Download Flanking Mutated A and T Nucleotides Reverse Complementarity of

This information is current as
of June 16, 2017.
Hypermutation at A-T Base Pairs: The A
Nucleotide Replacement Spectrum Is Affected
by Adjacent Nucleotides and There Is No
Reverse Complementarity of Sequences
Flanking Mutated A and T Nucleotides
Jo Spencer and Deborah K. Dunn-Walters
J Immunol 2005; 175:5170-5177; ;
doi: 10.4049/jimmunol.175.8.5170
http://www.jimmunol.org/content/175/8/5170
Subscription
Permissions
Email Alerts
This article cites 33 articles, 15 of which you can access for free at:
http://www.jimmunol.org/content/175/8/5170.full#ref-list-1
Information about subscribing to The Journal of Immunology is online at:
http://jimmunol.org/subscription
Submit copyright permission requests at:
http://www.aai.org/About/Publications/JI/copyright.html
Receive free email-alerts when new articles cite this article. Sign up at:
http://jimmunol.org/alerts
The Journal of Immunology is published twice each month by
The American Association of Immunologists, Inc.,
1451 Rockville Pike, Suite 650, Rockville, MD 20852
Copyright © 2005 by The American Association of
Immunologists All rights reserved.
Print ISSN: 0022-1767 Online ISSN: 1550-6606.
Downloaded from http://www.jimmunol.org/ by guest on June 16, 2017
References
The Journal of Immunology
Hypermutation at A-T Base Pairs: The A Nucleotide
Replacement Spectrum Is Affected by Adjacent Nucleotides
and There Is No Reverse Complementarity of Sequences
Flanking Mutated A and T Nucleotides1,2
Jo Spencer and Deborah K. Dunn-Walters3
T
he generation of a diverse Ig repertoire is initiated at the
IgH locus by recombination of V, D, and J gene segments.
If recombination is unsuccessful on the first allele of IGH,
rearrangement at the second allele is initiated. If this is successful,
V and J segments rearrange at the Ig L chain loci on each allele in
turn, until a successful rearrangement is achieved. As a consequence of this process, some B cells carry out-of-frame Ig gene
rearrangements. A second wave of diversification of IG genes is
generated by somatic hypermutation (SHM)4 in the germinal centres of B cell follicles. SHM is characterized by predominantly
single base substitutions and results in a characteristic spectrum of
mutations (1–3). Functional IG genes diversified by SHM are seDepartment of Immunobiology, King’s College London School of Medicine at Guy’s,
King’s and St. Thomas’ Hospitals, Guy’s Campus, London, United Kingdom
Received for publication April 14, 2005. Accepted for publication August 11, 2005.
The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance
with 18 U.S.C. Section 1734 solely to indicate this fact.
1
This work was supported by the Wellcome Trust (to J.S.) and the Medical Research
Council (to D.K.D.-W.).
2
The sequence(s) presented in this article has been submitted to GenBank under
accession number(s). DNA sequences are available in GenBank (具www.ncbi.nlm.
nih.gov/Genbank/index.html典) under the following accession numbers: AJ493283-88,
AJ493291, AJ508940-45, AJ508947-49, AJ535508-11, AJ535514, AJ535516,
AJ576317-19, X87032, Y13167, Y13172, Y16646-49, Z738020-21, Z80389,
Z80708, Z93132, Z93134, Z93138, Z93141, Z93153-54, Z93156, Z93158-59,
Z73863, Z93213-14, Z93216, Y13167-68, Z73839, Z73858, Z73860, Z80570-71,
Z80760, Z93198-99, Z93204, Z80673, X87013, X87064, X87075, X97784, Z80753,
X87035, X87082, Z80396, Z803564, and Z803719.
3
Address correspondence and reprint requests to Dr. Deborah K. Dunn-Walters, Department of Immunobiology School of Medicine at Guy’s, King’s and St. Thomas’
Hospitals, 2nd Floor New Guy’s House, Guy’s Hospital, London SE1 9RT, U.K.
E-mail address: [email protected]
4
Abbreviations used in this paper: SHM, somatic hypermutation; AID, activationinduced cytidine deaminase; BER, base excision repair; MMR, mismatch repair;
UNG, uracil N-glycosylase.
Copyright © 2005 by The American Association of Immunologists, Inc.
lected based on the affinity of the newly encoded Ig for Ag. Thus
the mutations observed in the functional genes are a result of both
SHM and selection. Any out-of-frame IG gene rearrangements
within the same cell also undergo SHM but, because these genes
are not expressed, the distribution of mutations is unaffected by the
selection process. These genes are therefore of use in the study of
the mechanism of hypermutation.
Analysis of the sequence context of mutations in out-of-frame Ig
genes has proved a powerful tool enabling several accurate deductions concerning the mechanism of SHM. Differences in the characteristics of sequences flanking mutations from G and C nucleotides and those flanking A and T nucleotides implied separate
mechanisms for G/C mutations and A/T mutations (4). Moreover,
the remarkable resemblance between the flanking sequence surrounding mutations from G and the reverse complement of flanking sequence surrounding mutation from C lead us to suggest that
either G or C nucleotides were likely to be mutated and not both
(5). Both of these predictions have now been confirmed because
considerable progress in the understanding of the SHM process has
been made. The current, broadly accepted, hypothesis is that SHM
is a two-step process (6). The initiating event is via the action of
activation-induced cytidine deaminase (AID), which is both necessary and sufficient for SHM, class switching and gene conversion (7). AID can directly deaminate C nucleotides in DNA, targeting deamination events to C nucleotides in WRC motifs (8, 9).
This results in the in vivo observations of the RGYW/WRCY hot
spot motifs for mutations from G and C. The second step of SHM
involves error prone repair enzymes and generates further mutations, most notably those from A and T.
Mismatched U-G base pairs caused by cytosine deamination can
be repaired either by base excision repair (BER) or by mismatch
repair (MMR). Both are excision repair processes that require resynthesis of excised nucleotides by polymerases— especially for
0022-1767/05/$02.00
Downloaded from http://www.jimmunol.org/ by guest on June 16, 2017
Hypermutation is thought to be a two-phase process. The first phase is via the action of activation-induced cytidine deaminase
(AID), which deaminates C nucleotides in WRC motifs. This results in the RGYW/WRCY hot spot motifs for mutation from G
and C observed in vivo. The resemblance between the hot spot for C mutations and the reverse complement of that for G mutations
implies a process acting equally on both strands of DNA. The second phase of hypermutation generates mutations from A and T
and exhibits strand bias, with more mutations from A than T. Although this does not concur with the idea of one mechanism acting
equally on both strands, it has been suggested that the AT mutator also has a reversible motif; WA/TW. We show here that the
motifs surrounding the different substitutions from A vary significantly; there is no single targeting motif for all A mutations.
Sequence preferences associated with mutations from A more likely reflect an influence of adjacent nucleotides over what the A
mutates “to.” This influence tends toward “like” replacements: Purines (A or G) in the 5ⴕ position bias toward replacement by
another purine (G), whereas replacement with pyrimidines (C or T) is more likely if the preceding base is also a pyrimidine. There
is no reverse complementarity in these observations, in that similar influences of nucleotides adjacent to T are not seen. Hence,
WA and TW should not be considered as reverse complement hot spot motifs for A and T mutations. The Journal of Immunology,
2005, 175: 5170 –5177.
The Journal of Immunology
a) 20
15
10
5
0
A-C T-G
A-G T-C
A-T T-A
C-A G-T
C-GG-C
C-T G-A
A-C T-G
A-G T-C
A-T T-A
C-A G-T
C-G G-C
C-T G-A
b) 20
15
10
5
0
FIGURE 1. Hypermutation spectra. The spectra of mutations, as percentage of total mutations, is shown for each of two different data sets. H
chain Ig genes (a) and L chain Ig genes (b).
Materials and Methods
Two different sets of data were used in the primary analysis. Both were of
Ig gene rearrangements that were nonproductive due to their V and J regions being in different reading frames. These genes had, however, been
isolated from postgerminal centre cells and therefore these nonproductive
rearrangements carried mutations. The advantage of using such data is that
the mutations have accumulated without any possible confounding effects
of selection. One set of data was of H chain VH-DH-JH rearrangements (4)
and the other was of ␭ L chain VL-JL rearrangements (24). The genes were
arranged according to their IGV gene usage. Eight groups of H chain genes
were collated, containing a total of 55 sequences with 732 mutations from
the germline IGHV sequence. Six groups of L chain genes were collated,
containing a total of 25 sequences with 548 mutations from the germline
IGLV sequence (3). In addition, mutated sequences from four additional
sources were analyzed: 750 mutations from a 580-bp region of the mouse
JH intron (13), 296 mutations from a 320-bp region of the human JH intron
in patients with XPV (26), 282 mutations from a 560-bp region of the S␮
switch region in a Pol␩ ⫺/⫺ mouse (27), and 2324 mutations from the
V␬Ox1 transgene mutated in vivo (916 mutations) or in vitro (1408 mutations) (18).
The mutated sequences were aligned beneath the appropriate germline
sequence and a raw text file of the alignment created. This file was imported into a Microsoft Excel spreadsheet and computations of the number
of each type of nucleotide substitution (e.g., A to T, A to C, A to G, etc.)
and the composition of the flanking sequences around these substitutions
were performed using macros in Excel (Visual Basic). Calculations were
extended to 6 bases either side of the mutated base.
Because the chances of seeing a particular base in each position are not
always 25%, and there are differences in the germline genes of individual
sequences, the composition of each germline gene used in the analysis was
determined in terms of the percentage composition of bases at each position around each separate base (A, C, G, T). The individual compositions
for each IGV gene were compiled into two sets of baseline data for H and
L chain genes, the resulting compositions being the outcome from an analysis of 15,003 and 7,740 bases of sequence respectively. For example, the
percentage composition at position ⫺1 from an A in the H chain IGHV
genes used in this analysis was 17% A, 34% C, 32% G, and 17% T as
compared with the same composition in L chain IGLV genes of 14% A,
41% C, 27% G, and 18% T.
The percentage composition of each base at each position flanking a
particular mutation was determined. The baseline percentage composition
of the germline sequences was subtracted from this to show any differences
particular to that mutation. ␹2 analysis was used to determine whether any
differences seen were significant. Computations of percentage differences
and ␹2 analysis were also performed using Excel.
Downloaded from http://www.jimmunol.org/ by guest on June 16, 2017
long patch repair. Evidence for the involvement of BER and MMR
is strong, both processes have been shown to affect the SHM spectrum if altered. In the case of BER, Uracil N-glycosylase (UNG) is
the key first step in removing mispaired uracil. The absence of
UNG (either in mice, in cultured B cells, or in hyper-IgM syndrome type 4 patients) causes a skewing in the pattern of SHM so
that more transitions at G:C pairs are seen, but the pattern of A and
T mutations appears normal (10, 11). The evidence for the involvement of MMR derives from observations of deficiencies in
the mismatch repair process. Deletions of mismatch repair proteins
MSH2 or MSH6 also causes skewing of the pattern of SHM—in
this case to exclude mutations from A and T (3). Similarly a deficiency in exonuclease 1, another component of the MMR process, can result in a reduction in the number of mutations from A
and T in Ig genes (12). Recent experiments combining deficiency
of both UNG and MSH2 showed an ablation of all mutations except transitions at G:C pairs (13). The error prone polymerase that
appears to be the most important for the secondary phase of SHM
is DNA polymerase ␩ (Pol␩). Pol␩ is absent in patients with
XP-V, and these patients have a reduced (but not completely absent) number of mutations from A and T in their Ig genes (14, 15).
Further in vitro analysis of the Pol␩ error spectrum reports that it
shares many features with the SHM spectrum found in vivo (16 –
18). However, other error prone polymerases such as Pol␨ (19, 20),
Pol␫ (21), and Pol␮ (22, 23) may also be involved.
Targeting of mutations to certain areas of the IGHV sequence is
biologically important because mutations are advantageous predominantly when they modify the hypervariable regions. However, because the mechanisms that mutate G/C and A/T are not the
same, the mechanisms that target mutations to the hypervariable
regions are not likely to be the same either. Unlike the RGYW/
WRCY motifs, which reflect target preferences of the AID complex (8, 9), the significance of the motifs that appear to focus
mutations to A/T nucleotides is unknown. The issue of whether
both processes act on one strand or two has still not been completely resolved. There is no strand bias in G vs C mutations in
most studies. In contrast, the AT mutator consistently shows strand
bias in that there are always more mutations from A than mutations
from T. The lack of strand bias in GC mutations, coupled with the
fact that the RGYW/WRCY motifs are reverse complements of
each other, shows that the GC mutator likely involves one mechanism that acts on both strands of DNA (5, 24). Strand bias in
AT mutations indicates a possible preference for the action of
the mutator for one strand over the other, or the existence of
separate mutators for A and T mutations. The hot spot for mutations from A is generally accepted as being WA (17). Evidence that mutation occurs by the same mechanism on both
strands of DNA would be that mutations from T occur in the
reverse complement hot spot TW. This has been found in some
cases, and has been used as a predictor of mutations from T
(25). However, we have previously shown that the hot spot
motifs for mutations from A and T vary depending whether the
mutations are transitions or transversions (4) and these are not
necessarily reverse complement.
To analyze influences on AT mutations we have analyzed the
hot spot motifs around each individual mutation (A-C, A-G, A-T,
T-A, T-C, T-G). This analysis has been done in out of frame Ig
sequences to exclude any effects of selection. We show that WA is
not a target motif for all mutations from A and that adjacent nucleotides bias the nature of replacements from A. Furthermore
there are qualitative, as well as quantitative, aspects to strand bias.
These results have been confirmed in mouse JH intronic
sequences.
5171
5172
Results
Different motifs favor individual mutations from A and T
transversions from A to T, and AAT for transitions from A to G
(where the mutated base is underlined). Looking at the flanking
bases with a negative influence on a particular mutation, a G in
position ⫺1 from an A has a negative influence on the transversions (A-C and A-T), the influence being much stronger for the
A-T mutations. A transitions (A-G) are negatively influenced by a
C in position ⫺1, as are the A-C transversions, but this is not
apparent in the A-T transversions (Fig. 3).
Adjacent bases influence the spectrum of replacements from A
The data in Fig. 3 are inconsistent with the notion that the flanking
sequences associated with mutations from A are ‘target’ motifs,
because a targeting motif would result in some aspect being common to all three different motifs for mutations from A and this is
not the case. We therefore considered the possibility that what is
perceived to be a motif is in reality a consequence of the influence
of adjacent bases on either the replacement spectra, or the probability of subsequent fixing of mutation or repair. The analysis used
in generating Figs. 2 and 3, in common with other analyses of
hypermutation motifs (4, 5, 28, 29) corrects for the germline composition of the Ig genes in question. To view the influence of
adjacent bases on nucleotide replacements directly, we looked at
the numbers of each substitution without correcting for germline
composition. Fig. 4 shows the nucleotide replacement spectrum
when A or T are flanked by different nucleotides in either the 5⬘ or
the 3⬘ positions.
There are significant differences in the proportions of transition
(e.g., A-G or T-C), complement (e.g., A-T or T-A), or transition
FIGURE 2. Reversibility of A-N and T-N hypermutation motifs. Base preferences at positions ⫺1 (5⬘ side) to ⫹1 (3⬘ side) around mutations from A
and T (a and b). In each case, the motif for one mutation is shown above the reverse complement of the motif for the matching mutation, i.e., the T-N motif
is shown in the opposite direction (3⬘ to 5⬘) with the complementary bases (A is read as T, C is read as G, etc.) as c underneath the A-N motif in a. Similarly,
the reverse complement of the motif for A-N (d) is compared with the T-N motif (b). The bars represent the percentage difference between the base
composition around the mutation compared with the normal germline base composition The levels of significance after ␹2 test are indicated by the shading
of the bars: u, p ⬍ 0.05; o, p ⬍ 0.005; and f, p ⬍ 0.0005. Where the percentage difference is significant this represents a positive or negative influence
of that base in that position over the mutation in question. For example, T and A in position ⫺1 both have a significant positive influence on an A-N
mutation, while G and C both have a significant negative influence at this position. In position ⫹1 there is no significant effect of A or C, but G has a
negative influence and T has a positive influence. Overall this results in a WAT motif for the A-N mutation (where W is A or T and N is any nucleotide).
Downloaded from http://www.jimmunol.org/ by guest on June 16, 2017
1251 nucleotide substitutions in out of frame Ig rearrangements
were analyzed, including 727 from IGVH, and 524 from IGVL.
Consistent with all previous studies, significant strand bias in mutations from A and T was observed in IGVH and IGVL sequences
(Fig. 1). Because the two different data sets revealed very similar
mutation spectra the data were pooled for further analyses. We
determined how the sequence flanking A and T nucleotides influenced the probability of their mutation by analysis of the bases
favored (or not) in positions surrounding a mutated A or T, after
correction for the natural bias present in the germline sequence.
The strongest, and most consistent, influences on A and T mutations were from the bases in positions ⫺1 and ⫹1. This analysis is
presented in graphs that show whether a particular base is over- or
underrepresented (compared with the germline sequence) in these
positions around a mutated A or T.
The motifs thus determined for all mutations from A or T appear
to be the reverse complement of each other, as illustrated in Fig. 2.
Having a T before a mutated A, or an A after a mutated T, are
consistent positive influences on whether A and T are mutated.
However, this reverse complementarity did not always hold true
when individual substitutions from A and T were analyzed separately. Although the motifs around the different substitutions from
T were very similar, those around mutations from the different
substitutions from A showed significant differences (Fig. 3). Looking at the flanking bases with a positive influence on mutation from
A, the motifs are: TAC for transversions from A to C, TAT for
HYPERMUTATION AT A-T BASE PAIRS
The Journal of Immunology
5173
complement (e.g., A-C or T-G) mutations depending on the base in
the 5⬘ or 3⬘ position from the mutations. Because the normal distribution of transition, complement and transition complement mutations in mutated Ig genes is not equal we used the average percentage distribution from all mutations (50.2, 30.9, and 18.9%,
respectively) as the expected values for comparison. The significant changes from these expected values are indicated in Fig. 4.
Strikingly, the normal transition bias observed when all replacements are pooled is not seen when A is preceded by a T, and the
number of transition complements is significantly increased.
Hence, in this sequence context the replacement of A with C, G or
T occurs with approximately equal frequency. The presence of a C
in the 3⬘ position has a similar effect. An A or a G in the 5⬘ position
predisposes the replacements toward transitional (A-G) mutations.
The 5⬘ G also prejudices against complement (A-T) mutations,
while a C in the 5⬘ position favors them. There are no such changes
in the spectra of T mutations, the only point of change that is
significantly different from the normal is that a T in the 5⬘ position
from a T prejudices against transition complement mutations
(T-G).
Microsequence effects around mutations from A and T are not
reverse complements of each other
Reverse complementarity of motifs around complementary mutations can indicate targeting of one base on both strands of DNA.
The data presented in Fig. 4 above shows quite clearly that there
are significant effects of the bases immediately flanking mutations
from A which are not seen for those around T. If the effect were
due to the same mechanism acting on both strands of DNA, then
one would expect the effects seen at the 5⬘ side of A mutations
(Fig. 4a) to be mirrored at the 3⬘ side of T mutations (Fig. 4d), this
is not the case. Moreover, when the individual motifs for the different mutations from A are considered alongside the reverse complement motifs for their corresponding reverse complement mutation it can be seen that there are more points that differ than there
are that match (Fig. 5).
Comparison with mutations from the mouse JH intronic
sequence
Human out-of-frame IGV gene data has the advantage that it is
made up of 14 different IGHV and IGLV genes so that the chances
Downloaded from http://www.jimmunol.org/ by guest on June 16, 2017
FIGURE 3. Hypermutation motifs around individual mutations from A and T. Base preferences at positions ⫺1 to ⫹ 1 around mutations from A and
T. The bars represent the percentage difference between the base composition around the mutation compared with the normal germline base composition.
The levels of significance after ␹2 test are indicated by the shading of the bars: ( ), p ⬍ 0.05, (_), p ⬍ 0.005, and (䡵), p ⬍ 0.0005.
5174
HYPERMUTATION AT A-T BASE PAIRS
of bias arising from multiple copies of a single gene are less.
However, because IG genes have been selected through evolution alongside the hypermutation mechanisms they have characteristics that may cause perceived biases compared with
non-Ig gene sequences. Data are available from the mouse JH
intronic region (13) and we have analyzed this with respect to
the two key findings in this paper. First, with respect to the
effect of the preceding base on the type of mutation that occurs,
our results show that an A in the position immediately preceding the mutated A causes a bias toward transition mutations,
whereas a T in this position causes a bias toward transversions
(Fig. 4a). We found that this was also true in a database of 750
mutations measured in the JH intronic region of the mouse (of
which 280 mutations were from A) and in 916 mutations measured in a VkOx1 transgene (of which 298 mutations were from
A). In both cases the ratio of transitions to transversions was ⬎1
in the motif AA and ⬍1 in the motif TA (Table I). Second, with
respect to the nonreversibility of complementary mutations, the
reverse complements of motifs for mutations at T are not the
same as the motifs for A (Fig. 5). The effect of 5⬘ T on the types
of mutation that occur from A is not mirrored by the effect of A
in the 3⬘ position from a T (Fig. 4a compared with Fig. 4d). This
is also illustrated in Fig. 6a, showing the distribution of the
different types of mutation from A in the TA motif compared
with the reverse complement situation—the distribution of the
different types of mutation from T in the TA motif. Mutations
in the TA motifs of the mouse JH intronic sequences (Fig. 6b)
have the same trend of nonreversibility for transition (A-G vs
T-C) and complement (A-T vs T-A) mutations. However, this
does not hold true for the transition complement (A-C and T-G)
mutations. We investigated whether this may be due to a skewing in the germline composition of the two data sets. The microsequence that favored A-C mutations was TAC and the effect of a 3⬘ C on the types of mutations from A is very striking
Downloaded from http://www.jimmunol.org/ by guest on June 16, 2017
FIGURE 4. Influence of neighboring bases on hypermutation. Base composition of the 5⬘ and 3⬘ positions for each of the three different types of
mutation: Transition (u), complement (f), and transition complement (䡺). In the case of a and b, transition ⫽ A-G, complement ⫽ A-T, and transition
complement ⫽ A-C. In the case of c and d, transition ⫽ T-C, complement ⫽ T-A, and transition complement ⫽ T-G. The germline distribution of bases
adjacent to A for this data is also shown (e and f). Significant differences from the average distribution of all mutations in these data (50.2% transitions,
30.9% complements, and 18.9% transition complements) are indicated as follows: ⴱ, p ⬍ 0.05; ⴱⴱ, p ⬍ 0.005; and ⴱⴱⴱ, p ⬍ 0.0005.
The Journal of Immunology
5175
(Fig. 4b). TAC occurs with a frequency of only 0.52% in the JH
intronic sequence, compared with 2.1% in the human IGV gene
sequences.
Comparisons with Pol␩ activity
Mutations from A and T have been ascribed to the error prone
polymerase activity of Pol␩. The sequence context of in vivo IGV
mutations from A and T has been compared favorably with that
shown by mutations caused by Pol␩ when copying a lac Z gene, or
a mouse transgene in vitro (16 –18), and Pol␩-deficient mice and
humans show a reduced level of mutation from A and T (14, 15,
26, 27). An analysis of the replacement spectrum of mutations
from A caused by Pol␩ in the AA and TA motifs shows a strong
bias toward transition mutations when synthesizing the nontranscribed strand of an IGVK transgene, and a bias toward transversions when synthesizing the transcribed strand. There are differences between the mutations in the AA motif compared with those
in the TA motif, the transition bias is very high for the AA motif
(Table I). In contrast, the transition bias normally seen in the AA
motif as compared with the TA motif is absent in the mutation
spectrum of mice and humans deficient in Pol␩ (Table I).
Discussion
We have analyzed the replacement spectra resulting from the SHM
of A and T nucleotides in the context of local microsequence. We
have observed that adjacent bases 5⬘ and 3⬘ consistently and significantly influence the replacement spectrum of mutations from A.
Most notably we observed random replacement of A (i.e., C, G, or
T occur with equal frequency) when A is flanked by a T in the 5⬘
position, or a C in the 3⬘ position. This is significantly different to
the pattern of mutations from A seen in other sequence contexts,
where the more usual bias toward A-G transitions occurs. Other
studies of hypermutation hot spots, including our data in Fig. 1,
involved the determination of the proportion of mutations occurring in a particular sequence context as compared with the frequency of that sequence context in the germline. Such analysis,
using proportions to control for germline sequence biases, can
mask more simple trends such as the random replacement spectrum from A in the context of TA described above and illustrated
in Fig. 6. Other biases in substitution were apparent, and indicated
an influence of the adjacent nucleotides over the incorporation of
mis-paired nucleotides. This preference tended toward “like” nucleotides: if a purine (A or a G) preceded the target A, then the
replacement base was more likely to be another purine (G). However, if a pyrimidine, T, preceded the target A, then the replacement base was more likely to be the pyrimidines C or T. Similarly,
the presence of a pyrimidine, C, in the 3⬘ position predisposes
toward replacement by pyrimidines C or T. The most striking observation in this regard is that there is a bias in the A-G transitions
from the second A of an AA motif, whereas the bias switches
Table I. Relative numbers of transitions and transversions from A occurring in motif AA as compared with motif TA
Human IGHV
Human IGLV
JH intron (13)
VkOx1 (18)
Transcribed strand (18)b
Nontranscribed strand (18)b
Human JH XPV (31)
Pol ␩⫺/⫺ S␮ (32)
AA
TA
Total No. of Mutations
No. of Mutations
Ts/Tv (ratio)a
Ts/Tv (ratio)a
732
548
750
916
632
776
296
282
209
182
280
298
141
421
25
31
28/20 (1.40)
21/9 (2.33)
56/33 (1.70)
52/25 (2.08)
5/16
147/0
4/7 (0.57)
4/12 (0.33)
19/42 (0.45)
19/36 (0.53)
53/77 (0.69)
32/57 (0.56)
3/48
65/0
1/1 (1)
2/3 (0.67)
a
Transcribed strand and nontranscribed strand are mutations arising during the (in vitro) synthesis (by polymerase ␩) of the transcribed strand and nontranscribed strand of
the VkOx1 gene, respectively (18).
b
The numbers of transitions (Ts) and transversions (Tv) are given. The ratio is of transitions over transversions.
Downloaded from http://www.jimmunol.org/ by guest on June 16, 2017
FIGURE 5. Nonreversibility of hypermutation motifs for A and T transitions. Base preferences at positions ⫺1 to ⫹1 around A-C, A-G, and A-T
mutations compared with the reverse complement of the corresponding T-A, T-C, and T-A motifs, respectively. The bars represent the percentage difference
between the base composition around each mutation, compared with the normal germline base composition. The levels of significance after ␹2 test are
indicated by the shading of the bars: u, p ⬍ 0.05; o, p ⬍ 0.005; and f, p ⬍ 0.0005.
5176
toward transversions (A-C and A-T) in the A of a TA motif. This
observation is consistent between different data sets, including
those from non-Ig sequence (Table I). The influence of adjacent
bases on replacements from A was not mirrored in mutations from
T. The only significant variability in the replacement spectrum
from T occurred when there was a C or a T in the 5⬘ position. The
levels of significance were much smaller in this instance but it is
still possible to see a trend toward replacement by pyrimidine C
when preceded by pyrimidines C or T.
The fact that adjacent bases can influence the type of mutation
is consistent with the involvement of error prone polymerases at
this stage of SHM. Different polymerases have widely differing
abilities to insert, and to extend from, a mispaired insertion and it
has been shown that sequence context can influence the specificity
of a polymerase (30). It is therefore likely that what has hitherto
been perceived as motifs targeting mutation from A and T may
actually be a consequence of the influence of adjacent bases on
substitution preferences by the error prone polymerases. Pol␩ is
thought to be the main error-prone polymerase causing mutation of
A and T nucleotides; the absence of Pol␩ causes a reduced number
of mutations at A and T (14, 15, 26, 27) and it has recently been
shown to have physical and functional links with MSH2 (31). This
polymerase tends to generate transitions as a consequence of misincorporation of dGMP opposite T (16, 32). Transitions from T
predominate when the transcribed strand is synthesized (18), and
are more likely to occur when the base pair preceding the error is
T.A or A.T than when it is G.C or C.G (16). The sequence contextdependent variations in replacement spectrum observed in our
study were not observed in the in vitro models of Pol␩ activity.
However, we did see that the transition bias of mutations from A
as a result of the in vitro Pol␩ activity was much higher in the AA
than in the TA motif. This is in accord with the transition bias seen
in vivo in the human IGV sequences as well as the data from the
VkOx1 mouse transgene (18) and mouse JH intron (13). Also consistent with a role for Pol␩ in AT mutation is the reduction in the
transition bias of mutations from A in the AA motif in the human
and mouse models of Pol␩ deficiency (26, 27), although the numbers of mutations available in this analysis are rather low. However, removal of Pol␩ does not completely abrogate mutations
from A and T and on balance it is likely that Pol␩ is not the only
error prone polymerase involved in the generation of mutations at
A and T in vivo. Other error-prone polymerases (Pol␨, Pol␫ and
Pol␮) have also been implicated (19 –23).
Strand bias, generating more mutations from A than from T, is
a consistent feature of the SHM process in vivo. The implication
has been that this reflects differences in activity of the AT mutator
on the transcribed and non-transcribed strands. The idea that the A
and T mutators are equivalent, but operate unequally on both
strands in vivo, was supported by the apparently reversible motifs WA and TW that are associated with hypermutation hot
spots. A previous study by Cowell and Kepler (33) reported
symmetry under complementation in support of this hypothesis.
However, in this study we show that WA and TW are not truly
reversible motifs for A and T mutations, but only appear so as
a result of pooling transitions and transversions. There are some
striking differences in motifs from A compared with the reverse
complement of the complementary motifs, most notably in the
transitions; A to G mutations occur preferentially in AAT, but
T to C mutations occur preferentially in TTA (Fig. 5). This
finding is in accord with the finding of Cowell and Kepler (33)
that a nucleotide within a homodimer is more likely to undergo
a transition mutation. However, these motifs are not reverse
complements of each other so do not indicate strand symmetry
under complementation. The actual numbers of mutations,
without the proportional correction for germline sequence,
clearly illustrate that the influences of adjacent nucleotides on
the types of mutations from A are not mirrored in mutations
from T (Fig. 4). This is also true for mutations in non-Ig sequence (Fig. 6). Thus there are two aspects to strand bias; a
quantitative bias indicating that a mechanism operates unequally on both strands, and a qualitative difference—in the
type of mutation that is created in a given context—indicating
that the mechanism itself may be different on each strand.
Instances of TA and TAC sequence motifs where the A is in the
second position of a codon are most commonly observed in the
complementarity determining regions of the Ig genes. Hence,
the mutation of A in these circumstances is likely to have functional consequences. For example, the TAC codon is often observed in complementarity determining regions where mutation of
the A would result in replacement of tyrosine with phenylalanine,
serine, or cysteine. This could potentially generate significant
changes in the Ag binding sites.
In summary, we have shown that adjacent nucleotides affect the
substitutions when A nucleotides are mutated. However, mutations
from T are not influenced in the same way. Equivalent, complementary, sequence contexts for A and T mutations result in different replacement spectra. Therefore, not only is there a quantitative difference between mutations from A and T, but there is also
a qualitative difference. This could be a consequence of differential
mutation and/or repair activity on the two DNA strands.
Acknowledgments
We are grateful to Laurent Boursier and Su Wen for producing Ig gene
sequence data, and to Mark Dunn for help with the programming.
Downloaded from http://www.jimmunol.org/ by guest on June 16, 2017
FIGURE 6. Qualitative nonreversibility of the TA motif; the distribution of the three different types of mutation in motif TA from both A and
T mutations. A-G transitions are shown in gray, A-T complement mutations are shown in black, and A-C transition complement mutations are
shown in white. The corresponding mutations from T (T-C transitions, T-A
complement, and T-G transition complement, respectively) are shown in
the adjacent striped bars. The values are given as a percentage of the total
numbers of mutations from A in TA (n ⫽ 116 for human IGV; n ⫽ 130 for
mouse JH intron) or from T in TA (n ⫽ 72 for human IGV; n ⫽ 62 for
mouse JH intron).
HYPERMUTATION AT A-T BASE PAIRS
The Journal of Immunology
Disclosures
The authors have no financial conflict of interest.
References
17. Rogozin, I. B., Y. I. Pavlov, K. Bebenek, T. Matsuda, and T. A. Kunkel. 2001.
Somatic mutation hotspots correlate with DNA polymerase ␩ error spectrum.
Nat. Immunol. 2: 530 –536.
18. Pavlov, Y. I., I. B. Rogozin, A. P. Galkin, A. Y. Aksenova, F. Hanaoka, C. Rada,
and T. A. Kunkel. 2002. Correlation of somatic hypermutation specificity and
A-T base pair substitution errors by DNA polymerase ␩ during copying of a
mouse immunoglobulin ␬ light chain transgene. Proc. Natl. Acad. Sci. USA 99:
9954 –9959.
19. Diaz, M., L. K. Verkoczy, M. F. Flajnik, and N. R. Klinman. 2001. Decreased
frequency of somatic hypermutation and impaired affinity maturation but intact
germinal center formation in mice expressing antisense RNA to DNA polymerase
zeta. J. Immunol. 167: 327–335.
20. Zan, H., A. Komori, Z. Li, A. Cerutti, A. Schaffer, M. F. Flajnik, M. Diaz, and
P. Casali. 2001. The translesion DNA polymerase zeta plays a major role in Ig
and Bcl-6 somatic hypermutation. Immunity 14: 643– 653.
21. Faili, A., S. Aoufouchi, E. Flatter, Q. Gueranger, C. A. Reynaud, and J. C. Weill.
2002. Induction of somatic hypermutation in immunoglobulin genes is dependent
on DNA polymerase ␫. Nature 419: 944 –947.
22. Ruiz, J. F., D. Lucas, E. Garcia-Palomero, A. I. Saez, M. A. Gonzalez,
M. A. Piris, A. Bernad, and L. Blanco. 2004. Overexpression of human polymerase mu (Pol ␮) in a Burkitt’s lymphoma cell line affects the somatic hypermutation rate. Nucleic Acids Res. 32: 5861–5873.
23. Lucas, D., T. Lain de Lera, M. A. Gonzalez, J. F. Ruiz, O. Dominguez,
J. C. Casanova, C. Martinez-A, L. Blanco, and A. Bernad. 2005. Polymerase mu
is up-regulated during the T cell-dependent immune response and its deficiency
alters developmental dynamics of spleen centroblasts. Eur. J. Immunol. 35:
1601–1611.
24. Boursier, L., W. Su, and J. Spencer. 2004. Analysis of strand biased ‘G’.C hypermutation in human immunoglobulin V(␭) gene segments suggests that both
DNA strands are targets for deamination by activation-induced cytidine deaminase. Mol. Immunol. 40: 1273–1278.
25. Michael, N., T. E. Martin, D. Nicolae, N. Kim, K. Padjen, P. Zhan, H. Nguyen,
C. Pinkert, and U. Storb. 2002. Effects of sequence and structure on the hypermutability of immunoglobulin genes. Immunity 16: 123–134.
26. Mayorov, V. I., I. B. Rogozin, L. R. Adkinson, and P. J. Gearhart. 2005. DNA
polymerase ␩ contributes to strand bias of mutations of A versus T in immunoglobulin genes. J. Immunol. 174: 7781–7786.
27. Delbos, F., A. De Smet, A. Faili, S. Aoufouchi, J.-C. Weill, and C.-A. Reynaud.
2005. Contribution of DNA polymerase ␩ to immunoglobulin gene hypermutation in the mouse. J. Exp. Med. 201: 1191–1196.
28. Shapiro, G. S., K. Aviszus, D. Ikle, and L. J. Wysocki. 1999. Predicting regional
mutability in antibody V genes based solely on di- and trinucleotide sequence
composition. J. Immunol. 163: 259 –268.
29. Dorner, T., H. P. Brezinschek, R. I. Brezinschek, S. J. Foster, R. Domiati-Saad,
and P. E. Lipsky. 1997. Analysis of the frequency and pattern of somatic mutations within nonproductively rearranged human variable heavy chain genes.
J. Immunol. 158: 2779 –2789.
30. Strauss, B. S. 2002. The “A” rule re-visited: polymerases as determinants of
mutational specificity. DNA Repair. 1: 125–135.
31. Wilson, T. M., A. Vaisman, S. A. Martomo, P. Sullivan, L. Lan, F. Hanaoka,
A. Yasui, R. Woodgate, and P. J. Gearhart. 2005. MSH2-MSH6 stimulates DNA
polymerase ␩, suggesting a role for A:T mutations in antibody genes. J. Exp.
Med. 201: 637– 645.
32. Kunkel, T. A., Y. I. Pavlov, and K. Bebenek. 2003. Functions of human DNA
polymerases ␩, ␬ and i suggested by their properties, including fidelity with
undamaged DNA templates. DNA Repair 2: 135–149.
33. Cowell, L. G., and T. B. Kepler. 2000. The nucleotide-replacement spectrum
under somatic hypermutation exhibits microsequence dependence that is strandsymmetric and distinct from that under germline mutation. J. Immunol. 164:
1971–1976.
Downloaded from http://www.jimmunol.org/ by guest on June 16, 2017
1. Golding, G. B., P. J. Gearhart, and B. W. Glickman. 1987. Patterns of somatic
mutations in immunoglobulin variable genes. Genetics 115: 169 –176.
2. Betz, A. G., C. Rada, R. Pannell, C. Milstein, and M. S. Neuberger. 1993. Passenger transgenes reveal intrinsic specificity of the antibody hypermutation mechanism: clustering, polarity, and specific hot spots. Proc. Natl. Acad. Sci. USA 90:
2385–2388.
3. Phung, Q. H., D.B. Winter, A. Cranston, R. E. Tarone, V. A. Bohr, R. Fishel, and
P. J. Gearhart. 1998. Increased hypermutation at G and C nucleotides in immunoglobulin variable genes from mice deficient in the MSH2 mismatch repair
protein. J. Exp. Med. 187: 1745–1751.
4. Spencer, J., M. Dunn, and D. K. Dunn-Walters. 1999. Characteristics of sequences around individual nucleotide substitutions in IgVH genes suggest different GC and AT mutators. J. Immunol. 162: 6596 – 6601.
5. Dunn-Walters, D. K., A. Dogan, L. Boursier, C. M. MacDonald, and J. Spencer.
1998. Base-specific sequences that bias somatic hypermutation deduced by analysis of out of frame genes. J. Immunol. 160: 2360 –2364.
6. Rada, C., M. R. Ehrenstein, M. S. Neuberger, and C. Milstein. 1998. Hot spot
focusing of somatic hypermutation in MSH2-deficient mice suggests two stages
of mutational targeting. Immunity 9: 135–141.
7. Durandy, A. 2003. Activation-induced cytidine deaminase: a dual role in classswitch recombination and somatic hypermutation. Eur. J. Immunol. 33:
2069 –2073.
8. Larijani, M., D. Frieder, W. Basit, and A. Martin. 2005. The mutation spectrum
of purified AID is similar to the mutability index in Ramos cells and in
ung⫺/⫺msh2⫺/⫺ mice. Immunogenetics 56: 840 – 845.
9. Pham, P., R. Bransteitter, J. Petruska, and M. F. Goodman. 2003. Processive
AID-catalysed cytosine deamination on single-stranded DNA simulates somatic
hypermutation. Nature 424: 103–107.
10. Rada, C., G. T. Williams, H. Nilsen, D. E. Barnes, T. Lindahl, and
M. S. Neuberger. 2002. Immunoglobulin isotype switching is inhibited and somatic hypermutation perturbed in UNG-deficient mice. Curr. Biol. 12:
1748 –1755.
11. Imai, K., G. Slupphaug, W. I. Lee, P. Revy, S. Nonoyama, N. Catalan, L. Yel,
M. Forveille, B. Kavli, H. E. Krokan, et al. 2003. Human uracil-DNA glycosylase
deficiency associated with profoundly impaired immunoglobulin class-switch recombination. Nat. Immunol. 4: 1023–1028.
12. Bardwell, P. D., C. J. Woo, K. Wei, Z. Li, A. Martin, S. Z. Sack, T. Parris,
W. Edelmann, and M. D. Scharff. 2004. Altered somatic hypermutation and reduced class-switch recombination in exonuclease 1 mutant mice. Nat. Immunol.
5: 224 –229.
13. Rada, C., J. M. Di Nola, and M. S. Neuberger. 2004. Mismatch recognition and
uracil excision provide complementary paths to both Ig switching and the A/Tfocused phase of somatic mutation. Mol. Cell 16: 163–171.
14. Zeng, X., D. B. Winter, C. Kasmer, K. H. Kraemer, A. R. Lehmann, and
P. J. Gearhart. 2001. DNA polymerase ␩ is an A-T mutator in somatic hypermutation of immunoglobulin variable genes. Nat. Immunol. 2: 537–541.
15. Yavuz, S., A. S. Yavuz, K. H. Kraemer, and P. E. Lipsky. 2002. The role of
polymerase ␩ in somatic hypermutation determined by analysis of mutations in
a patient with XPV. J. Immunol. 169: 3825–3830.
16. Matsuda, T., K. Bebenek, C. Masutani, I. B. Rogozin, F. Hanaoka, and
T. A. Kunkel. 2001. Error rate and specificity of human and murine DNA polymerase ␩. J. Mol. Biol. 312: 335–346.
5177