| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department for Antibody Discovery, Symphogen A/S, Lyngby, Denmark
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
The functional development of Ag-selected polyclonal Ab repertoires has mainly been studied by making mAbs from mice progressively exposed to Ag. By studying anti-hapten responses it has been concluded that primary responses are of low affinity, which increases as a consequence of somatic mutations upon multiple challenges with Ag (2, 3, 4). However, anti-hapten responses are known to be genetically restricted and involve only a limited set of V gene segments (discussed below), implying that the Ag has a significant role in shaping such Ab repertoires. In contrast, high-affinity Abs have been observed early in mice in response to a viral infection (5) and to hen egg lysozyme (6), implying that primary Ab responses toward more complex Ags are less restricted in terms of affinity. Foote and Eisen (7) proposed a theoretical limit for Ab affinities in developing B cell responses based on the consideration that the selection based on the off-rate must be limited by the rate of Ag endocytosis, leading to maximal off-rates in the range of 10–3 to 10–4 s–1 and that the on-rates cannot exceed the maximal rate of diffusion in the range of 105 to 106 M–1 s–1. Thus, the immune system should not be able to select for affinity constants (KD) beyond 0.1 nM, although such Abs might arise by chance.
Genetic diversity of Ag-selected polyclonal Ab repertoires toward complex Ags has mainly been studied by using phage display technology because of its ability to screen massive numbers of V gene sequences. As a recent example, 
55 uniquely rearranged H chain variable domain (VH)2 sequences were isolated by phage display from two human donors immunized with a rabies vaccine (8). However, phage display involves the random pairing of H and L chain genes, which often leads to a drop in affinity. Consequently, only about one-third of the anti-rabies Abs appear to be of high affinity. In addition, phage display preferentially selects for promiscuous VH domains, which further reduces the diversity of the Ag-specific repertoires isolated by phage display (9, 10). Thus, although it is currently the preferred method for studying the composition of Ag-specific repertoires, phage display is likely to underestimate the actual VH diversity generated by the immune system and, further, provides no information regarding cognate L chain usage and, hence, native affinities.
We have recently reported a novel strategy, the Symplex Technology, for the isolation of comprehensive repertoires of cognate pairs of Ab V genes directly from human plasma cells (10). The screening of such V gene repertoires allows for simultaneous analysis of the clonal and functional diversity of the Ag-selected human Ab repertoire. In this study we examine the genetic diversity and binding characteristics of two comprehensive panels of distinct mAbs obtained from single individuals immunized with a tetanus toxoid (TT) vaccine. Our results have implications for the understanding of the mechanisms that lead to a polyclonal response and for the limits of the selective forces responsible for affinity maturation.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
Two healthy volunteers were boosted with a TT vaccine (Statens Serum Institut, Copenhagen, Denmark) in the musculus deltoideus. Both donors were Caucasian adults (designated TT01, a male, and TT02, a female) residing in Denmark and both had been challenged with TT several times before this study. Six days after the vaccination, blood (200 ml) was collected in anticoagulant. The project was approved by the Regional Ethical Review Board in Copenhagen, Denmark, and informed consent was obtained from each donor.
Ab libraries were constructed according to Meijer et al. (10) from plasma blasts isolated on a FACSAria cell-sorting system (BD Biosciences) from freshly drawn blood. Shortly, PBMC were isolated using Lymphoprep (Axis-Shield) and enriched for B cells by a CD19+ MACS procedure. B cells that stained positive for CD19, highly positive for CD38, and intermediately positive for CD45 were single cell-sorted directly into wells of 96-well PCR plates and designated plasma cells. H chain and 
L chain variable region genes were linked and amplified by RT-PCR. Amplified genes were pooled and inserted into a bacterial Fab expression vector and transformed into Escherichia coli. Fabs were expressed and screened for activity against TT by ELISA. All positive clones were sequenced and the VH-V sequences were aligned to group clones according to sequence homology. For each group, the V-D-J usage and location of somatic mutations were determined by alignment with germline sequences using the ImMunoGeneTics sequence directory (www.ebi.ac.uk; http://imgt.cines.fr). The phylogenetic trees were generated using ClustalW (www.ebi.ac.uk).
Construction of deduced germline Fabs
Protein sequences of two Fabs without hypermutations were constructed by aligning mutated Ab sequences from common rearrangements in the two donor repertoires (VH1–69-JH6 V1–27-J3 and VH3–21-JH6 V1–39-J1; where JH is H chain joining region) to germline protein sequences of V and J segments (http://imgt.cines.fr). The more common CDR3 sequence motifs of each H and L chain rearrangement were inserted in each synthetic protein sequence (underlined in Figs. 2 and 3). Each protein sequence was back translated and AscI and XhoI restriction sites were added to the ends of VH sequences. Furthermore, an L chain constant domain was added to each L chain variable region and NheI and NotI restriction sites were added to the ends of L chain genes. Finally, the genes were optimized for expression in E. coli and synthesized (GeneArt). Each synthesized gene was cloned into a vector and plasmid DNA was purified from transformed E. coli (GeneArt). Each gene was released from its vector with the appropriate combination of restriction enzymes, and fragments were subsequently purified by preparative 1% (w/v) agarose gel electrophoresis. L chain genes were subcloned into pJSK301. Subsequently, VH regions were subcloned into the L chain gene-containing vectors. Plasmid DNA was sequenced (AGOWA) and verified. Expression of TT-specific Fab expression was verified by ELISA.
|
|
A genetic phylogeny analysis was conducted to assess the homology of all 206 Abs from donors TT01 and TT02. A phylogenetic cladogram (data not shown) was generated by submitting FASTA formats of variable region amino acid sequences (combined H and L chain sequences) of all Abs to the European Molecular Biology Laboratory-European Bioinformatics Institute online phylogeny analysis program, ClustalW (http://www.ebi.ac.uk/clustalw/) (11), using default settings. The cladogram containing only affinity-measured Abs (see Fig. 5) was generated similarly.
|
According to Behlke et al. (12) the probability of observing exactly d distinct gene rearrangements among a total number of r examined sequences for a fixed number n of distinct gene rearrangements is shown in Equation 1,
 | (1) |
For donor TT01 the maximum likelihood estimate of n was n = 29 for d and r fixed at d = 29 and r = 122, whereas the maximum likelihood estimate for donor TT02 was n = 48 for d = 40 and r = 84. A 95% one-sided confidence of the estimates were calculated by determining the smallest value of n for which the probability of observing d distinct gene rearrangements among r examined sequences was <5%. This returned a value of n = 31 for donor TT01, whereas the 95% confidence for donor TT02 was bound at n = 57.
Surface plasmon resonance (SPR)
Fabs were expressed in culture volumes of 100–300 ml depending on expression levels. Bacteria were grown at 37°C in 2x YT broth supplemented with 100 µg/ml carbenicillin and 0.1% glucose until the A600 value reached 1.3, after which isopropyl-
-D-thiogalactopyranoside was added to a final concentration of 0.1 mM and the culture was continued overnight at 30°C. Bacteria were pelleted by centrifugation and the periplasmic fraction was extracted by resuspending the pellet in 1–3 ml (depending on culture volume) of ice-cold PBS containing 0.94 M NaCl and 0.8 mM EDTA. Fabs were further purified on protein L spin columns according to the manufacturer’s instructions (Pierce). The concentration of Fab was determined by an indirect immunoassay using a purified Fab as standard. SPR analysis was performed using a Biacore 2000 apparatus (Biacore AB). TT was immobilized on a CM5 chip surface using standard amine coupling chemistry to a level resulting in a maximum response unit (RU) value of 100 RU or less. Purified Ab Fab were diluted serially in HBS-EP running-buffer (10 mM HEPES (pH 7.4), 0.15 M NaCl, 3 mM EDTA, 0.005% surfactant P20; Biacore) and passed over the chip at 50 µl/min. All measurements were conducted at 25°C. In most cases, association was measured for 5 min and dissociation for 20 min. However, for Fab with very slow off-rates (below 1 x 10–4 s–1), dissociation was measured for longer periods of time, up to 16 h, depending on the off-rate. Rate constants (kon and koff) and affinity constants (KD) were determined using the BIAevaluation software (Biacore) by the global fitting of 4–6 different concentrations passed over the same sensor surface. Global fits of kon and koff simultaneously gave average 
2 values of 1.8 and 0.8, respectively.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
Ab repertoires were, as reported previously (10), made from plasma cells obtained from blood of two healthy volunteers (donors TT01 and TT02) after a boost with TT. Using the Symplex Technology, which involves high throughput single-cell sorting, single-cell RT-PCR, and bulk cloning of the isolated IgG H chain and L chain variable region gene pairs (VH-V), repertoires were generated with preservation of the cognate H and L chain pairing. These repertoires were subsequently expressed in bacteria and screened for Ag reactivity. This led to the isolation of 122 and 84 unique tetanus-specific VH-V gene products from donor TT01 and TT02, respectively.
Statistical analysis of repertoire sizes
Based on V and J gene segment usage and CDR3 sequence homology, we estimated the numbers of rearrangement events in the isolated Ab repertoires to be 29 and 40 events for donor TT01 and TT02, respectively. Because the number of isolated sequences for each donor is much larger than the number of rearrangements (four and two times larger for TT01 and TT02, respectively), it appears reasonable to make statistical predictions of actual repertoire sizes based on the isolated libraries. The analysis is based on the assumption that no significant skewing of the data set by nonrandom gene segment usage occurred (12). It calculates the statistical number of distinct rearrangements (the statistical/theoretical repertoire size) from the number of examined sequences and the number of rearrangements in the data set. This analysis returned statistical maximum likelihood estimates of the repertoire size of 29 and 48 distinct rearrangement events for donor TT01 and TT02, respectively. Ninety-five percent one-sided confidences were calculated as the smallest number of rearrangements for which the probability of observing the distinct number of gene rearrangements among the number of examined sequences was <5%. These confidences returned values of 31 and 57 rearrangement events for donors TT01 and TT02, respectively.
Distribution of variable gene fragment usage in human anti-TT Ab repertoires
Major structural restrictions of an Ab response can be identified by a genetic analysis of the variable region gene segment usage. Hence, the distributions of V gene segments and J gene segments for the VH and V genes of donor TT01 and TT02 were analyzed and compared with a naive repertoire from peripheral IgM+ B cells (13, 14) (Fig. 1). Both TT repertoires showed extensive diversity in their V gene segment usage. The five most prevalent H chain V gene segment families were found with only minor differences relative to the naive repertoire (Fig. 1A). The H chain J gene segment usage also followed the naive repertoire except for the JH5 gene segments, which were less frequently used (Fig. 1B). The distribution of -chain V gene families followed the naive repertoire dominated by the V1 and V3 families, although the V2 gene family was less prevalent in the TT repertoires (Fig. 1C). Similarly, -chain J gene segment usage showed a similar distribution between the repertoires except for J5 which was less frequent in the TT01 repertoire (Fig. 1D). Thus, the TT repertoires showed extensive diversity without any major biases in V and J gene segment usage.
|
Binding site topology is highly dependent on the length and composition of the CDR3 of the H chain. Hence, restricted Ab responses have been shown to select for CDR3 loops of particular lengths (15, 16, 17). Fig. 1E shows the distributions of the CDR3 lengths of donor TT01 and TT02 in comparison to a randomly selected set of Abs with different specificities (18) and a highly restricted anti-rhesus D Ab repertoire (15). The distribution of the randomly selected Ab repertoire is relatively broad, going from 3 to 27 aa with an average CDR3 length of 14. In marked contrast, the highly restricted rhesus D Ab repertoire shows a narrow distribution ranging from 14 to 30 aa with a prominent peak of 18 aa. The CDR3 length distributions of donor TT01 and TT02 are both broad, ranging from 8 to 26 aa, indicative of an unrestricted response. However, both TT repertoires showed an increased preference for longer CDR3 lengths of 15 or 16 aa, leading to averages of 17 and 16 aa, respectively, which was higher than the unselected repertoire. Thus, although the TT responses were diverse in their requirement for CDR3 lengths, there was a bias toward more extensive contact regions.
Sequence homology among Abs within and between anti-TT repertoires
Structurally restricted Ab responses are known to lead to almost identical Abs from different individuals (15, 16). To investigate whether any of the V gene diversity was shared between the two donors, a multiple alignment was performed on the combined set of 206 VH and V sequences (data not shown). V gene sequences of the two donors were evenly distributed throughout the cladogram. This indicates that both donor repertoires possessed extensive diversity with no dominating preferences for particular V genes in agreement with the previous analyses. However, two clusters of homology contained V gene sequences from both the TT01 and TT02 repertoires. The VH and V sequences were aligned separately for each of the two clusters. Abs belonging to cluster 1 (Fig. 2) all had H chain CDR3 lengths of 15 aa containing the LGGTR consensus motif or single amino acid variants thereof. There was an exclusive usage of the JH6 gene segment and the V gene alleles VH1-69 or VH3-21. The -chain V gene usage was restricted to the alleles V1-27 and V3-20, whereas CDR3 length varied from 9 to 10 aa and -chain J usage involved J1, J2, and J3. In total, the different H and L chain rearrangements made up five different assemblies of Ab germline genes. Competitive binding studies among Abs belonging to cluster 1 by SPR revealed that they all recognized overlapping epitopes (data not shown). Abs using the VH1-69 all originated from the TT01 repertoire and could therefore potentially be of common clonal origin, and their differential usage of three different V genes could be explained by receptor revision (19). However, the Abs appeared to be the result of independent rearrangement events because only few somatic mutations were shared among the VH genes, which contrasts with previous examples of receptor revision of Ab L chains (20), and because differences within CDR3 were mainly located in the H chain V-D and D-J joining regions, which are encoded by random insertion/deletion of nucleic acids and thus typically unique for each clonal rearrangement. An analysis of silent mutations confirmed these observations (not shown). Selection of such highly homologous albeit clonally independent Abs was further demonstrated by the remaining Abs belonging to cluster 1, which were isolated from both the TT01 and the TT02 repertoires. Thus, it appeared that Abs from TT01 belonging to cluster 1 were the result of four independent rearrangement events rather than being the result of receptor revision of a single B cell progenitor. Cluster 2 (Fig. 3) consisted of Abs from the TT01 and TT02 repertoire characterized by having H chain CDR3 lengths of 8 aa containing the (S/T)SG(S/T)L motif and L chain CDR3 lengths of 9 aa containing a central QTYS consensus motif. There was an exclusive usage of the VH3-21 allele together with JH6 paired with a L chain having the V1-39 allele joined to J3. Furthermore, several somatic mutations were shared between the Abs independently of donor origin. This further demonstrated the selection of highly homologous albeit clonally independent Abs. Together, these observations indicated that structural restrictions were indeed shaping parts of the Ab repertoire against TT despite the extensive overall diversity of the TT01 and TT02 repertoires and the vast complexity of the antigenic surface.
Binding characteristics of Abs in germline configuration
To investigate whether the Abs of cluster 1 and 2 originated from germline Abs of particularly high affinity, Fab genes were constructed in their deduced germline configuration and their binding constants were determined by SPR (Table I). Cluster 1 appeared to originate from a germline Ab of fast association and slow dissociation leading to high affinity, whereas cluster 2 originated from an Ab of fast association and fast disassociation leading to a relatively modest affinity. One Ab, 60A02, isolated from TT01, was without any identifiable hypermutations and had a modest on-rate and a slow off-rate leading to high affinity, although not as high as that seen for cluster 1. These observations indicate that Abs of the primary repertoire can be of high affinity and have interactions of relatively high stability. Further, comparison of the binding constants of the hypermutated Abs with their deduced germline counterparts revealed that the positive effects of hypermutations were mainly acting by lowering the off-rates whereas the on-rates were both positively and negatively affected by the hypermutations. Moreover, in both cluster 1 and 2 hypermutations lead to an 100-fold increase in affinity.
|
To address the proposed limits for Ab affinity and kinetics, we determined the binding rate constants of representative sets of Abs from the TT01 and TT02 repertoires. The association rate (kon), dissociation rate (koff), and affinity constant (KD) values of 28 TT01 and 21 TT02 Fabs were determined by SPR (Fig. 4, A–C). Association rates showed a wide range of distributions covering almost 5 orders of magnitude. The medians for the TT01 and TT02 repertoires were of similar magnitude at 1.77 x 105 M–1 s–1 and 1.10 x 105 M–1 s–1, respectively. The dissociation rates of TT01 ranged >2 orders of magnitude with a median value of 1.6 x 10–4 s–1 whereas the TT02 repertoire was more restricted covering, 1.5 orders of magnitude with a median value of 3.6 x 10–4 s–1. Affinity constants showed the widest range covering up to 6 orders of magnitude with median values of 9.3 x 10 –10 M and 3.6 x 10–9 M for the TT01 and TT02 repertoires, respectively. Thus, affinities ranged from micromolar to picomolar values and the median affinity of the TT01 repertoire was 4-fold higher than the median of the TT02 repertoire due to a 2-fold difference in both kon and koff rate medians. To examine whether binding constants were randomly distributed, data sets were plotted as logarithmic normal fractile diagrams (Fig. 4, D–F). All data sets fitted well to linear regressions, with R2 ranging from 0.91 to 0.99 as expected for a log-normal distribution. Hence, no obvious biases were apparent and the data sets seemed evenly distributed around the median as expected for a random distribution.
|
To investigate the relations between genetic homology and affinity, VH and V sequences of Abs with known affinity were combined into one single continuous sequence for each Ab and aligned according to homology. The affinity constants were evenly distributed throughout the cladogram (Fig. 5), indicating no general correlation between affinity and V gene composition. However, the affinity between clonally related Abs varied in the order of 10- to 100-fold, which was significantly less than the 106-fold variation observed for the whole set.
Relationship between number of somatic mutations, affinity, and kinetics
The relationship between somatic amino acid replacements and binding constants was investigated by plotting for each TT repertoire the combined number of replacements in the VH and V genes against kon, koff, and KD (Fig. 6, A–C). The average number of replacements for the full data set was 21 per V gene pair, whereas the average number for donor TT01 and donor TT02 separately were 23 and 19, respectively. The scattered distributions of the data points on all plots indicated that within each repertoire there was no correlation between the number of amino acid replacements and the binding strength. This was also the case among clonally related Abs. Plotting replacement frequencies of VH and V separately did not change the appearance of the data (not shown). Hence, somatic mutations had multiple effects on Ag binding, reflecting the intrinsic properties of random somatic diversification. Regardless, the higher replacement average of TT01 relative to TT02 correlated with the increased average in binding strength, supporting the notion that somatic mutations, together with Ag selection, lead to enhancement of the overall affinity of Ab repertoires.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
The two TT repertoires showed extensive diversity in their V gene family and J gene segment usage for the VH and V genes without any major biases toward certain gene segments. The extent of the diversity was similar to a naive repertoire from peripheral IgM+ B cells (13, 14) and thus matched the extensive structural complexity of the TT surface, which creates a vast number of putative Ab epitopes. The TT responses were also diverse in their requirement for VH-CDR3 lengths, although there was a bias toward a more extensive CDR3 relative to a randomly selected repertoire of different specificities (18). This concurs with a previous report, which showed that secondary response repertoires on average use longer CDR3 than primary repertoires (21), leading to the hypothesis that large Ags may skew the VH repertoire toward longer CDR3 loops by allowing more contact points and, hence, higher affinity for the Ab combining site.
Despite the extensive genetic diversity found within each repertoire, we found two shared clusters of highly homologous V genes. Within each cluster, members were characterized by CDR3 consensus motifs, restricted V and J gene segment usage, and characteristic amino acid replacements. Further, within the TT01 repertoire it appeared that highly homologous albeit clonally unrelated Abs were selected from the naive repertoire. Thus, despite the vast complexity of the antigenic surface of TT there was a common preference for certain V gene configurations. To test whether this was caused by high native affinity of these particular rearrangement events, V gene pairs of each cluster were synthesized in their proposed germline configuration and expressed as functional proteins. The affinity constants (KD) of cluster 1 and 2 germline Ab Fabs were 1.1 x 10 –10 M and 2.7 x 10–7 M, i.e., of high and intermediate affinity, respectively (Table I). Thus, selection of these highly homologous V gene sequences appears to be caused by their relatively high native affinity for TT. Further, we found one Ab without somatic mutations in the TT01 repertoire with a KD of 1.4 x 10–8 M. Although this Ab did not belong to any of the shared clusters, it confirmed the possibility of naive Abs of high affinity from primary repertoires. This raises the question of whether such naive high affinity Abs are due to the evolutionary selection of certain innate V genes matching common pathogens (in this case tetanus toxin) or whether it simply reflects a random occurrence of rare high-affinity B cell clones within the naive repertoire. The latter possibility appears the most likely, because high-affinity Abs without hypermutations have also been observed in human responses against the rhesus D Ag (15), in short term responses in mice toward lysozyme (6, 22), and in mice responding toward a nonpathological viral infection (5).
Selection of high-affinity Abs in vitro from naive phage display libraries of 1010 clones or greater has resulted in Ab affinities in the subnanomolar range (23). Considering that the total number of naive B lymphocytes in the human organism is >1010, it appears reasonable to expect a similar selection of such high affinity Abs from the primary repertoire in vivo. Further considering the occurrence of naive high-affinity Abs a purely statistical event, it follows that structurally complex Ags harboring an exhaustive set of epitopes are more likely to select for early high-affinity Abs relative to less complex Ags harboring only a few epitopes. Such an argument would explain the lack of early high-affinity Abs in the first reports on anti-hapten responses in mice that contrasts the later observations on structurally more complex protein Ags. Alternatively, the absence of early high-affinity Abs in responses against haptens can be explained by the multivalency of the Ag when coupled to the carrier protein, which contrasts the monovalency of the protein Ags mentioned above. Hence, anti-hapten responses would be dominated by avidity effects favoring the selection of low-affinity Abs in the primary response (24).
Based on V gene sequence homology, we have estimated the number of clonal selection events creating the two isolated anti-TT IgG/ Ab repertoires to be 29 and 40 events, respectively. These repertoires can be considered representative of the native Ag-selected repertoires from the two donors for several reasons. Firstly, the isolation of clonally related Abs within a repertoire indicated good sampling of the existing repertoire. Secondly, the isolation of highly homologous albeit clonally unrelated Abs within a repertoire as well as between repertoires further indicated good sampling of the actual repertoires. Third, phage display on combinatorial libraries that was made using total RNA from TT01 blood cells (our unpublished data) gave, in 95% of the cases, VH rearrangements identical to those identified with Symplex. Statistical analysis of the data set of each donor returned 95% confidences at 31 and 57 rearrangement events for the IgG/ Ab repertoires of donor TT01 and TT02, respectively. However, actual repertoire sizes are likely to be somewhat larger than estimated by pure statistics because the statistical analysis is based on the assumption that all clones are equally represented within the repertoire. Thus, further assuming that the V
repertoire, which typically constitutes 40% of a human Ab response, is of equal complexity, the Ag-selected repertoires against TT is estimated to be in the order of 100 clonally unique Abs in addition to hypermutated variants thereof.
Dissociation and association rate constants of biologically active molecules are normally in the range of 100 to 10–5 M–1 s–1 and 10–2 to 10–7 M–1 s–1, respectively. Low affinity receptors are characterized by having slow on-rates in the range of 102 to 104 M–1 s–1 and/or fast off-rates in the range of 100 to 10–2 s–1, whereas high affinity receptors are characterized by having fast on-rates in the range of 105 to 107 M–1 s–1 and/or slow off-rates in the range of 10–3 to 10–5 s–1. Both TT repertoires essentially cover the entire biological range of kon values, indicating no special preference for a particular kon. In contrast, the koff range of both repertoires had a lower limit of 2 x 10–3 s–1, which is well beyond what is generally found for low-affinity ligands. Even though we found examples of unmutated Abs of high intrinsic affinity, it must be assumed that most of the Abs were initially of relatively low affinity (and fast dissociation) as exemplified by the fast dissociation rate of cluster 2 in germline configuration. Further, the ELISA used for the screening and selection of clones for this study were able to identify TT Abs with fast dissociation kinetics from phage display libraries (our unpublished results), verifying that the observed lower limit for dissociation is not a consequence of technical limitations. Thus, it appears that the off-rate is subjected to more stringent requirements for efficient clonal expansion and selection than the corresponding on-rate. Differential effects of the two kinetic parameters on clonal selection have previously been observed (2, 24, 25) and are also the rationale behind the proposed affinity limits for the maturing Ab response (7). Although studies using B and T cell hybridomas have demonstrated that Ab-mediated Ag presentation requires an off-rate of at least 0.5 M s–1 (25), our findings suggest that this limit is moved a further 250-fold to 2 x 10–3 s–1 in order for efficient clonal expansion and selection in a developing Ab response in vivo. Interestingly, this minimum half-life of Ab/Ag interactions of 5.7 min corresponds to the estimated half-life of 8 min for membrane-bound IgG, which offers an explanation for the requirement for high affinity for the Ab-mediated internalization of monovalent and soluble Ag by specific B cells for presentation to T cells at the beginning of a T cell-dependent humoral response (24). Further emphasizing the importance of off-rates on clonal selection, we found that the affinity maturation due to somatic mutations within cluster 1 and 2 mainly affected dissociation with little or even negative effect on association.
The even distribution of affinities and kinetic rate constants around their medians follows a log-normal distribution. This concurs with the random nature of somatic rearrangement and hypermutation events and is in agreement with clonal selection being based solely on ligand binding characteristics. Median values were all close to the proposed theoretical limits, which further supports the existence of an affinity ceiling based on positive selection. Because both repertoires yielded similar median values and showed extensive diversification by somatic mutation as expected for highly matured Ab repertoires, it appears reasonable to make predictions regarding the kinetic and affinity limits for positive selection of B lymphocytes: association rates accumulated at 105 and dissociation rates at 10–4, leading to an affinity ceiling of 10–9 M. These predictions are supported by a previous report finding a similar affinity ceiling of 0.5 x 10–9 M in Ab repertoires from mice responding to a viral infection (5). However, the affinity ceiling does not appear discrete, because evidence for affinity maturation beyond 10 –10 M was observed within cluster 1. Thus, the possibility remains that the ceiling is at an even higher affinity than suggested. One such additional factor that could push the affinity ceiling even further into the lower picomolar range could be interclonal competition for limited amounts of Ag (25).
Both repertoires were extensively mutated, indicative of a highly matured response. Within each repertoire it was not possible to observe any correlation between binding strength and level of amino acid replacements, even among clusters of Abs of identical clonal origin, indicating that the repertoire was saturated in terms of somatic alterations. The only positive correlation between the level of hypermutation and affinity was between the two repertoires. TT01 had a median affinity 4-fold above that of TT02, which correlated to the averages of 23 and 21 aa transitions per V gene pair, respectively. Thus, even though both repertoires were saturated with hypermutations, the overall effect on affinity was still positive.
The observation that for germline Abs of clusters 1 and 2 affinity maturation led to maximally a 100-fold increase in affinity and the observation that Ab affinity was generally restricted to vary 10- to 100-fold among clonally related Abs point to the possibility that each rearrangement carries a limited potential in regard to affinity maturation. This is in agreement with previous observations in which secondary responses of enhanced affinity involved changes in V gene usage (2) and also in agreement with the limitations on affinity maturation imposed by particular amino acid replacements in developing anti-hapten Ab responses in mice (26). It further implies that the large affinity range spanning six logarithmic units, observed for both TT repertoires, is mainly due to differences in native affinity among the Abs selected from the primary repertoire. An extensive range of affinities in the primary repertoire would also explain the observation of early high affinity Abs as discussed above.
We here present a molecular analysis of two human Ab responses against TT. Each repertoire is composed of extensive genetic diversity involving most V gene families and J gene segments. The repertoires appear unrestricted and representative, and the human polyclonal repertoire can thus be estimated to be in the order of 100 clonal selection events. Affinity and association constants are evenly distributed over a relatively large range, whereas dissociation rates cover a more restricted range of relatively slow dissociation. Further, we find evidence that affinity maturation mainly favors reduction in dissociation rates. Hence, the dissociation and association rates have a differential effect on the developing Ab response, with more strict requirements for the dissociation rates. Finally, identification of high-affinity Abs without hypermutations in combination with a relatively modest effect of hypermutations on affinity increases indicates that the primary repertoire has a relatively large affinity span into the subnanomolar range. Taken together, our results have implications for the understanding of the development of human Ab responses and for the understanding of polyclonality in an immunological context.
|
Acknowledgments |
|---|
|
Disclosures |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
|
Footnotes |
|---|
1 Address correspondence and reprint requests to Dr. Peter S. Andersen, Symphogen A/S, Elektrovej Bygning 375, Lyngby, Denmark. E-mail address: psa{at}symphogen.com 
2 Abbreviations used in this paper: VH, H chain variable domain; JH, H chain joining region; SPR, surface plasmon resonance; TT, tetanus toxoid; V, -chain variable domain; J, -chain joining region.
Received for publication March 29, 2007. Accepted for publication July 16, 2007.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
-chain expression: dependence on relatively few variable region genes. Science 229: 566-570. gene repertoire of IgM+ B cells. J. Clin. Invest. 99: 1614-1627. [Medline]
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
