nucleotides upstream and / or downstream of each of said CpG dinucleotides indicates the proportion of granulocytes.000 nucleotides upstream and / or downstream of said CpG dinucleotide indicates the proportion of monocytes, - the methylation level of at least one CpG dinucleotide selected from the group consisting of the CpG dinucleotides cg23054181 (FAM169BP), cg26076724 (RP11-146I2.2), cg12249234 (KSR1) and at least one CpG dinucleotide located within a region of 1,000 nucleotides upstream and / or downstream of each of said CpG dinucleotides indicates the proportion of lymphocytes, - the methylation level of the CpG dinucleotide cg02212339 (TRPV1) or of at least one CpG dinucleotide located within a region of 1.000 nucleotides upstream and / or downstream of said CpG dinucleotide indicates the proportion of B cells, - the methylation level of at least one CpG dinucleotide selected from the group consisting of the dinucleotides cg05074138 (CTLA4), cg15564619 (SKI), cg05705140 (LINC01237) and at least one CpG dinucleotide located within a region of 1,000 nucleotides upstream and / or downstream of each of said CpG dinucleotides indicates the proportion of CD4. + - T cells, - and the methylation level of the CpG dinucleotide cg11531557 (HMBOX1) or of at least one CpG dinucleotide located within a region of 1,000 nucleotides upstream and / or downstream of said CpG dinucleotide indicates the proportion of CD8 +- indicates T cells, and - the methylation level of at least one CpG dinucleotide selected from the group consisting of the dinucleotides cg22488278 (ZFYVE28), cg02240030 (MAD1L1), cg24408769 (JARID2), and at least one CpG dinucleotide located within a region of 1,000 nucleotides upstream and / or downstream of each of said CpG dinucleotides indicates the proportion of NK cells. The sample number "cg..." within the meaning of the invention refers to the position of the respective CpG dinucleotide on the "450k Illumina Bead Chip" (Illumina, San Diego, California, USA). Based on DNAm profiles from over 1,300 sorted cell samples from 40 different studies, the following CpG sites with cell type-specific methylation were identified: cg22381196 (DHODH), cg12483340 (TMEM87A), cg06270401 (DYRK4), cg23054181 (FAM169BP), cg26076724 (RP11-146I2.2), cg12249234 (KSR1), cg04468741 (MICAL2), cg05074138 (CTLA4), cg05705140 (LINC01237), cg15564619 (SKI), cg11531557 (HMBOX1), cg02212339 (TRPV1), cg22488278 (ZFYVE28), cg02240030 (MAD1L1) and cg24408769 (JARID2). Using these CpG sites, flow cytometric and epigenetic estimates of relative leukocyte counts correlated in venous blood for granulocytes (r=0.95), lymphocytes (r=0.96), monocytes (r=0.82), CD4 T cells (r=0.84), CD8 T cells (r=0.94), B cells (r=0.96), and NK cells (r=0.72). Similar correlations were obtained in dried blood samples. Spike-in with a reference plasmid enabled accurate epigenetic estimation of absolute leukocyte counts from dried blood samples, which correlated with conventional venous (r=0.85) and capillary (r=0.80) blood measurements.The novel genomic regions according to the invention are even more suitable for epigenetic blood counts than the CpG sites already known from EP 3382033 A1. The method was validated in a large cohort of healthy donors and patients with hematopoietic diseases. The measurements were performed using digital droplet PCR (ddPCR). Overall, the precision and sensitivity of the measurement method could be further improved using the new genomic regions and the ddPCR measurement. The advantages of the invention over conventional blood counts lie in particular in the fact that the method according to the invention is based on DNA, which is stable over a long period of time, so that the samples to be tested can be stored for an extended period (for decades in a frozen state). It has also been demonstrated that the method according to the invention can also be performed with dried blood.Patients could thus take a blood sample at home by pricking their finger. Furthermore, the method according to the invention is quantitative, easy to standardize, and correlates with conventional measurement methods. The invention thus opens up the possibility of analyzing patients' blood counts regardless of how long the samples have been stored. Even retrospective analyses years later (e.g., for scientific studies) are possible. Furthermore, DNA can also be isolated and analyzed from dried blood samples. In order to optimize the analysis method as much as possible, all possible public datasets with methylation data were searched on "Gene Expression Omnibus." A total of datasets from 40 studies, comprising 1,303 samples, were identified. Using a proprietary, newly developed data analysis pipeline called "CimpleG," the inventors created methylation data for each cell type (granulocytes, lymphocytes, monocytes, CD4.+ -T cells, CD8 + -T cells, B cells and NK cells) the 3 most important CpGs were identified, i.e. those with the largest differences in DNA methylation between a cell type compared to all other cell types for which the standard deviation is the smallest. For all of these 21 CpGs, ddPCR assays were developed and tested in patient samples with known cell counts. The following CpGs proved to be suitable in the ddPCR validation and have not yet been described in connection with a blood count determination (see Table 1): Table 1: List of CpG sites according to the invention Blood cell type CpG site Gene, position in the genome Correlation DHODH (“Dihydroorotate granulocyte cg22381196 dehydrogenase”), 0.83 chr16:72041376 TMEM87A (“Transmembrane protein ” Granulocytes cg06270401 phosphorylation regulated kinase 4), 0.81 chr12:4699085 FAM169BP (“Family with sequence” Lymphozyten cg26076724 RNA), 0.83 chr6:15090163 KSR1 (“Kinase suppressor of ras 1”), Lymphozyten cg12249234 0.83 Monozyten cg04468741 monooxygenase, calponin and LIM 0.5 domain containing 2”), chr11:12181467 CD4 + -T- CTLA4 (“Cytotoxic T-lymphocyte cg05074138 0.92 ” cg05705140 Protein Coding RNA 1237), 0.86 Zellen chr2:242945114 CD4 + -T- SKI (“SKI proto-oncogene”), cg15564619 0.71 Zellen chr8:28919234 TRPV1 (“Transient receptor potential ” containing 28), chr4:2321957 MAD1L1 (“Mitotic Arrest Deficient 1 Like NK-Zellen cg02240030 - ” NK cells cg24408769 Interaction Domain Containing 2), - chr6:15506085 It is noteworthy that to date, some cell types cannot be accurately predicted, whereby the selection of the genomic region plays a very important role. Since the CpG sites according to the invention are based on many different public data and are measured with ddPCR, a relatively new and very precise method, instead of quantitative PCR (qPCR) or pyrosequencing, the various cell types can be predicted much better than before and with even less volume (<30 µl blood). The epigenetic leukocyte determination according to the invention also represents an alternative measurement method for cases in which conventional methods do not provide a clear result (e.g., due to coagulation). Preferably, at least one CpG dinucleotide is located within a region of 500, 400, 300,200 or 100 nucleotides upstream and / or downstream of each of the mentioned CpG dinucleotides cg22381196, cg12483340, cg06270401, cg23054181, cg26076724, cg12249234, cg04468741, cg05074138, cg05705140, cg15564619, cg11531557, cg02212339, cg22488278, cg02240030 and cg24408769. In an advantageous embodiment of the invention, it is further provided that the sample is provided with a volume of at most or less than 50 µl, at most or less than 40 µl, at most or less than 30 µl, at most or less than 20 µl, at most or less than 10 µl, 10-40 µl, 10-30 µl, 10-20 µl, 20-40 µl, 30-40 µl, 20-30 µl, 10-20 µl, 25-35 µl or preferably 30 µl. The method according to the invention thus requires very small sample volumes, so that large sample quantities do not need to be available for a successful and accurate determination of the cellular composition. The biological sample can, for example, be a blood sample.which comprises mobilized peripheral blood of a human individual. This can advantageously also be frozen blood or dried blood spots. However, the biological sample can also comprise bone marrow, cerebrospinal fluid, and / or other tissue containing leukocytes. The determination of the degree of methylation or measurement of DNA methylation (DNAm) can preferably be carried out using digital PCR (dPCR), for example digital droplet PCR (ddPCR). The object is further achieved according to the invention by the use of at least one artificial nucleic acid molecule for determining at least part of the cellular composition of a biological sample comprising leukocytes, according to the method according to the invention, wherein the nucleic acid molecule comprises at least one nucleotide sequence selected from the group consisting of: a) A nucleotide sequence selected from the group consisting of SEQ ID NO: 1, SEQ ID NO: 2,SEQ ID NO: 3 and SEQ ID NO: 4 for determining the proportion of granulocytes; b) a nucleotide sequence selected from the group consisting of SEQ ID NO: 5, SEQ ID NO: 6, SEQ ID NO: 7 and SEQ ID NO: 8 for determining the proportion of granulocytes; c) a nucleotide sequence selected from the group consisting of SEQ ID NO: 9, SEQ ID NO: 10, SEQ ID NO: 11 and SEQ ID NO: 12 for determining the proportion of granulocytes; d) a nucleotide sequence selected from the group consisting of SEQ ID NO: 13, SEQ ID NO: 14, SEQ ID NO: 15 and SEQ ID NO: 16 for determining the proportion of monocytes; e) a nucleotide sequence selected from the group consisting of SEQ ID NO: 17, SEQ ID NO: 18, SEQ ID NO: 19 and SEQ ID NO: 20 for determining the proportion of lymphocytes; f) a nucleotide sequence selected from the group consisting of SEQ ID NO: 21, SEQ ID NO: 22, SEQ ID NO: 23 and SEQ ID NO: 24,for determining the proportion of lymphocytes; g) a nucleotide sequence selected from the group consisting of SEQ ID NO: 25, SEQ ID NO: 26, SEQ ID NO: 27 and SEQ ID NO: 28 for determining the proportion of lymphocytes; h) a nucleotide sequence selected from the group consisting of SEQ ID NO: 29, SEQ ID NO: 30, SEQ ID NO: 31 and SEQ ID NO: 32 for determining the proportion of B cells; i) a nucleotide sequence selected from the group consisting of SEQ ID NO: 33, SEQ ID NO: 34, SEQ ID NO: 35 and SEQ ID NO: 36 for determining the proportion of CD4, + - T cells; j) a nucleotide sequence selected from the group consisting of SEQ ID NO: 37, SEQ ID NO: 38, SEQ ID NO: 39 and SEQ ID NO: 40 for determining the proportion of CD4 + - T cells; k) a nucleotide sequence selected from the group consisting of SEQ ID NO: 41, SEQ ID NO: 42, SEQ ID NO: 43 and SEQ ID NO: 44 for determining the proportion of CD4 +- T cells; l) a nucleotide sequence selected from the group consisting of SEQ ID NO: 45, SEQ ID NO: 46, SEQ ID NO: 47 and SEQ ID NO: 48 for determining the proportion of CD8 +- T cells; m) a nucleotide sequence selected from the group consisting of SEQ ID NO: 49, SEQ ID NO: 50, SEQ ID NO: 51 and SEQ ID NO: 52 for determining the proportion of NK cells; n) a nucleotide sequence selected from the group consisting of SEQ ID NO: 53, SEQ ID NO: 54, SEQ ID NO: 55 and SEQ ID NO: 56 for determining the proportion of NK cells; o) a nucleotide sequence selected from the group consisting of SEQ ID NO: 57, SEQ ID NO: 58, SEQ ID NO: 59 and SEQ ID NO: 60 for determining the proportion of NK cells; p) a nucleotide sequence which differs from one of the nucleotide sequences according to a) to o) by the exchange of at most 10% of the nucleotides, preferably at most 5% of the nucleotides; and q) a nucleotide sequence which corresponds to the complementary strand of one of the nucleotide sequences according to a) to p).The invention includes, for example, a kit with a compilation of suitable reference DNAs, primer sets, as well as optional instructions and evaluation software. The invention consequently also comprises a kit, in particular a kit for determining at least part of the cellular composition of blood, preferably for carrying out the method according to the invention, which comprises at least one oligonucleotide primer (for example oligonucleotide primer set", see e.g. Table 2) for amplifying and / or sequencing at least one CpG dinucleotide of at least one nucleotide sequence which comprises at least one CpG dinucleotide from the group consisting of the CpG dinucleotides cg22381196, cg12483340, cg06270401, cg23054181, cg26076724, cg12249234, cg04468741, cg05074138, cg05705140, cg15564619, cg11531557, cg02212339, cg22488278, cg02240030, cg24408769 and at least one CpG dinucleotide located within a region of 1.000 nucleotides upstream and / or downstream of each of the CpG dinucleotides mentioned. The kit according to the invention can, for example, comprise at least one primer set according to the invention and optionally at least one buffer solution and / or at least one reagent for carrying out at least one method selected from the group consisting of DNA amplification, bisulfite treatment of DNA, DNA sequencing, preferably pyrosequencing of DNA, MassArray analysis, deep sequencing of bisulfite-converted DNA, semi-quantitative PCR, digital PCR, digital droplet PCR, Epityper, amplicon sequencing, nanopore sequencing and SMART sequencing. Table 2: List of primers and probes for the ddPCR CpG site Primer Sequence SEQ ID NO: cg22381196 Before GGTTTTATTGTGTTAGTTAGGAT 1. cg23054181 Before TTTGGTTTTAAAGAAGTTTTTTGAG 17 " c = cw rs- rmer "Sm-" = probe for unmethylated strand, 5' 6-FAM - [sequence] - 3' BHQ-1 "Sm+" = probe for methylated strand, 5' HEX - [sequence] - 3' BHQ-1 The invention is explained in more detail below with reference to the figures and tables. Brief description of the figures Figure 1: Identification of CpGs with cell type-specific DNAm in leukocyte subsets. (A) Flow diagram of how Illumina BeadChip measurements from samples of purified leukocyte subsets (n) were compiled in different studies (N). Data from 40 studies were eligible for reanalysis by CimpleG, which included methylation data from 1303 sorted cell samples from 726 patients. (B) Based on this data collection, candidate CpGs for each cell type were ranked by CimpleG based on mean difference and total variance. The top 3 CpGs with the highest CimpleG score for each cell type are highlighted.(C) The heatmap shows the DNAm (beta values) for the 27 cell-type-specific CpG sites. HSPC: hematopoietic stem and progenitor cells; nRBC: nucleated red blood cells. Figure 2: Epigenetic estimates of granulocytes, lymphocytes, and monocytes. (A) Simple linear regression models for monocytes, lymphocytes, and granulocytes were trained for 50 healthy donors based on DNAm measurements (ddPCR) compared to conventional cell counts (measured with Cell Dyn Emerald). An inverse correlation is shown for monocytes and granulocytes (hypomethylated CpGs) and a positive correlation for lymphocytes (hypermethylated CpGs). (B,C) These models were tested on 110 independent blood samples from healthy donors (B; also measured with Cell Dyn Emerald) and 150 patients (C; measured with Sysmex XS800i).Due to the systemic offset observed in epigenetic estimates from samples measured with the Sysmex XS800i, individual linear regression models for this instrument were retrained on 50 randomly selected samples. (D,E) ddPCR measurements were compared on 36 proficiency testing samples measured by approximately 400 laboratories using various conventional cell counters from different manufacturers. The epigenetic predictions trained on the Sysmex XS800i were compared with the average cell counts from the proficiency test (D). In addition, alternative models for epigenetic predictions trained on either Cell Dyn Emerald (green) or Sysmex XS800i (red) for each individual sample were compared (E). The Pearson correlation r and the mean absolute error (MAE) are shown for each cell type. Figure 3: Blood epigenetic values ​​correlate with flow cytometric measurements.(A) DNAm values ​​determined by ddPCR were correlated with the conventional blood count in blood samples from 50 randomly selected patients tested for granulocytes, monocytes, pan-lymphocytes (Sysmex XS800i) and CD4. + -T cells, CD8 +T cells, B cells, and NK cells (flow cytometry). A positive correlation is observed for lymphocytes (hypermethylated CpG) and a negative correlation for the other cell types (hypomethylated CpGs). (B) Linear regression models for each cell type, trained on these data, were then applied to an independent validation set (n = 100) to estimate cell counts. The Pearson correlation r and mean absolute error (MAE) are given for each leukocyte subset. Figure 4: Relative leukocyte counts for dried blood samples. (AC) Correlation of conventional cell counts (Sysmex XS800i and flow cytometry) with relative blood counts estimated by ddPCR for cryopreserved venous blood from 75 patients, which was either thawed and used directly (A) or dried overnight on either Mitra 30µl microsamplers (Neoterxy) (B) or Whatman protein saver cards (C).(D) Fresh capillary blood was collected from 31 donors and immediately counted using a cell counter (Cell Dyn Emerald) or dried on Whatman protein paper saver cards for 3–5 days prior to ddPCR analysis. The Pearson correlation r and mean absolute error (MAE) are shown for each cell type. Figure 5: Total leukocyte counts by ddPCR with a reference plasmid. (A) Workflow diagram for absolute leukocyte quantification. Venous blood, Whatman Protein Saver Card punches with dried blood, or Mitro microsampler tips with dried blood were mixed prior to DNA isolation and ddPCR with a reference plasmid. (B) Cell counts were estimated based on the ratio of genomic to plasmid DNA. (C) Combining estimates for relative leukocyte count with absolute cell count. The Pearson correlation between epigenetic estimates and conventional measurements is given.Figure 6: Principal component analysis based on the 27 cell type-specific CpGs. For each of the 9 cell types, 3 differentially methylated CpGs were identified. Principal component analysis of the DNAm data of these 27 CpGs showed that the cell types are clustered together, even though the data were derived from many different studies. Note that in PC1 / PC2, the nRBCs cluster very closely with the HSPCs, which is expected, as nRBCs also tend to resemble immature blood cells. HSPC: hematopoietic stem and progenitor cells; RBC: red blood cells. Figure 7: Evaluation of the cell type-specific CpGs. The behavior of the granulocyte-, lymphocyte-, and monocyte-specific CpG sites was evaluated in 92 healthy donors. The behavior of the CpG sites of lymphocyte subsets was evaluated in 21 patients. The Pearson correlation r is calculated for the DNAm values ​​compared to those obtained with automatic cell counters orCell counts determined by flow cytometry are given. Figure 8: Separation of methylated and unmethylated signals in the ddPCR analysis. Example results for ddPCR measurements. Individual droplets were scored as either methylated CpG (blue), unmethylated CpG (green), or both (orange). Based on this, the DNAm content is determined using the Fischer distribution. For absolute quantification, the droplets contain genomic DNA (blue), plasmid DNA (green), or both (orange). Figure 9: DNA methylation of genes with the top candidate CpGs. The candidate CpG sites were identified in DHODH, FAM169BP, MICAL2, TRPV1, CTLA4, CD8B, and MVD. This figure shows the DNAm values ​​(β values) measured with a 450k Illumina Beadchip for the sorted leukocytes from the publicly available dataset GSE35069.The genes are represented schematically according to Illumina annotations and show 1500 bp and 200 bp upstream of the transcription start site (TSS1500 and TSS200), the 5' and 3' untranslated regions (UTR), the first exon, and the gene body. Each dot represents a CpG site, with the red line indicating the selected CpG. Figure 10: Cell-type-specific DNAm is hardly reflected at the gene expression level. The single-cell transcriptome data available in the Human Protein Atlas (proteinatlas.org) show that most genes that have a cell-type-specific CpG site are expressed in different cell types. Only for CD8B (CpG for CD8). + -T cells) and CTLA4 (CpG for CD4 +-T- cells), a specific upregulation was observed in the corresponding cell types. DHODH (CpG granulocytes) was particularly expressed in hepatocytes. Figure 11: Accuracy of epigenetic estimates depending on age and gender. (A) The comparison between young (age ≤ 30 years, n = 24) and elderly patients (age ≥ 60 years, n = 58) showed that age had no significant influence on the accuracy of epigenetic cell type prediction. (B) No significant differences in prediction accuracy were found between female (n = 74) and male donors (n = 76). The data are presented as a whisker-box plot of the deviation between epigenetic estimates and conventional cell counts. The boxes indicate the first and third quartiles, the horizontal lines within the boxes the median, and the whiskers the 1.5-fold interquartile range. Figure 12: Summary of epigenetic estimates for individual cell types.To further assess the consistency of the epigenetic cell type predictions, it was estimated whether the proportions of the individual cell types deviated from the expected 100%. (A, B) For this purpose, the predictions for granulocytes, monocytes, and pan-lymphocytes (3 CpG score) for healthy donors (Cell Dyn Emerald model, A) and patients (Sysmex XN9000; B) were summed. The slightly higher deviation in patient samples could be due to hematological diseases. (C) Alternatively, the predictions for granulocytes, monocytes, CD4+ T cells, CD8 T cells, B cells, and NK cells (6 CpG score) were summed and compared with the patient samples. Figure 13: Absolute quantification based on genomic DNA copies. Digital droplet PCR is an absolute quantification method that provides an absolute number of DNA copies per reaction.In the context of the invention, a unique genomic region (R5), which was also alternatively used for the reference plasmid-based approach, was analyzed. While there is a clear correlation between the genomic copies of R5 detected by ddPCR and the conventional leukocyte count in venous blood, there are several outliers in the analysis of dried blood. The Pearson correlation r is given. Description of exemplary and preferred embodiments of the invention: DNAm profiles from 40 different studies were compiled to identify CG dinucleotides (CpGs) with cell type-specific DNAm using CimpleG. The DNAm concentrations at these CpGs were then measured using digital droplet PCR (ddPCR) in venous blood from 160 healthy donors and 150 patients with various hematological diseases.Deconvolution was further validated using venous blood (n=75) and capillary blood (n=31) dried either overnight (16 hours) or for up to 5 days on Whatman paper or on Mitra microsamplers. Differential blood counts are widely used to aid in the diagnosis of a variety of diseases or to monitor treatment. They are typically performed using either automated cell counters or flow cytometry (1, 2). Fresh blood must be drawn for this purpose, as cell counts within 24 hours are affected by loss of cellular integrity, coagulation, impaired electrical impedance, or poor antibody binding (3-5).In contrast, leukocyte counting based on dried blood spots would facilitate self-collection by finger prick without the need for skilled personnel, facilitate long-term storage and shipping of blood samples, and pose a significantly lower risk of bleeding compared to venipuncture (6). The deconvolution of leukocyte subsets is also possible by epigenetic means. DNA methylation (DNAm) patterns are consistently modulated in a cell-type-specific manner during differentiation (7). Epigenetic signatures that integrate DNAm values ​​at many CG dinucleotides (CpGs) can therefore be used to determine the composition of cell types in tissue (8, 9) or of leukocytes in blood (10-12). Epigenetic signatures for the deconvolution of hematopoietic subsets were initially developed for Illumina BeadChip data (10-12).However, since genome-wide DNAm profiling is hardly suitable for routine diagnostics, it is important to define targeted DNAm analysis of individual CpGs for clinical application (13). Many different methods can be used for this purpose, e.g., quantitative PCR (qPCR) (14), pyrosequencing (15), barcoded amplicon sequencing (16), or methylation-specific digital droplet PCR (ddPCR) (17). In particular, ddPCR can enable more reliable DNAm measurements because there is no PCR bias between methylated and unmethylated sequences (18). In addition to selecting a sensitive method, the selection of cell-type-specific candidate CpGs is crucial for the development of reliable targeted biomarkers. In a previous work, the inventors selected such genomic sites based on a dataset of sorted leukocyte subsets (15, 19).Although these signatures were validated in patient material, they did not show a very strong correlation with conventional blood counts for all leukocyte subsets (17). Therefore, the available Illumina BeadChip DNAm profiles of sorted leukocytes were compiled in the NCBI's Gene Expression Omnibus (GEO). Based on these profiles, new cell-type-specific CpGs were identified using a computer-assisted system called "CimpleG" (20). Epigenetic predictions based on ddPCR measurements at these CpGs significantly improved the accuracy of leukocyte counts in blood samples from healthy donors and patients with hematological diseases. Furthermore, the applicability to dried blood samples was demonstrated.Selection of differentially methylated sites The GEO database was searched for publicly available DNAm profiles (450k or EPIC Illumina BeadChip arrays) of sorted human hematopoietic cells (Supplementary Table S1). These profiles were subsequently analyzed using CimpleG (20). Blood samples All blood samples (cryopreserved venous blood from 160 healthy donors and 150 patients with hematological diseases, as well as fresh capillary blood from 31 healthy donors) were collected at the University Hospital of RWTH Aachen University after informed and written consent in accordance with the local ethics committee (EK041 / 15, EK206 / 09). For a subset of samples, dried blood spots were collected using two different microsamplers: Mitra® microsamplers (Neoteryx) and Whatman® Protein Saver cards (Cytiva). In addition, 36 cryopreserved blood samples were purchased from the Reference Institute for Bioanalytics (RfB, Bonn, Germany).Digital Droplet PCR: Primers were designed using the Bisulfite Primer Finder tool (Zymo Research), and fluorescent probes were manually designed with melting temperatures several degrees higher than those of the primers (Tables 2 and 3). Droplets containing 20 µl of reaction mixture containing 2x Supermix for probes (no dUTPs, Bio-Rad), 1 µM primer, 0.25 µM TaqMan probes (both Metabion), and 10–30 ng of bisulfite-converted DNA were generated using a QX200 droplet generator (Bio-Rad). The DNA was then amplified on a C1000 Touch Thermal Cycler (Bio-Rad) using the following program: 1 x 95°C for 10 min, 40 x 94°C for 30 s and 54°C for 30 s, 1 x 98°C for 10 min. The droplets were quantified using a QX200 Droplet Digital Reader and analyzed using QuantaSoft analysis Pro software (both Bio-Rad).Epigenetic Leukocyte Counts: To estimate relative leukocyte counts, 50 healthy donor or 50 patient samples were randomly selected to obtain individual linear regressions based on DNAm values ​​compared to conventional manual differential cell counts for the respective cell type. The remaining samples were used as independent validation sets. Absolute leukocyte counts were estimated for a subset of samples relative to the reference plasmid (17). Based on the detected plasmid copies and genomic copies, the absolute cell count was calculated as described below. CpG Selection: To identify leukocyte subset-specific CpG sites, the Gene Expression Omnibus (GEO) was searched for publicly available datasets. The following search term was used: "lymphocytes OR NK cells OR CD4." + OR CD8 +OR B cells OR monocytes OR granulocytes OR nucleated red blood cells OR CD34 +OR progenitor cells OR stem cells). The search was restricted to "Homo sapiens" as the organism. Datasets were filtered by the study types "Methylation Profiling by Array" or "Methylation Profiling by Genome Tiling Array," using either the Illumina HumanMethylation450 BeadChip (GPL16304 OR GPL13534) or the Infinium MethylationEPIC (GPL21145) platform. To enable better reanalysis, only studies that published their raw data in IDAT format were considered. This search was last performed on October 14, 2021, and yielded 150 datasets. Datasets containing samples from diseased patients, especially those with hematopoietic malignancies, were subsequently excluded from further analysis because they could influence the epigenetic profile (21, 22).In addition, a dataset containing a mixture of 450K and EPIC BeadChip data was excluded during the initial quality assessment because it was considered an outlier. This resulted in 40 candidate datasets containing 1303 DNAm profiles. These 1303 DNAm profiles were analyzed using the CimpleG pipeline (20). First, the data were processed using the R package minfi v1.30.0 (minfi, RRID:SCR_012830) and normalized using ssNoob, a within-sample normalization method. Subsequently, probes with poor mapping quality or those mapping to sex chromosomes were excluded. Finally, only CpGs present in both the 450K and EPIC platforms were retained. These remaining CpGs were examined for their cell type specificity based on a 10-fold stratified cross-validation.Within each fold or loop, each CpG is scored with a t-like statistic, and its performance on the classification task is measured using the area under the precision-recall (AUPR) curve. A final score for each best-performing probe was determined by combining the probe's average t-like statistic with the average training AUPR multiplied by the proportion of folds in which a CpG was selected as a top candidate. This ranking analysis for CpG selection was performed twice. The first analysis was used to identify CpGs with methylation specific to monocytes, granulocytes, and pan-lymphocytes. The second analysis was used to identify CpG sites for more specific blood cell subsets, including B cells and CD4. + T cells, CD8 + T cells, natural killer cells (NK), nucleated red blood cells (nRBC) and CD34 +Hematopoietic stem and progenitor cells (HSPC). Blood sample collection: The inventors analyzed cryopreserved venous blood samples (160 healthy donors and 150 patients with hematological diseases) that were stored at -80°C for up to five years prior to DNA isolation. These samples were collected at the RWTH Aachen University Hospital after informed and written consent in accordance with the local ethics committee of RWTH Aachen (EK041 / 15, EK206 / 09). For dried blood spot analysis, cryopreserved venous blood from 75 patients was thawed and subsequently dried to 30 µl using Mitra® microsamplers (Neoteryx, Torrance, USA) and Whatman® Protein Saver cards (Cytiva, Freiburg im Breisgau, Germany). In addition, fresh capillary blood was collected from 31 patients using a Microvette® (Sartstedt, Nümbrecht, Germany), which was immediately dried on Whatman® Protein Saver cards and analyzed up to 5 days later.At the time of blood collection, conventional leukocyte counts were performed immediately using either a Cell Dyn Emerald (Abbott, Wiesbaden, Germany) or a Sysmex XS800i (Sysmex, Norderstedt, Germany) in combination with flow cytometry, as indicated. DNA Isolation and Bisulfite Treatment: DNA was isolated from cryopreserved venous blood either directly from 150 µl using the QIAamp DNA Mini Kit (Qiagen, Hilden, Germany) in 200 µl of elution buffer, or alternatively, 30 µl was dried for at least 16 hours (overnight) on either Mitra® microsampling devices or Whatman® Protein Saver cards. DNA from these dried samples was isolated using the QIAamp DNA micro kit (Qiagen, Hilden, Germany). For this purpose, the 30 µl Mitra® tip was fragmented, and three areas (each with a diameter of 3 mm) were taken from the Whatman® Protein Saver cards using punch knives.For capillary blood, the volume applied to the Whatman paper varied, but DNA was always isolated from three areas (each 3 mm in diameter) using the QIAamp DNA micro kit (Qiagen, Hilden, Germany). The final elution volume for DNA from dried blood was 30 µl. For methylation analysis, the isolated DNA (50-500 ng) was bisulfited using the EZ DNA Methylation Kit (Zymo research) according to the manufacturer's instructions and eluted in a total volume of 20 µl. Additional information for digital droplet PCR: For relative leukocyte counts, three primer pairs were initially designed for each CpG and tested on a temperature gradient from 50°C to 60°C using a mixture of DNA from 10 patients. The best primer pairs performed equally well at 54°C, so they can be combined in one assay.Absolute quantification of leukocyte count was performed using a reference plasmid containing a genomic region unique and conserved in the human genome, altered at three positions (Table 5), as previously described (17). The sample (150 µl of blood, 3x3 mm Whatman paper, or the fragmented Mitra tip) is mixed with 15 µl of plasmid (corresponding to 0.00389 ng or 580,908 copies), followed by proteinase K and lysis buffer, and vortexed vigorously for 15 seconds. The isolation protocol is then performed according to the manufacturer's instructions.In ddPCR, the number of copies of genomic DNA and reference DNA is measured, and the absolute cell counts can then be calculated using the following formula: Cg × 580908 ^^^^^ / µl = × ^^ 2 × Cr × v where Cg is the measured copy number of genomic DNA (copies / µl), Cr is the measured copy number of the reference plasmid (copies / µl), and v is the starting volume (150µl for venous blood and 30µl for dry blood). The "2" is used to correct for the fact that each cell contains two copies of the genomic DNA region of interest.To account for plasmid DNA contamination (bacterial genome and other plasmids), technical variations in pipetting, DNA quantification during reference plasmid preparation, DNA isolation efficiency, and changes in the initial volume of dried blood, an empirically determined correction factor cf was included (for venous blood cf = 0.07; venous blood dried on Whatman cf = 5; venous blood dried on Mitra cf = 8; and capillary blood dried on Whatman cf = 10). Data Analysis and Statistics Scatterplots and principal component analysis were generated in R. The heatmap was generated using heatmapper (23). The Pearson correlation coefficient r and the mean absolute error (MAE) of the DNAm were analyzed and plotted in Windows Office 2016 Excel (Microsoft) and GraphPad Prism version 9 (Graph Pad Software Inc.).Identification of cell-type-specific methylation sites To select the best candidate CpGs for cell-type-specific DNAm, a dataset of 1303 DNAm profiles from 40 different studies was compiled (Figure 1A; Supplementary Table S1). Using CimpleG, the top three CpGs for granulocytes, monocytes, lymphocytes, CD4 T cells, CD8 T cells, B cells, NK cells, hematopoietic stem and progenitor cells (HSPCs), and nucleated red blood cells (nRBCs; Figure 1B) were selected. The selected CpGs showed cell-type-specific hypomethylation, with the exception of pan-lymphocytes and HSPCs, which were hypermethylated. Heatmap analysis also confirmed the consistency of the cell-type-specific DNAm differences (Figure 1C). Furthermore, principal component analysis using these CpG sites revealed distinct clusters for each cell type (Figure 6).Thus, the selected candidate CpGs appeared suitable for clearly distinguishing the respective cell types across many different DNAm datasets. Targeted DNA methylation analysis using digital droplet PCR. Subsequently, ddPCR assays were developed for targeted DNAm analysis at the selected CpGs. The CpGs for HSPCs and nRBCs were not considered for this analysis, as these cell types are generally rarely present in venous blood. On the other hand, ddPCR assays were also designed for cell-type-specific CpGs, which had previously been selected based on DNAm profiles from a single study (15, 17). Since the CD8A-associated cg25939861 overlapped in both selections for CD8 T cells, the inventors analyzed 27 CpGs. An initial screening was performed on 92 venous blood samples from healthy donors to select the best CpGs for granulocytes, monocytes, and pan-lymphocytes.For CD4 T cells, CD8 T cells, B cells, and NK cells, lymphocytes from 21 patients were further stratified by flow cytometry (Figure 7). Overall, the correlation of DNAm values ​​with the conventional blood count was very high (r > 0.8 or r < -0.8). The best-performing CpG candidates (Table 1) were selected based on these correlations and, if performance was similar, based on their rank in the CimpleG analysis or how well the positive ddPCR drops could be distinguished (Figure 8). The selected CpGs were located either in the 1500 bp region upstream of the transcription start site (DHODH), in the 5'UTR region (FAM169BP and MICAL2), in the first exon (CTLA4), or in the gene body (CD8B, TRPV1, and MVD; Figure 9). Analysis of gene expression in the human protein atlas showed that, with the exception of CTLA4 and CD8B, the corresponding genes do not show characteristic upregulation in the respective cell types (Figure 10) (24).Therefore, different DNAm in these epigenetic biomarkers is not necessarily reflected at the level of gene expression. Epigenetic estimation of granulocytes, monocytes, and lymphocytes For epigenetic estimation of the granulocyte, monocyte, and pan-lymphocyte fraction, 50 samples from healthy donors were randomly selected to train individual linear regression models for each cell type (Figure 2A, Table 4). These models could accurately predict the number of granulocytes (r = 0.81; mean absolute error [MAE] = 3.7) and pan-lymphocytes (r = 0.82; MAE = 3.4) and moderately predict the number of monocytes (r = 0.40; MAE = 2.1) in 110 independent donors measured with the same Cell Dyn Emerald cell counter (Figure 2B).However, when these predictors were tested on 150 patient samples analyzed in another department using a Sysmex XS800i, a systematic underestimation of granulocytes (MAE = 9.5) and a systematic overestimation of lymphocytes (MAE = 8.0) and monocytes (MAE = 3.7) was observed (Figure 2C). However, the very high correlations with conventional counts (r = 0.95, r = 0.96, and r = 0.86, respectively) suggest that this discrepancy is due to the reference measurements using different cell counters. In particular, the CpG for monocytes showed a better correlation with the automatic cell counts in the patient samples, suggesting that these measurements might be more precise. Therefore, the models were retrained on a randomly selected set of 50 patient samples measured using a Sysmex XS800i (Table 4).To further compare these assays, the inventors participated in a collaborative study with more than 400 other institutes using various conventional automated cell counters. The ddPCR measurements clearly correlated with the average total cell count (Figure 2D). However, the collaborative study results showed a large variation between the different instruments (Figure 2E). This demonstrates the importance of precise reference measurements for fine-tuning the linear regression models for ddPCR. While ddPCR measurements are robust, predictions trained on Cell Dyn measurements (green) overestimated lymphocytes and monocytes, whereas predictions trained on Sysmex XS800i measurements (red) tended to overestimate granulocytes.Epigenetic blood count correlates with flow cytometry in patient samples To compare the performance of CpGs specific for CD4 T cells, CD8 T cells, B cells, and NK cells, blood samples from 150 patients with various hematological diseases were analyzed. These blood samples were analyzed for granulocytes, monocytes, and pan-lymphocytes using a Sysmex XS800i, and the lymphocytes were further stratified using flow cytometry. As mentioned above, a subset of 50 patients was randomly selected for individual linear regression models for each of these cell types (Figure 3A; Table 4). Using these models for the independent set of 100 patient samples, the proportion of granulocytes (r = 0.95, MAE = 3.2), pan-lymphocytes (r = 0.97, MAE = 2.2), and pan-lymphocytes (r = 0.98, MAE = 2.2) could be determined.5), monocytes (r = 0.82; MAE = 1.9), CD4 T cells (r = 0.84; MAE = 2.5), CD8 T cells (r = 0.94; MAE = 1.7), B cells (r = 0.96; MAE = 1.2), and NK cells (r = 0.72; MAE = 1.4) (Figure 3B). The accuracy of the predictions was not significantly affected by age or sex (Figure 11). To further evaluate the consistency of the estimates independent of the conventional cell counters, it was determined whether the predictions for the individual cell types added up to 100% (Figure 12). The estimated proportions of granulocytes, monocytes, and pan-lymphocytes in patients totaled 99.3 ± 5.2%, which corresponded to the sum in healthy donors (100.1 ± 2.6%), and for both groups, the sum did not differ significantly from the expected 100% (p = 0.12 and p = 0.78, respectively).The sum also did not differ significantly from 100% (99.2 ± 6.8%; p = 0.18) when the pan-lymphocytes were further stratified into B cells, CD4 T cells, CD8 T cells, and NK cells. Thus, the epigenetic estimates are robust and can also provide consistent results for patient samples. Evaluation of leukocyte fractions from dried blood. Next, the feasibility of epigenetic blood counting in dried blood samples was investigated. For this purpose, 30 µl of venous blood from 75 patients was pipetted and dried overnight on two alternative sampling devices: Whatman Protein Saver cards and Mitra 30 µl microsamplers. Notably, the predictions of individual cell fractions showed a similar precision to that observed for 150 µl of venous blood, particularly when the blood was dried on Whatman Protein Saver cards, slightly outperforming the Mitra microsamplers (Figure 4A-C).Next, the applicability was tested for capillary blood collected from 31 healthy donors via fingerstick. A fresh sample was measured using a Cell Dyn Emerald device, and in parallel, blood spots were dried on Whatman Protein Saver cards for up to five days prior to epigenetic blood counting. Good correlations were observed, particularly for granulocytes (r = 0.79, MAE = 5.1) and pan-lymphocytes (r = 0.80, MAE = 4.6) (Figure 4D). A lower correlation was observed for monocytes (r = 0.16, MAE = 2.0), which could be due to the inconsistency of the reference counts on the Cell Dyn Emerald device, as also mentioned above. Overall, the results demonstrate that the inventive method is suitable for dried capillary blood collected via fingerstick.Absolute quantification of leukocyte count. Since each leukocyte possesses two DNA copies, the absolute cell count is expected to correlate with the DNA concentration. This can be estimated from the number of copies of a specific genomic region detected by ddPCR in non-biosulfite-converted DNA. Indeed, for DNA isolated from 150 µl of venous blood, the number of positive drops in ddPCR for the genomic region "R5" correlated with the leukocyte count (r = 0.95; Figure 13). However, several outliers were observed in dried blood, particularly for blood dried on Mitra microsamplers (r = 0.58).A direct comparison of positive ddPCR events in 30 µl of fresh blood compared to 30 µl of the same sample dried overnight revealed that significantly fewer copies were detected by both Whatman Protein Saver cards (67.7 ± 21.4%) and Mitra microsamplers (33.5 ± 27.6%), thus demonstrating lower and more variable DNA isolation efficiency in dried blood spots. Therefore, prior to DNA isolation, samples were spiked with a known amount of a reference plasmid, as previously described (Figure 5A)(17). Cell count can then be estimated using the ratio of genomic to plasmid DNA copies. This approach provides a higher correlation with conventional leukocyte counts of venous blood dried on either Whatman Protein Saver cards (r = 0.86) or Mitra microsamplers (r = 0.84) (Figure 5B). This was also true for capillary blood dried on Whatman paper (r = 0.80).By combining estimates of leukocyte counts with relative counts based on cell-type-specific DNAm, the absolute counts for each of the leukocyte subsets could be calculated. For each cell type, a clear correlation between the conventional absolute cell counts and the epigenetic estimates was observed (Figure 5C). Summary: Using the novel method of the invention, epigenetic leukocyte counts can be further improved through a comprehensive selection of cell-type-specific candidate CpGs and optimized ddPCR assays. The results exceeded the precision of previous models based on various CpGs and pyrosequencing measurements (17). Furthermore, it was shown that dried blood spots obtained by finger prick can be used for epigenetic leukocyte counts according to the invention.For the selection of candidate CpGs, a very large dataset of DNAm profiles was compiled (1303 cell-type-specific DNAm profiles from 726 donors, derived from 40 different studies). In contrast, the previous selection of cell-type-specific CpGs was based on DNAm profiles from only six donors (19). In another study, Baron et al. initially focused on functionally relevant genes to select CpG candidates, particularly for CD4 T cells (CD4 gene) and CD8 T cells (CD8B gene) (14). Notably, despite the very different selection approaches, a candidate CpG for CD8 T cells (cg04329870) was also found in this CD8B region. They also analyzed sorted cells from two donors to identify cell-type-specific CpGs for B cells, NK cells, and granulocytes (14). Remarkably, a CpG candidate for NK cells (cg05355684) was also selected in their work.More recently, the inventors have identified cell type-specific CpGs for HSPCs based on DNAm profiles of CD34. + Cells from five donors were identified (25) compared to DNAm profiles of other sorted leukocyte subsets from six donors (19). On this basis, the most important candidate CpGs for HSPCs were identified (26). Although the HPSCs were not validated in the current study, since CD34 + cells in normal peripheral blood, there was a striking overlap of two of the three top CpGs in the different selections: SP140 (cg17607231) and CD48 (cg13311440) were also identified in this independent selection and correlated with CD34 +Cells in mobilized peripheral blood and blast counts in leukemia (26). The finding that different studies select the same genomic region or even exactly the same CpGs suggests that genomic regions with well-suited candidate CpGs for cell-type-specific DNAm might be limited. For a more systematic selection of the top candidate CpGs, the new CimpleG pipeline was used (20). DNA from non-leukocyte cell types was not considered because the inventive application focuses specifically on the analysis of blood. While it is theoretically possible that skin cells, endothelial cells, or even cell-free DNA could influence the predictions (8, 9, 27), the vast majority of DNA in blood clearly originates from leukocytes. It is noteworthy that some of the selected CpGs did not show a clear correlation with cell number in the ddPCR measurements (PPM1F and TOP1MT for monocytes, LCN8 for B cells, and PLXND1 for NK cells).Thus, either the probe sets of the Illumina BeadChip array or the corresponding ddPCR assays did not provide specific or accurate DNAm measurements. Such discrepancies between arrays and targeted assays have been reported previously (28, 29) and underscore the need to validate such biomarkers. Several methods exist for targeted DNAm measurements (18). In previous work, epigenetic blood counts were mainly based on pyrosequencing (17, 19), whereas Baron et al. performed their assays based on methylation-specific qPCR (14). It has been shown that DNAm concentrations can vary considerably between pyrosequencing and qPCR measurements (30). The improved precision of epigenetic blood counts in the current study is likely due in part to the use of ddPCR.In contrast to qPCR, ddPCR has fewer PCR biases for methylated or unmethylated sequences, as individual droplets are only classified as positive or negative (31). Indeed, for epigenetic estimations of CD3 T lymphocytes, ddPCR has been shown to offer higher precision and greater accuracy, especially for samples with low copy numbers of target genes (32, 33). A major advantage of epigenetic blood counts is that blood samples can be collected with a simple finger prick and shipped as dried blood spots. This facilitates self-testing without trained medical personnel—e.g., for elderly patients who have difficulty consulting a doctor. Furthermore, the method is suitable for newborn screening for severe combined immunodeficiencies (14, 34).Other clinically relevant examples include CD4 T cell counts during HIV infection (14) or monitoring neutropenia after chemotherapy. Two alternative dried blood sampling devices were tested here. Filter paper, such as Whatman Protein Saver cards, are commonly used and less expensive, whereas microsamplers, such as Mitra microsamplers, can sample a more defined blood volume. Surprisingly, in the experiments conducted, epigenetic cell counts were more precise with Whatman Protein Saver cards, which may be due to less efficient DNA isolation from the Mitra tips. Although the dried blood estimates are promising, there are still limitations and challenges that need to be addressed before the method can be used clinically.1) One limitation is that red blood cells and platelets cannot be considered because they do not contain genomic DNA. 2) A particular challenge for epigenetic blood counts is the detection of rare cell types (<5%) or closely related cell types, as their differential DNA methylation has only a minor impact on the mean methylation levels in bulk DNA. 3) The method is relatively time-consuming. With the current protocol, the turnaround time is approximately 2 days because the bisulfite conversion occurs overnight (12–16 hours). However, with alternative bisulfite conversion methods that are directly applicable to blood, the entire workflow can be reduced to approximately 8 hours. 4) For absolute quantification, especially with dried blood, the efficiency of DNA extraction must be improved.Here, the variability in DNA isolation efficiency was addressed using a reference plasmid, but its stability at a defined concentration still needs to be further validated in independent cohorts.5) Although no significant influence on epigenetic blood levels was observed in healthy cohorts and patients, epigenetic composition can be influenced by specific diseases—particularly leukemia. (16) Furthermore, variations may occur between samples from children and adults. Therefore, the method must be specifically validated for specific clinical applications.6) Last but not least, regulatory hurdles must be overcome. The biomarkers must be validated and accredited according to local regulatory requirements—in Europe, according to the new directive for in vitro diagnostic medical devices (35).In general, it should be feasible to meet these demanding requirements, as similar procedures are clinically approved, the instruments for DNA isolation and ddPCR are already CE-certified, and the required analysis software, which in this case is based on relatively simple linear regression models, has also been certified accordingly. Despite these limitations and challenges, this study has demonstrated that epigenetic leukocyte counting by ddPCR is an accurate alternative to cryopreserved blood or dried fingerstick blood. Furthermore, a recent study, regarding possible discrepancies between samples from children and adults, demonstrated that the method of the invention can also be reliably applied to blood samples from children and is thus also suitable for pediatric applications (36).

[0002] Table 3: Other primers and probes used for ddPCR (see also Table 2). Gene, cg… Primer / Probe Sequence, SEQ ID NO:[…] PPM1F Forward Primer GTTTAGTTTAGATTTTGTTTAGGAAAGG

[0061] ] Q1 AT cg17117981 Reverse primer AAACAAAAATTAATCAAACACC

[0066] 6-FAM- probe not methylated TGTGAGGTTTTTTTGATTGTTTTTTTTT - BHQ1

[0067] HEX - TGTGAGGTTCGTTTTTGATTGTTTG - probe methylated BHQ1

[0068] LRP5 Forward primer GTTTTTTAGGTTGTAGGTGTTTATTG

[0069] ] cg14482811 Reverse primer CACCCCACATAACACTTATTTT

[0074] Probe not 6-FAM- AGTGTTGGAGGTTTTGGTTTGTTG - methylated BHQ1

[0075] HEX - TGTTGGAGGTTTTGGTTCGTTGT - Probe methylated BHQ1

[0076] CD8A Forward primer TGATGATGGTTAGATTTGGGG

[0077] GTT T - cg04329870 Reverse primer AACTCCTTCAAACCCTAAACC

[0082] Probe not 6-FAM- TTTAGTTATGGAGGTTTTTGAGTGT - methylated BHQ1

[0083] HEX - TAGTTATGGAGGTTTTCGAGTGT -BHQ1 Probe methylated

[0084] cg21333217 Reverse primer ACTAACCTTACCCAACTACAA

[0090] 6-FAM- probe not methylated TGTTTTTTAGTTATTGTGTAGGTTTTTAT - BHQ1

[0091] HEX - TTTTTAGTTATCGCGTAGGTTTT -BHQ1 probe methylated

[0092] AG R5-Reference Forward primer 5'-GTTTCCACACACAGAGGAAGTAG-3'

[0097] Reverse primer 5'-ACGGGATGAGGAGATTTTAAGCC-3'

[0098] Probe 5' 6-FAM- TGGCTGGCAGAGACCCAGGA - 3' genomic DNA BHQ-1

[0099] Probe plasmid- 5' HEX - TGGCTGCGAGAGACCGAGGA - 3' DNA BHQ-1

[0100]

[0003] Table 4: Simple linear regression models for epigenetic estimates of relative leukocyte count Cell Dyn Emerald - Sysmex XS800i measurements in measurements from healthy patients with hematological cell type donor diseases Granulocytes 82.0817 - 0.9072*DNAm 98.7825 - 1.2030*DNAm Lymphocytes 10.6012 + 1.0541*DNAm 1.0172 + 1.1007*DNAm CD4-T-KA 58.2602 - 0.6056*DNAm Cells B cells KA 45.4241 - 0.4825*DNAm DNAm: DNA methylation, expressed as a percentage (0-100) Table 5: Sequence in the reference plasmid for absolute leukocyte quantification by ddPCR Cloned sequence AGTACCGAATTCGCTGAGTTTCCACACAGGAAGTAGGTGGAAAATTTGACAA AACTCTGGAGATGGCTGCGAGACCGAGGAAAAATTAAATTATCTCCAGATTTG AGTTTTTCTGGTAGAGGCTTAAAATCTCCTCATCCCGTTGCATGGATCCCCAT GA [SEQ ID NO: 101] References: 1. Pitoiset F, Cassard L, El Soufi K, Boselli L, Grivel J, Roux A, et al. Deep phenotyping of immune cell populations by optimized and standardized flow cytometry analyses. Cytometry A 2018;93:793-802. 2. Bruegel M, Nagel D, Funk M, Fuhrmann P, Zander J, Teupser D. Comparison of five automated hematology analyzers in a university hospital setting: Abbott Cell-Dyn Sapphire, Beckman Coulter DxH 800, Siemens Advia 2120i, Sysmex XE-5000, and Sysmex XN-2000. Clin Chem Lab Med 2015;53:1057-71. 3. Navas A, Giraldo-Parra L, Prieto MD, Cabrera J, Gomez MA.Phenotypic and functional stability of leukocytes from human peripheral blood samples: considerations for the design of immunological studies. BMC Immunol 2019;20:5. 4. McGann LE, Yang HY, Walterson M. Manifestations of cell damage after freezing and thawing. Cryobiology 1988;25:178-85. 5. Herzenberg LA, Tung J, Moore WA, Herzenberg LA, Parks DR. Interpreting flow cytometry data: a guide for the perplexed. Nat Immunol 2006;7:681-5. 6. Jimenez Vera E, Chew YV, Nicholson L, Burns H, Anderson P, Chen HT, et al. Standardisation of flow cytometry for whole blood immunophenotyping of islet transplant and transplant clinical trial recipients. PLoS One 2019;14:e0217163. 7. Mattei AL, Bailly N, Meissner A. DNA methylation: a historical perspective. Trends Genet 2022;38:676-707. 8. Moss J, Magenheim J, Neiman D, Zemmour H, Loyfer N, Korach A, et al. Comprehensive human cell-type methylation atlas reveals origins of circulating cell-free DNA in health and disease. Nat Commun 2018;9:5068. 9.Schmidt M, Maie T, Dahl E, Costa IG, Wagner W. Deconvolution of cellular subsets in human tissue based on targeted DNA methylation analysis at individual CpG sites. BMC Biol 2020;18:178. 10. Houseman EA, Accomando WP, Koestler DC, Christensen BC, Marsit CJ, Nelson HH, et al. DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinformatics 2012;13:86. 11. Accomando WP, Wiencke JK, Houseman EA, Nelson HH, Kelsey KT. Quantitative reconstruction of leukocyte subsets using DNA methylation. Genome Biol 2014;15:R50. 12. Salas LA, Koestler DC, Butler RA, Hansen HM, Wiencke JK, Kelsey KT, Christensen BC. An optimized library for reference-based deconvolution of whole-blood biospecimens assayed using the Illumina HumanMethylationEPIC BeadArray. Genome Biol 2018;19:64. 13. Wagner W. How to Translate DNA Methylation Biomarkers Into Clinical Practice. Front Cell Dev Biol 2022;10:854797. 14. Baron U, Werner J, Schildknecht K, Schulze JJ, Mulu A, Liebert UG, et al.Epigenetic immune cell counting in human blood samples for immunodiagnostics. Sci Transl Med 2018;10. 15. Frobel J, Bozic T, Lenz M, Uciechowski P, Han Y, Herwartz R, et al. Leukocyte Counts Based on DNA Methylation at Individual Cytosines. Clin Chem 2018;64:566-75. 16. Bozic T, Kuo CC, Hapala J, Franzen J, Eipel M, Platzbecker U, et al. Investigation of measurable residual disease in acute myeloid leukemia by DNA methylation patterns. Leukemia 2022;36:80-9. 17. Sontag S, Bocova L, Hubens WHG, Nuchtern S, Schnitker M, Look T, et al. Toward Clinical Application of Leukocyte Counts Based on Targeted DNA Methylation Analysis. Clin Chem 2022;68:646-56. 18. Han Y, Franzen J, Stiehl T, Gobs M, Kuo CC, Nikolic M, et al. New targeted approaches for epigenetic age predictions. BMC Biol 2020;18:71. 19. Reinius LE, Acevedo N, Joerink M, Pershagen G, Dahlen SE, Greco D, et al.Differential DNA methylation in purified human blood cells: implications for cell lineage and studies on disease susceptibility. PLoS One 2012;7:e41361. 20. Maié T, Schmidt M, Erz M, Wagner W, Costa IG. CimpleG: Finding simple CpG methylation signatures. bioRxiv 2022:2022.09.12.507513. 21. Robertson KD. DNA methylation and human disease. Nat Rev Genet 2005;6:597-610. 22. Ehrlich M. DNA hypermethylation in disease: mechanisms and clinical relevance. Epigenetics 2019;14:1141-63. 23. Babicki S, Arndt D, Marcu A, Liang Y, Grant JR, Maciejewski A, Wishart DS. Heatmapper: web-enabled heat mapping for all. Nucleic Acids Res 2016;44:W147-53. 24. Karlsson M, Zhang C, Mear L, Zhong W, Digre A, Katona B, et al. A single- cell type transcriptomics map of human tissues. Sci Adv 2021;7. 25. Aranyi T, Stockholm D, Yao R, Poinsignon C, Wiart T, Corre G, et al. Systemic epigenetic response to recombinant lentiviral vectors independent of proviral integration. Epigenetics Chromatin 2016;9:29. 26.Bocova L, Hubens W, Engel C, Koschmieder S, Jost E, Wagner W. Quantification of hematopoietic stem and progenitor cells by targeted DNA methylation analysis. Clin Epigenetics 2023;15:105. 27. Neuberger EWI, Sontag S, Brahmer A, Philippi KFA, Radsak MP, Wagner W, Simon P. Physical activity specifically evokes release of cell-free DNA from granulocytes thereby affecting liquid biopsy. Clin Epigenetics 2022;14:29. 28. Cheung K, Burgers MJ, Young DA, Cockell S, Reynard LN. Correlation of Infinium HumanMethylation450K and MethylationEPIC BeadChip arrays in cartilage. Epigenetics 2020;15:594-603. 29. Roessler J, Ammerpohl O, Gutwein J, Hasemeier B, Anwar SL, Kreipe H, Lehmann U. Quantitative cross-validation and content analysis of the 450k DNA methylation array from Illumina, Inc. BMC Res Notes 2012;5:210. 30. De Chiara L, Leiro-Fernandez V, Rodriguez-Girondo M, Valverde D, Botana- Rial MI, Fernandez-Villar A.Comparison of Bisulfite Pyrosequencing and Methylation-Specific qPCR for Methylation Assessment. Int J Mol Sci 2020;21. 31. Warnecke PM, Stirzaker C, Melki JR, Millar DS, Paul CL, Clark SJ. Detection and measurement of PCR bias in quantitative methylation analysis of bisulphite-treated DNA. Nucleic Acids Res 1997;25:4422-6. 32. Wiencke JK, Bracci PM, Hsuang G, Zheng S, Hansen H, Wrensch MR, et al. A comparison of DNA methylation specific droplet digital PCR (ddPCR) and real time qPCR with flow cytometry in characterizing human T cells in peripheral blood. Epigenetics 2014;9:1360-5. 33. Malic L, Daoud J, Geissler M, Boutin A, Lukic L, Janta M, et al. Epigenetic subtyping of white blood cells using a thermoplastic elastomer-based microfluidic emulsification device for multiplexed, methylation-specific digital droplet PCR. Analyst 2019;144:6541-53. 34. Blom M, Pico-Knijnenburg I, Imholz S, Vissers L, Schulze J, Werner J, et al.Second Tier Testing to Reduce the Number of Non-actionable Secondary Findings and False-Positive Referrals in Newborn Screening for Severe Combined Immunodeficiency. J Clin Immunol 2021;41:1762-73. 35. Lubbers BR, Schilhabel A, Cobbaert CM, Gonzalez D, Dombrink I, Bruggemann M, et al. The New EU Regulation on In Vitro Diagnostic Medical Devices: Implications and Preparatory Actions for Diagnostic Laboratories. Hemasphere 2021;5:e568. 36. Wouter Hubens, Lara Kluge, Alexander Seitz, Eva Verjans, Lothar Rink, Wolfgang Wagner, Epigenetic Leukocyte Counts from Dried Blood Spots of Pediatric Patients, Clinical Chemistry, Volume 70, Issue 7, July 2024, Pages 997–999, https: / / doi.org / 10.1093 / clinchem / hvae066.

Claims

RWTH Aachen 23521WO Patent claims 1. Method for determining at least part of the cellular composition of at least one biological sample comprising leukocytes, the method comprising the following steps: - identifying DNA methylation levels of at least one region of the genomic DNA in the sample, wherein the specific region comprises at least one CpG dinucleotide, and wherein the specific region comprises at least one nucleotide sequence selected from the group consisting of cg22381196 (DHODH), cg12483340 (TMEM87A), cg06270401 (DYRK4), cg23054181 (FAM169BP), cg26076724 (RP11- 146I2.2), cg12249234 (KSR1), cg04468741 (MICAL2), cg05074138 (CTLA4), cg05705140 (LINC01237), cg15564619 (SKI), cg11531557 (HMBOX1), cg02212339 (TRPV1), cg22488278 (ZFYVE28), cg02240030 (MAD1L1) and cg24408769 (JARID2); and - determining the proportion of at least one leukocyte type based on the identified methylation levels in order to predict the relative distribution of that leukocyte type in the sample, wherein - the methylation level of at least one CpG dinucleotide selected from the group consisting of CpG dinucleotides cg22381196 (DHODH), cg12483340 (TMEM87A), cg06270401 (DYRK4) and at least one CpG dinucleotide located within a region of 1,000 nucleotides upstream and / or downstream of each of said CpG dinucleotides indicates the proportion of granulocytes, - the methylation level of the CpG dinucleotide cg04468741 (MICAL2) or at least one CpG dinucleotide located within a region of 1,000 nucleotides upstream and / or downstream of each of said CpG dinucleotides indicates the proportion of granulocytes.000 nucleotides upstream and / or downstream of the said CpG dinucleotide, indicates the proportion of monocytes, - the methylation level of at least one CpG dinucleotide selected from the group consisting of CpG dinucleotides cg23054181 (FAM169BP), cg26076724 (RP11-146I2.2), cg12249234 (KSR1) and. 2 at least one CpG dinucleotide located within a region of 1,000 nucleotides upstream and / or downstream of each of said CpG dinucleotides indicates the proportion of lymphocytes, - the methylation level of the CpG dinucleotide cg02212339 (TRPV1) or at least one CpG dinucleotide located within a region of 1,000 nucleotides upstream and / or downstream of said CpG dinucleotide indicates the proportion of B cells, - the methylation level of at least one CpG dinucleotide selected from the group consisting of the dinucleotides cg05074138 (CTLA4), cg15564619 (SKI), cg05705140 (LINC01237) and at least one CpG dinucleotide located within a region of 1,000 nucleotides upstream and / or downstream of each of said CpG dinucleotides, is selected to increase the proportion of CD4 +- T cells, - and the methylation level of the CpG dinucleotide cg11531557 (HMBOX1) or of at least one CpG dinucleotide located within a region of 1,000 nucleotides upstream and / or downstream of said CpG dinucleotide indicates the proportion of CD8 +- indicates T cells, and - the methylation level of at least one CpG dinucleotide selected from the group consisting of the dinucleotides cg22488278 (ZFYVE28), cg02240030 (MAD1L1), cg24408769 (JARID2) and at least one CpG dinucleotide located within a region of 1,000 nucleotides upstream and / or downstream of each of said CpG dinucleotides indicates the proportion of NK cells.

2. Method according to claim 1, characterized in that at least one CpG dinucleotide is located within a region of 500, 400, 300, 200 or 100 nucleotides upstream and / or downstream of each of said CpG dinucleotides cg22381196, cg12483340, cg06270401, cg23054181, cg26076724, cg12249234, cg04468741, cg05074138, 3 cg05705140, cg15564619, cg11531557, cg02212339, cg22488278, cg02240030 and cg24408769.

3. The method according to claim 1 or 2, characterized in that the sample is provided with a volume of at most or less than 50 µl, at most or less than 40 µl, at most or less than 30 µl, at most or less than 20 µl, at most or less than 10 µl, 10-40 µl, 10-30 µl, 10-20 µl, 20-40 µl, 30-40 µl, 20-30 µl, 10-20 µl, 25-35 µl, or preferably 30 µl.

4. The method according to one of the preceding claims, characterized in that the biological sample comprises blood, bone marrow, cerebrospinal fluid, and / or another leukocyte-containing tissue.

5. Use of at least one artificial nucleic acid molecule for determining at least part of the cellular composition of a biological sample comprising leukocytes according to the method according to one of claims 1 to 4,in which the nucleic acid molecule comprises at least one nucleotide sequence selected from the group consisting of: a) a nucleotide sequence selected from the group consisting of SEQ ID NO: 1, SEQ ID NO: 2, SEQ ID NO: 3 and SEQ ID NO: 4 for determining the proportion of granulocytes; b) a nucleotide sequence selected from the group consisting of SEQ ID NO: 5, SEQ ID NO: 6, SEQ ID NO: 7 and SEQ ID NO: 8 for determining the proportion of granulocytes; c) a nucleotide sequence selected from the group consisting of SEQ ID NO: 9, SEQ ID NO: 10, SEQ ID NO: 11 and SEQ ID NO: 12 for determining the proportion of granulocytes; d) a nucleotide sequence selected from the group consisting of SEQ ID NO: 13, SEQ ID NO: 14, SEQ ID NO: 15 and SEQ ID NO: 16 for determining the proportion of monocytes; e) a nucleotide sequence selected from the group consisting of SEQ ID NO: 17, SEQ ID NO: 18, SEQ ID NO: 19 and SEQ ID NO: 20, 4, for determining the proportion of lymphocytes; f) a nucleotide sequence selected from the group consisting of SEQ ID NO: 21, SEQ ID NO: 22, SEQ ID NO: 23 and SEQ ID NO: 24, for determining the proportion of lymphocytes; g) a nucleotide sequence selected from the group consisting of SEQ ID NO: 25, SEQ ID NO: 26, SEQ ID NO: 27 and SEQ ID NO: 28, for determining the proportion of lymphocytes; h) a nucleotide sequence selected from the group consisting of SEQ ID NO: 29, SEQ ID NO: 30, SEQ ID NO: 31 and SEQ ID NO: 32, for determining the proportion of B cells; i) a nucleotide sequence selected from the group consisting of SEQ ID NO: 33, SEQ ID NO: 34, SEQ ID NO: 35 and SEQ ID NO: 36 for determining the proportion of CD4 + - T cells; j) a nucleotide sequence selected from the group consisting of SEQ ID NO: 37, SEQ ID NO: 38, SEQ ID NO: 39 and SEQ ID NO: 40 for determining the proportion of CD4 +- T cells; k) a nucleotide sequence selected from the group consisting of SEQ ID NO: 41, SEQ ID NO: 42, SEQ ID NO: 43 and SEQ ID NO: 44 for determining the proportion of CD4 + - T cells; l) a nucleotide sequence selected from the group consisting of SEQ ID NO: 45, SEQ ID NO: 46, SEQ ID NO: 47 and SEQ ID NO: 48 for determining the proportion of CD8 +- T cells; m) a nucleotide sequence selected from the group consisting of SEQ ID NO: 49, SEQ ID NO: 50, SEQ ID NO: 51 and SEQ ID NO: 52 for determining the proportion of NK cells; n) a nucleotide sequence selected from the group consisting of SEQ ID NO: 53, SEQ ID NO: 54, SEQ ID NO: 55 and SEQ ID NO: 56 for determining the proportion of NK cells; o) a nucleotide sequence selected from the group consisting of SEQ ID NO: 57, SEQ ID NO: 58, SEQ ID NO: 59 and SEQ ID NO: 60 for determining the proportion of NK cells; p) a nucleotide sequence which differs from one of the nucleotide sequences according to a) to o) by the exchange of not more than 10% of the nucleotides, preferably not more than 5% of the nucleotides; and 5 q) a nucleotide sequence corresponding to the complementary strand of one of the nucleotide sequences according to a) to p).

Images