|Home | About | Journals | Submit | Contact Us | Français|
Delayed T-cell recovery and restricted T-cell receptor (TCR) diversity after allogeneic hematopoietic stem cell transplantation (allo-HSCT) are associated with increased risks of infection and cancer relapse. Technical challenges have limited faithful measurement of TCR diversity following allo-HSCT. Here we combined 5′-RACE PCR with deep sequencing, to quantify TCR diversity in 28 allo-HSCT recipients using a single oligonucleotide pair. Analysis of duplicate blood samples confirmed that the frequency of individual TCRs was accurately determined. After 6 months, cord blood graft recipients approximated the TCR diversity of healthy individuals, whereas recipients of T-cell-depleted peripheral blood stem cell grafts had a 28-fold and 14-fold lower CD4+ and CD8+ T-cell diversity, respectively. After 12 months, these deficiencies had improved for the CD4+, but not the CD8+ T-cell compartment. Overall, this method provides unprecedented views of T-cell repertoire recovery after allo-HSCT and may identify patients at high risk of infection or relapse.
Allo-HSCT is a potentially curative treatment for a variety of hematologic diseases, including lymphoid and myeloid malignancies. Prior to transplantation, patients undergo conditioning with chemotherapy with or without irradiation, which results in severe immunodeficiency that particularly for the T-cell compartment can take months or years to restore1,2. This prolonged T-cell deficiency predisposes patients to infection and cancer relapse3–6. Strategies that improve T-cell reconstitution and recovery of high TCR diversity could therefore greatly reduce transplant-associated morbidity and mortality7.
Restoration of TCR diversity after allo-HSCT heavily depends on the thymic generation of new naïve T cells8–10. Thymic function, however, diminishes markedly after the onset of puberty, and, in the allo-HSCT setting, is further impaired due to conditioning-associated damage and graft-versus-host disease (GVHD)11,12. Thus, it is unclear how well TCR diversity can be restored, particularly in older patients.
Over the past two decades, several strategies have been developed to probe human TCR diversity. One strategy aims to identify the presence of different TCR families, by using flow cytometry or PCR to determine the usage of different TCR variable (V) genes13,14. A second strategy, called CDR3 size spectratyping, aims to determine polyclonality of the repertoire, by using fluorescent primers to measure length variation of the CDR3 region within each TCR V family15,16. Spectratyping in particular has been useful to document substantial abnormalities in T-cell repertoire composition after allo-HSCT17–19. However, as neither of these strategies is able to measure the frequency of individual TCRs, they can only provide an estimate of repertoire complexity. With the advent of deep sequencing technology, it has now become possible to directly measure TCR diversity with high resolution20–26. Here, we have built on this approach to address two fundamental questions related to T-cell reconstitution after allo-HSCT: how TCR diversity recovers I) over time and II) as a function of different stem cell sources27,28.
28 patients underwent allo-HSCT at Memorial Sloan-Kettering Cancer Center (MSKCC) from April 2010 through September 2011. Patient and treatment characteristics are summarized in Supplementary Table 1. Pre-transplant conditioning varied according to patient age, diagnosis, remission status, extent of prior therapies, and co-morbidities; and consisted of high-dose, reduced-intensity myeloablative and nonmyeloablative regimens35. GVHD prophylaxis for peripheral blood stem cell transplantation was either with T-cell depletion34 or calcineurin inhibitor-based, and anti-thymocyte globulin was used according to protocol or physician preference. Cord blood recipients received mycophenolate mofetil and calcineurin inhibitors36; however, no patient received anti-thymocyte globulin33. Post-transplant granulocyte colony-stimulating factor was used in all patients. Acute and late acute or chronic GVHD were diagnosed clinically with histological confirmation when possible. Staging of GVHD was graded according to standard criteria37,38. All subjects provided Institutional Review Board-approved informed consent for collection of blood samples. Graft samples were not available for analysis.
From each ~8 ml heparinized blood sample, mononuclear cells were isolated by density centrifugation (Lymphocyte Separation Medium, MP Biomedicals). Recovered cells were lysed in RLT buffer (QIAGEN), homogenized using QIAshredder columns (QIAGEN) and stored at −80 °C. For CD4+ and CD8+ T-cell separation, two blood samples were pooled, followed by isolation of the mononuclear cell fraction. Recovered cells were split into two fractions and incubated with either CD4 or CD8 MicroBeads (Miltenyi Biotec). CD4+ and CD8+ T cells were separated using MS columns (Miltenyi Biotec). Eluted cells were lysed, homogenized and stored as above. To determine the efficiency of T-cell separation, eluted cells were stained with antibody to CD14 (clone M5E2, 1:5), antibody to CD4 (clone SK3, 1:20) and antibody to CD8 (clone RPA-T8, 1:5; all BD Pharmingen), and measured on an LSRII flow cytometer (BD Biosciences). Data was analyzed using FlowJo software (TreeStar). For separation of naïve and memory CD8+ T cells, isolated mononuclear cells were stained with antibody to CD45RA (clone HI100, 1:5), antibody to CD45RO (clone UCHL1, 1:5; both BD Pharmingen) and antibody to CD8. Cells were sorted using a FACSAria cell sorter (BD Biosciences) into CD8+CD45RA+CD45RO− (naïve) and CD8+CD45RA–CD45RO+ (memory) fractions.
Total RNA from frozen homogenates was extracted using an RNeasy mini kit (QIAGEN). RACE-Ready cDNA was generated using a SMARTer RACE cDNA Amplification kit (Clontech) and oligo(dT) or random (N–15) primers. 5′-RACE PCR was performed using Advantage 2 Polymerase (Clontech) with Clontech’s universal forward primer and a self-designed universal TCRβ-constant reverse primer compatible with both TRBC genes (5′-GCACACCAGTGTGGCCTTTTGGG-3′). Amplification was performed on a Mastercycler pro (Eppendorf) and was 1 min at 95 °C; 5 cycles of 20 sec at 95 °C and 30 sec at 72 °C; 5 cycles of 20 sec at 95 °C, 30 sec at 70 °C and 30 sec at 72 °C; 25 cycles of 20 sec at 95 °C, 30 sec at 60 °C and 30 sec at 72 °C; 7 min at 72 °C. PCR products were loaded on 1.2% agarose gels (Bio-Rad) and bands centered at ~600 bp were excised and purified using a MinElute Gel Extraction kit (QIAGEN). Purified products were subjected to a second round of amplification to introduce adaptor sequences compatible with uni-directional Roche/454 sequencing. 1/50th of first-round PCR product was amplified using Advantage 2 Polymerase with a hybrid forward primer consisting of Roche’s Lib-L primer B and Clontech’s nested universal primer (5′-CCTATCCCCTGTGTGCCTTGGCAGTCTCAGAAGCAGTGGTATCAACGCAGAGT-3′) and a hybrid reverse primer consisting of Roche’s Lib-L primer A and a self-designed nested universal TCRβ-constant primer (5′-CCATCTCATCCCTGCGTGTCTCCGACTCAG-MID-AACACAGCGACCTCGGGTGGGAA-3′, wherein MID represents the multiplex identifier used to separate pooled samples during sequence analysis). Multiplex identifiers were 6–7 bp long. Amplification was 1 min at 95 °C; 25 cycles of 20 sec at 95 °C, 30 sec at 68 °C and 30 sec at 72 °C; 7 min at 72 °C. PCR products were purified from agarose gels as above. Purified products were sequenced using the GS Junior 454 platform (Roche).
Raw sequence data was converted to FASTA format using MOTHUR software39. Sequences shorted than 125 bp, with uncalled bases, with a Phred quality score average below 30 (base call accuracy < 99.9%)30, or with no exact match to the TCRβ-constant primer or a multiplex identifier were discarded. Resulting FASTA files were uploaded to the IMGT/HighV-QUEST database (http://www.imgt.org/HighV-QUEST/index.action)40. Using IMGT summary files, sequences with out-of-frame rearrangements, with a V- and J-region identity < 80%, with V-region pseudogenes or a CDR3β amino acid junction lacking a 5′ cysteine and 3′ phenylalanine were discarded. Resulting sequences were sorted using Excel (Microsoft) and graphed using Prism 5 software (GraphPad). The inverse Simpson’s diversity index (1/Ds) was calculated using MOTHUR.
TCR diversity was compared using an unpaired Student’s t-test (two groups) or one-way ANOVA with Bonferroni's multiple comparison test (three groups or more). A P value of < 0.05 was considered statistically significant.
T cells typically express only one TCRβ chain, making sequence analysis of TCRβ cDNA a useful measure of TCR diversity. To faithfully evaluate the TCRβ repertoire we made use of 5′-RACE PCR, which allows amplification of all 48 Vβ genes using a single oligonucleotide pair. To test the reproducibility of this approach29, we separately amplified two ~8 ml blood samples from a single patient, 138 days following T-cell-depleted (TCD) peripheral blood stem cell transplantation (TCD no.1; Supplementary Table 1). For comparison, we also amplified two blood samples from four healthy donors (Healthy nos.1–4). All samples were analyzed by Roche/454 sequencing, using a Phred quality score average of 30 to minimize sequence errors30. Among total TCR sequences, all 48 Vβ genes were found (Supplementary Fig. 1), confirming that our approach covers the entire TCRβ repertoire. Comparison of the two patient blood samples showed a highly reproducible pattern of Vβ usage, which differed markedly from healthy donors (Fig. 1a). This suggested substantial clonal expansions in the patient’s repertoire, which were confirmed by digital CDR3 size spectratype profiles (Fig. 1b).
Next, we determined the frequency at which each distinct TCRβ chain, or TCRβ clonotype was present. In 15,902 reads obtained from both patient blood samples, we detected 1,097 clonotypes, with the most frequent clonotype comprising 11.7% of reads (Fig. 1c). In fact, 19 clonotypes were present above 1% and together constituted 70.8% of reads. In contrast, the most abundant clonotype in healthy donors comprised just 2.8 ± 2.6% of reads, and on average only 3 clonotypes were found above 1% (Supplementary Fig. 1). To establish the accuracy of clonotype frequencies, we compared the clonotype distribution of both patient blood samples. Importantly, abundant clonotypes found in one blood sample were found with almost identical frequency in the other blood sample, resulting in a near-perfect correlation of both clonotype distributions (r: 0.98; Fig. 1d). In healthy donors, expanded clonotypes were also detected with high reproducibility, however the average correlation between two blood samples was lower (r: 0.44; Supplementary Fig. 1), primarily because fewer clonotypes passed the threshold for physical presence in a second blood sample (8 ml of ~5 liter total = ~0.16%).
To quantify TCR diversity, we used the inverse Simpson’s diversity index (1/Ds), which sums the frequency of each clonotype31. This index ranges from 1 to ∞ (no to infinite diversity) and is highest when all clonotypes are equally distributed. To test the usefulness of this index, we sorted naïve and memory CD8+ T cells from Healthy no.1 and found a 20-fold higher TCR diversity in the naïve T-cell compartment (Supplementary Fig. 2). Analysis of the patient’s repertoire revealed a very low TCR diversity (1/Ds: 23), which was more than 100-fold lower than the average diversity of four healthy donors (1/Ds: 2,525; Fig. 1e). Therefore, at 138 days after transplant this patient’s T-cell repertoire was highly restricted.
To monitor repertoire recovery over time, we measured three additional timepoints of TCD no.1 at days 147, 194 and 377 after transplant. To our surprise, TCR Vβ usage was very different at each timepoint examined (Fig. 2a), indicating high variability of the T-cell repertoire. To evaluate individual clonotypes, we first determined whether clonotype frequencies were reliably measured. On day 147, we again found a near-perfect correlation between two blood samples (r: 0.99), and the same held true for days 194 and 377 (r: 1 and 0.98, respectively; Fig. 2b). Comparison of the repertoire on days 138 and 147, however, revealed dramatic shifts in clonotype frequencies (Fig. 2c). Although some clonotypes were present at roughly similar frequency on both days, several others differed by more than 100-fold; resulting in a low similarity between both T-cell repertoires measured just 9 days apart (r: 0.24). Importantly, these repertoire shifts coincided with Epstein-Barr virus (EBV) reactivation in the patient, which was first detected on day 145 and peaked on day 147 (Supplementary Fig. 3). We identified the 9th most abundant clonotype on day 147 (TRBV29–1, CDR3β CSVGTGGTNEKLFF) as being specific for the HLA-A2-restricted BMLF1280 epitope from EBV32. While undetectable on days 138 and 194, this clonotype comprised 2.9% of reads on day 147 and 0.1% on day 377, highlighting the potential of our method to track antigen-specific clonotypes. Despite apparent resolution of EBV reactivation by day 158, subsequent blood samples continued to reveal clonotype frequencies fluctuating over orders of magnitude. Thus, the repertoire of two samples on the same day was highly similar (r: 0.99), whereas on different days it was markedly distinct (r: 0.25; Fig. 2d). A notable exception was the repertoire of days 138 and 377, which revealed a surprisingly large degree of similarity (r: 0.87; Supplementary Fig. 3). Despite profound changes in repertoire composition, TCR diversity did not increase over time (1/Ds: 23 and 19 for days 138 and 377, respectively; Fig. 2e). Therefore, between 4.5 and 12.5 months after transplant, the complexity of this patient’s T-cell repertoire did not improve.
To evaluate whether our method might enable stratification of allo-HSCT recipients according to their T-cell repertoire, we measured TCR diversity in 27 patients at either 6 or 12 months after conventional (Conv) or TCD peripheral blood stem cell transplantation, or double-unit umbilical cord blood (DUCB) transplantation (Fig. 3a–c). Since analysis of TCD no.1 had suggested substantially greater TCR diversity in CD4+ compared to CD8+ T cells (Supplementary Fig. 4), we separately analyzed both T-cell compartments for all ensuing patients, as well as for five healthy donors (Fig. 3d). After determining the TCR diversity of each individual, we established the following: I) for all stem cell sources as well as for healthy donors, CD4+ T-cell diversity was ~50-times higher than CD8+ T-cell diversity (1/Ds: 4,665 and 81, respectively; Fig. 3e,f). II) Regarding CD4+ T cells, healthy donors had the highest TCR diversity (1/Ds: 15,470), followed by DUCB after 12 months (1/Ds: 5,069), DUCB after 6 months (1/Ds: 3,745), Conv after 12 months (1/Ds: 3,298), TCD after 12 months (1/Ds: 1,871), Conv after 6 months (1/Ds: 674) and TCD after 6 months (1/Ds: 132). Therefore, DUCB recipients had the highest TCR diversity of all patients and had a significantly (28-fold; P = 0.033) more diverse CD4+ T-cell repertoire compared to TCD recipients after 6 months. Importantly, this increased TCR diversity also correlated with a substantially greater fraction of naïve CD4+ T cells in DUCB compared to TCD recipients (Supplementary Fig. 5). Although TCD recipients had limited CD4+ T-cell diversity after 6 months, this diversity was 14-fold higher after 12 months, reducing the difference with DUCB recipients to 3-fold. III) Regarding CD8+ T cells, DUCB recipients again had the highest TCR diversity of all patients, which was 14-fold higher than TCD recipients after 6 months and 17-fold higher after 12 months, thereby reaching statistical significance (P = 0.012).
Using above data, we investigated several clinical parameters that could influence T-cell repertoire recovery. We found no significant impact of age or donor on TCR diversity (Fig. 4a,b). Interestingly, acute GVHD (grade 2 or 3) and systemic steroid treatment were associated with higher TCR diversity, suggesting that these variables do not restrict repertoire recovery (Fig. 4c,d). In contrast, cytomegalovirus (CMV) or EBV infection were associated with lower TCR diversity (Fig. 4e).
Within each stem cell group, we identified patients who had normal T-cell counts, but very low TCR diversity after one year (Supplementary Fig. 6). To investigate repertoire recovery during the second year of transplant, we reanalyzed TCR diversity in three of these patients after 19–21 months. For comparison, we also reanalyzed three healthy donors. We found stability in the CD4+ T-cell repertoires of Conv no.6 and TCD no.8, and the CD8+ T-cell repertoire of DUCB no.7 (Fig. 5a). Similar stability was observed in the CD8+ T-cell repertoires of healthy donors when measured over 109 days, whereas over 299 days there was somewhat greater divergence (Fig. 5b). Despite occasional changes in clonotype frequencies, TCR diversity in healthy donors was stable (Supplementary Fig. 7). Interestingly, in Conv no.6 the frequency of abundant clonotypes had substantially decreased over time, resulting in a 10-fold higher CD4+ T-cell diversity (Fig. 5c). In contrast, there was no diversification in the other two patients. Together, these data illustrate the potential of our method to identify patients as well as transplant protocols that are associated with either greater or lesser T-cell repertoire recovery.
In this study, we established a method to reproducibly and accurately measure human TCR diversity. By combining 5′-RACE PCR with deep sequencing, this method assesses the entire TCRβ repertoire using a single oligonucleotide pair, thereby eliminating amplification bias. This contrasts with TCR sequencing methods based on gDNA20–23, which have to use many different oligonucleotides for amplification, making some degree of bias unavoidable. While 5′-RACE PCR provides a clear advantage, a limitation is that it requires RNA, and thus changes in TCR transcription could skew the frequency of particular clonotypes. Although we used Roche/454 sequencing here, our approach should be readily adaptable to other platforms with greater sequence coverage, which could help to identify infrequent TCRs.
We validated our method by measuring T-cell repertoire recovery in allo-HSCT recipients, in whom limited TCR diversity is linked to susceptibility to infection and cancer relapse. Although we found significant improvement in TCR diversity over time, there was substantial variability in the rate of recovery between different stem cell sources. Most notably, DUCB recipients had a 28-fold higher CD4+ T-cell diversity compared to TCD recipients after 6 months. Importantly, this is highly consistent with clinical findings, which have shown that after 6 months, DUCB recipients have a low incidence of infection33, have higher CD4+ T-cell numbers than TCD recipients33,34 and have a lower rate of leukemia relapse27. Although many variables could contribute to this differential repertoire recovery, at least partially it can be explained by the fact that DUCB recipients receive ~7,000-fold more T cells in their graft and transplantation is performed without T-cell-depleting regimens33.
Besides allo-HSCT, this method should be useful to characterize T-cell immunity in other clinical settings of immune deficiency, autoimmunity and tumor immunity. Ultimately, the ability to measure T-cell repertoire complexity with great precision should guide the way for novel therapeutic approaches aimed at immune regeneration.
We are grateful to all patients and healthy donors who kindly volunteered to participate in this study. We thank G. Redelman-Sidi for collection of blood samples, I. Leiner for processing of blood samples, C. Ubeda-Morant for assistance with data analysis, J. Li and A. Viale of the MSKCC Genomics Core Laboratory for performing Roche/454 sequencing and R. Jenq for critical reading of the manuscript. This work was supported by the Netherlands Organization for Scientific Research (NWO) (Rubicon fellowship to J.W.J. van Heijst), the US National Institutes of Health (NIH) (grants HL069929, CA107096 and AI080455 to M.R.M. van den Brink, and AI042135, AI039031 and CA023766 to E.G. Pamer), and The Tow Foundation (to E.G. Pamer). Additional support was received from the US Department of Defense (USAMRAA Award W81XWH-09-1-0294), the Radiation Effects Research Foundation (RERF-NIAID), The Lymphoma Foundation, Alex’s Lemonade Stand, The Geoffrey Beene Cancer Research Center at Memorial Sloan-Kettering Cancer Center, and The Peter Solomon Fund (to M.R.M. van den brink), The Translational and Integrative Medicine Research Fund of Memorial Sloan-Kettering Cancer Center, Cycle for Survival, The New York Community Trust, and When Everyone Survives (to M.A. Perales), and The Experimental Therapeutics Center of Memorial Sloan-Kettering Cancer Center funded by Mr. W.H. Goodwin and Mrs. A. Goodwin (to M.R.M. van den brink and M.A. Perales).
AUTHOR CONTRIBUTIONSJ.W.J.v.H. designed and performed the study, analyzed data and wrote the manuscript. I.C. collected clinical data. L.B.L. performed Roche/454 sequencing. D.W.S. compared naïve and memory CD8+ T cells. G.D.W., A.M.R.G. and J.L.N. provided patient samples. M.R.M.v.d.B. designed and supervised the study. M.A.P. designed and supervised the study, and provided patient samples. E.G.P. designed and supervised the study, analyzed data and wrote the manuscript.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.