Narrowing of Human Influenza A Virus-Specific T Cell Receptor α and β Repertoires with Increasing Age

Ethics statement.

The study was approved by the Institutional Review Board of the University of Massachusetts Medical School and the local ethics committee of the Blood Center of Wisconsin. All individuals gave written informed consent.

Study population.

Blood samples were collected from HLA-A2.1-positive (HLA-A*0201) individuals, whose demographic and VA and VB family usage information is summarized in Table 1. HLA typing was performed by use of a Biotest sequence-specific primer (SSP) system (Biotest Diagnostics, Denville, NJ). Peripheral blood mononuclear cells (PBMCs) were obtained by use of Ficoll-Paque Plus medium (Amersham Biosciences, Piscataway, NJ) following the manufacturer's protocol.

Design of study.

To determine age-related changes, we performed a comparative analysis of the CD8⁺ T cell repertoires of eight middle-aged (mean ± standard deviation[SD] age, 42.9 ± 7.95 years) and eight older (mean ± SD age, 74 ± 7.76 years) individuals. Due to the fact that the frequency of memory CD8 T cells in blood specific for the M1_58–66 peptide is very low (0.2%) (24), in our studies we utilized CD8 T cells isolated from 2-week cultures. As a control for our cell culturing procedure, which enables the generation of antigen-specific T cells, we also included data for fluorescence-activated cell-sorted and M1 tetramer-stained T cells received from two middle-aged donors previously described (25).

An initial screening of the T cell repertoire in M1_58–66-specific CD8 T cell populations was done by CDR3 length analysis. In the spectratype technique, PCR was used to amplify the template cDNA corresponding to rearranged transcripts with CDR3 regions of different lengths. By using this method, we compared the CDR3 profiles of cultures of unstimulated and M1-stimulated CD8 T cells from middle-aged and older individuals. All samples were analyzed for the usage of the VB19 family. Vα was focused on the VA27, VA10, VA8.1, VA8.6, and VA12 families, which have previously been demonstrated to be commonly utilized by middle-aged donors (25). Representative examples of spectratypes are shown in Fig. S1A to D in the supplemental material.

Peptide synthesis.

The influenza A virus matrix protein-derived peptide M1_58–66 (GILGFVFTL) was from BioSource International and purified to 90% purity.

Cytotoxic T lymphocyte isolation, RNA isolation, and cDNA synthesis.

PBMCs were isolated from peripheral blood by the use of Ficoll-Paque Plus medium (Amersham Biosciences, Piscataway, NJ) according to the manufacturer's recommendations. Cells were cultured at a concentration of 1.5 × 10⁶/ml in the presence of 1 μM M1_58–66 peptide in 24-well plates for 1 week. On day 4, recombinant interleukin-2 (rIL-2; 10 U/ml) was added to the cell culture. After 7 days, the primed cells were restimulated with an equal number of irradiated autologous PBMCs (3,000 rads) that had previously been precoated with 1 μM M1_58–66 peptide. rIL-2 was added to the cell culture every 4 days. On day 14, the cells were collected and subjected to CD8⁺ T cell isolation. To demonstrate that cell culture stimulation with the M1 peptide resulted in cell proliferation, samples representing CD8⁺ T cells not treated with antigen (ex vivo) were included.

To isolate CD8⁺ T cells, a Dynal CD8 positive isolation kit (Invitrogen) was applied according to the manufacturer's recommendations. The mRNA was extracted from cultured cells using the kit (Invitrogen) following the manufacturer's recommendations. In the next step, the mRNA samples were converted into cDNA using 500 units of Moloney murine leukemia virus reverse transcriptase (Invitrogen), an RNAguard system (Amersham Biosciences), and poly(T) (Integrated DNA Technologies) as the primer for reverse transcription.

CDR3 spectratyping.

Before spectratyping analysis, cDNA titration was performed by PCR amplification with primers specific for the constant β-chain (CB) region in the presence of serially diluted cDNA samples. Next, amplified PCR products were visualized and quantified on a resolving gel using ImageQuant software. Aliquots which contained the same quantity of total TCR β-chain transcripts were used for spectratype analysis of CDR3 sizes within TCR β and α chains (TCRβ and TCRα, respectively). For this purpose, titrated cDNA samples were subjected to one-step PCR amplification with previously described primers specific for the VB19, VA12, and VA27 families (forward primers). Primers specific for CB or the constant α-chain (CA) region were labeled with 6-carboxyfluorescein at the 5′ end and utilized as reverse primers. The conditions of PCR amplification for the TCR β and α chains were previously described (25). The fluorescent PCR products and size marker (Applied Biosystems, Carlsbad, CA) were mixed with HiDi formamide (Applied Biosystems, Carlsbad, CA), heated to 90°C for 2 min, and run on an ABI 3730xl DNA analyzer (Applied Biosystems, Carlsbad, CA). All the data were analyzed with proprietary software (Flynn Creek Biosciences, Hubertus, WI).

PCR product subcloning and DNA sequencing.

The PCR products were subsequently subcloned into the pCR4-TOPO plasmid vector (Invitrogen). Ligated products were then transformed into One-Shot TOP10 chemically competent cells (Invitrogen) according to the manufacturer's instructions. Sequencing was performed by Agencourt Bioscience (Beckman Coulter, Beverly, MA) or DNA GeneWiz (South Plainfield, NJ). Data were analyzed using software developed in-house.

TCRα and TCRβ clonotype analysis.

The term “clonotype” denotes T cells with unique DNA sequences that result from TCRα or TCRβ rearrangements, and it is possible that a unique VA sequence may pair up with more than one VB sequence and vice versa. The total number of sequences (M) and total number of unique clonotypes (N) for each donor are shown in Table 1. To analyze the clonality of VA27, VA12, and VB19 family TCR repertoires for all donors, we calculated the percentage of unique clonotypes (N/M) and the percentage of dominant clonotypes (M_d/M, where M_d is the total number of sequences of the dominant clonotype) (see Tables 2 to 4 and Tables S1A to C in the supplemental material). A unique nucleotide sequence indicates a nucleotide sequence which was not repeated in the T cell repertoire of each donor, and a dominant clonotype was defined to be the most frequent nucleotide sequence identified among all sequences.

To test the diversity of the VA27, VA12, and VB19 family TCR repertoires for each donor analyzed, we calculated the previously described Simpson's diversity index (26).

To demonstrate the contribution of the JA42, JA37, and JA29 families in combination with the VA27 and VA12 families for each donor, we calculated the percentages of particular clonotypes (e.g., for VA27-JA42 clonotypes, the number of sequences of VA27-JA42 clonotypes/total number of sequences [M_JA42/M, where M_JA42 is the total number of sequences of the JA42 clonotype] was calculated). The contribution of the CARDGGGNTPLV sequence in four middle-aged and eight older donors was determined by calculation of the percentage of VA12-JA29 clonotypes encoding this sequence (the number of sequences of VA12-JA29 clonotypes/total number of sequences [M_CARD/M], were M_CARD is the total number of sequences of the clonotype encoding CARDGGGNTPLV).

CDR3 loop amino acid sequence analysis.

The length of the CDR3 region was defined by the number of amino acids found after the cysteine (C) of the V region and continued to the sequence of the J region. Age-dependent changes in the sizes of CDR3 of VA27-JA42 clonotypes were shown by calculation of their percentages in two CDR3 lengths: 10 aa and 12 aa. The glycine, alanine, and serine amino acid contents were determined by calculation of the numbers of each amino acid in shorter (10-aa) and longer (12-aa) CDR3 regions.

Statistical analysis of TCR repertoire.

Comparisons of two groups of individuals, middle-aged and older individuals, were done using the nonparametric Mann-Whitney U test. The relation between the percentage of unique and dominant clonotypes and the donor's age was analyzed using the Spearman correlation coefficient. To compare differences in the numbers of glycines, alanines, and serines in 10-aa and 12-aa-long CDR3 regions, an unpaired t test (two-tailed) was applied. P values of ≥0.05 were considered significantly different.

The usage of the VA27 and VA12 families differed in the two study populations.

All the middle-aged (8/8) and older (8/8) donors utilized the VB19 family in response to M1_58–66 peptide stimulation, as expected (21, 23) (Table 1). In the group of middle-aged donors, the VA27 family was used by all individuals (8/8), as expected (25), but surprisingly, VA27 was used in only some of the older individuals (5/8) (Table 1). Interestingly, VA12 family usage showed the reverse, being present in all older donors (8/8), and, as expected, its usage was more variable in younger individuals, where five out of eight donors utilized this family (Table 1). We therefore focused our more detailed TCR sequencing studies on these three dominant families, VB19, VA27, and VA12. The detailed TCR sequencing results for five representative older and three middle-aged donors are shown in Table 2, those for all eight older and two representative middle-aged donors are shown in Table 3, and those for all eight older and three representative middle-aged donors are shown in Table 4. The results for the other middle-aged donors are shown in Tables S1A to C in the supplemental material.

The distributions of the VA and VB TCR clonotypes differed in the two age groups.

Figure 1 depicts the clonotype distribution of the VA and VB TCR repertoires in the individual middle-aged and older donors, giving an overview of their general organization. Medium-frequency clonotypes and clonotypes observed only once (singletons) in either longer or shorter tails were observed. This depiction of the shape of the TCR repertoire for the three most commonly used families, VB19, VA27, and VA12, gives the impression that TCR repertoires differ in middle-aged and older donors. It appears that in contrast to middle-aged donors, older individuals have more dominant oligoclonal VA27 clonotypes. This might suggest that with age the TCR repertoire becomes skewed and more focused. Also, the distributions of the VA27 family clonotypes in middle-aged individuals were more similar to each other than to those in older donors. They did not contain highly dominant clonotypes and had long tails, consistent with increased diversity. These differences between the two groups of donors appeared to extend to the VB19 family as well. To determine if these differences in the patterns of VA and VB clonotype usage in the two age groups were significantly different, we performed statistical analyses, as outlined in the subsections below.

(i) Increased oligoclonality of the TCR VA27 family in older individuals.

Consistent with TCR repertoire narrowing with age, older donors showed a significant decrease in the percentage of unique VA27 clonotypes (Fig. 2Ai) in association with an increased usage of dominant VA27 clonotypes (Fig. 2Bi) compared to the percentage for middle-aged donors. For instance, for the VA27 family, in middle-aged donor 6, 36 clonotypes out of 127 total sequences (28%) were unique, whereas in older donor G, as few as 10 clonotypes of 178 sequences (6%) were unique (Table 2). Older donors F and G had the most focused repertoires of all the individuals studied, with 132 and 135 observations per dominant clonotype, respectively (Table 2).

(ii) The VA12 family analysis did not show significant differences between the two age groups.

There was a trend toward a decreased percentage of unique clonotypes in the VA12 family in older versus middle-aged donors, but this did not reach statistical significance (Fig. 2Aii). For the VA12 repertoire, there was no significant difference in the usage of dominant clonotypes between the two groups (Fig. 2Bii and Table 3; see also Table S1B in the supplemental material).

(iii) The VB19 family repertoire showed increased oligoclonality in older individuals.

The VB19 T cell repertoire followed a pattern similar to that for the VA27 T cell repertoire, with greater oligoclonality and a significantly decreased percentage of unique clonotypes (Fig. 2Aiii). This decreased percentage of unique VB19 clonotypes was associated with a significantly higher percentage of dominant clonotypes in older individuals (Fig. 2Biii). For instance, in one older donor (donor F), the frequency of the most dominant clonotype was 109 of 166 (66%) total sequences analyzed (Table 4). This was in contrast to the findings for middle-aged donors. For instance, in donor 3 the frequency of dominant clonotypes was 23 of 125 (18.5%) total sequences (Table 4), and the highest frequency of the most dominant clonotypes found in middle-aged donors was 41% (40/98) (see Table S1C in the supplemental material).

(iv) Decreased diversity of the TCR VA27 and VB19 families with increasing age.

To test if the TCR repertoire of older individuals is statistically significantly less diverse than that of middle-age donors, we calculated Simpson's diversity index for each donor (26). The value of this index ranges from 0 to 1, with 0 and 1 representing the minimal and maximal diversities, respectively. Median Simpson's diversity indices of 0.6 and 0.84 of the VA27 family repertoire in older and middle-aged individuals, respectively, demonstrated a significant decrease in VA27 diversity with age (Fig. 3A), but such a decrease was not found for the VA12 family (Fig. 3B). The VB19 repertoire was also significantly less diverse in older donors (n = 8), with a mean of 0.77 ± 0.09, than middle-aged individuals, with a mean of 0.9 ± 0.06 (n = 8) (Fig. 3C).

(v) Increased age correlated with increased oligoclonality and decreased diversity.

For the VA27 family there was an inverse correlation between age and the percentage of unique clonotypes (Spearman's r² = 0.41, P = 0.02) (Fig. 4Ai) and a direct correlation between increasing age and the percentage of dominant clonotypes (Spearman's r² = 0.42, P = 0.02) (Fig. 4Bi). Also, the percentage of unique clonotypes in the VA12 family decreased with age (Spearman's r² = 0.36, P = 0.03) (Fig. 4Aii), whereas the percentage of dominant clonotypes was less age dependent (Spearman's r² = 0.16, P = 0.2) (Fig. 4Bii). Similarly, the percentage of unique and dominant clonotypes of the VB19 family was inversely and directly correlated with age, respectively (Spearman's r² = 0.46 and P = 0.004 for unique clonotypes and Spearman's r² = 0.34 and P = 0.02 for dominant clonotypes) (Fig. 4Aiii and Biii, espectively). Taken together, these data demonstrate that increasing age directly correlates with repertoire narrowing.

The characteristic JA usage in the T cell response to the M1_58–66 peptide in middle-aged donors is not maintained in older individuals.

The M1_58–66-specific CD8⁺ T cell response of middle-aged individuals preferentially utilizes the VA27-JA42 and VA27-JA37 clonotypes (25). Thus, we determined if this preferential usage was maintained in older donors. In all donors who utilized the VA27 family, the contribution of two JA families, JA42 and JA37, was calculated. Since JA42 usage has also been demonstrated among donors who use other VA families (25), the analysis was extended to the VA12 family. The VA27-JA42 clonotypes were universally detected at high percentages in older and middle-aged individuals (47% and 55%, respectively) (Fig. 5A). The most significant difference observed between the two age groups was in the distribution of VA27-JA37 clonotypes, with a decrease in the usage of the VA27-JA37 clonotype being noted in older individuals (for older versus middle-aged individuals, 12.50% ± 26.59% versus 38.38% ± 31.56% [P = 0.05; n = 5 versus n = 8]) (Fig. 5A).

With age, the usage of the VA12-JA42 and VA12-JA37 clonotypes was reduced 2- and 3-fold, respectively. In older donors the remarkable new feature was the contribution of the JA29 family, which was detected only in combination with the VA12 family and not the VA27 family (Fig. 5A). By comparison to middle-aged individuals, there was a 2.5-fold increase in the utilization of VA12-JA29 clonotypes in older individuals.

One unique VA12 family clonotype was present in all older individuals.

In a further analysis of the VA12 repertoire, one clonotype with the amino acid sequence CARDGGGNTPLV was repeatedly present in all eight older donors and only four of the eight middle-aged donors (Fig. 5B and Table 3; see also Table S1B in the supplemental material). Not only was it present in all older donors (8/8), but also it was the most dominant VA12 clonotype in older individuals. Its usage was significantly increased to 36% in older donors and was detected in only 12.25% of the JA repertoire in middle-aged donors (for older versus middle-aged individuals, 36.5% ± 20.07% versus 12.25% ± 15% [P = 0.02; n = 8 versus n = 4]). These results suggest that the VA27 repertoire changes with increasing age, concomitant with decreased JA42 and JA37 usage and increased usage of other JA families, as well as a surprising predominant focus on one unique VA12 clonotype.

Increased usage of longer CDR3α loops and increased content of glycines and alanines in older individuals.

To determine if there were also changes at the amino acid level rather than the nucleotide level of the CDR3α regions in older donors, we performed further analysis. In our prior studies in all middle-aged donors, we have demonstrated the increased usage of AGA(G_n) GG-like amino acid motifs (where A is alanine and G is glycine) in VA27-JA42 clonotypes compared to that in naive thymocytes (25). It has been demonstrated that increased numbers of G/A amino acid residues in the CDR3 regions increase the flexibility of the TCR, leading to greater promiscuity in antigen recognition (27). First, we analyzed age-related changes in the size of the CDR3 regions of VA27-JA42 clonotypes. In contrast to the findings for middle-aged donors, older individuals had a tendency to utilize longer CDR3 loops (11 aa to 16 aa) (Fig. 6A). The percentage of clonotypes with shorter CDR3 regions (8 aa to 10 aa) was almost 1.5-fold higher in middle-aged donors.

Second, we were interested in whether the change in the size of the CDR3 region with age would also influence the amino acid composition, particularly the glycine and alanine content, as these amino acids promote flexibility in TCR structure. Therefore, we calculated the number of all glycines and alanines in 10-aa- and 12-aa-long CDR3 loops of the VA27-JA42 clonotypes (Fig. 6B; see also Table S2 in the supplemental material). We also included data for serine (S), which has been identified to be involved in the molecular flexibility of proteins (27). Interestingly, the number of glycines in the shorter CDR3 regions was significantly higher in older donors than in middle-aged individuals (for older versus middle-aged individuals, 3.75 ± 0.17 versus 3.2 ± 0.18 [P = 0.05; n = 12 versus n = 14]) (Fig. 6B). There was no difference in the alanine and serine residue compositions of the 10-aa-long CDR3 loops with age. However, the number of glycines was significantly increased in the longer CDR3 loops than in the shorter CDR3 loops in both the middle-aged individuals (3.2 ± 0.18) versus 4.08 ± 0.18 for the 10-aa versus 12-aa CDR3 loops [P = 0.01; n = 14 versus n = 15 and the older individuals (3.75 ± 0.18) versus 4.44 ± 0.17 for the 10-aa versus 12-aa CDR3 loops; [P = 0.003]; n = 12 versus n = 9. Correspondingly, significantly increased numbers of alanines were observed in the longer CDR3 loops of both the middle-aged donors (1 ± 0.16 versus 1.6 ± 0.13 for the 10-aa versus 12-aa CDR3 loops [P = 0.01; n = 14 versus n = 15]) and the older individuals (1.08 ± 0.14 versus 1.88 ± 0.11 for the 10-aa versus 12-aa CDR3 loops [P = 0.0006; n = 12 versus n = 9]). Perhaps the longer CDR3 loops already had maximal numbers of glycines and alanines, and therefore, no differences in the numbers of alanines and glycines between the two age groups were observed for these longer 12-aa CDR3 loops.