To determine whether cryopreservation altered the cellular composition of CD138⁺ myeloma cells, we performed cell cluster analysis for each patient sample. To determine the number of cell clusters, we performed principal component analysis using the top 2,000 genes with the highest cell-to-cell variation in expression levels (Supplementary Methods). This analysis resulted in the assignment of a total of 37,463 fresh and 49,203 frozen CD138⁺ cells from six patients into 22 clusters (Figure 2A and Supplementary Table 1). The number of clusters did not differ between fresh and frozen samples from the same patient (Figure 2B), apart from four small clusters that contained less than 0.1% of the total number of cells (Supplementary Table 2). As demonstrated in Figure 2C, the percentage of cells in different clusters was consistent between the fresh and frozen samples (R = 0.802). These findings indicate that cryopreservation maintained the overall cellular composition in CD138⁺ cells.

To evaluate whether cryopreservation maintained the gene expression levels in the CD138⁺ cells, we conducted pair-wise comparisons of the global gene expression levels for all fresh and frozen/thawed samples (Figure 3A). The global gene expression levels in each sample were calculated by averaging the normalized expression level of each gene in all cells (Supplementary Methods). Among all the comparisons, fresh and frozen samples from the same patient showed the highest correlation coefficient (R > 0.96). In addition, results from hierarchical clustering analysis showed that the fresh and frozen/thawed samples from the same patient were more similar to each other than they were to samples from different subjects (Figure 3B). This finding that inter-individual differences are preserved demonstrates that the freeze/thaw process had minimal effect on gene expression profiles.

We next asked whether the expression of key genes that define multiple myeloma cytogenetic subgroups were maintained following cryopreservation. Each of the six myeloma samples analyzed in this study belonged to one of the following three cytogenetic subgroups: hyperdiploid (HRD), translocation t(4;14), and translocation t(11;14). The chromosome 14 translocation subgroups are defined by high expression of key marker genes, including CCND2, FGFR3, and WHSC1 for t(4;14) and CCND1 and SLC8A1 for t(11;14) (Stein et al., 2017; Supplementary Table 3). We found that the expression levels of these marker genes showed high consistency between fresh and frozen/thawed cells from the same patient. For example, the expression level of FGFR3 showed highly similar expression profiles in fresh and frozen cells in both patients 2 and 6 (Figure 3C). Examples of the expression profiles of CCND2 and WHSC1 in patients 2 and 6, and of CCND1 and SLC8A1 in patient 3 are shown in Supplementary Figure 1. Together, these results demonstrate that cryopreservation had minimal effects on global gene expression levels or the expression levels of key genes defining subgroups with chromosome 14 translocations in CD138⁺ myeloma cells.

Unlike CD138⁺ cells, which are mainly composed of plasma cells, the CD138^– cell population is heterogeneous, consisting of hematopoietic stem cells and blood cells in various stages of maturation. To determine whether cryopreservation might affect cell type composition in the CD138^– compartment, we analyzed scRNA-seq data from 9,131 fresh and 21,586 frozen/thawed CD138^– cells from four of the same six patients described above. The CD138^– cells were classified into 19 clusters representing seven different cell types (Figures 4A,B). We confirmed that every cell population identified in the fresh cells was also detected in the cryopreserved cells (Figure 4C). Similar to what was observed for the CD138⁺ cells, the number of clusters in fresh vs. frozen CD138^– samples did not differ, except for patient 5 where clusters containing non-blood cells (e.g., epithelial cells, fibroblasts, and keratinocytes) were not found in the fresh cells. While the total number of cells differed between fresh and frozen samples (Supplementary Table 1), our results indicate that all cell types were preserved in the frozen/thawed CD138^– samples.

We further examined whether cryopreservation altered the gene expression patterns in each of the major cell types composing the myeloma microenvironment (CD138^–). Gene expression levels in the fresh and frozen cells from the same patient showed high correlation (R > 0.9) in all cell types (Figures 4D–I). Like the findings in the CD138⁺ cells, the correlation of gene expression profiles between fresh and frozen samples from the same patient were more similar to each other than to samples from different patients (Figures 4D–I). Based on these results, we conclude that cryopreservation also maintains gene expression patterns in CD138^– cell populations from bone marrow aspirates.

Because previous studies reported that cryopreservation of blood cells resulted in altered cell proportions of some immune cell types (Zheng et al., 2017; Bahsoun et al., 2019; Wohnhaas et al., 2019), we investigated whether the relative abundance of different immune cells was altered in cryopreserved CD138^– cells. We found that NK cells were more abundant in the frozen/thawed cells compared to the fresh samples; similarly, progenitor cells were also present at higher levels in frozen compared to fresh samples in three of the four patients (Figure 5A and Supplementary Table 2). In contrast, the proportion of B-cells and CD8⁺ T-cells were lower in the frozen/thawed cells compared with fresh cells and the proportion of CD4⁺ T-cells was lower in the frozen samples in 3 of 4 patients. Monocyte cell proportions also differed between fresh and frozen samples, although no consistent pattern was observed. Overall, the differences in cell proportions between fresh and frozen/thawed samples was on average less than 10% (Supplementary Table 2).

We next asked whether the observed differences in CD138^– cell proportions between fresh and frozen samples could be explained by differences in immune cell viability following thawing of the cryopreserved cells. One feature associated with cell stress and apoptotic cell death is a high number of mitochondrial transcripts (Ilicic et al., 2016). In our analysis, we considered cells having more than 15% mitochondrial reads as low-quality cells and removed them from the scRNA-seq data in the preprocessing quality control step. Therefore, we compared the percentage of mitochondrial reads in these low-quality cells and (Figures 5B–G) and the cells that passed the quality control filter (Supplementary Figures 2D–I) for each major cell type in the fresh and frozen/thawed samples from each patient. T-test was used to measure the statistical difference of the percentage of mitochondrial reads in the fresh and frozen/thawed samples from the same patient. We found the percentage of mitochondrial reads for each cell type was higher in the cryopreserved cells (Figures 5B–G), indicating that the freeze/thaw process likely generates a stress response in all cell types. These data suggest that the observed changes in cell type proportions (e.g., lower T-cells and higher progenitor cells) in the cryopreserved cells reflect the lower viability of T-cells relative to other cell types following freeze/thaw. Consequently, while cryopreservation maintains the cell populations present in fresh CD138^– cells, the freeze/thaw process can result in small changes to the relative abundance of each cell type.

Bone marrow aspirates were obtained from patients enrolled in The Indiana Myeloma Registry (Clinicaltrials.gov ID: NCT03616483). This is an IRB-approved prospective, non-interventional study collecting comprehensive clinical, genomic, demographic, social, environmental and quality of life data from subjects with plasma cell dyscrasias.

Human bone marrow mononuclear cells (BM MNCs) were collected in heparin tubes from multiple myeloma patients and were isolated by Ficoll-Paque PLUS (General Electric). Briefly, bone marrow aspirates were diluted twofold with phosphate-buffered saline (PBS) and layered over Ficoll-Paque PLUS in a 15 ml tube and centrifuged at 400 × g for 30 min at room temperature with the brake off. After removing the top layer of plasma, BM MNCs were collected and diluted with 3 volumes of PBS buffer, and washed once at room temperature. The red blood cells (RBC) were removed by addition of 7 ml RBC lysis buffer to the cell pellet and incubation at room temperature for 2 min, followed by centrifugation at 1,400 rpm for 5 min and one time wash with cold PBS. CD138⁺ and CD138^– cells were then sorted using CD138 Microbeads (Miltenyi Biotec), based on the standard MACS separation protocol (Miltenyi Biotec).

Freezing medium (90% FCS, 10% DMSO) was added to 5 × 10⁵ to 1 × 10⁶ cells/mL and cells were transferred to a cryogenic vial. The vials were put into a Mr. Frosty freezing container and stored at −80°C overnight. Vials were stored in liquid nitrogen until use.

Frozen cells were thawed slowly according to the Demonstrated Protocol for Fresh Frozen Human Mouse Cell line Mix, CG00014 RevF (10X Genomics, Pleasanton, CA). Briefly, a vial of cells was removed from liquid nitrogen storage and thawed in a water bath at 37°C until only a small ice crystal remained. Pre-warmed RPMI-1640 media containing 10% FBS (RPMI + FBS) was added to the cryovial and gently mixed with a wide-bore pipette tip. Cells were transferred to a 50 ml conical tube, followed by sequential additions of 2, 4, 8, and 16 ml of RPMI + FBS at a rate of 1 ml per 5 s. Cells were washed 3 times by pelleting cells at 300 × g for 5 min at room temperature and gently resuspending in 9 ml of RPMI + FBS. If cell aggregations were observed, the cell suspensions were filtered through a 40 μm Flowmi Cell Strainer (Bel-Art SP Scienceware, Wayne, NJ). Cell concentration and viability were determined, and cell concentration was adjusted to approximately 1,000 cells per μl. According to 10X Genomic protocol, the targeted cells per sample is 10,000 cells per μl for cell recovery.

Single-cell 3′ gene expression libraries were prepared for each sample targeting10,000 cells for cell recovery using the Chromium Single Cell 3′ Reagent Kits v2 and the Chromium single-cell system (10X Genomics, Inc). The resulting library was sequenced using a custom program for 26 bp plus 98 bp paired-end sequencing on a NovaSeq 6000 sequencer (Illumina, Inc, San Diego, CA). Approximately 50,000 reads per cell were generated.

Two rounds of data processing were performed using Seurat V3 (Stuart et al., 2019) and SingleR. In the first round, low-quality droplets were removed, and cell types were annotated. Low quality droplets were defined as those containing < 600 total unique molecular identifiers, or the percentage of mitochondrial reads were > 15% (Supplementary Table 1). Following cell-type annotation, we identified contaminating cells (i.e., non-plasma cells in CD138⁺ cells, and plasma cells and erythrocytes in CD138^– cells) and identified genes highly expressed in these cells using FindMarkers function with default parameters in Seurat. In the second round, the incorrect cells were removed before performing the standard processing procedure with Seurat to cluster cells. SingleR was used for cell type annotation. The detailed steps of data processing can be found in the Supplementary Methods.

Consistency of gene expression between fresh and preserved cells was calculated using Pearson correlation test function in R 4.0. To assess the overall consistency, we calculated the average gene expression profiles of each patient per cell type across samples for CD138⁺ cells and CD138^– cells separately. These average gene expression profiles were calculated for the entire gene set using the AverageExpression function in Seurat V3 and log2 transformed.