Single-cell and clonal analysis of AL amyloidosis plasma cells and their bone marrow microenvironment

Abstract

AL amyloidosis is a disorder characterized by expansion of clonal plasma cells in the bone marrow and distant end organ damage mediated by misfolded immunoglobulin free light chains. There are currently limited data regarding the functional characteristics of AL amyloidosis plasma cells and their surrounding bone marrow microenvironment. We performed 5' single-cell RNA sequencing on newly diagnosed, treatment-naïve patients with AL amyloidosis and healthy subjects. We identified generalized suppression of normal bone marrow hematopoiesis with distinct expansion of monocytes and subsets of CD4+ T cells in patients with AL amyloidosis. We detected significant transcriptional changes broadly occurring among immune cells with increased tumor necrosis factor-α signaling and interferon response accompanied by increased inflammatory response in bone marrow plasma, as measured via quantitative proteomics with specific elevation of costimulatory molecule soluble CD276 (sB7-H3). A transcriptionally distinct population of nonmalignant plasma cells was disproportionately expanded in patients with AL amyloidosis and characterized by increased expression of CRIP1. Finally, clonal AL amyloidosis plasma cells were identified based on their unique variable-diversity-joining. rearrangement and showed increased expression of genes involved in proteostasis when compared with autologous, polyclonal plasma cells. Interpatient transcriptional heterogeneity was evident, with transcriptional states reflective of common genomic translocations easily identifiable. This study defines the transcriptional characteristics of AL amyloidosis plasma cells and their surrounding bone marrow microenvironment with identification of altered genes previously involved in the pathogenesis of other protein deposition disorders. Our data provide the rationale for functional validations of these genes in future studies.

Figure 1.. Study design

A. Schema synthesizing the workflow from bone marrow aspirate and biopsy collection in AL patients and HDs until analysis. B. Key characteristics and allocation of material derived from AL amyloidosis patients included in this study. Further clinical metrics can be found in Supplemental Tables 1. C. Epidemiologic data and allocation of material derived from healthy donors (HD). Supplemental Table 3 reports on sample batching and number of cells analyzed per each AL and HD individual. D. Hematoxylin-eosin (first row), CD138 staining (second row) and congo red stain under direct and polarized light (third and fourth row, respectively) of bone marrow biopsies of DFCI/BWH AL amyloidosis patients included in scRNAseq analysis. For patient AL9 (*), a double stain of CD138 and cyclin D1 is shown in red in the second row. Congo red stained was performed on all patients and it is shown only if positive. For patients AL3, AL4, AL8 and AL9 amyloid deposition was minute (pink arrows).

Figure 2.. Bone marrow hematopoiesis in AL is suppressed and upregulation of TNF-α and inflammatory response is noted across several cell states.

A. Projection of CD138− transcriptomes of DFCI/BWH and Tufts cohorts onto a pre-annotated two-dimensional embedding that spans the hematopoietic hierarchy and mature BM cell types. B. Log-log plot of frequencies of annotated bone marrow cell states comparing AL patients to HDs. Points are colored by statistical significance of changes in frequency after beta-binomial testing and multiple hypothesis correction. Changes indicate a significant and substantial expansion in monocytes and distinct CD4+ T cell subsets, and a broad suppression of other hematopoietic states. C. Multiplex immunofluorescence staining for CD3, CD4, CD14 and CD16 of formalin fixed and paraffin embedded bone marrow biopsies from six AL patients (first and second row) and three healthy donors (last row, HD-A, -C and -E). Quantification of cell types is shown on the right. D. Rank-ordered plot of the number of differentially expressed genes (DEGs) between AL patients and HDs in each of the annotated CD138− cell states in panel B at a 5% false-discovery rate. CD14 monocytes and distinct CD4+ T cell subsets undergo the largest statistically significant changes in gene expression. E. Volcano plots of DEGs in selected cell populations, revealing recurrent and state-specific changes. F. Identification of recurrently elevated gene sets across CD138− cell states. The plots summarize the number of cell states with increased gene set expression, revealing increases in TNF-α signaling and inflammatory response. G. Fold-change expression (log2-scale) of a representative interferon response gene (IFITM3) and a TNF-α response gene (RELB) across annotated cell states. Red colors indicate that the changes are statistically significant (5% FDR). Plots are ordered by magnitude. H. Western blot showing increased expression of IFITM3 and RELB in protein lysates obtained from CD138− bone marrow monomuclear cells from AL patients as compared to healthy donors. GAPDH and ponceau staining (supplementary figure 2F) serve as loading control. Uncropped western images are shown in supplementary Figure 2F–G.

Figure 3.. Bone marrow plasma quantitative proteomics in AL amyloidosis patients and healthy individuals

A. Volcano plot showing significantly proteins enriched (right side, red) or depleted (left side, blue) in bone marrow plasma of AL amyloidosis patients as compared to healthy donors. Raw data available in Supplementary table 3 B. Hallmark pathway analysis showing significantly enriched pathways (green bars) in AL amyloidosis patients as compared to healthy donors. Adjusted p value <0.005. Black arrows point at enrichment in TNF α and inflammatory pathways. Red arrow points at enrichment in complement proteins C. CD276 levels in the plasma of healthy donors (HD) versus AL amyloidosis patients (AL) as measured by Luminex technology. P value: 0.01. Raw data are shown in Supplemental table 4. D. Reactome pathway analysis showing significantly enriched pathways (blue bars) in AL amyloidosis patients as compared to healthy donors. Adjusted p value <0.005. Arrows point at pathways involved in phagocytosis and protein organization/remodeling.

Figure 4.. Clonal and gene expression distributions of PCs in AL amyloidosis and healthy individuals

A. Indication of clusters annotated as plasma cells in the total data set embedding from Supplemental Figure 2A. B. scVI minimum distortion embedding (MDE) of PCs after further filtering steps. Asterisk (*) indicates states removed from subsequent analyses based on a signatures of stress/lysis. Colors indicate Leiden clusters. C. Rank plot of PC according to clone size, colored by condition (AL in blue, HD in orange). The top 10 largest clones are from AL patients, and represent all 9 AL clonal PC plus a κ-restricted, non-malignant, plasma cell population in patient AL3 (AL3κ). D. Cells belonging to the top 10 largest clones (Fig. 4C) indicated on the embedding with distinct colors and symbols, indicating that discrete clusters are enriched for expanded clones. Cells from all other detected, smaller clones are represented as black dots. E. Fraction of PCs in polyclonal clusters (PC1–3) among AL patients (AL) and healthy donors (HD). Relative expansion of PC3 population is observed in AL. P-values from beta-binomial testing. F. The filtered scRNA-Seq embedding, colored by dominant immunoglobulin light chain transcript expression. Light chain expression partitions plasma cell clusters PC1–2. G. Heatmap quantifying the immunoglobulin light chain usage across clusters (see also panel F). Color scale gives log-transformed normalized counts. H. Heatmap showing the top 20 genes enriched in each polyclonal PC cluster (PC1–3) after excluding light-chain VJ transcripts. Colorbar normalized to the median of the 3 clusters (CP10K_ref.) for each gene. I. Co-immunofluorescence staining using CD138 (AF549, red) and CRIP1 (AF488, green) to identify CRIP1+ plasma cells in representative fields of selected bone marrow biopsies from AL patients (AL) and healthy donors (HD). DAPI was used to counterstain nuclei. Yellow arrowheads point at CD138+/CRIP1+ cells. Pink arrows point at CRIP1+ blood vessels in healthy donors, suggesting CRIP1 staining of endothelium. Scale bar is shown. Images obtained from each AL and HD individuals are shown in Supplemental Figure 3H. J. Fraction of PC3 cells identified in scRNAseq datasets obtained from patients with AL amyloidosis (AL), monoclonal gammopathy of undetermined significance (MGUS), smoldering multiple myeloma (SMM), multiple myeloma (MM) and healthy donors (HD). Each dot represents an individual patient/healthy donor. Dots are color-coded according to dataset. Analysis pipeline is shown in Supplemental Figure 3I and detailed in Supplementary Methods. Data regarding patients included in this analysis are provided in Supplemental Table 5 and 6.

Fig. 5.. Signatures of dominant plasma cell clones in AL amyloidosis

A. Schematic representation of the analytical approach for identifying signatures of malignant PCs. B. Plasma cell clonal frequencies per individual (the fraction of all plasma cells with detectable and distinct BCRs per each individual). Each color represents a single clone, ordered by size (blue: largest clone; green: second largest clone; purple: third largest clone; etc..). Clones with ≤2 cells are grouped together in grey. C. Expression of the top 10 genes enriched in each of the AL-specific clusters, with 1–2 genes per cluster highlighted. For full list, see supplemental table 4. Colorbar gives log2-transformed gene expression after normalization to the 2nd-max value of the states plotted. D. Volcano plot of fold-change in gene expression and associated adjusted p-values in the malignant PC clones from all AL patients compared to non-clonal PCs as inferred from regression. E. As (D), but restricted to donors with t(11;14). F. Venn diagram showing the overlap of genes enriched in PC clones with a t(11;14), as compared to genes enriched in all AL-PCs. The genes included reflect a subset of hits from (D,E) at FDR<5% and which further show a minimum 1.5-fold change in at least 4/9 patients (all patient), and in all 3/3 patients for t(11;14). G. Normalized expression level of ITM2B and CCND1 transcripts as assessed via quantitative RT-PCR in AL amyloidosis patients harboring a t(11;14) (red) or not (black). **: P value=0.0037, ***: P value= 0.0003. H. Heatmap showing expression of the 19 genes enriched in AL malignant PC clones per the criteria in (F), as well as the three additional genes for the subset of patients with t(11;14). Colorbar gives log2-transformed gene expression after normalization to the mean of non-AL states PC1–2 (CP10K_PC). I. Co-immunofluorescence staining using CD138 (AF549, red) and FKBP2 (AF488, green) identified colocalization and strong expression of FKBP2 in plasma cells from AL amyloidosis patients (AL), but not healthy donor (HD) plasma cells. DAPI was used to counterstain nuclei. Scale bar is shown.

Fig. 6.. Characteristics of immunoglobulin sequence in AL-PCs and clonal analysis of T and B cells in AL amyloidosis patients

A. The table outlines the IgLV and IgLJ genotype of clonal PCs for each individual AL patient. B-C. Frequency of use of IgHV (B) and IgLV (C) genotypes across all plasma cells (clonal and non clonal) in AL (blue) and HD (yellow) individuals. D. Gene-specific substitution profiles of IGHV2–14 germline genes for the 3 patients with IgHV2–14 genotype (AL1, AL3 and AL4). The germline sequence is listed at the bottom in bolded letters, followed by patient AL1, AL3 and AL4 sequences and published AL light chain dimer (PDBID: 2OLD), serves as a positive control. Dots indicate no substitution. Black letters indicate mutations and red letters indicate mutations that are rarely observed during the affinity maturation process (frequency <0.5%). Outlined in yellow are amino acid residues at the dimer-dimer interface, identified from the PDB structure (2OLD). Outlined in green is the consensus sequence for N-glycosylation (NXS/T motif, X represents any amino acid except proline). The length of an amino acid letter in the graph is proportional to the frequency of such substitution in an established healthy donor antibody repertoire. Gene-specific substitution profiles of IgHV locus of remainder of AL patients are shown in Supplemental Figure 6F-K. E Specificity of BCR and TCR sequencing in CD138− cells is confirmed by inspecting the localization of cells with detectable BCR and TCR sequences on the scRNA-Seq embedding from Fig. 2B. Over 98% of cells with detectable BCR/TCR sequences localize to B or T cell lineages, respectively. F. T cells belonging to expanded clones (≥10 cells/clone detected) show an expected increase in gene expression for genes associated with T cell activation. Scores shown are evaluated on each clone. G. Size distributions of T cell clones aggregated from all individuals. Healthy and AL patients have T cell clone size distributions with characteristic power-law tails, with different exponents (shown) but no evidence of outlier expanded clones. H. Comparison of the normalized size of the largest T cell clones shows no evidence of consistent expansion in AL patients compared to HDs. I. Size distributions of B cell clones aggregated from all individuals show no evidence of expanded clones in AL patients. Clones are identified by common light-chain (VJ) sequences. J. Box and whisker plots showing the distribution of IgL CDR3 length across each individual AL amyloidosis patient (blue) and HD (yellow). Distribution of CDR3 length in AL patients (blue box and whisker plots) and HDs (yellow box and whisker plots). Each dot represents a distinct CD138− B cell clone. K. Box and whisker plots showing the percentage of somatic hypermutation at the IgH locus across each individual AL amyloidosis patient (blue) and HD (yellow). Each dot represents a distinct CD138− B cell clone.