Global change in the gene expression profile in human T cells induced by anti-CD3 treatments
To compare the effects of each of the three anti-CD3 antibodies on human T cells, the gene expression profiles were analyzed. T cells were obtained from 72-hour untreated or treated PBMCs with one of the three anti-CD3 antibodies: OKT3, FvFc M (OKT3 scFv fused to human IgG1 Fc), and a humanized version of this FvFc (FvFc R). Anti-CD3 was used as the sole stimulus. To avoid any further stimulation, T cells were obtained by negative selection, using magnetic beads for cell surface markers. The purity of the T cell population was assessed by flow cytometry and was above 96% (Supplementary Figure S1). The transcriptomes of stimulated and unstimulated T cells from a single individual were obtained by performing sequencing in two replicates. More than 55 million paired-end reads of 150 bp length were obtained. The reads were mapped to the human reference genome (hg19); of the total reads, 84% to 94% were mapped (Table 1).
Subsequently, we assessed differentially expressed genes (DEGs) by comparing each anti-CD3 antibody-treated sample with the control of unstimulated T cells. The gene sets found to be differentially expressed in the different treatments are shown as a MA plot in Figure 1A and as a Venn diagram in Figure 1B. OKT3 treatment resulted in a larger set of differentially expressed genes (7089) with a fold change of less than -0.8 or above 0.8, followed by FvFc R treatment with 2425 DEG and FvFc M treatment with 1406 DEG. We found 860 genes that were equally regulated among the treatments, considering a padj ≤ 0.05. Except for FvFc R treatment, DEGs were mostly downregulated. FvFc R induced the most unbalanced DEG dataset, with 58% (1,419) upregulated over 41% (1,006) downregulated DEGs. The gene regulation profile promoted by FvFc R was more similar to OKT3 than FvFc M, even though the cluster analysis suggested a similar DEG profile for each treatment (Figure 1C).
Associations of DEGs with Gene Ontology categories
Anti-CD3 stimulation was shown to affect different set of genes18,19. Therefore, functional characterization of the differentially expressed genes was performed using GO term enrichment analysis. Anti-CD3 activated and repressed DEGs were separately classified for the GO category “biological process”. Upregulated genes were dominated by terms associated with cell proliferation (Figure 2), reflecting the anti-CD3 associated activation of T cells. To visualize changes in GO term enrichment and coverage (completeness), immune-associated terms were selected among up- and downregulated DEGs for each antibody treatment, focusing on those associated with the immune response and inflammation typically associated with anti-CD3 therapy (Figure 3).
All antibodies induced a similar profile of GO term enrichment, coverage and FDR adjusted p-value, shown by radar plots (Figure 3). Among the upregulated genes, the predominance of OKT3-induced GO term coverage was less obvious. Between selected terms, the most enriched GO term among the upregulated genes was the Regulation of Regulatory T Cell Differentiation (GO:0045589), but terms for the regulation of IFNγ (GO:0032729), IL-10 (GO:0032653) and IL-12 production (GO:0032655) were also highlighted.
The downregulated DEG set enriched terms reflected categories that fade after antibody treatment. It is notable that, among the GO terms enriched by genes repressed after treatment, the term “regulation of inflammatory response” (GO:0050727), was the most conspicuous. Furthermore, the terms “immune response-regulating signaling pathway” (GO:0002433) and “activation of immune response” (GO:0002253) were also evident (Figure 3).
Regulation of cytokines and their receptors by anti-CD3 stimulation
Anti-CD3 antibody therapy is strongly associated with an over secretion of cytokines, also known as a “Cytokine Storm”4. The deleterious consequences of the cytokine production are assumed to be promoted by the Fc part of the molecule, and novel humanized antibodies can circumvent these consequences by inducing a nonmitogenic effect. Our data suggest that the in vitro administration of all three anti-CD3 antibodies induce the upregulation of several cytokine genes, including INFG, IL17A, IL17F, LIF and TNF (Figure 4). However, when we analyzed the expression of IL17 in human donors by RT-qPCR, we noticed that even though the IL17A gene expression was consistently expressed along all treatments in the NGS panel, its induction was variable among antibody-treated donor T cells (Figure 5A). The FvFc R and OKT3 treatment also induced upregulation of IL6 and IL32. OKT3 treatment induced additional interleukins such as IL1B, IL2, IL3, IL9, IL13, IL12B, IL21 and IL22 (Figure 4).
Cytokine receptors were also induced after antibody treatment, including strong upregulation of the IL2 receptor subunit genes, IL2RA and IL2RB (Figure 4). IL2RA expression was also tested in the qPCR panel of treated donor T cells, suggesting that any form of anti-CD3 induces the expression of the IL2 receptor α-chain, also known as CD25 (Figure 5B). Moreover, all antibody treatments induced the expression of IL1R2, IL12RB2, IL18R1, IL21R, IL23R (Figures 4 and 5C). However, as suggested by the NGS panel, anti-CD3-treated T cells increased their sensitivity toward IL1, IL2, IL12, IL18, IL21 and IL23.
Anti-CD3 antibody treatment induced the upregulation of several interleukin and interleukin receptors genes, but only a few interleukins and receptors were downregulated due to antibody treatment. IL10 and IL24 expression was significantly repressed after OKT3 and FvFc R treatment, while IL18BP was repressed by OKT3 and FvFc M. In addition, OKT3 treatment also reduced the expression of IL18 (Figure 4). IL10 was further investigated by qPCR. Notwithstanding, the qPCR panel suggested that OKT3 treatment had a variable effect on IL10 expression among treated donor cells, and the FvFc-based antibody had no significant effect (Figure 5D).
Downregulation of interleukin receptors makes T cells less sensitive to their cognate cytokine. The NGS panel suggested that OKT3 treatment might interfere with signaling of interleukins IL10, IL11 and IL13, due to the downregulation of IL10RA, IL10RB, IL11RA and IL13RA1 (Figure 4). IL6R was downregulated after treatment with OKT3 and FvFc R. The IL17RA codes for IL17A specific receptor and was found to be downregulated after OKT3 treatment, with a barely significant q-value (0.0069); nevertheless, the qPCR panel confirmed this tendency for downregulation after treatment with any of the antibodies (Figure 5E). The IL17RC gene, which codes for a receptor for both IL17A and IL17F, was found to be downregulated after both OKT3 and FvFc M treatment. The receptor for IL7, IL7R, was shown to be downregulated with both FvFc R and OKT3 treatment. The qPCR panel corroborated these results, suggesting that most donor T cells respond to any anti-CD3 antibody format, reducing the IL7R expression levels (Figure 5F).
Anti-CD3 stimulation regulates phenotypic marker genes
Activation of resting T cells by anti-CD3 antibodies can induce cell differentiation, and indeed, several phenotypic markers are modulated after antibody treatment. Resting T cells can differentiate in several lineages of effector and regulatory phenotypes, and specific genetic markers can characterize these T cell phenotypes. We compared several markers for CD4 and CD8 subpopulations depicted as panels to visualize their possible differentiation (Figure 6). To confirm prototype marker expression levels found in the NGS panel, qPCR analyses were performed using anti-CD3 treated T cells (Figure 7). Some expression markers are key for charactering T cell subpopulations. The Th1 marker TBX21, which codes for the TBET transcription factor, was shown to be significantly induced only with OKT3 treatment in the NGS panel (Figure 4). The qPCR panel corroborated the NGS data (Figure 7A), suggesting a minimal effect of FvFc antibodies on TBX21 expression. STAT4, another Th1 marker, was also only induced by OKT3 in the NGS experiment, but qPCR data suggests that FvFc R could also affect the expression levels of STAT4 in stimulated cells20,21 (Figures 4 and 7B). GATA3, a Th2 phenotypic marker, was not significantly induced in NGS or qPCR data (Figures 4 and 7C). However, other characteristic markers of this subtype were induced21,22 (Figure 6).
In addition, we also analyzed markers for the Th17 subpopulation23,24 (Figures 4, 5, 6 and 7). The gene that codes for RORϒt, RORC was found to be slightly upregulated after treatment with both OKT3 and FvFc M antibodies (Figure 4), but without significance (padj > 0.01). In the qPCR panel, RORC was shown to be barely activated in all three treatments (Figure 7D). IL17A, known to be produced by Th17 cells, was upregulated in the NGS panel, but these data were not supported by qPCR, which suggests a variable and mild regulation of this gene (Figure 5A). The third marker, STAT3, was found to be induced by OKT3 in the NGS data and was induced by OKT3 and FvFc R treatments, as measured by qPCR (Figure 7E). Interestingly, the FvFc M antibody induced a very contrasting effect on different donors. Half of the donors showed an upregulated profile, while the other half showed a downregulated profile.
T cells can assume a regulatory phenotype, and many regulatory markers were found in this analysis25,26 (Figure 6). FOXP3, a major transcription factor that is associated with the human T regulatory phenotype, was upregulated in the NGS data for all antibody treatments. These data were corroborated by qPCR (Figure 7F). GITR (TNFRSF18) was strongly upregulated by all antibodies (approximately 16-fold, Figure 4), and this effect was also observed for all donors in qPCR (Figure 7G). CTLA4 and LAG3 were similarly upregulated in the NGS (Figure 4), and qPCR data supported this finding (Figures 7H and I), but the effect was less pronounced for FvFc antibodies compared to OKT3. The gene that codes for PD-1, namely, PDCD1, was also consistently induced by all antibody treatments (approximately 5-fold, Figure 4), and qPCR data confirmed this observation (Figure 7J).
Modulations of CD8 T cell markers were also observed after anti-CD3 treatment, suggesting changes in the CD8 T cell population27,28 (Figure 6). Among these markers, EOMES and KLRG1 were repressed after all the anti-CD3 treatments, but GMZB was strongly induced by anti-CD3. These three markers were also tested by qPCR, which confirmed the tendency of the NGS data (Figure 7K, L and M). Moreover, markers of regulatory CD8 T cell29, such as IL2RA, were sharply induced by OKT3 and FvFc R but to a lesser extent by FvFc M. CD274 (PD-L1) was only marginally induced in all treatments (Figure 6), and FOXP3 showed a variable profile (Figure 7F). ICOS was weakly induced only by OKT3 and FvFc M (Figure 4).
Phenotypic markers associated with T cell activation, cell death and apoptosis pathways were also affected by anti-CD3 treatment. Figure 4 resumes the induction/repression of these markers after anti-CD3 treatment. Overall, OKT3 induced most activation markers except EOMES and AIF1, while FvFc-based antibodies had a milder profile. Among the activation molecules, IFGN, GZMB, IL2RA, TNFRSF4 and TNFRSF9 showed remarkable induction. Cell death was the fate of activated cells, and the FAS/FASLG pathway was induced after T cell activation. The anti-CD3 effect on FAS induction was slight (Figure 6) and variable among donors (Figure 7N), and FASLG was very consistent among donors with the treatment of FvFc R (Figure 7O). GITR, along with PDCD1, was consistently induced by all the treatments (Figures 4, 6, 7G and J).
Anti-CD3 stimulation modulates genes that encode nuclear receptor transcription factors
Nuclear receptors integrate a family of transcription factors that respond to hormones and hydrophobic molecules that have been associated with the control of the immune response30. Thus, the PFAM family for Nuclear Receptor (PF00104), was used to probe antibody-induced DEGs. Anti-CD3 treatment induced the expression of PF00104-associated genes. OKT3 induced 7 genes, while FvFc R induced 3 and FvFc M induced 2 genes. The orphan nuclear receptor gene NR4A1 was activated in all treatments at a padj < 10-5. Three other PF00104 annotated genes were found in two of three treatments: NR4A3, RORC, and VDR (Figure 4). NR4A3 codes for a mitogen-associated nuclear receptor (http://www.uniprot.org/uniprot/Q92570). RORC is mentioned above as a marker for lymphocyte lineages. VDR codes for the vitamin D3 receptor, and its overexpression was detected in all antibody treatments by qPCR (Figure 7P).
Among the downregulated DEGs, peroxisome proliferator-activated receptor gamma (PPARG), a gene associated with the development of Tregs, was found to be 4- to 9-fold less expressed than that in the unstimulated T cells (Figure 4). Moreover, the THRA gene that codes for thyroid hormone receptor alpha was also repressed in all treatments.
Effect of an exclusive anti-CD3 stimulation
To compare the global gene expression profile under the effect of anti-CD3 antibodies with that of activated T cells, we paralleled our results with those described by Zhao and colleagues (2014), who probed DEGs of immortalized T cells cultured in the presence of anti-CD3 and anti-CD28 antibodies. Their DEG dataset after 72 hours treatment was compared with NGS data generated in the present work focusing on DEGs regulated after anti-CD3 treatment without the costimulatory anti-CD28 stimulus. Among the 12 most opposite DEG (Supplementary Table 3), three genes were selected for qPCR analysis: AIF1, XCL1 and IDO1 (Figures 7Q, R and S). XCL1 and IDO1 were induced by all of the anti-CD3 treatments, as observed in the qPCR panel, while AIF1 was found to be repressed after anti-CD3 treatment.