Progression of AITL-like tumors in mice is driven by Tfh signature proteins and T-B cross talk

Abstract

Angioimmunoblastic T-cell lymphoma (AITL) is an aggressive peripheral T-cell lymphoma driven by a pool of neoplastic cells originating from T follicular helper (Tfh) cells and concomitant expansion of B cells. Conventional chemotherapies for AITL have shown limited efficacy, and as such, there is a need for improved therapeutic options. Because AITL originates from Tfh cells, we hypothesized that AITL tumors continue to rely on essential Tfh components and intimate T-cell-B-cell (T-B) interactions. Using a spontaneous AITL mouse model (Roquinsan/+ mice), we found that acute loss of Bcl6 activity in growing tumors drastically reduced tumor size, demonstrating that AITL-like tumors critically depend on the Tfh lineage-defining transcription factor Bcl6. Because Bcl6 can upregulate expression of signaling lymphocytic activation molecule-associated protein (SAP), which is known to promote T-B conjugation, we next targeted the SAP-encoding Sh2d1a gene. We observed that Sh2d1a deletion from CD4+ T cells in fully developed tumors also led to tumor regression. Further, we provide evidence that tumor progression depends on T-B cross talk facilitated by SAP and high-affinity LFA-1. In our study, AITL-like tumors relied heavily on molecular pathways that support Tfh cell identity and T-B collaboration, revealing potential therapeutic targets for AITL.

Graphical abstract

Figure 1.

An increase in plasmablasts in tumor lymph nodes of Roquin^san/+mice. (A) Pictogram illustrating differences between tumor-free and -bearing mice, as well as providing a pictorial explanation of tumor-free, nontumorous (non-tumor), and tumorous (tumor) lymph nodes. (B-D) Representative flow cytometric analyses and frequencies of B-cell populations using the surface markers Fas and GL7 (n = 10 tumor-free; n = 11 nontumorous; and n = 15 tumorous samples). (E-F) Representative flow cytometry plots and histograms comparing the frequency of plasmablasts (n = 5 tumor-free, n = 6 nontumorous; and n = 9 tumorous samples). (G-I) Expression profile of Bcl6 and IRF4 in GL7^hiFas⁺ B cells as depicted through representative flow cytometry plots and frequencies (n = 5 tumor-free; n = 5 nontumorous; and n = 9 tumorous samples). Error bars in panels C-D,F,H-I represent the standard error of the mean (SEM). Data are pooled from at least 2 independent experiments. *P < .05; **P < .01; ***P < .001; ****P < .0001. ns, not significant.

Figure 2.

Disruption of functional Bcl6 gene in growing AITL-like tumors leads to tumor regression. (A) Exemplary immunohistochemistry sections of AITL patient samples depicting CD3 and Bcl6 expression. (B) Representative flow cytometric plots showing subsets of CD4 cells based on expression of PD-1 and CXCR5. (C) Comparing frequencies of CD4⁺CXCR5⁺PD-1⁺ cells between tumor-free, nontumorous (non-tumor), and tumorous (tumor) samples (n = 11 tumor-free; n = 15 nontumorous; and n = 21 tumorous samples). (D) MFI of CXCR5 and PD-1 on CD4⁺CXCR5⁺PD-1⁺ cells (n = 11 tumor-free; n = 15 nontumorous; and n = 21 tumorous samples). Bcl6 expression in CXCR5⁻PD1⁻, CXCR5⁺PD-1⁻, and CXCR5⁺PD-1⁺ cells from tumor samples (E) and Bcl6 MFI of CD4⁺CXCR5⁺PD-1⁺ cells (F) (n = 9 tumor-free; n = 12 nontumorous; and n = 13 tumorous). (G) Representative histogram and frequency of Ki-67 levels in tumor CD4⁺ subsets (n = 21 tumorous). (H) Roquin^san/+ mice were bred with a CD4-specific, tamoxifen-inducible Cre-recombinase to target the Bcl6 gene (Roquin^san/+; Cd4-Cre^ERT2+/−; Bcl6^f/f). Representative time course (I) and sonograms (J) demonstrating tumor regression in mice with Bcl6 gene deletion (n = 5 Roquin^san/+; Cd4-Cre^ERT2+/−; Bcl6^+/+; n = 6 Roquin^san/+; Cd4-Cre^ERT2+/−; Bcl6^f/f). Error bars in panels C-G,I represent the SEM. Data are pooled from at least 2 independent experiments. *P < .05; **P < .01; ***P < .001; ****P < .0001.

Figure 3.

Abrogation of SAP expression from AITL-like tumors leads to tumor regression. (A) Increased expression of the SLAM adaptor protein, SAP in CD4 cells from AITL-like tumors (n = 5 tumor-free; n = 8 nontumorous [non-tumor]; and n = 12 tumorous [tumor] samples). (B) Frequency and MFI of SAP in CXCR5⁺PD-1⁺ cells from tumor-free, nontumorous, and tumor samples (n = 5 tumor-free; n = 8 nontumorous; and n = 12 tumorous samples). (C) SAP^hi CD4 cells from tumors are more proliferative, as depicted through increased frequencies of Ki-67⁺ cells (n = 4 tumor-free; n = 7 nontumorous; and n = 11 tumorous samples). (D) SAP^hi CD4⁺ tumor cells are more proliferative than SAP^lo as shown in representative histogram and frequencies. (E) Roquin^san/+ mice were bred with a CD4 specific tamoxifen-inducible Cre recombinase and conditional Sh2d1a allele (Roquin^san/+; Cd4-Cre^ERT2+/−; Sh2d1a^{f/y or f/f}). Representative time course (F) and sonograms (G) demonstrating tumor regression in mice with CD4-specific abrogation of SAP (n = 5 Roquin^san/+; Cd4-Cre^ERT2+/−; Sh2d1a^+/+; n = 9 Roquin^san/+; Cd4-Cre^ERT2+/−; Sh2d1a^{f/y or f/f}). In panel F, we reused CD4-CreERT2 control data shown in Figure 2I to perform statistical analysis for different time points. Error bars in panels A-D,F represent the SEM. Data are pooled from at least 2 independent experiments. *P < .05; **P < .01; ***P < .001; ****P < .0001.

Figure 4.

Loss of SAP-Fyn signaling pathway does not greatly alter AITL-like disease. (A) Abrogation of downstream signaling between SAP and Fyn kinase does not prevent tumor incidence, but may have a partial effect (n = 17 Roquin^san/+; Sh2d1a^+/y and n = 29 Roquin^san/+; Sh2d1a^R78A/y). (B-E) Tumors from Roquin^san/+; Sh2d1a^+/y or Roquin^san/+; Sh2d1a^R78A/y show comparable T-B-cell expression patterns and frequencies (n = 3 Roquin^san/+; Sh2d1a^+/y tumors and n = 7 from Roquin^san/+; Sh2d1a^R78A/y tumors). Error bars in panels B-E represent the SEM. Data are pooled from at least 2 independent experiments. **P < .01.

Figure 5.

Elevated high-affinity LFA-1 and ICAM-1 levels in Roquin^san/+tumors. (A) Tumor samples have increased frequency of high-affinity LFA-1. Representative flow cytometric analyses and frequency of high-affinity LFA-1 expression on CD4⁺ cells from recombinant ICAM-1 binding assay (10 μg/mL of murine recombinant ICAM-1; n = 5 tumor-free; n = 4 nontumorous [non-tumor]; and n = 4 tumorous [tumor] samples). (B) Concomitant increased expression of ICAM-1 on B220⁺ cells in Roquin^san/+ tumors as shown in representative flow cytometric analyses and frequencies. (B-C) MFI of ICAM-1 on B-cell subsets (n = 5 tumor-free; n = 5 nontumorous; and n = 7 tumorous samples). (D) Incubating wild-type Peyer’s patch cells with Lovastatin during ICAM-1 binding assay verifies reduction in high-affinity LFA-1 levels on CD4⁺ T cells with minimal differences in cell viability (n = 5, wild type). (E) Left graph shows a time course of change in tumor size after treatment with lovastatin (2 mg/kg, intraperitoneal, every other day for 2 weeks) compared with the untreated control group. Lovastatin-treated mice were further subgrouped based on the following criteria: no response (<25% reduction in area or growth), partial response (25% to 60% reduction in area), or full response (>60% reduction in area) at the final time point (week 6). The right graph further compares each response type with the untreated control group. (F) Exemplary regressed tumor is shown in sonograms. Error bars in panels A-E represent the SEM. Data are pooled from at least 2 independent experiments. *P < .05; **P < .01; ***P < .001; ****P < .0001.