Alternative splicing of its 5'-UTR limits CD20 mRNA translation and enables resistance to CD20-directed immunotherapies

Abstract

Aberrant skipping of coding exons in CD19 and CD22 compromises the response to immunotherapy in B-cell malignancies. Here, we showed that the MS4A1 gene encoding human CD20 also produces several messenger RNA (mRNA) isoforms with distinct 5' untranslated regions. Four variants (V1-4) were detected using RNA sequencing (RNA-seq) at distinct stages of normal B-cell differentiation and B-lymphoid malignancies, with V1 and V3 being the most abundant. During B-cell activation and Epstein-Barr virus infection, redirection of splicing from V1 to V3 coincided with increased CD20 positivity. Similarly, in diffuse large B-cell lymphoma, only V3, but not V1, correlated with CD20 protein levels, suggesting that V1 might be translation-deficient. Indeed, the longer V1 isoform contained upstream open reading frames and a stem-loop structure, which cooperatively inhibited polysome recruitment. By modulating CD20 isoforms with splice-switching morpholino oligomers, we enhanced CD20 expression and anti-CD20 antibody rituximab-mediated cytotoxicity in a panel of B-cell lines. Furthermore, reconstitution of CD20-knockout cells with V3 mRNA led to the recovery of CD20 positivity, whereas V1-reconstituted cells had undetectable levels of CD20 protein. Surprisingly, in vitro CD20-directed chimeric antigen receptor T cells were able to kill both V3- and V1-expressing cells, but the bispecific T-cell engager mosunetuzumab was only effective against V3-expressing cells. To determine whether CD20 splicing is involved in immunotherapy resistance, we performed RNA-seq on 4 postmosunetuzumab follicular lymphoma relapses and discovered that in 2 of them, the downregulation of CD20 was accompanied by a V3-to-V1 shift. Thus, splicing-mediated mechanisms of epitope loss extend to CD20-directed immunotherapies.

Graphical abstract

Figure 1.

Healthy and malignant B cells express 4 CD20 transcript variants (V1-V4) with distinct 5′-UTRs. (A) Oxford Nanopore long-read direct RNA sequencing of mRNA from Raji cells. Sequence alignments corresponding to full-length, cap-to-poly(A) CD20 mRNAs with 4 5′-UTR splice variants and 2 alternative 3′-UTRs are shown. Alignments were extracted and visualized using an Integrative Genomics Viewer. (B) Diagram depicting the CD20 pre-mRNA and 4 distinct 5′-UTR splice variants, from V1 to V4. (C-D) Relative abundance of the V1 (red), V2 (green), V3 (blue), and V4 (yellow) isoforms in OCI-Ly8, Raji, and MEC-1 cells. Panel C is based on RNA-seq data, with the stack plot showing the ratio of sequencing reads mapped to the unique exon junctions found in each 5′-UTR variant of CD20. Here and below, these variants are color-coded as shown in the Reads spanning panel. Here and below, pan-isoform reads mapping to any exon of CD20 are shown on the right for comparison as TPM. Panel D shows RT-qPCR-mediated quantification of pan- and 5′-UTR variant–specific CD20 levels in OCI-Ly8, Raji, and MEC-1 cells. RNA levels are normalized to the reference gene RPL27. Each bar represents the average from each repeated experiment (N = 4). (E-I) Relative abundance of the V1 (red), V2 (green), V3 (blue), and V4 (yellow) isoforms in primary samples corresponding to healthy and malignant B cells. Each bar represents data from a single donor. The corrected TPM (cTPM) values at the top of panel H are the TPM values corrected for potential batch effects between the different data sets. Healthy B-cell subsets were fluorescence-activated cell sorting–enriched from the human bone marrow and tonsils (E) and peripheral blood (F). In panel E, CD19⁺IgM⁺IgD⁺ naïve tonsillar B cells were treated with anti-IgM or an isotype control. Western blots of phosphorylated SYK are displayed at the bottom right corner as a marker of B-cell activation, with actin serving as a loading control.

Figure 2.

CD20 protein levels positively correlate with the abundance of V3 and V4 but not V1, V2, or pan-isoform CD20 mRNA. (A-C) Levels of pan-isoform and 5′-UTR variant–specific mRNAs and CD20 protein (expressed as molecules of soluble fluorochrome or MESF) in primary or immortalized human samples. They were quantified using RT-qPCR and flow cytometry, respectively. RNA levels are normalized to those of 3 reference genes: RPL27, B-actin (ACTB), and glyceraldehyde 3-phosphate dehydrogenase (GAPDH). Each bar or point represents data from a single donor. Panel A displays data from EBV-transformed lymphoblastoid B-cell lines (LCLs) and peripheral blood B cells from healthy donors and patients with DLBCL, CLL, or FL. In panel B, peripheral blood B cells from 4 different donors (N = 4) were cultured with the B95.8 strain of EBV (EBV⁺) or without EBV (EBV⁻) for 3 days before CD20 expression quantification. (C) Regression analysis of 7 LCL samples from panel A, in which CD20 protein levels are independently correlated with each of the known CD20 mRNA isoforms. (D) A similar regression analysis of 27 DLBCL samples from TCGA consortium for which reverse phase protein array (RPPA) and RNA-seq data are available. The shaded area around the regression line indicates the standard error. The Spearman rank correlation coefficient, R, and the 2-tailed P value, P, as calculated using GraphPad Prism, are shown. APC, allophycocyanin.

Figure 3.

CD20 mRNA isoforms V3 and V4, but not V1 and V2, are efficiently translated into protein. (A) Polysome profiling of OCI-Ly8 and Raji cells. The top panel shows the ribosomal content measured at an absorbance of 254 nm, whereas the bottom panels show the relative distribution of specific transcripts across sucrose gradient fractions, as measured by RT-qPCR (supplemental Methods). (B) CD20 mRNA and protein measurements in HEK293T cells transiently transfected to express V1, V2, V3, or V4 mRNA isoforms. Pan-isoform CD20 transcripts were measured using RT-qPCR. Total CD20 protein was measured by flow cytometry and expressed as delta median fluorescence intensity (ΔMFI) (supplemental Methods). Representative scatter plots (bottom). ∗∗∗P < .001 per 1 way analysis of variance test (pairwise comparison [pwc]: Bonferroni). (C) Immunoblotting analysis of CD20 protein levels in cells from the previous panel. ACTB served as a loading control. Data from 2 independent experiments are shown. (D) CD20 mRNA and protein measurements in HEK293T stably expressing the empty vector (Ctrl-Puro) or the 5′-UTR and CDSs of V1, V2, or V3 (V1-, V2-, or V3-Puro). The kz-Puro Ctrl has the GCCACC Kozak consensus as its sole 5′-UTR element. The levels of CD20 mRNA were quantified by RT-qPCR. The MFI of total cellular PE-CD20 were determined by flow cytometry. The representative histograms are shown. (E) The same experiment was performed using OCI-Ly8 cells. (F) CD20 mRNA and protein measurements in OCI-Ly8 cells stably expressing the V1 or V3 5′-UTR sequences followed by a green fluorescence reporter (GFP) reporter (V1- and V3-GFP). The kz-GFP Ctrl has the GCCACC Kozak consensus as its sole 5’-UTR element. The GFP mRNA levels were measured using RT-qPCR. The MFI of GFP was measured using flow cytometry. Representative histograms are shown. In panels B,D-F, RNA levels are normalized to the reference gene RPL27. Each bar or dot represents the average of repeated experiments (N ≥ 2). RNA and MFI values are expressed as log₂ fold change relative to the Ctrl or Ctrl-Puro samples, as indicated. LTR, lentiviral tong terminal repeats.

Figure 4.

An RNA stem-loop structure and uORFs repress translation of the CD20 V1 isoform. (A) Diagram depicting the uORFs and the stem-loop in the 5′-UTR of the CD20 V1 isoform. The stem loop was predicted by the RNAfold web server. (B) Diagrams of mRNA products corresponding to various derivatives of V1-Puro. All AUG start codons found in the 5′-UTR of V1-Puro were mutated to AUC in the V1AUC-Puro and V1AUCDel-Stem-Puro constructs. Stem and Del-Stem refer to mutations stabilizing and destabilizing the predicted secondary structure, respectively. (C-F) Expression of V1- and V3-Puro and the V1-Puro mutants in stably transduced HEK293T cells. Pan-isoform CD20 RNA levels were quantified using RT-qPCR. The MFI of the total cellular CD20 protein was measured using flow cytometry. Representative histograms are shown on the right. Each bar or dot represents the average of repeated experiments (N = 2). All values are expressed as log₂ fold change relative to the Ctrl-Puro sample. (G) Diagram depicting putative Sam68 binding sites found within exons 2 and 3 of the V1. (H-I) Sam68 and CD20 expression measurements in Raji cells electroporated with a mixture of 2 Cas9-gRNA ribonucleoproteins targeting the CDS of Sam68. Shown in panel H is immunoblotting analysis of Sam68 protein expression, with ACTB serving as the loading control. Shown on the left side of panel I are Sam68 and CD20 mRNA levels, as quantified using RT-qPCR. Two PCR primer pairs measured nonoverlapping regions within the Sam68 transcript. The second primer pair is denoted with an asterisk (∗). Shown on the right side of panel I are MFI of total CD20 protein, as determined using flow cytometry. Representative histograms are shown on the far right. In both panels H and I, each row in the immunoblotting image and the heat map represent an independent experiment (N = 4).

Figure 5.

Shifting CD20 splicing toward V3 and V4 increases CD20 protein expression and rituximab-mediated cytotoxicity. (A-C) CD20 mRNA and protein measurements in OCI-Ly8, Raji, and MEC-1 cells treated with either the solvent control (no-Morpholino treatment [NT]), Ctrl-viMO, or 5ex3-viMO. (D) The same experiment performed on OCI-Ly8 variants with the intact MS4A1 gene (WT, top), knocked out MS4A1 gene (CD20-KO, middle), or knocked out MS4A1 gene replaced with the V3 expression cassette (V3-rCD20, bottom). CD20 RNA levels were quantified using RNA-seq (leftmost heat map in panel A and sashimi plot in panel B) and RT-qPCR (middle heatmaps in panel A). For the RNA-seq data in panel A, V1∗-4∗ are reads that mapped to the unique exon junctions found in each 5′-UTR variant of CD20, color-coded as shown in the RNA-seq reads panel. The MFI of total CD20 protein was determined using flow cytometry (rightmost heat map in panel A, top panel in panel C, and left and middle panels in panel D). Representative histograms are shown on the far right of panel A and far left of panel D. At the bottom panel of panel C and the right panel of panel D are shown cells that were additionally treated with increasing concentrations of rituximab (RTX), and cell viability (plotted) was measured using the WST-1 assay. All values are normalized to the “no rituximab Ctrl.” In panels A,C-D, each dot in a graph or value in a heat map represents independent experiments (N ≥ 2), with mean values indicated by red lines in the graphs.

Figure 6.

Shifting CD20 splicing from V3-to-V1 increases resistance to mosunetuzumab but not CART-20. (A) Diagram depicting the lentiviral vector transductions performed on OCI-Ly8 cells expressing a luciferase reporter. The MS4A1 gene was knocked out (CD20KO) and reconstituted with V1-, V2-, or V3-rCD20. The resistant CD20 (or rCD20) constructs have silent mutations in their CDS to avoid recognition by the CD20-targeting gRNA used in the knockout. (B) Isoform-specific RT-qPCR measurements of CD20 mRNA isoform relative to the RPL27 reference gene. (C) Measurement of CD20 protein levels by flow cytometry. On the left, it is quantitated as a log2 fold change in the MFI of total APC-CD20 relative to the unstained Ctrls. On the right, it is quantitated as the cell surface CD20 levels expressed as MESF. Each spot is an independent experiment. Representative histograms are included in each subpanel. (D) Viability of OCI-Ly8 cells after 24 hours of coculturing with CART-20 or untransduced donor T cells at 1:4 effector-to-target ratio. Cell viability (in percentage) is shown relative to that of Ctrl OCI-Ly8 cells that were not cocultured with either CART-20 or donor T cells. (E) OCI-Ly8 cell viability after 24 hours of coculturing with donor T cells and the indicated concentrations of mosunetuzumab. Cell viability (in percentage) is shown relative to that of the NT Ctrl. For panel D, data from 2 independent experiments using T cells from a single donor are shown. For panel E, data from 4 independent experiments with 2 unrelated donors are shown. In panel D and on the right of panel E, each spot represents the results from the replicate wells, with different colors denoting independent experiments. Horizontal bars represent the mean of all replicates. ∗∗P < .01; ∗∗∗P < .001 per the Kruskal-Wallis test (pwc: the Dunn test). (F) CD20 and PAX5 status per immunohistochemistry (IHC) (top) and transcript read abundance per RNA-seq (bar graphs) in paired pre- and postmosunetuzumab FL. TPM values quantify RNA-seq reads mapping to any exon in CD20 and PAX5. The relative abundance of V1 (red), V2 (green), V3 (blue), and V4 (yellow) is the ratio of sequencing reads mapping to the unique exon-exon junctions found in each 5′-UTR variant of CD20, as color-coded in the Reads spanning panel. (G) Micrographs corresponding to P29 pre and postmosunetuzumab FL samples. Formalin-fixed paraffin-embedded sections were IHC stained for CD20 (brown colorimetric detection with 3, 3'-diaminobenzidine [DAB]) using hematoxylin nuclear counterstain (blue). (H) Changes in ΔPSI values or in the ratio between reads including or excluding CD20 exons in the P29-post sample relative to the P29-pre sample. RNA-seq data were analyzed using the MAJIQ algorithm for all possible splicing changes in CD20, but only ΔPSI values above or below 0.05 are shown. (I) Sashimi plots depicting the density of exon-including and exon-skipping reads in the RNA-seq data corresponding to P29-pre and -post samples. FACS, fluorescence-activated cell sorting; H&E, hematoxylin and eosin.