Fig 1.
Isolation of human in vivo-generated PC and B-cells.
(A) Tonsil PC (TPC), (B) peripheral blood PC (BPC), (C) bone marrow PC (BMPC) and (D) peripheral blood naïve B cells (BC) were isolated by magnetic selection and flow cytometry sorting according to their surface markers expression as was indicated in Material and methods section. Representative dot plots are depicted.
Fig 2.
Human B-cell maturation into PC produces distinctive signatures.
(A) Principal Component Analysis (PCA) of the whole transcriptomes of BC, TPC, BPC and BMPC. (B) Heatmap of the average expression of differentially expressed probesets (adj. P-value < 0.1) in any pair-wise comparison. (C) PC-down signature. Left, Venn diagram showing the down-regulated genes in each PC compared to naïve BC: the central number represents the PC-down signature. Right, GO enrichment analysis of PC-down signature (P< 0.05, DAVID), in which the number of genes represents the mean ± SD of the genes contained in the GO terms within the indicated functional categories. (D) PC-up signature. Left, Venn diagram showing the up-regulated genes in each PC compared to naïve BC: the central number of the diagram represents the PC-up signature. Right, GO analysis as in (C) for the PC-up signature. (E) Box-and-whisker plots of the PC-down, BPC-up and PC-up genes that overlapped with the subset defined in GenomicScape as those genes with the highest deviation from average expression among naïve BC (NB), memory BC (MB) and BMPC from the resource [20]. Numbers of overlapping genes are in brackets in the color legend. For clarity, outliers have been removed. *, P < 0.05 in Student’s t-tests between signatures within the same GenomicScape cell subtype; §, P< 0.05 Student’s t-tests between each cell subtype and NB within the same signature.
Fig 3.
The PC signature is largely conserved in human and mouse.
(A) Distribution of the genes from the mouse ASC signature previously defined [30] that were unequivocally comparable to human orthologs in the Agilent arrays (final number: 260 genes), across the whole transcriptome of human BC, ordered from more to less significance, without applying any filter. Data are expressed as mean ± SEM of the number of genes in each pair-wise comparison (any PC vs BC). The circles indicate the number of genes showing an overlap between the mouse ASC signature and the most up- and down-regulated genes (top 2000) in naïve BC compared to the three PC. In brackets it is denoted the number of overlapping genes with the human PC-down and PC-up signatures. Note that the ASC overlap specifically with the PC-up but not with the PC-down signature, as expected. (B) The results of the GO enrichment analysis for the genes in the mouse ASC signature that were either up-regulated genes in human PC (upper panel) or defined in the human PC-up signature (lower panel) are represented as in Fig 2. Most of the enriched functions observed in the human PC-up signature (Fig 2D) are also retrieved in this analysis. Ub, Ubiquitylation, Glycos, Glycosylation.
Fig 4.
Transcriptional markers of PC subtypes.
(A) The number of significantly up-regulated genes in one PC-subtype (Specific), or in more than one PC-subtype (Common) are indicated. (B) RT-qPCR assays in independent samples (n = 3 for BPC, TPC and n = 4 for BMPC) for selected genes. *, P ≤ 0.05, Mann-Whitney test related to BPC. (C) GO enrichment analysis for up-regulated genes in BPC (BPC-up) (P< 0.05, DAVID) as represented in Fig 2. (D) Box-and-whisker plots of the PC-down, BPC-up and PC-up genes that overlapped with the successive stages obtained in an in vitro BC differentiation system defined in GenomicScape; this included pre-plasmablasts (prePB), plasmablasts (PB) and early plasma cells (EPC) stages [20]. *, P < 0.05 in Student’s t-tests between signatures within the same GenomicScape cell subtype; §, P< 0.05 Student’s t-tests between each cell subtype and prePB within the same signature.
Table 1.
Functional enrichment analysis of PC signatures.
Fig 5.
Putative regulation of the PC transcriptional program.
(A) The 282 TFBS of the TRANSFAC database contained in the Pscan tool are ranked according to their Z-scores in the analysis of PC-down and PC-up signatures. The black lines separate the significantly and non-significantly enriched TFBS (P < 0.05). Coloring notes the same TFBS in adjacent ranks: blue and red for significant motifs in the PC-down and PC-up list, respectively. Divergent motifs (significant in one rank but non-significant in the other) are indicated. Letter intensity denotes the degree of divergence, according to a delta (Δ) calculation of the rank position between both lists: bold for Δ > 150, normal for 50 < Δ < 150 and grey for Δ < 50. (B) Number of genes in the PC-down (blue) and PC-up (red) signatures that are altered in conditional knockout BMPC for Prdm1 and Xbp1 [40]. The genes are divided in concordant and discordant, dependent on the expected direction of change. (C) The same analysis as in A was performed for specific BPC and TPC/BMPC genes. Most of the resulting TFBS can be redundant due to sequence similarities, and can be grouped by transcription factor families: AP-2: V$AP2ALPHA_01, V$AP2GAMMA_01; Ahr and Arnt-like (PAS domain): V$AHRARNT_02, V$ARNT_01, V$ARNT_02; ATF/CREB: V$ATF_01, V$ATF6_01, V$CREB_01, V$CREB_02, V$CREB_Q2, V$CREB_Q4, V$CREBP1_01, V$CREBP1_Q2, V$CREBP1CJUN_01; other ATF/CREB related: V$AP4_01, XBP1_01; E-box (bHLH-ZIP): V$MAX_01, V$MYCMAX_01, V$MYCMAX_03, V$NMYC_01, V$SPZ1_01, V$USF_01, V$USF_Q6; E2F-related (Forkhead): V$E2F_02, V$E2F_03; ETS-related: V$ELK1_02, V$CETS1P54_01; NF-κB-related: V$CREL_01, V$NFKB_C, V$NFKAPPAB_01, V$NFKAPPAB50_01, V$NFKAPPABP65_01; NF-Y (CCAAT-binding): F$HAP234_01, V$NFY_01; Oct-like (POU domain): V$OCT_C, V$OCT_05, V$OCT_Q6; PAX: V$PAX4:_03, V$PAX5_02; SRF (MADS-box): V$SRF_C, V$SRF_01, V$SRF_Q6; STAT: V$STAT1_01, V$STAT3_01, V$STAT5A_02, V$STAT6_02.
Fig 6.
Transcriptomic distribution of PC signatures in healthy BMPC and MM.
The whole transcriptomes of BMPC and MM from four datasets (see Materials and Methods) were ranked according to the level of gene expression, and the number of the genes contained in the PC-down (upper panel) and PC-up (lower panel) signatures were counted and averaged in each bin. The profile of our in-house BMPC is depicted in gray for comparative purposes. Data are expressed as mean ± SD. The % of genes in the 90th percentile is also depicted.
Fig 7.
The PC transcriptional program is imbalanced in monoclonal gammopathies.
Genes of the PC-down and PC-up signatures that were also dysregulated genes in MGUS, SMM and MM related to healthy BMPC are represented as total number (A) or as number and % in each signature and direction of change (B, C). The same representation is depicted for BPC-up and TPC/BMPC-up signatures (D-F). Data are expressed as mean ± SD. (A, B, D, E) or as proportion (C, F). (G, H) Prediction analysis of TFBS enrichment in the dysregulated genes in MM and MGUS. Similar to Fig 5, The TFBS of the TRANSFAC database in the Pscan tool are ranked according to their Z-scores in the analysis of PC-down and PC-up signatures (G) and BPC-up and TPC/BMPC-up signatures (H). The black lines separate the significantly and non-significantly enriched TFBS (P < 0.05). Coloring follows the same code as in Fig 5. The 95th percentile (the most significantly enriched TFBS) is shown in a dash line for comparison purposes.
Table 2.
Differentially expressed genes of the PC signatures between MM and MGUS.
Table 3.
Genes of the PC signatures as potential MM risk markers.