of April 15, 2014.
This information is current as Normalized Sequence Profiling
Diversity Measured with Deep and Age-Related Decrea in TCR Repertoire Dmitriy M. Chudakov
A. Bogdanova, Ilgar Z. Mamedov, Yuriy
B. Lebedev and Staroverov, Dmitriy A. Bolotin, Sergey Lukyanov, Ekaterina Ekaterina M. Merzlyak, Maria A. Turchaninova, Dmitriy B. Olga V. Britanova, Ekaterina V. Putintva, Mikhail Shugay,/content/192/6/2689
doi: 10.4049/jimmunol.1302064
英语口试考什么February 2014;
2014; 192:2689-2698; Prepublished online 7J Immunol Material
Supplementary
4.DCSupplemental.html
/content/suppl/2014/02/07/jimmunol.130206References /content/192/6/2689.full#ref-list-1
, 26 of which you can access for free at: cites 58 articles This article Subscriptions /subscriptions is online at: The Journal of Immunology Information about subscribing to Permissions /ji/copyright.html
Submit copyright permission requests at: Email Alerts
/cgi/alerts/etoc
Receive free email-alerts when new articles cite this article. Sign up at: Print ISSN: 0022-1767 Online ISSN: 1550-6606.
Immunologists, Inc. All rights rerved.
Copyright © 2014 by The American Association of 9650 Rockville Pike, Bethesda, MD 20814-3994.The American Association of Immunologists, Inc., is published twice each month by
The Journal of Immunology at Merck & Co on April 15, 2014
/Downloaded from
The Journal of Immunology Age-Related Decrea in TCR Repertoire Diversity Measured with Deep and Normalized Sequence Profiling
Olga V.Britanova,*,1Ekaterina V.Putintva,*,1Mikhail Shugay,*,†,1
Ekaterina M.Merzlyak,*Maria A.Turchaninova,*Dmitriy B.Staroverov,*
Dmitriy A.Bolotin,*Sergey Lukyanov,*,†Ekaterina A.Bogdanova,*Ilgar Z.Mamedov,* Yuriy B.Lebedev,*and Dmitriy M.Chudakov*,‡
The decrea of TCR diversity with aging has never been studied by direct methods.In this study,we combined high-throughput Illumina quencing with unique cDNA molecular identifier technology to achieve deep and precily normalized profiling of TCR b repertoires in39healthy donors aged6–90y.We demonstrate that TCR b diversity per106T cells decreas roughly linearly with age,with significant reduction already apparent by age40.The percentage of naive T cells showed a strong correlation with measured TCR diversity and decread linearly up to age70.Remarkably,the oldest group(average age82y)was characterized by a higher percentage of naive CD4+T cells,lower abundance of expanded clones,and incread TCR diversity compared with the previous age group(average age62y),suggesting the influence of age lection and association of the three rela
ted parameters with longevity.Interestingly,cross-analysis of individual TCR b repertoires revealed a t>10,000of the most reprentative public TCR b clonotypes,who abundance among the top100,000clones correlated with TCR diversity and decread with aging.The Journal of Immunology,2014,192:2689–2698.
I n humans,aging is associated with prominent decline of the
adaptive immune respon,caud by both altered function-ality of aged T cells(1,2)and overall decrea in naive T cell abundance and TCR diversity.The latter results from thymus in-volution(3,4);stochasticity,lectivity,and exhaustion of periph-eral proliferation of naive T cells(5–9);and expansion of effector memory clones that take over the homeostatic space(6,10–13). It is well acknowledged that high TCR repertoire diversity is a necessary prerequisite for an effective adaptive immune respon against new Ags(6).Loss of naive T cells and constriction of the TCR repertoire result in impaired immunity to viral and bacterial infections(6,14–16),poor respon to vaccination(17,18),poor immune system recovery after chemotherapy and hematopoietic stem cell transplantation(19–21),and poor control of autoreactive T clones and autoimmunity(22–24).
The phenomenon of aging-associated TCR repertoire constric-tion,although widely accepted,has ne
ver been studied by direct methods.Oligonucleotide hybridization assays(25)and spectra-typing analysis(26)have suggested that early and midadulthood are not associated with significant contraction of the TCR reper-toire and that this occurs only after the age of75(25,27).
In this study,we have ud an advanced deep TCR b quencing approach coupled with a sample normalization strategy that allowed us to directly quantify and compare T cell repertoire diversity across samples obtained from numerous individuals of different ages.We achieved deep and accurate profiling of individual TCR b CDR3 repertoires by performing unique barcoding of cDNA molecules (28),followed by paired-end Illumina quencing and rational bioinformatic analysis of the output data for preci sample nor-malization(29,30).This approach enabled us to track age-related changes in the human TCR repertoire with unprecedented accu-racy,revealing key parameters of adaptive immunity aging and pro-viding high-quality reference data for future studies of adaptive im-munity in health and dia.
Materials and Methods
Sample collection
This study was approved by the local ethics committee and conducted in accordance with the Declar
ation of Helsinki.All donors were informed of thefinal u of their blood and signed an informed connt.Ten milliliters of peripheral blood was obtained from39systemically healthy Caucasian donors from Russia,aged6–90y old(Table I).Peripheral blood was collected into EDTA-treated Vacutainer tubes(BD Biosciences,Franklin Lakes,NJ).PBMCs(at least73106/sample)were isolated by Ficoll-Paque(Paneco,Russia)density-gradient centrifugation.Total RNA(at least6m g/sample)was isolated using Trizol reagent(Invitrogen,Carlsbad, CA),according to the manufacturer’s protocol.
cDNA synthesis
First-strand cDNA was synthesized using the Mint cDNA synthesis kit (Evrogen,Russia),according to the manufacturer’s protocol(Supplemental Fig.1).RNA(1.5m g)was collected per15-m l reaction volume,with a total of four to six tubes per sample,such that all RNA was ud for each
*Shemiakin-Ovchinnikov Institute of Bioorganic Chemistry,Russian Academy of Science,117997Moscow,Russia;†Pirogov Russian National Rearch Medical University,117997Moscow,Russia;and‡Central European Institute of Technology, Masaryk University,60177Brno,Czech Republic
1O.V.B.,E.V.P.,and M.S.are jointfirst authors who contributed equally to this work. Received for public
ation August6,2013.Accepted for publication December27, 2013.
This work was supported by the Molecular and Cell Biology Program from the Russian Academy of Science,Russian Foundation for Basic Rearch Grants12-04-33139,12-04-00229(to D.M.C.),and14-04-01247(to E.M.M.),and European Regional Devel-opment Fund CZ.1.05/1.1.00/02.0068.
O.V.B.,E.V.P.,M.A.T.,D.B.S.,E.M.M.,E.A.B.,and I.Z.M.isolated PBMC,gen-erated TCR b libraries,and performedflow cytometry analysis;M.S.and D.A.B. performed data analysis;S.L.and Y.supervid the project;and D.M.C.su-pervid the project,designed,and interpreted all experiments and wrote the paper. Address correspondence and reprint requests to Dr.Dmitriy M.Chudakov,Shemiakin-Ovchinnikov Institute of Bioorganic Chemistry,Russian Academy of Science,Miklukho-Maklaya16/10,117997Moscow,Russia.E-mail address:ChudakovDM@mail.ru
The online version of this article contains supplemental material.
Abbreviation ud in this article:repPC,reprentative public clonotype. CopyrightÓ2014by The American Association of Immunologists,Inc.0022-1767/14/$16.00 /cgi/doi/10.4049/jimmunol.1302064 at Merck & Co on April 15, 2014 www.jim
blood sample.Synthesis was primed with the BC_R4_short oligonucleo-tide (59-GTATCTGGAGTCATTGA-39),which is specific to both gene variants of the human TRBC gment.To denature RNA and anneal the priming oligonucleotide,the RNA was incubated at 70˚C for 2min and then at 42˚C for 2min.The SmartNNNNa 59-adaptor (59-AAGCAGUG-GTAUCAACGCAGAGUNNNNUNNNNUNNNNUCTTrGrGrGrG-39),which carries a molecular identifier (12random “N”nucleotides)and dU nucleotides (U),was added for the template switch.This reaction was carried out at 42˚C for 130min,with 5m l IP solution added after the first 40min.Products of cDNA synthesis were treated with fresh uracil-DNA glycosyla (New England Biolabs,Ipswich,MA)to degrade the SmartNNNNa adaptor (5U/15m l reaction)at 37˚C for 30min.To capture the maximum number of input cDNA molecules,the whole volume of the treated cDNA synthesis reaction was ud for the first PCR.
First PCRwith的用法
The first PCR amplification was performed using Encyclo PCR mix (Evrogen)with a pair of universal primers:M1SS (59-AAGCAGTGGT-ATCAACGCA-39)and BC2_uni_R (59-TGCTTCTGATGGCTCAA
AC-AC-39).The PCR mixture contained 13Encyclo polymera buffer,0.125mM of each 29-deoxynucleoside 59-triphosphate,10pmol of each primer,1m l Encyclo polymera mix,and 3m l of undiluted first-strand cDNA/50m l reaction volume.The reaction was performed on an ABI 9700Thermal Cycler with a gold block in multiple 0.2ml PCR tubes,such that all cDNA was ud (up to 32tubes/sample),for 18cycles with the fol-lowing temperature regimen:94˚C for 10s,62˚C for 20s,and 72˚C for 30s.
Second PCR
Sample products from the first PCR amplification were combined into one tube and mixed.A 100-m l aliquot of undiluted PCR product was purified by QIAquick PCR purification kit (Qiagen)and eluted in 20m l TE buffer.For the cond PCR amplification,we ud the M1S primer ((N)2–4(XXXXX )CAGTGGTATCAACGCAGAG;XXXXX is a sample barcode,and (N)2–4are random nucleotides that generate diversity for better cluster identifi-cation on Illumina quencer)on the 59-end of the library and J-b primer mix (31)on the 39-end.1m l of purified product from the first PCR was ud per 50m l PCR,with four tubes for each sample,for 7–11cycles with the following temperature regimen:94˚C for 10s,62˚C for 20s,and 72˚C for 30s.
Sequencing
The PCR product concentration in each library was determined using a QuBit fluorometer (Invitrogen).PCR products of 9–20samples from different donors were mixed together in an equal ratio.To avoid experi-mental bias,donors of different age were randomly distributed across the three lanes.Illumina adapters were ligated according to the manufacturer’s protocol.The libraries were analyzed using three Illumina HiSEquation 2000lanes,with 100+100-nt paired-end quencing with Illumina primers.
Raw quencing data analysis
Preliminary analysis showed that at least 33106CDR3-containing paired-end quencing reads were obtained for each sample.From tho,we randomly lected 13106high-quality quencing reads so that only one read was prent with a given unique molecular identifier.No binning of CDR3quences with identical molecular identifiers was performed,and reads with duplicate molecular identifiers were filtered out.Thus,for each donor we obtained quences of 13106unique TCR b cDNA molecules.To check for possible effects of this random lection,we performed a repetitive sampling of molecular identifiers that yielded no change in our overall results,including sample diversity (data not shown).Further analysis (final CDR3identification,clonotype clusterization,and correc-tion for rever transcription,PCR,and quencing errors)was performed as
previously described (29)using our MiTCR software (30)(/).The threshold on quencing quality for each nucle-otide within the CDR3region was t as .Q25(phred).To eliminate a maximal number of accumulated errors,we employed the strictest “eliminate the errors”correction algorithm,which eliminates 98%of artificial CDR3clonotypes,including variants that ari from rare errors in minor clonotypes.At the same time,this algorithm los ,2%of input TCR b diversity,according to control experiments and in silico modeling.
Lower-bound estimate of total TCR b CDR3diversity in blood
go west
The total number of unique TCR b clonotypes was estimated as in Ref 32,using a nonparametric unen species model as described previously (33).Briefly,the Poisson process assumption,which relates the expected
number of times a species is going to appear in the sample (x s )and the sampling depth (t ),leads to an estimator for total number of species in the form of an infinite ries:where n x is the number of species found exactly x times.Unfortunately,the ries begin to verely oscillate with in-creasing sampling depth,which renders it impossible to estimate total diversity.This problem can be dealt with
月夜忆舍弟using Euler transformation,which forces ries to converge rapidly.We therefore calculated the first x 0terms in modified ries,where x 0is the largest number that yields a co-efficient of variation for estimating S(t)that is ,0.1.This provides a tradeoff between bias and variance,as suggested by Efron et al.(33).
Abs and flow cytometry
An aliquot of PBMC from each sample was stained with Abs for flow cytometry analysis.The following anti-human Abs were ud:CD3-PC7(clone UCHT1;eBioscience),CD27-PC5(clone 1A4CD27;Invitrogen),CD4-PE (clone 13B8.2;Beckman Coulter),CD45RA FITC (clone JS-B3;eBioscience).Cells were incubated with Abs for 20min at room tem-perature and washed twice with PBS.Flow cytometry was performed with the Cytomics FC 500(Beckman Coulter),with data analysis carried out using the Cytomics RXP Analysis program (Beckman Coulter).
Results
Normalization of samples using unique molecular identifiers Blood samples from different donors—even quential blood samples from the same donor—may contain different numbers of T cells.Furthermore,none of the TCR gene library preparation and -quencing procedures can be ma
de absolutely uniform,regardless of whether genomic DNA-bad (32,34)or cDNA-bad (29,31,35)technique is ud.Therefore,one can never be sure of the quantity of T cells that are efficiently covered by the output of quencing analysis of large TCR libraries.This makes it impos-sible to perform accurate and reproducible deep quencing-bad comparison of relative TCR diversity in blood samples obtained from different donors or at different times.
To overcome this basic limitation,we combined a cDNA-bad protocol for preparing quantitative TCR b libraries (31,35)with a unique molecular barcoding technique similar to that reported in Ref 28.In this technique,the template-switching effect (36,37)is ud to introduce a 59-adaptor that carries 12random nucleotides.As a result,each synthesized cDNA molecule is specifically la-beled with one of 412(.16.7million)unique identifier variants.Such molecular identifiers allow robust estimation of the number of cDNA templates in a deeply quenced library,which cannot be accurately deduced from read data alone (38).The whole TCR b library was further amplified using universal primers (e Materials and Methods and Refs.29and 31)and analyzed using paired-end Illumina quencing,which provided information about both the TCR b CDR3quence and the unique molecular identifier of the starting cDNA molecule (Supplemental Fig.1).
This method allows normalization of any two or more libraries,even if they were obtained from differe
nt numbers of T cells and with different quencing coverage,by analyzing equal numbers of unique cDNA molecules labeled with different molecular identi-fiers.Becau each read reprents a distinct cDNA molecule,such normalization eliminates systematic library preparation bias,re-sulting in a prominent reduction of variance between samples and a notable increa in detectable TCR b diversity per fixed number of analyzed quencing reads (Fig.1).This normalization also reduces bias within the sample;becau of the stochastic and biad nature of PCR and quencing,cDNA variants with the same abundance may have quencing read counts that differ by veralfold.Sampling equal numbers of reads will therefore lead to some variants being overreprented and some lost,especially the low-abundance variants that primarily contribute to sample diversity.In contrast,extraction of quences with unique cDNA identifiers allows uniform sampling and accurate comparison of the samples.
2690
MEASURING AGE-RELATED DECREASE IN TCR DIVERSITY
贫富差距英文at Merck & Co on April 15, 2014
/Downloaded from
Model experiments demonstrated that the average efficiency of the template-switch technique that we ud is about one molecular event per two T cells.For example,deep quencing analysis of libraries prepared from ∼23104or 13106T cells produced output data comprising ∼13104or 53105uniquely barcoded variants,respectively.Therefore,each analyzed TCR b cDNA labeled with a unique molecular identifier is equivalent to a single randomly captured T cell,and 1million such molecular events is equivalent to 1million T cells randomly captured from a 10ml peripheral blood sample.
This powerful tool for normalizing analyzed quencing datats enabled us to perform deep and accurate comparison of TCR b CDR3diversity between samples with sufficient but not neces-sarily equal numbers of T cells or levels of quencing coverage.Furthermore,this approach minimized the impact of individual donor blood characteristics or unavoidable bottlenecks and bias during blood sampling,library preparation,and quencing.Obrved TCR b CDR3diversity declines throughout life Our 39systemically healthy human donors were divided into four age groups reprenting young (average 16y),middle-aged (av-erage 39y),aged (average 62y),and long-lived (average 82y)donors,respectively (Table I).
For each donor,a molecular-barcoded cDNA-bad TCR b li-brary was generated and analyzed by d
eep paired-end Illumina quencing.We procesd the output for normalization via unique molecular identifiers,with further clonotype clustering and error correction using MiTCR software (30)as described in Materials and Methods .The resulting normalized datats,which are available at /datats/,contained full quantitative information on the clonal composition and di-versity of TCR b CDR3repertoires per 106T cells from each donor.
The obrved diversity per 106T cells decread almost linearly with age (R =20.75,p =431028),ranging from 7.83105clonotypes for the youngest down to 1.33105clonotypes for one of the aged donors.We obrved a significant decrea in the number of unique TCR b clonotypes between young and middle-age (p ,0.01,1.3-fold)and between middle-age and aged (p ,0.01,1.6-fold)groups.However,a nonsignificant increa in TCR diversity between the aged and long-lived groups (Fig.2A,Table I)was obrved.
水嶋あい
Notably,analysis of 13106randomly chon CDR3-containing quencing reads without using molecular identifiers distinguished poorly between different age cohorts (p .0.05,two-tailed t test for age groups 1and 2).This is most likely becau group 1is char-acterized by the most diver repertoire,whereas read-bad data generally underestimate sample diversity (e Fig.1,Supplemental Fig.2A and below).
We analyzed absolute T cell counts in blood for lected donors,and obrved no significant age-related changes (data not shown),in agreement with previous data showing either no significant change or only a marginal decrea in either absolute count or proportion of CD3+cells with age (39–41).Thus,we believe that the obrved changes in the TCR b repertoire generally refer both to the diversity per 1million T cells and to the absolute T cell counts,albeit with a possible bias in the latter due to individual variance in T cell counts in the
blood.
FIGURE 1.Normalization via unique molecular barcoding increas detected TCR b diversity and decreas sample variance.(A )TCR b libraries prepared from three independent 2ml peripheral blood draws from the same donor.Columns show the number of unique TCR b clonotypes identified among 23105randomly lected,CDR3-containing quencing reads without considering unique molecular identifiers (left )versus the same number of reads characterized with unique molecular identifiers (right ).(B )TCR b libraries prepared from unequal PBMC samples (53106and 13107cells)obtained from a single blood draw.Columns show the number of unique TCR b clonotypes identified from 13106randomly-lected,CDR3-containing quencing reads without considering unique molecular identifiers (left )versus the same number of reads characterized with unique molecular identifiers (right ).TCR b libraries were prepared and analyzed using Illumina quencing as described in Materials and Methods .CV ,coefficient of variance.
Table I.Characteristics of donors by age group
Group No.
Average Age (Range),y No.of Patients (Female/Male)
Obrved TRBV CDR3Diversity per 106T Cells,3105
Lower Bound for Total TRBV CDR3Diversity,3106
%CD3Naive,Average,+/2%CD4Naive,Average,+/2%CD8Naive,Average,+/2
西安翻译学院是几本CD4/CD8T Cells Ratio,Average,+/2
Group 1,young 16(6–25)11(4/7) 6.461.0 5.061.354.3610.349.869.659.7614.0 1.3860.49Group 2,middle-age 39(34–43)10(6/4) 5.061.3 4.161.035.269.433.7612.437.568.1 1.9760.56Group 3,aged 62(61–66)7(4/3) 3.161.0 2.466.517.368.119.967.918.8612.6 2.4560.95Group 4,long-lived
82(71–90)
11(5/6)
3.660.9
2.666.2
27.5614.8
34.6620.9
11.266.1
2.5461.08
The Journal of Immunology 2691
at Merck & Co on April 15, 2014
/Downloaded from
Estimating the lower bound for total individual TCR b CDR3diversity
Analysis of species accumulation curves (Fig.2B)showed that the number of TCR b clonotypes detected in each donor sample in-cread proportionally with the number of cDNA molecules an-alyzed,with an almost direct relation.The species accumulation curves were similar within age-groups but differed between them.For young individuals,each additional 100,000cDNA molecules analyzed consistently yielded ∼60,000additional TCR b CDR3clonotype variants.In general,most curves remained far from sat-uration,confounding estimation of total TCR b diversity.
To estimate the lower bound of total TCR b diversity for the studied donors,we applied an unen species model (e Materials and Methods ).For young cohort (group 1),the estimated lower bound on total TCR b diversity was ∼73106different clonotypes,
and this estimate declined to ∼43106for middle-aged individuals.In group 3,the lower bound was ∼2.43106,significantly lower than that of group 2(1.6-fold,p =0.001)(Fig.2C).
showgood更夫It should be noted that this is only the lower bound estimate,which would be predicted to increa with greater sampling depth.To verify this,we further analyzed two independent replicas consisting of 13106unique cDNA reads each from two donors from our study,aged 25and 87,who blood samples contained 44and 15%naive T cells,respectively (starting from the same blood draw,cells parated at the level of purified PBMC).The additional replicas,which were added to the full analysis pipe-line starting from joint reads,incread the directly obrved di-versity 1.91-and 1.74-fold,and incread the lower-bound estimate for total TCR b diversity 2.360.4-and 2.460.1-fold for young and long-lived donors,respectively,confirming that
线上小游戏determination
FIGURE 2.Age-dependent trends in TCR repertoire diversity.(A )The number of unique TCR b CDR3clonotypes per 106T cells is shown as a function of donor age.Obrved diversity declines rou
ghly linearly with age (R =20.75).The trend is significant (p ,0.0001;Kruskal–Wallis test).**p ,0.01;two-tailed t test.(B )Species accumulation curves for TCR b clonotypes for all four age groups.We analyzed subts of a sample of 106cDNA molecules in increments of 13105.Plots show the average number of unique clonotypes obtained from random samplings (n =5)of cDNA molecules from each donor.(C )Estimates of the lower bound of total TCR b diversity in human donors.See Materials and Methods for details.
2692MEASURING AGE-RELATED DECREASE IN TCR DIVERSITY
at Merck & Co on April 15, 2014
/Downloaded from
