High-Quality Nuclei Isolation from Postmortem Human Heart Muscle Tissues for Single-Cell Studies

Single-cell approaches have become an increasingly popular way of understanding the genetic factors behind disease. Isolation of DNA and RNA from human tissues is necessary to analyze multi-omic data sets, providing information on the single-cell genome, transcriptome, and epigenome. Here, we isolated high-quality single-nuclei from postmortem human heart tissues for DNA and RNA analysis. Postmortem human tissues were obtained from 106 individuals, 33 with a history of myocardial disease, diabetes, or smoking, and 73 controls without heart disease. We demonstrated that the Qiagen EZ1 instrument and kit consistently isolated genomic DNA of high yield, which can be used for checking DNA quality before conducting single-cell experiments. Here, we provide a method for single-nuclei isolation from cardiac tissue, otherwise known as the SoNIC method, which allows for the isolation of single cardiomyocyte nuclei from postmortem tissue by nuclear ploidy status. We also provide a detailed quality control measure for single-nuclei whole genome amplification and a pre-amplification method for confirming genomic integrity.

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  Catalog No. 14-282-301) and homogenized with a mechanical dissociator with use of reusable 131 tips (Omni 30750H) at 24,000 rpm with 2 × 30 sec pulses per sample (Figure 1c). 132 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 5, 2023. ; https://doi.org/10.1101/2023.02.05.526322 doi: bioRxiv preprint

Cell-type identification from 10X RNAseq data 185
After sequencing, the resulting sample FASTQ files from all samples were processed using 186 CellRanger (v2.1.1) and Seurat package (v3.1.5) pipeline. Using the "mkfastq" and "count" 187 commands we generate raw gene-barcode matrices and align them to GRCh38 Ensembl (v1.2.0). 188 Combining multiple strategies, we compile a list of genes that are expressed in each type of cell 189 in a specific way. By categorizing each nucleus as either coming from the type of target cell or 190 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 5, 2023. ; https://doi.org/10.1101/2023.02.05.526322 doi: bioRxiv preprint type), we eliminated genes whose counts were too low for testing. To determine marker genes, 193 we selected protein-coding genes that were expressed in at least 25% of nuclei from the target 194 cell type, with AUC for the target cell type greater than 0.60, a log-fold change, and an FDR 195 adjusted P-value < 0.01 16 . The cell-type labels for each cluster were assigned based on enriched 196 ontologies. Based on the mean expression of the top 1,000 most variable genes (the top 10 genes 197 are shown in Table 2), cell-type centroids were grouped together. This process was repeated with undigested tissue until all the cells were dissociated from the 232 tissue blocks. Cells were filtered through a 100 μm filter and centrifuged at 300xg for 2 minutes 233 to pellet the cardiomyocytes. Pelleted cardiomyocytes were resuspended in 1-3 ml (depending 234 on the yield) of 1X PBS. Cells were fixed for 5 minutes, with 100% methanol precooled to -235 20 o C, and the volume was adjusted so that the final concentration was 95%. Fixed cells were 236 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 5, 2023. ; https://doi.org/10.1101/2023.02.05.526322 doi: bioRxiv preprint centrifuged at 300xg for 5 minutes and the supernatant was discarded. Cardiomyocytes were 237 permeabilized with 0.5% Triton X-100 in 1X PBS at room temperature for 10 minutes and 238 centrifuged at 300xg for 5 minutes, blocked with 1% BSA for 30 minutes at room temperature. 239 The cells were incubated with a-actinin as the primary antibody for 2 hours at room temperature 240 and washed in 1% BSA in PBS twice. Secondary antibody incubation was done for 2 hours at 241 room temperature, and cells were washed twice after incubation with 1% BSA in PBS. For 242 evaluating the myocardium tissue quality, we also performed immunostaining on intact 243 myocardium. The tissue was fixed and embedded in a paraffin block. After deparaffinization, 244 tissues were heated in citrate buffer pH 6.0 (Millipore Sigma) for 20 minutes, permeabilized 245 using 1% donkey serum in PBS plus 0.5% Triton X100, blocked in PBS-T containing 5% 246 donkey serum for 30 minutes at room temperature and incubated with a monoclonal antibody 247 against a-actinin at 1:500 dilution overnight at 4°C. After rinsing in PBST (PBS plus 0.1% 248 Tween-20), sections were incubated with secondary donkey IgG Alexa Fluor 647-conjugated 249 antibody for 1 hour and stained with Wheat Germ Agglutinin and Syto 13 at 1:1000 dilutions for 250 30 minutes at room temperature. Images were captured on an LSM 880 confocal microscope 251 (Zeiss) and processed using ImageJ (NIH). 252 253

Statistical analysis 254
Linear regression analysis was conducted to assess the association between DIN or RIN and PMI 255 or age. The Mann-Whitney U test was performed to assess the association between DIN or RIN 256 and gender or race. The Kruskal-Wallis test was performed to assess the association between 257 DIN or RIN and disease status. Scatterplots and boxplots were also created. 258 259 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made   7 to be an optimal quality of tissue for further biological work 25 . We also tested the DIN from 279 multiple other organs (heart, liver, kidney) for a small number of cases (n=4). Our analysis 280 indicates liver tissue from the same case had a trend of lower DIN with some less than the 281 optimal DIN value of 7 (Table 1b). We found that decreased DIN value was associated with 282 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 5, 2023.

290
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 5, 2023. ; https://doi.org/10.1101/2023.02.05.526322 doi: bioRxiv preprint association with DIN values for Black/African American versus White/Caucasian donors ( Figure  292 2g, p-value=0.570), or for female versus male donors (Figure 2h, p-value=0.551). Furthermore, 293 we tested the association of RIN and DIN with disease status, characterized as donors with 294 history of atherosclerosis or hypertension as compared to control members; of note, all the DIN 295 values were greater than 7 (Table 1a). We performed the Kruskal-Wallis test for testing the 296 dependence of disease status versus RIN/DIN (Figure 2i, 2j) and found no association (p-297 value=0.924 and p-value=0.467, respectively). This is notable because many single-cell analysis 298 experiments aim to compare the presence of genome level mutations in control versus disease 299 tissues. If researchers run into problems with their sample preparation, including unsuccessful 300 attempts at whole genome amplification, lack of sufficient amplified sample concentration, or 301 uneven amplification across loci, it is possible that the tissue quality has exacerbated these 302 problems. Therefore, it may be prudent to check the DNA integrity prior to sample preparation to 303 help ensure better results for downstream analysis. 304 305

Determination of quality of cardiomyocytes following single-cell isolation 306
In this study, the heart tissues we used were collected from the NeuroBioBank. Thus, it was 307 important to evaluate the quality of the heart muscle cells from the tissue bank, where the heart is 308 not the main organ collected in the tissue bank. We tested the quality of isolated cardiomyocytes 309 (Figure 3a, b) from frozen tissue by staining with a-actinin (Millipore-Sigma Catalog No. 310 A7811) as well as a GAP junction protein, connexin 43 (Abcam Catalog No. ab87645). Our 311 immunofluorescence staining with intact connexin in cardiomyocytes ( Figure 3c) and a-actinin 312 ( Figure 3d) indicated superior quality heart cells could be isolated from these postmortem heart 313 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 5, 2023. ; https://doi.org/10.1101/2023.02.05.526322 doi: bioRxiv preprint tissues that were collected at the NeuroBioBank. These isolated cardiomyocytes are not for live 314 culture, but they could be used for cell-specific protein expression analysis. 315 316 with a-actinin (red) to check for intact myocardium before isolation methods. 321 322

346
Determination of DNA quality from single-nuclei isolation 347 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  Table 3). MAPD analysis indicates that 84.8% 374 of nuclei amplified by MDA had even genome coverage for case 5657 where the DIN was 8.8 375 and 68.75% for case 5657 where the DIN was 5.8. Together our analysis indicates that SoNIC 376 method could be utilized to isolate superior-quality nuclei from postmortem heart tissue even 377 when the DIN score is 5.8, which is below-average tissue quality. 378

Conclusions: 379
Our modified single-nuclei isolation from cardiac tissue (SoNIC) method allowed for the 380 isolation of single cardiomyocyte nuclei from postmortem tissue for single-nuclei whole genome 381 amplification as well as RNAseq. Single-nuclei sorting criteria based on the ploidy of nuclei 382 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 5, 2023. ; https://doi.org/10.1101/2023.02.05.526322 doi: bioRxiv preprint race, whereas the same subset of tissue samples indicated a negative correlation between DIN 387 and PMI or age. Our study emphasizes the inclusion of DIN along with RIN for tissue quality 388 measures prior to the performance of whole genome amplification in human postmortem heart 389 tissue. Our study indicates that postmortem frozen tissue with a DIN over 5.8 could be used for Limitation of the study: 411 We restricted our analysis to available heart tissue from NIH NeuroBioBank, the University of 412 Maryland (between 2016-2020), which had a DIN value between 5.8 and 9.8. 413 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 5, 2023. ; https://doi.org/10.1101/2023.02.05.526322 doi: bioRxiv preprint