doi:10.1182/blood-2007-07-099564Prepublished online October 24, 2007;   

 Belessi, Paolo Ghia, Fred Davi, Richard Rosenquist and Kostas StamatopoulosAchilles Anagnostopoulos, Federico Caligaris-Cappio, Dominique Vaur, Christos Ouzounis, ChrysoulaScielzo, Nikolaos Laoutaris, Karin Karlsson, Fanny Baran-Marzsak, Athanasios Tsaftaris, Carol Moreno, Fiona Murray, Nikos Darzentas, Anastasia Hadzidimitriou, Gerard Tobin, Myriam Boudjogra, Cristina in leukemogenesis

selectionchronic lymphocytic leukemia: implications for the role of antigen Stereotyped patterns of somatic hypermutation in subsets of patients with

(4217 articles)Neoplasia   �Articles on similar topics can be found in the following Blood collections

http://bloodjournal.hematologylibrary.org/site/misc/rights.xhtml#repub_requestsInformation about reproducing this article in parts or in its entirety may be found online at:

http://bloodjournal.hematologylibrary.org/site/misc/rights.xhtml#reprintsInformation about ordering reprints may be found online at:

http://bloodjournal.hematologylibrary.org/site/subscriptions/index.xhtmlInformation about subscriptions and ASH membership may be found online at:

digital object identifier (DOIs) and date of initial publication. theindexed by PubMed from initial publication. Citations to Advance online articles must include

final publication). Advance online articles are citable and establish publication priority; they areappeared in the paper journal (edited, typeset versions may be posted when available prior to Advance online articles have been peer reviewed and accepted for publication but have not yet

Copyright 2011 by The American Society of Hematology; all rights reserved.20036.the American Society of Hematology, 2021 L St, NW, Suite 900, Washington DC Blood (print ISSN 0006-4971, online ISSN 1528-0020), is published weekly by    

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

1

STEREOTYPED PATTERNS OF SOMATIC HYPERMUTATION IN

SUBSETS OF PATIENTS WITH CHRONIC LYMPHOCYTIC LEUKEMIA:

IMPLICATIONS FOR THE ROLE OF ANTIGEN SELECTION IN

LEUKEMOGENESIS

Fiona Murray1*, Nikos Darzentas2*, Anastasia Hadzidimitriou2,3*, Gerard Tobin1,

Myriam Boudjogra4, Cristina Scielzo5, Nikolaos Laoutaris6, Karin Karlsson7, Fanny

Baran-Marzsak8, Athanasios Tsaftaris2, Carol Moreno9, Achilles Anagnostopoulos3,

Federico Caligaris-Cappio5, Dominique Vaur10, Christos Ouzounis2, Chrysoula

Belessi6, Paolo Ghia5, Fred Davi4, Richard Rosenquist1, and Kostas Stamatopoulos3

From: (1) Department of Genetics and Pathology, Rudbeck Laboratory, Uppsala

University, Uppsala, Sweden; (2) Computational Genomics Unit/ Institute of

Agrobiotechnology, Centre for Research and Technology, Thessaloniki, Greece; (3)

Hematology Department and HCT Unit, G. Papanicolaou Hospital, Thessaloniki,

Greece; (4) Laborary of Hematology and Université Pierre et Marie Curie, Hôpital

Pitié-Salpètrière, Paris, France; (5) Laboratory and Unit of Lymphoid Malignancies,

Department of Oncology, Università Vita-Salute, San Raffaele and Instituto

Scientifico San Raffaele, Milano, Italy; (6) Hematology Department, Nikea General

Hospital, Athens, Greece; (7) Department of Hematology, Lund University Hospital,

Lund, Sweden; (8) Laboratory of Hematology, Hôpital Avicenne and EA 3046

Université Paris 13, Bobigny, France; (9) Institute of Hematology and Oncology and

Institut d’Investigacions Biomediques August Pi i Sunyer (IDIBAPS), Hospital Clinic,

University of Barcelona, Barcelona, Spain; (10) Laboratory of Biology and Oncology,

Centre François Baclesse, Caen, France.

*Fiona Murray, Nikos Darzentas and Anastasia Hadzidimitriou contributed equally to

this work as first authors.

Running title: Somatic hypermutation patterns in CLL

Blood First Edition Paper, prepublished online October 24, 2007; DOI 10.1182/blood-2007-07-099564

Copyright © 2007 American Society of Hematology

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

2

Corresponding author:

Paolo Ghia, M.D., Ph.D.

Università Vita-Salute San Raffaele

Via Olgettina 58, 20132 Milano, Italy

Phone: +39-02-2643.4797/ Fax : +39-02-2643.4723

e-mail: ghia.paolo@hsr.it

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

3

ABSTRACT

We have examined somatic hypermutation (SHM) features in a series of 1967

immunoglobulin heavy chain gene (IGH) rearrangements obtained from patients with

chronic lymphocytic leukemia (CLL) and compared them with IGH sequences from

non-CLL B cells available in public databases. SHM analysis was performed for all

1290 CLL sequences in this cohort with <100% identity to germline. At the cohort

level, SHM patterns were typical of a canonical SHM process. However, important

differences emerged from the analysis of certain subgroups of CLL sequences

defined by: (i) IGHV gene usage; (ii) presence of stereotyped heavy chain

complementarity-determining region 3 (HCDR3) sequences; and (iii) mutational load.

We demonstrate that recurrent, “stereotyped” amino acid changes occur across the

entire IGHV region in CLL subsets carrying stereotyped HCDR3 sequences,

especially those expressing the IGHV3-21 and IGHV4-34 genes. These mutations

are under-represented among non-CLL sequences and thus can be considered as

CLL-biased. Furthermore, we show that even a low level of mutations may be

functionally relevant, given that stereotyped amino acid changes can be found in

subsets of minimally mutated cases. The very precise targeting and distinctive

features of SHM in selected subgroups of CLL patients provide further evidence for

selection by specific antigenic element(s).

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

4

INTRODUCTION

Developing B cells generate a vast repertoire of antibody specificities through

somatic recombination of distinct variable (V), diversity (D) (heavy chain only), and

joining (J) genes to form the variable domain exons of immunoglobulins (IG)1. Unlike

heavy chain complementarity determining regions (HCDR) 1 and 2, which are

entirely encoded by the IGHV gene, HCDR3 is created de novo by the VDJ

recombination process1. The skewing of diversity to the HCDR3 implies that HCDR3

sequences are the principal determinants of specificity, at least in the primary

repertoire2-3. However, HCDR3 diversity is not enough to realize the full potential of

antibody diversity4. Furthermore, unconventional antigens, such as B cell

superantigens, may be recognized not via the CDRs but rather via the framework

regions (FRs) 5.

Somatic hypermutation (SHM) of IG variable genes forms a second round of

diversification after somatic recombination which increases antibody diversity6. SHM

has long been thought to occur mainly in the germinal centers (GCs) after antigen

stimulation and in a manner dependent on T cell help7. Recent reports, however,

suggest that SHM can be T cell independent and may also occur outside classical

GCs8-13.

In recent years, the mutational status of IGHV genes has been established as one of

the most important molecular genetic markers in defining prognostic subgroups of

chronic lymphocytic leukemia (CLL). CLL patients who carry IGHV genes with ≥98%

identity to the closest germline gene (“unmutated”) follow a more aggressive clinical

course and have strikingly shorter survival than patients carrying IGHV genes with

<98% identity to germline (“mutated”)14-15. The 98% cut-off was chosen as a short cut

to exclude potential polymorphic variants16-19 and has been used by the vast majority

of studies to make the clinically relevant distinction between “mutated” and

“unmutated” cases. Initially it was assumed that CLL cells expressing unmutated

IGHV genes derived from naïve B cells. Nevertheless, it was subsequently

demonstrated that all CLL cells, irrespective of IGHV gene mutation status, have a

surface phenotype typical of antigen-experienced B cells and show gene expression

profiles similar to memory B cells14, 20-23.

The CLL IG repertoire is characterized by over-representation of selected IGHV

genes, in particular IGHV1-69, IGHV4-34, IGHV3-7 and IGHV3-21, although their

relative frequencies vary between cohorts14, 24-27. SHM does not appear to occur

uniformly among IGHV genes: for example, the IGHV1-69 gene is consistently

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

5

reported to carry very few mutations as opposed to the IGHV3-7, IGHV3-23 and

IGHV4-34 genes, which typically show a high load of mutations14, 24-27.

Recently, multiple CLL subsets with distinctive IG heavy and light chain gene

rearrangements were characterized and found to have remarkably stereotyped

HCDR3 sequences within their B cell receptors (BCRs)27-34. The expression of

stereotyped BCRs was reported as significantly more frequent among CLL patients

with unmutated vs. mutated IGHV genes32, 34. CLL cases expressing stereotyped

BCRs may also share unique molecular and clinical features, suggesting that a

particular antigen-binding site can make a difference in terms of clinical presentation

and possibly prognosis30, 34. For instance, the IGHV3-21/IGLV3-21 subset should be

regarded as unfavorable whatever the degree of mutation35, whereas the IGHV4-

34/IGKV2-30 subset seems to be associated with an indolent course of the disease34,

36.

Shared replacement mutations (“stereotyped” amino acid changes) at particular

codon positions have been reported for a few subsets34, 37. These selective

hypermutations may thus be interpreted as further evidence of antigen selection in

CLL. That notwithstanding, relatively little is known about the pattern of SHM in CLL

using certain IGHV genes or in subsets with stereotyped BCRs, in relation to that of

B cells from healthy individuals or patients with autoreactive diseases.

In this study, we examined the IGHV/IGHD/IGHJ rearrangements of 1939 patients

with CLL and compared them with a large panel of IGH sequences from various

types of normal and autoreactive B cells available in public databases. We

demonstrate striking repertoire biases and HCDR3 features in unmutated or

minimally mutated sequences, suggesting that, at least in some cases, the lack of

mutations could be interpreted in the context of antigenic pressure to maintain the

BCR in a germline state. While SHM patterns were, for the most part, typical of a

canonical SHM process, we report that groups of CLL cases expressing the IGHV3-

21 and IGHV4-34 genes exhibit unique SHM patterns. Remarkably, we also

demonstrate that recurrent, “stereotyped” amino acid changes may often be evident

across the entire IGHV gene sequence of patients with CLL expressing mutated

BCRs with stereotyped HCDR3 sequences, even among minimally mutated cases.

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

6

PATIENTS AND METHODS Patient group

A total of 1939 patients with CLL from collaborating institutions in Finland (n=33),

France (n=756), Greece (n=452), Italy (n=178), Spain (n=59) and Sweden (n=461)

were studied for IGHV repertoire and mutational status. All cases displayed the

typical CLL immunophenotype as described earlier25, 27 and met the diagnostic

criteria of the National Cancer Institute Working Group (NCI-WG)38. Written informed

consent was obtained according to the Helsinki declaration and the study was

approved by the local Ethics Review Committee of each institution.

PCR amplification of CLL IGH rearrangements

In the vast majority of cases (1797/1939 cases; 93%), peripheral blood samples were

analyzed; bone marrow (105 cases), lymph nodes (28 cases) and spleen specimens

(9 cases) were also analyzed. Amplification and sequence analysis of IGH

rearrangements were performed on either DNA or cDNA as previously described25, 27,

34, 37 or using the BIOMED-2 protocol39. Sequence data was analyzed using the IMGT

database and tools40-41. All sequences were in-frame; any partial sequences that did

not include the entire HCDR1 were excluded from the analysis.

Collection of non-CLL sequence data

Non-CLL IGH sequences were retrieved from the IMGT/LIGM-DB database in

August 2006. Stringent criteria were followed so that redundant, poorly annotated,

out-of-frame, incomplete or clonally-related sequences were excluded from the

analysis. The non-CLL cohort was intentionally diverse in order to offer the

opportunity for comparisons with various types of B cells. The final collection of 5303

unique IGHV-D-J sequences included (i) 447 sequences from B cell

lymphoproliferative disorders, (ii) 3235 sequences from normal B cells, (iii) 499

sequences from “immune dysregulation” disorders (allergy, asthma, various types of

immunodeficiency), and (iv) 1122 sequences from autoreactive cells (Supplemental

Table 1).

Sequence analysis and data mining

Both CLL and non-CLL sequence sets were submitted to the IMGT V-QUEST

analysis software41 to obtain gene and allele usage and mutation data. The following

information was extracted:

1. IGHV gene usage, percentage of identity to germline, and HCDR3 length

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

7

Output data from IMGT V-QUEST for both CLL and non-CLL sequence sets were

parsed, re-organized, and exported to a spreadsheet through the use of computer

programming with the Perl programming language. IGHV, IGHD and IGHJ gene

usage, allele usage, percentage of identity to germline, and the HCDR3 length were

recorded for each sequence.

2. Somatic hypermutation characteristics

Each nucleotide mutation in every sequence was recorded, as was the change or

preservation of the corresponding amino acid (AA), identified as replacement (R) or

silent (S), respectively. Amino acids were grouped into one of five categories,

compiled according to standardized biochemical criteria42 and based on

physicochemical properties (hydropathy, volume, chemical characteristics)43: (i) non-

polar/aliphatic: G, A, P, V, L, I, M; (ii) polar, uncharged: S, T, C, N, Q; (iii) basic: K, R,

H; (iv) acidic: E, D; (v) aromatic: F, Y, W.

In order to account for the fact that a mutation is more likely to occur in a HFR than a

HCDR simply due to its greater length, each mutation was ‘weighted’, or normalized,

by the codon length of the region in which it occurred, e.g. an AA mutation in a

HCDR1 of length 8 would be assigned a weight of 1/8, or 0.13. Subsequently, in

order to compare mutation distributions between groups (IGHV genes, subsets etc),

the sum of the normalized mutation counts per HFR/HCDR was expressed as a

percentage of the total normalized mutation counts in the group. We describe these

values as the “normalized distribution percentages” throughout Results.

Consequently, it was possible to compare mutation data (e.g. total mutations/R

mutations/S mutations) per region (e.g. HCDR2, HFR3) or combinations of regions

(HCDR1 and HCDR2), within/across different groupings of sequences (e.g. individual

IGHV genes, homologous subsets and CLL vs. non-CLL sequences).

We extracted additional information on all AA changes codon-by-codon and

examined whether the somatically introduced AA belonged to the same biochemical

category as the mutating AA (“conservative” change) or not (“non-conservative”

change).

3. Hotspot targeting.

Mutated sequences were also analyzed for targeting to the tetranucleotide (4-NTP)

motifs RGYW/WRCY (R=A/G, Y=C/T, and W=A/T)44 and DGYW/WRCH (D=A/G/T,

H=T/C/A)45. To account for differences in germline composition, counts were

normalized by evaluating the number of 4-NTP mutations per HCDR/HFR nucleotide

length per 4-NTP position for each sequence.

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

8

Statistical analysis

Descriptive statistics for discrete parameters included counts and frequency

distributions. For quantitative variables, statistical measures included means,

medians, standard deviation and min–max values. Significance of bivariate

relationships between factors was assessed with the use of Chi-square and Fisher’s

exact tests. For all comparisons, a significance level of p=0.05 was set and all

statistical analyses were performed with the use of the Statistical Package SPSS

Version 12.0 (SPSS Inc, Chicago, USA).

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

9

RESULTS

IGHV repertoire and mutation status

A total of 1967 in-frame IGHV-D-J sequences obtained from 1939 CLL patients were

included in the analysis; 28 patients carried double in-frame rearrangements. Overall,

this large and geographically diverse series confirmed previously published IGHV

repertoire data obtained in smaller series24-27, 33 (Supplemental Table 2).

Following the 98% identity cut-off value, which is used to make the clinically relevant

distinction between “mutated” and “unmutated” CLL cases15-19, 1064/1967 sequences

(54%) from our series were defined as “mutated”, whereas the remainder (903/1967

sequences, 46%) had “unmutated” IGHV genes. Of note, concordant mutational

status was observed in both IGHV-D-J rearrangements in 15/28 cases with double

in-frame rearrangements; in the remaining 13 cases, the two rearrangements had

different mutational status.

We subdivided “unmutated” sequences into a “truly unmutated” subgroup, which

included 677/1967 sequences (34.4%) with IGHV genes in germline configuration

(100% identity), a “minimally mutated” subgroup, which included 133/1967

sequences (6.8%) with 99-99.9% identity to germline, and a “borderline mutated”

subgroup, which included 93/1967 sequences (4.7%) with 98-98.9% identity to

germline. The IGHV repertoires of the “mutated”, “minimally mutated”, “borderline

mutated” and “truly unmutated” subgroups differed (Supplemental Table 3), in

keeping with previous reports24-27, 33. At the individual gene level, the distribution of

rearrangements of IGHV genes according to mutation status varied significantly

(Figure 1 / Supplemental Table 4). In particular, the IGHV1-69 and IGHV1-2 genes

predominated among, respectively, “truly unmutated” and “minimally mutated”

sequences. In contrast, other IGHV genes were mostly utilized in “mutated” (<98%

identity) rearrangements (e.g., IGHV4-34, IGHV3-23, IGHV3-7). Finally, the IGHV3-

21 and IGHV3-48 genes had the highest proportion of “borderline mutated” (98-

98.9% identity) rearrangements. Significant differences were also observed with

regard to mutation status among groups of sequences utilizing different alleles (39) of

certain IGHV genes, in particular IGHV1-69, IGHV4-39 and IGHV3-30 (Supplemental

Table 5).

“Truly unmutated” sequences had significantly longer HCDR3s (median 21 AA, range

4-32 AA) than all other sequences; a significant difference in HCDR3 length was also

observed among “minimally mutated” (median 19 AA, range 9-29) and “borderline

mutated” or “mutated” sequences (median 15 AA for both groups, range 9-30 AA)

(Figure 2) (p<0.001 for all comparisons).

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

10

Targeting of somatic hypermutation

Nucleotide substitution analysis was performed for all CLL sequences of the present

series with <100% identity to germline. Of the 18149 mutations analyzed, transitions

predominated (10219/18149, or 56.3%), in keeping with a canonical SHM process.

However, at the level of individual IGHV genes, IGHV3-21 rearrangements showed

distinctive features. In particular, compared to all other IGHV3 subgroup genes,

IGHV3-21 rearrangements showed: (i) significantly fewer G-to-A substitutions (12.6%

vs. 17.2%; p<0.01); (ii) significantly more T-to-A substitutions (14% vs. 7.8%;

p<0.001). As revealed by comparison to non-CLL IGHV3-21 sequences, the over-

representation of the T-to-A substitution was “IGHV3-21/CLL-biased”.

SHM frequencies in the HFRs and HCDRs were calculated for all IGHV subgroups.

Here, as in all analyses, the normalized distribution percentages (as described in the

Patients and Methods section) were employed. Examination of the three largest

IGHV subgroups (IGHV1/3/4) revealed markedly different SHM targeting. Overall,

there was a greater targeting of R mutations to the HCDRs (especially HCDR2) of

IGHV3 sequences compared to IGHV1 and IGHV4 sequences (Supplemental Table

6). At the level of individual genes of the IGHV1/3/4 subgroups, the highest

normalized R/S mutation ratios in HCDRs were observed among sequences utilizing

the IGHV4-59, IGHV3-15, IGHV4-4, IGHV3-21 and IGHV3-33 genes. In contrast, the

lowest R/S mutation ratios in HCDRs were seen among IGHV4-39, IGHV4-34 and

IGHV3-48 sequences (Supplemental Tables 7-8).

In particular, within the HCDR2, IGHV3-21 sequences had the highest R mutation

targeting and the lowest S mutation targeting relative to all other genes. IGHV3-21

sequences also carried the lowest R mutation frequencies in all three FRs.

Conversely, IGHV4-34 sequences displayed the lowest R mutation frequency as well

as the lowest R/S mutation ratio in HCDR2. As revealed by comparison to IGHV4-34

sequences from normal and autoreactive cells, the paucity of R mutations in HCDR2

is a “CLL-biased” feature (Figure 3).

A significantly higher clustering of R mutations to 4-NTP motifs in the HCDR2 were

observed among IGHV3- vs. IGHV1- or IGHV4-expressing sequences (p<0.01). A

significant bias for R mutation targeting to 4-NTPs was also evident in HFR3 of

IGHV4-expressing sequences, as exemplified by markedly different targeting for AA

changes of two consecutive, alternative serine codons. In particular, the AGC codon

(“the hottest of SHM hotspots”46-47) at IMGT/HFR3-92 carried an AA change in 59%

of mutated IGHV4 sequences, whereas the TCT codon at position IMGT/HFR3-93

carried an AA change in only 4% of sequences. Of note, the targeting of the AGC

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

11

serine codon at IMGT/HFR3-92 was significantly higher in CLL vs. normal vs.

autoreactive IGHV4 sequences (59% vs. 39% vs. 23.6%; p<0.05).

Recurrent amino acid changes in subsets of CLL cases expressing stereotyped

HCDR3 sequences

Analysis of sequences from the present series following previously described

criteria34 allowed us to identify 530/1967 sequences (26.9%) as belonging to 110

different subsets with stereotyped HCDR3 (Supplemental Table 9), of which 48 have

been reported previously27-34; each subset included from two up to 56 cases. The

frequency of sequences carrying a stereotyped HCDR3 was significantly higher

among “truly unmutated” or “minimally mutated” (43.4% and 36.7%, respectively) vs.

“borderline mutated” (24.7%) vs. “mutated” (15.5%) sequences (p<0.001 for all

comparisons).

Shared (“stereotyped”) AA changes (i.e., the same AA replacement at the same

position) across the whole IGHV gene sequence were identified for subsets of CLL

sequences with stereotyped HCDR3s. As revealed by comparison of the CLL vs.

non-CLL datasets, certain AA changes could be considered as “CLL-biased”.

Furthermore, for certain IGHV genes, many stereotyped AA changes occurred

significantly more frequently in cases with stereotyped rather than heterogeneous

HCDR3 sequences and, therefore could be considered as “subset-biased” (Table 1).

A comprehensive list of such stereotyped AA changes is provided in Supplemental

Table 10. The most striking “CLL-biased” hypermutations were observed in the

following subsets of sequences with stereotyped HCDR3s:

(1) Nineteen sequences from the present series utilizing allele *02 of the IGHV1-2

gene belonged to two subsets with stereotyped HCDR3s32-34. The first subset (#1)

included 53 minimally mutated / truly unmutated sequences which utilized IGHV

genes of the same clan (IGHV1-2/IGHV1-3/IGHV1-18, IGHV5-a, IGHV7-4-1). Among

15 IGHV1-2*02-expressing sequences of this subset, 9 had 100% identity to

germline whereas 6 were found to carry a single replacement mutation, leading to a

W-to-R change at IMGT/HFR2-55 (Figure 4A). The second subset (#28) included 5

IGHV1-2 sequences with stereotyped HCDR3s of which one utilized allele *01 and

had 100% identity to germline, whereas four utilized allele *02 (as previously

described33-34) and carried the same single replacement mutation as described above

for the subset #1. Comparison of “subset” IGHV1-2*02 sequences to CLL IGHV1-

2*02 sequences with heterogeneous HCDR3 or non-CLL IGHV1-2*02 sequences

demonstrated that the W-to-R change was “subset-biased”. In two cases of this

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

12

subset, germline sequence analysis of the IGHV1-2 gene confirmed that the W-to-R

change was generated somatically and, thus, did not represent a polymorphism.

(2) Fifty-six IGHV3-21 sequences with stereotyped HCDR3s belonged to

subset#227,29,32-34. In this subset, four different recurrent mutations were observed at a

frequency of 15-32% (Figure 4B). Comparison to CLL IGHV3-21 sequences with

heterogeneous HCDR3s or non-CLL IGHV3-21 sequences demonstrated that AA

changes (3) and (4) were “subset-biased” (Table 1). Remarkably, within CLL, subset

#2 cases had a higher targeting of the HCDR2 than non-subset-#2 IGHV3-21 cases

(Supplemental Table 11).

(3) Among a group of 27 IGHV4-34 sequences with stereotyped HCDR3s which

belonged to two different subsets (#4, #16)32-34, 36, four different recurrent mutations

were observed at a frequency of 35-100% (Figures 4C-4D). Noticeably, comparison

to CLL IGHV4-34 sequences with heterogeneous HCDR3 or non-CLL IGHV4-34

sequences demonstrated that three of the four stereotyped AA changes were

“subset-biased” (Table 1). Similar to subset #2, subset #4 and subset#16 sequences

also showed distinctive SHM distribution “profiles” in the HCDRs/HFRs compared to

IGHV4-34 sequences with heterogeneous HCDR3s. In particular, subset #4 IGHV4-

34 sequences displayed a notably higher targeting of HFR2 and HCDR1 than

IGHV4-34 sequences with heterogeneous HCDR3s; subset #16 cases also

demonstrated a notably higher targeting of the HCDR1 than IGHV4-34 sequences

with heterogeneous HCDR3s (Supplemental Table 11).

(3) Among a subset of four IGHV4-4-expressing sequences with stereotyped

HCDR3s (subset #14)34, six different recurrent mutations were observed in 75-100%

cases (Figure 4-E). Comparison to CLL IGHV4-4 sequences with heterogeneous

HCDR3s or non-CLL IGHV4-4 sequences demonstrated that all the above AA

changes were “subset-biased” (Table 1).

Mutation targeting of superantigenic-binding motifs

(1) A total of 706 IGHV3-expressing cases with <100% identity to germline were

examined for SHM targeting to the IGHV3-specific motif responsible for

Staphylococcal protein A (SpA) binding, which is mediated by a conformational

surface generated by AAs at 13 positions in the V region of IGHV3 subgroup genes5.

Non-conservative residue variations at 2 or more positions of this motif result in loss

of SpA binding activity5. Overall, such variations were observed in 80/706 IGHV3-

expressing cases (11.3%). Remarkably, significantly fewer changes were identified in

rearrangements utilizing the IGHV3-21 vs. all other IGHV3 subgroup genes [13/79

(16%) vs. 377/627 cases (60%), p<0.01]. Furthermore, the few AA changes that did

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

13

occur in IGHV3-21 rearrangements (in particular those carrying a stereotyped

HCDR3) tended to be conservative; only 2.5% of IGHV3-21 sequences (2/79) carried

two or more non-conservative AA changes of the motif and neither of these belonged

to subset #2. In contrast, though also relatively infrequent, up to three quarters of AA

changes identified in rearrangements of other IGHV3 genes (even those with a

similar mutation load as the IGHV3-21 rearrangements) could be non-conservative.

(2) A total of 126 IGHV4-34 sequences with <100% identity to germline were

examined for SHM targeting to the IGHV4-34-specific motif responsible for

carbohydrate I binding, which is mediated by a hydrophobic patch in HFR1 involving

residue W7 on β-strand A and the AVY motif (residues 24–26) on β-strand B48.

Notably, few IGHV4-34 sequences were altered at the four positions of the anti-I/i

motif. Overall, there were only 0.9-4.9% non-conservative AA changes at these

codon positions and only one sequence had an AA change at more than one of the

motif positions.

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

14

DISCUSSION

In the present study, 1967 IGHV-D-J sequences from 1939 patients with CLL were

analyzed for SHM patterns and compared to public non-CLL sequences from the

IMGT database. Our series consisted of mutated and unmutated sequences at a

frequency reported as typical for CLL18-19,24, 26-27.

The gene repertoire of “truly unmutated” (100% identity to germline) CLL sequences

of the present series (n=677) was extremely skewed and also characterized by

significantly longer HCDR3s. Furthermore, 43.4% of “truly unmutated” sequences

were found to belong to a subset with stereotyped HCDR3s. These observations

suggest that the unmutated state in CLL could reflect selective pressures for

maintaining germline configuration28,49.

Unmutated BCRs of CLL B cells have recently been shown to be associated with

autoreactivity and polyreactivity against molecules such as DNA, insulin and LPS,

whereas BCRs in mutated CLL did not exhibit these polyreactive properties50.

Furthermore, as previously shown, the antigen binding site excluding the HCDR3 is

exceptionally cross-reactive, at least until acted on by SHM51-52. Based on the

findings of the aforementioned studies and the results of the present study, it could

perhaps be reasonable to speculate that unmutated BCRs with multiple specificities

may provide CLL progenitors with a selective advantage because they widen the

spectrum of potential antigenic stimuli53-54.

Previous studies in both normal and autoreactive B cells have shown that even a few

mutations may be functionally relevant55-57. Along these lines, in the present study,

we also explored potential biological implications of low mutational “load” in CLL.

Therefore, SHM analysis was undertaken for the cohort of all 1290 sequences of the

present series with <100% identity to germline. At the cohort level, SHM patterns

were typical of a canonical SHM process6,46-47,58-59. However, important differences

emerged from the analysis of SHM in subgroups of CLL sequences defined by: (i)

IGHV gene usage; (ii) HCDR3 length and degree of HCDR3 stereotypy; and, (iii)

minimal vs. borderline vs. high mutation load.

Evidence for very precise SHM targeting was obtained by the evaluation of SHM

patterns in different alleles of certain IGHV genes, indicating preferential selection of

one allele over another. Remarkably, within the group of rearrangements utilizing the

IGHV1-69 gene, 87% of sequences expressing the *01 allele were “truly unmutated”

vs. only 50% of sequences expressing the *06 allele; yet, these two alleles differ from

each other by only one AA at codon 82 (glutamic acid in IGHV1-69*01 / lysine in

IGHV1-69*06). Furthermore, all “minimally mutated” IGHV1-2 sequences of subsets

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

15

#1 and #28, which carried as a single mutation the tryptophane-to-arginine (W-to-R)

change at IMGT-HFR2 codon 55, expressed allele *02 of the IGHV1-2 gene. This

change causes the IG sequence to become more like the IGHV1-2*01 allele, since

an arginine at that position is only present in the germline configuration of the IGHV1-

2*01 allele. Of note, within the comparable non-CLL group, 10/17 IGHV1-2*02

sequences carrying this mutation encoded autoantibodies, of which 7 were

rheumatoid factors (Supplemental Table 12). These findings illustrate that even very

slight alterations in IG sequence appear to be selected for, perhaps because they

may confer a clonal advantage.

At the level of individual IGHV genes, the most distinctive, often “CLL-biased”, SHM

patterns were observed in groups of sequences utilizing the IGHV3-21 and IGHV4-34

genes. Although frequently mutated, almost a quarter of IGHV3-21 cases in our

series had a low mutation load and fell into the ‘borderline/minimally mutated’ group.

The distribution of R mutations and the nucleotide substitution spectra of IGHV3-21

sequences differed significantly from other IGHV3 genes. Of note, IGHV3-21

sequences with stereotyped HCDR3s belonging to subset #2 showed 0.8-2.4 fold

lower targeting of all regions (except HCDR2) than non-subset-#2 IGHV3-21

sequences. Furthermore, several recurrent AA changes were observed among

subset #2 IGHV3-21 sequences, in particular at HCDR2 codons. Remarkably, a

serine deletion at IMGT/HCDR2 codon 59 was detected in 18 IGHV3-21 CLL

sequences, all expressing stereotyped BCRs. This finding confirms and extends a

recent report from our group, which first suggested that this deletion is “CLL-biased”

37. Therefore, while IGHV3-21 sequences are generally less targeted by SHM than

other IGHV3 genes, the observed mutations appear to be very precisely and

effectively targeted, indicating selection by specific antigen(s). Along these lines, it is

also perhaps relevant that IGHV3-21 sequences from our series, in particular those

carrying stereotyped HCDR3s, showed a strong tendency to retain germline

configuration in the binding motif for Staphylococcal protein A, the prototype for a

class of naturally arising proteins that have the properties of model B-cell

superantigens5. At present, the biological and clinical implications of this observation

(if any) remain unknown.

The IGHV4-34 gene encodes antibodies which are intrinsically autoreactive in the

germline state by virtue of recognition of the N-acetyllactosamine (NAL) antigenic

determinant of the I/i blood group antigen60. Anti-I/i IGHV4-34 antibodies also bind

the linear poly-NAL in the B cell isoform of CD4560. The I/i antigen may be expressed

in oxidized apoptotic cells and CD45 is expressed by pre-apoptotic T cells61-62; these

findings explain why IGHV4-34 antibodies bind apoptotic cells63. B cells whose

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

16

surface receptors bind to apoptotic cells may serve ‘‘housekeeping’’ functions by

removing cellular debris64. Thus, it is possible that immature B cells expressing

IGHV4-34 participate in the removal of apoptotic cell remnants. However, given the

remarkable cross-reactivity of IGHV4-34 antibodies against several auto- and exo-

antigens65-67, if immature IGHV4-34-expressing B cells participate in the uptake of

apoptotic cell remnants in the bone marrow, at the same time, they must be

undergoing modifications to ablate self-reactivity68. These modifications may be

introduced by somatic diversification mechanisms, such as SHM and receptor

editing66, 69-70. In the present study, 79% of IGHV4-34 CLL sequences were mutated,

in keeping with previous reports in smaller series24,26-27,34. In line with the reasoning

presented above, this trend might reflect the fact that IGHV4-34 sequences must

undergo SHM in order to negate their autoreactivity and be sufficiently “safe” to be

allowed into the functioning IG repertoire.

Previous studies have demonstrated that the region of the IGHV4-34 molecule that

cross-reacts with the I antigen is a hydrophobic patch in HFR1 created by a

discontinuous sequence involving a W residue at codon 7 and the AVY triplet at

codons 24-2648. On examination of the anti-I/i-binding motif in the HFR1 of IGHV4-

34 CLL sequences from our series, we observed that each of the four positions of the

W-AVY motif was very infrequently mutated. Most interesting, however, was the fact

that none of subset #4 or subset #16 IGHV4-34 sequences were among those

carrying an altered motif. Thus, in theory, these IGHV4-34-expressing CLL cells

could still be bound (and stimulated for clonal expansion) by I/i antigens or the CD45

on B cells, similar to what has been reported previously for normal B cells71. In this

context, Catera et al recently demonstrated that three IGHV4-34 recombinant CLL

antibodies with stereotyped BCRs, similar to our subset #4 sequences, bound to

viable B cells via the NAL epitope72.

HCDR3 sequence motifs enriched in basic amino acids have been shown to

correlate strongly with reactivity of IGHV4-34 antibodies against both B cells and

DNA73-75. All subset #4 IGHV4-34 CLL sequences from our series have high HCDR3

isoelectric point values and all carry a couplet of basic residues (arginine-arginine or

arginine-lysine) at the IGHD-IGHJ junction. High isoelectric point, overall positive

charge, and increased numbers of arginine residues are frequent features of many

pathogenic anti-DNA antibodies57,76-78. Although it is not possible to accurately predict

IG specificity by sequence analysis alone, these findings suggest that subset #4

BCRs may have anti-DNA specificity.

In transgenic mouse model systems, introduction of acidic residues (particularly

aspartic acid) by SHM is a means to edit anti-DNA reactivities56,69,79. A remarkable

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

17

analogy can be drawn with SHM patterns observed in CLL sequences of subsets #4

and #16 from our series. Aspartic and glutamic acid residues introduced by SHM

were observed with a high frequency in the HCDR1 of these sequences. Along these

lines, it would be tempting to speculate that modification of subset #4 and #16

IGHV4-34 sequences by SHM in precursors of the CLL clones significantly reduced

or eliminated the postulated anti-DNA reactivity. This hypothesis is supported by the

study of Herve et al50, in which unmutated revertant antibodies engineered from

mutated IGHV4-34 recombinant antibodies of CLL patients, similar to subset #4

antibodies from the present series, showed increased HEp-2 reactivity and/or

acquired polyreactivity. Therefore, the SHM patterns observed among IGHV4-34 CLL

sequences, in particular those expressed by subset #4 and #16 cases, may induce a

state of diminished responsiveness towards a selecting antigenic element. However,

these IGHV4-34 clones could retain the ability to engage in superantigen-like

interactions with various auto- and exo-antigens via their preserved (non-mutated)

HFR1 motifs. Therefore, in principle, CLL progenitors could be activated or “kick-

started” on infection or reactivation by certain microbial pathogens (CMV or EBV

might be such pathogens80-84) and thus receive signals promoting survival,

expansion, malignant transformation, and potentially clonal evolution.

In conclusion, groups of patients with CLL utilizing certain IGHV genes - in particular

subsets grouped according to HCDR3 composition - evidently carry shared,

“stereotyped” mutations across the entire IGHV gene sequence. Furthermore, the

mutation pattern within these subgroups was not only gene- and subset-biased, but

also, in most cases, “CLL-biased”. The finding of such “stereotyped” mutations in

mutated CLL sequences carrying stereotyped HCDR3s indicates that the leukemic

progenitor cells may have responded in a similar fashion to the selecting antigen(s).

Remarkably, as shown in the present study, selection for individual mutations is

evident even in subsets with minimally mutated sequences, indicating a functional

purpose for these modifications. Finally, the presence of stereotyped mutations is

strong evidence that not only the HCDR3 but also other regions of the IG molecule

could actively participate in antigen recognition and thus be involved in the

development and evolution of the CLL clone.

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

18

ACKNOWLEDGEMENTS

The authors would like to sincerely thank Prof. Marie-Paule Lefranc and Dr.

Veronique Giudicelli, Laboratoire d’Immunogenetique Moleculaire, LIGM, Universite

Montpellier II, Montpellier, France, for their enormous support and help with the large

scale immunoglobulin sequence analysis throughout this project. We also wish to

thank Prof. Göran Roos, Department of Medical Biosciences, Umeå University,

Sweden, Prof. Christer Sundström, Department of Genetics and Pathology, Uppsala

University, Sweden), Dr. Mats Merup, Department of Medicine, Karolinska University

Hospital, Huddinge, Sweden, Dr. Lyda Osorio, Department of Microbiology, Tumour

and Cell Biology, Karolinska Institutet, Stockholm, Sweden and Prof. Juhani Vilpo,

Laboratory Centre, Tampere University Hospital, Tampere, Finland for providing

samples and clinical data concerning Swedish and Finnish CLL patients; and, Dr.

Hedda Wardemann, Max-Planck Institute for Infection Biology, Berlin, Germany, for

her provision of antibody sequences of IgG+ memory B cells from healthy donors.

We also acknowledge the contribution of Dr Ulf Thunberg, Dr. Tatjana Smilevska,

Maria Norberg, Arifin Kaderi, Ingrid Thörn and Kerstin Willander to the sequence

analysis.

This work was supported by the Swedish Cancer Society, the Swedish Research

Council, Medical Faculty of Uppsala University, Uppsala University Hospital, Lion's

Cancer Research Foundation in Uppsala, Sweden; the Networks of Excellence

BioSapiens (contract number LSHG-CT-2003-503265) and ENFIN (contract number

LSHG-CT-2005-518254), both funded by the European Commission (Computational

Genomics Unit, Thessaloniki, Greece); the Associazione Italiana per la Ricerca sul

Cancro (AIRC, Milano), the Italian Ministry of Foreign Affairs, the CLL Global

Research Foundation (Milano, Italy). ND is supported by an ENTER career

development award from the General Secretariat for Research and Technology,

Greek Ministry of Development. CS is a recipient of a fellowship from the Foundation

Anna Villa e Felice Rusconi, Varese, Italy.

FM, ND and AH performed research, analyzed data, and wrote the paper. GT

performed research and wrote the paper. MB, CS, KK, FB-M, CM and DV performed

research. NL, AA and FC-C provided samples and associated data. AT and CO

supervised research. CB, PG, FD, RR and KS designed and supervised the research

and wrote the paper. The authors declare no competing financial interests.

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

19

REFERENCES

1. Maizels N. Immunoglobulin gene diversification. Annu Rev Genet. 2005;39:23-46. 2. Davies DR, Padlan EA, Sheriff S. Antibody-antigen complexes. Annu Rev Biochem.

1990;59:439-473. 3. Davies DR, Cohen GH. Interactions of protein antigens with antibodies. Proc Natl

Acad Sci USA. 1996;93:7-12. 4. Xu JL, Davis MM. Diversity in the CDR3 region of V(H) is sufficient for most antibody

specificities. Immunity. 2000;13:37-45. 5. Silverman GJ, Goodyear CS. A model B-cell superantigen and the immunobiology of

B lymphocytes. Clin Immunol. 2002;102:117-134. 6. Di Noia JM, Neuberger MS. Molecular Mechanisms of Antibody Somatic

Hypermutation. Annu Rev Biochem. 2007;76:1-22. 7. Manser T. Textbook germinal centers? J Immunol. 2004;172:3369-3375. 8. De Vinuesa CG, Cook M, Ball C, et al. Germinal centers without T cells. J Exp Med.

2000;191:485-494. 9. Weller SA, Faili C, Garcia MC, et al. CD40-CD40L independent Ig gene

hypermutation suggests a second B cell diversification pathway in humans. Proc Natl Acad Sci USA. 2001;98:1166-1170.

10. William J, Euler C, Christensen S, Shlomchik MJ. Evolution of autoantibody responses via somatic hypermutation outside of germinal centers. Science. 2002;297:2066-2070.

11. Weller S, Braun MC, Tan BK, et al. Human blood IgM "memory" B cells are circulating splenic marginal zone B cells harboring a prediversified immunoglobulin repertoire. Blood. 2004;104:3647-3654.

12. Kruetzmann S, Rosado MM, Weber H, et al. Human immunoglobulin M memory B cells controlling Streptococcus pneumoniae infections are generated in the spleen. J Exp Med. 2003;197:939-945.

13. Weller S, Reynaud CA, Weill JC. Vaccination against encapsulated bacteria in humans: paradoxes. Trends Immunol. 2005;26:85-89.

14. Chiorazzi N, Rai KR, Ferrarini M. Chronic lymphocytic leukemia. N Engl J Med. 2005;352:804-815.

15. Dighiero G. CLL Biology and Prognosis. Hematology Am Soc Hematol Educ Program. 2005:278-284.

16. Schroeder HW Jr, Dighiero G. The pathogenesis of chronic lymphocytic leukemia: analysis of the antibody repertoire. Immunol Today. 1994;15:288-94.

17. Ghia P, Stamatopoulos K, Belessi C, et al. European Research Initiative on CLL. ERIC recommendations on IGHV gene mutational status analysis in chronic lymphocytic leukemia. Leukemia. 2007;21:1-3.

18. Damle RN, Wasil T, Fais F, et al. Ig V gene mutation status and CD38 expression as novel prognostic indicators in chronic lymphocytic leukemia. Blood. 1999;94:1840-1847.

19. Hamblin TJ, Davis Z, Gardiner A, Oscier DG, Stevenson FK. Unmutated Ig V(H) genes are associated with a more aggressive form of chronic lymphocytic leukemia. Blood.1999;94:1848-1854.

20. Damle RN, Ghiotto F, Valetto A, et al. B-cell chronic lymphocytic leukemia cells express a surface membrane phenotype of activated, antigen-experienced B lymphocytes. Blood. 2002;99:4087-4093.

21. Stevenson FK, Caligaris-Cappio F. Chronic lymphocytic leukemia: revelations from the B-cell receptor. Blood. 2004;103:4389-4395.

22. Rosenwald A, Alizadeh AA, Widhopf G, et al. Relation of gene expression phenotype to immunoglobulin mutation genotype in B cell chronic lymphocytic leukemia. Journal of Experimental Medicine. 2001;194:1639-1647.

23. Klein U, Tu Y, Stolovitzky GA, et al. Gene expression profiling of B cell chronic lymphocytic leukemia reveals a homogeneous phenotype related to memory B cells. J Exp Med. 2001;194:1625-1638.

24. Fais F, Ghiotto F, Hashimoto S, et al. Chronic lymphocytic leukemia B cells express restricted sets of mutated and unmutated antigen receptors. J Clin Invest. 1998;102:1515-1525.

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

20

25. Tobin G, Thunberg U, Johnson A, et al. Somatically mutated Ig V(H)3-21 genes characterize a new subset of chronic lymphocytic leukemia. Blood. 2002;99:2262-2264.

26. Chiorazzi N, Ferrarini M. B cell chronic lymphocytic leukemia: lessons learned from studies of the B cell antigen receptor. Annu Rev Immunol. 2003;21:841-894.

27. Ghia P, Stamatopoulos K, Belessi C, et al. Geographic patterns and pathogenetic implications of IGHV gene usage in chronic lymphocytic leukemia: the lesson of the IGHV3-21 gene. Blood. 2005;105:1678-1685.

28. Widhopf GF 2nd, Kipps TJ. Normal B cells express 51p1-encoded Ig heavy chains that are distinct from those expressed by chronic lymphocytic leukemia B cells. J Immunol. 2001;166:95-102.

29. Tobin G, Thunberg U, Johnson A, et al. Chronic lymphocytic leukemias utilizing the VH3-21 gene display highly restricted Vlambda2-14 gene use and homologous CDR3s: implicating recognition of a common antigen epitope. Blood. 2003 Jun 15;101:4952-4957.

30. Ghiotto F, Fais F, Valetto A, et al. Remarkably similar antigen receptors among a subset of patients with chronic lymphocytic leukemia. J Clin Invest. 2004;113:1008-1016.

31. Widhopf GF 2nd, Rassenti LZ, Toy TL, Gribben JG, Wierda WG, Kipps TJ. Chronic lymphocytic leukemia B cells of more than 1% of patients express virtually identical immunoglobulins. Blood. 2004;104:2499-2504.

32. Messmer BT, Albesiano E, Efremov DG, et al. Multiple distinct sets of stereotyped antigen receptors indicate a role for antigen in promoting chronic lymphocytic leukemia. J Exp Med. 2004;200:519-525.

33. Tobin G, Thunberg U, Karlsson K, et al. Subsets with restricted immunoglobulin gene rearrangement features indicate a role for antigen selection in the development of chronic lymphocytic leukemia. Blood. 2004;104:2879-2885.

34. Stamatopoulos K, Belessi C, Moreno C, et al. Over 20% of patients with chronic lymphocytic leukemia carry stereotyped receptors: Pathogenetic implications and clinical correlations. Blood. 2007;109:259-270.

35. Thorselius M, Krober A, Murray F, et al. Strikingly homologous immunoglobulin gene rearrangements and poor outcome in VH3-21-utilizing chronic lymphocytic leukemia independent of geographical origin and mutational status. Blood. 2006;107:2889-2894.

36. Potter KN, Mockridge CI, Neville L, et al. Structural and functional features of the B-cell receptor in IgG-positive chronic lymphocytic leukemia. Clin Cancer Res. 2006;12:1672-1679.

37. Belessi CJ, Davi FB, Stamatopoulos KE, et al. IGHV gene insertions and deletions in chronic lymphocytic leukemia: "CLL-biased" deletions in a subset of cases with stereotyped receptors. Eur J Immunol. 2006;36:1963-1974.

38. Cheson BD, Bennett JM, Grever M, et al. National Cancer Institute-sponsored Working Group guidelines for chronic lymphocytic leukemia: revised guidelines for diagnosis and treatment. Blood. 1996;87:4990-4997.

39. van Dongen JJ, Langerak AW, Bruggemann M, et al. Design and standardization of PCR primers and protocols for detection of clonal immunoglobulin and T-cell receptor gene recombinations in suspect lymphoproliferations: report of the BIOMED-2 Concerted Action BMH4-CT98-3936. Leukemia. 2003;17:2257-2317.

40. Lefranc MP, Giudicelli V, Kaas Q, et al. IMGT, the international ImMunoGeneTics information system. Nucleic Acids Res. 2005;33(Database issue):D593-597.

41. Giudicelli V, Chaume D, Lefranc MP. IMGT/V-QUEST, an integrated software program for immunoglobulin and T cell receptor V-J and V-D-J rearrangement analysis. Nucleic Acids Res. 2004;32(Web Server issue):W435-440.

42. Nelson DL, Cox MM, eds. Lehninger Principles of Biochemistry. New York, NY: WH Freeman & Co; 2005.

43. Pommie C, Levadoux S, Sabatier R, Lefranc G, Lefranc MP. IMGT standardized criteria for statistical analysis of immunoglobulin V-REGION amino acid properties. J Mol Recognit. 2004;17:17-32.

44. Rogozin IB, Pavlov YI. Theoretical analysis of mutation hotspots and their DNA sequence context specificity. Mutat Res. 2003;544:65-85.

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

21

45. Rogozin IB, Diaz M. Cutting edge: DGYW/WRCH is a better predictor of mutability at G:C bases in Ig hypermutation than the widely accepted RGYW/WRCY motif and probably reflects a two-step activation-induced cytidine deaminase-triggered process. J Immunol. 2004;172:3382-3384.

46. Rogozin I, Kolchanov N. Somatic hypermutagenesis in immunoglobulin genes. II. Influence of neighbouring base sequences on mutagenesis. Biochim Biophys Acta.1992;1171:11-18.

47. Goyenechea B, Milstein C. Modifying the sequence of an immunoglobulin V-gene alters the resulting pattern of hypermutation. Proc Natl Acad Sci USA.1996; 93:13979–13984.

48. Potter KN, Hobby P, Klijn S, Stevenson FK, Sutton BJ. Evidence for involvement of a hydrophobic patch in framework region 1 of human V4-34-encoded Igs in recognition of the red blood cell I antigen. J Immunol. 2002;169:3777-3782.

49. Longo NS, Lipsky PE. Why do B cells mutate their immunoglobulin receptors? Trends Immunol. 2006;27:374-380.

50. Herve M, Xu K, Ng YS, et al. Unmutated and mutated chronic lymphocytic leukemias derive from self-reactive B cell precursors despite expressing different antibody reactivity. J Clin Invest. 2005;115:1636-1643.

51. Mouthon L, Nobrega A, Nicolas N, et al. Invariance and restriction toward a limited set of self-antigens characterize neonatal IgM antibody repertoires and prevail in autoreactive repertoires of healthy adults. Proc Natl Acad Sci USA. 1995;92:3839-3843.

52. Jang YJ, Stollar BD. Anti-DNA antibodies: aspects of structure and pathogenicity. Cell Mol Life Sci. 2003;60:309-320.

53. Claflin JL, Berry J. Genetics of the phosphocholine-specific antibody response to Streptococcus pneumoniae. Germ-line but not mutated T15 antibodies are dominantly selected. J Immunol. 1988;141:4012-4019.

54. Hangartner L, Zinkernagel RM, Hengartner H. Antiviral antibody responses: the two extremes of a wide spectrum. Nat Rev Immunol. 2006;6:231-243.

55. Barbas SM, Ditzel HJ, Salonen EM, Yang WP, Silverman GJ, Burton DR. Human autoantibody recognition of DNA. Proc Natl Acad Sci USA. 1995;92:2529-2533

56. Cocca BA, Seal SN, D'Agnillo P, et al. Structural basis for autoantibody recognition of phosphatidylserine-beta 2 glycoprotein I and apoptotic cells. Proc Natl Acad Sci USA. 2001;98:13826-13831.

57. Rahman A, Giles I, Haley J, Isenberg D. Systematic analysis of sequences of anti-DNA antibodies-relevance to theories of origin and pathogenicity. Lupus. 2002;11:807-23.

58. Zheng NY, Wilson K, Jared M, Wilson PC. Intricate targeting of immunoglobulin somatic hypermutation maximizes the efficiency of affinity maturation. J Exp Med. 2005;201:1467-1478.

59. Clark LA, Ganesan S, Papp S, van Vlijmen HW. Trends in antibody sequence changes during the somatic hypermutation process. J Immunol. 2006;177:333-340.

60. Silberstein LE, George A, Durdik JM, Kipps TJ. The V4-34 encoded anti-i autoantibodies recognize a large subset of human and mouse B-cells. Blood Cells Mol Dis. 1996;22:126-138.

61. Beppu M, Ando K, Saeki M, Yokoyama N, Kikugawa K. Binding of oxidized Jurkat cells to THP-1 macrophages and antiband 3 IgG through sialylated poly-N-acetyllactosaminyl sugar chains. Arch Biochem Biophys. 2000;384:368-374.

62. Renno T, Attinger A, Rimoldi D, Hahne M, Tschopp J, MacDonald HR. Expression of B220 on activated T cell blasts precedes apoptosis. Eur J Immunol. 1998;28:540-547.

63. Pugh-Bernard A, Hocknell K, Cappione A, Anolik J, Sanz I. VH4–34 anti-I/i autoantibodies recognize apoptotic cells. Arthritis Rheum. 2000;46:S126.

64. Shaw PX, Horkko S, Chang MK, et al. Natural antibodies with the T15 idiotype may act in atherosclerosis, apoptotic clearance, and protective immunity. J Clin Invest. 2000;105:1731-1740.

65. Bhat NM, Bieber MM, Chapman CJ, Stevenson FK, Teng NN. Human antilipid A monoclonal antibodies bind to human B cells and the i antigen on cord red blood cells. J Immunol. 1993;151:5011-5021.

66. Spellerberg MB, Chapman CJ, Mockridge CI, Isenberg DA, Stevenson FK. Dual recognition of lipid A and DNA by human antibodies encoded by the VH4-21 gene: a

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

22

possible link between infection and lupus. Hum Antibodies Hybridomas. 1995;6:52-56.

67. Thomas MD, Clough K, Melamed MD, et al. A human monoclonal antibody encoded by the V4-34 gene segment recognises melanoma-associated ganglioside via CDR3 and FWR1. Hum Antibodies. 1999;9:95-106.

68. Pugh-Bernard AE, Silverman GJ, Cappione AJ, et al. Regulation of inherently autoreactive VH4-34 B cells in the maintenance of human B cell tolerance. J Clin Invest. 2001;108:1061-1070.

69. Li Y, Li H, Ni D, Weigert M. Anti-DNA B cells in MRL/lpr mice show altered differentiation and editing pattern. J Exp Med. 2002;196:1543-1552.

70. Li H, Jiang Y, Prak EL, Radic M, Weigert M. Editors and editing of anti-DNA receptors. Immunity. 2001;15:947-957.

71. Bhat NM, Bieber MM, Spellerberg MB, Stevenson FK, Teng NN. Recognition of auto- and exoantigens by V4-34 gene encoded antibodies. Scand J Immunol. 2000;51:134-140.

72. Catera R, Hatzi K, Chu C, et al. Polyreactive Monoclonal Antibodies Synthesized by Some B-CLL Cells Recognize Specific Antigens on Viable and Apoptotic T Cells [abstract]. Blood 2006;108:2813.

73. Li Y-C, Spellerberg MB, Stevenson FK, Capra JD, Potter KN. The I binding specificity of human VH4-34 (VH4.21) encoded antibodies is determined by both VH framework region 1 and complementarity determining region 3. J Mol Biol 1996;256:577–589.

74. Bhat NM, Bieber MM, Hsu FJ, et al. Rapid cytotoxicity of human B lymphocytes induced by VH4-34 (VH4.21) gene encoded monoclonal antibodies, II. Clin Exp Immunol. 1997;108:151–159.

75. Bhat NM, Bieber MM, Spellerberg MB, Stevenson FK, Teng NN. Recognition of auto- and exoantigens by V4-34 gene encoded antibodies. Scand J Immunol. 2000;51:134-140.

76. Jang YJ, Stollar BD. Anti-DNA antibodies: aspects of structure and pathogenicity. Cell Mol Life Sci. 2003;60:309-320.

77. Li Z, Schettino EW, Padlan EA, Ikematsu H, Casali P. Structure-function analysis of a lupus anti-DNA autoantibody: central role of the heavy chain complementarity-determining region 3 Arg in binding of double- and single-stranded DNA. Eur J Immunol. 2000;30:2015-2026.

78. Krishnan MR, Jou NT, Marion TN. Correlation between the amino acid position of arginine in VH-CDR3 and specificity for native DNA among autoimmune antibodies. J Immunol. 1996;157:2430-2439.

79. Behrendt M, Partridge LJ, Griffiths B, Goodfield M, Snaith M, Lindsey NJ. The role of somatic mutation in determining the affinity of anti-DNA antibodies. Clin Exp Immunol. 2003;131:182-189.

80. Chapman CJ, Spellerberg MB, Hamblin TJ, Stevenson FK. Pattern of usage of the VH4-21 gene by B lymphocytes in a patient with EBV infection indicates ongoing mutation and class switching. Mol Immunol 1995;32:347–353.

81. McClain MT, Heinlen LD, Dennis GJ, Roebuck J, Harley JB, James JA. Early events in lupus humoral autoimmunity suggest initiation through molecular mimicry. Nat Med. 2005;11:85-89.

82. Sundar K, Jacques S, Gottlieb P, et al. Expression of the Epstein-Barr virus nuclear antigen-1 (EBNA-1) in the mouse can elicit the production of anti-dsDNA and anti-Sm antibodies. J Autoimmun. 2004;23:127-140.

83. Isenberg D, Spellerberg M, Williams W, Griffiths M, Stevenson F. Identification of the 9G4 idiotope in systemic lupus erythematosus. Br J Rheumatol. 1993;32:876-882.

84. Isenberg DA, McClure C, Farewell V, et al. Correlation of 9G4 idiotope with disease activity in patients with systemic lupus erythematosus. Ann Rheum Dis. 1998;57:566-570.

85. Schneider T, Stephens R. Sequence logos: a new way to display consensus sequences. Nucleic Acids Res.1990;18:6097-6100.

86. Beitz E. Subfamily logos: visualization of sequence deviations at alignment positions with high information content. BMC Bioinformatics. 2006;7:313.

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

23

FIGURES

Figure 1. Distribution of rearrangements of the 10 most frequent IGHV genes of the

present series according to mutational status.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

IGHV1-2 IGHV1-69 IGHV3-21 IGHV3-23 IGHV3-30 IGHV3-33 IGHV3-48 IGHV3-7 IGHV4-34 IGHV4-39

<98%

98-98.9%

99-99.9%

100%

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

24

0

2

4

6

8

10

12

14

16

18

4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

<98%

98-98.9 %

99-99,9%

100%

Figure 2. Distribution of HCDR3 lengths according to mutational status. The striking

peak at codon length 9 is predominantly comprised of IGHV3-21 subset #2 cases

which carry a distinctively short, stereotyped HCDR3.

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

25

Figure 3. R/S normalized mutation ratios in the HCDR2 of rearrangements utilizing

the IGHV4-34 gene. Statistically significant differences were observed between CLL

vs. normal (N) or autoreactive (AU) clones.

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

26

A.

B.

C.

D.

E.

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

27

Figure 4. Amino acid sequence alignments of 5 selected subsets defined by HCDR3

stereotypy. Sequence alignments for (A) Subsets #1 and #28; (B) subset #2; (C)

subset#4; (D) subset #14; (E) subset #16 are represented as sequence logos85-86 to

summarize a total of 106 sequences belonging to these selected subsets

(Supplemental Table 10). In each subset representation (i.e. sequence logo), the

colored letters above the line represent the amino acids used in that particular

subset, while the grey letters shown upside down below the line represent the

germline amino acid composition of the relevant IGHV gene. Each colored letter

indicates an amino acid position where a mutation occurred. When more than one

change was observed in a position, the letters representing each change are

displayed as a stack. Thus, the size of the AA symbol represents the relative

frequency of that AA at that position relative to all other mutations at that position in

that subset. The height of the inverted germline amino acid symbol is the sum of the

heights of the upright amino acids. Blank spaces represent amino acids which are

unchanged in the CLL IGHV sequence as compared to the germline sequence.

Amino acids are colored based on their similarity in terms of their physicochemical

properties: [GAPVLIM], blue; [FYW], purple; [STCNQ], green; [KRH], red; and, [DE],

orange. Sequence logos are vertically stretched so that the tallest upright stacks are

of the same size, irrespective of the number of sequences. For example, in subset

#4, 9 of 20 sequences carry E whereas 5 of 20 sequences carry D at position

IMGT/HCDR1-28 (see supplemental Table 10); therefore, E is taller than D at that

position in the sequence logo for subset #4 (C), while the height of the inverted

germline G is the sum of the heights of the upright D and E. Further information

about number of sequences with a certain amino acid change out of total number of

sequences in each subset can be found in Table 1 and Supplemental Table 10. For

clarity, only codons 27 to 104, corresponding to HCDR1-HFR3 of the V region, are

shown. In (B), the letter X denotes the Serine deletion at IMGT/HCDR2 codon 59.

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom

28

TABLES

Table 1. “Stereotyped” amino acid changes. The frequency of changes among

mutated sequences utilizing the same IGHV gene was recorded in CLL sequences

with stereotyped HCDR3s (sequences belonging to subsets), CLL sequences with

heterogeneous HCDR3s and sequences from normal or autoreactive clones. For this

comparison, non-CLL sequences were pooled regardless origin (i.e. whether they

derived from normal or autoreactive B cells). That notwithstanding, full details about

non-CLL sequences (including origin) are provided in Supplemental Tables 1 and 12.

IGHV1-2*02 sequences Change CLL-Subset#1 CLL-heterogeneous Non-CLL IMGT-HFR2, codon 55 W-to-R 6/15 0/30 17/119 IGHV1-2*02 sequences Change CLL-Subset#28 CLL-heterogeneous Non-CLL IMGT-HFR2, codon 55 W-to-R 4/4 0/30 17/119 IGHV3-21 sequences Change CLL-Subset#2 CLL-heterogeneous Non-CLL IMGT-HCDR1, codon 32 S-to-Ta 11/56 6/29 12/95 IMGT-HCDR1, codon 34 S-to-Na 9/56 2/29 7/95 IMGT-HCDR2, codon 61 S deletion 18/56 0/29 1/95 IMGT-HFR3, codon 66 Y-to-H 7/56 2/29 3/95 IGHV4-34 sequences Change CLL-Subset#4 CLL-heterogeneousb Non-CLL IMGT-HCDR1, codon 28 G-to-D 5/20 0/108 6/320 IMGT-HCDR1, codon 28 G-to-E 8/20 6/108 18/320 IMGT-HCDR1, codon 32 G-to-Da 7/20 20/108 49/320 IMGT-HFR2, codon 40 S-to-T 10/20 29/108 45/320 IMGT-HFR2, codon 45 P-to-S 10/20 17/108 33/320 IGHV4-34 sequences Change CLL-Subset#16 CLL-heterogeneousb Non-CLL IMGT-HCDR1, codon 28 G-to-E 6/7 6/108 18/320 IMGT-HFR2, codon 40 S-to-T 4/7 29/108 45/320 IMGT-HFR2, codon 45 P-to-S 3/7 17/108 33/320

IGHV4-4 sequences Change CLL-Subset#14 CLL-heterogeneous Non-CLL

IMGT-HCDR1, codon 33 S-to-N 3/4 0/17 8/90

IMGT-HFR2, codon 40 S-to-T 4/4 4/17 12/90

IMGT-HCDR2, codon 57 Y-to-H 4/4 3/17 7/90

IMGT-HCDR2, codon 58 H-to-P 3/4 0/17 0/90

IMGT-HFR3, codon 78 I-to-M 4/4 6/17 8/90

IMGT-HFR3, codon 92 S-to-N 3/4 4/17 10/90 aThese mutations, though very frequent among sequences of a subset, were also identified at a high frequency among either CLL sequences with heterogeneous HCDR3 or non-CLL sequences and, thus, were not considered as “subset-biased”. bAll IGHV4-34 cases except for those belonging to subsets#4 and #16.

For personal use only. by guest on May 30, 2013. bloodjournal.hematologylibrary.orgFrom