D H reading frame bias: evolutionary selection, antigen selection or both? Evolutionary selection
Table 1. Comparison of DH reading frame usage in nonproductive VHDHJH joints and DHJH joints DH rf according to VH 1 2 3 10 Nonproductive VHDHJH joints from pre-B cells 6 Nonproductive VHDHJH joints from B cells
DH rf according to JH 1 2 3
Refs
11
10
15
3
7
5
1(3a)
12
10
0
11
9
The process of antibody variable (V) region gene assembly allows for variability at the borders of the Total 16 14 17 20 1(3~') 23 b recombining V, D and J elements, because the recombination break- DHJ H joints from 20(31 c) 3 15b 4,6 points can vary and, in the case of the pre-B and B cells heavy (H) chain, N sequences can be inserted at the borders of D H to the ~lTwojoints in rf2 contain stop codons in the N-region which would prevent DI* protein neighbouring JH and VH elements 1. expression, b15/23 and 10/15 of the joints in rf3 contain stop codonsin DH-the basis of Because of the sequence variability counterselection of rf3 in murine antibodies, qn 11 cases joining involved sequence at the recombination breakpoints homologiesbetweenDHandJH which favoursrfl usage4'6.This was rarelyseenin the nonthe D H elements can be used in any productive VHDHJHjoints from Refs 9 and 10 which mostly carried N sequences at the of the three possible reading frames DH--JHborder. Table 1 summarizes the data from (rfs) in productive VHDHJH joints. (this mechanism may be related to In contrast to the D H elements, VH that mediating allelic exclusion6,8). the analysis of Feeney, Decker et al. andJH segments can only be read in a However, in recent papers by and ourselves. It is apparent that, if Feeney9 and byDecker et al. 1°, evi- DH rfs are correctly assigned, the single rf. Inspection of VHDHJH joints in dence was presented that appears to three sets of data are perfectly comroutine antibodies reveals a striking be in conflict with the above con- patible. (The analysis in Table 1 does overrepresentation of one of the clusion. These authors confirmed the not include the DH dements DQ52 three D H rfsz,3, namely rfl in the classical bias of D H rf usage in pro- and DFl16.2. So far there is no evinomenclature of Ichihara et al. 3 On ductive VHDHJH joints, but claimed dence that these elements can be the basis of sequence analysis of that D H rfs were equally represented expressed in D~ proteins.) We VHDHJ H genes amplified from pre- in nonproductive rearrangements. conclude that the bias of DH rf usage B- and B-cell populations, we have Such a result, while expected for rf3 is indeed established at the stage of recently proposed that this bias of usage, would clearly indicate that DH--JH joining, before antigenic D H rf usage is inherent in the mech- the bias against D H rf2 usage must be selection of antibodies in the soma. anism of the somanc generation of due to selection at the level of proHua Gu antibody diversity, that is, selected ductive VHDHJ H joints, that is celluDaisuke Kitamura through evolution and independent lar selection on the basis of receptor Klaus Rajewsky of antigenic selection in the on- specificity, in accord with what had togeny4. Proof of this hypothesis earlier been the accepted view 2,11. came from subsequent work in How can this conflict of data be Institute for Genetics, University of Cologne, Weyertal 121, D-5000 which DHJ H joints were molecularly explained? Cologne 41, FRG. analysed5,6. This analysis showed The simple answer to this question that the D H rf bias seen in antibodies is that Feeney and Decker et al. is already established at the stage of assign DH rfs in nonproductive References 1 Tonegawa, S. (1983) Nature 302, DH--JH joining, i.e. before antigenic VHDHJ H joints from the context of 575 selection. Three mechanisms are re- DH with V , instead of JH. However, 2 Kaartinen, M. and M~ikel~, O. sponsible for this. First, promotion the bias of D H rf usage is based upon (1985) Immunol. Today 6, 324-327 of rfl usage in the absence of N- the joining of DH to JH in the latter's 3 Ichihara, Y., Hayashida, H., region insertion, through preferen- appropriate rf, that is the one with- Miyazaw, S. and Kurosawa, Y. (1989) tial joining of D H to JH in areas of out stop codons. This is immediately Eur. J. lmmunol. 19, 1847-1854 sequence overlap. Second, counter- apparent from the counterselection 4 Gu, H., F6rster, I. and Rajewsky, K. selection of D H rf3 in productive of DH rf2 through the expression of (1990) EMBO J. 9, 2133-2140 VHDHJH joints on the basis of the D~ protein, which requires an open 5 Meek, K. (1990) Science 250, frequent occurrence of stop codons rf of both D H and JH. By assigning 820-823 6 Gu, H., Kitamura, D. and Rajewsky, in D H elements in this rf. Third, DH rfs in VHDHJH joints from the K. (1991) Cell 65, 47-54 counterselection of rf2 on the basis context of D H with VH, that is by 7 Reth, M. and Air, F.W. (1984) of the expression of a truncated p~ counting from the 5' instead of the 3' Nature 312, 418 chain (DI~ protein 7) on the cell sur- end, stop codons in JH are excluded 8 Retb, M., Ammirati, P., Jackson, S. face once the cell has assembled a D H only in productive, but not in non- and Alt, F.W. (1985) Nature 317, 353-355 element to JH in this particular rf productive joints.
Immunology Today
420
vol 12 No. 11 1991
9 Feeney, A.J. (1990)J. Exp. Med. 172, 1377-1390 10 Decker, D.J., Boyle, N.E., Koziot, J.A. and Klinman, R. (1991) J. Immunol. 146, 350-361 11 Langman, R.E. and Cohn, M. (1987) Moi. Immunol. 24, 675-678
Reply: developmental selection complements evolutionary determination It is generally accepted that D Hencoded termination codons and DIx expression contribute to the overrepresentation of DFL and DSP2 D H gene segments in reading frame (rf) 1 among productively rearranged H chain V genes ~-8. Gu, Kitamura and Rajewsky now propose that, in the presence of N additions, D H rfl preference is entirely dependent on these effects4Q However, an analysis of genomic DNA sequences (Table 1) clearly evidences additional selection for D H rfl during B-cell development3,s,7,9. For example, D H rf3-encoded termination codons only partially account for the greater-than-four-fold preference of
DH rfl over D H rf3 among productive rearrangements (Table 1) since an almost twofold preference of D H rfl over D H rf3 exists, even among sequences that include only D H segments that could not have encoded termination codons. The elegant demonstration by Gu et al. s that cells expressing D~t are disfavored is consistent with the underrepresentation of rf2 in both productive and nonproductive D-J and V-D-J rearrangements. However, as is illustrated by numerous productive and nonproductive rearrangements of VH to D-J H in rf2, the expression of DIx does not preclude further gene segment rearrangement 1°. Thus, DIx expression would often be extinguished by subsequent D-J or VH gene segment rearrangements. Furthermore, a comparison of productive versus nonproductive rearrangements (Table 1) clearly demonstrates that, superimposed on the effect of D> (seen in nonproductive re.arrangements) is an additional preference of almost twofold for D H in rfl versus rf2 among productively rearranged genes. Gu, Kitamura and Rajewsky (above) have also proposed, from an analysis of productive VH--DI4-JH rearrangements and of DFr-JH
Table 1. The relative contribution to D H reading frame usage of termination codons and DIx versus developmental selection a The contribution of termination codons versus developmental selection ratio D H rf according to VH rfl:rf3 1 3 4.3:1 28 Total productive sequences 121 1.9:1 19 Productive sequences excluding 37 termination codons b
The contribution of DIx versus developmental selection D H rf according to JH 1 2 121 32 Productive sequences 33 15 Nonproductive sequences
ratio rfl:rf2 3.8:1 2.2:1 c
~lPublishedgenomicDNA sequences5'7'9 includingonly those that contain N additionsat the DH--JHjoint and DSP2 or DFLDHgenesegmentsidentifiedby at leastfivenucleotides2. Sequenceswere obtained from adult BALB/cspleen and bone marrow cellsand include74 sequences from slg bone marrow 'pre-B cells' that yieldedsimilar ratios when analysed independently,bSequencesincludingonly DH gene segments that do not encode termination codons (DSP2.3,4,6, DFL16.2) and other DSP2 and DFL sequences, wherein all potential termination codons had been excised during recombination. CApparentdifferences betweenthis ratio and that obtainedby Gu et al. (above),are due (1) to our inclusion of additionalsequences9 not analysedby Gu et al.; (2) to our inclusionof sequenceshaving rearrangementsto DFL16.2which Gu et al. excludedbecausetheymightnot haveencoded DI*, and (3) to our exclusion of sequences without N additions at the D~r-JH joint, wherein rfl would have been favoredat the levelof recombination. tmm,~otogy'roday
421
joints, that the preferred usage of DH rfl is, in part, due to sequence overlaps of D H and JH that facilitate rearrangements in the absence of N additions. An extensive analysis of both nonproductive and productive rearrangements that exclude N additions (authors' unpublished observations), confirms this and also supports the expected role for termination codons and Dfx expression. However, as with sequences containing N additions, there is a further selection of D H translated in rfl among productively rearranged H chain V genes. The additional selection for DH rfl, as well as selection against productive rearrangements of certain VH gene segments (for example VH81X), occurs prior to surface immunoglobulin (slg) expression 3'7'9. We have suggested that the recognition of the V region amino acid sequence of nascent H chains (perhaps by 'surrogate L chains') is essential for clonal maturation. By this scenario, cells whose nascent H chains are appropriate for L chain interaction would be selected for clonal expansion (at least six divisions) prior to L chain gene segment rearrangement and slg expression. Debra J. Decker Norman R. Klinman Dept of Immunology, The Scripps Research Institute, 10666 North Torrey Pines Road, La Jolla, CA 92037, USA. References 1 Kaartinen, M. and Mfikelfi,O. (1985) lmmunol. Today 6,324-327 2 lchihara, Y., Hayashida, H., Miyazaw, S. and Kurosawa, Y. (1989) Eur. J. lmmunol. 19, 1847-1854 3 Klinman, N.R. and Decker, D.J. (1990) Immunol. Rev. 115,211-224 4 Gu, H., F6rster, I. and Rajewsky, K. (1990) EMBOJ. 9, 2133-2140 5 Feeney, A.J. (1990)J. Exp. Med. 172, 1377-1390 6 Meek, K. (1990) Science 250, 820-823 7 Decker, D.J., Boyle, N.E., Koziol, J.A. and Klinman, N.R. (199l) J. 1mmunol. 146, 350-361 8 Gu, H., Kitamura, D. and Rajewsky, K. (1991) Cell 65, 47-54 9 Decker, D.J., Boyle, N.E. and Klinman, N.R.J. Immunol. (in press) 10 Reth, M. and Alt, F.W. (1984) Nature 312, 418
vol 12 No. 11 1991
Table 1. Comparison of DH reading frame usage in nonproductive VHDHJH joints and DHJH joints DH rf according to VH 1 2 3 10 Nonproductive VHDHJH joints from pre-B cells 6 Nonproductive VHDHJH joints from B cells
DH rf according to JH 1 2 3
Refs
11
10
15
3
7
5
1(3a)
12
10
0
11
9
The process of antibody variable (V) region gene assembly allows for variability at the borders of the Total 16 14 17 20 1(3~') 23 b recombining V, D and J elements, because the recombination break- DHJ H joints from 20(31 c) 3 15b 4,6 points can vary and, in the case of the pre-B and B cells heavy (H) chain, N sequences can be inserted at the borders of D H to the ~lTwojoints in rf2 contain stop codons in the N-region which would prevent DI* protein neighbouring JH and VH elements 1. expression, b15/23 and 10/15 of the joints in rf3 contain stop codonsin DH-the basis of Because of the sequence variability counterselection of rf3 in murine antibodies, qn 11 cases joining involved sequence at the recombination breakpoints homologiesbetweenDHandJH which favoursrfl usage4'6.This was rarelyseenin the nonthe D H elements can be used in any productive VHDHJHjoints from Refs 9 and 10 which mostly carried N sequences at the of the three possible reading frames DH--JHborder. Table 1 summarizes the data from (rfs) in productive VHDHJH joints. (this mechanism may be related to In contrast to the D H elements, VH that mediating allelic exclusion6,8). the analysis of Feeney, Decker et al. andJH segments can only be read in a However, in recent papers by and ourselves. It is apparent that, if Feeney9 and byDecker et al. 1°, evi- DH rfs are correctly assigned, the single rf. Inspection of VHDHJH joints in dence was presented that appears to three sets of data are perfectly comroutine antibodies reveals a striking be in conflict with the above con- patible. (The analysis in Table 1 does overrepresentation of one of the clusion. These authors confirmed the not include the DH dements DQ52 three D H rfsz,3, namely rfl in the classical bias of D H rf usage in pro- and DFl16.2. So far there is no evinomenclature of Ichihara et al. 3 On ductive VHDHJH joints, but claimed dence that these elements can be the basis of sequence analysis of that D H rfs were equally represented expressed in D~ proteins.) We VHDHJ H genes amplified from pre- in nonproductive rearrangements. conclude that the bias of DH rf usage B- and B-cell populations, we have Such a result, while expected for rf3 is indeed established at the stage of recently proposed that this bias of usage, would clearly indicate that DH--JH joining, before antigenic D H rf usage is inherent in the mech- the bias against D H rf2 usage must be selection of antibodies in the soma. anism of the somanc generation of due to selection at the level of proHua Gu antibody diversity, that is, selected ductive VHDHJ H joints, that is celluDaisuke Kitamura through evolution and independent lar selection on the basis of receptor Klaus Rajewsky of antigenic selection in the on- specificity, in accord with what had togeny4. Proof of this hypothesis earlier been the accepted view 2,11. came from subsequent work in How can this conflict of data be Institute for Genetics, University of Cologne, Weyertal 121, D-5000 which DHJ H joints were molecularly explained? Cologne 41, FRG. analysed5,6. This analysis showed The simple answer to this question that the D H rf bias seen in antibodies is that Feeney and Decker et al. is already established at the stage of assign DH rfs in nonproductive References 1 Tonegawa, S. (1983) Nature 302, DH--JH joining, i.e. before antigenic VHDHJ H joints from the context of 575 selection. Three mechanisms are re- DH with V , instead of JH. However, 2 Kaartinen, M. and M~ikel~, O. sponsible for this. First, promotion the bias of D H rf usage is based upon (1985) Immunol. Today 6, 324-327 of rfl usage in the absence of N- the joining of DH to JH in the latter's 3 Ichihara, Y., Hayashida, H., region insertion, through preferen- appropriate rf, that is the one with- Miyazaw, S. and Kurosawa, Y. (1989) tial joining of D H to JH in areas of out stop codons. This is immediately Eur. J. lmmunol. 19, 1847-1854 sequence overlap. Second, counter- apparent from the counterselection 4 Gu, H., F6rster, I. and Rajewsky, K. selection of D H rf3 in productive of DH rf2 through the expression of (1990) EMBO J. 9, 2133-2140 VHDHJH joints on the basis of the D~ protein, which requires an open 5 Meek, K. (1990) Science 250, frequent occurrence of stop codons rf of both D H and JH. By assigning 820-823 6 Gu, H., Kitamura, D. and Rajewsky, in D H elements in this rf. Third, DH rfs in VHDHJH joints from the K. (1991) Cell 65, 47-54 counterselection of rf2 on the basis context of D H with VH, that is by 7 Reth, M. and Air, F.W. (1984) of the expression of a truncated p~ counting from the 5' instead of the 3' Nature 312, 418 chain (DI~ protein 7) on the cell sur- end, stop codons in JH are excluded 8 Retb, M., Ammirati, P., Jackson, S. face once the cell has assembled a D H only in productive, but not in non- and Alt, F.W. (1985) Nature 317, 353-355 element to JH in this particular rf productive joints.
Immunology Today
420
vol 12 No. 11 1991
9 Feeney, A.J. (1990)J. Exp. Med. 172, 1377-1390 10 Decker, D.J., Boyle, N.E., Koziot, J.A. and Klinman, R. (1991) J. Immunol. 146, 350-361 11 Langman, R.E. and Cohn, M. (1987) Moi. Immunol. 24, 675-678
Reply: developmental selection complements evolutionary determination It is generally accepted that D Hencoded termination codons and DIx expression contribute to the overrepresentation of DFL and DSP2 D H gene segments in reading frame (rf) 1 among productively rearranged H chain V genes ~-8. Gu, Kitamura and Rajewsky now propose that, in the presence of N additions, D H rfl preference is entirely dependent on these effects4Q However, an analysis of genomic DNA sequences (Table 1) clearly evidences additional selection for D H rfl during B-cell development3,s,7,9. For example, D H rf3-encoded termination codons only partially account for the greater-than-four-fold preference of
DH rfl over D H rf3 among productive rearrangements (Table 1) since an almost twofold preference of D H rfl over D H rf3 exists, even among sequences that include only D H segments that could not have encoded termination codons. The elegant demonstration by Gu et al. s that cells expressing D~t are disfavored is consistent with the underrepresentation of rf2 in both productive and nonproductive D-J and V-D-J rearrangements. However, as is illustrated by numerous productive and nonproductive rearrangements of VH to D-J H in rf2, the expression of DIx does not preclude further gene segment rearrangement 1°. Thus, DIx expression would often be extinguished by subsequent D-J or VH gene segment rearrangements. Furthermore, a comparison of productive versus nonproductive rearrangements (Table 1) clearly demonstrates that, superimposed on the effect of D> (seen in nonproductive re.arrangements) is an additional preference of almost twofold for D H in rfl versus rf2 among productively rearranged genes. Gu, Kitamura and Rajewsky (above) have also proposed, from an analysis of productive VH--DI4-JH rearrangements and of DFr-JH
Table 1. The relative contribution to D H reading frame usage of termination codons and DIx versus developmental selection a The contribution of termination codons versus developmental selection ratio D H rf according to VH rfl:rf3 1 3 4.3:1 28 Total productive sequences 121 1.9:1 19 Productive sequences excluding 37 termination codons b
The contribution of DIx versus developmental selection D H rf according to JH 1 2 121 32 Productive sequences 33 15 Nonproductive sequences
ratio rfl:rf2 3.8:1 2.2:1 c
~lPublishedgenomicDNA sequences5'7'9 includingonly those that contain N additionsat the DH--JHjoint and DSP2 or DFLDHgenesegmentsidentifiedby at leastfivenucleotides2. Sequenceswere obtained from adult BALB/cspleen and bone marrow cellsand include74 sequences from slg bone marrow 'pre-B cells' that yieldedsimilar ratios when analysed independently,bSequencesincludingonly DH gene segments that do not encode termination codons (DSP2.3,4,6, DFL16.2) and other DSP2 and DFL sequences, wherein all potential termination codons had been excised during recombination. CApparentdifferences betweenthis ratio and that obtainedby Gu et al. (above),are due (1) to our inclusion of additionalsequences9 not analysedby Gu et al.; (2) to our inclusionof sequenceshaving rearrangementsto DFL16.2which Gu et al. excludedbecausetheymightnot haveencoded DI*, and (3) to our exclusion of sequences without N additions at the D~r-JH joint, wherein rfl would have been favoredat the levelof recombination. tmm,~otogy'roday
421
joints, that the preferred usage of DH rfl is, in part, due to sequence overlaps of D H and JH that facilitate rearrangements in the absence of N additions. An extensive analysis of both nonproductive and productive rearrangements that exclude N additions (authors' unpublished observations), confirms this and also supports the expected role for termination codons and Dfx expression. However, as with sequences containing N additions, there is a further selection of D H translated in rfl among productively rearranged H chain V genes. The additional selection for DH rfl, as well as selection against productive rearrangements of certain VH gene segments (for example VH81X), occurs prior to surface immunoglobulin (slg) expression 3'7'9. We have suggested that the recognition of the V region amino acid sequence of nascent H chains (perhaps by 'surrogate L chains') is essential for clonal maturation. By this scenario, cells whose nascent H chains are appropriate for L chain interaction would be selected for clonal expansion (at least six divisions) prior to L chain gene segment rearrangement and slg expression. Debra J. Decker Norman R. Klinman Dept of Immunology, The Scripps Research Institute, 10666 North Torrey Pines Road, La Jolla, CA 92037, USA. References 1 Kaartinen, M. and Mfikelfi,O. (1985) lmmunol. Today 6,324-327 2 lchihara, Y., Hayashida, H., Miyazaw, S. and Kurosawa, Y. (1989) Eur. J. lmmunol. 19, 1847-1854 3 Klinman, N.R. and Decker, D.J. (1990) Immunol. Rev. 115,211-224 4 Gu, H., F6rster, I. and Rajewsky, K. (1990) EMBOJ. 9, 2133-2140 5 Feeney, A.J. (1990)J. Exp. Med. 172, 1377-1390 6 Meek, K. (1990) Science 250, 820-823 7 Decker, D.J., Boyle, N.E., Koziol, J.A. and Klinman, N.R. (199l) J. 1mmunol. 146, 350-361 8 Gu, H., Kitamura, D. and Rajewsky, K. (1991) Cell 65, 47-54 9 Decker, D.J., Boyle, N.E. and Klinman, N.R.J. Immunol. (in press) 10 Reth, M. and Alt, F.W. (1984) Nature 312, 418
vol 12 No. 11 1991
