Reconstitution of EBV-directed T cell immunity by adoptive transfer of peptide-stimulated T cells in a patient after allogeneic stem cell transplantation for AITL

Reconstitution of the T cell repertoire after allogeneic stem cell transplantation is a long and often incomplete process. As a result, reactivation of Epstein-Barr virus (EBV) is a frequent complication that may be treated by adoptive transfer of donor-derived EBV-specific T cells. We generated donor-derived EBV-specific T cells by stimulation with peptides representing defined epitopes covering multiple HLA restrictions. T cells were adoptively transferred to a patient who had developed persisting high titers of EBV after allogeneic stem cell transplantation for angioimmunoblastic T-cell lymphoma (AITL). T cell receptor beta (TCRβ) deep sequencing showed that the T cell repertoire of the patient early after transplantation (day 60) was strongly reduced and only very low numbers of EBV-specific T cells were detectable. Manufacturing and in vitro expansion of donor-derived EBV-specific T cells resulted in enrichment of EBV epitope-specific, HLA-restricted T cells. Monitoring of T cell clonotypes at a molecular level after adoptive transfer revealed that the dominant TCR sequences from peptide-stimulated T cells persisted long-term and established an EBV-specific TCR clonotype repertoire in the host, with many of the EBV-specific TCRs present in the donor. This reconstituted repertoire was associated with immunological control of EBV and with lack of further AITL relapse.


Part II -Major Issues
"1-The expanded T cells were directed against antigens from BZLF1, EBNA3A and EBNA1. In vivo, BZLF1-specific T cells expanded more that the EBNA-1 specific T cells (which was not the case ex vivo). Did the AITL at relapse (or at diagnosis) express the targeted proteins? Not all AITL express latency type 3 proteins and, as such, the AITL expression of BZLF1 should be confirmed (especially in a context in which AITL relapse clearly appears to be the cause of the rising EBV viral load)? Otherwise, what can account for this difference (exhaustion markers expression, poor functional features, immunodo minance, etc)? Additional characterization of the product along these lines would improve the manuscript." To verify EBV expression of AITL, we used EBER in situ hybridization (IsH) of the lymph node at diagnosis. Only very rarely EBER + cells were seen (new S2A Fig). AITL is often accompanied by activated EBV + B cells in the tumor microenvironment; however, we did not observe it in this case (new S2B Fig.). Unfortunately, no lymph node biopsies from relapse were available for further analysis, as this relapse appeared to be leukemic.
Adoptive transfer of EBV-specific T cells was given as a curative treatment specifically for high EBV viremia . The leukemic relapse after transplant was initially treated with a conventional DLI; however, the EBV load was high and associated with severe symptoms. In this study, we decided to focus on the expansion and in vivo survival of EBV-specific T cells to treat EBV reactivation post-transplant. The fact that the AITL was EBVindicates there was a separate infection site which produced the EBV viremia. However, simultaneous AITL relapse and increased EBV viral titers in blood suggest a strong association between AITL and EBV infection.
We feel that potential mechanisms to expla in differences in expansion between BZLF1-and ENBA1-specific T cells are their differentiation stages and the availability of their target epitope. Although cells for the 3 specificities mostly lack CCR7 expression (CCR7 -CD45RA -), both lytic BZLF-1 EPL-and RAK-specific T cells had a stronger CD62L expression than latent EBNA1 HPV-specific T cells (new S5 Fig). This difference and CD62L association with central memory may indicate a higher proliferative strength, a stronger expansion (as seen in Fig 3C), less exhaustion, and better homing ability of the lytic-specific compartment. Otherwise, an alternative hypothesis to the expansion difference between lytic and latent-specific cells is availability of target epitope: The strong expansion of EPL-and RAK-specific T cells in peripheral blood and cytokine release (new S4 Fig) happened directly after transfer, when ongoing EBV viremia may have provided enough lytic epitope presentation. EBV viremia decrease over time and lesser epitope availability would explain for few clonotypes to remain and survive long-term. On the other hand, HPV-specific T cells strongly expanded ex vivo but not as strongly in peripheral blood in vivo, either due to its homing to an active infection site, such as a lymph node, or the unavailability of its target antigen EBNA1.
We have included these points in the Results ( "2-The clonotype composition of sorted cells did show significant overlap between Ag -specific populations. Although resolved by the application of "filters," doubts remain regarding the true Agspecificity of the sorted cells and whether sorting was stringent enough. Functional (cytokine secretion, degranulation) tests using sorted cells in the presence of their target peptides vs non-target peptides would convince further." Our wording in the manuscript text may have suggested that the overlap between the three sorted populations was larger than it actually was; however, these overlapping clonotypes had only minor frequencies, compared to very dominant clonotypes per specificity. As an example, HPV-specific TCR VJ-4001.53.1 had a 38.8% frequency in HPV-sorted cells, compared to 0.1% in EPL and 0.2% in RAK. For this reason, we believe that the composition of each epitope-specific compartment was dominated by higher-frequency clonotypes, while we can always expect to find cells with very low frequencies as contaminants.
To confirm this point, we have included in the revised manuscript: • Multimer sorting purity: Gating for multimer sorting was stringent enough to achieve an efficiency above 98% for all epitopes (new S6 Fig). We used multimer sorting in the article as party of quality control to analyze the epitope-specific compartment in greater detail. This was not used as part of the manufacturing protocol, as multimer binding might alter T cell functionality and lead to artificial activation. For this reason, the functional status of the sorted populations wasn't explored. Instead, we showed functionality of the T cell product by restimulation with EPL, RAK, and HPV peptides, which lead to IFN-γ production as an equivalent (Fig 1C).

Part III -Minor Issues
"1-The description of the case. The report is at times confusingto help with the flow of the manuscript, the content of Table S1 should be described at the beginning and a timeline provided (time from chemotherapy to transplant unclear, various treatments, etc) in the body of the manuscript." We thank the reviewer for this suggestion: We have rearranged the manuscript to start with the description of the case in Results (lines 98-109), a new figure with histological information (S2 Fig), and a timeline (S3 Fig) as a supplement to S1 Table. We cannot show particular dates of chemotherapy due to privacy regulations; however, we have assigned the day of allogeneic transplantation as day 0 and calculated chemotherapy dates accordingly.

"2-The occurrence of severe GVHD after adoptive transfer should be discussed (role of DLI, ATCT?, impact of steroid-therapy on reconstitution and in this case, the occurrence of a fatal infectious outcome). Likewise, DLI and Rituximab seem to have had a greater therapeutic impact than AT CT (not clear that it was administered in the context of a rising EBV viral load). The potential implications of that should be discussed as well."
Thank you for having raised this issue. GvHD is an important, yet slightly confusing point, which we deliberately left out in the initial manuscript. Given that both reviewers asked for this information, we have added a section in the Discussion (lines 378-389) and a new figure (S9 Fig) and hope this clarifies the issue.
We expected a decrease in EBV viremia after DLI and Rituximab infusion; however, EBV viremia kept increasing until it reached its peak simultaneously with the last dose of Rituximab on day 89 and EBV-related symptoms remained. While EBV titers started declining after this point, we saw an expansion of the EBVspecific T cells and cytokine release directly after adoptive transfer. This expansion points to the availability of their target antigens still being present. We altered Fig.3A to reflect more precisely the days of Rituximab infusion and have included these points in the Discussion (lines 352-356).

"3-Fig 4 shows that the donor PBMCs contain several of the abundant product clonotypes. Although the author provides a rationale to link ATCT to EBV-specific immune reconstitution, they do not rule out the possibility that clonotypes in the DLI contributed to the immune reconstitution. This should be mentioned and discussed."
Thank you for raising this important point. DLI may indeed have contributed to the surviving EBV memory T cell pool and we cannot formally exclude this. However, we were able to track EBV-specific T cell clonotypes coming from the adoptive transfer that persisted long-term in the patient.
We included these points in the Discussion (lines 352-356).

"4-The EBV viral load measurement method is not specifiedplasma/serum or whole blood? Commercial assay or LDT?"
We have included this information in the Materials and Methods section (lines 450-454).

"5-As opposed to what is described in the text, Fig S1 does not show data on IFNg."
We apologize for this error. We have now altered the figure to show cytokine secretion as S3 Fig and rearranged the other supplementary figures accordingly.

Part I -Summary "The authors demonstrate that Epstein Barr virus (EBV) derived peptide stimulation of peripheral blood mononuclear cells (PBMCs) derived from a patient with recurrent angioimmunoblastic T cell lymphoma
(AITL) after transplantation leads to the expansion of EBV specific CD8 + T cells. These are primarily directed against peptides from BZLF1 and EBNA1. Particularly EBNA1 specific CD8 + T cells were dominated by one TCR clonotype. Several of the identified EBV specific TCR clonotypes expand after transfer into the patient and are maintained for at least 8 months. From these data the authors conclude that a diverse set of EBV specific clonotypes can be expanded with their protocol and transferred for treatment of AITL, reestablishing EBV specific immune control after stem cell transplantation. Med 1996). Therefore, the authors should address in more detail how the adoptively transferred T cell products control the pathogenic T cell expansion in this patient and if there are differences between the detected EBV specific TCR clonotypes."

Even so the authors apply TCR clonotype tracing to follow adoptively transferred EBV specific T cells in one bone marrow transplant patient for the first time, it is unclear what new insights they gain from these studies. Persistence of such adoptively transferred T cells has previously been documented for up to 18 months (Heslop et al., Nat
We thank the reviewer for pointing this out. The novel aspects of this article are the molecular characterization at a clonotype level of adoptively-transferred EBV-specific T cells against selected epitopes of known HLArestriction Therefore, we have emphasized these aspects of this article in the Abstract (lines 30-33, 36-37), Introduction (lines 90-95), and Discussion (lines 330-337).
The a doptive transfer of EBV-specific T cells was decided as a curative treatment for high EBV viremia and was not aimed to control AITL. However, simultaneous AITL relapse and increased EBV viral titers in blood suggest a strong association between AITL and EBV, as mentioned in the literature.
We feel potential mechanisms to explain differences in expansion between BZLF1 -and EBNA1-specific T cells are their differentiation stages and availability of their target epitope. Although cells for the 3 specificities mostly lack CCR7 expression (CCR7 -CD45RA -), both lytic BZLF-1 EPL-and RAK-specific T cells had a stronger CD62L expression than latent EBNA1 HPV-specific T cells (new S5 Fig). This difference and CD62L association with central memory may indicate a higher proliferative strength, a stronger expansion (as seen in Fig 3C), less exhaustion, and better homing ability of the lytic-specific compartment. Otherwise, an alternative hypothesis to the expansion difference between lytic and latent -specific cells is availability of target epitope: The strong expansion of EPL-and RAK-specific T cells in peripheral blood and cytokine release (new S4 Fig) happened directly after transfer, when ongoing EBV viremia may have provided enough lytic epitope presentation. EBV viremia decrease over time and lesser epitope availability would explain for few clonotypes to remain and survive long-term. On the other hand, HPV-specific T cells strongly expanded ex vivo but not as strongly in peripheral blood in vivo, either due to its homing to an active infection site, such as a lymph node, or the unavailability of its target antigen EBNA1.