http://www.diva-portal.org
This is the published version of a paper published in Molecular Immunology.
Citation for the original published paper (version of record):
Hauser, J., Grundström, C., Kumar, R., Grundström, T. (2016) Regulated localization of an AID complex with E2A, PAX5 and IRF4 at the Igh locus. Molecular Immunology, 80: 78-90 https://doi.org/10.1016/j.molimm.2016.10.014
Access to the published version may require subscription.
N.B. When citing this work, cite the original published paper.
Permanent link to this version: http://urn.kb.se/resolve?urn=urn:nbn:se:umu:diva-130546
Molecular Immunology 80 (2016) 78–90
Contents lists available at ScienceDirect
Molecular Immunology
j ournal homepage: www.elsevier.com/locate/molimm
Regulated localization of an AID complex with E2A, PAX5 and IRF4 at the Igh locus
∗
Jannek Hauser, Christine Grundström, Ramesh Kumar, Thomas Grundström
Department of Molecular Biology, Umeå University, SE-901 87 Umeå, Sweden
a r t i c l e i n f o a b s t r a c t
Article history: Activation-induced cytidine deaminase (AID) is the key mutagenic enzyme that initiates somatic hyper-
Received 2 May 2016
mutation (SH) and class switch recombination (CSR) by deaminating cytosine to uracil. The targeting of
Received in revised form 25 October 2016
AID and therefore SH and CSR to Ig genes is a central process of the immune system, but the trans-acting
Accepted 27 October 2016
factors mediating the specific targeting have remained elusive. Here we show that defective calmodulin
Available online 12 November 2016
inhibition of the transcription factor E2A after activation of the B cell receptor (BCR) leads to reduced
BCR, IL4 plus CD40 ligand stimulated CSR to IgE and instead CSR to other Ig classes. AID that initiates CSR
Keywords:
is shown to be in a complex with the transcription factors E2A, PAX5 and IRF4 on key sequences of the
Class switch recombination
Antibodies Igh locus. Calmodulin shows proximity with each of them after BCR stimulation. BCR signaling reduces
binding of the proteins to some of the target sites on the Igh locus, and calmodulin resistance of E2A
Activation-induced cytidine deaminase
Transcription factors blocks these reductions. AID binds directly to the bHLH domain of E2A and to the PD domain of PAX5.
E2A E2A, AID, PAX5 and IRF4 are components of a CSR complex that is redistributed on the Igh locus by BCR
Calmodulin signaling through calmodulin binding.
© 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
1. Introduction expression of AID leads to increased mutagenesis at non-Ig sites in
the genome and development of a wide range of tumours (Casellas
Upon infection, B lymphocytes undergo affinity maturation to et al., 2016; Morisawa et al., 2008; Okazaki et al., 2003). Mech-
optimize the antibody response to the specific pathogen. Point anistically, deamination by AID requires active transcription, and
mutations are introduced in the variable regions of the antibody proteins associated with the transcription process or to splicing
genes by somatic hypermutation (SH) in the germinal centers. have been shown to interact with AID (Kothapalli and Fugmann,
In addition, antibody effector functions can also be changed by 2011; Stavnezer, 2011).
class switch recombination (CSR), which changes the expressed The ability to switch to specific isotypes by CSR depends on a
constant region exons of the immunoglobulin heavy chain (IgH). complex, wide-ranging integration of signals resulting from acti-
These antibody diversification processes are initiated by targeted vation of the B-cell receptor (BCR), cytokine receptors, Toll-like
DNA deaminations made by activation-induced cytidine deaminase receptors, and tumor necrosis factor family members. Ultimately
(AID) (Chen and Wang, 2014; Fear, 2013; Stavnezer, 2011; Wang, this leads to transcription of the switch region of the appropri-
2013). The activity of various DNA repair enzymes on the deamina- ate antibody class (Pone et al., 2010). While this transcription is of
tions made by AID subsequently leads to mutations and deletions in critical importance for CSR to a particular isotype, the trans-acting
the DNA (Chen and Wang, 2014; Fear, 2013; Stavnezer, 2011; Wang, factors mediating selective targeting of AID to antibody genes and
2013). AID is highly regulated to permit SH and CSR of Ig genes therefore SH and CSR remain elusive (Chandra et al., 2015; Chen and
while retaining genome integrity more widely (Chen and Wang, Wang, 2014; Fear, 2013; Kothapalli and Fugmann, 2011; Stavnezer,
2014; Fear, 2013; Kothapalli and Fugmann, 2011; Stavnezer, 2011; 2011; Wang, 2013). To date, the best candidate of AID targeting
Wang, 2013). This tight regulation of AID is critical since aberrant is the basic-helix-loop-helix (bHLH) class of transcription factors,
since E-boxes, which most bHLH’s bind, have been shown to be
important in the AID mediated diversification process (Chen and
Wang, 2014; Fear, 2013; Stavnezer, 2011; Tanaka et al., 2010;
Abbreviations: CSR, class switch recombination; SH, Somatic hypermutation;
Wang, 2013). However, while E-boxes have been implicated in tar-
AID, activation-induced cytidine deaminase; PLA, proximity ligation assay; BCR, B-
cell receptor; CaM, calmodulin. geting of AID, to date, no direct association between AID and any
∗
Corresponding author. protein known to bind to E-boxes has been documented. Thus, how
E-mail address: [email protected] (T. Grundström).
http://dx.doi.org/10.1016/j.molimm.2016.10.014
0161-5890/© 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
J. Hauser et al. / Molecular Immunology 80 (2016) 78–90 79
AID is recruited to its specific targets in antibody genes is still among vated by incubation with goat F(ab’)2 anti-mouse IgM (Southern
the biggest mysteries in the field (Chandra et al., 2015; Chen and Biotechnologies) at 2.5 g/ml for the indicated times. Retrovirus
Wang, 2014; Fear, 2013; Kothapalli and Fugmann, 2011; Stavnezer, concentrated by centrifugation was added with 5 g/ml polybrene
6
2011; Wang, 2013). to 0.5 × 10 purified B cells after activation with CD40L plus IL-4
2+
We have previously reported that Ca -loaded calmodulin for 24 h. After 12 h incubation, the infection was repeated for 12 h,
(CaM) can inhibit the E-box binding basic-helix-loop-helix tran- followed by incubation for a further 22 h post-infection in fresh
scription factor E2A and that this inhibition is essential for certain complete medium with the stimulants to allow for expression of
important responses to activation of the membrane bound anti- E12, AID, PAX5, IRF4 and/or GFP. The medium was supplemented
body of the BCR (Corneliussen et al., 1994; Hauser et al., 2008b, with a minimal 0.2 g/ml of LPS (Calbiochem), i.e. more than 100
2009; Saarikettu et al., 2004; Verma-Gaur et al., 2012). Mecha- times less LPS than usually used in LPS-driven CSR studies, during
2+
nistically, Ca -loaded CaM inhibits E2A by binding as a dimer to retroviral infection incubations to enable infection. Where indi-
the two DNA-binding basic sequences of the bHLH domain and cated, anti-mouse IgM was added for the indicated times.
thereby inhibits the binding of the transcription factor to E-box
DNA (Larsson et al., 2005; Onions et al., 2000; Saarikettu et al., 2.3. FACS analysis
2004). Here we use the sensitivity of E2A to BCR activation as a
tool to assess if E2A is important in CSR. We analyzed whether E2A Immunostaining of harvested cells to analyze class switch-
is important in targeting of CSR by determining if mutation of E2A ing was performed on cells fixed with 2% paraformaldehyde and
rendering it resistant to CaM changes BCR dependent CSR. The CaM- permeabilized with 0.5% saponin using the following antibodies:
resistant mutant (m8N47) carries substitution of three amino acids anti-IgM-PE-Cy7, anti-IgD-PE and anti-IgG1 from BD Pharmin-
in the DNA binding basic sequence of the bHLH domain. This muta- gen, anti-IgG3 and anti-IgG2b from BioLegend, anti-IgA-PE from
tion of E2A inhibits CaM binding without inhibiting binding to DNA eBioscience, and anti-IgG2c and anti-IgE from Southern Biotech.
or to Id proteins (Hauser et al., 2008a,b). We show that defective When using primary antibody biotinylated using an antibody
CaM inhibition of E2A leads to reduced BCR, IL4 plus CD40 ligand- biotinylation kit (Miltenyi Biotech), the secondary antibody was
stimulated CSR to IgE and instead directs CSR to other Ig classes. Streptavidin-APC or Streptavidin-PerCP from BioLegend. Anti-
We demonstrate that AID forms a complex together with the tran- rat-PE and anti-goat-APC were from Jackson ImmunoResearch.
scription factors E2A, PAX5 and IRF4 on key sequences of the Igh Stainings were for 30 min at RT in the dark. Flow Cytometry was
locus in activated mouse splenic B cells. Also CaM is in proximity with a FACSCalibur instrument and analysis was with CellQuest
with them after BCR stimulation. Upon BCR signaling, binding of software (Becton Dickinson Biosciences).
the AID, E2A, PAX5 and IRF4 proteins to some of the target sites is
reduced, and CaM resistance of E2A prevents these BCR stimulated 2.4. Proximity ligation assay
reductions and instead increases binding to other target sites. We
have previously reported a direct interaction between purified het- Harvested B-lymphocytes were fixed with 4% paraformalde-
erologously expressed E2A and PAX5 transcription factors in vitro hyde, permeabilized with 0.2% Triton X-100, cytospun, and blocked
(Hauser et al., 2014). Here we show direct interactions between with 5% FCS in PBS. Proximity ligation assay (PLA) (Jarvius et al.,
AID, the PD domain of PAX5 and the bHLH domain of E2A. Taken 2007; Söderberg et al., 2006, 2008, 2007) was performed with the
together, these data demonstrate that E2A, AID, PAX5 and IRF4 form Duolink system using in situ PLA kits from Olink Biosciences. In
a complex that is redistributed on the Igh locus upon BCR signal- brief, cells were labelled with antibodies raised against intracellu-
ing. Mechanistically, this is achieved through CaM binding to E2A, lar proteins (antibodies listed in Supplementary Table 1). Secondary
therefore directing the specificity of CSR. antibodies conjugated with oligonucleotides (PLA probes) were
subsequently used according to the manufacturer’s protocol to gen-
erate fluorescence signals only when the two PLA probes were
2. Materials and methods
in close proximity (≤40 nm). The fluorescence signal from each
detected pair of PLA probes was visualized as a distinct individual
2.1. Plasmids and viruses
red dot in a Nikon 90i epifluorescence microscope. The antibod-
ies were used at dilutions that yielded a low number of dots/cell
The retrovirus encoding E12 splice form of E2A and C-terminally
to reduce the risk of missing dots in the quantifications. Nuclei
Flagged PAX5 has been described previously (Hauser et al., 2014).
were counterstained with Hoechst 33342 dye. To enhance detec-
For the AID and IRF4 constructs, the GFP-encoding DNA of the
tion of infected cells, anti-GFP-FITC antibody (Novus biologicals)
MSCV-IRES-GFP plasmid was replaced by oligonucleotides com-
was added during the PLA signal-detection step. The AID and PAX5
prising a tandem repeat of the Flag epitope coding sequence
antibodies were used after biotinylation with an antibody biotiny-
(Saarikettu et al., 2004) and restriction enzyme sites for inser-
lation kit (Miltenyi Biotech) where necessary.
tion of full-length cDNAs of AID and IRF4 obtained from GenScript.
E12 wild-type and CaM-resistant mutant with or without the C-
2.5. Chromatin immunoprecipitation
terminal Flag epitope were cloned into the MSCV-IRES constructs
as described previously (Hauser et al., 2008b). Production of retro-
Purified and infected primary splenic B-cells from mice het-
viruses was in 293T packaging cells as previously described (Hauser
erozygous for knockout of the E2A gene (Zhuang et al., 1994) were
et al., 2008b).
cross-linked with 1% formaldehyde for 10 min at room tempera-
ture. Cross-linking was quenched by adding glycine to 120 mM, and
2.2. Activation and infection of B-lymphocytes from mouse cells were washed with Phosphate-buffered saline. Cells were lysed
spleens on ice with 0.25 ml of RIPA buffer (50 mM Tris-HCl, pH 8.0, 150 mM
NaCl, 1% Nonidet P-40, 0.1% SDS, 0.5% sodium deoxycholate, and
Primary B-lymphocytes were purified from mouse spleens and 5 mM EDTA) supplemented with protease inhibitor mixture (Roche
maintained as previously described (Hauser et al., 2008b) and Applied Science). The lysate was sonicated at 60% amplitude for
stimulated with 5 ng/ml IL-4 (Pepro-Tech) and 50 ng/ml CD40L 10 times 10 s on a VibraCell instrument (Sonics), giving an average
6
(R&D Systems). The BCR of stimulated cells at 10 cells/ml, or size of sheared chromatin fragments of 180 bp, and clarified by cen-
5
where indicated at 2 × 10 cells/ml, was (where indicated) acti- trifugation. 5 l was saved as input, and the lysate was incubated
80 J. Hauser et al. / Molecular Immunology 80 (2016) 78–90
◦
overnight at 4 C with anti-Flag-tag antibody (anti-DYKDDDDK- indicated lane of Fig. 7C. The AID-Sepharose, E12-Sepharose, PAX5-
tag) (GenScript). Immunocomplexes were bound for 1 h to 20 l Sepharose and control Sepharose were washed twice with binding
◦
of magnetic Dynabeads-Protein A (Invitrogen) at 4 C. Beads were buffer and bound proteins were eluted by boiling in Laemmli load-
washed three times with RIPA buffer and three times with Tris- ing buffer. Samples were separated by SDS-PAGE and analyzed by
EDTA, pH 8.0, on a magnet (Invitrogen). Cross-links were reversed western blot using Tetra-His antibody (Roche) for E. coli-produced
◦
with 1% SDS in Tris-EDTA, pH 8.0, at 65 C overnight, and samples E12, PAX5 and deletion derivatives of these and with CaM antibody
◦
were treated with Proteinase K (20 g/ml) for 1 h at 45 C. DNA (EP799Y; Novus Biologicals) for E. coli-produced CaM.
was phenol-chloroform extracted, ethanol-precipitated, and ana-
lyzed by real-time PCR. The Real-time PCR analysis was performed
2.7. Statistical analysis
as previously described using glyceraldehyde-3-phosphate dehy-
drogenase (GAPDH) as internal control (Hauser et al., 2008b) and
All experiments were performed at least three times. All data
each sample was quantified relative to its input chromatin DNA. The
are expressed as means ± SD. The results were analyzed using the
specificity of the Real-time PCR was analyzed by melt-curve anal-
Student’s t-test with two-tailed distribution. Significant P values
ysis as described by the manufacturer. The Real-time PCR primer
are shown at *P < 0.05, **P < 0.01, or ***P < 0.001.
pairs used are listed in Supplementary Table 1. The E and hs3b
primers were from (Greenbaum and Zhuang, 2002), the V1, V11,
V13, VH7183, VHJ558, C and C␦ primers were from (Zhang et al., 3. Results
2006). The primers for hs5 and hRE1 were designed based on pos-
itive E2A ChIP signals in mature B cells within these amplicons in 3.1. CaM resistance of E2A reduces BCR stimulated CSR to IgE and
the 3 regulatory region of Igh (Lin et al., 2010). The switch region instead directs CSR to other ig classes
primers were designed based on sequences found to be mutage-
nized in class switching of s (Rajagopal et al., 2009; Xue et al., To analyze whether E2A has a role in controlling CSR, we infected
2006), and S␥1, S␥2b, S␥2c, S␥3, S␣ and S (Rajagopal et al., 2009; splenic mouse B cells with retrovirus expressing green fluorescent
Xue et al., 2006). protein (GFP) plus either wild type or CaM-resistant E12 splice
form of E2A. We have systematically found in previous studies that
2.6. Protein expression and binding assay CaM-resistant E12 protein is dominant over all endogenous CaM-
sensitive E2A, i.e. both proteins containing the E12 splice form and
The expression of full-length murine PAX5 and amino acids the E47 splice form (Corneliussen et al., 1994; Hauser et al., 2014,
431–652 of murine E12 in E. coli BL21(DE3)pLysS and the purifica- 2008a,b, 2013, 2009, 2010; Saarikettu et al., 2004; Verma-Gaur
tion of the proteins have been described previously (Hauser et al., et al., 2012; Wallenius et al., 2014). This is as expected, since dimers
2014). Expression plasmids for the E12 derivatives with deletion of between a sensitive and a resistant E2A lack half of the calmodulin
the bHLH domain, the sequences N-terminal or C-terminal to the interaction sites and therefore will be essentially resistant. Mice
bHLH domain, or deletion of everything except the bHLH domain, homozygous for knock-out of E2A could not be used to assess the
and the PAX5 derivatives with deletion of the paired domain, the role of E2A in CSR since such mice have blocked B cell develop-
homeodomain or everything C-terminal to the homeodomain or ment and lack a spleen (Bain et al., 1994; Zhuang et al., 1994). To
C-terminal to amino acid 176 (Fig. 7F) were purchased from Gen- reduce background from endogenous CaM-sensitive E2A, B-cells
Script. The deletion derivatives were expressed and purified as from mice heterozygous for knock-out of the E2A gene (Zhuang
previously described for full-length PAX5 and the E12 (Hauser et al., 1994) were used instead. Intracellular FACS staining for E2A
et al., 2014). E. coli-produced CaM was purchased from Millipore. showed that the expression level of E2A was similar to endoge-
The cDNA for full-length murine AID was obtained from Genscript nous wild type E2A levels (less than two fold difference) in the vast
and subcloned into pET21b+ (2), a derivative of pET21b+ (Novagen) majority of infected splenic B-cells. This observation is consistent
where the N-terminal T7-tag is replaced with the pelB leader from with data from pre-B cells infected with these constructs reported
pET20b+ (Novagen). The His6-tagged protein was expressed in in a previous study (Hauser et al., 2014). CSRs occur during the G1
E. coli BL21(DE3)pLysS. AID expressing cells were centrifuged and phase of the cell cycle in cells that are rapidly dividing (Hodgkin
lysed by sonication in 50 mM Sodium phosphate, pH 8.0, 300 mM et al., 1996; Schrader et al., 2007; Sharbeen et al., 2012; Stavnezer
NaCl, 2 mM MgCl2, 0.5 mM DTT, 1 mM PMSF. Urea was added to et al., 2010). To ensure that the analysis included conditions that
8 M and AID was purified by Ni-NTA agarose chromatography (Qia- facilitated fast growth, we seeded the cells at two different densi-
6 5
gen) according to the manufacturer’s instructions. The AID was ties, 10 /ml and 2 × 10 /ml. The cells were activated with IL-4 and
eluted with 100 mM Sodium phosphate, 10 mM Tris, pH 4.5, 8 M CD40 ligand (CD40L), a key activating cytokine from T helper cells,
Urea and dialysed against 6 M Guanidine, 40 mM DTT. Purified that together induce CSR predominantly to IgG1 but also to IgE
AID was then renatured by addition of 20 vol of 1% myristyl- (Pone et al., 2010). The cells were given a time-limited treatment
sulfobetanaine (surfactant SB3-14; TaKaRa) and dialysis against with a low amount of LPS (0.2 g/ml) during the retroviral infec-
75 mM Sodium phosphate, pH 8.0, 68 mM NaCl, 0.7% Sarkozyl, 15% tion incubation step to enable infection, and permeabilized cells
Glycerol, 0.3 M Arginine. Purified AID and the E12 were dialysed were labelled to enable measurement of cytoplasmic antibodies
against 0.1 M Sodium phosphate, 0.5 M NaCl (pH 8.0) and coupled to and FACS analysis was performed. Under these conditions, many
CNBr-activated Sepharose 4 B (GE Healthcare) following the man- cells had class-switched from IgM and IgD to IgG1 or IgE after 4 days
ufacturer’s instructions and using 0.1 M Sodium phosphate, 0.5 M both for cells infected with retrovirus expressing wild type E12 and
NaCl, pH 8.0, as coupling buffer. Purified PAX5 were coupled to CaM resistant E12 as well as the vector control retrovirus (Fig. 1A-B
CNBr-activated Sepharose 4 B as previously described (Hauser et al., and Supplementary Fig. 1A). As previously reported (Hauser et al.,
2014). Purified E12, PAX and deletion derivatives of these (0.2 g) 2009), stimulation of the BCR after 2 days increased the number of
were incubated with 10 l AID-Sepharose, E12-Sepharose, PAX5- large cells at day 4 (Supplementary Fig. 1A). It also led to a strong
Sepharose or control Sepharose beads as indicated in a binding reduction in cells expressing IgM or IgD both among cells infected
buffer containing 50 mM Tris-HCl, pH 8, 100 mM NaCl, 1.5% Tri- to express wild type E12 and cells infected with vector control
ton X-100, 1 mM DTT and protease inhibitor cocktail tablet without retrovirus at both the higher and the lower cell density (Fig. 1A-B
EDTA (Roche) and 0.1 mM EGTA with rotation for 30 min at room and Supplementary Fig. 1A). As previously reported (Hauser et al.,
temperature. 0.1 mM CaCl2 was used instead of the chelator in the 2009), CaM-resistance of E12 reduced the shift away from expres-
J. Hauser et al. / Molecular Immunology 80 (2016) 78–90 81
Fig. 1. CaM resistant E2A reduces IL4, CD40L plus BCR stimulated CSR to IgE and increases CSR to several other Ig classes. B lymphocytes purified from spleen of mice
5 6
heterozygous for knock-out of the E2A gene were activated at low cell density (2 × 10 cells/ml) or high cell density (10 cells/ml) as indicated with CD40L plus IL-4. After
1 day, cells were infected with MSCV retroviruses for 1 day with addition of a low amount of LPS at this step, followed by culture for 2 days with or without stimulation of
the BCR with anti-IgM. Harvested cells were immunostained with antibodies against the indicated Ig class and analyzed by FACS. Large live cells expressing GFP from the
virus were selected and the number of cells expressing the indicated Ig class was determined. (A) Summary of FACS analyses as in Supplementary A after infections with
R
MSCV virus expressing wild-type E12 (WT) or CaM-resistant mutant (CaM ) E12 splice form of E2A and infections with the corresponding empty MSCV vector (Vec.). The
bars show mean percentages of gated cells ± s.d. for the different Ig classes based on experiments with three different mice in the upper panels and seven different mice in
R
the lower panels. The P values of the difference between wild-type E12 (WT) and CaM-resistant (CaM ) mutant E12 after stimulation of the BCR with anti-IgM (gray bars)
are indicated. (B) Effect of stimulation of the BCR with anti-IgM on expression in the experiments summarized in (A) and corresponding experiments for different other Ig
classes. The results are based on three experiments in the upper panels and on seven experiments for IgM, IgD, IgG1 and IgE and four experiments for IgG2b, IgG2c, IgG3 and
IgA in the lower panels. The bars show mean fold changes in percentage of gated cells ± s.d. after stimulation of the BCR compared to without the BCR stimulation. Significant
P values for differences between wild-type and CaM-resistant E12 are shown at *P < 0.05, **P < 0.01, or ***P < 0.001. The FACS results from a representative experiment with
the B lymphocytes of one mouse is shown in Supplementary A.
sion of IgM and IgD upon BCR stimulation at both cell densities BCR stimulation had at both cell densities no significant effect on
(Fig. 1A-B and Supplementary Fig. 1A). Importantly, the frequency the frequency of cells positive for IgG2c but strongly reduced the
of IgE positive B cells was much lower when CD40L and IL-4 stim- frequency of cells positive for IgG2b, IgG3 and IgA (Fig. 1B). Impor-
ulation was combined with BCR stimulation for the CaM-resistant tantly, and in strong contrast, BCR stimulation did not reduce the
mutant than for wild-type E2A at both cell densities (Fig. 1A-B and fraction of the cells with CaM-resistant E12 that were positive for
Supplementary Fig. 1A), whereas no significant difference between any of the IgG2b, IgG3 and IgA classes (Fig. 1B). Taken together,
wild-type and mutant was seen for IgG1. The differences between this supports the notion that CaM resistance of E2A that inhibited
wild-type and mutant in the percentage of gated cells after BCR appropriate CSR to IgE in CD40L, IL-4 plus BCR stimulated B cells
stimulation were statistically significant at both cell densities for at the same time blocked BCR-mediated reduction in CSR to IgG2b,
IgM, IgD and IgE (Fig. 1A-B and Supplementary Fig. 1A). Thus, CaM IgG3 and IgA.
resistance of E2A inhibited class-switching from IgM and IgD to Since the effects of CaM resistance of E2A on CSR implied a possi-
IgE in the BCR stimulated B cells. We next measured the effect of ble coupling between E2A and the CSR-inducing mutagenic enzyme
the CaM resistance on class switching to the other Ig classes. The AID, we next examined whether over-expression of AID, which
82 J. Hauser et al. / Molecular Immunology 80 (2016) 78–90
could lead to ectopic localization of AID in the locus, also could interactions by microscopy of cells. Mouse splenic B-cells grown
lead to changed CSRs. We infected B cells from spleens of mice het- with CD40L and IL-4 were labelled using antibodies against the
erozygous for knock-out of the E2A gene with retrovirus expressing two proteins and subsequently with PLA probes to generate flu-
both AID and wild type or CaM-resistant E12 from the same virus. orescence signals seen as distinct individual red dots when the two
Cells were immunostained with antibodies against both AID and proteins were in close proximity (≤ 40 nm) (Jarvius et al., 2007;
the indicated Ig class. Cells with and without clear over-expression Söderberg et al., 2006, 2008, 2007). We observed specific proxim-
of AID were gated separately. In cells with moderate expression of ity signals between AID and E2A, and the background signals were
AID and E2A from the virus, BCR stimulation reduced the frequency minimal when one or both of the primary antibodies were omit-
of cells expressing IgM, IgD, IgG3 and IgA, whereas the frequency ted (Fig. 2A). The proximity measurements between AID and E2A
of IgE expression increased (Supplementary Fig. 1B). Interestingly, were observed with two independent AID antibodies confirming
over-expression of AID resulted in a reduction in CSR to IgE and this finding. BCR stimulation transiently potentiated the proxim-
a corresponding increase in other CSRs similar to CaM-resistance ity signal between AID and E2A by 2-fold to 3-fold in 30 min, and
of E2A (cf. Fig. 1B and Supplementary Fig. 1B). Furthermore, the after 3 h the proximity returned to the level before BCR stimu-
effects of over-expression of AID and of CaM-resistance of E2A were lation (Fig. 2A and quantifications in Fig. 2B and Supplementary
synergistic, leading to an enhanced phenotype greater than the Fig. 2A). It has been previously shown that the transcription fac-
effects of only expressing CaM-resistant E2A or over-expressing tor IRF4 is critical for CSR (Lu, 2008) and the PAX5 transcription
AID together with wild type E2A (Fig. 1B and Supplementary Fig. factor is important at the Igh locus (Medvedovic et al., 2011). Both
1B). In conclusion, both over-expression of AID and CaM-resistance PAX5 and IRF4 showed proximity with AID significantly over back-
of E2A changes CSR when combining the activation of the B-cells ground (Fig. 2B and Supplementary Fig. 2). BCR stimulation further
with BCR-stimulation. increased the proximity between AID and PAX5 approximately 4-
fold in 30 min, and after 3 h most of this increase had disappeared. In
contrast, BCR stimulation reduced the proximity between AID and
3.2. AID is together with the transcription factors E2A, PAX5 and
IRF4 after 30 min (Fig. 2B and Supplementary Fig. 2). The specificity
IRF4 in activated B cells
of the PLA signals was verified by omitting one or both of the pri-
mary antibodies and in several cases by using different antibodies
The related effects of either mutating E2A or over-expressing
against the same protein (Fig. 2 and Supplementary Figs. 2 and 3),
AID on CSR indicated a possible connection between E2A and AID
and by comparing proximity signals in B cells used in this study with
in vivo. To analyze if E2A and AID were physically together, we per-
those in cells that do not express PAX5, IRF4 or AID. The background
formed proximity ligation assays (PLA) in situ using the Duolink
was minimal in several control PLAs of PAX5 and of IRF4 in cells
system, which enables visualization and quantification of protein
Fig. 2. In vivo proximities between AID, the transcription factors E2A, PAX5 and IRF4, CaM and chromatin before and after BCR stimulation. (A) Purified primary mouse
splenic B-cells grown with CD40L and IL-4 were stimulated with anti-IgM for the indicated times. The detection of proximity between the proteins (shown as red dots)
was by in situ proximity ligation assay (PLA). Cells used in controls with only one of the antibodies, or with no primary antibody (-Ab) were stimulated with anti-IgM for
0.5 h. Samples were counterstained with Hoechst 33342 dye (Blue) to visualize cell nuclei. (B-C) Summary of quantification of the numbers of AID-E2A proximities/cell in
experiments such as those shown in (A) with three different mice and corresponding analyses of the other protein pairs indicated. The numbers of dots/cell after 0.5 h of
BCR stimulation were set as 100%. A representative experiment of different protein pairs and bar diagrams with the mean ± s.d. from three different mice are shown in
Supplementary Figs. 2, 3. Different antibodies against the same protein are distinguished by superscript numbers in (A) as specified in Supplementary Table 1. Results from
PLA with different antibodies against the same protein are shown as averages in (B-C), but shown separately in the bar diagrams of Supplementary.
J. Hauser et al. / Molecular Immunology 80 (2016) 78–90 83
lacking the probe-marked protein and also in AID PLAs in resting tion by BCR stimulation of proximity of E2A with CaM was almost
B-cells not stimulated to induce expression of AID (Supplementary completely blocked in the CaM-resistant mutant (Fig. 3B and Sup-
Fig. 3C). For the PAX5 and E2A antibodies used, we have previously plementary Fig. 4B). The remaining small increases could be due to
shown that the PLA background is minimal in control cells lacking not complete CaM resistance of the mutant (Hauser et al., 2008a,
the protein (Hauser et al., 2014). The proximity identified between 2008b) and to intermediately sensitive dimers, and perhaps higher
AID and each of E2A, PAX5 and IRF4 prompted us to analyze prox- multimers, between the CaM-resistant E12 and endogenous CaM-
imity between these transcription factors. The PLA showed specific sensitive E2A. The induction by BCR stimulation of the proximities
E2A-PAX5, E2A-–IRF4 and PAX5–IRF4 proximity. BCR stimulation of AID and IRF4 with CaM were also inhibited when E12 was CaM-
increased the E2A–PAX5, E2A-–IRF4 and PAX5-–IRF4 proximities 2- resistant (Fig. 3B and Supplementary Fig. 4B), which shows that
fold or more in 30 min, and after 3 h the proximity returned to close these proximities are through the proximities of AID and IRF4 with
to the level before BCR stimulation (Fig. 2B and Supplementary Fig. E2A. However, the BCR stimulated increase in proximity between
2). In summary, the PLAs show pair-wise proximity between each PAX5 and CaM was not blocked by CaM-resistant E12 (Fig. 3B
of AID, E2A, PAX5 and IRF4 in activated B cells, and BCR stimulation and Supplementary Fig. 4B), indicating that these two proteins
temporarily increased the levels of the proximities. can come in proximity through an E12-independent mechanism.
The increases in proximities of AID, E2A, PAX5 and IRF4 with
3.3. In vivo regulation of an AID complex with E2A, PAX5 and H2B(Ser14P)-containing and H3K4me3-containing chromatin after
IRF4 through CaM inhibition of E2A 30 min of BCR stimulation were all unaffected by CaM-resistance of
E12, showing that they occurred through another mechanism. The
We also analyzed the proximity of AID, E2A, PAX5 and IRF4 with reductions of these proximities after 3 h were instead blocked by
CaM, since CaM resistance of E2A affected the CSR (Fig. 1 and Sup- CaM-resistant E12 (Fig. 3C and Supplementary Fig. 4C-D), show-
plementary Fig. 1). The proximities were at background before the ing that these reductions are through CaM inhibition of E2A. Taken
BCR stimulation with the only exception that the PAX5-CaM pair together, these data suggest the existence of a protein complex dur-
was moderately increased over background for one of the CaM anti- ing class-switching consisting of at least AID, E2A, PAX5 and IRF4
bodies tested (Supplementary Fig. 2C). The findings that PLA pairs at H2B(Ser14P)- and H3K4me3-containing chromatin that remains
with CaM in almost all cases did not give significant PLA signals when CaM binds to the complex after BCR stimulation.
before stimulation of the BCR provides additional negative con-
trols supporting the notion that the PLA did not give unspecific 3.4. Localization of a complex of AID, E2A, PAX5 and IRF4 at the
signals over background. BCR stimulation for 30 min increased the Igh locus is affected by CaM inhibition of E2A
proximity of all protein pairs with CaM. Proximities clearly over
background endured over extended time periods (3 h), albeit in To analyze the effects of CaM resistance of E12 on the localiza-
most cases at a reduced level. AID-dependent phosphorylation of tion of E12, AID, PAX5 and IRF4 in vivo, we performed chromatin
serine 14 of histone H2 B (H2B(Ser14P)) and trimethylation of lysine immunoprecipitation (ChIP) before and after BCR stimulation using
4 of histone 3 (H3K4me3) at the Igh locus are coupled to SHM and splenic B cells of mice heterozygous for knock-out of the E2A
CSR (Begum et al., 2012; Odegard et al., 2005; Stanlie et al., 2010). gene. The cells were infected with retrovirus expressing E12 with a
We therefore analyzed whether there was proximity between the C-terminal tandem Flag epitope, and the ChIP was against the Flag-
histone modifications H2B(Ser14P) and H3K4me3 and AID, E2A, tag. The recovered DNA was analyzed by quantitative real-time
PAX5 and IRF4. We observed proximity between both H2B(Ser14P) PCR using primer pairs amplifying regions of the Igh locus contain-
and H3K4me3 and each of AID, E2A, PAX5 and IRF4. Furthermore, ing target sites for E2A (E-box sequences), PAX5, and AID, and the
as found above for AID and transcription factor proximities, BCR transcribed C and C␦ parts of the locus. Some of the amplicons
stimulation for 30 min increased these proximities (Fig. 2C and with E2A, PAX5 or AID sites also have an IRF4 site. For the E-box-
Supplementary Fig. 2D-E). Notably, BCR stimulation did not inhibit containing Igh enhancers E and hs3b (Greenbaum and Zhuang,
the proximity of E2A, AID, PAX5 or IRF4 with H2B(Ser14P)- and 2002), we observed as expected Flag ChIP signals in cells express-
H3K4me3-containing chromatin (Fig. 2C and Supplementary Fig. ing E12 over the background of cells infected with empty vector
2+
2D-E) despite that the Ca signal after BCR stimulation makes (Fig. 4A). In cells expressing E12 we also observed Flag ChIP sig-
CaM inhibit the DNA binding of E2A (Corneliussen et al., 1994; nals significantly higher than the background of cells infected with
Hauser et al., 2008b, 2009; Larsson et al., 2005; Onions et al., 2000; empty vector for the hs5 and hRE1 amplicons in the 3 regulatory
Saarikettu et al., 2004; Verma-Gaur et al., 2012). This finding indi- region of Igh that also contains sites of positive E2A ChIP signals in
cates that when the stimulation induces CaM to bind to and inhibit mature B cells (Lin et al., 2010). The hs3a and hs4 amplicons did not
DNA binding of E2A, then E2A stays at the chromatin through a result in signals enough over vector background to give meaningful
direct or indirect interaction of AID, PAX5, IRF4 or another pro- data and were therefore not included. Several nucleotides in addi-
tein(s). tion to the canonical E-box sequence CANNTG are important for
Next we analyzed the dependence of the protein proximities on binding of E2A, and the genome has only a few in vivo E2A bind-
the CaM-sensitivity of E2A. B cells from spleens of mice heterozy- ing sites/million bp (Lin et al., 2010). Nevertheless, we observed
gous for knock-out of the E2A gene were infected with retrovirus anti-Flag signals for E12-Flag far over the vector background also
expressing wild-type or CaM-resistant E12 together with GFP to for almost all amplicons containing AID or PAX5 target sequences
identify the infected cells. The increases in E2A-PAX5 and E2A- but not for control sequences from the C and C␦ regions of the
IRF4 proximities after 30 min of BCR stimulation were all blocked gene, which do not contain a target sequence for E2A, AID, IRF4
by CaM-resistant E12, whereas the increases in PAX5-IRF4, AID- or PAX5. Thus, E2A binds not only to E2A sites but also to many
E2A and AID-PAX5 proximities remained with CaM-resistant E12 PAX5, IRF4 and AID sequences. This supports the existence of a
(Fig. 3A and Supplementary Fig. 4A). In contrast to the wild-type complex between E2A, AID, IRF4 and PAX5 on the Igh locus. To fur-
E12, when E12 was CaM-resistant the PAX5-IRF4, AID-E2A and ther examine potential interactions between E2A, AID, IRF4 and
AID-PAX5 proximities that were induced after 30 min of BCR stim- PAX5, we next infected with retrovirus expressing non-tagged E12
ulation were not reduced in the mutant after 3 h (Fig. 3A and together with either AID-Flag, PAX5-Flag or IRF4-Flag. As expected,
Supplementary Fig. 4A). This latter difference is probably through we observed anti-Flag signals over the background of empty virus
the previously reported (Hauser et al., 2008b, 2009) reductions in in the ChIP for target sequences of AID, PAX5 and IRF4, respec-
AID and PAX5 expression after CaM inhibition of E2A. The induc- tively (Fig. 4B–D). Importantly, we observed for all three Flagged
84 J. Hauser et al. / Molecular Immunology 80 (2016) 78–90
Fig. 3. CaM resistance of E2A affects in vivo proximities of AID, E2A, PAX5, IRF4 and CaM in chromatin after BCR stimulation. Purified splenic B-cells of mice heterozygous for
R
knock-out of the E2A gene were infected with retrovirus expressing wild-type (WT) or CaM-resistant (CaM ) E12 together with green fluorescent protein (GFP) as in Fig. 1A.
Infected cells were identified by GFP fluorescence in the PLA microscopy and used to quantify the numbers of proximities/cell before BCR stimulation and after stimulation
for the indicated times. The figure is a summary of the PLA results with three different mice for the protein pairs indicated. PLA results with different antibodies against the
same protein are shown as an average, whereas results with different antibodies against the same protein are shown separately in the bar diagrams of Supplementary Fig. 4.
proteins also signals far over the background for all amplicons that and 2 h of BCR stimulation (Fig. 5A). In contrast, at the sites that
had a target sequence for any one of the three other proteins of the contain an IRF4 target, 30 min or 2 h of BCR stimulation instead
complex that were without a Flag (Fig. 4B–D). The anti-Flag signals increased the ChIP signal 2-fold to 5-fold for the CaM-resistant E12
were much weaker for control sequences from the C and C␦ for mutant (Fig. 5A). These data suggest that the AID complex with E2A,
all four viruses (Fig. 4A–D). These data strongly support the notion PAX5 and IRF4 is partially changing location at the Igh locus upon
that E12, AID, PAX5 and IRF4 interact in vivo in a complex at the Igh BCR stimulation through CaM inhibition of E2A. Next we examined
locus through the DNA target sequences of these proteins. interactions using virus expressing wild-type or CaM-resistant E12
To analyze the role of CaM inhibition of E2A in the effects of BCR plus AID-Flag (Fig. 5B), PAX5-Flag (Fig. 5C) or IRF4-Flag (Fig. 5D).
stimulation on the E2A, AID, IRF4 and PAX5 protein complex, we BCR stimulation reduced the anti-Flag ChIP signal for each of the
infected the B cells with retrovirus expressing either wild-type or AID-Flag, PAX5-Flag and IRF4-Flag viruses that expressed wild-type
CaM-resistant E12 and performed anti-Flag ChIP before and after E12 when the site did not contain an IRF4 target site, independent
BCR stimulation for 30 min or 2 h (Fig. 5). Since the VHJ558 site of the presence of an E2A, AID or PAX5 target site at the DNA seg-
from analysis with the E12-Flag virus did not result in signal clearly ment. In all cases the ChIP signal was 2-fold to 3-fold reduced after
over the background (Fig. 4A), we excluded it from our analysis. both 30 min and 2 h (Fig. 5B–D). In contrast, no reduction of the
BCR stimulation for 30 min reduced the ChIP signal with anti-Flag anti-Flag ChIP signal after BCR stimulation was observed for any
between 2-fold and 8-fold in cells infected with the retrovirus one of the corresponding CaM-resistant viruses at any of these
expressing wild-type E12-Flag for all amplicons with E2A, AID or sites, except the E2A site in the hs3b enhancer where in some
PAX5 target sequence that do not contain an IRF4 target site. This cases a partial reduction was observed. The anti-Flag ChIP signal
reduction remained after 2 h (Fig. 5A). Thus, the binding of E2A stayed or was only relatively little reduced after BCR stimulation for
to the Igh locus is rapidly reduced upon BCR stimulation both on the viruses expressing AID-Flag or PAX5-Flag together with wild-
E2A binding sites and on target sequences for AID and PAX5 where type E12 on the amplicons containing an IRF4 site (Fig. 5B–C). In
the localization of E2A must be indirect. In contrast, with ampli- contrast, for the CaM-resistant E12-Flag virus, the anti-Flag ChIP
cons containing IRF4 target sites, the anti-Flag ChIP signal showed signal increased significantly after BCR stimulation for the viruses
very little or no reduction, independent of the presence of an E2A, expressing AID-Flag or PAX5-Flag together with CaM-resistant E12
AID or PAX5 target sequence (Fig. 5A). Importantly, the reductions on the amplicons containing an IRF4 site (Fig. 5B–C). An exception
in anti-Flag ChIP signals for the amplicons with E2A, PAX5 or AID to this was infection with viruses expressing IRF4-Flag, since there
sequences that do not contain an IRF4 target site were all abol- was relatively little difference in the anti-Flag ChIP signal between
ished by the expression of CaM-resistant E12-Flag after both 30 min the IRF4-Flag viruses co-expressing wild-type E12 or CaM-resistant
J. Hauser et al. / Molecular Immunology 80 (2016) 78–90 85
Fig. 4. Chromatin immunoprecipitation analysis of AID, E2A, PAX5 and IRF4 at the Igh locus. Purified splenic B-cells of mice heterozygous for knock-out of the E2A gene were
infected with empty retrovirus (vector) or viruses expressing E12-Flag (A), or E12 plus AID-Flag (B), E12 plus PAX5-Flag (C) or E12 plus IRF4-Flag (D) from the MSCV-IRES
plasmid and subjected to chromatin immunoprecipitation using the Flag epitope of the tagged protein. The recovered DNA was analyzed by real-time quantitative PCR using
primer pairs amplifying regions of the Igh locus containing target sites for E2A (E-box sequences), PAX5, IRF4 and AID, and transcribed C and C␦ parts of the Igh locus
as indicated. The identical amount of DNA, in mass, was used in the analysis of all samples from the same mouse and all quantifications were from the early log phase
of the PCR. Three mice were analyzed and all PCR measurements were performed in triplicate. The figure shows the amount of PCR product obtained for each site after
immunoprecipitation with anti-Flag (expressed as per mille (‰) of the amount of PCR product before immunoprecipitation). All data are presented as mean ± s.d.
E12 for any of the amplicons containing an IRF4 site (Fig. 5D). For AID at S, the switch region for IgE, was significantly lower in the
four of these sites, the signal increased 1.5-fold to three-fold after cells expressing CaM-resistant E12 than for the wild-type (Fig. 6A),
BCR stimulation and on the fourth site, V1, the signal was relatively which directly correlates with the reduced CSR to IgE in these
modestly affected by the stimulation. The reason for the excep- cells (Fig. 1). Conversely, localization of AID at the switch regions
␥
tion is unknown. However, since this occurred only for IRF4 on S 2b, S␥2c, S␥3 and S␣ was higher in the cells expressing CaM-
amplicons with IRF4 sites, it indicates that the interaction of this resistant E12 than in the wild-type (Fig. 6; P values <0.02 except for
␥
particular protein to its own site in contrast to the other proteins S 3), which corresponds to the increased CSR to IgG2b, IgG2c, IgG3
can remain unaffected after BCR signaling activates CaM to block and IgA in the mutant cells or when over-expressing AID (Fig. 1
the DNA binding of E2A. In summary, the ChIP analysis provides evi- and Supplementary Fig. 1B). Thus, the CaM-resistant mutant has
dence supporting a complex of AID, E2A, PAX5 and IRF4 at target a re-localization of AID between the different switch regions that
sequences of these proteins in the Igh locus and that some com- directly correlates with the shift in CSR. We observed anti-Flag
plexes leave some of the target sites upon BCR stimulation when ChIP signals with each of E2A-Flag, PAX5-Flag and IRF4-Flag over
E12 is wild-type, whereas when E12 is CaM-resistant the com- the vector virus background for the amplicons containing differ-
plexes instead remain bound at these sites and accumulate at other ent switch regions (Fig. 6B-D). Localization of each of E12, PAX5
target sites. and IRF4 was also significantly lower at the switch region S in
To further study whether the reduced CSR to IgE and the inap- the cells expressing CaM-resistant E12 than the wild-type (Fig. 6),
propriate CSRs to IgG2b, IgG3 and IgA in B-cells with CaM-resistant which directly correlates with the reduced CSR to IgE (Fig. 1 and
E12 were due to a shift in the localization of AID, E12, PAX5 Supplementary Fig. 1A). Conversely, for all three viruses localiza-
and IRF4 at the Igh locus, we next analyzed binding of AID and tion of E12, PAX5 and IRF4 was at vector background or close to it
the transcription factors to the switch regions in the locus after for S␥2b, S␥2c, S␥3 and S␣ in cells expressing wild-type E12. These
BCR stimulation by ChIP. We observed anti-Flag ChIP signals with localizations were all higher in the cells expressing CaM-resistant
AID-Flag over the vector background for the amplicons contain- E12 than in the wild-type (Fig. 6; all P values <0.05, except 0.07
ing different switch regions (Fig. 6A). Importantly, localization of for S␣ with IRF4-Flag), which corresponds to the increased CSR to
86 J. Hauser et al. / Molecular Immunology 80 (2016) 78–90
Fig. 5. CaM inhibition of E2A participates in redistribution of AID, E2A, PAX5 and IRF4 at the Igh locus after BCR stimulation. Purified splenic B-cells of mice heterozygous for
R
knock-out of the E2A gene were infected with retrovirus constructs as in Fig. 4 using either wild-type (WT) or CaM-resistant (CaM ) E12. Infected cells were BCR-stimulated
with anti-IgM for the indicated times and then subjected to chromatin immunoprecipitation using the Flag-epitope of the tagged protein and analyzed by quantitative PCR
as in Fig. 4. The Flag-tagged protein was (A) E12, (B) AID, (C) PAX5 and (D) IRF4. The figure shows the amount of PCR product obtained for each site after immunoprecipitation
with anti-Flag compared with the amount of PCR product before immunoprecipitation. These values for anti-Flag–immunoprecipitated DNA are expressed relative to the
R
immunoprecipitated DNA from the corresponding unstimulated cells expressing either WT or CaM E12 that were set as 100% and indicated as a dotted line (bars not shown).
All data are presented as mean ± s.d. based on three different mice. Significant P values for differences between wild-type and CaM-resistant E12 are shown at *P < 0.05,
**P < 0.01, or ***P < 0.001. Some of the PCR amplified segments are in addition to sites for either E2A, AID or PAX5 also sites for IRF4 as indicated.
these classes (Fig. 1). In summary, the CD40L, IL-4 plus BCR stim- to AID Sepharose beads but not to control Sepharose beads with
ulated CaM-resistant mutant has a re-localization of each of AID, Protein A or without coupled protein (Fig. 7A). Analysis of dele-
E2A, PAX5 and IRF4 from S to other switch regions that directly tion derivatives revealed that the AID binding was in the Paired
correlates with the shift in class-switching between these Ig classes homology Domain (PD) of PAX5, since all deletion derivatives of
in the mutant. PAX5 with the PD domain reproducibly bound efficiently to the AID
beads, whereas PAX5 with deletion of only the PD domain repro-
ducible showed no binding over the background (Fig. 7A and F).
3.5. Direct interactions between AID, the PD domain of PAX5 and
E12 bound clearly less than PAX5 to the AID beads but it bound
the bHLH domain of E2A
more than to the background control beads without coupled pro-
tein or with Protein A (Fig. 7B). E12 bound to the AID beads through
To analyze whether AID could interact directly with a protein
its bHLH domain, since E12 lacking the bHLH domain reproducibly
in the identified complex, we next determined if purified recom-
showed no binding over the background, whereas the bHLH domain
binant E12 splice form of E2A (C-terminal 222 amino acids) and
alone was sufficient for binding to the AID beads (Fig. 7B and
full-length recombinant PAX5 and CaM expressed in E. coli bound
F). Notably, E12 lacking the part N-terminal to the bHLH domain
to purified recombinant full-length AID. PAX5 bound efficiently
J. Hauser et al. / Molecular Immunology 80 (2016) 78–90 87
reproducibly bound much more than wild type E12 to the AID
beads, approximately as much as PAX5 to the AID beads. Thus,
sequences N-terminal to the bHLH domain act negatively in the
binding of the bHLH domain to AID. CaM showed no binding to
2+
the AID beads neither in the absence nor in the presence of Ca
(Fig. 7C). The interaction of AID with the PD domain of PAX5 and
the bHLH domain of E12 was specific and does not represent an
unspecific “stickiness” of our AID beads, since three proteins, CaM,
PAX5 without PD domain and E12 without bHLH domain, did not
bind.
To localize the direct interaction between PAX5 and E2A that
we have reported previously (Hauser et al., 2014), we analyzed the
interaction of the deletion derivatives of E12 with PAX5 beads and
the interaction of the deletion derivatives of PAX5 with E12 beads.
PAX5 bound to E12 with the PD domain, since all deletion deriva-
tives of PAX5 with the PD domain reproducibly bound efficiently to
the E12 beads, whereas PAX5 with deletion of only the PD domain
reproducibly showed no binding over the background (Fig. 7D, F
and G). E12 bound to PAX5 with the bHLH domain, since E12 lacking
the bHLH domain reproducibly showed no binding over the back-
ground, whereas the bHLH domain alone was sufficient for binding
to the PAX5 beads (Fig. 7E, F and G). In summary, AID can interact
directly with the PD domain of PAX5 and with the bHLH domain
of E2A, and PAX5 and E2A can interact directly with each other
through the same domains (Fig. 7G).
4. Discussion
CSR to the appropriate isotype depends on the extensive inte-
gration of signals from the BCR, cytokine receptors, Toll-like
receptors, and tumor necrosis factor family members (Pone et al.,
2010). However, the trans-acting factors mediating specific target-
ing of AID to antibody genes and appropriate segments of them
and therefore SH and the appropriate CSR have remained elusive
(Chen and Wang, 2014; Fear, 2013; Kothapalli and Fugmann, 2011;
Stavnezer, 2011; Wang, 2013). To date, no direct coupling between
any gene regulatory transcription factor and the targeting of AID
has been previously demonstrated. In this study we showed that
defective CaM inhibition of E2A reduced CD40L, IL-4 plus BCR stim-
ulated CSR to IgE and instead increased CSR to other Ig classes.
Furthermore, we found that AID and the transcription factors E2A,
PAX5 and IRF4 were in a complex that is redistributed on the Igh
locus upon BCR signaling through CaM binding. Is this a CSR com-
plex and/or an SH complex? In support of a potential role in CSR
we found a shift in physical localisation of AID, E2A, PAX5 and
IRF4 when E2A was CaM-resistant from the usually used switch
sequences of IgE towards localisation at the switch sequences used
in the other CSRs. This strongly argues that this protein complex
directs CSR to the appropriate classes. Is it also used in directing
SH? Several of the target sites where AID, E2A, PAX5 and IRF4 were
localised (Fig. 5) have no known role in CSR but are rather at sites of
SH or close to such sites. This favours the idea that this complex also
Fig. 6. Distribution of AID and transcription factors at the switch regions of the directs the SH. It is hard to envisage a separate set of transcription
Igh locus after BCR stimulation is affected by CaM resistance of E2A. Chromatin
factors directing SH, since E2A, PAX5 and IRF4 are already all known
immunoprecipitation (ChIP) analysis of binding of (A) AID, (B) E12, (C) PAX5, and
key transcription factors in creating the repertoire of antibodies by
(D) IRF4 to the switch regions in the Igh locus. B cells from spleen of mice het-
rearrangements of the Ig loci during B cell development (Johnson
erozygous for knock-out of the E2A gene were activated with CD40L plus IL-4 as in
Fig. 1 and infected with retrovirus expressing either wild-type E12 (WT) or CaM- et al., 2009). The identification of a complex of AID with E2A, PAX5
R
resistant (CaM ) E12 and where indicated another transcription factor or AID with and IRF4 in directing CSR and presumably also SH does however not
a C-terminal Flag on the indicated protein. ChIP against the Flag-epitope was per-
imply that these are the sole gene-regulatory transcription factors
formed on lysates after 2 h of BCR stimulation with anti-IgM, and the recovered involved.
DNA was analyzed by quantitative RT-PCR as in Fig. 5 of segments from the differ-
The finding that each of AID, E2A, PAX5 and IRF4 could be
ent switch regions. The bars show mean fold enrichment of the tagged protein ± s.d.
based on three different mice compared to control cells infected with the retrovirus immunoprecipitated with target sites for each of the others in ChIP
vector. The vector background is indicated as a dotted line.
(Figs. 4 and 5) argues that the proximities between these proteins
shown in the PLA are due to interactions between the proteins.
Furthermore, the effects of the CaM resistance mutation of E12 on
88 J. Hauser et al. / Molecular Immunology 80 (2016) 78–90
Fig. 7. Direct interactions of AID with the basic-helix-loop-helix domain of E12 and the paired domain of PAX5. His6-tagged E12 splice form of E2A (C-terminal 222 amino
acids) and full-length PAX5 and AID were produced in E.coli and the purified proteins were immobilized on cyanogen bromide-activated Sepharose 4B. The binding of the E12
and PAX5 and deletion derivatives of them and E.coli produced CaM to the Sepharose with immobilized proteins was analyzed and representative experiments are shown
in (A)-(E). (A) Binding of PAX5 and deletion derivatives to AID Sepharose. (B) Binding of the E2A and deletion derivatives to AID Sepharose. (C) No binding of CaM to AID
Sepharose. (D) Binding of PAX5 and deletion derivatives to E2A Sepharose. (E) Binding of the E2A and deletion derivatives to PAX5 Sepharose. 0.2 g of the purified proteins
were allowed to interact with 10 l of Sepharose derivatives for 30 min in the presence of 0.1 mM EGTA except in (C) where 0.5 g CaM was used and 0.1 mM CaCl2 was used
instead of the chelator in the indicated lane. After washings, the bound proteins were eluted, separated by SDS-PAGE and detected by western blot against the His6-tags or
with anti-CaM antibody. Preloads were 10% of the amount of protein added to the binding reactions. Controls were with empty Sepharose (Seph.) and Protein-A Sepharose
(A-Seph.) and without addition of protein to Sepharose with the immobilized protein. The lower band in panels with E. coli produced E12 is due to partial protease cleavage
N-terminal to the bHLH domain during E. coli production. (F) Summary of the results of binding of PAX5, E12 and derivatives of them in experiments such as the representative
ones shown in (A)-(B) and (D)-(E). The basic-Helix-Loop-Helix (bHLH) domain, the paired domain (PD) and the homeodomain (HD) are indicated as well as the parts of the
proteins remaining in the various deletion constructs of the proteins. Weak binding is indicated with + and stronger binding with ++. (G) Illustration summarizing that AID
interacts directly with the PD domain of PAX5 and with the bHLH domain of E2A, and PAX5 and E2A interact directly with each other through the same domains. The minus
signs illustrate the negative effect of sequences N-terminal to the bHLH domain of E12 on the binding of the bHLH domain to AID.
the level of this co-immunoprecipitation, even when AID sites are and E2A (Fig. 7). This finding does not exclude that there also are
analyzed with PAX5 or IRF4 and when PAX5 binding sites are ana- interactions between these proteins through another protein in the
lyzed with AID or IRF4, further supports the notion that all four complex. This is likely considering that other proteins coupled to
proteins are in a complex together on the Igh locus. We have pre- the transcription process or to splicing also interact with AID and
viously reported direct interaction between E2A and PAX5 (Hauser are required for the deamination (Kothapalli and Fugmann, 2011;
et al., 2014) and between E2A and CaM (Corneliussen et al., 1994; Stavnezer, 2011). We showed that AID interacts directly with the
Hauser et al., 2008b, 2009; Saarikettu et al., 2004; Verma-Gaur PD domain of PAX5 and with the bHLH domain of E2A (Fig. 7A, B,
et al., 2012), whereas direct interaction between any of the other F and G). Notably, the interaction with AID was much stronger for
proteins had not been previously reported. The findings that the E12 with deletion of the sequences N-terminal to the bHLH domain
proximities between most of these protein pairs, and between CaM than for E12 having these sequences. This indicates a possible regu-
and IRF4 or AID, are influenced when E2A is CaM-resistant (Fig. 3) latory mechanism that the E12-AID interaction might be negatively
suggests direct contact between at least some of AID, IRF4, PAX5 regulated by the N-terminal domain unless an appropriate pro-
and E2A. Here AID was shown to interact directly with both PAX5 tein interaction(s) impedes this inhibition of the interaction. That
J. Hauser et al. / Molecular Immunology 80 (2016) 78–90 89
interaction(s) impeding the inhibition could be by a protein(s) par- References
ticipating in the activation of expression of immunoglobulins, but
Bain, G., Maandag, E.C., Izon, D.J., Amsen, D., Kruisbeek, A.M., Weintraub, B.C., Krop,
at present the identity of such a protein remains to be elucidated.
I., Schlissel, M.S., Feeney, A.J., van Roon, M., et al., 1994. E2A proteins are
What is the biological significance of the AID/E2A/PAX5/IRF4
required for proper B cell development and initiation of immunoglobulin gene
complex changing physical location at the Igh locus after BCR stim- rearrangements. Cell 79, 885–892.
Begum, N.A., Stanlie, A., Nakata, M., Akiyama, H., Honjo, T., 2012. The histone
ulation? A strong stimulation of the membrane-bound antibody
chaperone Spt6 is required for activation-induced cytidine deaminase target
of the BCR indicates to the cell that SH was successful and should
determination through H3K4me3 regulation. J. Biol. Chem. 287, 32415–32429.
be arrested. This signal has been shown to shut-off AID expres- Bolland, D.J., Wood, A.L., Corcoran, A.E., 2009. Large-scale chromatin remodeling at
the immunoglobulin heavy chain locus: a paradigm for multigene regulation.
sion (Hauser et al., 2008b). In contrast to SH, such BCR stimulation
Adv. Exp. Med. Biol. 650, 59–72.
increases CSR (Fig. 1A, B). It is therefore not surprising that the
Bossen, C., Mansson, R., Murre, C., 2012. Chromatin topology and the regulation of
appropriate localization of the AID/E2A/PAX5/IRF4 complex on antigen receptor assembly. Annu. Rev. Immunol. 30, 337–356.
the Igh locus is different before and after BCR stimulation. It is Casellas, R., Basu, U., Yewdell, W.T., Chaudhuri, J., Robbiani, D.F., Di Noia, J.M., 2016.
Mutations, kataegis and translocations in B cells: understanding AID
notable that AID, IRF4, PAX5 and E2A were all found to be in phys-
promiscuous activity. Nat. Rev. Immunol. 16, 164–176.
ical contact with many dispersed segments of the Igh chromatin
Chandra, V., Bortnick, A., Murre, C., 2015. AID targeting: old mysteries and new
distributing over most of the locus, which would require exten- challenges. Trends Immunol. 36, 527–535.
Chen, Z., Wang, J.H., 2014. Generation and repair of AID-initiated DNA lesions in B
sive DNA looping. This is reminiscent of the state of the Igh locus
lymphocytes. Front. Med. 8, 201–216.
when the gene is rearranged to create the repertoire of antibodies
Corneliussen, B., Holm, M., Waltersson, Y., Onions, J., Hallberg, B., Thornell, A.,
during B cell development. When Igh is accessible for the recom- Grundström, T., 1994. Calcium/calmodulin inhibition of basic-helix-loop-helix
transcription factor domains. Nature 368, 760–764.
bination machinery, chromosomal contraction and DNA looping
Fear, D.J., 2013. Mechanisms regulating the targeting and activity of activation
enables recombination, whereas chromosomal expansion inhibits
induced cytidine deaminase. Curr. Opin. Immunol. 25, 619–628.
the recombination events (Bolland et al., 2009; Roldan et al., 2005). Greenbaum, S., Zhuang, Y., 2002. Identification of E2A target genes in B
lymphocyte development by using a gene tagging-based chromatin
During RAG-mediated recombination, the chromatin of the locus
immunoprecipitation system. Proc. Natl. Acad. Sci. U. S. A. 99, 15030–15035.
is organized in compartments containing clusters of loops sepa-
Hauser, J., Saarikettu, J., Grundström, T., 2008a. Calcium regulation of myogenesis
rated by linkers, and the entire repertoire of variable regions merge by differential calmodulin inhibition of basic helix-loop-helix transcription
factors. Mol. Biol. Cell 19, 2509–2519.
and juxtapose to other DNA elements of the gene (Jhunjhunwala
Hauser, J., Sveshnikova, N., Wallenius, A., Baradaran, S., Saarikettu, J., Grundström,
et al., 2008, 2009). The large-scale chromatin structures reported
T., 2008b. B-cell receptor activation inhibits AID expression through
(Bossen et al., 2012; Jhunjhunwala et al., 2008, 2009; Medvedovic calmodulin inhibition of E-proteins. Proc. Natl. Acad. Sci. U. S. A. 105,
et al., 2013; Seitan et al., 2012; Stubbington and Corcoran, 2013) 1267–1272.
Hauser, J., Verma-Gaur, J., Wallenius, A., Grundström, T., 2009. Initiation of antigen
and the changes in them upon productive recombination indicate
receptor-dependent differentiation into plasma cells by calmodulin inhibition
the existence of protein complexes binding to multiple places in Igh
of E2A. J. Immunol. 183, 1179–1187.
and thereby enabling looping between the critical DNA elements Hauser, J., Wallenius, A., Sveshnikova, N., Saarikettu, J., Grundström, T., 2010.
Calmodulin inhibition of E2A stops expression of surrogate light chains of the
during that RAG-mediated recombination. The results presented
pre-B-cell receptor and CD19. Mol. Immunol. 47, 1031–1038.
here suggest that there are corresponding protein complexes dur-
Hauser, J., Verma-Gaur, J., Grundstrom, T., 2013. Broad feedback inhibition of
ing AID-mediated CSR and SH, and that these contain at least AID, pre-B-cell receptor signaling components. Mol. Immunol. 54, 247–253.
Hauser, J., Grundström, C., Grundström, T., 2014. Allelic exclusion of IgH through
E2A, PAX5 and IRF4.
2+ inhibition of E2A in a VDJ recombination complex. J. Immunol. 192, 2460–2470.
BCR stimulation that leads to binding of Ca -loaded CaM to
Hodgkin, P.D., Lee, J.H., Lyons, A.B., 1996. B cell differentiation and isotype
the AID/E2A/PAX5/IRF4 complex changed the distribution of the switching is related to division cycle number. J. Exp. Med. 184, 277–281.
Jarvius, M., Paulsson, J., Weibrecht, I., Leuchowius, K.J., Andersson, A.C., Wählby, C.,
complex at the Igh locus. It is interesting in this context that the pre-
Gullberg, M., Botling, J., Sjöblom, T., Markova, B., et al., 2007. In situ detection
viously reported direct interaction between E2A and PAX5 (Hauser
of phosphorylated platelet-derived growth factor receptor beta using a
et al., 2014) in this study was found to be through the same domains generalized proximity ligation method. Mol. Cell. Proteomics 6, 1500–1509.
Jhunjhunwala, S., van Zelm, M.C., Peak, M.M., Cutchin, S., Riblet, R., van Dongen, J.J.,
as the interaction with AID, the PD domain of PAX5 and the bHLH
Grosveld, F.G., Knoch, T.A., Murre, C., 2008. The 3D structure of the
domain of E2A (Fig. 7). Thus, the DNA interacting PD domain of
immunoglobulin heavy-chain locus: implications for long-range genomic
PAX5 also interacts with both AID and E12 and the DNA interacting interactions. Cell 133, 265–279.
bHLH domain of E2A interacts with each of CaM, AID and PAX5. Jhunjhunwala, S., van Zelm, M.C., Peak, M.M., Murre, C., 2009. Chromatin
architecture and the generation of antigen receptor diversity. Cell 138,
Thus, it is easy to envisage that binding of CaM to the complex of
435–448.
the DNA binding PD and bHLH domains with AID could have an
Johnson, K., Reddy, K.L., Singh, H., 2009. Molecular pathways and mechanisms
effect on the fine DNA binding preferences of one of the domains regulating the recombination of immunoglobulin genes during B-lymphocyte
development. Adv. Exp. Med. Biol. 650, 133–147.
or both of them. The complex interactions between E2A, PAX5 and
Kothapalli, N.R., Fugmann, S.D., 2011. Targeting of AID-mediated sequence
IRF4, their target sites on DNA, and AID and CaM and presumably
diversification to immunoglobulin genes. Curr. Opin. Immunol. 23, 184–189.
other proteins in the complex and how they influence each other Larsson, G., Schleucher, J., Onions, J., Hermann, S., Grundström, T., Wijmenga, S.S.,
2005. Backbone dynamics of a symmetric calmodulin dimer in complex with
will be an important focus of further research.
the calmodulin-binding domain of the basic-helix-loop-helix transcription
factor SEF2-1/E2-2: a highly dynamic complex. Biophys. J. 89, 1214–1226.
Lin, Y.C., Jhunjhunwala, S., Benner, C., Heinz, S., Welinder, E., Mansson, R.,
Sigvardsson, M., Hagman, J., Espinoza, C.A., Dutkowski, J., et al., 2010. A global
Acknowledgements network of transcription factors, involving E2A, EBF1 and Foxo1: that
orchestrates B cell fate. Nat. Immunol. 11, 635–643.
Lu, R., 2008. Interferon regulatory factor 4 and 8 in B-cell development. Trends
We thank Tanzeel Ahmed for a part of the PLA measurements.
Immunol. 29, 487–492.
This work was supported by grants from the Swedish Research
Medvedovic, J., Ebert, A., Tagoh, H., Busslinger, M., 2011. Pax5: a master regulator
Council and the Swedish Cancer Society. of B cell development and leukemogenesis. Adv. Immunol. 111, 179–206.
Medvedovic, J., Ebert, A., Tagoh, H., Tamir, I.M., Schwickert, T.A., Novatchkova, M.,
Sun, Q., Huis In ‘t Veld, P.J., Guo, C., Yoon, H.S., et al., 2013. Flexible long-range
loops in the VH gene region of the Igh locus facilitate the generation of a
diverse antibody repertoire. Immunity 39, 229–244.
Appendix A. Supplementary data
Morisawa, T., Marusawa, H., Ueda, Y., Iwai, A., Okazaki, I.M., Honjo, T., Chiba, T.,
2008. Organ-specific profiles of genetic changes in cancers caused by
activation-induced cytidine deaminase expression. Int. J. Cancer 123,
Supplementary data associated with this article can be found,
2735–2740.
in the online version, at http://dx.doi.org/10.1016/j.molimm.2016.
Odegard, V.H., Kim, S.T., Anderson, S.M., Shlomchik, M.J., Schatz, D.G., 2005. Histone
10.014. modifications associated with somatic hypermutation. Immunity 23, 101–110.
90 J. Hauser et al. / Molecular Immunology 80 (2016) 78–90
Okazaki, I.M., Hiai, H., Kakazu, N., Yamada, S., Muramatsu, M., Kinoshita, K., Honjo, Sharbeen, G., Yee, C.W., Smith, A.L., Jolly, C.J., 2012. Ectopic restriction of DNA
T., 2003. Constitutive expression of AID leads to tumorigenesis. J. Exp. Med. repair reveals that UNG2 excises AID-induced uracils predominantly or
197, 1173–1181. exclusively during G1 phase. J. Exp. Med. 209, 965–974.
Onions, J., Hermann, S., Grundstrom, T., 2000. A novel type of calmodulin Stanlie, A., Aida, M., Muramatsu, M., Honjo, T., Begum, N.A., 2010. 2010: Histone3
interaction in the inhibition of basic helix-loop-helix transcription factors. lysine4 trimethylation regulated by the facilitates chromatin transcription
Biochemistry 39, 4366–4374. complex is critical for DNA cleavage in class switch recombination. Proc. Natl.
Pone, E.J., Zan, H., Zhang, J., Al-Qahtani, A., Xu, Z., Casali, P., 2010. Toll-like receptors Acad. Sci. U. S. A. 107, 22190–22195.
and B-cell receptors synergize to induce immunoglobulin class-switch DNA Stavnezer, J., Björkman, A., Du, L., Cagigi, A., Pan-Hammarström, Q., 2010. Mapping
recombination: relevance to microbial antibody responses. Crit. Rev. Immunol. of switch recombination junctions, a tool for studying DNA repair pathways
30, 1–29. during immunoglobulin class switching. Adv. Immunol. 108, 45–109.
Rajagopal, D., Maul, R.W., Ghosh, A., Chakraborty, T., Khamlichi, A.A., Sen, R., Stavnezer, J., 2011. Complex regulation and function of activation-induced cytidine
Gearhart, P.J., 2009. Immunoglobulin switch mu sequence causes RNA deaminase. Trends Immunol. 32, 194–201.
polymerase II accumulation and reduces dA hypermutation. J. Exp. Med. 206, Stubbington, M.J., Corcoran, A.E., 2013. Non-coding transcription and large-scale
1237–1244. nuclear organisation of immunoglobulin recombination. Curr. Opin. Genet.
Roldan, E., Fuxa, M., Chong, W., Martinez, D., Novatchkova, M., Busslinger, M., Skok, Dev. 23, 81–88.
J.A., 2005. Locus ‘decontraction’ and centromeric recruitment contribute to Tanaka, A., Shen, H.M., Ratnam, S., Kodgire, P., Storb, U., 2010. Attracting AID to
allelic exclusion of the immunoglobulin heavy-chain gene. Nat. Immunol. 6, targets of somatic hypermutation. J. Exp. Med. 207, 405–415.
31–41. Verma-Gaur, J., Hauser, J., Grundström, T., 2012. Negative feedback regulation of
Söderberg, O., Gullberg, M., Jarvius, M., Ridderstråle, K., Leuchowius, K.J., Jarvius, J., antigen receptors through calmodulin inhibition of E2A. J. Immunol. 188,
Wester, K., Hydbring, P., Bahram, F., Larsson, L.G., et al., 2006. Direct 6175–6183.
observation of individual endogenous protein complexes in situ by proximity Wallenius, A., Hauser, J., Aas, P.A., Sarno, A., Kavli, B., Krokan, H.E., Grundstrom, T.,
ligation. Nat. Methods 3, 995–1000. 2014. Expression and recruitment of uracil-DNA glycosylase are regulated by
Söderberg, O., Leuchowius, K.J., Kamali-Moghaddam, M., Jarvius, M., Gustafsdottir, E2A during antibody diversification. Mol. Immunol. 60, 23–31.
S., Schallmeiner, E., Gullberg, M., Jarvius, J., Landegren, U., 2007. Proximity Wang, J.H., 2013. The role of activation-induced deaminase in antibody
ligation: a specific and versatile tool for the proteomic era. Genet. Eng. (N. Y.) diversification and genomic instability. Immunol. Res. 55, 287–297.
28, 85–93. Xue, K., Rada, C., Neuberger, M.S., 2006. The in vivo pattern of AID targeting to
Söderberg, O., Leuchowius, K.J., Gullberg, M., Jarvius, M., Weibrecht, I., Larsson, L.G., immunoglobulin switch regions deduced from mutation spectra in msh2-/-
Landegren, U., 2008. Characterizing proteins and their interactions in cells and ung-/- mice. J. Exp. Med. 203, 2085–2094.
tissues using the in situ proximity ligation assay. Methods 45, 227–232. Zhang, Z., Espinoza, C.R., Yu, Z., Stephan, R., He, T., Williams, G.S., Burrows, P.D.,
Saarikettu, J., Sveshnikova, N., Grundström, T., 2004. Calcium/calmodulin Hagman, J., Feeney, A.J., Cooper, M.D., 2006. Transcription factor Pax5 (BSAP)
Inhibition of transcriptional activity of E-proteins by prevention of their transactivates the RAG-mediated V(H)-to-DJ(H) rearrangement of
binding to DNA. J. Biol. Chem. 279, 41004–41011. immunoglobulin genes. Nat. Immunol. 7, 616–624.
Schrader, C.E., Guikema, J.E., Linehan, E.K., Selsing, E., Stavnezer, J., 2007. Zhuang, Y., Soriano, P., Weintraub, H., 1994. The helix-loop-helix gene E2A is
Activation-induced cytidine deaminase-dependent DNA breaks in class switch required for B cell formation. Cell 79, 875–884.
recombination occur during G1 phase of the cell cycle and depend upon
mismatch repair. J. Immunol. 179, 6064–6071.
Seitan, V.C., Krangel, M.S., Merkenschlager, M., 2012. Cohesin, CTCF and
lymphocyte antigen receptor locus rearrangement. Trends Immunol. 33, 153–159.