Rosetting T cells in Hodgkin lymphoma are activated by immunological synapse components HLA class II and CD58

A unique Hodgkin lymphoma (HL) characteristic is the tight adherence of rosetting CD4⁺ T cells to Hodgkin tumor cells.^1,2  Rosetting T cells are extremely important for tumor cell survival.^3,4  Previous studies have described the involvement of several adhesion molecules in HL rosette formation.^5-7  These molecules are important components of the immunological synapse (IS), which enables antigen-dependent communication between T cells and antigen-presenting cells (APCs). Adhesion molecule pairs, such as CD2-CD58 and CD11a-CD54, induce a transient adhesion of T cells to APCs, after which the T-cell receptor (TCR) scans the cell surface for antigens presented in the context of human leukocyte antigen (HLA) molecules. Upon antigen recognition, F-actin polymerization enables stabilization of the mature IS with a central region that contains the TCR-HLA molecules and interacting adhesion molecules in the periphery.^8-10  IS formation and the specific role of HLA-II-TCR interaction has never been studied in HL. To address this, we developed a novel short-term in vitro coculture model system with peripheral blood mononuclear cells (PBMCs) that are matched to the HLA-II type of HL cell lines.

Conditions for the short-term in vitro model system were optimized by using HLA-II unmatched PBMCs (supplemental Methods, available on the Blood Web site). This approach with unmatched PBMCs is consistent with most previous studies, but has the disadvantage that T-cell responses are largely attributable to alloreactivity.^6,11  We used HLA-II–matched PBMCs to more closely resemble the in vivo situation (supplemental Tables 1 and 2). HL cell lines were cocultured with PBMCs at a ratio of 1:20. After incubation, cytospins were made (at 30 minutes or 2 hours) or supernatant was collected (at 18 hours). Rosettes were defined as tumor cells with 3 or more adherent lymphocytes and were quantified. The interleukin-2 (IL-2) concentration in supernatants was used as a measure of T-cell activation. Localization of IS molecules was studied by immunofluorescence. Results were statistically significant at P < .05. Additional details are provided in the supplemental Methods.

HL cell lines were characterized for cell-surface expression of IS components HLA-I, HLA-II, CD58, and CD54 (supplemental Figure 1). Expression of CD54 was observed in all HL cell lines, whereas HLA-II was absent only in DEV and HLA-I in both DEV and L428. There was no expression of CD58 in KMH2, SUPHD1, and DEV; low expression in L540 and HDLM2 (referred to as CD58⁻); intermediate expression in L591; and high expression in L1236 and L428 (referred to as CD58⁺). Lack of HLA-II or CD58 has been described in cell lines and in advanced-stage or relapsed disease, possibly because the tumor cells progress to become independent of the interaction with rosetting T cells.^12-14 

CD58⁺ HL cell lines had a significantly higher percentage of tumor cells with rosettes compared with CD58⁻ cell lines (Figure 1A). The rosetting cells were mainly CD4⁺ T cells (supplemental Figure 2). In addition, T cells in coculture with CD58⁺ HL cell lines showed a clear induction of IL-2, whereas T cells cocultured with CD58⁻ HL cell lines did not (Figure 1B). Thus, expression of CD58 plays a role in both T-cell adhesion and activation.

To further study the roles of CD58, HLA-II, and CD54, we generated knockouts (KOs), by using the CRISPR-Cas9 approach in the HLA-I–deficient L428 cell line (supplemental Figure 3). In comparison with wild-type (WT), the percentage of tumor cells with rosettes decreased significantly after CD58 KO (20% to 25%), whereas HLA-II downregulation (generated by class II transactivator knockout [CIITA-KO]) or CD54-KO did not affect rosette formation (Figure 1C). However, activation of rosetting T cells was dependent on both HLA-II and CD58, as interaction with CIITA-KO or CD58-KO cells significantly reduced IL-2 production compared with WT cells (Figure 1D). The KOs did not affect the T-cell composition of the rosettes (supplemental Figure 4A). In accordance with the CD58-KO results, addition of a CD2-blocking antibody to cocultures with WT cells reduced both the percentage of tumor cells with rosettes and IL-2 production (supplemental Figure 4B-C).

Next, IS assembly was studied, focusing on relocalization of IS components and F-actin polymerization. CD58⁺ HL cell lines induced a distinct relocalization of CD2 on most of the rosetting T cells to the physical interface with the tumor cells, which was not observed in response to CD58⁻ HL cell lines or CD58-KO cells (Figure 1E-F; supplemental Figure 4D). There was hardly any relocalization of CD11a and TCR, whereas F-actin polymerization was prominent. This observation is in line with those in previous studies of normal IS assembly, reporting fast and massive relocalization of CD2 towards the APC, whereas the extent of TCR relocalization is less pronounced.^10,15  Because CD2 is not involved in the earliest steps of T-cell activation,^10,16-19  we suggest that activation of rosetting T cells in our model occurs in 2 steps: engagement of the TCR with antigen-HLA-II complexes which initiates activation, followed by signal amplification through CD2-CD58. Because we used HLA-II–matched PBMCs, the observed T-cell activation should be the result of specific TCR-mediated recognition of the HLA-antigenic peptide complex.

We next used proximity ligation assays (PLAs) to confirm physical interactions between CD2-CD58 and TCR-associated CD4-HLA-II. Clear interaction signals at the IS were observed between CD2 on T cells and CD58 on WT, CD54-KO, and CIITA-KO L428 cells, but not on CD58-KO L428 and CD58⁻ KMH2 cells (Figure 2A-B). Protein-protein interactions were also seen between CD4 and HLA-II at the IS with WT, CD58-KO, or CD54-KO L428 cells and, as expected, not with the CIITA-KO L428 and HLA-II⁻ DEV cells (Figure 2A,C). Extension of our study to primary HL tissue showed results compatible with the in vitro model, that is, a strong rim of CD2 staining in T cells at the tumor cell interface, whereas this was not observed for the TCR (Figure 2D). The PLA assays in HL tissue showed clear interaction signals between CD2 and CD58 and between CD4 and HLA-II at the interface between T cells and tumor cells (Figure 2E).

Taken together, these results show that rosetting CD4⁺ T cells in HL are activated after formation of a functional IS, with essential roles for interaction between TCR-HLA-II and CD2-CD58. A large proportion of the rosetting cells are CD4⁺ T cells. The previously shown polyclonal nature of these T cells suggests that many different antigenic peptides are being recognized, which is consistent with the high mutational burden in HL.^20-22  Although the rosetting T cells are activated, they do not expand, because of immune regulatory mechanisms exerted by the tumor cells, such as expression of immune checkpoint molecules and production of immunosuppressive cytokines.^3,4,23  Instead, they become anergic, provide growth stimuli for the tumor cells and form a physical barrier between the tumor cells and other, more hostile immune cells, such as cytotoxic T cells and NK cells.^24,25  This knowledge of the cross talk of HL cells with their immediate neighbors provides a better understanding of early lymphoma development and may help improve immunotherapy in HL.

J.V. was funded by the Graduate School of Medical Sciences, University of Groningen, University Medical Center Groningen, Groningen, The Netherlands. The modified lentiCRISPR v2 vector was the kind gift of Ryan O’Connell (University of Utah, Salt Lake City, UT).

Contribution: J.V., A.D., L.V., and A.v.d.B. designed the study and wrote the manuscript; B.H. and M.H.-K. performed the HLA typing and selected the donors; J.V., N.M., and L.V. performed the in vitro experiments; and J.V. and L.V. performed the data analysis.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Arjan Diepstra, University Medical Center Groningen, Hanzeplein 1, code EA10, PO Box 30.001, 9700 RB Groningen, The Netherlands; e-mail: a.diepstra@umcg.nl.