B cell-specific and stimulation-responsive enhancers derepress Aicda by overcoming the effects of silencers.

Abstract

Activation-induced cytidine deaminase (AID) is essential for the generation of antibody memory but also targets oncogenes, among other genes. We investigated the transcriptional regulation of Aicda (which encodes AID) in class switch-inducible CH12F3-2 cells and found that Aicda regulation involved derepression by several layers of positive regulatory elements in addition to the 5' promoter region. The 5' upstream region contained functional motifs for the response to signaling by cytokines, the ligand for the costimulatory molecule CD40 or stimuli that activated the transcription factor NF-kappaB. The first intron contained functional binding elements for the ubiquitous silencers c-Myb and E2f and for the B cell-specific activator Pax5 and E-box-binding proteins. Our results show that Aicda is regulated by the balance between B cell-specific and stimulation-responsive elements and ubiquitous silencers.

Full text

Page 1

Title
B cell‒ specific and stimulation-responsive enhancers derepress
Aicda by overcoming the effects of silencers
Author(s)
Honjo, Tasuku; Nagaoka, Hitoshi; Tran, Thinh Huy; Nakata,
Mikiyo; Suzuki, Keiichiro; Begum, Nasim A; Shinkura, Reiko;
Fagarasan, Sidonia
CitationNature Immunology (2009)
Issue Date2009-12-06
URLhttp://hdl.handle.net/2433/88755
Right
c 2009 Nature America, Inc. All rights reserved. 許諾条件に
より論文本文は2010-06-07に公開.
TypeJournal Article
Textversionauthor
KURENAI : Kyoto University Research Information Repository
Kyoto University

Page 2

1
B cell–specific and stimulation-responsive enhancers derepress the Aicda gene by
overcoming the effects of silencers

Thinh Huy Tran1,2, Mikiyo Nakata1, Keiichiro Suzuki2, Nasim A. Begum1, Reiko
Shinkura1, Sidonia Fagarasan 2, Tasuku Honjo1 & Hitoshi Nagaoka1

1Department of Immunology and Genomic Medicine, Graduate School of Medicine,
Kyoto University, Yoshida Sakyo-ku, Kyoto 606-8501, Japan
2Laboratory for Mucosal Immunity, RIKEN, Yokohama 1-7-22, Tsurumi, Yokohama,
230-0045, Japan.

Correspondence should be addressed to T.H., (honjo@mfour.med.kyoto-u.ac.jp).
Department of Immunology and Genomic Medicine, Graduate School of Medicine, Kyoto
University, Yoshida Sakyo-ku, Kyoto 606-8501, Japan.
Phone: 81-75-753-4371; Fax: 81-75-753-4388

Abbreviations: Aicda, activation-induced cytidine deaminase gene; CSR, class switch
recombination; SHM, somatic hypermutation; TSS, transcription start site.

Page 3

2
Abstract
Activation-induced cytidine deaminase (AID) is essential for the generation of antibody
memory but also targets oncogenes among others. We investigated the transcriptional
regulation of the Aicda gene, which encodes AID, in the class switch–inducible CH12F3-2
cells, and found that the Aicda regulation involves derepression by several layers of
positive regulatory elements in addition to the 5′ promoter region. The 5′ upstream region
contains functional motifs for the response to signaling by cytokines, CD40-ligand or
stimuli that activate NF-κB. The first intron contains functional binding elements for the
ubiquitous silencers c-Myb and E2f and for B cell–specific activator Pax5 and E-box-
binding proteins. The results revealed that the Aicda is regulated by the balance between
B-cell-specific and stimulation-responsive elements and ubiquitous silencers.

Page 4

3
Introduction
Activation-induced cytidine deaminase (AID) is essential for the physiological alterations
in immunoglobulin genes (Igs) that generate antibody memory, namely, class switch
recombination (CSR) and somatic hypermutation (SHM)1,2. SHM introduces non-
templated point mutations at a high frequency in the variable (V) regions of antibody
genes, which, in conjunction with cellular selection mechanisms, give rise to high-affinity
antibodies. CSR is a region-specific DNA recombination that occurs between two switch
(S) regions located 5′ to each heavy-chain constant region (CH) gene. This recombination
juxtaposes a downstream CH gene with the V region gene by excising the intervening CH
genes. CSR results in immunoglobulin class switching without changing antigen
specificity3,4.
AID is responsible for DNA cleavage in both the V and S regions5–8. Since DNA
cleavage by AID can be deleterious to the genome, its is tightly regulated; virtually no
expression is observed in non-B cells or even in B cells, unless they have been activated
by appropriate stimuli, such as lipopolysaccharide (LPS), CD40-ligand (CD40L;
http://www.signaling-gateway.org/molecule/query?afcsid=A000536), or certain cytokines,
including interleukin 4 (IL-4; http://www.signaling-
gateway.org/molecule/query?afcsid=A001262), transforming growth factor-β (TGF-β;
http://www.signaling-gateway.org/molecule/query?afcsid=A002271) and interferon-γ
(IFN-γ)9,10. Consequently, the transcriptional regulatory system of Aicda should include
both lineage-specific and stimulus-specific response elements.
Complicating this picture, AID appears to target not only Ig genes but also the genes
for several oncogenes that are frequently mutated or translocated to Ig loci in B cell
malignancies (reviewed in 11). Studies on AID-deficient mice suggest that AID may be
involved in the pathogenesis of B cell malignancy12–16. Aberrant AID expression in human

Page 5

4
B cell lymphomas was also speculated to correlate with degree of malignancy11,17.
Interestingly, infection by certain viruses or bacteria that are potentially tumorigenic can
induce AID expression in B and non-B cells18–20. Such ectopic expression of AID may
contribute to tumorigenesis when AID expression persists during chronic infection20–22. In
support of this idea, systemic overexpression of AID in transgenic animals leads to tumor
development in various organs22–24.
Several studies have addressed the mechanism for the transcriptional regulation of
Aicda. The putative promoter region has been identified immediately upstream of the
transcription start site (TSS), but this promoter is not lymphocyte-specific25. Two tandem
E-boxes, recognized by E proteins such as E2A, located in the first intron have been
proposed to be important for the induction of AID in B cells26. The transcription factor
Pax5 [http://www.signaling-gateway.org/molecule/query?afcsid=A000403] may play a
role, in cooperation with E proteins, in the B lineage-specific control of AID expression27.
However, the location of the functional Pax5 motif is controversial25,27. In addition, the
functional contributions of proposed putative binding sites for the signal transducer and
transactivator 6 (STAT6; http://www.signaling-
gateway.org/molecule/query?afcsid=A002236) and nuclear factor kappa B (NF-κB;
http://www.signaling-gateway.org/molecule/query?afcsid=A001645), located upstream of
the Aicda promoter, are not yet clear28. Finally, most previous studies were done either in
cell lines whose AID expression is constitutive, or in primary B cells that had been fully
stimulated by mitogens. Additional information requires detailed and thorough studies on
the regulatory elements of Aicda in stable cell lines in which AID expression is inducible.
In this study, we performed an extensive analysis using luciferase reporter assays to
determine the functional regulatory elements of the Aicda locus in CH12F3-2 cells. These
cells carry out CSR with an efficiency as high as 70% upon stimulation with CD40L, IL-4,

Page 6

5
and/or TGF-β (CIT)29. Because each of these CIT factors is also important for CSR
stimulation in primary B cells, the control of the induction of AID expression in the
CH12F3-2 cells is likely to reflect the physiological regulation of Aicda. We identified
and characterized two clusters of regulatory elements, one about 8 kb upstream of the TSS
and the other within the first intron. The former contains the major enhancers that respond
to CIT stimulation, and the latter confers B cell specificity through a combination of B
cell–specific positive elements and ubiquitous negative regulators. Our results revealed a
sophisticated regulatory network in which AID expression is controlled by the balance
among more than a dozen enhancers and silencers.

Results
Conserved regions in and around the Aicda gene
To identify candidates for cis-regulatory elements of the AID gene, we compared the
mouse and human genomic DNA sequences within 50 kb upstream and downstream of the
Aicda TSS by using PipMaker and mVISTA pairwise alignments. Consistent with similar
analyses by others25, we found four relatively well conserved non-coding regions (Fig. 1a
and Supplementary Fig. 1). Region-1 (–1500 to +101), located immediately upstream of
the TSS, contains a ubiquitously active promoter25 and the following putative transcription
factor binding motifs: one for NF-κB, one for STAT6, two for Sp that can bind both Sp1
and Sp3, and one for HoxC4-Oct (Supplementary Fig. 1b).
Region-2 (+121 to +2221), located in the first intron, part of which shows more than
70% local nucleotide sequence identity between mouse and human. This region contains
the following binding motifs: two for NF-κB, two for Mzf1, two for Cp-2, three for c-Myb,
one for Nkx2.5, one for Pax5, two E-boxes, and one E2f-binding site (Supplementary Fig.
1b). Only the two high-affinity E-box have been well characterized26,27.

Page 7

6
Region-3 (+16278 to +18378) is located downstream of Aicda at approximately 6
kb and 25 kb from exon 5 in mouse and human, respectively. This region was reported to
be required for normal AID expression by BAC (bacterial artificial chromosome)
transgenic mouse system30. Region-4 (–9224 to –7424), located approximately 8 kb
upstream of the TSS, has not been studied before, although it contains candidate binding
motifs for positive regulatory transcription factors, including two for NF-κB, two for
STAT6, three for enhancer binding protein (C/EBP), and one for Smad3/4
(Supplementary Fig. 1b).

The Aicda promoter is not CIT-responsive
To study the cis regulatory elements of Aicda by the luciferase reporter assay, we used
CH12F3-2 cells, in which AID expression is barely detectable and strongly up-regulated
by stimulation with CIT, resulting in efficient CSR to IgA9. First, we generated plasmids
containing serial 5′ deletions of region-1, to examine whether region-1 contributes to AID
induction by CIT. This region contains several elements that were proposed or
demonstrated to be involved in Aicda regulation, including an Sp-binding motif25 that was
originally reported as a putative Pax5 site27, STAT6 and NF-kB sites28, and HoxC4-Oct
motifs31. We found all the tested constructs showed an approximately 4- to 8-fold increase
in the luciferase activity, compared with pGL3, the promoter-less luciferase vector. We
also found weak positive and negative effects in Sp motif and GA-rich sequence of which
binding factor is unkown, respectively (Fig. 1b). Unexpectedly, however, none of these
constructs showed any response to the CIT stimulation. The HoxC4-Oct element was
previously demonstrated to be involved in AID expression using Hoxc4–/– B cells31, but
this fragment alone did not significantly respond to CIT. The HoxC4-Oct element may be
involved in the basal expression of AID.

Page 8

7
Therefore, we extended our analysis to the other conserved regions in order to identify
CIT-responsive enhancer elements. We also sought stronger B cell–specific enhancer
elements, because Sp1 and Sp3 are general transcription factors and the activity of the
HoxC4-Oct motif was minimal, although it is lymphoid specific31,32. Since the 100-bp
fragment immediately upstream of TSS, containing the Sp and HoxC4-Oct motifs, still
activated transcription 5-fold, to a level almost equivalent to that induced by the entire
region-1, we used this fragment as the minimal promoter.

Regulation of the Aicda promoter by regions-2 and -4
To delineate the role of the conserved regions in the regulation of AID expression, we
constructed reporter vectors harboring region-1 in combination with regions -2, -3, or -4.
The addition of region-2 to region-1 decreased the luciferase activity by 50%, and this
construct did not respond to CIT stimulation (Fig. 1c). On the other hand, a construct
containing both region-1 and region-4 induced enhanced luciferase activity in response to
CIT stimulation, and its basal activity was also slightly higher than that of the region-1-
only construct, suggesting that region-4 contains DNA elements responsive to IL-4, TGF-
β and/or CD40L.
By contrast, the addition of region-3 did not result in any significant change in either
the basal or CIT-induced AID promoter activity, in contrast to a previous report30 (Fig. 1c).
Serially deleted region-3 fragments did not show any significant effects on the luciferase
activity driven by region-1, either (data not shown). Taken together, our data indicate that
the basal expression of AID in CH12F3-2 is regulated negatively by region-2 and
positively by region-4 while no enhancer activity was detectable in region-3 by the
luciferase assay. In addition, region-4 was primarily responsible for the up-regulation of
the Aicda gene in response to CIT stimulation.

Page 9

8

Region-2 confers B cell specificity
To dissect the regulatory elements in region-2, serially deleted region-2 fragments were
ligated downstream of the luciferase reporter construct driven by the minimal Aicda
promoter (–101 to +1) (Fig. 2a,b). CIT stimulation did not cause a significant increase in
the luciferase activity of CH12F3-2 cells expressing any of these constructs, supporting
the idea that the CIT responsive element is located outside region-2.
Further study of region-2, to look for B cell–specific regulators, showed that a deletion
of the 3′ sequence (+1709 to +1783) increased the luciferase activity approximately six
times over that induced by full-length region-2, suggesting negative regulatory elements in
the deleted sequence (Fig. 2a, line 4). The further serial deletion of the two E-boxes
reduced the activity to almost equal that of the minimal promoter (Fig. 2a, lines 5, 6), in
agreement with a previous report on the enhancer function of these E-boxes26. When the
Pax5 element was also deleted, the luciferase activity was reduced further (Fig. 2a, line 7).
To confirm the positive roles of the Pax5 motif and E-boxes, we made constructs with
only these motifs because the suppressing effect from the adjacent sequences masked their
activity. The fragment containing only the Pax5 site and the two E-boxes enhanced the
luciferase activity of the minimal promoter about 3-fold (Fig. 2c, line 4). Mutating any
one of the three motifs obviously impaired the enhancer activity, indicating that all three
contributed to the positive regulation of Aicda (Fig. 2c, lines 5-7). Since Pax5 and E
proteins are well-established B cell–specific co-activators, which actually bind to this
region in activated B cells26, 27, they could be responsible for the activation of Aicda in B
cells. In contrast, the NF-κB-3 site (see Fig. 2a for the positions of the numbered motifs)
did not seem to play a relevant role in either basal or CIT induced- AID expression (Fig.
2a, line 10 and Fig. 2b, line 4).

Page 10

9
The negative regulatory region (+1709 to +1789) contains the binding motifs for NF-
κB-4, E2f, Cp2-2, c-Myb-3, and Mzf1-2, but only the E2f-binding site appeared to be
responsible for the negative regulation by mutagenesis experiments (Fig. 2c, lines 4, 8 and
9, and Supplementary Fig. 2, lines 8, 9). Deletion of the upstream sequence (+1203 to
+1530) that contains the Mzf1-1, Cp2-1, c-Myb-1, c-Myb-2, and Nkx2.5 binding motifs
enhanced the luciferase activity, suggesting the presence of other negative regulatory
motifs within this region (Fig. 2a, line 9; Fig. 2b, line 7). We excluded the contribution
from the Mzf1-1 and Cp2-1 motifs by mutagenesis experiments as no significant
differences were observed relative the intact region (Supplementary Fig. 2, compare lines
6, 7 with line 5). Although the suppressor activity of a 350-bp CT-rich sequence from
+874 to +1221 was obvious only in the 5′ deletion construct (Fig. 2a,b), a similar
sequence has a reported suppressor function in another gene, Cyp1a1, in epidermal
keratinocytes33. Notably, the activity of the B cell-specific enhancers (the Pax5 site and E-
boxes) was not sufficient to counter the effect of the suppressive elements, including c-
Myb-1, c-Myb-2, Nkx2.5, E2f, and the CT-rich sequence of region-2 (Fig. 1c, Fig. 2a,b).

Suppressor elements in region-2 counteract enhancers
The interaction between regions-2 and -4 was examined. Region-2 suppressed the
enhancer activity of region-4 regardless of CIT stimulation (Fig. 2d, line 4). Since region-
4 amplified the luciferase activity strongly, we could confirm suppressor activities
observed in the deletion experiments. Deletion of the CT-rich sequence together with the
inactive NF-κB-3 motif led to a slightly higher luciferase activity (Fig. 2d, line 5).
Additional point mutations in one of the three motifs in this region, namely c-Myb-1, E2f,
and c-Myb-2 (overlapping the Nkx2.5 site), reduced the suppression activity (Fig. 2d,
lines 6–8). The combined disruption of all these motifs along with the CT-rich region

Page 11

10
almost completely abolished the suppressor activity of region-2 (Fig. 2d, line 9). Although
the involvement of Nkx2.5 cannot be formally excluded, the knockdown experiment with
siRNA oligonucleotides for c-Myb but not for Nkx2.5 increased the luciferase activities
(Supplementary Figs. 3,4). Taken together, these results indicate that the binding motifs
in region-2 for c-Myb and E2f, and the CT-rich sequence, function independently as
suppressor elements to counteract the positive elements in region-2 and region-4.

Elements responsible for CIT stimulation in region-4
IL-4, TGF-β and CD40L induce AID expression independently and additively in
CH12F3-2 cells9. To confirm that region-4 contains independent response elements for
IL4, TGF-β or CD40L, each stimulation was separately examined. IL-4, TGF-β, and
CD40L individually up-regulated the luciferase activity of the construct carrying region-4
(Fig. 3a). Combining the three stimuli resulted in the highest activity, and the increment
appeared to be additive.
To identify the functional elements in region-4, a series of deletion constructs was
generated (Fig. 3b,c). The inductive response to CIT stimulation greatly decreased by the
deletions of the DNA region containing STAT6-1 (see Fig. 3b line 5), STAT6-2 (Fig. 3c,
line 4), NF-kB-2 (Fig. 3b, line 6), Smad3/4 (Fig. 3c, line 6), C/EBP-1 (Fig. 3b, line 7),
and C/EBP-2 (Fig. 3c, line 7). Among these elements, NF-kB-2 and two tandem C/EBP
motifs seem to have strong contribution to the basal activity as well. In contrast, the
deletions of the DNA region containing NF-kB-1 (Fig. 3b, line 4) and C/EBP-3 (Fig. 3c,
line 5) elements had no effect on the luciferase activity.
The mutagenesis experiments confirmed the involvement of the region-4 elements.
Mutations of either or both of the STAT6-binding sequences strongly reduced the
luciferase activity induced by IL-4 alone as well as by CIT, indicating that both sites are

Page 12

11
response elements for the IL-4 signal (Fig. 4a, lines 4–7). STAT6 knockdown almost
completely blocked the luciferase induction by IL-4 stimulation (Fig. 4b and
Supplementary Fig. 5). We conclude that the IL-4 signal activates the Aicda promoter
through the STAT6-1 and –2 binding sites on region-4.
Destruction of the NF-κB-2 motif severely impaired the response to both CD40L alone
and to CIT (Fig. 4c, line 4), and the NF-κB-specific inhibitor, sulfasalazine, almost
completely suppressed the luciferase induction by CD40L through region-4 (Fig. 4d and
Supplementary Fig. 6). The inactivation of NF-κB also reduced the basal activity.
When the Smad3/4 site was mutated, the TGF-β response and the basal activity were
clearly suppressed (Fig. 4e, line 4). Furthermore, in agreement with the deletion
experiment results, the two C/EBP binding sites were required for the response to both
TGF-β alone and to CIT (Fig. 4f, lines 4, 5). Thus, the pair of C/EBP sites was required
for the region-4 response to all three CIT components: IL-4, CD40L and TGF-β. However,
the C/EBP sites alone did not respond to CIT stimulation (Fig. 4g), indicating that the pair
of C/EBP sites is required but not sufficient.
Finally, we examined whether the CIT response was dependent on the B cell–specific
enhancers (Pax5-binding site and E-boxes) of region-2 when the repressor elements co-
exist. Deletion of Pax5-binding site and E-boxes from the reporter construct containing
regions-2 and -4 still showed clear induction by CIT although the luciferase activity was
reduced to about 35% (Fig. 5a, lines 4, 5). These results indicate that the B cell–specific
enhancers are not absolutely essential for the CIT response through region-4, at least in
our luciferase system.
The same constructs used above were tested in two non-B cells, 2B4.11 T cells and
NIH3T3 fibroblastic cells. The basal luciferase activity in CH12F3-2 cells by the region-4
reporter alone was approximately twice that observed in the non-B cells, indicating that

Page 13

12
CH12F3-2 cells express more intrinsic factors for AID expression than non-B cells (Fig.
5b, line 3). Addition of region-2 repressed the basal activity not only in the B cells but also
in the other lineage of cells, indicating that the region-2 silencers are active in non-B as
well as B cells (Fig. 5b, line 4). Importantly, the deletion of the Pax5-binding site and E-
boxes reduced the luciferase activity only in B cells (Fig. 5b, lines 5).

Involvement of candidate transcription factors
To examine in vivo binding of the transcription factors to the candidate regulatory
elements, we performed chromatin immunoprecipitation (ChIP) assay of region-2 or -4
(Fig. 6a). Twelve-hour cultivation with CIT significantly induced the specific binding of
transcription factors, NF-κB, C/EBPβ, STAT6 and Smad4 to region-4 in CH12F3-2 cells
(Fig. 6b). We tried c-Myb and some of E2f proteins (E2f-1, -4 and -5) without success
partly because of the absence of high-quality antibodies (data not shown). C/EBPβ and
STAT6 also bound to region-4 in spleen B cells activated by LPS and IL-4 (Fig. 6c).
Binding of Pax5 and E-proteins to region-2 was already demonstrated26,27.
To confirm involvement of these transcription factors in Acida regulation, we examined
their expression profile in stimulated spleen B cells (Fig. 7). mRNAs of STAT6, NF-kB,
Smads, C/EBPs, Pax5, E2a, cMyb, E2f-2 and -4 were rapidly induced by stimuli of LPS,
IL-4 and TGF-β, declined by 12 h, and maintained thereafter. Expression of E2f-7 and -8,
which can repress by binding to the E2f motif34, increased 20 h after stimulation in parallel
with AID expression. We also compared the expression of these factors in naïve and
germinal center (GC) B cells. Most of the activators were already well-expressed in naïve
cells, suggesting the activation of their function by the protein modification after
stimulation (Supplementary Fig. 7). Expression of repressive E2fs, namely E2f-5, -7 and
-8, and cMyb were increased in GC B cells. Since the suppressive elements were

Page 14

13
functional in CIT stimulated CH12F3-2 cells, the increased expression of candidate
repressor proteins (E2f and c-Myb) in activated B cells may be a feedback response to
avoid the over-expression of AID (Supplementary Fig. 8).

Discussion
In the present study, we identified three functionally important regions in the 10-kb region
upstream of the Aicda gene and within it. Region-1 serves as the basic promoter, and
contains positive elements, Sp-binding sites, and a HoxC4-Oct motif, and a weak negative
element, the GA-rich sequences. Region-2, located in the first intron of Aicda, also
contains both positive (Pax5-binding site and E-boxes) and negative (c-Myb- and E2f-
binding sites) elements. On balance, region-2 functions as a negative regulator when
combined with region-1. Here we showed that the Pax5-binding site and E-boxes in
region-2, contribute to the restriction of AID expression to the B cell lineage. Region-4
contains the functional binding sites for NF-κB, STAT6, and Smad3/4, which are,
respectively, response elements for CD40L, IL-4 and TGF-β. A tandem pair of C/EBP
binding sites is required for each of these elements to respond to CIT stimulation. A
similar requirement for a tandem pair of C/EBP elements has been reported for the gene
encoding the common β-chain of the IL-3, IL-5, and GM-CSF receptors35.
Because AID promotes genome-wide damage and leads tumorigenesis, the promoter
activity should be finely regulated. In fact, AID expression virtually restricted to activated
B cells. How could such a restricted expression be achieved by the elements we found in
this study? Perhaps, the key is the negative elements in region-2. These counteract against
positive signals mediated from cytokine- or Toll-like-receptors (TLR) which are widely
expressed. They also work against B cell-specific enhancers, of which binding proteins are
functional even in non-activated B cells, thereby would confer the active transcription

Page 15

14
only in activated B cells.
C-Myb often functions as a transcription activator, but it also represses many
genes, including Cd4, Nras, c-erbB-2 (refs. 36-38). Interestingly the repressive c-Myb-
binding site of the Cd4, whose expression is tightly regulated during T cell development,
is also located in the first intron38. Some E2f proteins such as E2f-4, E2f-5, E2f-6 can
reversibly repress a promoter in conjunction with other co-repressors39. The E2f protein
can also induce stable repression by recruiting the polycomb complex40. Such a
mechanism might be involved in the complete silencing of Aicda in non-lymphoid cells.
Interestingly, we found increase of E2f-7 and -8 in in vitro activated and GC B-cell. E2f-7
and -8 counteract activating E2fs 34. Given activating E2fs were also up-regulated in GC
cells, they might bind to region-2 in certain condition, and E2f-7 and -8 would counteract
against them. We could not detect any enhancer activity to the E2f site in region-2, but it
could be due to the expression pattern of E2fs in CH12F3-2 which is repressor type
dominant (not shown). It still remains unsolved which E2f proteins are actually
responsible to the repressive function in region-2.
We found discrepancies between our results and previous studies on AID regulation. An
active Pax5 binding site in region-1 was proposed, but later another group concluded the
site is in fact an Sp-binding site25,27. Our mutagenesis experiments strongly indicated the
Pax5 involvement, but the responsive site is the one in region-2. Another discrepancy was
our inability to detect enhancer activity in region-3, which was reported to be required for
AID expression in the BAC transgenic mouse30. Region-3 might contain a positive
regulator which works only in the chromatin context.
Our study suggests that the region-4 is essential for the AID response in B cells to the
environmental stimulation delivered by T cells, dendritic cells, and other cytokine-
producing cells. Immature B cells were reported to express low amounts of AID by