Donor CD4 T Cell Diversity Determines Virus Reactivation in Patients After HLA-Matched Allogeneic Stem Cell Transplantation.

Delayed reconstitution of the T cell compartment in recipients of allogeneic stem cell grafts is associated with an increase of reactivation of latent viruses. Thereby, the transplanted T cell repertoire appears to be one of the factors that affect T cell reconstitution. Therefore, we studied the T cell receptor beta (TCRβ) gene rearrangements of flow cytometry-sorted CD4(+) and CD8(+) T cells from the peripheral blood of 23 allogeneic donors before G-CSF administration and on the day of apheresis. For this purpose, TCRβ rearrangements were amplified by multiplex PCR followed by high-throughput amplicon sequencing. Overall, CD4(+) T cells displayed a significantly higher TCRβ diversity compared to CD8(+) T cells irrespective of G-CSF administration. In line, no significant impact of G-CSF treatment on the TCR Vβ repertoire usage was found. However, correlation of the donor T cell repertoire with clinical outcomes of the recipient revealed that a higher CD4(+) TCRβ diversity after G-CSF treatment is associated with lower reactivation of cytomegalovirus and Epstein-Barr virus. By contrast, no protecting correlation was observed for CD8(+) T cells. In essence, our deep TCRβ analysis identifies the importance of the CD4(+) T cell compartment for the control of latent viruses after allogeneic stem cell transplantation.

Introduction

Allogeneic hematopoietic stem cell transplantation (aHSCT) represents a curative treatment option for patients with hematological malignancies. Today, G-CSF mobilized peripheral blood stem cell (PBSC) products, rather than bone marrow, are the predominant source of stem cells used in transplantation. Both stem cell sources differ in their cellular composition. In particular, PBSC grafts contain approximately 10 times more T cells than those derived from bone marrow (1). Despite the higher number of T cells transfused along with PBSCs, the incidence of acute graft-versus-host disease (aGvHD) after transplantation is not significantly higher with PBSC grafts compared to bone marrow transplants. This fact has been explained by a Th2 polarization of T cells in the graft induced by G-CSF (2). However, the influence of G-CSF on the donor T cell repertoire remains unknown.

The success of aHSCT is, however, compromised by two additional factors, namely, relapse of the underlying malignancy and infectious complications. While aGvHD can be eliminated by T cell depletion of the graft, the incidence of relapse and infectious complication increases dramatically due to delayed recovery of the donor T cell compartment. Even if the PBSC graft is not T cell depleted, quantitative and qualitative reconstitution of the T cell compartment in the host requires several months and tends to remain incomplete (3,4). Delayed T cell reconstitution and restricted T cell receptor (TCR) diversity after transplantation are associated with increased reactivation of latent viruses such as cytomegalovirus (CMV) and Epstein–Barr virus (EBV) (5–8). Once full donor chimerism in recipients is achieved and donor hematopoiesis is the only source for T cells, the donor T cell repertoire is solely responsible for controlling latent viruses. Therefore, we hypothesized that the T cell repertoire diversity of the donor after G-CSF mobilization determines later clinical outcome of recipients.

The high diversity of the human T cell repertoire is mainly generated by somatic recombination among the variable (V), diversity (D), and joining (J) TCRβ gene segments and the addition of random nontemplated bases at recombination junctions. The hypervariable complementarity-determining region 3 (CDR3) of the rearranged TCRβ represents the antigen-binding site and can generate more than 106 unique TCRβ receptor rearrangements (9).

Several strategies have been developed over the last two decades to study the composition and diversity of the T cell repertoire, including flow cytometry and spectratyping (10–13). However, none of these attempts has managed to generate individual sequences or frequencies of unique TCRβ rearrangements. This has changed with technical advances in high-throughput sequencing (HTS), which now allow the analysis of T cell repertoires based on sequence analysis of TCRβ rearrangements (14–16).

To test our hypothesis that T cell diversity of stem cell donors is important for the clinical course of recipients, we employed an approach that combines multiplex TCRβ PCR employing DNA extracted from flow cytometry–sorted CD4+ and CD8+ T cells with HTS. To this end, we analyzed peripheral blood samples of 23 donors obtained before G-CSF–induced stem cell mobilization and at the day of stem cell apheresis. By correlation with clinical outcomes of the stem cell recipients, we demonstrated that reduced TCRβ diversity of the CD4+ T cell compartment after G-CSF mobilization of the donor is associated with reactivation of CMV and EBV in the recipients.

Materials and Methods

Samples acquisition and recipient data

Peripheral blood samples of six healthy volunteers were used after informed consent to determine the robustness and reliability of our HTS-based TCRβ analysis (Table 1).

Table 1

Characteristics of healthy control individuals

Control individual		Age	No. of clonotypes CD4+ cells	No. of clonotypes CD8+ cells
1		24	2555	2566
2		29	2529	2513
3		38	2464	1800
4		22	2383	2547
5	A	44	3002	1904
	B		2937	1338
6	A	29	2830	1734
	B		2943	2376

Number of different clonotypes in the CD4+ and CD8+ T cell compartments measured by HTS as well as the age of six healthy control individuals. A and B represent two different flow cytometry sorts of the same blood sample.

To study T cell diversity in donors before and after G-CSF induced stem cell mobilization, peripheral blood samples of 23 allogeneic, HLA-matched (10/10) donors were analyzed before G-CSF mobilization and on the day of apheresis. Gender, age, donor type, virus (CMV, EBV) status of stem cell donors and recipient outcomes were assessed (Table 2). Recipients (see also Table S1) were transplanted and monitored at the Department of Hematology and Oncology (University of Erlangen) after aHSCT.

Table 2

Characteristics of 23 stem cell donors and corresponding recipients

Donor										Recipient
				Serostatus (IgG)		No. of clonotypes CD4+ cells preMOB	No. of clonotypes CD4+ cells postMOB	No. of clonotypes CD8+ cells preMOB	No. of clonotypes CD8+ cells postMOB	Virus reactivation
No.	Gender	Age	Donor type	CMV	EBV					CMV	EBV
1	M	57	SIB	neg.	pos.	2454	2288	887	840	No	Yes
2	M	30	MUD	neg.	pos.	3034	3110	2350	2369	No	No
3	M	20	MUD	neg.	pos.	2198	2604	1938	2231	Yes	No
4	M	33	SIB	neg.	neg.	2786	2825	1984	1254	No	No
5	F	51	SIB	neg.	pos.	2562	2596	1871	2140	No	No
6	M	23	MUD	pos.	pos.	2380	1542	1489	1400	Yes	Yes
7	M	58	SIB	neg.	pos.	2410	3061	NA	NA	No	No
8	M	50	MUD	pos.	pos.	2617	2674	1503	1295	No	No
9	M	44	MUD	pos.	pos.	NA	NA	995	1085	Yes	No
10	F	61	SIB	neg.	pos.	2020	2629	1433	1577	No	No
11	F	51	SIB	neg.	pos.	2773	2810	2592	1519	No	No
12	M	45	MUD	pos.	pos.	2655	2552	1903	2027	Yes	Yes
13	M	30	MUD	neg.	pos.	2575	2687	2296	2427	Yes	Yes
14	M	51	MUD	neg.	pos.	2403	2143	941	1520	Yes	No
15	M	48	SIB	neg.	pos.	2493	2065	508	1088	Yes	No
16	M	20	MUD	pos.	neg.	2373	2084	2054	1729	Yes	No
17	M	32	MUD	pos.	pos.	2859	2503	1642	1666	Yes	No
18	M	48	MUD	pos.	pos.	2189	2292	941	1380	Yes	Yes
19	M	40	MUD	neg.	neg.	2674	2602	2141	1668	No	No
20	M	41	MUD	neg.	neg.	2054	2795	1764	2381	No	No
21	M	49	MUD	neg.	pos.	2853	2577	1813	1508	No	No
22	M	40	SIB	pos.	pos.	2245	1866	2400	2021	No	Yes
23	M	45	MUD	pos.	pos.	2294	2477	1434	1414	Yes	Yes

Gender, age, donor type, virus (CMV and EBV) status, and number of clonotypes of the CD4+ and CD8+ T cell compartments of stem cell donors before and after G-CSF mobilization as well as clinical data (CMV and EBV reactivation) of corresponding recipients.

CMV, cytomegalovirus; EBV, Epstein–Barr virus; F, femal; G-CSF, granulocyte colony-stimulating factor; MUD, matched unrelated donor; M, male; NA, not available; SIB, sibling donor.

Virus reactivation was monitored by specific quantitative real-time PCR (for details see Supplementary Methods in Supporting Information). Ethical approval for this study was given by the local ethics review board of the University of Erlangen (Re. No. 4224).

Cell sorting

Whole blood samples were processed by density gradient centrifugation (Ficoll, GE Healthcare, UK) to obtain mononuclear blood cells (PBMCs). For flow cytometric sorting, PBMCs were stained with CD4, CD8, CD14 and TCRαβ antibodies (for details see Supplementary Methods in Supporting Information). Cells were gated on (i) vital lymphocytes in forward/side scatter, (ii) exclusion of doublets and (iii) TCRαβ+ CD14negative T cells. CD4+ and CD8+ T cells were sorted into separate tubes. A purity of >98.0% was achieved.

DNA isolation

DNA was extracted from CD4+ and CD8+ flow cytometry–sorted T cells using Qiagen AllPrep DNA/RNA Mini Kit (Qiagen, Hilden, Germany). Quantification of the extracted DNA was done employing Qubit® 1.0 Fluorometer (Invitrogen, Carlsbad, CA).

Amplification and sequencing of TCRβ gene rearrangements

TCRβ gene rearrangements were amplified by a multiplex PCR approach, which includes 58 primers initially described by Robins et al (14), but was redesigned for our purpose: (i) all primer sequences were re-evaluated and modified in order to allow a more homogeneous amplification of the various TCRβ rearrangements (Table S3); (ii) employing a two-step PCR strategy, the TCRβ amplicons were tagged with universal Illumina adapter sequences, including an additional barcode during a second amplification step, allowing parallel sequencing of several samples on Illumina HiSeq2000 (Illumina, San Diego, CA); and (iii) using the Illumina paired-end technology (2 × 100 bp) our amplicon sequences covered the entire CDR3 length and Vβ and Jβ segments in parts providing higher sequence accuracy.

The multiplex primers used contain a universal adapter sequence as a tail at the 5′ end complementary to the 3′ ends of second amplification adaptor primers. The adaptor PCR primers contained universal sequences that permitted solid-phase PCR on the Illumina Genome Analyzer (HiSeq2000 Sequencing System).

Primary amplification (final volume: 50 µL) was processed, including 100 ng DNA, 1.0 µM equimolar Vβ and Jβ primer pools, 1× PCR buffer, 3 mM MgCl2, 0.2 mM of each dNTP and 1 U AmpliTaq Gold DNA Polymerase (Applied Biosystems, Foster City, CA). The amplification was performed on a DNA thermal cycler (GeneAmp® PCR System 9700, Applied Biosystems) for 34 cycles at 62°C annealing temperature.

All PCR products were purified using the QIAquick PCR Purification Kit (Qiagen) and diluted (final amount: 500 pg) for further amplifications. Adapter PCRs were set up with 1× Phusion HF Buffer, 1.0 µM forward and reverse adapter primers, 0.05 mM of each dNTP and 1 U Phusion High-Fidelity DNA Polymerase (Finnzymes, Espoo, Finland). Secondary amplification was performed for 12 cycles at 58°C annealing temperature. Products were isolated from a 2% agarose gel using the Wizard® SV Gel and PCR Clean-Up System (Promega, Mannheim, Germany). DNA concentration was determined via the Qubit® 1.0 Fluorometer (Invitrogen).

Generated Illumina libraries were sequenced on a HiSeq2000 analyzer using paired-end modus with 2 × 100 bp read length and joining of read-pairs was performed by an in-house developed algorithm. Paired-end amplicon sequencing resulted in highly accurate sequence reads due to the overlap of related single sequence reads which is of great advantage over single-read-based clustering approaches. After read-pair joining, all identical and related sequence reads that deviated up to 1 bp mismatch were clustered and designated as unique TCRβ clonotypes. Clonotypes below a cut-off of 0.01% of all reads were regarded as background signals and neglected. According to the ImMunoGeneTics information system (http://www.imgt.org), the clonotype-specific Vβ and Jβ segments were assigned and the TCRβ CDR3 sequence was translated to in silico protein sequence. The TCRβ CDR3 sequence was defined as all amino acids (AAs) starting from the conserved 5′ cysteine in the Vβ segment and ending at the conserved 3′ phenylalanine in the Jβ segment.

Statistical analyses

Nonparametric analysis between data groups was performed by the Mann–Whitney U-test using GraphPad PRISM 6 software. Linear correlations between T cell diversity and patient age were tested by calculation of the Pearson's correlation coefficient. Contingency tables estimating Pearson's Chi square were generated using IBM SPSS Statistics software (version 22). A p-value < 0.05 was considered significant.

The TCRβ repertoire diversity was validated by employing the Shannon entropy for which a maximum entropy of 14 500 clonotypes per sample (equal to the number of T cells in 100 ng DNA) was assumed (17).

Results

Validation of the TCRβ composition in healthy control individuals

To demonstrate the reliability and robustness of our approach, we amplified the TCRβ gene rearrangements of flow cytometry–sorted CD4+ and CD8+ T cells derived from six healthy control volunteers (designated as “controls”; Table 1). For comparability, the same amount of DNA (100 ng DNA extracted from sorted T cells) representing the same number of T cells (approximately 14 500 T cells) was employed in each TCRβ PCR multiplex assay. After Illumina sequencing of the PCR products, a mean of approximately 4 610 000 joined sequence reads were generated per sample. In a first step, reads with identical TCRβ gene rearrangements were clustered leading to a mean of 2705 (range: 2383–3002) rearrangements for the CD4+ T cells and 2097 (range: 1338–2566) rearrangements for the CD8+ T cells. Clusters of identical TCRβ gene rearrangements were designated as clonotypes. The comparison of these clonotypes with germline Vβ and Jβ segments revealed very similar Vβ and Jβ usage in all samples (Figure 1A), which is in harmony with published data (18). Additional proof for the robustness of our data was given by the CDR3 size distribution (Figure 1B), which revealed a Gaussian-like curve almost identical to size distribution generated by conventional capillary electrophoresis (GeneScan) (19). Individual CDR3 length distribution graphs are given in Figure S2.

Figure 1

TCRβ analysis in healthy control individuals. (A) Vβ subgroup usage of TCRs from CD4+ (left panel) or CD8+ (middle panel) T cells of six untreated healthy control individuals (designated as “C1” to “C6”). Seven of the most frequent Vβ subgroups were illustrated individually. The remaining subgroups were compiled. The nomenclature according to the international ImMunoGeneTics information system (IMGT) was used. The right panel shows mean frequencies of all six controls for CD4+ and CD8+ T cells. (B) HTS-derived CDR3 length distribution of CD4+ (left panel) or CD8+ (right panel) TCRβ sequences for each control donor. The right panel shows mean frequencies of all donors. TCRβ CDR3 size was defined as all amino acids (AAs) from the conserved cysteine in the Vβ segment to the conserved phenylalanine in the Jβ segment. (C) The left and middle panels show the individual clonotype frequency distribution of CD4+ or CD8+ T cells of each control donor. The right panel shows the mean values of CD4+ and CD8+ T cells of all control donors. Each dot represents the occurrence (percentage) of a distinct TCRβ clonotype. Only the clonotype frequencies for the first 250 clonotypes were illustrated. (D) Diversity dot plot of CD4+ and CD8+ T cells for six healthy control individuals. Each dot shows the number of unique clonotypes detected in one donor. The bars illustrate the mean value of the groups. (E) Overlaps of individual clonotypes in CD4+ or CD8+ T cells of one control donor sequenced twice out of one T cell preparation (5A) and of two separate cell sorts (5A/B). (F) Overlaps of individual clonotypes in CD4+ or CD8+ T cells of distinct control individuals.

Although the Vβ and Jβ usage was highly similar in the two different T cell populations, the frequency and number of clonotypes (designated as diversity) in CD4+ and CD8+ T cells in our control samples revealed a statistically significant difference. In the CD4+ T cell fractions, the most frequent clonotype comprised 0.8% of all sequence reads (mean), whereas in the CD8+ T cell compartments the mean of the most common clonotype was 4.7% of all reads (Figure 1C). In consequence, a higher number of different clonotypes could be demonstrated in the CD4+ T cell fraction (Figure 1D). Taken together, CD4+ T cells showed a significantly higher diversity in their TCRβ repertoire compared to the CD8+ T cell compartment.

In order to determine the re-occurrence of identical clonotypes within individual control donors, we analyzed aliquots of the same DNA preparations of the same CD4+ and CD8+ sorted T cells in duplicate. In addition, we employed DNA preparations derived from different flow cytometry sorts of the same peripheral blood sample. The analysis of two different aliquots of the same DNA preparation revealed an overlap of 4.1% identical clonotypes for CD4+ T cells and an overlap of 13.7% for CD8+ T cells (Figure 1E). Comparing the different flow cytometry sorts of the same peripheral blood sample, the overlap was marginally lower leading to 3.2% identical clonotypes in CD4+ T cells and 9.2% in CD8+ T cells, respectively. Strikingly, the overlap of clonotypes between two different individuals in our study was less than 0.2% in CD4+ and CD8+ T cells (Figure 1D). Whereas the extremely rare coincidence of identical TCRβ rearrangements in different individuals is expected, the low overlap within the same DNA preparation or same T cell sort of the same individual implies that the T cells analyzed are composed to a large extent of unique TCRβ rearrangements. This is in line with the results of Warren et al (16), who demonstrated a similar low overlap between the TCRβ repertoires of different whole peripheral blood samples from one individual.

Analysis of TCRβ repertoire in stem cell donors before and after G-CSF mobilization

To monitor changes in the TCRβ repertoire during stem cell mobilization, we investigated the presence and frequency of TCRβ gene rearrangements in CD4+ and CD8+ T cells of 23 stem cell donors before G-CSF administration (pre) and on the days of stem cell apheresis (post) consisting of 88 samples in total (Table 2).

Analysis of the Vβ segment usage showed neither in CD4+ nor in CD8+ T cells a significant modification after G-CSF administration (Figure 2A) and the CDR3 length distribution remained unaffected (Figure S3).

Figure 2

TCR analysis in stem cell donors before and after G-CSF induced mobilization. (A) Mean Vβ subgroup usage of TCRs from CD4+ or CD8+ T cells of 23 donors before (pre) and after (post) G-CSF induced stem cell mobilization. The seven most frequent Vβ subgroups were illustrated individually; the remaining subgroups were grouped together. The nomenclature according to the international ImMunoGeneTics information system (IMGT) was used. (B) Mean percentage of overlaps of individual clonotypes in CD4+ or CD8+ T cells of the 23 donors pre– and post–G-CSF–induced stem cell mobilization. (C and D) Diversity dot plots of CD4+ and CD8+ T cells pre– and post–G-CSF mobilization. Each dot represents the number of unique clonotypes detected in individual donors. (C) The number of clonotypes in CD4+ T cells differs significantly before and after G-CSF mobilization from the number of clonotypes in the CD8+ compartment (****p < 0.0001; n = 44). Bars show mean values. (D) The number of clonotypes in CD4+ and CD8+ T cells is not affected by G-CSF administration. (E) Scatter plots of CD4+ and CD8+ clonotype numbers and the age for stem cell donors pre– and post–G-CSF mobilization. The number of individual CD8+ clonotypes decreases with increasing donor age irrespective of G-CSF mobilization (preMOB: r = −0.42; p = 0.052; postMOB: r = −0.42; p = 0.051; n = 22). For CD4+ T cells, no negative correlation was found. CD4+ T cells preMOB; CD4+ T cells postMOB; CD8+ T cells preMOB; CD8+ T cells postMOB.

The pairwise comparison of individual T cell repertoires before and after G-CSF mobilization in the 23 donors revealed a mean overlap of 1.6% for CD4+ and 6.2% for CD8+ T cells (Figure 2B), which is in harmony with the results of our six control individuals. The lower percentage is most likely explained by the fact that the samples of the controls were taken at the same time, whereas the samples of the stem cell donors were drawn at different time points before and after G-CSF mobilization. Regarding the number of different TCRβ clonotypes in CD4+ and CD8+ T cells, we found a significantly higher diversity of TCRβ gene rearrangements in CD4+ T cells compared to CD8+ T cells (Figure 2C; q < 0.0001; n = 44). However, the overall diversity of CD4+ and CD8+ T cells remained unaffected during G-CSF mobilization (Figure 2D) and was independent of the increased number of T cells in the peripheral blood after G-CSF administration (Figures S4 and S5).

Interestingly, in CD8+ T cells the TCRβ diversity correlated negatively with increasing donor age independent of mobilization state (Figure 2E; preMOB: r = −0.42; p = 0.052; postMOB: r = −0.42; p = 0.051; n = 22). Thus, our data confirm the notion that the TCRβ repertoire in CD8+ T cells is decreasing as a function of age (20).

Impact of the donor's viral status on TCR diversity during G-CSF induced mobilization

Since a preceding infection with herpes viruses (Table 2) may impact TCRβ diversity, we analyzed the repertoire of CMV-seropositive and CMV-seronegative donors separately (Figure 3A). As a result, we could demonstrate a significantly higher number of clonotypes in the CD4+ T cell compartment after G-CSF mobilization of CMV-seronegative donors compared to CMV-seropositive donors (p = 0.0158). On average, G-CSF per se does not influence the TCR repertoire in both T cell subsets irrespective of the virus status of the donors (Figure S6).

Figure 3

Donor TCRβ diversity and clinical correlation. (A) Diversity dot plots of CD4+ and CD8+ T cells of cytomegalovirus (CMV)-seropositive or -seronegative donors before (pre) and after (post) G-CSF mobilization. Only for the CD4+ T cell compartment postmobilization a significant difference could be displayed between CMV-seropositive and -seronegative donors (*p < 0.05; neg. n = 14; pos. n = 8). (B) Diversity dot plots of CD4+ and CD8+ TCRβ rearrangements of G-CSF–mobilized donors premobilization and postmobilization according to patients either reactivated (“Yes”) or not (“No”) with CMV (“Yes”: CD4+ n = 10; CD8+ n = 11) or Epstein–Barr virus (EBV; “Yes”: CD4+ n = 7; CD8+ n = 7). Cases with CMV and/or EBV reactivation presented significantly lower numbers of unique clonotypes in CD4+ T cells post–G-CSF mobilization (*p < 0.05). In CD8+ T cells, no significant difference could be detected for virus reactivation. (C) Diversity dot plots of CD4+ and CD8+ T cells of donor T cells premobilization and postmobilization according to patients who either exhibited acute graft-versus-host disease (aGvHD; “Yes”) or not (“No”). No significant correlation could be detected between donor T cell diversity and aGvHD appearance in patients. CD4+ T cells preMOB; CD4+ T cells postMOB; CD8+ T cells preMOB; CD8+ T cells postMOB.

However, four donors were EBV-seronegative. Interestingly, three of these donors were double negative for CMV and EBV and therefore presented higher numbers of clonotypes in their CD4+ T cells as shown above. In line, the only EBV-seronegative but CMV-seropositive donor displayed a reduced number of clonotypes in the CD4+ T cell compartment after G-CSF administration.

Correlation of clinical outcomes after stem cell transplantation with TCRβ diversity of the stem cell donors

Next we correlated the CD4+ and CD8+ T cell repertoires before and after G-CSF mobilization with various clinical outcomes of the patients after aHSCT. Remarkably, a reduced diversity of the TCRβ repertoire of the donors CD4+ T cell compartment after G-CSF mobilization was significantly correlated with EBV and CMV reactivation (Figure 3B; CMV: p = 0.0137; EBV: p = 0.0377). Employing the cumulative incidence of viral reactivation revealed that significantly more patients receiving grafts with lower TCRβ diversity reactivated CMV and EBV, although the serostatus of patients and donors for both viruses was different (Figure S1). In contrast, the diversity of the CD8+ T cell compartment has no impact on the viral reactivation irrespective of G-CSF mobilization. In addition, we could not identify a correlation between the development of aGvHD and the transferred TCRβ repertoire diversity at any mobilization stage (Figure 3C). These observations remained stable despite the inclusion of the donor age in our analyses (data not shown). Furthermore, neither the correlation of the Vβ segment usage of donor T cells nor the reactivation of CMV and/or EBV showed any association with the development of aGvHD in the corresponding recipients (Figure S7; Table S2).

In order to verify our observations, we generated contingency tables using the donor TCRβ diversity in the CD4+ and CD8+ T cell compartment after G-CSF mobilization and various clinical outcomes of the stem cell recipients. The T cell diversity was categorized into high and low based on its deviation from the mean value. According to this classification, the correlation between lower donor CD4+ TCRβ diversity and virus reactivation (CMV and EBV) was again significant (Table 3; CMV: χ2-test, p = 0.035; EBV: χ2-test, p = 0.020). In addition, a significant difference between lower CD8+ TCRβ diversity and the age of the donor (χ2-test, p = 0.035) could be demonstrated. Thus, our observation of a negative correlation between donor CD4+ T cell repertoire diversity and virus reactivation in stem cell–transplanted patients as well as a restricted CD8+ TCRβ repertoire in elderly donors was confirmed via contingency tables (Table 3).

Table 3

TCRβ repertoire diversity and clinical correlation

		CD4+ T cells				CD8+ T cells
Donor TCRβ repertoire		>Mean	<Mean	Total	p-value χ2-test	>Mean	<Mean	Total	p-value χ2-test
Donor type	MUD	9	5	14		8	7	15
	SIB	5	3	8	0.933	2	5	7	0.277
Donor age	<40	5	2	7		5	2	7
	>40	8	5	13	0.658	3	10	14	0.035
Patient
Survival	Alive	9	7	16		8	9	17
	Dead	5	1	6	0.240	2	3	5	0.781
TRM	Alive	11	7	18		9	10	19
	Dead	3	1	4	0.601	1	2	3	0.650
Relapse	No	10	6	16		6	10	16
	Yes	4	2	6	0.856	4	2	6	0.221
EBV reactivation	No	12	3	15		7	8	15
	Yes	2	5	7	0.020	3	4	7	0.867
CMV reactivation	No	10	2	12		5	6	11
	Yes	4	6	10	0.035	5	6	11	1.000
aGvHD	No	4	2	6		2	4	6
	Yes	10	6	16	0.856	8	8	16	0.484

Contingency table of donor CD4+ and CD8+ T cell diversity post–G-CSF mobilization in correlation with clinical parameters of the recipients. Groups were divided according to their clonotype numbers above or below the compartment mean (CD4+ mean = 2490 clonotypes; CD8+ mean = 1661 clonotypes). According to this classification, a correlation between lower donor CD4+ TCRβ diversity and virus reactivation (CMV and EBV) was significant. In addition, a significant correlation could be demonstrated between lower CD8+ TCRβ diversity and the age of the donor.

aGvHD, acute graft-versus-host disease; χ2-test, chi-squared test; TRM, transplant-related mortality.

Finally, to independently validate the diversity of the TCRβ repertoires, we employed the Shannon entropy for each sample. Thereby, it could be confirmed that a higher CD4+ T cell repertoire in donors after G-CSF–induced stem cell mobilization is associated with a lower virus reactivation in the related recipients (Figure S8A; Mann–Whitney U-test: CMV: p = 0.034; EBV: p = 0.003) and that the CD8+ TCRβ repertoire is restricted in elderly donors (Figure S8B; Pearson's correlation: preMOB: r = −0.55; p = 0.008; postMOB: r = −0.52; p = 0.013; n = 22).

Discussion

After aHSCT, delayed reconstitution of the donor T cell compartment with respect to cell number but also T cell repertoire is regarded as a risk factor for relapse and infectious complications (3,5,7,21). So far, previous studies focused on the impact of T cell development in the recipients after aHSCT not considering the T cell repertoire of the donors. However, the donor-derived repertoire—especially after G-CSF mobilization—lays the basis for T cell composition in the stem cell recipient and therefore may have great impact for the clinical course after transplantation. Hence, we analyzed the T cell composition of the stem cell donors before G-CSF mobilization and on the day of apheresis. Thanks to the advances in HTS techniques, we were able to assess the TCR beta (TCRβ) repertoire diversity of the donors and correlated the results with the clinical outcome of the corresponding stem cell recipients.

For a more adequate interpretation of the data generated from the stem cell donors, we additionally analyzed flow cytometry–sorted CD4+ and CD8+ T cells isolated from a control cohort of six healthy volunteers. Consistent with published findings, we demonstrated a very similar usage of the Vβ and Jβ gene segments and a Gaussian-like distribution of the CDR3 length in all six individuals (Figure 1A and B) (18). Furthermore, our analysis led to the identification of more dominant clonotypes in CD8+ T cells in comparison to CD4+ T cells (Figure 1C) (22). In consequence, the analyzed CD4+ T cell compartment displayed a higher T cell diversity (Figure 1D). This observation is also in perfect harmony with published data and confirms the reliability of our approach (23). Of special interest are the results derived from the comparison of the T cell composition of (i) different DNA aliquots from the same DNA preparation of the same T cell sort, (ii) different T cell sorts of the same individual at the same time point and (iii) sorted T cells of different individuals. Merely 4.1% and 3.2% of the TCRβ rearrangements were found to be identical after bioinformatics analysis of the HTS data when the two DNA aliquots of the same CD4+ T cell sort or the different CD4+ T cell sorts of the same individual, respectively, were compared. Consistent with the lower diversity of the CD8+ T cell compartment, the percentage of overlapping TCRβ rearrangements was higher in the two DNA aliquots and in both CD8+ T cell sorts of the same individual (13.7% and 9.2%; Figure 1E). In contrast, there was almost no overlap in the TCR repertoires of the CD4+ or CD8+ T cell compartment of different individuals (0.03% and 0.19%; Figure 1F). Much higher overlaps might be anticipated in repeated analysis of samples derived from same individuals. However, the low percentage of identical T cell rearrangements in the same DNA preparation or in identical cell sorts is very logical in light of the enormous T cell heterogeneity in healthy individuals (9). This implies that the DNA aliquots (100 ng equivalent to approx. 14 500 T cells) subjected to multiplex TCRβ PCR and HTS largely consist of individual TCRβ rearrangements not present in the second DNA aliquot or parallel T cell sorts. Only those rearrangements that occur at higher frequencies will be identified in two independent PCRs. This is precisely the reason as to why in CD8+ T cells the overlap is higher in the two DNA aliquots or in the same T cell sort of the same individual as compared to the CD4+ T cells. We are well aware of the fact that we analyzed only a very small fraction of the entire T cell repertoire of a given individual, but we are very much convinced that our HTS approach is representative enough to reliably reflect T cell composition and T cell diversity.

Our TCR analysis approach was then applied to monitor the T cell diversity of the CD4+ and CD8+ T cells of 23 allogeneic stem cell donors before G-CSF administration and on the day of apheresis. These analyses confirmed the higher diversity in the CD4+ T cell compartment, which was in line with our control experiments and indicated no significant changes in the Vβ and Jβ usage, the CDR3 length distribution and the diversity of the CD4+ and CD8+ T cell compartment under G-CSF administration (Figures 2A and D and S2). Furthermore, we neither observed a correlation of diversity with the increase of T cell numbers after G-CSF mobilization in the peripheral blood nor with T cell numbers in the graft as a large randomly drawn sample of the peripheral blood (Figure S5), indicating that G-CSF indiscriminately mobilizes T cells into the peripheral blood. Moreover, we were able to illustrate that the TCRβ repertoire in CD8+ T cells is reduced as a function of age (Figure 2E; Table 3). This observation is supported on a molecular level by previous data showing that CD8+ T cells lose their naïve T cell compartment more rapidly than CD4+ T cells during mid-adulthood and most recently published data confirmed age-related decreases in TCRβ repertoire diversity by HTS (24,25).

The idea of our study is based on the assumption that the T cell composition of the donor after G-CSF mobilization is of crucial importance for the clinical outcome of the stem cell recipient. To this end, we compared the TCRβ rearrangements of the CD4+ and CD8+ T cells of 23 donors after stem cell mobilization with various clinical outcomes of the recipients. Those comparisons revealed no statistically significant correlations with one striking exception: We found that a lower CD4+ T cell diversity in donors after G-CSF mobilization was associated with reactivation of CMV and EBV in the stem cell recipients (Figure 3B; Table 3). These results complement very nicely a previously described association of CMV or EBV infection with lower TCR diversity, which was only analyzed in recipients and not in donors (23). However, it remains unclear to which extent the observed association between lower CD4+ T cell diversity and EBV reactivation could be a side effect of CMV reactivation.

Our data clearly demonstrate that HTS of the amplicons generated by TCRβ multiplex PCR provides a tool of crucial importance to study T cell repertoire and T cell diversity at the level of individual TCRβ gene rearrangements. This is also valid for small sample sizes with limited numbers of T cells. Thus, our approach is also applicable in clinical situations where merely specimens with low T cell counts are available.

In summary, the concept that the T cell repertoire before stem cell transplantation impacts the clinical outcome of recipients after aHSCT could be confirmed by our data. Reduced T cell diversity in the CD4+ T cell compartment of the donors after G-CSF mobilization and, in consequence, limited T cell diversity in the patient after transplantation is associated with viral reactivation. This notion supports concepts that would allow for rapid and early increase of the donor T cell repertoire by adoptive T cell therapy (26–28). Finally, since reduced TCRβ repertoires in donors correlate with an increased likelihood of viral reactivation in stem cell recipients, our TCRβ approach might pave the way for risk assessment of virus reactivation in patients.