CMV reactivation drives posttransplant T-cell reconstitution and results in defects in the underlying TCRβ repertoire

Abstract

Although cytomegalovirus (CMV) reactivation has long been implicated in posttransplant immune dysfunction, the molecular mechanisms that drive this phenomenon remain undetermined. To address this, we combined multiparameter flow cytometric analysis and T-cell subpopulation sorting with high-throughput sequencing of the T-cell repertoire, to produce a thorough evaluation of the impact of CMV reactivation on T-cell reconstitution after unrelated-donor hematopoietic stem cell transplant. We observed that CMV reactivation drove a >50-fold specific expansion of Granzyme B(high)/CD28(low)/CD57(high)/CD8(+) effector memory T cells (Tem) and resulted in a linked contraction of all naive T cells, including CD31(+)/CD4(+) putative thymic emigrants. T-cell receptor β (TCRβ) deep sequencing revealed a striking contraction of CD8(+) Tem diversity due to CMV-specific clonal expansions in reactivating patients. In addition to querying the topography of the expanding CMV-specific T-cell clones, deep sequencing allowed us, for the first time, to exhaustively evaluate the underlying TCR repertoire. Our results reveal new evidence for significant defects in the underlying CD8 Tem TCR repertoire in patients who reactivate CMV, providing the first molecular evidence that, in addition to driving expansion of virus-specific cells, CMV reactivation has a detrimental impact on the integrity and heterogeneity of the rest of the T-cell repertoire. This trial was registered at www.clinicaltrials.gov as #NCT01012492.

© 2015 by The American Society of Hematology.

Figures

Figure 1

Viral reactivation and T-cell reconstitution in CMV-reactivating and nonreactivating patients. (A) CD8+ Tem counts at day +365 posttransplant. (B) CMV viral load measured longitudinally. Each CMV-reactivating patient is shown as a unique colored line. Donor/recipient (D/R) pretransplant CMV serostatus is also indicated. Also shown is the mean day of CMV reactivation (±standard error of the mean [SEM]), depicted as a purple bar. (C) Longitudinal analysis of WBC, granulocytes, NK cells, monocytes, lymphocytes, total T cells, and total B cells posttransplant. Data are mean ± SEM in +CMV patients (red traces, n = 7), −CMV patients (blue traces, n = 10), and healthy controls (green diamond longitudinally extended as gray area, n = 10). *P ≤ .05; Wilcoxon rank-sum test. (D) Longitudinal analysis of absolute numbers ± SEM of CD4+ and CD8+ T-cell reconstitution (+CMV [n = 7] and −CMV [n = 10]). Also shown are the means ± SEM of 10 healthy controls (green diamond extended longitudinally as gray area). (E) Analysis of CD4:CD8 ratio at baseline and days 100, 180, and 365 posttransplant as mean ± SEM in +CMV, −CMV, and healthy controls (green diamond). *P ≤ .05; **P ≤ .01; ***P ≤ .001; NS, nonsignificant, Wilcoxon rank-sum test. (F) Analysis of CD4:CD8 ratio at day 100 of an extended cohort of 46 HSCT patients: +CMV (n = 26) and −CMV (n = 20).

Figure 2

Longitudinal analysis of CD8 naive and memory T-cell subpopulation reconstitution in +CMV and −CMV patients. (A) Representative flow cytometry analysis of memory subset marker expression in CD8+ T cells. Cells were gated as follows: lymphocytes were identified by FSC and SSC, CD14−CD3+ T lymphocytes were identified, further distinguished into CD8+ and CD4+ T cells, and analyzed for their expression of CCR7 and CD45RA memory markers. Tnaive were identified as CCR7+/CD45RA+, TCM: CCR7+/CD45RA−, TEM: CCR7−/CD45RA−, TEMRA: CCR7−/CD45RA+. (B-C) Longitudinal analysis of CD8+ naive and Tem subsets depicted in percentage frequency (B) or absolute cell numbers (C). Data are mean ± SEM in +CMV patients (n = 7), −CMV patients (n = 10), and healthy controls (n = 10). Also shown is the mean day of CMV reactivation (±SEM), depicted as a purple bar. (D) Left, Representative flow cytometry analysis at day +365 shows CD28 expression (left panel) and Granzyme B expression (right panel) on CD8+ Tem. Right, Longitudinal analysis of mean ± SEM CD28−CD8+ Tem (left y-axis) and Granzyme B+CD8+ Tem (right y-axis). (E) Longitudinal analysis of PD-1−/CD57+CD8+ Tem (solid lines) and naive T cells (dotted lines) of +CMV (n = 7), −CMV (n = 10) patients, and healthy controls. All data are mean ± SEM. *P ≤ .05; **P ≤ .01 Wilcoxon rank-sum test. CCR, C-C chemokine receptor; FSC, forward scatter; SSC, side scatter; TCM, central memory T cells; TEMRA, effector memory-RA T cells.

Figure 3

Contraction of Tem TCR diversity but not Tnaive TCR diversity in CMV-reactivating patients. (A) Representative flow cytometry analysis illustrates the purity of CD8+ Tnaive and CD8+ Tem cells before (left plot) and after (middle and right plots) cell sorting. (B) Shown are representative graphs of the CD8+ naive (left column) and CD8+ Tem (right column) TCR landscape (showing frequencies of V and J gene combinations detected through deep sequencing) from 1 −CMV (blue) and 1 +CMV (red) patient. (C) Clonality of PBMC, CD8+ Tnaive, and CD8+ Tem are shown for −CMV and +CMV patients as measured by the inverse of the normalized Shannon entropy number. All data are mean ± SEM. *P ≤ .05; Wilcoxon rank-sum test. (D) The clonality of Tnaive and Tem CD8+ cells for each patient is compared with the percentage of Tem CD8+ cells detected via flow cytometry in each patient. Linear regression (R2 = 0.506) shows a correlation between expansion of Tem and increased Tem clonality in +CMV patients (red). No such relationship could be detected in −CMV patients (blue). (E) TCR diversity of CD8+ Tnaive and CD8+ Tem are shown for −CMV and +CMV patients as measured by the Gini coefficient. All data are mean ± SEM. *P ≤ .05; Wilcoxon rank-sum test. #Sorting purity of Tnaive from patient 001-008 could not be confirmed due to low yield.

Figure 4

Analysis of CMV-specific CD8+ T cells using peptide stimulation and tetramer binding. (A) Patients who were positive for HLA-A2, -B7, or -B8 were analyzed for the frequency with which they bound to appropriate CMV-specific tetramers. Representative flow cytometry analysis is shown for −CMV (left) and +CMV (right) patients. Cells were gated as follows: lymphocytes were identified in a FSC/SSC plot, out of these, singlets were isolated. From these, CD14−/CD20−/CD3+ T cells were identified and further gated on CD8+ T cells, which were analyzed for their binding to matching tetramers. (B) Left, bar graph illustrating the frequency of tetramer-binding cells from −CMV (blue) and +CMV (red) patients. Right graph, Tetramer+ cells from +CMV patients were further separated into their respective memory subsets. (C) Representative flow cytometry analysis from −CMV and +CMV patients of intracellular cytokine expression after stimulation with CMV-specific peptides. CD8+ T cells were gated as described in Figure 2A and then analyzed for the production of IFNγ and TNF. (D) Frequency of IFNγ+/TNF+ CD4+ and CD8+ T cells of +CMV (n = 6) and −CMV (n = 4) patients (left). Also the frequency of IFNγ+/TNF+ CD8 T cells in each of the 4 memory subsets is shown for +CMV and −CMV patients (right). *P ≤ .05; **P ≤ .01 Wilcoxon rank-sum test.

Figure 5

Inter- and intrapatient evaluation of shared TCRβ clones between purified Tem and tetramer+ cells. (A) Color-coded pairwise TCRβ clone sharing between samples. Colors range from yellow (indicates no sharing) to blue (indicates 100% sequence sharing). This matrix depicts the degree of clone sharing between all identified TCRβ sequences in the sorted Tem and tetramer+ cells. (B) Enrichment of Tetramer+ cells in the clonally expanded Tem. Shown are dot plots for 4 +CMV patients which depict the binned frequencies for each of the TCR clones in the sorted CD8+ Tem (“TEM”; black) and sorted tetramer+ cells (“TETPOS”; green). These graphs depict the change in frequencies of productive clones, each compared with a sample-specific median frequency value. Bins represent 0.25 log(10) intervals, with binned data reported using a log10 scale. Tem clones are shown as black circles and the Tetramer+ clones are shown as green circles.

Figure 6

Impact of CMV reactivation on CD8 Tem TCR repertoire holes. (A-B) For panels A-B, V, and J genes are represented along the x-y axes, whereas the difference in the V-J–specific proportions of unique clones are represented along the z-axis and labeled “Freq.” TCE and holes analysis was performed as described in “Methods.” Panels A and B depict wireframe 3-dimensional graphs (created using the Lattice graphics package in R) for 2 representative patients (001-008, panel A and 001-005, panel B). Both TCEs and holes in the TCR repertoire are depicted in these graphs, with V-J families in orange depicting TCEs and those in green depicting holes. Those families shown in light gray are already sparse in the VJr and thus were not evaluated for holes. (C) Summary analysis showing the mean number of holes in the CD8 Tem from the +CMV (red squares) and −CMV (blue circles) patients using the VJr. **P < .01. (D) Summary analysis showing the mean number of holes in the CD8 Tem from the +CMV (red) and −CMV (blue) patients using the VJr-transplant. **P < .01. (E) Analysis of the number of holes in −CMV patients that either did not develop GVHD (green squares, left) or did develop GVHD (purple hexagons, right) compared with the VJr. P = 1.0. (F) Analysis of the number of holes in +GVHD patients that either did reactivate CMV (red squares) or did not reactivate CMV (blue circles) compared with the VJr. *P < .05.

Figure 7

Impact of CMV reactivation and GVHD on the reconstitution of CD4+ Tnaive and CD31+ recent thymic emigrants. (A) Longitudinal analysis of the reconstitution of CD4+ Tnaive. Data are mean ± SEM in +CMV patients (n = 7), −CMV patients (n = 10), and healthy controls (n = 10). Also shown is the mean day of CMV reactivation (±SEM), depicted as a purple bar. (B) Longitudinal analysis of the reconstitution of CD31+/CD4+ recent thymic emigrants. Data are mean ± SEM in +CMV (n = 7), −CMV patients (n = 10), and healthy controls (n = 10). *P ≤ .05; **P ≤ .01; NS, nonsignificant; Wilcoxon rank-sum test. (C) Longitudinal analysis of the reconstitution of CD4+ Tnaive dichotomized based on GVHD status. Data are mean ± SEM in +GVHD patients (purple circles, n = 10), −GVHD patients (green squares, n = 7), and healthy controls (green circle and gray bar, n = 10). (D) Longitudinal analysis of the reconstitution of CD31+/CD4+ recent thymic emigrants dichotomized based on GVHD status. Data are mean ± SEM in +GVHD patients (purple circles, n = 10), −GVHD patients (green squares n = 7), and healthy controls (green circle and gray bar, n = 7). (E) Longitudinal analysis of the reconstitution of naive CD8+ T cells dichotomized based on GVHD status. Data are mean ± SEM in +GVHD patients (purple circles, n = 10), −GVHD patients (green squares, n = 7), and healthy controls (green circle and gray bar, n = 10).