Dynamics of Individual T Cell Repertoires: from Cord Blood to Centenarians Olga V

Home , Junctional diversity

Dynamics of Individual T Cell Repertoires: From Cord Blood to Centenarians Olga V. Britanova, Mikhail Shugay, Ekaterina M. Merzlyak, Dmitriy B. Staroverov, Ekaterina V. Putintseva, Maria A. This information is current as Turchaninova, Ilgar Z. Mamedov, Mikhail V. Pogorelyy, of September 27, 2021. Dmitriy A. Bolotin, Mark Izraelson, Alexey N. Davydov, Evgeny S. Egorov, Sofya A. Kasatskaya, Denis V. Rebrikov, Sergey Lukyanov and Dmitriy M. Chudakov

J Immunol published online 13 May 2016 Downloaded from http://www.jimmunol.org/content/early/2016/05/12/jimmun ol.1600005

Supplementary http://www.jimmunol.org/content/suppl/2016/05/12/jimmunol.160000 http://www.jimmunol.org/ Material 5.DCSupplemental

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists by guest on September 27, 2021

• Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2016 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published May 13, 2016, doi:10.4049/jimmunol.1600005 The Journal of Immunology

Dynamics of Individual T Cell Repertoires: From Cord Blood to Centenarians

Olga V. Britanova,*,†,‡,1 Mikhail Shugay,*,†,‡,1 Ekaterina M. Merzlyak,*,1 Dmitriy B. Staroverov,*,1 Ekaterina V. Putintseva,* Maria A. Turchaninova,*,†,‡ Ilgar Z. Mamedov,*,‡ Mikhail V. Pogorelyy,* Dmitriy A. Bolotin,*,†,‡ Mark Izraelson,*,‡ Alexey N. Davydov,‡ Evgeny S. Egorov,*,†,‡ Sofya A. Kasatskaya,* Denis V. Rebrikov,†,x Sergey Lukyanov,*,† and Dmitriy M. Chudakov*,†,‡

The diversity, architecture, and dynamics of the TCR repertoire largely determine our ability to effectively withstand infections and malignancies with minimal mistargeting of immune responses. In this study, we have employed deep TCRb repertoire sequencing

with normalization based on unique molecular identiﬁers to explore the long-term dynamics of T cell immunity. We demonstrate Downloaded from remarkable stability of repertoire, where approximately half of all T cells in peripheral blood are represented by clones that persist and generally preserve their frequencies for 3 y. We further characterize the extremes of lifelong TCR repertoire evolution, analyzing samples ranging from umbilical cord blood to centenarian peripheral blood. We show that the fetal TCR repertoire, albeit structurally maintained within regulated borders due to the lower numbers of randomly added nucleotides, is not limited with respect to observed functional diversity. We reveal decreased efﬁciency of nonsense-mediated mRNA decay in umbilical cord

blood, which may reflect specific regulatory mechanisms in development. Furthermore, we demonstrate that human TCR rep- http://www.jimmunol.org/ ertoires are functionally more similar at birth but diverge during life, and we track the lifelong behavior of CMV- and EBV- specific T cell clonotypes. Finally, we reveal gender differences in dynamics of TCR diversity constriction, which come to naught in the oldest age. Based on our data, we propose a more general explanation for the previous observations on the relationships between longevity and immunity. The Journal of Immunology, 2016, 196: 000–000.

he native human TCR repertoire changes throughout life specificities that is available to respond to new challenges and to based on Ag experience, exhaustion, and nonuniformity sustain a balanced network of regulatory interactions (1, 7–12). T of T cell proliferation at the periphery, as well as on the The potential for revealing a detailed view of individual TCR by guest on September 27, 2021 aging-associated evolution and involution of the thymus (1–6). At repertoires, which opened with development of high-throughput any given moment, our accumulated population of expanded clonal sequencing techniques (13–16), has attracted many laboratories to T cells represents our up-to-date Ag experience and determines the this presently superficially explored field. Multiple studies have current state of our adaptive immune defenses, whereas the remaining been performed in recent years to characterize TCR repertoires diversity of the naive T cell population represents the reservoir of new either in general (13–22) or at the level of functional T cell subsets (17, 18, 23–33). However, these new opportunities have also brought new hidden pitfalls, which have not always been predicted, *Shemyakin–Ovchinnikov Institute of Bioorganic Chemistry of the Russian Acad- emy of Sciences, Moscow 117997, Russia; †Pirogov Russian National Research recognized, and proportionally addressed in the works published to Medical University, Moscow 117997, Russia; ‡Central European Institute of Tech- date. As long as the analyzed material is of sufficient quality and x nology, Masaryk University, 625 00 Brno, Czech Republic; and Vavilov Institute of quantity, shallow profiling at the scale of up to several thousands of General Genetics of the Russian Academy of Sciences, Moscow 119991, Russia 1 expanded clonotypes can usually be reliably interpreted. At the O.V.B., M.S., E.M.M., and D.B.S. contributed equally to this work. same time, deep and reliable profiling of hundreds of thousands and ORCIDs: 0000-0001-7826-7942 (M.S.); 0000-0001-6789-9442 (D.B.S.); 0000-0003- millions of analyzed T cells and sequencing reads remains chal- 4421-3844 (E.V.P.); 0000-0002-8484-6067 (D.A.B.); 0000-0002-6495-7173 (E.S.E.); 0000-0002-5522-7443 (S.A.K.); 0000-0002-1884-1807 (D.V.R.); 0000-0003-0430- lenging. This is compounded by biased amplification and accu- 790X (D.M.C.). mulation of PCR and sequencing errors (34–39), which must be Received for publication January 6, 2016. Accepted for publication April 16, 2016. distinguished from the true highly homologous TCR variants This work was supported by Russian Science Foundation Project 16-15-00149. M.S. originating from convergent recombination events (18, 19, 40). and E.V.P. are supported by individual fellowship Russian Foundation for Basic Potential cross-sample contamination and sequencing artifacts can Research Grants 16-34-60179 and 16-34-60178. further hamper accurate comparison of TCR repertoires in different The raw sequencing data in this article have been submitted to the National Center for Biotechnology Information Sequence Read Archive database (https://www.ncbi. samples. As we have accumulated more experience with deep TCR nlm.nih.gov/Traces/sra/) under accession number PRJNA316572. repertoire profiling, these limitations have become clearer, along Address correspondence and reprint requests to Dr. Dmitriy M. Chudakov, Shemyakin– with the possible solutions for obtaining more quantitative data on Ovchinnikov Institute of Bioorganic Chemistry of the Russian Academy of Sciences, clonotype diversity, minimizing the generation of artificial and loss Miklukho-Maklaya 16/10, Moscow 117997, Russia. E-mail address: [email protected] of real receptor diversity (35, 37–39, 41, 42). The online version of this article contains supplemental material. In this study, we employed our present molecular and software Abbreviations used in this article: MIG, molecular identifier group; NMD, nonsense- solutions to track both deeply and quantitatively the individual mediated decay; UCB, umbilical cord blood. dynamics and stability of T cell clones during 3 y. We also reveal Copyright Ó 2016 by The American Association of Immunologists, Inc. 0022-1767/16/$30.00 characteristic features of TCR repertoire at the polar extremes of

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1600005 2 DYNAMICS OF INDIVIDUAL T CELL REPERTOIRES the human life, including umbilical cord blood (UCB) and peripheral described previously (5) and analyzed via Illumina HiSeq 2500 blood samples obtained from centenarians, and we describe the paired-end 100 + 100 nt sequencing. In our downstream bio- general dynamics and extent of changes in the structure of the TCR informatics analysis, we grouped sequencing reads labeled with repertoire throughout life. Furthermore, we reveal gender differ- the same unique cDNA identifier to the MIG. Each nucleotide ences in adaptive immunity aging and link this finding to previous within the TCRb CDR3 sequence in each MIG was identified data on population studies. according to the cumulative quality in the assembled sequencing reads. This approach improves sequencing quality and eliminates Materials and Methods most PCR and sequencing errors according to the Safe-seqS logic Sample collection described by Kinde et al. (44). Because MIG-based analysis counts initial cDNA events, it also eliminates amplification and This study was approved by the local ethics committee and conducted in ac- sequencing biases and artifacts and thus significantly improves cordance with the Declaration of Helsinki. All donors were informed of the final use of their blood and signed an informed consent document. The cohort in- quantitative power in repertoire data analyses (5, 39, 45–48). cluded 65 healthy individuals aged 6–103 y (see Table II). We excluded indi- Importantly, we have accounted for artificial molecular identi- viduals who had previously been diagnosed with cancer or autoimmune disease. fier variants that result from errors within the identifier sequence We obtained 6–10 ml peripheral blood from each donor. Each sample was itself (Supplemental Fig. 1A). To remove those, we filtered out collected into a number of EDTA-treated Vacutainer tubes (BD Biosciences, Franklin Lakes, NJ). PBMCs were extracted using Ficoll-Paque (Paneco, Kirov, the MIGs in which the molecular identifier had a single-mismatch Russia) density gradient centrifugation. Before RNA extraction, all PBMC parent with a $10-fold relative quantity of sequencing reads. samples were initially frozen, then thawed and incubated overnight in RPMI Such correction is obligatory for accurate quantification of true,

1640 with 10% human serum (First Link, Birmingham, U.K.) in the presence of distinct cDNA events. We also counted only those uniquely la- Downloaded from 15 U/ml IL-2 (Roche, Basel, Switzerland). According to our experience, such beled cDNA molecules that were sequenced at least twice, ﬁlter- treatment increases the RNA expression levels of all TCR genes. For total RNA extraction, TRIzol reagent (Life Technologies, Carlsbad, CA) was used. ing out cDNA molecules covered by just a single sequencing read. This threshold additionally protects from artifacts and cross-sample Flow cytometry contaminations, because at least two independent sequencing read Flow cytometry analysis was performed on a Cytomics FC 500 (Beckman events on the solid phase of Illumina are analyzed for each cDNA

Coulter). Data analysis was carried out using the Cytomics RXP analysis (Supplemental Fig. 1B). Notably, this filter is blind with respect to http://www.jimmunol.org/ program (Beckman Coulter). For staining, aliquots of PBMCs were in- clonotype size, and it does not result in selective loss of naive cubated for 20 min at room temperature with anti-human Abs CD3-PC7 (clone UCHT1, eBioscience), CD27-PC5 (clone 1A4CD27, Invitrogen), T cells diversity, which would be the case if we filtered out the true CD4-PE (clone 13B8.2, Beckman Coulter), and CD45RA-FITC (clone JS- singletons, that is, clones represented by a single cDNA molecule. B3, eBioscience) and then washed twice with PBS. Because we sought to collect the maximum number of cDNA TCRb libraries preparation and data analysis events for each sample in the deep profiling, we did not apply the full MiGEC error-correction logic (39), which would require over- cDNA synthesis and template switch to the SmartNNNa 59-adaptor with a sequencing with a threshold of at least five reads per cDNA. In- unique molecular identifier (59-AAGCAGUGGTAUCAACGCAGAGUN- NNNUNNNNUNNNNUCTT(rG)(rG)(rG)(rG)-39), two-stage seminested stead, we employed the “eliminate these errors” mode in MiiTCR by guest on September 27, 2021 PCR amplification, multiplexing, and sequencing were performed as de- software with MIG data for the assembly of clonotypes and addi- scribed (5). Raw TCR sequencing data were processed using the MiGEC tional frequency-based error correction (37). pipeline (39). Briefly, checkout utility was used for data sample barcode To understand the quantitative power of the method, we first matching and unique molecular identifier sequence extraction. The data analyzed a frozen sample of a single blood draw from individual 1, were then assembled using the assemble utility with the erroneous identifiers filtering option and at least two reads per molecular identifier group (MIG) which we divided into two unequal portions of PBMCs, and two oversequencing threshold. MIG oversequencing and erroneous identifiers separately frozen unequal PBMC samples obtained from a single histograms were calculated using the histogram utility. As in the present blood draw of individual 2. Starting from 3 to 10 million PBMCs, experimental setup, the deep repertoire sequencing was chosen in favor of TCRb sequencing with MIG-based analysis yielded 1–2.6 million high sequence coverage of individual cDNA molecules, and additional error b correction was performed at the final stage of data processing. This was distinct, reliably sequenced TCR cDNA molecules per sample, carried out using MiTCR software (37) with the “eliminate these errors” which included 0.4–0.6 million distinct TCRb clonotypes per option for frequency-based error correction. All statistical analyses, clus- sample (Table I). Analysis of data on the two split portions of a tering procedures, model fitting, and plotting were performed in R. For the single thawed PBMC sample of individual 1 demonstrated almost cluster analysis andpffiffiffiffiffiffiffiffiffiffiffiffi multidimensional scaling, the similarity measure 2 F 52 +N f f absolute correlation in clonotype frequencies (r = 1.00, Fig. 1A). 2 log10 i¼1 xi yi was used, where fxi and fxy are the frequencies of an overlapped clonotype i in sample x and y, respectively, and N is the total We therefore concluded that library preparation, sequencing, and number of overlapped clonotypes. This measure accounts for both overall data analysis pipeline introduce minimal quantitative differences overlap size and correlation of overlapping clonotype frequencies (42). between samples. However, we also observed that sampling, freez- Normalization of repertoires overlap ing, and thawing procedures may introduce some variance, because clonotype frequencies in the two separately processed and frozen The number of intersecting clonotypes (see Fig. 2E) for a fixed pair of PBMC samples of individual 2 demonstrated lower correlation (r2 = samples d12 is proportional to the product of sample diversities d1 and d2, d = A 3 d 3 d (20). Still, on the population level the coefficient A may 0.92, Fig. 1A). In other analogous experiments, correlation between 12 1 2 ∼ ∼ vary depending on sample pair, that is, A = A(d1,d2). With this in mind, we the two independent replicate samples ranged from 0.8 to 0.9 for have fitted an empirical power model to our data and determined that A = three sample pairs (data not shown). Correspondingly, the relative 20.2 2 A0 3 (d1 3 d2) with r = 0.96 (n = 2628 pairs). Therefore, we have 0.8 homeostatic space occupied by shared clonotypes and the size of the selected D = d12/(d1 3 d2) as the measure for normalized overlap of unique TCRb amino acid variants. top clonotypes were nearly identical between replicas, confirming the high accuracy of the method (Fig. 1B). Results Stability of the TCR repertoire Optimized approach for deep cDNA-based TCR repertoire Earlier studies have demonstrated that human virus-specific memory profiling T cell clones may be stable for many years (49–54). However, as far RNA-based TCRb libraries, where each cDNA molecule was la- as we know, the full extent of changes that occur in a healthy in- beled with a unique molecular identifier (43), were prepared as dividual’s TCR repertoire over the years has not yet been tracked The Journal of Immunology 3

Table I. Normalized TCRb proﬁling data for the two replicas or time points

Raw Paired-End Analyzed cDNA Common Correlation of Year of Age Sequencing Molecules Nucleotide TCRb Nucleotide Overlap Shared Clonotype Donor Sampling (y) PBMCs Reads with CDR3 with CDR3a CDR3 Clonotypes Clonotypes (%)b Frequencies (r2) Individual 1 2013 87 5 3 106 23,918,642 2,343,918 402,378 102,903 72 1.00 3 3 106 25,684,335 2,649,250 401,771 Individual 2 2013 25 10 3 106 6,207,486 1,260,339 570,534 37,917 43 0.92 5 3 106 6,101,552 1,032,585 467,651 Individual 3 2011 27 20 3 106 47,765,136 7,038,180 1,828,488 78,009 48 0.66 2014 30 9 3 106 10,464,702 1,531,400 760,839 Individual 4 2011 47 20 3 106 44,028,675 6,040,218 965,644 36,854 64 0.82 2014 50 7.9 3 106 2,713,450 518,116 137,682 acDNA labeled by distinct molecular barcodes, each sequenced at least twice. bRelative overlap was calculated as average percentage occupied by shared clonotypes (see Materials and Methods). with deep sequencing. To directly visualize how stable the human Thus, we observed minimal differences in the expanded T cell T cell repertoire is over time, we performed deep TCRb profiling repertoires across this time period, comparable with the differ- for peripheral blood samples obtained from two systemically healthy ences between biological replicas obtained at the same time point. adult individuals at two different time points 3 y apart (Table I). Moreover, persistent clones cumulatively constituted from 45 to Downloaded from Comparison of the datasets obtained from the two time points 75% of the total T cell counts in peripheral blood (Fig. 1B). The revealed 78,009 and 36,854 clones present in both samples for percentage occupied by persistent clones was in a good agreement individuals 3 and 4, respectively. The overlap calculated as the sum with the percentage occupied by Ag-experienced T cells measured of geometric means of frequencies of shared clonotypes between by flow cytometry (41 and 79% for individuals 3 and 4, respec- the two time points was comparable to that of the two replicate tively), indicating that most clonal expansions persist for years. To samples obtained for individual 2. reveal the exact nature of persistent clones, we additionally per- http://www.jimmunol.org/ Most persistent clones generally preserved their relative fre- formed deep TCRb repertoire sequencing for the FACS-sorted quencies during this 3-y span (R = 0.861 and 0.905 for individuals T cells of individual 3. This analysis revealed that tracked 3 and 4, respectively; Fig. 1A). On average, the top 1000 clones clones are represented by both CD8+ and CD4+ Ag-experienced exhibited a 2- and 1.8-fold change in concentration for individuals T cells in comparable proportions (∼5:3). Notably, tracked CD8+ 3 and 4, respectively. Of the top 1000 clones observed at the first clones were mostly represented by large clonal expansions, time point, 952 and 882 were identified at the second time point whereas CD4+ clones were characterized by lower expansion and for individuals 3 and 4, respectively. Similarly, of the top 1000 ∼5-fold higher diversity. Therefore, some portion of stable but clones observed at the second time point, 962 and 966 clones were low-frequency CD4 clones could be missed in our analysis due to identified at the first time point for individuals 3 and 4, respectively. sampling limitations. by guest on September 27, 2021

FIGURE 1. Stability of individual TCR repertoire over time. (A) Scatter plots of clonotype frequencies in two replicas obtained from individuals 1 and 2, and in two sampling time points for individuals 3 and 4. Clonotypes shown are present in both sampling time points and replicas. Point size is scaled according to clonotype frequency. (B) Stack plots of clonotype concentrations for the four individuals. Two replicas are shown for individuals 1 and 2, and two time points are shown for individuals 3 and 4. The concentration changes of shared clonotypes are shown with colored areas. 4 DYNAMICS OF INDIVIDUAL T CELL REPERTOIRES

The space occupied by persistent clones enlarged with age in the (Fig. 2C). The highest individual TCRb diversity observed in elder individual, but shrank in the younger. Hypothetically, this UCB correlated with the highest percentage of naive T cells, and could reflect faster turnover of T cell clones at younger ages and both parameters declined at later ages with similar kinetics continuing expansion of long-lived memory T cell clones in the (Fig. 2D). Therefore, importantly, low TdT activity during the elderly, although this has to be confirmed in further studies using prenatal period lowers the entropy but does not limit the observed deep and long-term tracking of TCR repertoire dynamics on a TCR diversity. 3) At the same time, low TdT activity in the fetal cohort of donors of different ages. period influenced relative cross-individual similarity of UCB repertoires. We studied the average number of TCRb clonotypes Dynamics of TCR repertoires shared between two people within each age group, normalized In order to characterize the lifelong changes that affect TCR based on sample diversity (see Materials and Methods). These repertoires, we analyzed TCRb repertoires for 8 UCB and 65 metrics do not account for clonal size and reflect the general peripheral blood samples of healthy individuals of different ages similarity of naive T cell repertoire diversities. Correspondingly, (Table II, raw sequencing data deposited in the Sequence Read this overlap was high between UCB samples, where low TdT Archive database, accession number PRJNA316572). This cohort activity limits theoretical diversity of produced TCR b variants included the 39 individuals previously published in Britanova (67), but further decreased to lower levels and remained stable et al. (5), which we reanalyzed in the present study for better with aging (Fig. 2E). 4) Based on flow cytometry data, we ob- resolution as described above. To equalize and correctly compare served different dynamics in the percentages of naive CD4 and these 73 samples, we analyzed 300,000 randomly chosen TCRb naive CD8 T cells within the total T cell population. Naive CD4+ cDNA molecules per sample, each labeled with a unique molec- T cells represented more than half of all T cells in UCB, whereas Downloaded from ular identifier. Considering the excess (millions) of T cells used naive CD8+ T cells represented ∼20%. The ratio of CD4 and CD8 for RNA purification, that is approximately equivalent to 300,000 naive subsets tend to equalize in the young age (Fig. 2F), along randomly chosen T cells. Additionally, we performed flow cytom- with the increase in proportion of Ag-experienced CD8+ T cells etry analysis for the population of naive T cells in the CD8+ and (Supplemental Fig. 3A, 3B), although additional analysis for a CD4+ populations (CD27high/CD45RAhigh,in68of73samples, larger cohort is required to confirm statistical significance of these +

Supplemental Fig. 2) and counted absolute numbers of CD3 cells changes. The percentage of naive T cells decreased with age (Fig. http://www.jimmunol.org/ (in36of73samples). 2D), with a more rapid decline observed within the CD8+ subset This comparative analysis revealed a number of hallmarks that (Fig. 2F). As a result, naive CD4+ T cells dominated over the naive characterize T cell immunity at different life periods. 1) When CD8+ T cells in UCB and in old age, but not in the young age compared with any other age group, UCB samples were charac- (Fig. 2F, Supplemental Fig. 3C). Considering the decrease of total terized by a significantly higher percentage of nonfunctional TCRb T cell counts in blood with aging (68) (Supplemental Fig. 3E), the mRNA molecules representing either out-of-frame TCR variants decrease of the absolute counts of naive T cells per microliter of or those containing a stop codon within their CDR3 (p , 0.05 peripheral blood was even more dramatic (Supplemental Fig. 3F). versus 6–25 y age group and p , 0.001 versus other age groups, We further analyzed the ratio of the proportion of naive T cells two-tailed t test, Holm correction; Fig. 2A). Because the level of within the CD4+ subset to the proportion of naive T cells within by guest on September 27, 2021 nonfunctional TCRb mRNA is determined by nonsense-mediated the CD8+ subset. This parameter is not dependent on the CD4+/CD8+ decay (NMD) mechanisms (55–57), this should be a consequence cell ratio and thus reflects to which extent the CD4+ subset is naive of NMD suppression during the prenatal period. Interestingly, it compared with the CD8+ subset, irrespective of their relative abun- has been demonstrated that low oxygen slows down NMD (58), so dance. This analysis confirmed our previous observation (5) that the one of the possible explanations is the lower oxygen in UCB T cell population of the oldest individuals is specifically character- compared with peripheral blood (59). At the same time, this could ized by a relatively more naive state of CD4+ relative to CD8+ Tcell also be the consequence of more specific NMD regulatory mecha- subset (Supplemental Fig. 3D). 5) We further examined how the nisms that may play important roles in development (58, 60–62). 2) functional similarity of TCR repertoires changed with age. To this UCB samples were characterized by a significantly lower numbers end, we analyzed how the share of repertoire represented by amino of added N nucleotides within CDR3 (average per clonotype) than acid CDR3 clonotypes that are common for the two individuals 2 other age groups (p , 5 3 10 13, two-tailed t test, Holm correction; correlates with their age (accounting for relative clonotype size, as Fig. 2B). Similar observations were previously reported in both opposed to the normalized number of shared sequence variants dis- mice and humans, based on a few sequenced clonotypes and TdT cussed above and shown in Fig. 2E). This analysis revealed that RNA expression analysis (63–66). In this study, we confirm this functionally TCR repertoires are more similar in childhood, and they feature of human TCR repertoires at the deep sequencing scale. subsequently diverge from this common core repertoire with aging Notably, this feature did not influence TCRb diversity observed per (r2 = 0.54, n =73,p =43 10215; Fig. 3A–C). This high share of 300,000 T cells, which was the highest in UCB compared with other common T cells in UCB samples and in childhood probably re- samples. Almost each captured cDNA encoded a distinct TCRb sults from multiple public clonotypes produced due to recombi- variant, resulting in .250,000 distinct clonotypes per sample natorial biases (19, 69), which are abundantly represented within

Table II. Characteristics of donors by age group

Average Age Number of Patients Observed TCRb CDR3 Diversity Donor Groups (Range) (y) (Female/Male/NK) per 3 3 105 T Cells (3105) UCB 0 8 (2/3/3) 2.7 6 0.1 Young 16 (6–25) 11 (4/7/0) 2.2 6 0.3 Middle age 40 (30–50) 13 (7/6/0) 1.6 6 0.5 Aged 60 (51–75) 18 (11/7/0) 1.4 6 0.6 Long-lived 92 (85–103) 23 (15/8/0) 0.9 6 0.4 NK, not known. The Journal of Immunology 5 Downloaded from http://www.jimmunol.org/

FIGURE 2. Overall changes in T cell immunity. Data are shown for UCB samples and peripheral blood samples collected from donors of different age groups (denoted by different colors). (A) Percentage of nonfunctional TCR b mRNA molecules (h2 = 0.27, p =1024). (B) Average number of added N nucleotides within CDR3 per clonotype (h2 = 0.44, p =23 1028). (C) TCRb diversity observed per 300,000 T cells (h2 = 0.65, p =23 10215). (D) Percentage of naive T cells among all T cells as measured by FACS analysis (h2 = 0.72, p =93 10217). (E) Normalized number of TCRb clonotypes shared between pairs of samples within each age group (h2 = 0.18, p , 2 3 10216). (F) Percentage of naive CD4+ (light) and naive CD8 (dark) T cells among all T cells as measured by FACS analysis (h2 = 0.52 and 0.79, p =23 1029 and 4 3 10220). All parameters have a significant association with age 2 as shown by ANOVA test. Effect size h and p values are given in brackets. by guest on September 27, 2021 naive but not Ag-experienced T cells. Furthermore, input of pre- clones, or from age-selection barrier favoring longevity of indi- natal TdT-free TCR repertoire in UCB and potentially in child- viduals with either lower CMV and EBV burden (74, 75), or from hood should add to abundance of high-frequency public clonotypes lower clonal dominance in CMV- and EBV-specific responses. 7) (Fig. 2B). With aging, the proportion of naive T cells goes down. We performed comparison of TCRb repertoire diversity between Additionally, high-frequency public clonotypes, initially abun- donors of different sex within each age group. This analysis dant, gradually decrease their frequency within the naive T cell revealed that TCRb diversity decreases more rapidly in middle subset (M. Pogorelyy and D. Chudakov, unpublished observations age for males than for females (Fig. 4A, p = 0.02, two-tailed t test for several individuals), because peripheral homeostatic prolifer- with adjustment for multiple comparison). This corresponds with ation that supports naive T cell numbers after decline of the thy- a faster decline in the percentage of naive T cells in middle- mic function (70) is not uniform (71–73). Both processes should aged males within both the CD4+ and CD8+ subsets (Fig. 4B, result in a decrease of the share occupied by common T cells with Supplemental Fig. 4). At the same time, no gender differences aging. These considerations are supported by a decrease in the were observed at very old age. This is aligned with the fact that degree of convergence, measured as the average number of nu- the survival advantage of females versus males vanishes with cleotide CDR3 sequences per each amino acid CDR3 variant aging (76) (see below). (Fig. 3D). 6) Next we analyzed repertoires to identify amino acid TCRb clonotypes annotated as specific for common pathogens, Discussion including CMV and EBV, collected via literature mining (Supple- Capable of undergoing complex dynamic changes and intelligent mental Table I). The number of such clonotypes decreased with adaptation to various challenges, our T cell immunity can also be age (r2 = 0.31, n = 73, p =13 1027; Fig. 3E). At the same time, remarkably stable over time. In this study, we demonstrate that, in the average size of known viral-specific clonotypes increased with middle age, persistent clones may occupy more than half of the age, with a 7-fold growth observed for the 51–75 y age group homeostatic space in the peripheral blood, barely changing their compared with UCB samples (p = 0.02, Holm correction; Fig. 3F). frequency over several years (Fig. 1). This might be important for This observation suggests that naive T cells from young individ- not only protecting against those infections that the immune system uals comprise a substantial number of virus-specific precursors is already familiar with (49–53, 77), but also for sustaining balanced due to the evolutionary programmed probabilities of recombina- regulation of the existing social network of immune cells (78, 79). tion events, which are further supplanted by expansion of sto- The establishment of this social network starts at a very early age chastically selected primed clones. Interestingly, the average size and is manifested by highly dynamic changes in T cell immunity of viral-specific clonotypes decreased in very old age (Fig. 3F). and TCR repertoire. Indeed, comparative analysis of UCB relative This could result either from the exhaustion of Ag-specific T cell to peripheral blood samples from young children reveals that the 6 DYNAMICS OF INDIVIDUAL T CELL REPERTOIRES

FIGURE 3. Repertoire divergence with age. (A–C) Cluster analysis of repertoire pairs using a distance measure proportional to the sum of frequencies of overlapping clonotypes. (A) Dendrogram constructed using hierarchical clustering. (B) Multidi- mensional scaling of samples. The plot in Downloaded from (C) shows the distance metric computed as the Euclidean distance of a sample in multidimensional scaling coordinates to the centroid coordinates obtained by averaging the x- and y-coordinates of all samples. (D) Convergence of V(D)J recombination, as calculated by the average number of nu- http://www.jimmunol.org/ cleotide CDR3 sequences per amino acid CDR3 sequence. (E) The total number of CMV- and EBV-speciﬁc clonotypes, annotated based on sequences derived from the literature. (F) Average size of a pathogen- speciﬁc clonotype. Sample labels are colored according to age, color bar. by guest on September 27, 2021

native T cell repertoire that we are born with is characterized by a of naive CD4+ T cells of all T cells decreases from ∼50% in UCB number of specific features that disappear in the earliest years of to ∼30% in childhood (Supplemental Fig. 3B). life (Fig. 2). Later in life, involution of thymus, limited proliferative capacity of Cord blood samples are characterized by lower numbers of added naive T cells, and expansion of Ag-experienced clones (1–5) lead to a nucleotides within CDR3, reflecting low TdT activity in the fetal consistent decrease in percentages of T cell homeostasis occupied by period (Fig. 2B). This feature limits the overall theoretical diversity naive T cells (Fig. 2D), and consequently in a decrease of the ob- of TCR variants (80, 81), which was reflected in the relatively high served TCR diversity (Fig. 2C). Furthermore, a decrease of intrinsic numbers of clonotypes shared among UCB samples (Fig. 2E). At the diversity (Ref. 6 and M. Pogorelyy and D. Chudakov, unpublished same time, this limitation did not fundamentally alter the observed observations for several individuals) and smearing of the ini- TCRb diversity (Fig. 2C). This indicates, importantly, that the fetal tially high-frequency clonotypes within the naive T cells pool TCRb repertoire, albeit structurally maintained within regulated (M. Pogorelyy and D. Chudakov, unpublished observations) occur borders, is not actually limited in the number of diverse sequence with aging as a result of nonuniform homeostatic proliferation on the variants that may encounter Ag in terms of defined volume. periphery (70–73). We demonstrate in the present study that the CD4+ T cells dominate over CD8+ T cells, and naive CD4+ dynamics of the TCR diversity decrease differ for males and fe- T cells dominate over naive CD8+ T cells in the cord blood, but both males. By the age of 40, the the difference in TCR diversity becomes ratios tend to equalize in early childhood (Fig. 2F, Supplemental prominent (Fig. 4A). The difference is mainly determined by the Fig. 3A–C), which may partially result from rapid expansion of the more rapid shrinkage of naive T cells pool (Fig. 4B) and might be a newly primed CD8+ T cell clones. Correspondingly, the percentage consequence of gender variability of thymic output (82) and other The Journal of Immunology 7 Downloaded from

FIGURE 4. Age-dependent changes in T cell repertoire in males and females. (A) Observed TCRb diversity per 300,000 T cells. Note gender difference http://www.jimmunol.org/ in the 30–50 y age group. (B) Percentages of naive CD4+ and CD8+ T cells among all T cells for the 30–50 y age group. F, females; M, males. gender-specific factors. Paradoxically, however, men and women in Finally, we have demonstrated that the repertoires of different the oldest age cohort are characterized by similar TCRb diversity individuals are functionally more similar at birth and early and percentages of naive T cells (Fig. 4A, Supplemental Fig. 4). childhood and subsequently diverge during aging (Fig. 3A–D). Additionally, individuals in this age group are characterized by Paraphrasing Leo Tolstoy, all young people’s TCR repertoires are higher percentages of naive T cells within the CD4+ subset com- alike; each old individual TCR repertoire is different (89). At the pared with that within the CD8+ subset (Supplemental Figs. 3D, 4). same time, the observation that individuals are born with repertoires Both of these phenomena could be explained by the age-selection containing a high number of public pathogen-specific clonotypes by guest on September 27, 2021 barrier, which introduces bias to the oldest cohort. In this view, (Fig. 3E) could be indicative of human–pathogen coevolution. balanced function of the adaptive immune system in older age de- One could speculate that we start our lives with roughly similar pends on the availability of sufficient counts of naive CD4+ T cells, as and evolutionary predesigned immune repertoires, which then well as their diversity and domination over the counts of naive CD8+ diverge into a multitude of states as we age. The specific nature T cells. Broadly speaking, potential regulators should dominate over of these divergent states is determined by environmental experi- potential effectors. Given that those individuals whose immune ences throughout life and the stochastic nature of particular im- system fulfills these criteria have higher chances of living beyond mune responses. The latter means that if two hypothetical the age of ∼70 y, this would influence the observed characteristics of individuals with an identical TCR repertoire (never existed, be- the oldest age group and contribute to a leveling of the previously cause even TCR repertoires of monozygotic twins are different) observed gender differences. This model suggests a common ex- (22) encounter exactly the same Ag, their repertoire will still di- planation both for the observation by Ferguson and colleagues (83) verge, as was recently suggested by Walczak and colleagues (90) that an inverted CD4+/CD8+ ratioinagedpeopleisassociatedwith on a theoretical basis. increased 2-y mortality and for that of Pawelec and colleagues (84) Our study has several limitations. In particular, for most indi- that there is, paradoxically, an inverse correlation between the viduals we have not sorted naive and memory, CD4+ and CD8+ overall percentage of naive CD8+ T cells and survival. It is also in populations for TCR repertoire profiling, as should be preferably agreement with the point that the survival advantage of females performed (6), but ideally performed with molecular barcoding (5) versus males declines at very old age (76), when we observe and with oversequencing that allows to eliminate PCR and se- equalization of TCRb diversity and percentages of naive T cells in quencing errors without losing the true diversity of convergent males and females. These results may indicate that relative abun- TCR variants (39). We also have not analyzed TCRa repertoire dance of naive CD4+ T cells in middle age essentially contributes to diversity. Per se, TCRa repertoire profiling would not add much the generally increased longevity observed in women (85, 86). because TCRb repertoire is more diverse. However, combinatorial Unfortunately, we have no information on CMV infection status of diversity of TCR a- and b-chains (91) is important. Paired anal- the individuals in our study, which is known to influence immuno- ysis of TCR a- and b-chain repertoires that should ultimately senescence (87). We could expect that the cohort of oldest individ- become possible (92, 93) will strengthen the studies of adaptive uals who overcame the age-selection barrier is enriched with CMV- immunity dynamics, allowing to differentiate between clonal (a negative individuals, for which the percentage of naive CD4+ and b) versus clonotype (a or b) fates. Future works employing T cells was shown to be more stable with aging (88). This could more targeted strategies to track the long-term fate of T cell ultimately lead to conclusion that, although gender differences de- subsets and Ag-specific clones in a high-throughput manner termine the difference in life expectancy in middle age, in older age should reveal the whole picture of life-long adaptive immunity the factor of CMV status prevails. dynamics in detail. 8 DYNAMICS OF INDIVIDUAL T CELL REPERTOIRES

Acknowledgments 23. Venturi, V., M. F. Quigley, H. Y. Greenaway, P. C. Ng, Z. S. Ende, T. McIntosh, T. E. Asher, J. R. Almeida, S. Levy, D. A. Price, et al. 2011. A mechanism for The work was carried out in part using equipment provided by the TCR sharing between T cell subsets and individuals revealed by pyrosequencing. Shemyakin–Ovchinnikov Institute of Bioorganic Chemistry Core Facility. J. Immunol. 186: 4285–4294. 24. Fo¨hse, L., J. Suffner, K. Suhre, B. Wahl, C. Lindner, C. W. Lee, S. Schmitz, J. D. Haas, S. Lamprecht, C. Koenecke, et al. 2011. High TCR diversity ensures Disclosures optimal function and homeostasis of Foxp3+ regulatory T cells. Eur. J. Immunol. The authors have no financial conflicts of interest. 41: 3101–3113. 25. Estorninho, M., V. B. Gibson, D. Kronenberg-Versteeg, Y. F. Liu, C. Ni, K. Cerosaletti, and M. Peakman. 2013. A novel approach to tracking antigen- experienced CD4 T cells into functional compartments via tandem deep and References shallow TCR clonotyping. J. Immunol. 191: 5430–5440. 1. Messaoudi, I., J. Lemaoult, J. A. Guevara-Patino, B. M. Metzner, and 26. Edholm, E. S., L. M. Albertorio Saez, A. L. Gill, S. R. Gill, L. Grayfer, J. Nikolich-Zugich. 2004. Age-related CD8 T cell clonal expansions constrict N. Haynes, J. R. Myers, and J. Robert. 2013. Nonclassical MHC class I- CD8 T cell repertoire and have the potential to impair immune defense. J. Exp. dependent invariant T cells are evolutionarily conserved and prominent from Med. 200: 1347–1358. early development in amphibians. Proc. Natl. Acad. Sci. USA 110: 14342–14347. 2. Goronzy, J. J., and C. M. Weyand. 2005. T cell development and receptor di- 27. Greenaway, H. Y., B. Ng, D. A. Price, D. C. Douek, M. P. Davenport, and versity during aging. Curr. Opin. Immunol. 17: 468–475. V. Venturi. 2013. NKT and MAIT invariant TCRa sequences can be produced 3. Naylor, K., G. Li, A. N. Vallejo, W. W. Lee, K. Koetz, E. Bryl, J. Witkowski, efficiently by VJ gene recombination. Immunobiology 218: 213–224. J. Fulbright, C. M. Weyand, and J. J. Goronzy. 2005. The influence of age on 28. Lepore, M., A. Kalinichenko, A. Colone, B. Paleja, A. Singhal, A. Tschumi, T cell generation and TCR diversity. J. Immunol. 174: 7446–7452. B. Lee, M. Poidinger, F. Zolezzi, L. Quagliata, et al. 2014. Parallel T-cell cloning 4. Goronzy, J. J., W. W. Lee, and C. M. Weyand. 2007. Aging and T-cell diversity. and deep sequencing of human MAIT cells reveal stable oligoclonal TCRb Exp. Gerontol. 42: 400–406. repertoire. Nat. Commun. 5: 3866. 5. Britanova, O. V., E. V. Putintseva, M. Shugay, E. M. Merzlyak, 29. Linnemann, C., B. Heemskerk, P. Kvistborg, R. J. Kluin, D. A. Bolotin, X. Chen, M. A. Turchaninova, D. B. Staroverov, D. A. Bolotin, S. Lukyanov, K. Bresser, M. Nieuwland, R. Schotte, S. Michels, et al. 2013. High-throughput E. A. Bogdanova, I. Z. Mamedov, et al. 2014. Age-related decrease in TCR identification of antigen-specific TCRs by TCR gene capture. Nat. Med. 19: Downloaded from repertoire diversity measured with deep and normalized sequence profiling. J. 1534–1541. Immunol. 192: 2689–2698. 30. Gros, A., P. F. Robbins, X. Yao, Y. F. Li, S. Turcotte, E. Tran, J. R. Wunderlich, 6. Qi, Q., Y. Liu, Y. Cheng, J. Glanville, D. Zhang, J. Y. Lee, R. A. Olshen, A. Mixon, S. Farid, M. E. Dudley, et al. 2014. PD-1 identifies the patient-specific C. M. Weyand, S. D. Boyd, and J. J. Goronzy. 2014. Diversity and clonal se- CD8+ tumor-reactive repertoire infiltrating human tumors. J. Clin. Invest. 124: lection in the human T-cell repertoire. Proc. Natl. Acad. Sci. USA 111: 13139– 2246–2259. 13144. 31. Zhu, J., T. Peng, C. Johnston, K. Phasouk, A. S. Kask, A. Klock, L. Jin, K. Diem, 7. Yager, E. J., M. Ahmed, K. Lanzer, T. D. Randall, D. L. Woodland, and D. M. Koelle, A. Wald, et al. 2013. Immune surveillance by CD8aa+ skin-

M. A. Blackman. 2008. Age-associated decline in T cell repertoire diversity resident T cells in human herpes virus infection. Nature 497: 494–497. http://www.jimmunol.org/ leads to holes in the repertoire and impaired immunity to influenza virus. J. Exp. 32. Ye, L., J. C. Goodall, L. Zhang, E. V. Putintseva, B. Lam, L. Jiang, W. Liu, Med. 205: 711–723. J. Yin, L. Lin, T. Li, et al. 2016. TCR usage, gene expression and function of two 8. Arnold, C. R., J. Wolf, S. Brunner, D. Herndler-Brandstetter, and B. Grubeck- distinct FOXP3+Treg subsets within CD4+CD25hi T cells identified by expres- Loebenstein. 2011. Gain and loss of T cell subsets in old age—age-related sion of CD39 and CD45RO. Immunol. Cell Biol. 94: 293–305. reshaping of the T cell repertoire. J. Clin. Immunol. 31: 137–146. 33. Feng, Y., J. van der Veeken, M. Shugay, E. V. Putintseva, H. U. Osmanbeyoglu, 9. Buchholz, V. R., M. Neuenhahn, and D. H. Busch. 2011. CD8+ T cell differ- S. Dikiy, B. E. Hoyos, B. Moltedo, S. Hemmers, P. Treuting, et al. 2015. A entiation in the aging immune system: until the last clone standing. Curr. Opin. mechanism for expansion of regulatory T-cell repertoire and its role in self- Immunol. 23: 549–554. tolerance. Nature 528: 132–136. 10. O’Hara, G. A., S. P. Welten, P. Klenerman, and R. Arens. 2012. Memory T cell 34. Nguyen, P., J. Ma, D. Pei, C. Obert, C. Cheng, and T. L. Geiger. 2011. Identi- inflation: understanding cause and effect. Trends Immunol. 33: 84–90. fication of errors introduced during high throughput sequencing of the T cell 11. Goronzy, J. J., and C. M. Weyand. 2013. Understanding immunosenescence to receptor repertoire. BMC Genomics 12: 106. improve responses to vaccines. Nat. Immunol. 14: 428–436. 35. Bolotin, D. A., I. Z. Mamedov, O. V. Britanova, I. V. Zvyagin, D. Shagin, by guest on September 27, 2021 12. Nikolich-Zugich, J. 2014. Aging of the T cell compartment in mice and humans: S. V. Ustyugova, M. A. Turchaninova, S. Lukyanov, Y. B. Lebedev, and from no naive expectations to foggy memories. J. Immunol. 193: 2622–2629. D. M. Chudakov. 2012. Next generation sequencing for TCR repertoire profiling: 13. Robins, H. S., P. V. Campregher, S. K. Srivastava, A. Wacher, C. J. Turtle, platform-specific features and correction algorithms. Eur. J. Immunol. 42: 3073– O. Kahsai, S. R. Riddell, E. H. Warren, and C. S. Carlson. 2009. Comprehensive 3083. assessment of T-cell receptor b-chain diversity in ab T cells. Blood 114: 4099– 36. Baum, P. D., V. Venturi, and D. A. Price. 2012. Wrestling with the repertoire: the 4107. promise and perils of next generation sequencing for antigen receptors. Eur. J. 14. Freeman, J. D., R. L. Warren, J. R. Webb, B. H. Nelson, and R. A. Holt. 2009. Immunol. 42: 2834–2839. Profiling the T-cell receptor b-chain repertoire by massively parallel sequencing. 37. Bolotin, D. A., M. Shugay, I. Z. Mamedov, E. V. Putintseva, M. A. Turchaninova, Genome Res. 19: 1817–1824. I. V. Zvyagin, O. V. Britanova, and D. M. Chudakov. 2013. MiTCR: software for 15. Mamedov, I. Z., O. V. Britanova, D. A. Bolotin, A. V. Chkalina, T-cell receptor sequencing data analysis. Nat. Methods 10: 813–814. D. B. Staroverov, I. V. Zvyagin, A. A. Kotlobay, M. A. Turchaninova, 38. Mamedov, I. Z., O. V. Britanova, I. V. Zvyagin, M. A. Turchaninova, D. A. Fedorenko, A. A. Novik, et al. 2011. Quantitative tracking of T cell clones D. A. Bolotin, E. V. Putintseva, Y. B. Lebedev, and D. M. Chudakov. 2013. after haematopoietic stem cell transplantation. EMBO Mol. Med. 3: 201–207. Preparing unbiased T-cell receptor and antibody cDNA libraries for the deep 16. Warren, R. L., J. D. Freeman, T. Zeng, G. Choe, S. Munro, R. Moore, next generation sequencing profiling. Front. Immunol. 4: 456. J. R. Webb, and R. A. Holt. 2011. Exhaustive T-cell repertoire sequencing of 39. Shugay, M., O. V. Britanova, E. M. Merzlyak, M. A. Turchaninova, human peripheral blood samples reveals signatures of antigen selection and a I. Z. Mamedov, T. R. Tuganbaev, D. A. Bolotin, D. B. Staroverov, directly measured repertoire size of at least 1 million clonotypes. Genome Res. E. V. Putintseva, K. Plevova, et al. 2014. Towards error-free profiling of immune 21: 790–797. repertoires. Nat. Methods 11: 653–655. 17. Robins, H. S., S. K. Srivastava, P. V. Campregher, C. J. Turtle, J. Andriesen, 40. Venturi, V., D. A. Price, D. C. Douek, and M. P. Davenport. 2008. The molecular S. R. Riddell, C. S. Carlson, and E. H. Warren. 2010. Overlap and effective size basis for public T-cell responses? Nat. Rev. Immunol. 8: 231–238. of the human CD8+ T cell receptor repertoire. Sci. Transl. Med. 2: 47ra64. 41. Shugay, M., S. Lukyanov, and D. M. Chudakov. 2015. Sequencing rare T-cell 18. Quigley, M. F., H. Y. Greenaway, V. Venturi, R. Lindsay, K. M. Quinn, populations. Oncotarget 6: 39393–39394. R. A. Seder, D. C. Douek, M. P. Davenport, and D. A. Price. 2010. Convergent 42. Shugay, M., D. V. Bagaev, M. A. Turchaninova, D. A. Bolotin, O. V. Britanova, recombination shapes the clonotypic landscape of the naive T-cell repertoire. E. V. Putintseva, M. V. Pogorelyy, V. I. Nazarov, I. V. Zvyagin, V. I. Kirgizova, Proc. Natl. Acad. Sci. USA 107: 19414–19419. et al. 2015. VDJtools: unifying post-analysis of T cell receptor repertoires. PLOS 19. Li, H., C. Ye, G. Ji, X. Wu, Z. Xiang, Y. Li, Y. Cao, X. Liu, D. C. Douek, Comput. Biol. 11: e1004503. D. A. Price, and J. Han. 2012. Recombinatorial biases and convergent recom- 43. Kivioja, T., A. Va¨ha¨rautio, K. Karlsson, M. Bonke, M. Enge, S. Linnarsson, and bination determine interindividual TCRb sharing in murine thymocytes. J. J. Taipale. 2011. Counting absolute numbers of molecules using unique mo- Immunol. 189: 2404–2413. lecular identifiers. Nat. Methods 9: 72–74. 20. Shugay, M., D. A. Bolotin, E. V. Putintseva, M. V. Pogorelyy, I. Z. Mamedov, 44. Kinde, I., J. Wu, N. Papadopoulos, K. W. Kinzler, and B. Vogelstein. 2011. and D. M. Chudakov. 2013. Huge overlap of individual TCR beta repertoires. Detection and quantification of rare mutations with massively parallel se- Front. Immunol. 4: 466. quencing. Proc. Natl. Acad. Sci. USA 108: 9530–9535. 21. Putintseva, E. V., O. V. Britanova, D. B. Staroverov, E. M. Merzlyak, 45. Vollmers, C., R. V. Sit, J. A. Weinstein, C. L. Dekker, and S. R. Quake. 2013. M. A. Turchaninova, M. Shugay, D. A. Bolotin, M. V. Pogorelyy, I. Z. Mamedov, Genetic measurement of memory B-cell recall using antibody repertoire se- V. Bobrynina, et al. 2013. Mother and child T cell receptor repertoires: deep quencing. Proc. Natl. Acad. Sci. USA 110: 13463–13468. profiling study. Front. Immunol. 4: 463. 46. Vander Heiden, J. A., G. Yaari, M. Uduman, J. N. Stern, K. C. O’Connor, 22. Zvyagin, I. V., M. V. Pogorelyy, M. E. Ivanova, E. A. Komech, M. Shugay, D. A. Hafler, F. Vigneault, and S. H. Kleinstein. 2014. pRESTO: a toolkit for D. A. Bolotin, A. A. Shelenkov, A. A. Kurnosov, D. B. Staroverov, processing high-throughput sequencing raw reads of lymphocyte receptor rep- D. M. Chudakov, et al. 2014. Distinctive properties of identical twins’ TCR ertoires. Bioinformatics 30: 1930–1932. repertoires revealed by high-throughput sequencing. Proc. Natl. Acad. Sci. USA 47. Egorov, E. S., E. M. Merzlyak, A. A. Shelenkov, O. V. Britanova, 111: 5980–5985. G. V. Sharonov, D. B. Staroverov, D. A. Bolotin, A. N. Davydov, E. Barsova, The Journal of Immunology 9

Y. B. Lebedev, et al. 2015. Quantitative profiling of immune repertoires for 2008. TCR b-chain sharing in human CD8+ T cell responses to cytomegalovirus minor lymphocyte counts using unique molecular identifiers. J. Immunol. 194: and EBV. J. Immunol. 181: 7853–7862. 6155–6163. 70. Takada, K., and S. C. Jameson. 2009. Naive T cell homeostasis: from awareness 48. Best, K., T. Oakes, J. M. Heather, J. Shawe-Taylor, and B. Chain. 2015. Com- of space to a sense of place. Nat. Rev. Immunol. 9: 823–832. putational analysis of stochastic heterogeneity in PCR amplification efficiency 71. Takeda, S., H. R. Rodewald, H. Arakawa, H. Bluethmann, and T. Shimizu. 1996. revealed by single molecule barcoding. Sci. Rep. 5: 14629. MHC class II molecules are not required for survival of newly generated CD4+ 49. Day, E. K., A. J. Carmichael, I. J. ten Berge, E. C. Waller, J. G. Sissons, and T cells, but affect their long-term life span. Immunity 5: 217–228. M. R. Wills. 2007. Rapid CD8+ T cell repertoire focusing and selection of high- 72. Brocker, T. 1997. Survival of mature CD4 T lymphocytes is dependent on major affinity clones into memory following primary infection with a persistent human histocompatibility complex class II-expressing dendritic cells. J. Exp. Med. 186: virus: human cytomegalovirus. J. Immunol. 179: 3203–3213. 1223–1232. 50. Iancu, E. M., P. Corthesy, P. Baumgaertner, E. Devevre, V. Voelter, P. Romero, 73. Rudd, B. D., V. Venturi, G. Li, P. Samadder, J. M. Ertelt, S. S. Way, D. E. Speiser, and N. Rufer. 2009. Clonotype selection and composition of M. P. Davenport, and J. Nikolich-Zugich. 2011. Nonrandom attrition of the naive + human CD8 T cells specific for persistent herpes viruses varies with differen- CD8 T-cell pool with aging governed by T-cell receptor:pMHC interactions. tiation but is stable over time. J. Immunol. 183: 319–331. Proc. Natl. Acad. Sci. USA 108: 13694–13699. 51. Maini, M. K., N. Gudgeon, L. R. Wedderburn, A. B. Rickinson, and 74. Solana, R., R. Tarazona, A. E. Aiello, A. N. Akbar, V. Appay, M. Beswick, P. C. Beverley. 2000. Clonal expansions in acute EBV infection are detectable in J. A. Bosch, C. Campos, S. Cantisan, L. Cicin-Sain, et al. 2012. CMV and the CD8 and not the CD4 subset and persist with a variable CD45 phenotype. J. immunosenescence: from basics to clinics. Immun. Ageing 9: 23. € Immunol. 165: 5729–5737. 75. Fulo¨p, T., A. Larbi, and G. Pawelec. 2013. Human T cell aging and the impact of 52. Weekes, M. P., M. R. Wills, J. G. Sissons, and A. J. Carmichael. 2004. Long- persistent viral infections. Front. Immunol. 4: 271. term stable expanded human CD4+ T cell clones specific for human cytomeg- 76. Jacobs, J. M., A. Cohen, E. Ein-Mor, and J. Stessman. 2014. Gender differences alovirus are distributed in both CD45RAhigh and CD45ROhigh populations. J. in survival in old age. Rejuvenation Res. 17: 499–506. 77. Clambey, E. T., L. F. van Dyk, J. W. Kappler, and P. Marrack. 2005. Non- Immunol. 173: 5843–5851. + 53. Klarenbeek, P. L., E. B. Remmerswaal, I. J. ten Berge, M. E. Doorenspleet, malignant clonal expansions of CD8 memory T cells in aged individuals. B. D. van Schaik, R. E. Esveldt, S. D. Koch, A. ten Brinke, A. H. van Kampen, Immunol. Rev. 205: 170–189. F. J. Bemelman, et al. 2012. Deep sequencing of antiviral T-cell responses to 78. Bradley, L. M., J. Harbertson, G. C. Freschi, R. Kondrack, and P. J. Linton. 2000. Downloaded from HCMV and EBV in humans reveals a stable repertoire that is maintained for Regulation of development and function of memory CD4 subsets. Immunol. Res. many years. PLoS Pathog. 8: e1002889. 21: 149–158. 54. Kedzierska, K., S. A. Valkenburg, P. C. Doherty, M. P. Davenport, and 79. Dittel, B. N. 2008. CD4 T cells: balancing the coming and going of autoimmune- V. Venturi. 2012. Use it or lose it: establishment and persistence of T cell mediated inflammation in the CNS. Brain Behav. Immun. 22: 421–430. 80. Cabaniols, J. P., N. Fazilleau, A. Casrouge, P. Kourilsky, and memory. Front. Immunol. 3: 357. ab 55. Carter, M. S., J. Doskow, P. Morris, S. Li, R. P. Nhim, S. Sandstedt, and J. M. Kanellopoulos. 2001. Most T cell receptor diversity is due to terminal J. Exp. Med. M. F. Wilkinson. 1995. A regulatory mechanism that detects premature nonsense deoxynucleotidyl transferase. 194: 1385–1390. 81. Zemlin, M., R. L. Schelonka, K. Bauer, and H. W. Schroeder, Jr. 2002. Regu- codons in T-cell receptor transcripts in vivo is reversed by protein synthesis

lation and chance in the ontogeny of B and T cell antigen receptor repertoires. http://www.jimmunol.org/ inhibitors in vitro. J. Biol. Chem. 270: 28995–29003. Immunol. Res. 26: 265–278. 56. Wang, J., V. M. Vock, S. Li, O. R. Olivas, and M. F. Wilkinson. 2002. A quality 82. Pido-Lopez, J., N. Imami, and R. Aspinall. 2001. Both age and gender affect control pathway that down-regulates aberrant T-cell receptor (TCR) transcripts thymic output: more recent thymic migrants in females than males as they age. by a mechanism requiring UPF2 and translation. J. Biol. Chem. 277: 18489– Clin. Exp. Immunol. 125: 409–413. 18493. 83. Olsson, J., A. Wikby, B. Johansson, S. Lo¨fgren, B. O. Nilsson, and 57. Bhalla, A. D., J. P. Gudikote, J. Wang, W. K. Chan, Y. F. Chang, O. R. Olivas, F. G. Ferguson. 2000. Age-related change in peripheral blood T-lymphocyte and M. F. Wilkinson. 2009. Nonsense codons trigger an RNA partitioning shift. subpopulations and cytomegalovirus infection in the very old: the Swedish J. Biol. Chem. 284: 4062–4072. longitudinal OCTO immune study. Mech. Ageing Dev. 121: 187–201. 58. Gardner, L. B. 2008. Hypoxic inhibition of nonsense-mediated RNA decay 84. Derhovanessian, E., A. B. Maier, K. Ha¨hnel, H. Zelba, A. J. de Craen, regulates gene expression and the integrated stress response. Mol. Cell. Biol. 28: H. Roelofs, E. P. Slagboom, R. G. Westendorp, and G. Pawelec. 2013. Lower 3729–3741. proportion of naı¨ve peripheral CD8+ T cells and an unopposed pro-inﬂammatory 59. Soothill, P. W., K. H. Nicolaides, C. H. Rodeck, and H. Gamsu. 1986. Blood

response to human cytomegalovirus proteins in vitro are associated with longer by guest on September 27, 2021 gases and acid-base status of the human second-trimester fetus. Obstet. Gynecol. survival in very elderly people. Age (Dordr.) 35: 1387–1399. 68: 173–176. 85. Nathanson, C. A. 1984. Sex differences in mortality. Annu. Rev. Sociol. 10: 191– 60. Bhuvanagiri, M., A. M. Schlitter, M. W. Hentze, and A. E. Kulozik. 2010. NMD: 213. RNA biology meets human genetic medicine. Biochem. J. 430: 365–377. 86. Reade, M. C., S. Yende, G. D’Angelo, L. Kong, J. A. Kellum, A. E. Barnato, 61. Bruno, I. G., R. Karam, L. Huang, A. Bhardwaj, C. H. Lou, E. Y. Shum, E. B. Milbrandt, C. Dooley, F. B. Mayr, L. Weissfeld, and D. C. Angus, Genetic H. W. Song, M. A. Corbett, W. D. Gifford, J. Gecz, et al. 2011. Identification of a and Inflammatory Markers of Sepsis Investigators. 2009. Differences in immune microRNA that activates gene expression by repressing nonsense-mediated RNA response may explain lower survival among older men with pneumonia. Crit. decay. Mol. Cell 42: 500–510. Care Med. 37: 1655–1662. 62. Lykke-Andersen, S., and T. H. Jensen. 2015. Nonsense-mediated mRNA decay: 87. Sansoni, P., R. Vescovini, F. F. Fagnoni, A. Akbar, R. Arens, Y. L. Chiu, an intricate machinery that shapes transcriptomes. Nat. Rev. Mol. Cell Biol. 16: L. Cicin-Sain, J. Dechanet-Merville, E. Derhovanessian, S. Ferrando-Martinez, 665–677. et al. 2014. New advances in CMV and immunosenescence. Exp. Gerontol. 55: 63. Feeney, A. J. 1990. Lack of N regions in fetal and neonatal mouse immuno- 54–62. globulin V-D-J junctional sequences. J. Exp. Med. 172: 1377–1390. 88. Wertheimer, A. M., M. S. Bennett, B. Park, J. L. Uhrlaub, C. Martinez, V. Pulko, 64. Bogue, M., S. Gilfillan, C. Benoist, and D. Mathis. 1992. Regulation of N-region N. L. Currier, D. Nikolich-Zugich, J. Kaye, and J. Nikolich-Zugich. 2014. Aging diversity in antigen receptors through thymocyte differentiation and thymus and cytomegalovirus infection differentially and jointly affect distinct circulating ontogeny. Proc. Natl. Acad. Sci. USA 89: 11011–11015. T cell subsets in humans. J. Immunol. 192: 2143–2155. 65. George, J. F., Jr., and H. W. Schroeder, Jr. 1992. Developmental regulation of D 89. Tolstoy, L. 2014. Anna Karenina. Random House, London. beta reading frame and junctional diversity in T cell receptor-beta transcripts 90. Mayer, A., V. Balasubramanian, T. Mora, and A. M. Walczak. 2015. How a well- from human thymus. J. Immunol. 148: 1230–1239. adapted immune system is organized. Proc. Natl Acad. Sci. USA 112: 5950– 66. Cherrier, M., A. Cardona, I. Rosinski-Chupin, F. Rougeon, and N. Doyen. 2002. 5955. Substantial N diversity is generated in T cell receptor alpha genes at birth despite 91. Arstila, T. P., A. Casrouge, V. Baron, J. Even, J. Kanellopoulos, and P. Kourilsky. low levels of terminal deoxynucleotidyl transferase expression in mouse thymus. 1999. A direct estimate of the human ab T cell receptor diversity. Science 286: Eur. J. Immunol. 32: 3651–3656. 958–961. 67. Elhanati, Y., A. Murugan, C. G. Callan, Jr., T. Mora, and A. M. Walczak. 2014. 92. Turchaninova, M. A., O. V. Britanova, D. A. Bolotin, M. Shugay, Quantifying selection in immune receptor repertoires. Proc. Natl. Acad. Sci. USA E. V. Putintseva, D. B. Staroverov, G. Sharonov, D. Shcherbo, I. V. Zvyagin, 111: 9875–9880. I. Z. Mamedov, et al. 2013. Pairing of T-cell receptor chains via emulsion PCR. 68. Yan, J., J. M. Greer, R. Hull, J. D. O’Sullivan, R. D. Henderson, S. J. Read, and Eur. J. Immunol. 43: 2507–2515. P. A. McCombe. 2010. The effect of ageing on human lymphocyte subsets: com- 93. Howie, B., A. M. Sherwood, A. D. Berkebile, J. Berka, R. O. Emerson, parison of males and females. Immun. Ageing 7: 4. doi:10.1186/1742-4933-7-4 D. W. Williamson, I. Kirsch, M. Vignali, M. J. Rieder, C. S. Carlson, and 69. Venturi, V., H. Y. Chin, T. E. Asher, K. Ladell, P. Scheinberg, E. Bornstein, H. S. Robins. 2015. High-throughput pairing of T cell receptor a and b se- D. van Bockel, A. D. Kelleher, D. C. Douek, D. A. Price, and M. P. Davenport. quences. Sci. Transl. Med. 7: 301ra131.