Induction of somatic hypermutation by antigen-specific B cell receptors in the human BL2 cell line

Authors


Abstract

The role of the B cell antigen receptor in the induction of somatic hypermutation is presently unclear. We established stable transfectants of the human BL2 cell line expressing hen-egg lysozyme-specific IgM or IgA and compared their ability to induce somatic hypermutation of the endogenous rearranged heavy-chain gene. We found that IgM and IgA were both able to induce somatic hypermutation in an antigen dose-independent manner. The mutations displayed most of the characteristics of somatic hypermutation in vivo. Notably, some replacements introduced stop codons in the coding region. Our data suggest that class-switched memory B cells may undergo somatic hypermutation. They also suggest that the transmembrane/cytoplamic domains of the class-switched isotypes modulatethe signaling and down-modulation activities of the BCR in an antigen dose-dependent manner.

Abbreviations:
HEL:

Hen-egg lysozyme

CDR:

Complementarity determining region

FR:

Framework

BCR:

B cell antigen receptor

SHM:

Somatic hypermutation

HC:

Heavy chain

LC:

Light chain

1 Introduction

The hallmark of acquired immunity is the generation and selection of somatic antibody mutants essential for affinity maturation and the generation of memory in response to T cell-dependent antigens. The antibody mutants are generated through somatic hypermutation (SHM), which introduces mutations at a high rate (10–4 to 10–3 per base pair per generation) in the rearranged variable (V) regions of Ig heavy-chain (HC) and light-chain genes (LC) 1. SHM takes place in histologically defined substructures of peripheral lymphoid organs called germinal centers 2.

SHM is restricted to a region of ∼2 kb that mainly includes the whole of V(D)J segments. However, large variations in mutation frequency were found in the rearranged segments, with mutational hotspots occurring in the complementarity determining regions (CDR). Analysis of large databases of mutated Ig genes enabled the identification of the RGYW motif (where R = A or G, Y = C or T and W = A or T) as a preferred target for SHM 3.

Although the mechanism of SHM is not fully known, much progress has been made in defining cis- and trans-acting elements involved in the catalysis and regulation of this process. Thus, SHM and transcription are closely linked and require similar elements 48. Furthermore, several enzymes involved in the generation and/or repair of theDNA breaks are being analyzed in order to delineate their precise role in SHM 9, 10. A major advance was the discovery of activation-induced cytidine deaminase (AID) and the demonstration of its requirement for both SHM and class switching 1113, although the target of AID, its cofactors and its exact mechanism of action are still debatable 1416. Recent work strongly suggests that SHM and class switching involve different domains of AID 17, 18.

The description of cell lines permissive to SHM has been a valuable tool for unraveling the mechanisms and the requirements of SHM 19. Specifically, the Burkitt's lymphoma cell line BL2 can be induced to mutate in vitro following cross-linking of some surface receptors 20, 21.

The basic structure of a B cell antigen receptor (BCR) is made up of a membrane-bound Ig (mIg) noncovalently associated with the heterodimer Igα/Igβ, with a stoichiometry of 1:1 2225. Igα and Igβ contain a conserved immunoreceptor tyrosine-based activation motif (ITAM) in their cytoplasmic tails and thus serve as the signaling subunitsof the BCR 26, 27. The cytoplasmic domains of IgM and IgD consist of 3 amino acids, whereas class-switched isotypes have more extended cytoplasmic domains (28 amino acids for IgG and IgE and 14 amino acids for IgA). Despite their association with a common signaling component, IgM, IgD and IgG seem to transmit distinct signals 2831. However, the impact of these differences on SHM is presently unclear.

In this study, we analyzed the ability of Hen-egg lysozyme (HEL)-specific mIgM and mIgA isotypes to potentiate an accumulation of somatic mutations in the endogenous rearranged HC gene of the BL2 cell line. We found that both isotypes were able to induce SHM in an antigen dose-independent manner.

2 Results

2.1 Signal transduction by IgM and IgA

To check that IgM and IgA can both transmit signals in conditions that better mimic the situation in vivo, we made use of the ability of the murine A20 B cell lymphoma to secrete IL-2 into the medium following cross-linking of its BCR 32. The BL2 cell line did not produce IL-2 under the same conditions (not shown). We established stable A20 transfectants expressing either IgM or IgA (μ and α HC constructs depicted in Fig. 1A), which had the same basic structure and bore the same V region specific for HEL but differed in their constant region genes. The HC constructs were co-transfected with a HEL-specificκ LC gene to yield stable transfectants that were screened by FCM to select clones that displayed a matched surface expression of the BCR (Fig. 1B).

HEL-specific IgM- or IgA-expressing transfectants were incubated with increasing concentrations of antigen for 24 h. The supernatants were then tested for IL-2 production by ELISA. We found that at 10 and 100 ng/ml of antigen, IgM was more potent in transmitting the signal than IgA. However, at higher concentrations, both BCR were equally efficient. A similar result was obtained with three independent IgM- and IgA-expressing clones or A20 transfectants expressing HEL-specific IgG3 (not shown). No IL-2 production was detected in unstimulated transfectants (Fig. 1C) or in untransfected A20 cells incubated with the same concentrations of antigen (not shown). Thus, the differential efficiency in signaling by HEL-specific IgM and IgA depends on the antigen concentration.

Figure 1.

 (A) Maps of the μ and α HC constructs. The α HC gene transcription is driven by a PVH promoter, while the μ HC gene is driven by a β-globin promoter (pβG). (B) FCM profiles of IgM- and IgA-expressing transfectants. A20 transfectants were stained with mAb E5.2 anti-idiotype and FITC-conjugated anti-γ1. Controls included staining with FITC-conjugated anti-γ1 alone. (C) Signaling by IgM- and IgA-expressing transfectants. A20 transfectants were incubated with increasing concentrations of HEL for 24 h, and the culture supernatants were then assayed for IL-2 production by ELISA (n=7). (NS: non-stimulated; for 0.01 μg/ml, p=0.014; for 0.1 μg/ml, p=0.038; for 1 μg/ml, p=0.321; for 10 μg/ml, p=0.242; for 100 μg/ml, p=0.128)

2.2 Down-modulation of IgM and IgA

Exposure of the BCR to its specific ligand leads to a down-modulation of the receptor. Since our SHM assay was essentially based on a continuous exposure of the transfectants to antigen, we wished to compare the down-modulation rates of IgM and IgA in conditions close to those used for the SHM assay. The constructs described above were transfected into the BL2 cell line in order to establish stable transfectants expressing IgM and IgA of the same specificity and affinity for HEL (Fig. 2A). Surface expression of the transfected HEL-specific IgMκ or IgAκ BCR had no obvious effect on surface expression of the endogenous IgMλ (not shown).

BL2 transfectants displaying matched surface expression of IgM and IgA were incubated with increasing concentrations of HEL for 6, 18 and 24 h. The residual fluorescence on the cell surface was then monitored by FCM 33. IgM was predominantly down-modulated as compared to IgA, with an especially marked difference at low antigen concentrations (10 ng/ml and 100 ng/ml) (Fig. 2B). To check that this differential down-modulation was an inherent feature of IgM and IgA and not an artifact due to cell line or clonal variations, we repeated the same experiment with independent BL2 and A20 clones and found kinetics similar to those described above (not shown). Thus, upon exposure to a specific antigen, IgM displays a more prolonged down-modulation than IgA, especially at low antigen concentration.

Figure 2.

(A) FCM profiles of IgM- and IgA-expressing transfectants. BL2 transfectants were stained with mAb E5.2 anti-idiotype followed with FITC-conjugated anti-γ1. Controls included staining with FITC-conjugated anti-γ1 alone. (B) Down-modulation of IgM and IgA in BL2 transfectants. IgM- and IgA-expressing transfectants were incubated with increasing concentrations of HEL for the indicated time points and stained with mAb E5.2 anti-idiotype and FITC-conjugated goat anti-γ1 (n=10). (C) Statistical evaluation of the difference between IgM and IgA in the down-modulation assay.

2.3 IgM and IgA are equally potent in inducing SHM at low concentrations of antigen

We then asked if the differences seen between IgM and IgA in the signaling and down-modulation assays would be reflected in their ability to induce SHM in the BL2 cell line. We randomly picked one IgM and one IgA transfectant with matched surface expression and compared their ability to induce SHM in the endogenous rearranged μ gene. The antigen (1 ng/ml HEL) was added to the coculture system at 37°C. At days 4 and 7, the cells were washed and incubated again with 1 ng/ml HEL. At day 10, the culture was stopped, and total RNA from B cells was prepared for an RT-PCR. As a control, we used the above coculture system in the same conditions except that no antigen was added. For each transfectant, 32 independent clones were sequenced. The few mutations found in the controls were considered as the background of the PCR and subtracted from the total number of mutations found in the corresponding assay with HEL. The 32 sequences (corresponding to 13,056 bp sequenced) derived from IgM transfectants had accumulated 13.02×10–4 mutations per base pair, whereas the 32 sequences derived from IgA transfectants accumulated 9.2×10–4 mutations per base pair (Table 1). However, the χ2 test revealed that the difference was not statistically significant (p=0.45). Therefore, at low antigen concentrations, both IgM and IgA are equally efficient in inducing SHM.

Table 1. Somatic mutations induced by IgM and IgA at 1 ng/ml of HELa)
 No. mut./no. seq.No. mut./104 bp
  1. a) Total number of bp corresponding to the total number of sequences is shown in parentheses; the difference is not statistically significant, p=0.45.

IgM17/32 (13,056)13.02
IgA12/32 (13,056)9.2

2.4 Increasing the concentration of antigen does not lead to a significant difference between IgM and IgA in their abilities to potentiate accumulation of mutations

We then sought to detect a difference between IgM and IgA isotypes by varying the antigen concentration. Incubation of the cocultures with 10 ng/ml HEL at 37°C led to a slight but not statistically significant (p=0.62) increase in the number of mutations in IgA-expressing transfectants when compared with their IgM-expressing counterparts. Thus, 32 sequences derived from IgM transfectants accumulated 5.36×10–4 mutations per base pair, whereas the 32 sequences derived from IgA transfectants accumulated 7.66×10–4 mutations per base pair (Table 2).

Increasing the antigen concentration to 1 μg/ml led to the same tendency towards an increase, but still statistically insignificant (p=0.18) in the number of mutations induced by IgA. We found 11.60×10–4 mutations per base pair (19 sequences and 7,752 bp) and 21.17×10–4 mutations per base pair (22 sequences and 8,976 bp) for IgM- and IgA-transfectants, respectively (Table 2). With 10 μg/ml HEL, we found 6.54×10–4 mutations per base pair (30 sequences and 12,240 bp) and 7.09×10–4 mutations per base pair (38 sequences and 15,504 bp) for IgM- and IgA-transfectants, respectively (Table 2) (p>0.99). We note, however, that at 10 μg/ml, cell death was more pronounced (up to 82%) so that we cannot ascertain if the occurrence and accumulation of all the mutations followed normal paths. Thus, contrasting with the results found with the signaling and down-modulation assays, and within the range of our data set, increasing the antigen concentration does not lead to a significant difference between IgM and IgA in inducing SHM.

Figure 3.

Compilation of the mutations found in the VDJ region of the endogenous μ gene. Identical mutations were considered independent if they carried other different mutations. Replacement mutations are shown in upper case letters and silent mutations in lower case letters. AGC/T motif and the mutations leading to stop codons are shown in bold.

Table 2. Somatic mutations induced by IgM and IgA at increasing concentrations of HELa)
 No. mut./no. seq.No. mut./104 bp
  1. a) Total number of base pairs corresponding to the total number of sequences is shown in parentheses.

  2. b) p=0.62; c)p=0.18; d)p=0.99]

10 ng/mlb)
IgM 7/32 (13,056) 5.36
IgA10/32 (13,056) 7.66
1 μg/mlc)
IgM 9/19 (7,752) 11.6 
IgA19/22 (4,896)21.17
10 μg/mld)
IgM 8/30 (12,240) 6.54
IgA11/38 (15,504) 7.09

2.5 Bona fide somatic mutations, including stop codons, induced by IgM and IgA isotypes

A compilation of the mutations induced by IgM and IgA within the coding region of the endogenous μ HC gene is shown in Fig. 3. Consistent with previous findings 20, the mutations were distributed over the entire V region, and no insertion or deletion was found. In contrast to the framework region (FR), the distribution of mutations among CDR did not correlate with the size of these regions. Thus, CDR1 was more mutated than CDR2, which was more mutated than CDR3 (Table 3). Among the 13 AGT/C intrinsic hot spots, 7 were mutated. Three hot spots that do not conform to AGT/C were found, one in CDR1 and two in FR1 (Fig. 3). We considered these to be independent mutations because they occurred in sequences that carried additional mutations. There was no bias towards pyrimidines or purines. In contrast, a clear bias towards transitions over transversions was found (82% and 18%, respectively) (Table 4).

Figure 4.

Genealogical trees derived from four related sequences. Lower case letters indicate the shared mutations. CDR and/or FR domains are shown.

Table 3. Distribution of the mutations induced by IgM and IgA within the variable region domains
 LFR1CDR1FR2CDR2FR3CDR3FR4
IgM218 83 31942
IgA329125 91933
Total547208123875
Table 4. Transitions versus transversions and pyrimidines versus purines in the mutations induced by IgM and IgA
 TransitionsTransversionsPyrimidinesPurines
IgM 51 82534
IgA 65184637
Total116267171
Percentage 82%18%50%50%
Table 5. The R/S ratio within the variable region domains
 RSR/S
L 5 0 
FR117300.57
CDR118 29  
FR2 6 23  
CDR2 8 42  
FR327112.45
CDR3 6 16  
FR4 4 14  
Total91511.78

3 Discussion

Any comparison between a precursor IgM and a class-switched isotype specific for the same antigen during the maturation of a T cell-dependent immune response must accommodate at least two facts: the class-switched isotype has a higher affinity for the antigen because of SHM and positive selection, and class switching provides new transmembrane/cytoplasmic domains conferring novel properties. In this study, we focused on the transmembrane/cytoplasmic domains. We reasoned that if any difference in SHM induction could be detected in our system, it was likely to be due to these domains and not to other factors relating to different affinities or kinetic parameters 3436.

We used two isotypes with the same specificity and affinity for HEL but different constant regions. Stable transfectants of the BL2 cell line expressing either IgM or IgA were established, and we looked at their ability to potentiate an accumulation of somatic mutations in the V region of the endogenous rearranged μ gene. One advantage of this system is the use of somatic mutations in the endogenous gene instead of the transfected constructs as a readout system 37. This avoids the problems associated with the integration site, variable transcription rates, copy number, lack of regulatory sequences and repressive effects of heterochromatin. In addition, BL2 represents a developmental stage that normally mutates in vivo20, which is not the case of several cell lines used in SHM studies.

Using matched BL2 transfectants and increasing concentrations of antigen, we found that both isotypes were able to induce SHM in an antigen dose-independent manner, contrasting with a previous report 37. The mutations that accumulated in the V region of the BL2 μ gene had many of the features associated with SHM in vivo20. Notably, four independent mutations led to stop codons, which contrasts with previous findings where extensive sequencing did not enable detection of non-sense mutations 20. An interesting possibility may be related to what has been shown previously 38, where in vivo deletion of the mature BCR ultimately led to cell death, pointing to the central role of the surface expression of BCR per se for B cell survival 39, 40. In the case of untransfected BL2, surface expression of its IgM may be so vital that any mutation resulting in a stop codon will lead to cell death. This might be the basis of some form of selection suspected to operate during the culture independently of any antigen-driven process 20. In contrast, non-sense mutations could be found in the endogenous gene of our transfectants because surface expression of the transfected BCR could substitute for the endogenous BCR. Of note, this does not apply to another human Burkitt's lymphoma cell line (Ramos), in which IgM-loss variants could accumulate mutations 41, 42. It could be that BL2 and Ramos are not representative of exactly the same developmental stage; notably, whereas Ramos mutates constitutively, BL2 requires stimulation through the BCR.

Despite their dissimilarities in their transmembrane/cytoplasmic domains, the five classes of Ig associate with the Igα/Igβ heterodimer 23, which functions as the transducing unit of the BCR 4345. However, the role of the IgG, IgE and IgA cytoplasmic domains remains elusive. Gene-targeted deletion of the cytoplasmic domains of γ1 and ϵ HC led to a severe deficiency in IgG1 and IgE responses, respectively, and to a severe decrease in BCR affinity for γ1 tail mutant mice 46, 47, but the mechanism underlying such phenotypes has not been elucidated. Other in vitro studies concluded that the cytoplasmic domains have a role in internalization and antigen presentation independent from the Igα/Igβ heterodimer but play no part in signaling 43, 48, 49, but up to now there is no evidence that naked BCR exist in vivo. A recent study of mice expressing HEL-specific transgenes of the same affinity but with different cytoplasmic domains concluded that the γ cytoplasmic tail has a central role in enhancing T cell-driven formation of plasmablasts by selectively increasing cell survival during cell divisions 30. However, as discussed 50, the transgenic approach led to contradictory results with regard allelic exclusion and the capacity of the transgenes with different specificities and cell surface densities to support normal primary differentiation.

In this study we show that, affinity aside, modulation of BCR activity by its transmembrane/cytoplasmic domains depends on the antigen concentration. Thus, the threshold for IgM triggering proved lower than for IgA with regard signaling and down-modulation. Increasing antigen concentration led to comparable kinetics. The biochemical basis for this phenomenon is presently unclear. Another possibility pointed to by the downmodulation experiments relates to the stability of the BCR. Recent work suggests that stimulation of the BCR leads to a desensitization of the receptor, which is still able to bind antigen but fails to transduce signal, and that this desensitization correlates with the dissociation of the Igα/Igβ transducing unit from the mIg at the cell surface 51. Furthermore, due to the transmembrane domain of IgD, Igα/Igβ seems more stably associated with mIgD than with mIgM 52. Whether the heterodimer is more or less tightly associated with class-switched isotypes than with IgM remains to be shown.

The difference in the number of mutations induced by IgM and IgA in our assay does not seem to be statistically significant and therefore doesn't mirror the difference seen with the antigen dose-dependent signaling and down-modulation functions of IgM and IgA. It could be that, in our assay, the mutations accumulate along cell divisions so that a subtle difference between IgM and IgA is not detected. It would be interesting to compare the ability of IgM and IgA to induce early DNA breaks by using the recently described system 21.

Earlier work led to conflicting results as to whether memory B cells could accumulate further mutations 5355. Our finding that IgA can potentiate accumulation of somatic mutations suggests, but does not prove, that class-switched memory B cells may accumulate additional mutations, probably by reentering germinal centers. Centroblasts down-modulate their surface receptors, undergo a massive clonal expansion and activate SHM mechanism 1, 2. One might speculate that the differential down-modulation rates observedwith IgM and IgA may have an impact during this phase, which might be related to the accelerated intracellular targeting of antigen by the heterodimer Igα/Igβ 33 or to differential signaling abilities.

The centrocytes are believed to be the targets of the positive selection for high-affinity BCR. From our results, it might be inferred that for limiting amounts of antigen, the transmembrane/cytoplasmic domains confer no advantage for IgA over IgM. However, the situation might be completely different if we assume an IgA BCR with higher affinity than IgM. The class-switched BCR, which would then out compete IgM for limiting concentrations of antigen, may have an advantage because of better signaling or a higher capacity for antigen presentation . It must be stressed that in our assay, the mutations accumulate in the absence of the complex selection mechanisms operating in the germinal centers. Whether the BCR isotype directly regulates the mutation rate in vivo or a preset mutation program followed by selection exists 56 requires further investigations.

4 Materials and methods

4.1 Constructs and transfectants

The μ HC and κ LC (D1.3) constructs have been described 33. For the α HC construct, a previously described μ HC gene 57 was shortened bySacI digestion followed by the insertion of a NotI linker in the unique remaining SacI site. The μ constant gene was then removed as a NotI-XhoI fragment andreplaced by the murine α constant gene as a NotI-SalI fragment. The latter was made by inserting a NotI linker into the XhoI site upstream of Cα1, while SalI was picked from the pBluescript II KS (Stratagene, La Jolla, Ca). Both μ and α HC constructs encode the membrane forms of the murine μ and α HC, respectively.

Cells (2×107) of the human BL2 cell line (IgMλ) or of the murine A20 lymphoma cell line (IgG2aκ) were cotransfected with either the μ or the α HC construct and the κ LC gene by electroporation (250 V, 960 μF) and selected in 96-well plates. The resistant clones were screened by FCM using mAb E5.2 anti-idiotype (IgG1) 58 and FITC-conjugated goat anti-mouse γ1 HC (Southern Biotechnology, Birmingham, AL).

4.2 Interleukin-2 measurement

Untransfected or transfected A20 cells (5×104) were incubated in triplicate with increasing concentrations of HEL (Sigma, St. Louis, MO) in 200 μl in 96-well plates. After 24 h the plates were centrifuged, and 100 μl of the supernatants were used in an IL-2 ELISA assay according to the supplier's instructions (PharMingen, San Diego, CA).

4.3 Induction of SHM

We used a protocol similar to 20 with the following modifications: the cocultures of BL2 transfectants (500 cells) and irradiated phytohemagglutinin-activated PBMC from a healthy donor were performed in triplicate in the presence of increasing concentrations of HEL (i.e without preincubation on ice or washing). After 4 and 7 days, the B cells were washed and counted, and 500 B cells were seeded per well to which increasing concentrations of HEL were again added. At day 10, the cocultures were stopped and the B cell transfectants collected and pooled for RNA preparation. The protocol included as a control BL2 transfectants cocultured with T cells in the absence of antigen.

4.4 RT-PCR and sequencing

Total RNA was prepared by the Tripure technique (Roche, Mannheim, Germany) according to the supplier's instructions and retrotranscribed by addition of reverse transcriptase (Invitrogen, Groningen, The Netherlands). Primers and PCR conditions were as described 20. The PCR products were eluted and cloned in Topo I (Invitrogen). TG1 bacteria were transformed with the ligations and plated without preculture at 37°C. Recombinant clones were sequenced on both strands by using M13 universal primers. Sequences were aligned using Clustalw software (Institut Pasteur,Paris) focusing on 408 bp of the coding region.

4.5 Down-modulation of the BCR

BL2 or A20 transfectants were cultured in the presence of increasing concentrations of HEL at 37°C for 0, 6, 18 or 24 h. At the end of the time course, the cells were washed with ice-cold 0.01% sodium azide-containing PBS. The cells were stained with mAb E5.2 followed by FITC-conjugated goat anti-γ1, and the decrease in mean fluorescence at the cell surface was estimated by FCM.

4.6 Statistical analysis

Statistical significance was evaluated by χ2 test for the data sets indicated in the results section. A difference between IgM and IgA with p<0.05 was considered to be significant.

Acknowledgements

We are indebted to Michael S. Neuberger for the kind gift of material, to Roberto J. Poljak for mAb E5.2 and to Stéphane Denépoux for the BL2 cell line. Work in our laboratory is funded by the Association pour la Recherche sur le Cancer (Grant N° 4403), the Ligue Nationale Contre le Cancer, the Switch Network and the Conseil Régional du Limousin.

Footnotes

  1. 1

    WILEY-VCH

  2. 2

    WILEY-VCH

  3. 3

    WILEY-VCH

  4. 4

    WILEY-VCH