Immunoglobulin class switching appears to be regulated by B cell antigen receptor-specific T cell action

Summary

Antigen affinity is commonly viewed as the driving force behind the selection for dominant clonotypes that can occur during the T cell-dependent processes of class switch recombination (CSR) and immune maturation. To test this view, we analyzed the variable gene repertoires of natural monoclonal antibodies to the hapten 2-phenyloxazolone (phOx) as well as those generated after phOx protein carrier-induced thymus-dependent or Ficoll-induced thymus independent antigen stimulation. In contrast to expectations, the extent of IgM heterogeneity proved similar and many IgM from these three populations exhibited similar or even greater affinities than the classic Ox1 clonotype that dominates only after CSR among primary and memory IgG. The population of clones that were selected during CSR exhibited a reduced VH/VL repertoire that was enriched for variable domains with shorter and more uniform CDR-H3 lengths and almost completely stripped of variable domains encoded by the large VH1 family. Thus, contrary to the current paradigm, T-cell dependent clonal selection during CSR appeared to select for VH family and CDR-H3 loop content even when the affinity provided by alternative clones exhibited similar to increased affinity for antigen.

Introduction

Antibodies, the effector molecules of humoral immunity, are generated as a consequence of B lymphocyte stimulation. The nature of the antigen tends to influence which of the various differentiation pathways is activated.

Natural IgM (but not IgG and IgA) Ab are produced in `antigen-free´ animals [1, 2]. These natural Abs (nAbs) tend to be produced by autonomously `activated´ B1 cells, which predominantly reside in the peritoneal and pleural cavities. These cells produce low affinity Ab for a multitude of autoantigens, including intracellular components (e.g. DNA and nuclear and cytoskeleton proteins) as well as self-antigens found in cell membranes and serum (e.g. Ig and cytokines). Some nAb cross-react with bacterial polysaccharides, proteins and lipids, and thus can be induced by exogenous antigenic stimuli. IgM predominates, but IgA and IgG3 are also produced. NAb often express a restricted V region repertoire and show a high degree of idiotypic connectivity [1, 2].

In contrast to the B1 cell, autonomous activation or self-renewal is not a feature of the conventional B2 cell. Antibody production by B2 cells typically reflects exogenous stimulation by either thymus-independent (TI) or thymus-dependent (TD) Ag. Classic TI Ag can be subdivided into mitogenic TI-1 Ag, with LPS as the prototype, and TI-2 Ag that exhibit repetitive epitopes which allow multimeric binding to and direct activation of the B2 cell. TI-2 Ag include bacterial polysaccharides, synthetic complexes of such carbohydrate carriers or polysucrose (Ficoll) with low molecular weight haptens like 2-phenyloxazolone (phOx) [3] and certain polymerized proteins. Despite the T cell-independence of the direct activation of the B cell, the response to TI-2 Ag is influenced in various ways by T cells. For example, CD4⁺ T_H cells promote the switch from IgM to other isotypes [4] while suppressor/regulatory T cells are able to suppress the overall response and may restrict the response to IgM [5].

Classic TD responses require a series of complex cellular interactions between B cells and T cells and other APC. These interactions are mediated by a multitude of cellular co-stimulatory molecules and secreted cytokines. Shortly after immunization, the TD response starts at the edge of the T cell zone of the white pulp when antigen-binding B cells appear in the outer region of the periarteriolar lymphoid sheath (PALS). The Ag-dependent interaction of T and B cells leads to the development and enlargement of extra-follicular antibody-producing foci at the edge of the PALS and formation of germinal centers in follicles. Both pathways contribute to Ab production during the first 12 days of the primary response [6–10].

A number of B2 responses to exogenous Ag are heavily enriched for use of dominant clonotypes and a selected set of VH/VL genes. These restrictions tend to be genetically based, often strain-specific and include responses to protein-coupled haptens such as azophenylarsonate, phOx, (4-hydroxy-3-nitrophenyl)acetyl (NP), phosphorylcholine, DNP or phosphatidylcholine; responses to carbohydrates such as dextran and galactan; and responses to some protein Ags [11].

In order to test the extent to which the restricted repertoire generated by TD responses reflects an initial repertoire bias versus affinity-dependent clonal selection, we compared V region sequences of phOx-reactive natural and TI-2- and TD-induced mAb. The primary TD-induced IgM repertoire proved just as heterogeneous as that of natural and TI-2-induced Ab. A genetic restriction of the TD response was not observed until after class switch recombination (CSR) to IgG. The restriction in VH/VL gene use was accompanied by a focusing of the lengths of the 3^rd CDR of the heavy chain (CDR-H3). The population of CDR-H3 intervals was of heterogeneous length among the IgM antibodies, but of shorter and more homogeneous in IgG. Moreover, the natural, TI-2- and TD-induced IgM repertoires all contained considerable proportions of Ab with similar or even higher affinity than Ox1 idiotypic (Id_Ox1) Ab, which among the primary TD IgG response have highest affinity and first dominate that response. Finally, Ab encoded by VH1 family genes were frequently used by IgM, but almost absent among IgG anti-phOx. Collectively, these findings led us to conclude that clonal regulation of T-dependent CSR is heavily influenced by the sequence content of the BCR, including both V_H family and CDR-H3.

Results

Heterogeneity of natural anti-2-phenyloxazolone antibodies

Because murine pre-immune sera contain considerable amounts of phOx-reactive Ab [12] and the dominant NP^b Id of the TD anti-NP response is also observed among natural Ig [13], we asked how exogenous Ag-induced anti-phOx Ab with the dominant Id_Ox1 relate to their naturally occurring counterparts by comparing the genetic repertoires of both Ab populations. Although nAb-producing B1 cells mainly reside in the peritoneal cavity, anti-phOx activity could not be detected in five separate fusions of peritoneal cells, each with substantial cell growth. However, in 14 out of 27 successful fusions of spleen cells from non-immunized mice, one to seven exclusively IgM-anti-phOx-secreting hybridomas were obtained from single mice (Table 1). Since nAb predominantly react with a variety of evolutionarily conserved autoantigens [14] and cross-react with exogenous Ag, the anti-phOx nAbs were also tested for reactivity with actin, myosin, dsDNA, tubulin and keyhole limpet hemocyanin (KLH) (Supporting Information Table 1; Table 1). Four nAb appeared to be phOx-`specific´ within the panel of Ag tested, while another (nat11) showed an as strong reaction for both phOx and dsDNA. The remaining nAbs showed unique combinations of multiple reactivities.

Table 1.

VH/VL gene combinations of natural IgM anti-phOx antibodies

mAb nata	Fusb	Ag Specc	rel. Affd	VH chain genes						VL chain genes
				Fame	Clf	IGHVg	CDR3-H3h	D	J	Fame	Clf	IGKVg	CDR-L3h	J
01	3	po	n.d.	1	2	690	C_ARSSYYYGSSHADY	FL16.1	2	4/5	1	050	C_QQWSSNPLTF	5
02	7	po	~ 1	1	1	668	C_ASNYFDY	n.f.	2	4/5	1	143	C_QQRSSPRTF	1
03	8	mo	~ 0.1	1	2	643j	C_AGYGNYGFFY	SP2.5,7	2	2	1	097	C_WQGTHFPxTF	2
04	2	pol	n.d.	1	2	627	C_ARVTTVVARGYAMDY	FL16.1	4	4/5	1	157	C_QQWSSKPLTF	5
05	1	n.d.	~ 0.1	1	2	627	C_ARDYGIYYYAMDY	SP2.5,7	4	8	1	189	C_QQSYPWTF	1
06	9	n.d.	~ 0.1	1	2	588	C_ARWVIWYPFYAMDY	SP2.5,7	4	2	1	097	C_WQGTHFPGTF	1
07	9	n.d.	~ 0.1	1	2	510j	C_ARGDYYAMDY	n.f.	4	23	1	168	C_QQSNSWPTTF	5
08	14	n.d.	~ 1	1	3	398	C_ATGYRYDRVWFAY	SP2.11	3	8	1	188	C_QNDYSYPTWTF	1
09	4	n.d.	~ 0.1	1	3	286	C_ARRARATGAMDY	ST4	4	2	1	097	C_WQGTHFPQTF	1
10	6	po	n.d.	2	1	171	C_ARDRGAY	SP2.11	3	19/28	1	175	C_QQYSSYPLTF	2
11	10	bi	~ 14	2	1	171	C_ARDYYGSRAWFAY	FL16.1	3	1	1	115	C_FQGSHVPYTF	2
12	10	mo	~ 1	2	1	171	C_ARDRGAY	SP2.11	3	4/5	2	041	C_HQRSSYPYTF	2
13	13	po	~ 1	2	1	171	C_ARDRGAY	SP2.11	3	4/5	1	082	C_QQRSSYPPMYTF	2
14	12	n.d.	< 0.01	2	1	171	C_ARDRGAY	SP2.11	3	12/13	1	172	C_QHHYGTPYTF	2
15	9	n.d.	~ 0.1	2	1	159	C_AIYYYGKDYFDY	FL16.1	2	23	1	179	C_QNGHSFPYTF	2
16	1	n.d.	~ 0.1	3	1	138	C_ERLRNWYFDV	SP2.3,4,6	1	8	1	192	C_HQYLSSFTF	4
17	9	mo	~ 0.1	3	1	120	C_ARAGAYYRSWFAY	SP2.11	3	10	1	129	C_QQGQSYPFTF	4
18	2	po	~ 1	5	1	192	C_ARHNYGSSYYAMDY	FL16.1	4	12/13	1	172	C_QHHYGTPRTF	1
19	9	po	n.d.	5	1	176	C_ASLYDGYY	SP2.9	2	4/5	1	154	C_HQYHRSPPTF	2
20	6	po	~ 0.3	5	1	176	C_ARGAGGYYAMDY	n.f.	4	12/13	1	172	C_QHYGTPWTF	1
21	12	po	n.d.	5	1	165	C_ARELFFAY	n.f.	3	1	1	122	C_SQSTHVPYTF	2
22	11	mo	~ 1	5	1	160	C_ARPHEG	n.f.	3	4/5	1	050	C_QQWSSNPYTF	2
23	11	po	~ 1	5	1	139	C_ASQTGTLYAMDY	Q52	4	4/5	1	154	C_HQYHRSPPTF	1
24	3	n.d.	n.d	5	1	139	not complete			1	1	115	C_FQGSHVPLTF	5
25	5	n.d.	n.d.	5	1	135	C_ARHYYGSSWFAY	FL16.1	3	19/28	1	201	C_QQYNSYTF	4
26	9	n.d.	~ 1	5	1	131	C_ARHRGLRRPFDY	SP2.2	2	12/13	1	170	C_QHFWGTPPTF	1
27	9	po	~ 0.1	7	1	158	C_ARDKASSYSFRYYAMDY	SP2.6,7	4	8	1	192	C_HQYLSRTF	1
28	6	po	~ 0.1	9	1	156	C_ARNSVRAMDY	SP2.5,7,8	4	4/5	1	163	C_QQWSGYPFTF	4
29	10	po	~ 0.07	5	1	160	C_ARPHVG	Q52	3	n.d.
30	13	po	n.d.	5	1	139	C_AREIYYGSSYHWYFDV	FL16.1	1	n.d.
31	2	po	n.d.	n.d.						4/5	2	166	C_QQFTSSPSTF	2
32	7	po	n.d.	n.d.						19/28	1	201	C_QQYNYPYTF	2

Legend

^aThe annotation of antibodies indicates their generation from non-immunized mice and is followed by a sequential number (GenBank accession no. HQ446528-57 and HQ446558-87).

^bFusions from which antibodies were obtained are indicated by numbers.

^cCross-reactivity of antibodies was tested with varies autologous antigens (see text and Supporting Information Table I) and is indicated: po – poly-, mo – mono-, bi – bispecific.

^dThe relative affinities were measured in a hapten inhibition test in comparison to Id_Ox1 IgM anti-phOx antibody H11.5 (μ,κ). Equal and enhanced affinities are indicated in bold.

^eIndicates the VH or VL gene family.

^fIn the integrative database VBASE2, the genes are classified: for class 1 genes there is genomic and rearranged evidence, for class 2 genes only genomic evidence and for class 3 genes only rearranged evidence.

^gVH and VL genes numbers according to the integrative database VBASE2.

^hThe amino acid sequences of the third hypervariable regions of the heavy and light chains are given in the one-letter code. Amino acids in bold and underlined are derived from D gene-segments while those in italics are generated by N region insertions and P nucleotides.

^jThe VH gene shows several deviations from the indicated VH1 family Class 2 reference gene. Since the entire VH1 locus is not known and the corresponding VL gene is in germline configuration, we assume that this antibody nevertheless has to be regarded as a natural antibody.

n.d. – not determined; n.f. – not found

It is a common view that nAbs bind with low affinity to their respective Ag. Thus we compared the relative affinities of the anti-phOx nAb with that of the primary TD Ag-induced Id_Ox1 mAb H11.5 (μ,κ) (Table 1). As expected, most natural anti-phOx showed much lower affinities. However, eight nAbs displayed an affinity comparable to mAb H11.5, and nat11 actually exhibited a 14-times higher affinity. The nat11 nAb was encoded by VHOx1 (VH171 in the integrative database VBASE2), but it used Vκ115 and it did not contain the Ox1-typical DRG amino acid sequence in CDR3-H3. Moreover, nat11 did not react with our Id_Ox1-specific mAb [15]. These results demonstrated that, contrary to the common view, the natural BALB/c anti-phOx repertoire does contain a number of Ab of equal or even higher affinity than the classic clonotype Ox1, which exhibits the highest affinity in the primary TD antigen-induced anti-phOx response [3]. This is reminiscent of the observation that the affinity of human nAbs can be similar to tetanus vaccine-induced immune antibodies [14].

We obtained complete VH/VL sequences from 28 nAbs. We also obtained either a VH or a VL sequence from two additional nAbs (Table 1). Together, these Ab were encoded by 21 VH and 20 VL genes, and every VH/VL combination proved unique (Table 1). Five of the 28 nAb used the VHOx1 gene VH171. None used the VκOx1 gene Vκ072. Hence, since the Ox1 clonotype was not found among anti-phOx nAbs, it would appear that this clone contributes minimally, if at all, to the anti-phOx nAb repertoire. This would support the view that Ox1 Ab VH171-Vκ072 clones are more likely to be drawn from the conventional B2 B cell subset, as previously suggested in the case of influenza virus [16].

Composition of primary anti-phOx antibodies induced by the TI-2 antigen phOx-Ficoll

Immunization with TD and TI Ag typically leads to an early appearance of activated proliferating B cells within the PALS [17], and both responses are presumed to derive from the same shared B2 repertoire [18]. Because of these similarities and the observation that T cells can have a regulatory influence on TI-2 responses [4], we analyzed the clonal repertoire of 19 TI-2-induced mAb to see if the Id_Ox1 is as dominant as in the TD response, or whether the TI repertoire is similarly heterogeneous as the NP-specific repertoire in C57BL/6 mice [19]. Our TI-2-induced anti-phOx mAb were encoded by 13 VH and 10 VL genes (Table 2). This level of heterogeneity was similar to that observed in the NP-Ficoll response [19]. Compared with Id_Ox1 Ab, 12 (63%) of the 19 non-Id_Ox1 mAb exhibited an equal affinity to phOx. Two additional mAb actually exhibited an enhanced affinity for the antigen. In contrast to the anti-phOx nAb, eight (42%) of the 19 TI-induced anti-phOx mAb were encoded by Vκ072. Of these, only two (11%) were of the Id_Ox1, both with a VH171-encoded heavy chain and a DRG amino acid sequence in CDR-H3. And, a large proportion of these TI-2 anti-phOx IgM exhibited similar or even higher affinities than Id_Ox1 IgM H11.5. Hence, the TI-2 antigen phOx-Ficoll can activate Id_Ox1 B cells, but this anti-phOx response is just as heterogeneous as anti-phOx nAb, with Id_Ox1 Ab representing a minor subset of the TI-2 response (Table 2).

Table 2.

VH/VL gene combinations of primary IgM anti-phOx antibodies induced by immunization with the thymus-independent type 2 antigen phOx-Ficoll

mAb TI7a	rel. Aff.b	VH chain genes						VL chain genes
		Famc	Classd	IGHVe	CDR3-H3f	D	J	Famc	Classd	IGKVe	CDR-L3f	J
01	~ 0.5	1	2	643	C_ARDYGTY	n.f.	2	4/5	1	072	C_QQWSSNPLTF	5
02	~ 1	1	1	585	C_ASYYGHY	n.f.	2	4/5	1	072	C_QQWSSNPPFTF	5
03	~ 0.1	1	2	547	C_ASWLLRKGYFDY	SP2.9	2	1	1	115	C_FQGSHVPRTF	1
04	~ 1	1	1	532	C_AKGVGDY	SP2.6inv	4	4/5	1	050	C_QQWSSNPLTF	5
05	~ 0.5	1	1	532	C_ARSRGDY	n.f.	4	4/5	1	072	C_QQWSSNPLTF	5
06	~ 1	1	3	396	C_ARDYGDY	n.f.	2	2	1	099	C_VQGTHFPLTF	5
07	~ 1	2	1	183	C_ARNSGDY	ST4	4	4/5	1	072	C_QQWSSNPLTF	5
08	~ 1	2	1	171	C_ARDHGAY	SP2.2,9inv	3	9A	1	108	C_LQYASYPPTF	5
09g	~ 1	2	1	171	C_ARDRGLD	ST4	2	4/5	1	072	C_QQWSSNPLTF	5
10	~ 1	2	1	171	C_ARDRGAY	SP2.11	3	4/5	1	072	C_QQWSSNPLTF	5
11	~ 0.5	2	1	171	C_ARDGGAY	SP2.9	3	21	1	210	C_QQSKEVPYTF	2
12	~ 8	2	1	171	C_ARDWADY	Q52	3	9a	1	112	C_LQYASSPPTF	1
13	~ 1	2	1	171	C_ARDYYGSSYETY	FL16.1	3	9b	1	106	C_LQHGESPFTF	4
14	~ 0.5	3	1	120	C_ARGSHYYGYDY	FL16.2	2	23	1	173	C_QQSNSWPLTF	5
15	~ 0.5	5	1	181	C_ARVGRLLRYAMDY	FL16.1,2	4	4/5	1	072	C_QQWSSNPLTF	5
16	~ 0,5	5	1	163	C_ARYDYYAMDY	SP2.11	4	1	1	122	C_SQSTHVPLTF	5
17	~ 3	5	1	139	C_ARDYRYDDPFAY	SP2.11	3	4/5	1	072	C_QQWSSNPLTF	5
18	~ 0.5	7	1	158	C_ARGDYYGSWFAY	FL16.2	3	1	1	115	C_FQGSHVPYTF	2
19	~ 1	2	1	137	C_VRLRGFAY	SP2.11	3	not determined

Legend

^aThe annotation of antibodies indicates their generation after TI-2 immunization and fusion on day 7 followed by a sequential number (GenBank accession no. HQ446588-606 and HQ446607-624).

^bThe relative affinities were measured in a hapten inhibition test in comparison to Id_Ox1 IgM anti-phOx antibody H11.5. Equal and enhanced affinities are indicated in bold.

^cIndicates the VH or VL gene family.

^dIn the integrative database VBASE2, the genes are classified: for class 1 genes there is genomic and rearranged evidence, for class 2 genes only genomic evidence and for class 3 genes only rearranged evidence.

^eVH and VL genes numbers according to the integrative database VBASE2.

^fThe amino acid sequences of the third hypervariable regions of the heavy and light chains are given in the one-letter code. Amino acids in bold and underlined are derived from D gene-segments while those in italics are generated by N region insertions and P nucleotides.

^gThe Id_Ox1 gene combination VH171/Vκ072 is on gray background.

Since direct TI-2-mediated activation of B cells with resulting proliferation and Ab secretion critically depends on clustering of membrane-bound BCR by the multivalent ligand [20], the avidity of this interaction is presumably an important factor for clonal selection in this response. Since it has been shown that memory cells, induced with the TD form of the TI-2 antigen dextran, can be activated with plain dextran [18] we induced high-affinity phOx-specific memory cells by tertiary immunization with phOx-coupled chicken serum albumin (CSA) and eight weeks later re-stimulated the mice with phOx-Ficoll. Hybridoma anti-phOx mAb were analyzed to see if they resembled either those from TD-induced memory cells or those from the TI-2 response. In contrast to the primary TI-2 response, no IgM but 16 mAb of the IgG class were obtained. These IgG mAb exhibited a greatly restricted repertoire of VH and VL genes (Table 3). Eight (50%) of the 16 mAb used VH171 and eleven (69%) of the mAb were encoded by Vκ072. However, only five (31%) of the 16 mAb used the canonical Id_Ox1 VH171/Vκ072 gene combination. The eleven Vκ072-derived V_L domains all included either the H33N (2 mAb) or H33Q (9 mAb) classic early affinity-enhancing mutations; and all of them also showed the Y35F mutation (Supporting Information Fig. 1). Among the V_H domains, five of the eight VH171-using mAb exhibited the well-known mutations S31T (3 mAb) or S31N (2 mAb) [3]. Hence, all these TI-2 re-stimulated IgG anti-phOx mAb exhibited well-known characteristics of TD memory Ab, as reported earlier [3]. This applied also to the appearance of gene combinations of one highly mutated variable gene with the other in either germline or near germline configuration as previously observed in anti-phOx memory responses [3]. We have argued that such gene combinations might be generated by receptor revision [21]. These results suggest that, in the presence of high affinity TD-induced IgG memory cells, a TI-2 immunization appears to stimulate these cells with seemingly no concomitant activation of naïve TI-2-inducible B cells.

Table 3.

VH/VL gene combinations of IgG-anti-phOx antibodies after tertiary TD and a following primary TI immunization

mAb 3D1Ia	Isotype	VH chain genes						VL chain genes
		Famb	Classc	IGHVd	CDR-H3e	D	J	Famb	Classc	IGKVd	CDR-L3e	J
01	γ1,κ	2	1	183	C_ARDGGDY	SP2.9	4	4/5	1	072	C_QQWSSDPLTF	5
02	γ1, κ	2	1	183	C_ARDGGDY	SP2.9	4	4/5	1	072	C_QQWSSNPLTF	5
03	γ1,κ	2	1	175	C_ARDWGDY	Q52	4	4/5	1	072	C_QQWSSNSPLTF	5
04	γ1,κ	2	1	171	C_ARDGGDY	SP2.9	2	12/13	1	174	C_QHFWSPPWTF	1
05	γ2b,κ	2	1	171	C_ARDRGDY	SP2.11	4	1	1	110	C_LQVTHVPWTF	1
06f	γ1,κ	2	1	171	C_ARDGGDY	SP2.9	2	4/5	1	072	C_HQWSGHPPITF	4
07	γ1,κ	2	1	171	C_ARDGGDY	SP2.9	2	4/5	1	072	C_QQWSGHPPITF	4
08	γ1,κ	2	1	171	C_ARDGGDY	SP2.9	2	4/5	1	072	C_QQWSGSPPITF	4
09	γ2b,κ	2	1	171	C_ARDRGDY	SP2.11	4	4/5	1	072	C_QQWTSNPLTF	5
10	γ1,κ	2	1	171	C_ARDGGDY	SP2.9	2	4/5	1	072	C_QQWSGHPPITF	4
11	γ1,κ	2	1	171	C_ATYDGFAY	SP2.9	3	4/5	1	050	C_QQWISNPPMLTF	5
12	γ1,κ	5	1	139	C_ARDGGDY	FL16.1	4	4/5	1	072	C_QQWSSNPLTF	5
13	γ1,κ	5	1	139	C_ARDGGDY	FL16.1	4	4/5	1	072	C_QQWSSNPLTF	5
14	γ1,κ	5	1	139	C_ARDGGDY	FL16.1	4	4/5	1	072	C_QQWSSNPLTF	5
15	γ1,κ	6	1	114	C_ISVSPWFVY	SP2.11inv	3	9A	1	108	C_LQYASYPFTF	4
16	γ1,κ	2	1	186	C_ARDWGDY	Q52	4	not determined

Legend

^aThe annotation of antibodies indicates their generation after 3 TD and a final TI-2 immunization, fusion 3 days later plus a sequential number (GenBank accession no. HQ446694-446709 and HQ446710-446724).

^bIndicates the VH or VL gene family.

^cIn the integrative database VBASE2, the genes are classified: for class 1 genes there is genomic and rearranged evidence, for class 2 genes only genomic evidence and for class 3 genes only rearranged evidence.

^dVH and VL genes numbers according to the integrative database VBASE2.

^eThe amino acid sequences of the third hypervariable regions of the heavy and light chains are given in the one-letter code. Amino acids in bold and underlined are derived from D gene-segments while those in italics are generated by N region insertions and P nucleotides.

^fThe Id_Ox1 gene combination VH171/Vκ072 is on gray background.

Dichotomy of primary TD antigen-induced IgM and IgG anti-phOx antibodies

To compare these results with the primary TD response, in a first fusion we isolated only primary IgM anti-phOx mAb after immunization with phOx-CSA (Table 4). To our surprise, this early primary TD-induced IgM repertoire turned out to be just as heterogeneous as that of the TI-2 response (Table 2). Together with 12 mAb of a second fusion, 17 different VH and 12 different VL genes contributed to this IgM response, a heterogeneity which is comparable to that of the TI-2 response (Table 2). A small percentage of these mAb used the VH171 gene and was of the Ox1 clonotype. Importantly, a quarter of these 32 primary non-Id_Ox1 IgM exhibited an identical, and two (6%) even higher, affinities for phOx than the Id_Ox1 mAb H11.5. Conversely, VH171 usage as well as expression of the Id_Ox1 gene combination increased among IgG mAb which also exhibited a genetic restriction with only 4 VH and 4 VL coding genes (Table 5).

Table 4.

VH/VL gene combinations of early day 7 primary TD-induced IgM anti-phOx antibodies

mAb TD7a	Fub	rel. Aff	VH chain genes						VL chain genes
			Famc	Classd	IGHVe	CDR-H3f	D	J	Famc	Classd	IGKVe	CDR-L3f	J
01	1	~ 1	1	1	702	C_TRLDAFAY	Q52	3	9A	1	108	C_LQYASYPPTF	1
02	1	~ 1	1	1	668	C_ARDGGDY	SP2.9	4	4/5	1	050	C_QQWSSNPLTF	5
03	2	~ 0.2	1	2	627	C_VRYYFDY	n.f.	2	9B	2	121	C_LQYDEFPYTF	2
04	1	< 0.05	1	1	532	C_TNLRLQAY	FL16.2	3	9A	1	108	C_LQYASYPPTF	1
05	1	~ 1	1	1	532	C_ANLRLQAY	FL16.2	3	9A	1	108	C_LQYASYPPTF	1
06	1	~ 0.2	1	1	532	C_AKRRGDY	SP2.3,4,6	4	4/5	1	072	C_QQWSSNPLTF	5
07	1	n.d.	1	3	398	C_AGKYGGAY	SP2.3,4	3	19/28	1	203	C_QQYSSYPYTF	2
08	1	~ 0.05	1	3	398	C_AGKYGGAY	SP2.2,3	3	19/28	1	203	C_QQYSSYPYTF	2
09	2	~ 0.2	1	1	364	C_ARSYDGYWRFDY	SP2.9	2	4/5	1	072	C_QQWSSNPPTF	5
10	1	~ 0.2	1	3	286	C_ARWGYGNAY	ST4	3	4/5	1	078	C_QQYSGYPLTF	1
11	2	~ 8	1	2	034	C_ARPDGNYVAWFAY	SP2.8	3	1	1	115	C_FQGSHVPYTF	2
12	1	~ 0.05	1	3	023	C_AYYYGAY	FL16.1	3	10	1	129	C_QQGQSYPLTF	5
13	2	~ 9.0	2	1	190	C_AGGRSFAY	SP2.11	3	23	1	168	C_QQSNSWPTLTF	5
14	2	~ 0.1	2	1	190	C_AGGRSFAY	SP2.11	3	10	1	139	C_QQYSKLPYTF	2
15g	2	~ 1	2	1	171	C_ARDRGAY	SP2.11	3	4/5	1	072	C_QQWSSNPLTF	5
16	1	~ 0.05	2	1	171	C_ARDRGAYh	ST4	3	4/5	1	072	C_QQWSSNPLTF	5
17	1	~ 1	2	1	171	C_ARDRGAY	SP2.11	3	4/5	1	072	C_QQWSSNPLTF	5
18	1	n.d.	2	1	171	C_ARDRGAY	ST4inv	3	4/5	1	072	C_QQWSSNPLTF	5
19	2	~ 1	2	1	171	C_ARDRGDY	SP2.11	2	4/5	1	072	C_QQWSSNPP	n.d.
20	2	~ 0.2	3	1	128	C_ARATARAVAWFAY	ST4	3	9B	1	121	C_LQYDEFPYTF	2
21	2	~ 0.3	5	1	181	C_ARVGGRGHYFDY	SP2.11	2	23	1	168	C_QQSNSWPRTF	1
22	1	~ 0.2	5	1	163	C_ARNYLYAMDY	SP2.2–2.7	4	23	1	173	C_QQSNSWPYTF	2
23	1	~ 0.3	7	1	158	C_ARGILGGWFAY	Q52	3	2	1	097	C_WQGTHFYTF	2
24	1	~ 1	7	1	158	C_ARGYDYGAWFAY	SP2.2	3	1	1	115	C_FQGSHVPYTF	2
25	?	n.d.	1	1	532	C_ARSDYYGYDEGAWFAY	SP2.3,4,6	3	n.d.
26	2	n.d.	2	1	137	C_VRVAYYGSSW	FL16.1	1	n.d.
27	2	~ 0.2	6	1	114	C_TRPNWDSPWFAY	Q52	3	n.d.
28	1	n.d.	n.d.						4/5	1	072	C_QQWSSNPLTF	5
29	1	n.d.	n.d.						1	1	115	C_FQGSHVPRTF	1
30	1	~ 1	n.d.						9A	1	108	C_LQYASYPPTF	1
31	1	< 0.05	n.d.						9B	1	121	C_LQYDEFPYTF	2
32	1	~ 0.5	n.d.						9A	1	108	C_LQYASYPPTF	1

Legend

^aThe annotation of antibodies indicates their generation after primary TD immunization and fusion on day 7 followed by a sequential number (GenBank accession no. HQ446625-446651 and HQ446659-446687).

^bThe antibodies are derived from 2 fusions. Since fusion 1 was only screened for IgM-anti-phOx antibodies there are no IgG antibodies.

^cIndicates the VH or VL gene family.

^dIn the integrative database VBASE2, the genes are classified: for class 1 genes there is genomic and rearranged evidence, for class 2 genes only genomic evidence and for class 3 genes only rearranged evidence.

^eVH and VL genes numbers according to the integrative database VBASE2.

^fThe amino acid sequences of the third hypervariable regions of the heavy and light chains are given in the one-letter code. Amino acids in bold and underlined are derived from D gene-segments while those in italics are generated by N region insertions and/or P nucleotides.

^gThe Id_Ox1 gene combination VH171/Vκ072 is on gray background.

^hThis antibody showed mutations in VH: 1 in CDR2 and 1 in FR3.

n.d. not determined

Table 5.

VH/VL gene combinations of primary TD-induced day 7 IgG anti-phOx antibodies

mAb TD7a	Fub	VH chain genes						VL chain genes
		Famc	Classd	IGHVe	CDR-H3f	D	J	Famc	Classd	IGKVe	CDR-L3f	J
33	2	2	1	201	C_AKDSGDY	SP2.2	4	4/5	1	072	C_QQWSSTPLTF	5
34	2	2	1	171	C_ARDGGAY	SP2.9	3	4/5	1	157	C_QQWSSNPLTF	5
35g	2	2	1	171	C_ARDRGAY	SP2.11	3	4/5	1	072	C_QQWSSNPLTF	5
36	2	2	1	171	C_ARDRGDY	ST4inv	3	4/5	1	072	C_QQWSSNPLTF	5
37	2	3	1	437	C_ARDYYYGSKLLYYFDV	FL16.1	1	10	1	137	C_QQGNTLPPTF	1
38	2	6	1	114	C_TRPNWDSPWFAY	Q52	3	1	1	115	C_FQGSHVPRTF	1
39	2	2	1	171	C_ARDRGDY	ST4inv	4	not determined

Legend

^aThe annotation of antibodies indicates their generation after TD immunization and fusion on day 7 followed by a sequential number (GenBank accession no. HQ446652-446658 and HQ446688-446693).

^bThe antibodies are derived from the second fusion the IgM antibodies of which are depicted in Table 4.

^cIndicates the VH or VL gene family or subgroup, respectively.

^dIn the integrative database VBASE2, the genes are classified: for class 1 genes there is genomic and rearranged evidence, for class 2 genes only genomic evidence and for class 3 genes only rearranged evidence.

^eVH and VL genes numbers according to the integrative database VBASE2.

^fThe amino acid sequences of the third hypervariable regions of the heavy and light chains are given in the one-letter code. Amino acids in bold and underlined are derived from D gene-segments while those in italics are generated by N region insertions and P nucleotides.

^gThe Id_Ox1 gene combination VH171/Vκ072 is on gray background.

The VH/VL combinations of all these mAb and, for completion of the anti-phOx repertoires of both Ig classes, VH/VL combinations of anti-phOx mAb from a previous investigation [21] are depicted in Tables 6 and 7. The previously reported IgM and IgG mAb include those generated after (a) TI-2 pre-immunization with phOx-Ficoll and TD immunization 1 month later and fusion on day 7 (1°TI>1mo>1°TD day 7), (b) TI-2 pre-immunization and TD immunization 3 months later and fusion day 7 (1°TI>3mo>1°TD day 7), (c) TI-2 pre-immunization, TD immunization 3 months later and fusion on day 14 (1°TI>3mo>1°TD day 14), and (d) TI-2 pre-immunization, TD primary immunization after 3 months followed by a secondary TD immunization 8 weeks later and fusion on day 3 (1°TI>3mo>2°TD). In addition, Table 7 contains VH/VL combinations of TD-induced quaternary IgG anti-phOx mAb (4°TD) the genetic make-up of which is shown in Supporting Information Fig. 2 and Supporting Information Table 2. Importantly and irrespective of the immunization scheme, IgM antibodies of both the natural and the TI-2- and TD-induced repertoires proved to be highly diverse (41 VH; 37 Vκ), including non-Id_Ox1 with equal or even higher affinities than Id_Ox1 mAb (Table 6). In contrast, IgG mAb use a restricted repertoire (14 VH; 12 Vκ) encoded largely by either the Id_Ox1 genes VH171 or Vκ072 or a combination of these genes and a limited number of other gene sets (Table 7). Intriguingly, a large proportion of IgM is encoded by genes of the VH1 family, whereas usage of VH1 gene family members among primary as well as memory IgG was only found in one particular fusion after a 1°TI>3m>2°TD challenge regimen. Thus B cells using VH1 family-encoded IgM appeared to rarely produce class switched anti-phOx IgG (Fig. 1).

Table 6.

Gene usage of IgM antibodies in anti-2-phenyl-oxazolone immune responses of BALB/c mice

Legend

Each filled cell symbolizes an IgM anti-phOx Ab with a particular VH/VL combination. The numbers indicate a particular Ab within a group of Ab the generation of which is described below. Numbers after a slash indicate how many Ab of a particular gene combination were found. Cells of non-Id_Ox1 Ab exhibiting identical or similar affinities as Id_Ox1 Ab are marked with bold rectangles. Non-Id_Ox1 exhibiting higher affinities than Id_Ox1 Ab are marked with a bold upright arrow. The bold dashed line highlights the border between VH1 and VH2 family-encoded genes.

- natural IgM obtained from non-immunized mice.

- primary IgM day 7 Ab induced with phOx-Ficoll (TI-2 antigen)

- primary day 7 TD-(phOx-CSA)-induced Ab

- Ab obtained after initial TI-2 immunization and a primary TD immunization 1 month later and fusion on day 7

- Ab were obtained after an initial TI-2 immunization and a primary TD immunization 3 months later and fusion on day 7

Table 7.

Gene usage of IgG antibodies in anti-2-phenyl-oxazolone immune responses of BALB/c mice

Legend

Each cell marked with a colored ellipse symbolizes an IgG anti-phOx Ab with a particular VH/VL combination. The numbers indicate a particular Ab within a group of Ab the generation of which is described below. Numbers after a slash indicate how many Ab of a particular gene combination were found. The bold dashed line highlights the border between VH1 and VH2 family-encoded genes.

- primary day 7 TD-(phOx-CSA)-induced Ab

- Ab obtained after initial TI-2 immunization and a primary TD immunization 1 month later and fusion on day 7

- Ab were obtained after an initial TI-2 immunization and a primary TD immunization 3 months later and fusion on day 7

- initial TI-2 immunization, 3 months pause, followed by a primary TD immunization and fusion on day 14

- initial TI-2 immunization, 3 months pause, primary TD followed by a secondary TD immunization 8 weeks later and fusion on day 3

- tertiary TD-immunized mice received a final immunization with TI-2 antigen (phOx-Ficoll) 8 weeks after the last TD immunization, fusion on day 3

- mice received a quaternary TD immunization and spleen cells were fused 3 days after the last immunization

Figure 1. Percentages of VH1 gene family-encoded IgM and IgG anti-phOx antibodies.

The following groups of mAb were analyzed: (A) natural IgM obtained from non-immunized mice; (B) primary IgM (on day 7) after induction with phOx-Ficoll (TI-2 antigen); (C) primary TD-(phOx-CSA)-induced mAb obtained on day 7 after induction; (D) mAb obtained after initial TI-2 immunization, followed by a primary TD immunization 1 month later and fusion thereafter on day 7; (E) mAb were obtained after an initial TI-2 immunization, followed by a primary TD immunization 3 months later and fusion thereafter on day 7; (F) IgG mAb obtained after initial TI-2 immunization, followed by a 3 month pause, followed by a primary TD immunization and fusion thereafter on day 14; (G) IgG mAb obtained after initial TI-2 immunization, followed by a primary TD immunization after a 3 month pause, followed by a further secondary TD immunization 8 weeks later and fusion thereafter on day 3; (H) IgG mAb obtained after tertiary TD-immunized mice received a final immunization with TI-2 antigen (phOx-Ficoll) 8 weeks after the last TD immunization, fusion thereafter on day 3; (J) IgG mAb obtained after mice received a quaternary TD immunization and spleen cells were fused 3 days after the last immunization. Black bars show values for IgM mAbs, white bars show values for IgG. n indicates the numbers of IgM/IgG mAb, respectively. For further information see legend to Fig. 2.

Shortening and unification of CDR3-H3 lengths during class switch recombination

In mice [22] and humans [23, 24] it has been observed that Ab containing somatic mutations in VH and VL or memory Ab exhibit shorter CDR-H3 intervals than Ab in germline configuration. This poses the question as to when these Ab with truncated CDR-H3 are selected during the sequential steps of immune maturation. We compared the CDR-H3 lengths of all anti-phOx mAb (Tables 6 and 7, Fig. 2) and found that the CDR-H3 lengths of natural as well as TI-2- and TD-induced IgM are rather heterogeneous, ranging from 15 to 51 nucleotides. In contrast, the CDR-H3 lengths of all the IgG populations, which ranged from primary to quaternary responses, are much more homogeneous and show a dominant short length of 21 nucleotides throughout. The only exception was seen among secondary Ab, half of which showed CDR-H3 lengths of 33–45 nucleotides. However, this appears to be a transient phenomenon, and these Ab appear to be counter selected since they are not retrieved in tertiary and quaternary responses (Fig. 2H and J). Thus, the composition of the CDR-H3 loop appears to influence the selection, survival or proliferation of the anti-phOx B cells that underwent CSR.

Figure 2. Distribution of CDR-H3 nucleotide lengths among anti-phOx antibodies.

CDR-H3 nucleotide numbers of anti-phOx mAb contained in Tab. 6 were counted and assorted according to their length. IgM and IgG mAb are indicated by black and white bars, respectively. The following groups of mAb were analyzed: (A) natural IgM obtained from non-immunized mice (n = 30, 14 spleens); (B) primary IgM obtained 7 days after induction with the TI-2 antigen phOx-Ficoll (n = 19, 3 spleens); (C) primary IgM (n = 27, left) and IgG (n = 7, right) obtained 7 days after induction with TD antigen phOx-CSA (2 spleens); (D) IgM (n = 20, left) and IgG (n = 3, right) were obtained after initial TI-2 immunization and a primary TD immunization 1 month later and fusion thereafter on day 7 (1 spleen); (E) IgM (n = 8, left) and IgG (n = 6, right) were obtained after an initial TI-2 immunization and a primary TD immunization 3 months later and fusion thereafter on day 7 (1 spleen); (F) IgG (n = 17, 1 spleen) were produced after initial TI-2 immunization, followed by a 3 month pause, followed by a primary TD immunization and fusion thereafter on day 14; (G) IgG (n= 34, 1 spleen) were obtained after initial TI-2 immunization, followed by a primary TD immunization after a 3 month pause, followed by a further secondary TD immunization 8 weeks later and fusion thereafter on day 3; (H) IgG (n = 16, 3 spleens) were obtained after tertiary TD-immunized mice received a final immunization with TI-2 Ag phOx-Ficoll 8 weeks after the last TD immunization, fusion thereafter on day 3; (J) IgG (n = 10, 3 spleens) were obtained after mice received a quaternary TD immunization and spleen cells were fused 3 days after the last immunization.

All VH region amino acid sequences of antibodies depicted in Table 6 and 7 were analyzed with the RANKPEP program [25] (available under http://imed.med.ucm.es/Tools/rankpep.html.) for the prediction of peptides which can be bound by I-A^d and I-E^d MHC-II molecules of the BALB/c strain. Strikingly, a large proportion of these antibodies contained especially CDR-H3-related, MHC-II-presentable peptides (data not shown). However, because of the likely involvement of different idiotypically related T cells, it met our expectation that particular related sequences could not be deduced.

Discussion

These data shed new light on the relationship between activated B cells of the primary and memory responses and the role of VH family and CDR-H3 in influencing CSR. Ab from the early phases of the primary immune response are encoded by germline VH and VL genes and their generally low affinity is improved through somatic hypermutation [26, 27]. In transgenic mouse models, even very low affinity clones can be activated during TI-2 and TD immune responses, but competition for Ag favors high affinity clones during the earliest phases of the response [28, 29]. Since the heterogeneity of the early primary Ab repertoire is reduced in the course of the response, it is the common perception that affinity-driven competition is the mechanism that assures that only B-cell clones expressing high affinity BCR will survive, undergo CSR and enter the memory cell pool [30, 31]. However, our finding that a large fraction of non-Id_Ox1 IgM Ab exhibiting identical or even higher affinities than Id_Ox1 Ab (Table 6) are not found in the pool of IgG-secreting hybridomas (Table 7) argues against the view that affinity is the only factor in determining which B cells that respond to initial immunization will successfully transition to long-lived IgG producing memory B cells. If high affinity played the assumed role, then we would have expected that some non-Ox1 clones would have become prevailing contributors to the IgG anti-phOx repertoire, at least in some mice. Since this, however, was not the case, factors other than affinity must be playing a role in determining which B cells preferentially undergo IgM>IgG class switch recombination, or which B cells survive and prosper after CSR.

One possibility might still be that the CSR-related differences of anti-phOx Ab repertoires reflect the activation of different B cell subpopulations, namely nAb-producing B1 cells or B2 cells with the subpopulations of follicular (FO) and marginal zone (MZ) B cells. Thus, are there indications that one of these subpopulations is preferentially involved in the differentiation to Ab-secreting plasma cells, class switching and formation of memory cells? 1.) In one experimental system the primary but not the secondary TD response developed from B cell precursors that expressed high levels of heat-stable antigen (HSA^high) [32], whereas the majority of HSA^low virgin B cells proved responsible for GC formation and generation of memory cells for the secondary response [33]. Moreover, after in vitro stimulation, only the progeny of HSA^low B cells exhibited somatic mutations [34]. 2.) Blocking with mAb of the interaction of co-stimulatory membrane molecules with their respective ligands at the beginning of GC formation reduced the number of GC B cells and impaired GC maturation and affinity-maturation, while levels of low-affinity antibodies remained unaffected [35]. Since memory B cells were also of low affinity and showed reduced rates of somatic mutations, it was concluded that memory B cells are not only generated in the course of the GC route, but also in a GC-independent manner [35]. 3.) While blood-derived antigens quickly activate innate-like splenic MZ B cells to secretion of weakly self-reactive IgM Ab [36] which are responsible for the transport of the antigen in the form of immune-complexes to the follicles, isotype-switched antibodies are supposed to be mainly derived from FO B cells [9]. Hence, it seems possible that early MZ B cell-derived IgM anti-phOx Ab are eliminated before CSR because of a weak auto-reactivity. A somewhat contradictory finding to this view is the observation that MZ B cell-derived antibodies exhibit shorter CDR-H3 [37–39]. However, while the largest difference of CDR-H3 lengths with 8 bp was observed in MZ B cell-derived Ab that lack N region nucleotides [39], the average difference of 1–2 bp seems to be hardly comparable with the CSR-related reduction of CDR-H3 lengths in the present investigation. Moreover, since all of our 29 nAbs contained N nucleotides and only 1 N nucleotide negative mAb was found in each case among 19 phOx-Ficoll-induced primary IgM (TI7-16) and 27 IgM from the primary TD response (TD7-01), it seems rather unlikely that innate-like MZ B cells contributed to an important extent to our IgM anti-phOx repertoire. 4.) In addition, MZ B cells comprise heterogeneous populations which not only participate in the early but also in late antibody formation and generation of memory cells, and MZ and FO B cells exhibited similar VH gene repertoires and somatic mutations [40]. Analogous results were obtained in the murine Ab response of MZ and FO B cells to virus-like particles, except that somatic mutations were reduced in MZ as compared to FO B cells [41]. Nevertheless, a functional physiological difference between MZ and FO B cell repertoires is indicated by the finding that both populations differed in the expression of 181 genes [42]. Hence, a consistent view that could attribute the CSR-related change of the anti-phOx repertoire to a particular B cell subpopulation seems not to be available.

In contrast, the difference in CDR-H3 lengths between mutated and non-mutated Ab found in overall repertoire analyses could be as high as 10 nucleotides [22–24]. This value is better comparable to the respective range of CDR-H3 lengths in the anti-phOx response. However, it still might be possible that the CSR-association of this effect is unique for the anti-phOx response. We therefore re-examined, as another example, the CDR-H3 lengths of hapten NP-reactive IgM and IgG mAb. These mAb have been described in previous publications by Bothwell and co-workers. The IgM mAb (Supporting Information Table 3) are representative of a TI-2 response induced with NP-Ficoll [19]. They use a diverse repertoire of VH genes and D segments and their CDR-H3 amino acid sequences are encoded by 18–42 nucleotides. A comparable set of IgM mAb induced with the TD Ag NP-coupled chicken gamma globulin (NP-CGG) could not be found in the literature. Primary TD-induced IgG [43], however, are encoded by very few VH genes and use the DFL16.1 segment almost exclusively (Supporting Information Table 4). In CDR-H3, a coding length of 33 nucleotides dominated. Shorter sequences could not be found, but longer ones were still observed. However, in immune-maturated secondary TD-induced anti-NP IgG [44] a further restriction of VH gene and D segment usage as well as CDR-H3 lengths was noticed (Supporting Information Table 5). Strikingly, all IgG anti-NP show a common mostly DFL16.1-derived stretch of tyrosine-tyrosine-glycine (YYG) in the middle of CDR-H3. We regard this as an indication that it is the CDR-H3 sequence that is important for the observed clonal restrictions and not the CDR-H3 length as such. These results from the anti-NP response would suggest that the CSR-associated focusing of CDR-H3 lengths is not restricted to the anti-phOx response, and could be a more general phenomenon.

It has been proposed that a T-independent phase generally precedes the cognate MHC-mediated T-B collaboration during TD responses [7]. According to this `two-phase model of B cell activation´, B cells presumably of the extrafollicular foci are initially expanded and activated to early IgM secretion which, after complexation with antigen, assists strong primary and secondary Ab responses via binding to complement receptors CR2/CR1 on B cells. However, neither the two pathways of extrafollicular and GC-mediated generation of Ab-forming plasma cells and memory cells nor the proposal of a T cell-independent step of induction of TD Ag-induced IgM offer an explanation for the difference of IgM and IgG repertoires. Since an affinity-based selection is also not fully consistent with the data, we presume that another mechanism is likely responsible for the restricted IgG repertoire.

CSR induces a replacement of the constant domains while the variable domains of the initial IgM and the resulting IgG are retained. Therefore, a BCR-related positive clonal selection driven through improved affinity or negative selection as the consequence of new auto-reactive specificities induced by somatic hypermutation [45] would seem to be unlikely explanations for the massive restriction in the repertoire that we have observed in the initial CSR event. However, other experimental findings suggest that T-B cell collaboration during TD responses is not solely mediated by interaction of the TH cell receptor with MHC-II-presented antigenic peptides on B cells. After endocytosis, B cells not only process the bound protein Ag but, at the same time, their intrinsic BCR is also degraded and BCR-derived peptides are presented in a MHC-II bound form on the cell membrane and these peptides can also be recognized by T cells (reviewed in [46]). Processing and presentation of peptides of the intrinsic BCR is strongly enhanced by B cell activation [47]. While presentation of prevalent BCR peptides derived from constant and germline-encoded variable regions induces T cell tolerance [48, 49], T cells specific for rare non-germline peptides remain fully reactive. Such peptides are clonotypic/idiotypic markers (idiotopes) which are somatically generated either during V(D)J recombination, especially in CDR-H3, or by somatic hypermutation in the course of TD immune responses.

It has been argued that idiotope-mediated T-B cell interactions can take place without antigenic stimulation and participate in induction of B cell lymphomas and certain autoimmune diseases [50]. In a transgenic experimental system, idiotypic T-B collaboration provoked a switch to IgG autoantibodies of various specificities [51]. Moreover, application of soluble Id to anti-Id expressing B lymphoma cells directly induced signal transduction as revealed by tyrosine phosphorylation of a variety of intracellular proteins. In vitro experiments demonstrated that idiotope-presentation of this exogenous Id to CD4 T cells is very efficient and that the Id-specifically activated T cells in turn regulate the B lymphoma cells [52]. Hence, we are inclined to assume that endocytosis of the antigen-BCR complex during normal TD immune responses leads to presentation of an array of both antigenic as well as idiotypic BCR peptides, some derived from the sequence of the V_H gene segment itself and others from CDR-H3. Through such a double- or multi-specificity, antigen-specific and idiotope-specific T cells could both be involved in CSR. Depending on the T cell repertoire, particular B cell clones could receive mono-, bi- or even multi-specific T cell help or else this help could be suppressed by the action of regulatory T cells (Fig. 3). Support for this latter possibility in particular comes from studies of the TI-2 response to α (1–3) dextran of BALB/c mice (reviewed in [9, 46]). In this response, class switching of the dominating J558 IgM Id is inhibited by the prototypic suppressive regulatory T cell clone 178–4 which has an exquisite specificity for the CDR-H3 peptide of J558 [5, 53].

Figure 3. Possible involvement of CDR-H3 specific T cells in thymus-dependent antigen-induced class switch recombination.

After endocytosis of the BCR-bound Ag, enzymatic degradation not only leads to fragmentation of the Ag but also of the BCR itself. This leads to MHC class II-mediated presentation of antigenic as well as various BCR peptides of constant and variable regions. Peptides of CDR-H3 are clonotypic / idiotypic markers which can be perceived by functionally active anti-idiotypic T cells. Conceivably, such T cells can either promote or impede class switch recombination. Both types of CDR-H3-specific T cells are probably subjected to the influence of regulatory T cells.

If MHC-II-mediated recognition of BCR-derived Id is decisive or even just involved in clone-specific and thus Ag affinity-independent switching to IgG, then this offers one possible simplifying and unifying explanation for the observed repertoire restriction, for the almost complete absence of VH1 family genes among IgG anti-phOx mAb (Table 7), and for the dramatic shortening and concomitant approximation of CDR-H3 during CSR (Fig. 2). In addition, even if it remains possible that the shortening of CDR-H3 in particular responses conveys a unique and preferential binding benefit to the antigen over and above affinity, only the excellent T cell-specificity for particular CDR-H3 peptides could explain the frequently demonstrated recognition and induction of T cell responses to viral peptides artificially grafted into CDR-H3 of autologous Ig [54].

In conclusion, the data presented show that, in a normal and non-transgenic system, the restrictions in the repertoire observed after CSR do not depend solely or even primarily on either an initial repertoire bias or on the affinity of activated early primary IgM. The data strongly support a T cell-dependent clonotypic regulation of immune maturation, apparently reflecting T cell-specificity for BCR-derived peptides, including those of CDR-H3. A test of this hypothesis through examination of mice that differ only in their representation of the CDR-H3 repertoire is now possible [55–57].

Materials and methods

Animals

Female BALB/c (H-2^d) mice (Harlan-Winkelmann; Borchen, Germany) were reared under clean conventional conditions in the animal house of the University. Mice were maintained in a 12-h light-dark cycle under standard conditions and were provided with food and water ad libitum. Experimental handling of mice and their care were performed in accordance with relevant institutional and national guidelines and regulations. They were approved by the `Ministry of Agriculture, Environment and Rural Areas´ of the local government of Schleswig-Holstein, permission no. V 312-72241.121-3 (35-3/06).

Antigens

BALB/c mice were immunized either with the type 2 thymus-independent (TI-2) Ag phOx-Ficoll (molar ratio ~ 60) or with the TD Ag phOx-CSA (molar ratio ~ 11) [21]. Natural anti-phOx mAb were tested for reactivity with evolutionary-conserved molecules (all obtained from Sigma-Aldrich, Taufkirchen; Germany): actin from bovine muscle (no. A3653), myosin from rabbit muscle (no. 70045), dsDNA from calf thymus (no. D1501), tubulin from bovine brain (no. D4925) and keyhole limpet hemocyanin (KLH) from Megathura crenulata (no. H8283).

Immunizations and production of monoclonal antibodies

BALB/c mice (3–4 months old) received primary i.p. immunizations with 25 µg of phOx-Ficoll or 80 µg of phOx-CSA adsorbed to Al(OH)₃ and the spleen cells were fused with the non-secretor Ag8.653 myeloma cells by the conventional PEG-mediated hybridization technique after 4 and 7 days, respectively. The natural anti-phOx repertoire of BALB/c mice was analyzed from mAb which were produced from non-immunized animals. These mAb were selected for reactivity with phOx-BSA, but negativity or a much weaker reaction (100-times lower titers) with BSA alone.

Determination of relative affinities of anti-phOx antibodies

As previously described [21], the relative affinities of our antibodies were determined with a hapten-inhibition test in comparison to two prototypic Ox1-idiotypic mAb, namely H11.5 (μ,κ) [58] for IgM and NQ2/16.2 (γ, κ) [59] for IgG antibodies. Briefly, the binding of comparable amounts of anti-phOx Ab to surface-bound phOx-BSA was inhibited with graded concentrations of soluble phOx-caproic acid and those values giving 50% inhibition were taken as relative affinity measures. An affinity factor was generated as the quotient of the relative affinity of H11.5 (μ,κ) for IgM and NQ2/16.2 (γ, κ) for IgG mAb divided by that of a particular Ab. Hence, Ab with an affinity factor of >l exhibited a higher while those with an affinity factor of <l were of lower affinity than NQ2/16.2. The validity of our measurements is indicated by several findings. (i) The present and previous measurements of relative affinities of early and late primary [15, 21] as well as of secondary anti-phOx antibodies (unpublished data) correspond to other published anti-phOx mAb [3]. (ii) Although Id_Ox1 mAb H11.5 [15] does not perfectly match the VH/VL sequences of Ox1-idiotypic mAb (JH4 instead of JH3 and aspartic acid instead of alanine in position 98 of the heavy chain), all perfectly matching Id_Ox1 IgM of the present investigation gave identical relative affinities as H11.5. (iii) The affinity of NQ2/16.2 was also checked with fluorescence quenching and gave an almost identical value as obtained by Foote and Milstein [60]. Hence, our determinations of relative affinities are comparable to relevant published data.

Sequencing of antibody V region genes

Sequence analysis of mAb V regions was performed as described [61] with some modifications. Briefly, total RNA of anti-phOx Ab secreting hybrid cell lines was isolated with TRIZOL^® (GIBCO-BRL, Eggenstein, Germany) and transcribed into cDNA with SuperScript™ II RNAse H reverse transcriptase (GIBCO-BRL, Eggenstein, Germany) using pd(N)6 random and pd(T)_12–18 primers. The V_H and V_L mRNA sequences were first amplified by PCR using 10 primer sets for each of the V_H-regions and 7 primer sets for each of the V_L-regions and two forward primers specific for the 3´-end for the first domain of the V_H and V_L constant regions, respectively. Then, a second semi-nested amplification at 3’-end was performed with the relevant primers coupled with M13 oligonucleotides which were afterwards used for the sequencing reaction (MWG Biotech; Ebersberg, Germany). V region sequences were analyzed with the integrative database VBASE2 (http://www.vbase2.org/) [62].

Accession numbers

GenBank accession numbers for all Ab are indicated in the table legends of each group of antibodies.

Abbreviations

phOx

2-phenyl-oxazolone

CSR

class switch recombination

nAb

natural antibody

CSA

chicken serum albumin

TD

thymus-dependent

TI

thymus-independent

NP

(4-hydroxy-3-nitrophenyl)acetyl