Age-related alterations of somatic hypermutation and CDR3 lengths in human Vκ4-expressing B lymphocytes

Abstract

The lower avidity and/or affinity of antibodies generated by an aged immune system could be attributed to two major changes in the antibody repertoire: a shift in germline gene usage and a decrease in the rate of immunoglobulin hypermutation. In an attempt to identify the mechanisms involved in the observed humoral immune deficiency in the elderly, we studied whether differences in the somatic diversity of a particular Vκ region occurred with ageing. By using the polymerase chain reaction and sequencing, we analysed and compared Vκ4–Jκ rearrangements isolated from young (mean age 21 years) and aged (mean age 83 years) healthy adults. Mutations in the Vκ4 gene compared with the germline sequence were determined as well as the length and structure of the CDR3 sequence. We analysed in detail various mechanisms contributing to CDR3 and Vκ variability in rearrangements involving the Vκ4 gene. Our data revealed that, despite strong individual variations, significantly lower levels of somatic mutation were found in the aged group, both for complementarity-determining regions (CDRs) and framework regions (FRs) encoding Vκ4 sequences. This decrease mostly affected mutations responsible for replacements and thus resulted in a lowered somatic diversification of the encoded Vκ4 proteins in aged individuals. Moreover, comparison of the CDR3 regions of the Vκ4–Cκ cDNA revealed changes in light-chain junctional diversity that correlated with age. Altogether these data suggest an impaired light-chain somatic diversity in connection with human senescence.

INTRODUCTION

Antibodies consist of two heavy chains and two light chains (either κ or λ). Each polypeptide chain contains a variable (V) domain for antigen recognition and a constant domain, which is responsible for effector functions. The variable domain is assembled at the DNA level by splicing together V, diversity (D, heavy chain only) and joining (J) gene segments. Immunoglobulin variable domains contain three intervals of sequence hypervariability (complementarity-determining regions, CDRs), which are separated from each other by four intervals with more conserved sequences called framework regions (FRs). The V gene segment encodes CDR1 and CDR2, whereas CDR3 is the product of V–D–J joining. In the mature protein, the heavy- and light-chain CDRs are juxtaposed to form the antigen (Ag)-binding site.^1–3

The human Vκ locus contains ≈32 potentially functional Vκ gene segments, grouped into seven families based on shared nucleotide sequence homology (VκI–VII); VκIV is located in a single gene and is the most Jκ proximal. Random combination of these germline Vκ gene segments with one of the five Jκ gene segments can produce a large array of different light-chain rearrangements.^4,5

The antibodies produced in immediate response to antigen challenge are usually of low affinity, but their affinity increases as the response progresses (affinity maturation), particularly after secondary challenge. The immunoglobulin genes of those B-cell clones amplified during the primary immune response are subjected to a number of nucleotide substitutions, leading to the generation of a population of daughter cells carrying mutated immunoglobulin V genes (somatic hypermutation). These mutations may either be silent or result in amino acid replacements that change the affinity of the immunoglobulin for the Ag, occasionally improving antigen binding; such cells expressing higher-affinity antibodies, usually with mutations clustered in the CDRs, are selected by Ag. This selection also leads to B-cell clones that are poorly represented in the primary response becoming enriched in later responses.^6,7

Humoral immune responsiveness and antibody-mediated defence mechanisms are markedly reduced with ageing. The age-associated changes in humoral immunity affect the quality more than the quantity of the antibody response; these changes are manifested by a shift from adaptative humoral immunity (production of highly specific, high-affinity, IgG antibody response to foreign Ag) to a process of natural antibody-mediated immunity (dominated by low-affinity, polyreactive, IgM antibodies that react with auto-Ag).^8–11 Although T-cell (T_H) function impairments appear to be the basis for this shift rather than an intrinsic primary B-cell deficit, ageing-associated alterations in B-cell repertoire expression have been reported in old mice.^11,12 An impaired affinity maturation and differential D–JH gene usage has been recently demonstrated in human VH6-expressing B lymphocytes from older individuals.¹³

The aim of this study was to determine whether such impaired diversification mechanisms in heavy chains during human senescence could be associated with similar alterations in the light-chain repertoire, which might relate to the decline of immunocompetence in aged humans. By using the reverse transcription–polymerase chain reaction technique (RT–PCR) with Vκ4- and Cκ-specific primers, followed by nucleotide sequencing, we analysed and compared Vκ4–Jκ rearrangements isolated from young (mean age 21 years) and aged (mean age 83 years) healthy adults. We chose to work at the mRNA level in order to consider mainly actively expressed functional genes, and to focus our study on B cells that had already encountered antigen. Mutations in the Vκ4 and Jκ genes were determined in order to study in detail how the various mechanisms contributing to the somatic diversification of the light-chain repertoire, somatic hypermutation and junctional diversity are affected in the elderly.

MATERIALS AND METHODS

Donor samples and RNA preparation

Peripheral blood samples were obtained from healthy young (Y) and aged (O) volunteers (three female and one male in each group) (Table 1). Peripheral blood mononuclear cells (PBMC) were isolated by centrifugation over Ficoll, and total RNA was extracted using RNA NOW (Ozyme, Montigny-le-Bretonneux, France).

Table 1.

Sample population

PMBC, peripheral blood mononuclear cells.

RT–PCR and PCR primers

cDNA was reverse transcribed from 1 μg of total RNA using an oligo dT primer (Pharmacia, Uppsala, Sweden) and reverse transcriptase (Gibco-BRL, Cergy Pontoise, France). Amplification of the cDNA (1 μl) was performed in 50 μl of Taq 1×PCR buffer containing 200 μm of each dNTP, 100 ng of each primer and 1 U Taq polymerase (Pharmacia); PCR was performed using a multistep programme: 1 cycle at 94° for 5 min; 35 cycles at 94° for 30 seconds, 51° for 30 seconds and 72° for 30 seconds; and 1 cycle at 72° for 7 min.¹⁴ Vκ4–Cκ rearrangements were detected by PCR amplification using the following primers: sense leader Vκ4 primer 5′-AAGTCGACATGGTGTTGCAGACCCA-3′; and antisense Cκ primer 5′-GCGGGAAGATGAAGACAGATGG TGCAG-3′ (Genosys, Cambridge, Cambs, UK).

Cloning and sequencing

Amplified cDNA was separated on a 1.2% agarose gel, isolated through electroelution and cloned into the pCRII-TOPO vector (Invitrogen, Leek, the Netherlands) according to the manufacturer’s instructions. After transformation of Escherichia coli, DNA from putative recombinants was analysed by restriction to confirm that a fragment of the appropriate size (≈450 bp) had been cloned. Plasmids were isolated using a standard Triton/lysozyme method. All inserts were sequenced from both ends using the dideoxynucleotide termination method,¹⁵ with Taq DNA polymerase using the M13 Reverse and M13(–20) forward primers and an automated laser fluorescent DNA ABI 310 sequencer (Perkin-Elmer, Branchburg, NJ).

Analysis of Vκ4 and Vκ4–Jκ junctional regions

Sequences were compared with the corresponding regions of germline gene segments.⁴ Levels of somatic hypermutation were assessed by comparing regions from FR1 to CDR3 (codons 1–95 according to reference 16) of each complete transcript with the appropriate germline sequence and calculating mean rates for each sample. Nucleotide mismatches at the 3′ end of the Vκ4 gene segment or the 5′end of the Jκ gene segment were assumed to represent N-region addition rather than somatic mutation.

Statistical analysis

Differences in the rate of somatic mutation of clones from older and younger individuals, and CDR3 length heterogeneity, were analysed using the χ²-test or unpaired Student’s t-test, as appropriate, considering for each group the total number of Vκ4 base pairs sequenced and the length of the sequences (i.e. 69 bp for FR1, 51 bp for CDR1, 45 bp for FR2, 21 bp for CDR2, 96 bp for FR3, 21 bp for CDR3 and 303 bp for the total Vκ4 gene). The number (N) of expected R (replacement) mutations in CDRs or FRs for Vκ4 was calculated for each group using the formula described by Chang & Casali;¹⁷ this formula is based on the total observed mutations (n =Replacement+Silent), the relative lengths of the CDR or FR (CDR_rel or FR_rel), and the replacement frequency inherent to CDR or FR sequences (CDR R.f. = 0·7836 or FR R.f. = 0·7529):¹⁷

RESULTS

Sequence analysis of Vκ4 transcripts

Vκ4–Jκ transcripts from PBMC of young and old healthy volunteers were amplified using an RT–PCR assay (Table 1). A total of 86 clones containing Vκ4–Jκ recombinations were sequenced and analysed: 41 for the young and 45 for the aged individuals. The cDNA sequences were compared with the respective germline genes and the mutation frequencies were determined. The nucleotide substitution ratio (R:S, i.e. replacement versus silent mutations) was, in general, lower in the FR and CDR than expected randomly (3·0 and 3·6, respectively, according to reference 17). Large interindividual variations were observed among the aged subjects (Table 2); in the case of V1 and V4, a higher than expected ratio of R:S was found in the CDR and FR, respectively.

Table 2.

Distribution of mutations in the Vκ4 gene*

*The total number of nucleotide changes are given for each group; significantly different values are in bold (P <=0·01, x²-test)

†Framework regions (FR) are located between residues 1 and 23, 35 and 49, and 57 and 88.

‡Complementarity-determining regions (CDR) are located between residues 24 and 34,50 and 56, and 89 and 95, according to the terminology of Kabat et al¹⁶

§The numbers in parentheses indicate the number of R mutations that occurred by chance (calculations according to Chang & Casali¹⁷). The distribution of R mutations in the FR and CDR was found to be different from that expected by chance for each group(P <0·001; x²-test).

However, when all sequences were pooled in the two age groups and compared, our results showed statistically significant differences (P < 0·001) in mutational frequencies between the age groups (Fig. 1). The mutational frequency of the Vκ4 rearrangements was lower in PBMC from aged compared with young individuals. The pattern of this age-associated decreased frequency of mutations in the rearranged Vκ4 affected both the CDRs and the FRs (Fig. 1a); nevertheless, there was a more pronounced effect for the CDR1, CDR3 and FR1 regions where the average number of mutations was significantly lower in the old age group by comparison with the young age group (Fig. 1b). As for Jκ regions, the rate of somatic mutations did not differ significantly between the two groups (Table 3).

Figure 1.

Mutations in the Vκ4 gene of young and aged adults. (a) The frequency of mutations was calculated as a percentage for the framework regions (FR) and/or complementarity-determining regions (CDR) of the immunoglobulin V regions. (b) Average number of mutations within each of the regions encoded by the Vκ4 gene segment is given when normalized for length. *P < 0·001 between young and older individuals (χ² analysis).

Table 3.

CDR3 length, Jκ length and sequence analysis in Vκ4–Cκ transcripts from young and old age groups

Data represent mean ±SEM.

Statistical analysis between young and old age-groups was performed using the unpaired Student’s t-test.

The distribution of mutations in the Vκ4 gene were analysed among the two groups (Table 2). Replacement mutations occurred more frequently in the CDR (43% and 46% for the pooled young and old sequences, respectively) and less frequently in the FR (26% and 23% for the pooled young and old sequences, respectively) than those expected from random mutation. Moreover, a comparison of the pooled young to the pooled aged sequences showed that R mutations were significantly reduced in the CDR- and FR-encoded regions of the Vκ4 gene in the old sample as compared with the young (Table 2). Analysis of deduced translated amino acid sequences of the Vκ4 light chain thus revealed a significant decrease (P < 0·001, χ² analysis) of the variability of residues in the aged sample (mean 5·1 replacements±0·7 versus 7·1±0·9 in the old and young groups, respectively) although the residues that have the highest variability are clustered as expected within the CDRs (in particular the CDR1) for the two groups (Fig. 2).

Figure 2.

Variability of mutations in the deduced Vκ4 light chain of young (a) and old (b) adults. The ordinate represents the number of mutations at each amino acid position. Limits of complementarity-determining regions (CDRs) are delineated according to the terminology of Kabat et al.¹⁶

Jκ usage

Comparison of rearrangements derived from PBMC in the two age groups revealed variation in Jκ gene usage (Fig. 3). Although the previously reported under-representation of Jκ5 and, to a lesser extent, Jκ3 joints was found in both groups, preferential usage of Jκ1 (40%) appeared in aged individuals, whereas there was a co-dominant usage of both Jκ2 and Jκ4 (33%) in the young age group (Jκ1 and Jκ5 together accounted for 47% of rearrangements in the aged group versus 27% in the young, P =0·057, χ² analysis).

Figure 3.

Jκ gene usage in Vκ4 transcripts. Results are expressed as percentage±SEM, calculated as the number of sequences utilizing each Jκ gene divided by the total number of recombination events within each age group multiplied by 100.

CDR3 regions

Vκ4–Jκ junctional regions were sequenced from the 86 clones (41 young and 45 aged) (Fig. 4); nucleotides not assignable to germline Vκ4 or Jκ sequences are shown as N-region addition. CDR3 limits were determined according to Kabat et al.¹⁶ Total CDR3 lengths demonstrated relative homogeneity in each group; non-functional rearrangements (as defined by an out-of-frame CDR3 region) were occasionally found to be expressed as sterile transcripts (four clones among the old age group, two among the young age group). A global comparison of ‘young’ versus ‘aged’ junctions showed a significant difference and an increased length of the CDR3 domains in the sample from the old age group (Table 3) (P =0·038); this difference resulted from a significant increase (P =0·032) of the Jκ length in the aged as compared with the young group, fewer nucleotides being deleted at the 5′ end of Jκ segments in the aged (1·4±0·2 versus 2·3±1·4 bp deleted in old and young Jκ joints, respectively).

Figure 4.

Nucleotide sequences of CDR3 regions of clones from peripheral blood mononuclear cells (PBMC) of young (a) and aged (b) adults. All sequences are compared with the corresponding germline Vκ4 or Jκ gene segment, with a dot indicating identity. Nucleotides not assignable to germline Vκ4 or Jκ sequence are shown as N-region addition. CDR3 lengths are shown as the number of residues from codon 89 to codon 97, as assigned by Kabat et al.¹⁶*indicates an out-of-frame V–J junction.

DISCUSSION

An age-associated decrease in the average affinity of antibody responses has been known for many years.¹⁸ However, the underlying molecular cause(s) of these changes remain poorly understood. High-affinity antibodies are produced by somatic mutation of the rearranged genes, followed by selection for improved binding.¹⁹ Somatic mutations decline in the primary immune response to the hapten (4-hydroxy-3nitrophenyl)acetyl (NP) in aged mice²⁰ but little is known about antibody diversification in aged humans. Conflicting data are reported in the literature with regard to somatic mutation of antibody V regions throughout life.²¹ Recently, studies on VH6–D–JH rearrangements showed impaired affinity maturation in older humans with a significant decrease of the mutational frequency of the VH6 gene.¹³

In this work, we examined Vκ4–Jκ rearrangements in aged humans and compared them with those found in young adults. We chose to focus on transcripts derived from the Vκ4 gene (B3 in Zachau’s nomenclature⁵) because it is the only representative gene of the VκIV family and it displays no allelic polymorphism.²²

Statistically significant differences in mutational frequency were found between the age groups (Fig. 1 and Table 2). Analysis of the Vκ4 gene showed decreased levels of somatic mutation for both CDRs (CDR1 and CDR3) and FRs (FR1) in the pooled aged individuals when compared with the pooled young sequences.

Clonal expansion takes place in the germinal centre cells, which have acquired mutations that improve antigen binding whilst preserving the correct folding of the immunoglobulin V region. B cells subjected to several rounds of antigen selection are expected to exhibit clustering of R (replacement) mutations in the CDRs (positive selection for amino acid changes), whereas, on the contrary, R mutations would be less frequent in the FR (negative selection for destabilizing substitutions).²³ A binomial model tests whether the observed distribution of R mutations follows this pattern.^17,23,24

We studied, among the two age groups, whether or not mutations would lead to an amino acid replacement (Table 2). Consistent with selection by antigen, higher numbers of R mutations were found in the CDRs of both groups than those theoretically expected (P < 0·001); furthermore, in accordance with the preservation of residues shaping the FRs, the sequences displayed lower numbers of R mutations in the FR than those expected from random mutation (P < 0·001). However, the R:S ratio was lower in the FR and CDR than expected randomly (3·0 and 3·6, respectively, according to reference 17) suggesting the action of negative selection limiting the number of R mutations in the FR, presumably to maintain immunoglobulin structure, and in the CDR to preserve affinity; these results were found to be similar in both groups.

In addition, data revealed individual variation among the aged subjects; two of the four aged individuals showed much higher (V1) or lower (V4) average mutation frequencies (15 and four, respectively) as compared to the others. Analysis of the R:S ratio suggested a memory response for V1 (a higher R:S ratio was found in the CDR than randomly expected; Table 2).

Analysis of the distribution of the mutations between the pooled groups of the two ages revealed that the decreased levels of somatic mutation with increasing age mostly affected mutations responsible for replacements, and thus resulted in a significantly lowered somatic diversification of the Vκ4 proteins in aged individuals.

Present results suggest an age-related impaired affinity maturation in human Vκ4-expressing lymphocytes, which resembles previous findings on the VH6 gene.¹³ Differences in the proportions of B-cell subsets^21,25 as well as differences in immunoglobulin mRNA expression among B cells at different stages of differentiation between older and younger individuals may, in part, explain the apparent age-associated decrease of somatic mutation in PBMC Vκs. Healthy aged individuals have been found to have fewer circulating B lymphocytes and higher levels of IgG and IgA, but not IgM, than younger individuals.²⁶ In addition, as for T cells, accumulation of circulating memory B cells in the elderly has been suggested;^4,25 however, such accumulated memory B cells in the periphery would be expected to increase the level of mutations encountered in PBMC immunoglobulin mRNAs, but are thus unlikely to explain the herein reported decrease of somatic hypermutation in old individuals.

There are five functional Jκ segments in humans, albeit used with different frequencies.⁴ The Jκ segment used in each Vκ4–Cκ cDNA was identified by maximal alignment with the most closely related germline Jκ segment.¹⁶ Data reported in this study seem to indicate a preferential usage of Jκ1 in aged individuals but no restricted usage (40, 20, 4, 29 and 7% for Jκ1, Jκ2, Jκ3, Jκ4 and Jκ5, respectively); furthermore, our results are in agreement with a previous report of Klein et al. concerning a cDNA library derived from an 86-year-old person⁴ where a similar expression of Jκ genes was observed (35, 25, 8, 26 and 6% for Jκ1, Jκ2, Jκ3, Jκ4 and Jκ5, respectively); the infrequent usage of Jκ5 has also been detected in a 67-year-old individual by Klein et al.²¹ J_H- but not D-gene usage changes have been reported in human V_H6-expressing B lymphocytes during senescence.¹³ In contrast, in a recent study on VH5-Cμ cDNA sequences, Xue et al.²⁷ conclude that the process of ageing does not significantly affect D or J_H usage in the peripheral repertoire; our results suggest modest Jκ usage differences during senescence in Vκ4–Cκ light chains but no major shifts.

Junctional diversity generated in the assembly process by deletion and de novo addition of nucleotides (N-region addition) at the V–(D)–J junctions is another important source of CDR3 diversity; although terminal deoxynucleotidyl transferase (TdT) generally decreases in B cells before the κ genes are rearranged, the presence of short N segments in these genes has been repeatedly reported.^4,21,25 Although we observed more nucleotides added in the ‘young’ (15 in 41 sequences) than in the ‘old’ (five in 45 sequences), this difference is not significant and the increased additions did not balance the increase rate of junctional deletions in the young, so that they had overall shorter CDRs regions (Fig. 4 and Table 3).

Alterations in CDR3 diversity with ageing could be involved in the age-associated decrease in the affinity of antibody responses known for many years.^18,27 As regards CDR3 diversity, aged mice show no significant changes in nucleotide addition or deletion at VH–D–JH junctions; however, a marked shift in D-segment family utilization between aged and young mice has been found.²⁸ In a recent study,²⁷ a marginally significant increase of a few nucleotides was found in average V_H5-Cμ CDR3 length of aged adults human (mean 41 bp) by comparison to young people (mean 37·5 bp). Our data show alterations in the Vκ4 light-chain junctional diversification process in aged individuals with a higher proportion of long CDR3 regions (10 amino acid codons, Fig. 1), which is reminiscent to data obtained for the V_H5-Cμ heavy chain.²⁷

In conclusion, our results show a decreased Vκ4 mutation frequency affecting, in particular, replacement mutations; they also show significant changes in the CDR3 region of the Vκ4–Cκ transcripts during senescence. These long CDR3s suggest alterations in the molecular processes involved in immunoglobulin CDR3 diversification during B-cell development in the elderly, with less N-nucleotide insertion and with a lower frequency of deletions mediated by exonuclease activity at the 5′ end of Jκ. Altogether, CDR3 are longer in sequences from aged individuals and the process of ‘junctional diversity’ appears to be decreased.

This decrease in somatic diversification is consistent with results obtained by van Dijk-Härd et al.¹³ as regards the heavy-chain repertoire diversity and reveal impaired affinity maturation in immunoglobulin chains of older humans, which might explain, in part, humoral immunosenescence in humans.