Pre-B Cell Receptor Signaling Induces Immunoglobulin κ Locus Accessibility by Functional Redistribution of Enhancer-Mediated Chromatin Interactions

Chromatin conformation analyses provide novel insights into how variable segments in the immunoglobulin light chain gene become accessible for recombination in precursor B lymphocytes.


Introduction
B lymphocyte development is characterized by stepwise recombination of immunoglobulin (Ig), variable (V), diversity (D), and joining (J) genes, whereby in pro-B cells the Ig heavy (H) chain locus rearranges before the Igk or Igl light (L) chain loci [1,2]. Productive IgH chain rearrangement is monitored by deposition of the IgH m chain protein on the cell surface, together with the preexisting surrogate light chain (SLC) proteins l5 and VpreB, as the pre-B cell receptor (pre-BCR) complex [3]. Pre-BCR expression serves as a checkpoint that monitors for functional IgH chain rearrangement, triggers proliferative expansion, and induces developmental progression of large cycling into small resting Ig m + pre-B cells in which the recombination machinery is reactivated for rearrangement of the Igk or Igl L chain loci [3,4].
During the V(D)J recombination process, the spatial organization of large antigen receptor loci is actively remodeled [5]. Overall locus contraction is achieved through long-range chromatin interactions between proximal and distal regions within these loci. This process brings distal V genes in close proximity to (D)J regions, to which Rag (recombination activating gene) protein binding occurs [6] and the nearby regulatory elements that are required for topological organization and recombination [5,7,8]. The recombination-associated changes in locus topology thereby provide equal opportunities for individual V genes to be recombined to a (D)J segment. Accessibility and recombination of antigen receptor loci are controlled by many DNA-binding factors that interact with local cis-regulatory elements, such as promoters, enhancers, or silencers [7][8][9]. The long-range chromatin interactions involved in this process are thought to be crucial for the regulation of V(D)J recombination and orchestrate changes in subnuclear relocation, germline transcription, histone acetylation and/or methylation, DNA demethylation, and compaction of antigen receptor loci [5,10].
The mouse Igk locus harbors 101 functional V k genes and four functional J k elements and is spread over .3 Mb of genomic DNA [11]. Mechanisms regulating the site-specific DNA recombination reactions that create a diverse Igk repertoire are complex and involve local differences in the accessibility of the V k and J k genes to the recombinase proteins [12]. Developmental-stage-specific changes in gene accessibility are reflected by germline transcription, which precedes or accompanies gene recombination [13]. In the Igk locus, germline transcription is initiated from promoters located upstream of J k (referred to as k 0 transcripts) and from V k promoters [14]. Deletion of the intronic enhancer (iEk), located between J k and C k , or the downstream 39k enhancer (39Ek), both containing binding sites for the E2a and Irf4/Irf8 transcription factors (TFs), diminishes Igk locus germline transcription and recombination [15][16][17][18][19]. On the other hand, the Sis (silencer in intervening sequence) element in the V k -J k region negatively regulates Igk rearrangement [20]. This Sis element was shown to target Igk alleles to centromeric heterochromatin and to associate with the Ikaros repressor protein that also colocalizes with centromeric heterochromatin. Sis contains a strong binding site for the zincfinger transcription regulator CTCC-binding factor (Ctcf) [21,22]. Interestingly, deletion of the Sis element or conditional deletion of the Ctcf gene in the B cell lineage both resulted in reduced k 0 germline transcription and enhanced proximal V k usage [21,23]. Very recently, a novel Ctcf binding element located directly upstream of the Sis region was shown to be essential for locus contraction and recombination to distal V k genes [23]. In addition, the Igk repertoire is controlled by the polycomb group protein YY1 [24].
Induction of Igk rearrangements requires the expression of the Rag1 and Rag2 proteins, the attenuation of the cell cycle, and transcriptional activation of the Igk locus, all of which are thought to be crucially dependent on pre-BCR signaling [4,25]. At first, pre-BCR signals synergize with interleukin-7 receptor (IL-7R) signals to drive proliferative expansion of IgH m + large pre-B cells [4]. In these cells, transcription of the Rag genes is low and the Rag2 protein is unstable due to cell-cycle-dependent degradation [26]. Subsequently, signaling through the pre-BCR downstream adapter Slp65 (SH2-domain-containing leukocyte protein of 65 kDa, also known as Blnk or Bash) switches cell fate from proliferation to differentiation [4]. Importantly, Slp65 (i) induces the TF Aiolos, which down-regulates l5 expression [27]; (ii) binds Jak3 and thereby interferes with IL-7R signaling [28]; and (iii) reduces inhibitory phosphorylation of Foxo TFs [29]. All these changes result in attenuation of the cell cycle and thus Rag protein stabilization. Moreover, Rag gene transcription is induced by Foxo proteins [30].
Although rearrangement and expression of the Igk locus can occur independently of IgH m chain expression [31,32], several lines of evidence indicate that pre-BCR signaling is actively involved in inducing Igk and Igl locus accessibility and gene rearrangement. First, surface IgH m chain expression correlates with germline transcription in the Igk locus [33]. Second, in the absence of Slp65, k 0 germline transcription is reduced [34]. Third, mice deficient for Bruton's tyrosine kinase (Btk), which is a pre-BCR downstream signaling molecule interacting with Slp65, show reduced Igl L chain germline transcription and reduced Igl usage [35]. Fourth, transgenic expression of the constitutively active E41K-Btk mutant in IgH m chain negative pro-B cells induces premature rearrangement and protein expression of Igk L chain [34]. Based on fluorescence in situ hybridization (FISH) studies, it has been proposed that in pro-B cells distal Vk and Ck genes are separated by large distances and that the Igk locus specifically undergoes contraction in small pre-B and immature B cells actively undergoing V k -J k recombination [36]. However, it remains unknown how pre-BCR-induced signals affect the accessibility, contraction, and topology of the Vk region, or how they affect the long-range interactions of the k regulatory elements involved in organizing these events.
In this study, we identified the effects of pre-BCR signaling on germline Vk transcription and on the expression of TFs implicated in the regulation of Igk gene rearrangement. We found that the decrease in pre-BCR signaling capacity in wildtype, Btk-deficient, Slp65-deficient, and Btk/Slp65 doubledeficient pre-B cells was paralleled by a gradient of decreased expression of many TFs including Ikaros, Aiolos, Irf4, and (to a lesser extent) E2a, as well as by a decreased Igk locus accessibility for recombination. Several of these factors can mediate longrange chromatin interactions and are known to occupy k regulatory elements that regulate locus accessibility [37][38][39][40]. We therefore sought to analyze the effect of pre-BCR signaling on the higher order chromatin structure organized by these regulatory sequences at the Igk locus. To this end, we performed chromosome conformation capture and sequencing (3C-seq) analyses [41] on pro-B cells and pre-B cells from mice single or double deficient for Btk or Slp65 to evaluate the effects of this pre-BCR signaling gradient on Igk locus topology. These 3C-seq experiments demonstrated that already in pro-B cells the k enhancers robustly interact with the ,3.2 Mb Vk region and its flanking sequences, and that pre-BCR signaling induces accessibility by a functional redistribution of enhancer-mediated chromatin interactions within the V k region.

Author Summary
B lymphocyte development involves the generation of a functional antigen receptor, comprising two heavy chains and two light chains arranged in a characteristic ''Y'' shape. To do this, the receptor genes must first be assembled by ordered genomic recombination events, starting with the immunoglobulin heavy chain (IgH) gene segments. On successful rearrangement, the resulting IgH m protein is presented on the cell surface as part of a preliminary version of the B cell receptor-the ''pre-BCR.'' Pre-BCR signaling then redirects recombination activity to the immunoglobulin k light chain gene. The activity of two regulatory k enhancer elements is known to be crucial for opening up the gene, but it remains largely unknown how the hundred or so Variable (V) segments in the k locus gain access to the recombination system. Here, we studied a panel of pre-B cells from mice lacking specific signaling molecules, reflecting absent, partial, or complete pre-BCR signaling. We identify gene regulatory changes that are dependent on pre-BCR signaling and occur via long-range chromatin interactions between the k enhancers and the V segments. Surprisingly the light chain gene initially contracts, but the interactions then become more functionally redistributed when pre-BCR signaling occurs. Interestingly, we find that the two enhancers play distinct roles in the process of coordinating chromatin interactions towards the V segments. Our study combines chromatin conformation techniques with data on transcription factor binding to gain unique insights into the functional role of chromatin dynamics.

Identification of Genes Regulated by Pre-BCR Signaling
Whereas mice deficient for the pre-BCR signaling molecules Btk and Slp65 have a partial block at the pre-B cell stage [42,43], in Btk/Slp65 double-deficient mice, only very few pre-B cells show progression to the immature B cell stage characterized by functional IgL chain gene recombination [44]. To enable analysis of the effects of pre-BCR signaling on (i) the expression of genes involved in Igk gene rearrangement and on (ii) long-distance chromatin interactions in the Igk locus in pre-B cells in the absence of Igk gene recombination events, we bred Btk and Slp65 single-and doubledeficient mice on the Rag1 2/2 background. In these mice, progression of B cell progenitors to the pre-B cell stage was conferred by the transgenic, functionally rearranged VH81x IgH m chain, which ensures pre-BCR expression and cellular proliferation.
The absence of functional Rag1 protein precludes IgL chain gene rearrangement and cells are completely arrested at the small pre-B cell stage ( Figure 1A).
We performed genome-wide expression profiling of FACSpurified B220 + CD19 + pre-B cell fractions from wild-type (WT), Btk, and Slp65 single-and double-deficient VH81x transgenic Rag1 2/2 mice ( Figure 1A). In these experiments non-VH81x transgenic Rag1 2/2 pro-B cells served as controls. One-way ANOVA analysis using MeV software (p,0.01) [45] revealed that 266 genes were differentially expressed between the five groups of pro-B/pre-B cells ( Figure 1B). When compared with WT VH81x transgenic Rag1 2/2 pre-B cells, 174 genes were up-regulated, whereby the average values of the fold increase were ,1.70, ,3.28, ,3.36, and ,3.47 for Btk 2/2 , Slp65 2/2 , Btk 2/2 Slp65 2/2 VH81x transgenic Rag1 2/2 pre-B cells and non-VH81x transgenic Rag1 2/2 pro-B cells, respectively (see Table S1). A similar gradient of gene expression FACS sorting strategy for purification of pre-B cell fractions from the indicated mice on a VH81x transgenic Rag1 2/2 background. Lymphocytes were gated on the basis of forward/side scatter and B220 + CD19 + pre-B cell fractions were sorted. Virtually all B220 + CD19 + cells were cytoplasmic m heavy chain positive [34], but showed genotype-dependent levels of expression of the CD2 differentiation marker, in agreement with previous findings [34]. (B) DNA microarray analysis of total mRNA from FACS-purified B220 + CD19 + pre-B/pro-B cell fractions from the indicated mice. One-way ANOVA analysis (p = 0.01) identified 266 significantly different genes. MeV hierarchical clustering of gene expression differences are represented in the heatmap. (C) Validation of the expression of TFs implicated in Igk gene rearrangement. Total mRNA isolated from FACS-sorted B220 + CD19 + pre-B/ pro-B cell fractions from the indicated mice was analyzed by quantitative RT-PCR for expression of TFs. Expression levels were normalized to those of Gapdh, whereby the values in WT pre-B cells were set to one. Bars represent mean values and error bars denote standard deviations for four independent mice per group. doi:10.1371/journal.pbio.1001791.g001 changes was apparent from the average values of the fold change for the 192 significantly down-regulated genes, which were ,1.65, ,2.29, ,3.79, and ,4.15 in the four groups of pre-B/pro-B cells, respectively (see Table S2). In a hierarchical clustering analysis of the five groups of B cell precursors, the expression profiles of Btk 2/2 Slp65 2/2 VH81x transgenic Rag1 2/2 pre-B cells and non-VH81x transgenic Rag1 2/2 pro-B cells were very similar ( Figure 1B). This implies that expression of the 266 genes is not substantially influenced by pre-BCR-mediated proliferation, which is still induced in pre-B cells lacking both Btk and Slp65 [44,46] but not in Rag1 2/2 pro-B cells. Consistent with these findings, gene distance matrix analysis revealed a clear gene expression gradient among the five groups of pre-B/pro-B cells, in which Btk 2/2 Slp65 2/2 pre-B and Rag1 2/2 pro-B cells again showed highly comparably expression signatures ( Figure S1).
Next, we used quantitative RT-PCR to confirm the observed differential expression of several TFs. Expression levels of these genes were indeed significantly reduced in a pre-BCR signalingdependent manner, especially for Aiolos, Ikaros, and Irf4, with residual expression levels in Btk 2/2 Slp65 2/2 VH81x transgenic Rag1 2/2 pre-B cells that were ,1%, ,20%, and ,9% of those observed in WT VH81x Rag1 2/2 mice, respectively ( Figure 1C). In addition, we found moderate effects on Obf1 (Oca-B) and E2a with residual expression levels of ,28% and ,44%, respectively. In chromatin immunoprecipitation (ChIP) assays, we observed in pre-B cells substantial binding of E2a protein to the intronic and 39 k enhancer regions and to the three V k regions analyzed. Under conditions of reduced pre-BCR signaling activity, E2a binding to the enhancers was essentially maintained (39Ek) or reduced (iEk), but E2a binding to the V k regions was lost (Table S3). Consistent with the significant reduction of Ikaros expression in Slp65 2/2 pre-B cells, Ikaros binding to both k enhancers and V k regions was undetectable in these cells (Table S3).
Taken together, from these findings we conclude that the five groups of pro-B/pre-B cells, representing a gradient of progressively diminished pre-BCR signaling, show in parallel a gradient of diminished modulation of many genes that signify pre-B cell differentiation, including key genes implicated in Igk gene recombination.
Progressively Diminished V k and J k GLTs in Btk 2/2 , Slp65 2/2 , and Btk 2/2 Slp65 2/2 Pre-B Cells In these expression profiling studies, we only detected limited differences in germline transcription (GLT) over unrearranged J k and V k gene segments, which is thought to reflect locus accessibility [12]. However, we previously showed by serialdilution RT-PCR that the levels of k 0 0.8 and k 0 1.1 germline transcripts, which are initiated in different regions 59of J k and spliced to the C k region [49], are apparently normal in Btk 2/2 pre-B cells, modestly reduced in Slp65 2/2 pre-B cells, and severely reduced in Btk 2/2 Slp65 2/2 pre-B cells [34]. We could confirm these findings for k 0 GLT by quantitative RT-PCR assays on FACS-purified B220 + CD19 + pro-B/pre-B cell fractions ( Figure 2A). In agreement with our reported findings [34], we also found that Btk 2/2 and Slp65 2/2 pre-B cells have defective l 0 transcription, which is initiated 59 of the J l segments ( Figure 2B) [49].
GLT across the V k region showed a similar pattern of sensitivity to pre-BCR signaling: decreased transcription of six individual V k regions tested (V k 3-7, V k 8-24, V k 4-55, V k 10-96, V k 1-35, and V k 2-137) correlated with decreased pre-BCR signaling activity ( Figure 2C) in the pre-B cells of the four groups of mice. GLT over unrearranged V l 1 and V l 2 segments was strongly reduced in the absence of Btk or Slp65, as detected by the expression arrays (Table 1).
These observations indicate that Igk locus accessibility, a hallmark of recombination-competent antigen receptor loci, is progressively reduced under conditions of diminishing pre-BCR signaling.

Pre-BCR Signaling Induces Modulation of Long-Range Chromatin Interactions at the Igk Locus
Accessibility of antigen receptor loci for V(D)J recombination is thought to be initiated by enhancers, in part through long-range chromatin interactions with promoters of noncoding transcription, resulting in the activation of germline transcription [8]. Because pre-BCR signaling affects the expression of GLT and various nuclear proteins that mediate long-range chromatin interactions and bind the k enhancers, it is conceivable that pre-BCR signaling induces changes in the enhancer-mediated higher order chromatin structure of the Igk locus that facilitates V k gene accessibility.
In WT pre-B cells, all three regulatory elements showed extensive long-range chromatin interactions within the V k region and substantially less interactions with regions up-or downstream of the ,3.2 Mb Igk domain ( Figure 3A; see Figure S2, Figure S3, and Figure S4 for line graphs), confirming previous observations [21]. Under conditions of reduced pre-BCR signaling activity, the three Igk regulatory elements still showed strong interactions with the V k region. Surprisingly, even in the complete absence of pre-BCR signaling in Rag1 2/2 pro-B cells, long-range interactions were still observed at frequencies well above those seen in nonlymphoid cells, suggesting that a contracted Igk locus topology is not strictly dependent on pre-BCR signaling ( Figure 3A, Figure S2, Figure S3, and Figure S4). Next, we used 3D DNA FISH analyses using BAC probes hybridizing to the distal V k and C k /enhancer regions to confirm that Igk locus contraction was similar in Rag-1 2/2 pro-B cells and VH81x transgenic Rag-1 2/2 pre-B cells (both showing a contracted topology, compared with noncontracted pre-pro-B cells deficient for the TF E2a; Figure 3B). Nevertheless, we did observe that pre-BCR signaling induced clear differences in interaction frequencies. Whereas an increase in pre-BCR signaling was associated with a decrease in the interaction frequencies between the two k enhancers and regions flanking the Igk locus (as also revealed by more detailed images of selected regions upstream and downstream of the Igk domain; see Figure S5), the overall interaction frequency within the Igk domain appeared unchanged ( Figure S3, Figure S4, and Figure S5). Remarkably, interactions with the Sis element showed quite an opposite pattern: pre-BCR signaling correlated with increased overall interactions within the Igk domain and did not substantially affect interaction frequencies in the Igk flanking regions ( Figure S2 and Figure S5).
Taken together, these analyses show that (i) the Igk locus is already contracted at the pro-B cell stage and that (ii) pre-BCR signaling induces changes in long-range chromatin interactions, both within the Igk locus and in the flanking regions.
Pre-BCR Signaling Enhances Interactions of 39Ek and Sis, But Not iEk, with V k + Fragments The differential effects of pre-BCR signaling on long-range chromatin interactions of the iEk, 39Ek, and Sis elements clearly emerged in a quantitative analysis of the 3C-seq datasets ( Figure 4A; see Materials and Methods for a detailed description of the quantification methods used). When pre-BCR signaling was absent (Rag1 2/2 pro-B cells) or very low (Btk 2/2 Slp65 2/2 pre-B cells), the average interaction frequencies were similar within the ,3.2 Mb V k region and the ,3.2 Mb downstream flanking region, for all three regulatory elements. Interaction frequencies with the upstream flanking region were lower, consistent with the larger chromosomal distance to the three viewpoints. The presence of increasing levels of Btk/Slp65-mediated pre-BCR signaling was associated with reduced interaction of iEk and 39Ek with the Igk flanking regions and with increased interaction of the Sis element and (to a lesser extent) 39Ek with the V k region ( Figure 4A). As a result, for all three regulatory elements pre-BCR signaling resulted in a preference for interaction with fragments inside the V k region over fragments outside the V k region ( Figure S7).
We next focused our analysis on the V k region and compared fragments that harbor a functional V k gene (V k + fragment) and those that do not (V k 2 fragment). When pre-BCR signaling was absent (Rag1 2/2 pro-B cells) or very low (Btk 2/2 Slp65 2/2 pre-B cells), the average interaction frequencies of the Sis or iEk elements with V k + fragments were higher than with V k 2 fragments. The average interaction frequencies of 39Ek with V k + and V k 2 fragments, however, were similar ( Figure 4B). Upon pre-BCR signaling, the Sis element showed an increase in interaction frequencies with both V k + Figure 2. Reduction of Btk/Slp65-mediated pre-BCR signaling is associated with progressive loss of Igk GLT. Quantitative RT-PCR analysis for k 0 (A), l 0 (B), and V k GLT (C) of FACS-sorted B220 + CD19 + pre-B/pro-B cell fractions from the indicated mice on a VH81x transgenic Rag1 2/2 background. Expression levels were normalized to those of Gapdh, whereby the values in WT pre-B cells were set to one. Bars represent mean values and error bars denote standard deviations for four independent mice per group. doi:10.1371/journal.pbio.1001791.g002 and V k 2 fragments, with nevertheless an interaction preference for V k + fragments. In contrast, interaction frequencies between the iEk element and V k + or V k 2 fragments were not modulated by pre-BCR signaling at all ( Figure 4B). The 39Ek element exhibited yet another profile: pre-BCR signaling induced increased interaction frequencies specifically with V k + fragments, while interactions with V k 2 fragments were not notably modulated by pre-BCR signaling ( Figure 4B). When we separately analyzed nonfunctional pseudo-V k genes, we found for the Sis and 39Ek elements that the interaction patterns with functional and nonfunctional V k genes were similar ( Figure S8). In contrast, the iEk enhancer did show an overall increased interaction frequency with V k functional genes, compared with nonfunctional V k genes, a phenomenon which was again independent from pre-BCR signaling ( Figure S8).
The finding that interactions of V k genes with the intronic enhancer are already robust in pro-B cells, while those with the 39k enhancer are dependent on pre-BCR signaling, suggested that for individual V k genes pre-BCR signaling may result in more similar interaction frequencies with the two enhancers. To investigate this, we examined for all individual V k genes the correlation between their 3C-seq interaction frequencies with the iEk and 39k elements and found that these were highly correlated in WT pre-B cells (R 2 = 0.68; Figure 4C). Correlation was severely reduced when pre-BCR signaling was low in Btk 2/2 Slp65 2/2 pre-B cells (R 2 = 0.26; Figure 4C). Similar pre-BCR signalingdependent correlations were observed between V k -interactions with the Sis element and those with the two enhancers ( Figure S9). As the Sis element particularly suppresses recombination of the proximal V k 3 family, we investigated interaction correlations specifically for this V k family. Similar to our findings for all V k genes, a subanalysis showed strong correlations for the interactions of V k 3 family genes with iEk, 39k, and Sis in WT pre-B cells, which were diminished when pre-BCR signaling was low, except for iEk-Sis correlations, which were pre-BCR signaling-independent ( Figure S9).
In summary, we conclude that pre-BCR signaling induces a redistribution of long-range interactions of the iEk, 39Ek, and Sis elements, thereby restricting interactions towards the V k gene region. Moreover, upon pre-BCR signaling the longrange interactions mediated by 39Ek and Sis-but not those mediated by iEk-become enriched for fragments harboring a V k gene, demonstrating increased proximity of 39Ek and Sis to V k genes. Finally, for individual V k genes, the interactions with iEk, 39Ek, and Sis become highly correlated upon pre-BCR signaling, indicating that pre-BCR signals result in regulatory coordination between these three elements that govern Igk locus recombination. In contrast, interactions between genes of the proximal V k 3 family, Sis and iEk-but not 39k-appear to be coordinated already in the absence of pre-BCR signaling.

Long-Range Chromatin Interactions of k Regulatory Elements Correlate with V k Usage
Next, we investigated the effects of pre-BCR signaling on the interaction frequencies of individual functional V k genes with the three k regulatory elements ( Figure 5A,B). The 3C-seq patterns of the majority (,91%) of the 101 individual V k + fragments showed evidence for interaction with one or more of the k regulatory elements (.25 average counts). When comparing Btk 2/2 Slp65 2/2 with WT pre-B cells, we observed that for a large proportion (,38-52%) of V k + fragments, interaction frequencies increased upon pre-BCR signaling ( Figure 5B). Smaller proportions of V k + fragments showed a decrease (,12-29%) or were not significantly affected by pre-BCR signaling (,17-25% with ,1.5-fold change). The observed increase or decrease was not related to proximal or distal location of the V k genes, nor to their sense or antisense orientation (not shown). Distributions of the three different classes of V k + fragments showed substantial differences between the k regulatory elements. For the Sis and 39Ek elements, more V k + fragments showed increased than decreased interactions ( Figure 5B), in agreement with the signaling-dependent increase in average interaction frequencies of all V k + fragments ( Figure 4B). In contrast, for the iEk viewpoint, V k + fragments showing increased and decreased interactions were more equal in number, consistent with the limited effects of pre-BCR signaling on overall iEk interaction frequencies of all V k + fragments ( Figure 4B). Although antigen receptor recombination is in principle regarded as a random process, a significant skewing of the primary Igk repertoire of C57BL/6 mice was recently reported: one third of the V k genes was shown to account for .85% of the V k segments used by B cells [54]. To assess whether a correlation exists between usage of V k genes and their interaction frequencies with k regulatory elements, we divided the V k genes into four usage categories (, 0.1%, 0.1-0.3%, 0.3-0.5%, and .0.5%) and calculated their average 3C-Seq interaction frequencies with Sis, iEk, and 39k ( Figure 5C). In WT pre-B cells, V k usage showed a strong positive correlation with 3C-Seq interaction frequencies for all three regulatory elements (R 2 = ,0.7-0.9; Figure 5C). These correlations were pre-BCR signaling-dependent, since in Btk 2/2 Slp65 2/2 pre-B cells, they were reduced (for iEk; R 2 = 0.33) or absent (for Sis and 39k; R 2 ,0.10 and R 2 ,0.16, respectively) ( Figure 5C).
Collectively, our results indicate that specifically the most frequently used V k genes are the main interaction targets of k regulatory elements, whereby pre-BCR signaling completely underlies this specificity for the Sis and 39Ek elements, and to a lesser extent for iEk.

Long-Range Interactions with k Regulatory Elements
Correlate with the Presence of Ctcf, Ikaros, E2a, and H3K4 Hypermethylation Next, we investigated whether long-range interactions between k regulatory elements and the V k region correlated with the presence of the TFs Ctcf [21], Ikaros [55], and E2a [56], which Mb region containing the Igk locus (yellow shading) and flanking regions (cyan shading) is shown and genes and genomic coordinates are given (bottom). The locations of the two BAC probes used for 3D DNA-FISH are indicated by a green (distal V k probe) and red (proximal C k /enhancer probe) rectangle (bottom). Pre-B cell fractions were FACS-purified from the indicated mice on a VH81x transgenic Rag1 2/2 background (see Figure 1 for gating strategy). Erythroid progenitor cells were used as a nonlymphoid control. (B) 3D DNA-FISH analysis comparing locus contraction in cultured bone-marrow-derived E2a 2/2 pre-pro-B, Rag1 2/2 pro-B, and VH81x Rag1 2/2 pre-B cells (see Figure S6 for phenotype of IL-7 cultured B-lineage cells). Locations of the BAC probes used are indicated at the bottom of panel A. Representative images for all three cell types are shown on the left, quantifications (.100 nuclei counted per cell type) on the right. The red lines indicate the median distance between the two probes. Statistical significance was determined using a Mann-Whitney U test (***p,0.001; n.s., not significant, p$0.05). doi:10.1371/journal.pbio. 1001791.g003 have been implicated in Igk locus recombination [21,37,55,57,58]. Notably, Ikaros and E2a both strongly bind all three k regulatory elements, while the Sis element is also occupied by Ctcf ( [21]; unpublished data).
Remarkably, we found similar striking correlations between the presence of in vivo binding sites for each of these TFs (as determined by ChIP experiments; see Materials and Methods for the relevant references) and long-range chromatin interactions with the k regulatory elements ( Figure 6A-C), even though Ctcf sites are mostly located in between V k genes [21] and Ikaros/E2a sites were frequently found close to V k gene promoter regions ([2]; Figure 7A). Even when pre-BCR signaling was absent (Rag1 2/2 pro B cells) or very low (Btk 2/2 Slp65 2/2 pre-B cells), the average interaction frequencies of the k regulatory elements with fragments containing Ctcf, Ikaros, or E2a bindings sites were higher than those without binding sites. Irrespective of the presence or absence of bindings sites for these TFs, we found that upon pre-BCR signaling interaction frequencies with the Sis element increased and those with the iEk did not change. In contrast, for the 39Ek we found that pre-BCR signaling specifically increased interaction frequencies with fragments occupied by Ctcf, Ikaros, or E2a.
Finally, we found that the presence of di-or trimethylation of histone 3 lysine 4 (H3K4Me2/3), an epigenetic signature associated with locus accessibility [59] and Rag-binding [60,61], also correlated with increased interaction frequencies with k regulatory elements, revealing a similar pre-BCR signaling dependency as seen for the TFs analyzed ( Figure 6D).
We conclude that the presence of essential TFs or H3K4Me2/3 in the V k region strongly correlates with the formation of longrange chromatin interactions with the k regulatory elements, and that for the Sis and 39Ek elements this interaction preference is further enhanced by pre-BCR signaling.

Proximity of V k Genes to E2a Binding Sites Correlates with High V k Usage and Increased Long-Range Chromatin Interactions
Since the long-range interactions with k regulatory elements correlated with the presence of TFs implicated in Igk recombination, we next asked whether the k regulatory elements preferentially interacted with V k genes that are in close proximity to binding sites for Ctcf, Ikaros, or E2a.
Strikingly, the majority of functional V k genes (95/101) was found to have an Ikaros binding site in close proximity-that is, located on the same 3C-seq restriction fragment (average length of ,3 kb, unpublished data) ( Figure 7A). Proximity of V k genes to an E2a binding site (37%) or H3K4Me2/3 positive region (,28%) is more selective, while only a small fraction of V k genes are close to Ctcf binding sites (,12%) ( [22]; Figure 7A). All V k genes marked by E2a, Ctcf, H3K4Me2/3, or a combination of these also contain an Ikaros binding site. Frequently used V k genes (.1.0% usage; 33/ 101 genes) were located in two separate regions, a proximal and a distal region, which also contained virtually all E2a and H2K4Me2/ 3-marked V k genes ( Figure 7A). We found that V k genes marked by both Ikaros and E2a were used substantially more often than those only bound by Ikaros ( Figure 7B), suggesting that these V k genes are preferentially targeted for V k -to-J k gene rearrangement. Our 3C-seq analyses showed that in WT pre-B cells, interaction frequencies with the three k regulatory elements were higher for Ikaros/E2a-marked V k genes compared to genes marked by Ikaros binding alone ( Figure 7C). In fact, V k + restriction fragments containing an Ikaros binding site but not an E2a binding site showed interaction frequencies similar to V k 2 restriction fragments. Under conditions of very low pre-BCR signaling (in Btk 2/2 Slp65 2/2 pre-B cells), we observed strongly reduced interaction frequencies of V k + E2a binding restriction fragments with the Sis and 39Ek elements. These interaction frequencies were in the same range as those of V k 2 fragments or V k + fragments that harbored an Ikaros site only ( Figure 7C). Interaction frequencies with the iEk enhancer, however, were independent of pre-BCR signaling. As shown in Figure 7D, for the majority of Ikaros/E2a-marked V k + fragments (65%), pre-BCR signaling was associated with increased interactions with the Sis and 39Ek elements (comparing wild-type and Btk 2/2 Slp65 2/2 pre-B cells). In these analyses, only ,13.5% and ,5.4% of Ikaros/E2a-marked V k + fragments showed a decreased interaction frequency upon pre-BCR signaling. In contrast, almost equal proportions of Ikaros/E2a-marked V k + fragments showed increased (,37%) and decreased (,30%) interactions with iEk upon pre-BCR signaling.
Taken together, these data reveal strong positive correlations between the presence of E2a binding sites, V k usage, and longrange chromatin interactions with k regulatory elements in pre-B cells. Remarkably, for the iEk element, these correlations are largely independent of Btk/Slp65-mediated pre-BCR signaling, whereas for the 39Ek they are completely dependent on signaling.

Discussion
During B-cell development the pre-BCR checkpoint is known to regulate the expression of many genes, part of which control the increase in Igk locus accessibility to the V(D)J recombinase complex. However, it remained unknown how pre-BCR signaling events affect accessibility in terms of Igk locus contraction and topology.
Here we identified numerous genes involved in IgL chain recombination, chromatin modification, signaling, and cell survival to be aberrantly expressed in pre-B cells lacking the pre-BCR signaling molecules Btk and/or Slp65. We found that GLT over the Vk region, reflecting Vk accessibility, is strongly reduced in these cells. We used 3C-Seq to show that in pro-B cells both the intronic and the 39 k enhancers frequently interact with the ,3.2 Mb Vk region, as well as with Igk flanking sequences, indicating that the Igk locus is already contracted at the pro-B cell stage. 3C-Seq analyses in wild-type and Btk/Slp65 single-and double-deficient pre-B cells demonstrated that pre-BCR signaling significantly affects Igk locus topology. First, pre-BCR signaling  Figure 3A, for the five B-cell precursor fractions representing a pre-BCR signaling gradient. Average interaction frequencies per region were calculated as the average number of 3C-Seq reads per restriction fragment within that region. See Materials and Methods section for more details. (B) Average interaction frequencies within the V k region were determined for fragments that do not (2) contain a functional V k gene and for those that do contain a functional V k gene (+). (C) Correlation plots of average interaction frequencies of the two enhancer elements with the 101 functional V k genes for WT pre-B cells (left) versus Btk 2/2 Slp65 2/2 pre-B cells (right). On the log scale, frequencies ,1 were set to 10 0 . Statistical significance was determined using a Mann-Whitney U test (*p,0.05; **p,0.01; ***p,0.001; n.s., not significant, p$0.05). doi:10.1371/journal.pbio.1001791.g004 reduces the interactions of the intronic and 39k enhancers with Igk flanking regions, effectively focusing enhancer action towards the V k region to facilitate V k -to-J k recombination. Second, pre-BCR signaling strongly increases nuclear proximity of the 39k enhancer to V k genes, whereby this increase is more substantial for more frequently used V k genes and for V k genes close to a binding site for the basic helix-loop-helix protein E2a. Third, pre-BCR signaling augments interactions between k regulatory elements and fragments within the V k region bound by the key B-cell TFs Ikaros and E2a and the architectural protein Ctcf. Fourth, pre-BCR signaling has limited effects on interactions of the intronic k enhancer with fragments within the Igk locus, as this enhancer already displays interaction specificity for functional V k genes and TF-bound regions in pro-B cells. Fifth, pre-BCR signaling has limited effects on the interactions between the intronic or 39k enhancers and fragments that do not contain a V k gene or an Ikaros, E2a, or Ctcf binding site, emphasizing the specificity of pre-BCR signaling-induced changes in Igk locus topology. Sixth, pre-BCR signaling appears to induce mutual regulatory coordination between the three regulatory elements, as their interaction profiles with individual V k genes become highly correlated upon signaling. Finally, pre-BCR signaling increases interactions of the Sis element with DNA fragments in the Igk locus, irrespective of the presence of a V k gene or TF. Collectively, our findings demonstrate that pre-BCR signals relayed through Btk and Slp65 are required to create a chromatin environment that facilitates proper Igk locus recombination. This multistep process is initiated by up-regulation of key TFs like Aiolos, Ikaros, Irf4, and E2a. These proteins are then recruited to or further accumulate at the Igk locus and its regulatory elements, resulting in a specific finetuning of enhancer-mediated locus topology that increases locus accessibility to the Rag recombinase proteins.
Importantly, the presence of strong lineage-specific interaction signals between the C k /enhancer region and distal V k genes in pro-B cells indicates that the Igk locus is already contracted at this stage. In contrast to a previous microscopy study indicating that Igk locus contraction did not occur until the small pre-B cell stage [36], our 3D DNA FISH analysis indeed detected similar nuclear distances between distal V k and the C k /enhancer region in cultured pro-B and pre-B cells. Recently Hi-C was employed to study global early B cell genomic organization whereby substantial interaction frequencies were found between the intronic k enhancer and the V k region in pro-B cells [40]. E2a-deficient pre-pro-B cells, which are not yet fully committed to the B-cell lineage [62], showed very few interactions among the iEk and the distal part of the V k region [40], resembling the interactions we observed in nonlymphoid cells ( Figure 3A). Accordingly, 3D-FISH analysis showed that the Igk locus adopted a noncontracted topology in these pre-pro-B cells ( Figure 3B). These data indicate that Igk locus contraction is already achieved in pro-B cells and depends on the presence of E2a. Supporting this notion, active histone modifications and E2a were already detected at the k enhancers and V k genes at the pro-B cell stage [56,63], whereby E2a was frequently found at the base of long-range chromatin interactions together with Ctcf and Pu.1, possibly acting as ''anchors'' to organize genome topology [40]. The observed correlation between E2a binding, V k gene usage and iEk proximity in pro-B cells ( Figure 5C, Figure 7C) further strengthens an early critical role for E2a in regulating Igk locus topology, V k gene accessibility, and recombination.
Our 3C-seq experiments revealed that pre-BCR signaling is not required to induce long-range interactions between the k regulatory elements and distal parts of the V k locus, indicating that TFs strongly induced by signaling-that is, Aiolos, Ikaros, and Irf4-are not strictly necessary to form a contracted Igk locus. Prime candidates for achieving Igk locus contraction at the pro-B cell stage are E2a and Ctcf, as they have been implicated in regulating Ig locus topology [21,40,64,65] and E2a already marks frequently used V k genes at the pro-B cell stage (Figure 7), although we did observe reduced E2a expression and binding to the iEk enhancer and V k genes when pre-B cell signaling was low ( Figure 1 and Table S3), suggesting that pre-BCR signaling is required for high-level E2a occupancy of the V k genes. We previously reported that Igk gene recombination can occur in the absence of Ctcf and that Ctcf mainly functions to limit interactions of the k enhancers with proximal V k regions and to prevent inappropriate interactions between these strong enhancers and elements outside the Igk locus [21]. Because at the pro-to-pre-B cell transition Aiolos, Ikaros, and Irf4 are recruited to the Igk locus and histone acetylation and H3K4 methylation increases [17,38,63,66], we hypothesize that pre-BCR-induced TFs act upon an E2a/Ctcf-mediated topological scaffold to further refine the long-range chromatin interactions of the k regulatory elements. Hereby, these TFs mainly act to focus and to coordinate the interactions of the two k enhancers to the V k gene segments, in particular to frequently used V k genes, thereby increasing their accessibility for recombination (see Figure 7E for a model of pre-BCR signaling-induced changes in Igk locus accessibility).
In this context, our 3C-seq data show that the two k enhancer elements have distinct roles. Both 39Ek and iEk elements manifest interaction specificity for highly used, E2a-marked, V k genes. However, whereas iEk already shows this specificity in pro-B cells (although pre-BCR signaling does augment this specificity), 39Ek only does so in pre-B cells upon pre-BCR signaling. These observations indicate that iEk is already ''prefocused'' at the pro-B cell stage and that pre-BCR signals are required to fully activate and focus the 39Ek to allow synergistic promotion of Igk recombination by both enhancers (see Figure 7E) [52]. In agreement with such distinct sequential roles, iEk and not the 39Ek was found to be required for the initial increase in Igk locus accessibility, which occurred upon binding of E2a only [37,38,67]. The 39Ek on the other hand requires binding of pre-BCR signaling-induced Irf4 to promote locus accessibility [19,38], followed by further recruitment of E2a to both k enhancers and highly used V k genes (Table S3 and [38,57]).
The Sis regulatory element was shown to dampen proximal V k -J k rearrangements and to specify the targeting of Igk transgenes to centromeric heterochromatin in pre-B cells [20]. As Sis is extensively occupied by the architectural Ctcf protein and deletion of Sis or Ctcf both resulted in increased proximal V k usage [21,23], it was postulated that Sis functions as a barrier element to prevent the k enhancers from too frequently targeting proximal V k genes for recombination. In this context, we now provide evidence that interactions between the proximal V k genes, Sis, and iEkbut not 39k-are already coordinated before pre-BCR signaling occurs ( Figure S9). Perhaps not surprisingly, Sis-mediated longrange chromatin interactions displayed a pattern and pre-BCR signaling response that was different from the k enhancers. Unlike for the enhancers, upon pre-BCR signaling, Sis-mediated interactions with regions outside the Igk locus were maintained and Figure 6. Long-range chromatin interactions of k regulatory elements correlate with TF binding and histone modifications. (A-D) For fragments within the V k region, average 3C-seq interaction frequencies were calculated for fragments that did (+) or did not (2) contain binding sites for TFs or H3K4 histone modifications (as determined by previous ChIP-Seq studies; see Materials and Methods for references). Data for the three viewpoint and the five B-cell precursor fractions representing a pre-BCR signaling gradient are shown for Ctcf (A), Ikaros (B), E2a (C), and H3K4 di-and tri-methylation (Me2/3). Statistical significance was determined using a Mann-Whitney U test (*p,0.05; **p,0.01; ***p,0.001; n.s., not significant, p$0.05). doi:10.1371/journal.pbio.1001791.g006 interaction within the V k region increased, irrespective of the presence of V k genes or TF binding sites. Because Sis is involved in targeting the nonrecombining Igk allele to heterochromatin [20], the observed interaction pattern of the Sis element might reflect its action in pre-B cells to sequester the nonrecombining Igk locus and target it towards heterochromatin. This might also explain the increased interaction frequencies of Sis with highly used V k genes upon pre-BCR signaling ( Figures 5C and 7C), as such highly accessible genes likely require an even tighter association with Sis and heterochromatin to prevent undue recombination.
Surprisingly, we observed a striking correlation between Ikaros binding and V k gene location (94% of V k genes were in close proximity to an Ikaros binding site; Figure 7A). Although Ikaros and Aiolos have a positive role in regulating gene expression during B-cell development [55,58] and Ikaros is required for IgH and IgL recombination [39,58], Ikaros has also been reported to silence gene expression through its association with pericentromeric heterochromatin [68] or through recruitment of repressive cofactor complexes [69,70]. Recruitment of Ikaros to the Igk locus was found increased in pre-B cells as compared to pro-B cells [63], in agreement with its up-regulation in pre-B cells (Figure 1). Furthermore, Ikaros binds the Sis element, where it was suggested to mediate heterochromatin targeting of Igk alleles by the Sis region [20]. Aiolos, although not essential for B-cell development like Ikaros [58,71], is strongly induced by pre-B cell signaling and has been reported to cooperate with Ikaros in regulation gene expression [27]. Although their synergistic role during IgL chain recombination has not been extensively studied, the Ikaros/Aiolos ratio changes upon pre-BCR signaling (Figure 1). Increased recruitment of Ikaros/Aiolos to V k genes and the k enhancers likely increases Igk locus accessibility and contraction (see Figure 6), as Ikaros was very recently shown to be essential for IgL recombination [58]. On the other hand, it is conceivable that on the nonrecombining allele, increased recruitment of Ikaros/ Aiolos to V k genes and the Sis region could facilitate silencing of this allele. Further investigations using allele-specific approaches [72] will be required to clarify the allele-specific action of the Sis element during Igk recombination.
In summary, by investigating the effects of a pre-BCR signaling gradient-rather than deleting individual TFs-we have taken a more integrative approach to study the regulation of Igk locus topology. Our 3C-Seq analyses in wild-type, Btk, and Slp65 singleand double-deficient pre-B cells show that interaction frequencies between Sis, iEk, or 39 Ek and the V k region are already high in pro-B cells and that pre-BCR signaling induces accessibility through a functional redistribution of long-range chromatin interactions within the V k region, whereby the iEk and 39Ek enhancer elements play distinct roles.

Quantitative RT-PCR and DNA Microarray Analysis
Extraction of total RNA, reverse-transcription procedures, design of primers, and cDNA amplification have been described previously [21]. Gene expression was analyzed using an ABI Prism 7300 Sequence Detector and ABI Prism Sequence Detection Software version 1.4 (Applied Biosystems). All PCR primers used for quantitative RT-PCR of TFs or k 0 , l 0 , and V k GLT are described in [21], except for Obf1 (forward 59-CCTGGCCACC-TACAGCAC-39, reverse 59-GTGGAAGCAGAAA CCTCCAT-39, obtained from the Roche Universal Probe Library).
Biotin-labeled cRNA was hybridized to the Mouse Gene 1.0 ST Array according to the manufacturer's instructions (Affymetrix); data were analyzed with BRB-ArrayTools (version 3.7.0, National Cancer Institute) using Affymetrix CEL files obtained from GCOS (Affymetrix). The RMA approach was used for normalization. The TIGR MultiExperiment Viewer software package (MeV version 4.8.1) was used to perform data analysis and visualize results [45]. One-way ANOVA analysis of the five experimental groups of B Figure 7. Proximity of V k genes to E2a binding sites correlates with frequencies of long-range interactions. (A) Schematic representation of the Igk locus, showing the location of all functional V k (grey, top), J k and C k gene segments, and the k regulatory elements Sis, iEk, and 39Ek. MAR, matrix attachment region. V k genes within close proximity (as defined by colocalization on the same 3C-Seq restriction fragment) to the indicated TFs or H3K4 hypermethylation (as detected by previous ChIP-seq studies; see Materials and Methods for references) are shown. At the bottom, highly used (.1.0% used) V k gene segments are depicted (orange), which cluster within two large high-usage domains (yellow shading). Primary V k gene usage data was taken from [54]. (B) Average usage of V k genes marked only by an Ikaros binding site or those marked by binding sites of both Ikaros and E2a. (C) Comparison of average interaction frequencies (for the three k regulatory elements indicated) between V k 2 fragments (no V k ), V k + fragments containing an Ikaros binding site only, and V k + fragments containing both an Ikaros and E2a binding site. Bars represent average frequencies for Btk 2/2 Slp65 2/2 pre-B cells (yellow) and WT pre-B cells (grey). (D) Classification of V k + fragments, containing an Ikaros binding site only (top) or containing both an Ikaros and E2a binding site (bottom), based on the effect of pre-BCR signaling on their interactions with the three k regulatory elements indicated. Increase and decrease were defined as .1.5-fold change of interaction frequencies detected in WT pre-B cells versus Btk 2/2 Slp65 2/2 pre-B cells. (E) Proposed model of pre-BCR signaling-mediated changes in k enhancer action. In pro-B cells (left) the enhancers show minimal coordination and their interactions are not yet (fully) focused on the V k genes. Upon pre-BCR signaling and differentiation to pre-B cells (right), TFs bind the locus to coordinate enhancer action and focus their interactions to the V k genes, inducing germline transcription (GLT) and accessibility to the V(D)J recombinase. See Discussion for more details. Statistical significance was determined using a Mann-Whitney U test (*p,0.05; **p,0.01; ***p,0.001; n.s., not significant, p$0.05). doi:10.1371/journal.pbio.1001791.g007 cells was used to identify genes significantly different from wildtype VH81X Tg Rag1 2/2 pre-B cells (p,0.01).

Chromatin Immunoprecipitation (ChIP)
ChIP experiments were performed as previously described [76] using FACS sorted bone marrow pre-B cell fractions (0.3-2.0 million cells per ChIP). Antibodies against E2a (sc-349, Santa Cruz Biotechnology) and Ikaros (sc-9861, Santa Cruz Biotechnology) were used for immunoprecipitation. Purified DNA was analyzed by quantitative RT-PCR as described above. Primer sequences are available on request.
Chromosome Conformation Capture Coupled to High-Throughput Sequencing (3C-Seq) 3C-Seq experiments were essentially carried out as described previously [21,41]. For 3C-Seq library preparation, BglII was used as the primary restriction enzyme and NlaIII as a secondary restriction enzyme. 3C-seq template was prepared from WT E13.5 fetal liver erythroid progenitors and FACS-sorted bone marrow pro-B cell or pre-B cell fractions (see above) from pools of 4-6 mice. In total, between 1 and 8 million cells were used for 3C-seq analysis. Primers for the Sis, iEk, and 39Ek viewpointspecific inverse PCR were described previously [21]. 3C-seq libraries were sequenced on an Illumina Hi-Seq 2000 platform. 3C-Seq data processing was performed as described elsewhere [41,77]. Two replicate experiments were sequenced for each genotype and viewpoint, and normalized interaction frequencies per BglII restriction fragment were averaged between the two experiments.
For quantitative analysis, the Igk locus and surrounding sequences were divided into three parts (mm9 genome build): a ,2 Mb upstream region (chr6:65,441,978-67,443,029; 759 fragments), a ,3.2 Mb V k region (chr6:67,443,034-70,801,754; 1,290 fragments) and a downstream ,3.2 Mb region (chr6:70,801,759-73,993,074; 1,143 fragments). For each cell type (as described above) sequence read counts within individual BglII restriction fragments were normalized for differences in library size (expressed as ''reads per million''; see [74]) and averaged between the two replicates before further use in the various calculations. Very small BglII fragments (,100 bp) were excluded from the analysis. Fragments in the immediate vicinity of the regulatory elements (chr6:70,659,392-70,693,183; 10 fragments) were also excluded because of high levels of noise around the viewpoint, a characteristic of all 3C-based experiments. V k gene coordinates (both functional genes and pseudogenes) were obtained from IMGT [11] and NCBI (Gene ID: 243469) databases. V k gene usage data (C57BL/6 strain, bone marrow) were obtained from [54]. ChIP-seq datasets were obtained from [21] (Ctcf), [55] (Ikaros), and [56] (E2a, H3K4Me2, and H3K4Me3). V k genes were scored positive for TF binding sites or for a histone modification, if they were located on the same BglII restriction fragment (corresponding to the 3C-Seq analysis).
3D DNA FISH was performed as described previously [79] with BAC clones RP23-234A12 and RP23-435I4 (located at the distal end of the V k region and at the C k /enhancer region, respectively; Figure 3A) obtained from BACPAC Resources (Oakland, CA). Probes were directly labeled with Chromatide Alexa Fluor 488-5 dUTP and Chromatide Alexa Fluor 568-5 dUTP (Invitrogen) using Nick Translation Mix (Roche Diagnostics GmbH).
Cultured primary cells were fixed in 4% paraformaldehyde, and permeabilized in a PBS/0.1% Triton X-100/0.1% saponin solution and subjected to liquid nitrogen immersion following incubation in PBS with 20% glycerol. The nuclear membranes were permeabilized in PBS/0.5% Triton X-100/0.5% saponin prior to hybridization with the DNA probe cocktail. Coverslips were sealed and incubated for 48 h at 37uC, washed, and mounted on slides with 10 ml of Prolong gold anti-fade reagent (Invitrogen).
Pictures were captured with a Leica SP5 confocal microscope (Leica Microsystems). Using a 636 lens (NA 1.4), we acquired images of ,70 serial optical sections spaced by 0.15 mm. The datasets were deconvolved and analyzed with Huygens Professional software (Scientific Volume Imaging, Hilversum, the Netherlands). The 3D coordinates of the center of mass of each probe were transferred to Microsoft Excel, and the distances separating each probe were calculated using the equation: !(Xa2 Xb)2+(Ya2Yb)2+(Za2Zb)2, where X, Y, and Z are the coordinates of object a or b.

Statistical Analysis
Statistical significance was analyzed using a nonparametric Mann-Whitney U test (IBM SPSS Statistics 20). The p values, 0.05 were considered significant.

Accession Numbers
3C-seq and microarray expression datasets have been submitted to the Sequence Read Archive (SRA, accession number SRP032509) and Gene Expression Omnibus (GEO, accession number GSE53896), respectively. Figure S1 Gene distance matrix analysis using gene expression profiling data from pre-B/pro-B cell fractions representing a pre-BCR signaling gradient. Microarray expression profiling was performed on three or four independent FACS-purified B220 + CD19 + pre-B cell fractions from wild-type (WT), Btk, and Slp65 single-and double-deficient VH81x transgenic Rag1 2/2 mice (see Figure 1 for gating strategy). The TIGR Multi Experiment Viewer software package (MeV version 4.8.1) was used to perform a one-way ANOVA analysis (p,0.01) and identify genes differentially expressed within the five B-cell fractions (versus VH81X Tg Rag1 2/2 pre-B cells). The software was subsequently used to create a gene distance matrix of highly significant genes, resulting in the depicted plot. Differences in gene expression profiles are depicted as a color code; darker colors indicate greater similarity, and brighter colors less similarity between groups. Consistent with the unsupervised clustering analysis shown in Figure 1B Figure S7 Pre-BCR signaling is associated with an increase in the ratio of interactions inside the Igk locus over interactions outside the Igk locus. The differential effects of pre-BCR signaling on long-range chromatin interactions of the iEk, 39Ek, and Sis elements, as measured in 3C-seq datasets, were quantified (see Figure 4A). For all three regulatory elements, the y-axis shows the ratio of the average interaction frequencies per BglII fragment inside the ,3.2 Mb Igk locus over the average interaction frequencies per BglII fragment in the flanking regions (,2.0 Mb upstream together with 3.2 Mb downstream). Analyses of the five groups of B cell precursors, representing a gradient of pre-BCR signaling, are shown, revealing that pre-BCR signaling is associated with a preference for interaction with fragments inside the V k region over fragments outside the V k region. (TIF) Figure S8 Interaction frequencies of k regulatory elements with nonfunctional V k genes as measured in B cell fractions representing a pre-BCR signaling gradient. Quantitative analysis of 3C-Seq datasets obtained for the five B cell precursor fractions representing a pre-BCR signaling gradient, using the three indicated k regulatory elements as viewpoints. Average interaction frequencies within the V k region were determined for fragments that do not contain any V k gene (''no V k ''), those that contain a nonfunctional V k gene (''pseudo V k ''), and those that contain a functional V k gene (''functional V k ''). See Materials and Methods section for more details on analysis methods. Statistical significance was determined using a Mann-Whitney U test (n.s., not significant, p$0.05). (TIF) Figure S9 Correlations between the V k interaction profiles of the three k regulatory elements. Correlation plots of average interaction frequencies of the three regulatory elements with the 101 functional V k genes (A) or only the V k 3 gene family (B) are shown for WT pre-B cells (top, gray labels) versus Btk 2/2 Slp65 2/2 pre-B cells (bottom, yellow labels). Note that under conditions of low pre-BCR signaling (Btk 2/2 Slp65 2/2 pre-B cells) correlation strength is significantly reduced, with the exception of the correlation between de V k 3 interaction profiles of the Sis and iEk elements. (TIF)