CD8+γδ T Cells Are More Frequent in CMV Seropositive Bone Marrow Grafts and Display Phenotype of an Adaptive Immune Response

The role of gamma delta (γδ) T cells in human cytomegalovirus (HCMV) immune surveillance has been the focus of research interest for years. Recent reports have shown a substantial clonal proliferation of γδ T cells in response to HCMV, shedding light on the adaptive immune response of γδ T cells. Nevertheless, most efforts have focused on Vδ2negγδ T cell subset while less attention has been given to investigate other less common γδ T cell subsets. In this regard, a distinct subpopulation of γδ T cells that expresses the CD8 coreceptor (CD8+γδ T cells) has not been thoroughly explored. Whether it is implicated in HCMV response and its ability to generate adaptive response has not been thoroughly investigated. In this study, we combined flow cytometry and immune sequencing of the TCR γ-chain (TRG) to analyze in-depth bone marrow (BM) graft γδ T cells from CMV seropositive (CMV+) and CMV seronegative (CMV-) donors. We showed that the frequency of CD8+γδ T cells was significantly higher in CMV+ grafts compared to CMV- grafts (P < 0.001). Further characterization revealed that CD8+γδ T cells from CMV+ grafts express Vγ9− and preferentially differentiated from a naive to terminal effector memory phenotype (CD27low/-CD45RO−). In line with these findings, TRG immune sequencing revealed clonal focusing and reduced usage of the Vγ9/JP gene segment in a CMV+ graft. Furthermore, CD8+γδ T cells showed an enhanced response to TCR/CD3 and cytokine stimulation in contrast to CD8−γδ T cells. We conclude that γδ T cells in BM grafts are reshaped by donor CMV serostatus and highlight the potential adaptive role of CD8+γδ T cells in HCMV immune response.


Introduction
Human cytomegalovirus (HCMV) is a DNA virus that belongs to the β-herpes virus family [1]. In immunocompetent individuals, HCMV establishes a lifelong latent infection that is usually asymptomatic. However, in conditions where the immune system is dampened, such following allogeneic Hematopoietic Cell Transplantation (HCT), HCMV can be life-threatening, rendering CMV infection/reactivation a major cause of morbidity and mortality after HCT [2].
Human γδ T cells are unconventional T cells that express a T cell antigen receptor (TCR) formed by γ and δ chains and fundamentally differ from αβ T cells in their major histocompatibility complex-(MHC-) independent antigen recogni-tion [3]. In allogeneic HCT, γδ T cell reconstitution occurs shortly after transplantation [4], a process that has been associated with a favorable outcome, indicating their crucial role in protection against tumors and pathogens [5][6][7].
The role of γδ T cells in HCMV immune surveillance has been shown previously [8]. However, the underlying immune mechanism and the ligand/s mediating γδ T cell activation are poorly understood [8,9]. Furthermore, whether γδ T cells respond to HCMV through innate or adaptive immune pathways is unclear. Vγ9 + Vδ2 + cells express a semi-invariant TCR and respond to a limited range of nonpeptide antigens such as phosphoantigens, rendering their response innate-like in nature. In contrast, Vδ2 neg γδ T cells have a wider range of ligands and display high diverse TCR repertoire at birth that become focused at adulthood, sharing more properties of adaptive immunity [10,11].
HCMV infection is associated with a remarkable proliferation of Vδ2 neg γδ subsets, particularly Vδ1 + cells [1]. Recently, next-generation sequencing of the TCR chains δ (TRD) and γ (TRG) has allowed an in-depth analysis of the γδ TCR repertoire reshaping in response to HCMV. Using this state-of-the-art technique, Ravens et al. and Davey et al. have revealed for the first time CMV-associated clonotypic changes in the γδ TCR repertoire [12][13][14]; their reports provide strong evidence for the ability of γδ T cells to mount a virus-specific nonconventional adaptive immune response.
The majority of human adults circulating γδ T cells are double negative for CD8 and CD4 coreceptors (CD4 -CD8 -), partially accounting for their MHC independence [13]. However, a small subpopulation of γδ T cells expresses the CD8 coreceptor (CD8 + γδ T cells). Reports from several research groups including ours suggested distinct immunobiology of this subset [15,16]. In the context of allogenic HCT, the role of this subset in HCMV infection has not yet been fully described. Whether CD8 + γδ T cells undergo clonal proliferation in response to HCMV and if they are capable of mounting adaptive function has so far not been shown. It is therefore fundamental to address their potential role in CMV immune response. In this study, we characterized γδ T cells in BM grafts from CMV+ and CMV-donors using multicolor flow cytometry in addition to immune sequencing of the TCR γ chain (TRG).

Donor Characteristics and Ethical Approval.
A total of 16 samples (13 males and 3 females) were obtained from BM grafts before allogeneic HCT at the Cell Therapy and Allogeneic Stem Cell Transplantation (CAST), Karolinska University Hospital, Sweden. Out of 16 donors, 7 were CMV seropositive (CMV+) and 9 were CMV seronegative (CMV-). The median age of the donors was 28 and 22 years for CMV+ and CMV-donors, respectively. Written informed consent for sample collection and subsequent analysis was provided. The study was approved by the regional ethical review board in Stockholm (2008/206-31, 2010/760-31/1, 2013/2215-32, and 2017/469-32).

Sample Preparation.
Mononuclear cells (MNC) were freshly isolated from BM grafts by density gradient centrifugation (Lymphoprep, Fresenius Kabi, Oslo, Norway) as described previously [17], were cryopreserved in RPMI-1640 media containing 10% DMSO and supplemented with 10% human AB serum, and were stored in liquid nitrogen freezer until time of analysis.
Manual gating was used to characterize individual samples, and subsequently, data were downsampled and merged (concatenated) for further visualization using dimensionality reduction algorithm plugin, t-Distributed Stochastic Neighbor Embedding (tSNE).

γδ Genomic DNA Extraction and Immunosequencing.
MNCs from one CMV+ and one CMV-BM grafts were thawed, γδ T cells were sorted using the TCR γ/δ T cell isolation kit (Miltenyi Biotec) according to the manufacturer's protocol, and γδ purity was confirmed by FACS. Next, genomic DNA was extracted using the EZ1® DNA Blood Kit and EZ1 instruments (Qiagen, Germany) according to the manufacturer's instructions. Concentration and purity of eluted DNA were analyzed using NanoDrop 2000 (Thermo Fisher Scientific), and DNA samples were stored at -20°C. An average of 1 μg of genomic DNA was used for high throughput sequencing of the CDR3 region of TRG using the Immuno-SEQ platform (Adaptive Biotechnologies, Seattle, WA) as described previously [19]. Briefly, amplification of V-J segments was performed in a bias-controlled multiplex PCR reaction using primers specific for Vγ and Jγ gene segments. A specific algorithm was then applied to correct for sequencing error. CDR3 segments were annotated according to the International ImmunoGeneTics collaboration.
2.5. TCR γδ CDR3 Spectratyping. CD8 + and CD8γδ T cells were sorted from CMV+ grafts on a cell sorter (Sony MA900, Sony Biotechnology Inc.). RNA was extracted (All-Prep DNA/RNA Mini Kit, Qiagen, Germany) and immediately converted to cDNA (SuperScript™ IV VILO™ Master Mix, Thermo Fisher Scientific) as previously described [17]. The CDR3 region for Vγ2, Vγ3, Vγ4, Vγ5, Vγ9, and Vδ1 was amplified by PCR using forward primers specific for each V gene segment and a common 5 ′ FAM-labeled reverse primer for the constant γ (Cγ) or δ (Cδ) genes as described elsewhere [20] (Table S1). PCR conditions and spectratyping were performed according to a protocol described in detail previously [16].
2.6. T Cell Culture and Proliferation Assays. CD3 + T cells from donor BM grafts were magnetically bead sorted (Pan T Cell Isolation Kit, Miltenyi Biotec) and labeled using CellTrace violet (Thermo Fisher Scientific) according to the manufacturer's instructions. Labeled T cells were cultured in a 96-well plate (1 × 10 6 cells/mL) in a complete RPMI-1640 medium (containing 10% human AB serum, 50 μg/mL penicillin/streptomycin) either alone (unstimulated) or in the presence of anti-CD3 (clone OKT3, BioLegend), IL-7, IL-15, or IL-18 at 30 ng/mL (PeproTech) and were incubated at 37°C and 5% CO 2 for 5 or 7 days (for anti-CD3 or cytokines, respectively). Cells were analyzed by FACS, and proliferating cells were defined as % of Cell-Trace violet (CTV) low cells compared to unstimulated conditions. In addition to proliferation assay, staining for activation/exhaustion surface markers (CD69, TIM3, and KIR2DL2/3) was performed.

Bioinformatics and Statistics.
Parametric test statistics were used throughout the study after confirming that assumptions of normality were not violated using the Shapiro test and Q-Q plots. When comparing two groups, the Student t-test or paired t-test was used as indicated. ANOVA followed by post hoc multiple comparisons (Tukey's correction) were used when three or more unrelated groups were compared, and repeated measures ANOVA when the compared groups were related (paired). A P value < 0.05 was considered statistically significant, and the following significance levels were used: * P < 0:05, * * P < 0:01, and * * * P < 0:001. GraphPad Prism version 6.00 for Windows (Graph-Pad Software, La Jolla, California, USA) and IBM SPSS Statistics for Windows, Version 24.0. (Armonk, NY: IBM Corp.) were used to perform statistics. The ImmunoSEQ tool was used for initial handling of sequencing data, and diversity, clonal space homeostasis, V-J segment usage, CDR3 spectratyping, and repertoire overlap were performed using specific packages as previously described [19].
To gain more insight, we used a dimensionality reduction algorithm (tSNE) to visualize clusters of Vγ9 + and Vγ9subpopulations from CMV+ and CMV-grafts. Vγ9subset represented a predominant distinct cluster of γδ T cells within CMV+ grafts compared to CMV-grafts (Figure 1(c)). Furthermore, tSNE-generated histograms from Vγ9and Vγ9 + subpopulations showed remarkable downregulation of CD45RO in Vγ9subpopulation when compared to Vγ9 + both in CMV+ and in CMV-grafts. Additionally, CD27 downregulation was more prominent in Vγ9subpopulation in CMV+ grafts (Figure 1(c)).
3.3. CD8 + γδ T Cells within CMV+ BM Grafts Express Vγ9and Preferentially Display Effector Phenotype. As CMV+ grafts showed significantly increased proportions of CD8 + γδ T cells, we sought to further characterize this subset. Interestingly, comparison between CD8 + and CD8γδ T cells revealed increased proportions of Vγ9in CD8 + γδ T cells from CMV+ grafts compared to both CD8 + and CD8γδ T cells from CMV-grafts ( Figure S1A).
Next, we investigated whether the increased frequency of CD8 + γδ T cells is linked to differentiation. Comparing the frequency of different memory phenotypes revealed increased proportions of CD27 low/-CD45RO -(TE) γδ T cells among CD8 + γδ T cells only in CMV+ grafts (Figures 3(a) and 3(b)). Importantly, the frequency of combined EM and TE phenotypes (CD27 low/-CD45RO -/ + ) was higher among CD8 + γδ T cells from CMV+ grafts as compared to either CD8 + or CD8γδ T cells from CMV-grafts (Figure 3(c)). Consistently, CD8 + γδ T cells from CMV+ grafts showed decreased proportions of naïve γδ T cells compared to CD8 + or CD8γδ T cells from CMV-grafts (Figure 3(d)). In line with this memory phenotype, γδ T cells from CMV-grafts tend to express more CCR7 + when compared to CMV+ grafts ( Figure S1B).

γδ TRG Repertoire is Clonally Focused in CMV+ Grafts.
As flow cytometry data indicated a CMV-driven proliferation of CD8 + γδ T cells that displayed effector phenotype and preferentially enriched with Vγ9γδ T cells, we, therefore, sought to characterize the TRG CDR3 clonotypes of γδ T cells to determine if there are differences regarding TCR diversity or clonal focusing driven by CMV infection. The CMV+ graft displayed several single clone expansions, as depicted by treemap (Figure 4(a)) and quantile plots (Figure 4(b)). Consequently, the TCR diversity was consistently lower in the CMV+ graft when compared to the CMV-, including inverse Simpson's D (85.19 vs. 302.50), Efron-Thisted estimator (5347.45 vs 57100.37), and iChao1 estimate (3090.92 versus 51054.66). Additionally, the CMV + graft presented reduced singleton frequency (clones met once in the repertoire, 0.51% versus 39.00%, Figure 4(b)), high space taken by the top 10 most abundant clones (31.72% versus 11.34%, Figures 4(b) and 4(c)), high frequency of hyperexpanded clones (30.85% vs. 8.38%, Figure 4(d)), and high clonality (0.30 vs. 0.15), altogether demonstrating a high clonal focusing in this sample.
Consistent with our previous work [19], the CMV-graft presented a high proportion of clones with a TRG consisting of 14 amino acids (59.44% vs. 24.81% in the CMV+ graft). In contrast, the CMV+ graft presented an enrichment of clones with a TRG length of 7 to 12 amino acids and a reduced frequency of those with 14 to 16 amino acid length  ( Figure 5(a)). These changes resulted in the shift from a Gaussian-distributed spectratype observed in the CMV-graft to a skewed repertoire in the CMV+ graft ( Figure 5(b)). Furthermore, by evaluating the V-J pairing, we found that Vγ9/JP segments were the most commonly used segments in the CMV-graft. In the CMV+ graft, the Vγ3/J2, Vγ4/J2, Vγ5/J2, Vγ8/J2, and Vγ9/J2 were more used, while the Vγ9-JP pair was dramatically reduced (Figures 5(c) and 5(d)).
As NGS data showed clonal focusing in CMV+ grafts, we hypothesized that CD8 + γδ T cells are more clonally focused. To investigate this further, we assessed the TCR repertoire in sorted CD8 + and CD8γδ T cells by CDR3 spectratyping. Analysis of two CMV+ grafts revealed a more focused TCR repertoire in CD8 + γδ T cells compared to CD8γδ T cells ( Figure S2).

TCR/CD3
Stimulation Triggers CD8 + γδ T Cells. As our results suggested an adaptive-like phenotype of CD8 + γδ T cells, we sought to alleviate the potential role of TCR in the activation and proliferation of CD8 + γδ T cells. TCR/CD3 stimulation resulted in significantly increased proliferation of CD8 + γδ T cells compared to CD8γδ T cells (Figures 6(a) and 6(b)). Furthermore, this TCR-driven proliferation was accompanied by increased frequencies of CD69 + , TIM3+, and KIR2DL2/L3+ γδ T cells in CD8 + γδ T cells compared to CD8γδ T cells (Figure 6(c) and 6(d)) indicating their activation.

γδ T Cell Proliferation in response to Cytokine
Stimulation. Next, we tested the impact of different cytokines on γδ T cell proliferation (Figure 6(e)). Interestingly, we    observed a remarkably increased proliferation of CD8 + γδ T cells in response to IL-7 and IL-15 but not to IL-18. In contrast, there was no significant difference in the proliferation of CD8γδ T cells upon stimulation with IL-7, IL-15, or IL-18 (Figures 6(e)-6(g)).

Discussion
Consistent with previous reports, we showed higher proportions of Vδ1 + γδ T cells in CMV+ grafts. In fact, it has been shown that Vδ1 + γδ subset can pair to any Vγ chains including Vγ9 [13]. Of note, FACS data alone cannot show whether Vδ1 couple to the semi-invariant or the noninvariant Vγ9 chain. Using NGS, we showed less prevalence of JγP-Vγ9 pairing in CMV+ grafts [19]. In their study, Vermijlen et al. showed that CMV-responsive γδ T cells were restricted to Vγ9subset, irrespective of the Vδ chain usage [21]. Furthermore, a recent study showed that a distinct subset of Vδ2 + γδ T cells expresses Vγ9 -(Vγ9 -Vδ2 + ) and displayed an adaptive-like phenotype [22]. Therefore, Vγ9subset of γδ  T cells can better represent the adaptive-like compartment of γδ T cells compared to Vδ2 neg subset. In this context, our results showed an increased frequency of Vγ9γδ T cells in BM grafts from CMV+ compared to CMV-donors (P = 0:05) and were preferentially differentiated to TE phenotype (CD27 low/-CD45RO -) supporting an adaptive role of Vγ9γδ T cells.
γδ T cells expressing the CD8 coreceptor represent an unusual subpopulation of γδ T cells. Compared to the more common CD4 -CD8γδ T cells, their development and function are poorly understood. Reports have shown that CD8 + γδ T cells selectively localize to intestinal epithelial tissue and are mostly Vδ1 + [23,24]. Furthermore, a potential role in intestinal inflammatory diseases has been recently described [15]. In line with previous report, we showed that CD8 + γδ T cells are more frequent in CMV+ grafts [25] and express Vγ9γδ T cells. This CMV-driven proliferation was accompanied by a remarkable transition from CD27 + to CD27 low/phenotype, indicating differentiation from naïve towards effector phenotype. Consistently, lymphoid homing receptor CCR7 in CD8 + γδ T cells from CMV-grafts was higher as compared to CMV+ grafts, inferring their potential for homing to secondary lymphoid tissues and supporting a naïve phenotype in CMV-grafts. Whether this entails their ability to be primed by antigenpresenting cells in an adaptive-like manner remains to be investigated.
Importantly, it has been reported that CD8 coreceptors expressed by γδ T cells are mostly CD8αα, in contrast to CD8αβ expressed by conventional T cells [25]. In this regard, our study is limited as we have not assessed whether CD8αα or CD8αβ was mainly expressed; given the recent evidence on CD8αβ + γδ T cells [15], further characterization will be required to alleviate the immunobiological role of the different CD8 molecules.
We showed in a recent study that CD8 + γδ T cells, in contrast to CD8γδ T cells, displayed higher proliferation and activation markers in response to allogeneic stimulation [16]. Furthermore, their proportions in stem cell grafts were associated with the incidence of acute graft-versus-host disease (GVHD), supporting potential alloreactivity [16]. In the present study, we extended our findings by showing that CD8 + γδ T cells were more responsive to TCR stimulation ( Figure 6) and their response pattern to cytokines was different from CD8γδ T cells, suggesting adaptive rather than innate response. Of note, the relationship between CMV reactivation and GVHD development after HCT is complex and can be bidirectional [26]. In light of our findings with regard to CD8 + γδ T cells in stem cell grafts, it is valid to address whether CMV immune response and alloreactivity represent the dual face of this subset. Consistent with previous reports, NGS results indicated repertoire perturbation in the form of TRG clonal focusing and higher usage of non-Vγ9 gene segments in γδ T cells from the CMV+ graft. Of note, CMV+ grafts were enriched with CD8 + γδ T cells that strongly displayed a terminal effector memory phenotype, indicating that clonal focusing revealed by NGS represents the clonal proliferation of CD8 +γδ T cells. Though we have demonstrated in small scale spectratype that TCR repertoire was more clonally focused in CD8 + γδ T cells, our study is limited as NGS analysis was not done on sorted CD8 + γδ T cells to confirm this.
In conclusion, we showed that γδ T cell repertoire within BM grafts is reshaped by donor CMV serostatus and has provided evidence for the implication of CD8 + γδ T cells in the HCMV immune response. Further studies are required to confirm our findings and to in-depth alleviate the impact of CMV-induced TCR repertoire and phenotypic changes of CD8 + γδ T cells.

Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.