| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
3' Enhancer Is Activated by Gradients of Chromatin Accessibility and Protein Association1
* Department of Medicine, Immunobiology Unit, Evans Memorial Department of Clinical Research, Boston Medical Center, Boston, MA 02118;
Department of Animal Biology, School of Veterinary Medicine, University of Pennsylvania, Philadelphia, PA 19104; and
Department of Microbiology, Boston University School of Medicine, Boston, MA 02118
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
locus is recombined following initiation of a signaling cascade during the early pre-B stage of B cell development. The Ig 3' enhancer plays an important role in normal B cell development by regulating locus activation. Quantitative analyses of 3' enhancer chromatin structure by restriction endonuclease accessibility and protein association by chromatin immunoprecipitation in a developmental series of primary murine B cells and murine B cell lines demonstrate that the enhancer is activated progressively through multiple steps as cells mature. Moderate 3' chromatin accessibility and low levels of protein association in pro-B cells are increased substantially as the cells progress from pro- to pre-B, then eventually mature B cell stages. Chromatin immunoprecipitation assays suggest transcriptional regulators of the 3' enhancer, specifically PU.1 and IFN regulatory factor-4, exploit enhanced accessibility by increasing association as cells mature. Characterization of histone acetylation patterns at the 3' enhancer and experimental inhibition of histone deacetylation suggest changes therein may determine changes in enzyme and transcription factor accessibility. This analysis demonstrates activation is a multistep process initiated in early B cell precursors before Igµ recombination and finalized only after the pre-B cell stage. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
. The mechanisms controlling temporal differences in µ vs recombination are unknown. Regulated locus activation is controlled by two interacting regions, the intronic and 3' enhancers (1, 2, 3, 4); however, in vivo footprint analyses clearly demonstrate that delayed Ig activation as compared with µ is not due to an inability of proteins to simply access enhancer sequences (1, 2, 3, 4). Analysis of a developmental series of B cell lines demonstrated similar levels of 3' enhancer chromatin accessibility in all B cells with at least partially recombined Igµ genes, suggesting activation is a temporally compact event beginning and ending at the pre-B cell stage (1, 5). Similarly, nonquantitative DNase I analyses showed little change in accessibility in a comparison between pro- and pre-B cell lines (6), additionally eroding the likelihood that changes in chromatin as the cells transition into pre-B cells is important for to become competent for recombination. In contrast, subsequent work in genetically modified primary B cells demonstrated that footprint patterns over the 3' enhancer, but not the intronic enhancer, changed during the pro- to pre-B transition, suggesting protein occupancy, rather than chromatin structure, changes to facilitate recombination (2). Specifically, hypersensitivity at the PU.1/IFN regulatory factor-4 (IRF-4)3 3' element increased, suggesting that protein occupancy at the 3' enhancer, but not the intronic enhancer, changes as becomes active recombinationally. Mechanisms explaining these likely changes in protein occupancy have not been elucidated. However, these data indicate that assembling a proper nucleoprotein complex, rather than changing chromatin structure per se, is key for regulating . Furthermore, these findings argue that the relevant target of the LPS-induced signaling cascade shown to activate the locus is not the intronic enhancer because NF-B (a LPS target) or a related protein is already occupying the NF-B site in the intronic enhancer before cellular stimulation (2). The latter finding is not unexpected given that the intronic enhancer does not impart stage or lineage-specific rearrangement of a transgene (7). On the basis of the implications of the above studies, we have focused our attention on more clearly understanding the relationship between an active 3' enhancer chromatin structure and the role protein/3' association plays in activating the locus during B cell development.
Transcriptional activation of the 3' locus has been studied extensively outside of the constraints of chromatin structure, usually in the pre-B cell stage or later (8, 9, 10). Elements implicated by these in vitro and in vivo analyses include the cAMP response element (11) and a site occupied by a B cell lineage-specific activator protein (Pax5), an antagonist of PU.1 transactivation (12). However, the most thoroughly characterized 3' enhancer element is the PU.1/IRF-4 composite element. This element is activated due to PU.1 phosphorylation, which facilitates IRF-4 recruitment and subsequent transcriptional stimulation (9, 13). As noted above, the PU.1/IRF-4 composite element is also the element whose occupancy is most dramatically changed in vivo during the pro- to pre-B cell transition (2). Although PU.1 is unable to activate the 3' enhancer from a closed chromatin structure in nonlymphoid cells (14), the data are consistent with the model that PU.1, and by extension, IRF-4, takes advantage of an accessible 3' chromatin structure to activate the locus. However, this model has never been tested in the context of chromatin.
To elucidate the roles chromatin structural changes and transcription factor/3' enhancer association play in appropriate temporal activation of the locus, we quantitatively examined 3' enhancer chromatin accessibility in a developmental series of B cell lines. Interestingly, pre- recombination B cell precursors package the 3' enhancer into chromatin of intermediate accessibility as compared with non-B or -expressing B cells, which package the enhancer into inaccessible or highly accessible chromatin structures, respectively. Acetylation of histones packaging the 3' enhancer is also increased in more mature cell types as compared with pre- recombination cells. Hence, 3' activation is a prolonged multistep process initiated even before Igµ recombination begins. Examination of protein association with the 3' enhancer in B cell lines and primary pro- and pre-B cells demonstrated that documented changes in 3' footprint patterns are likely due to changes in PU.1 and IRF-4 association during locus activation independent of changes in protein expression levels. Overall, our data showing increases in PU.1/3' and IRF-4/3' association by chromatin immunoprecipitation (ChIP) concomitant with increased 3' enhancer chromatin accessibility and histone acetylation support the model that PU.1 recruitment of IRF-4 to the 3' locus is in part facilitated by changes in accessibility of the 3' chromatin structure. locus activation is thereby a protracted multistep process ensuing before Igµ gene recombination.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
Characteristics of murine cell lines used in this study have been described previously (1). Lines are defined by recombination status at the Igµ and loci. Briefly, AH-7 or 63-12 pro-B cells are RAG 1–/– or RAG 2–/–, respectively; hence, these lines have no opportunity to initiate Igµ or Ig recombination and represent early pro-B cells. 38B9 pro-B cells are recombined D-J at the Igµ locus. These cells complete Igµ recombination and initiate Ig sterile transcription and recombination upon LPS stimulation (15). 3-1 are pre-B cells that have completed Igµ recombination (VDJ+, VDJ–) and initiate Ig transcription in response to LPS (16). Wehi 231 cells have both Igµ and recombined; hence, they are considered immature B cells. BAL-17 B cells express surface Ig and model mature splenic B cells in signaling assays. S194 are nonsecreting plasmacytoma cells obtained from American Type Culture Collection (TIB19). 2017 cells are pro-T cells (17) containing both µ and 3' enhancers packaged into closed chromatin structures by several criteria (18, 19, 20). All lymphocytes were grown in RPMI 1640 supplemented with 5% heat-inactivated FCS, 10–5 M 2-ME, penicillin, and streptomycin. The exception was S194 cells, which were grown in DMEM plus 10% horse serum. NIH 3T3 fibroblasts or RAW 264.7 macrophages were grown in DMEM supplemented with 5 or 10% heat-inactivated FCS, respectively, penicillin, and streptomycin. Primary splenic (B2) B cells were prepared from mature BALB/c mice by Thy-1.2-mediated complement lysis, followed by a lympholyte gradient. Cells are routinely >90% pure as analyzed by FACS. These cells are largely follicular B cells but include a small proportion of marginal zone B cells. Primary pro-B cells were isolated from the femurs of 15 BALB/c mice by fluorescence sorting for the B220+CD43+CD23– population (see Fig. 3, A and B). Dead cells were excluded based on forward vs side scatter. Pre-B cells were isolated similarly as the B220+CD43–CD23– bone marrow population. Although CD43 status is a relatively simplistic approach to identifying pro- vs pre-B cells, purifying sufficient numbers of cells for analysis after additional subdivision proved impractical. Primary B cells were formaldehyde fixed for ChIP immediately following purification. For some experiments, B cells were treated with LPS or trichostatin A (TSA) at a concentration of 10 µg/ml or 33 nM, respectively, for 18–24 h.
|
-PU.1 Ig on a PU.1 affinity column. A total of 1.5 µl of this Ab was used per ChIP sample. Similarly, Ab to the full-length IRF-4 protein was raised in rabbits (21) and was used without additional purification from serum. A total of 2.5 µl of -IRF-4 was optimal for ChIP assays. ChIP assays for histone tail modifications used -acetylated H3 (06-599) or H4 (06-598) Abs from Upstate Biotechnology (24 and 6 µl, respectively). Control precipitations were performed with preimmune rabbit serum or rabbit -histidine tag Ab as noted in each figure. ChIP
ChIP was completed on 1–10 x 106 cells as we have described in detail previously (22). Analysis was quantitated by real-time PCR using SYBR green incorporation. Melt peaks indicated formation of a single product in all template-containing reactions. Identity of all PCR products was verified on native 6–8% acrylamide gels. Two methods were used to quantify ChIP products: the 
(Ct) method (23) or the 2(input-ChIP) quantitative PCR method (24), as designated in the figure legends. The former method was used to compare results between primary cell subpopulations and the latter to judge binding in absolute terms compared with the amount of PCR target present in the sample input before enrichment by immunoprecipitation. Primer sequences used in quantitative PCR are listed below. Alternatively, ChIP products in Figs. 7 and 8 were quantified by dot blot after 25 cycles of PCR amplification. Five, 10, or 20 ng of template were amplified in independent reactions to ensure analysis was in the linear range of amplification. Primers for the amplification were 5'-GCACAGAGTACCCACCCATATCTC-3' and 5'-CTTGAAAGGGTGTGGAGTGCACCA-3'. Amplified samples loaded onto Hybond-N+ (Amersham Biosciences) were detected with a probe made from the 1.1-kb EcoRI-SacI fragment of the 3' enhancer. Signals were quantified by phosphorimage analysis to determine the percentage of DNA bound as a fraction of total input DNA. Note that absolute values from the PCR- vs blot-based ChIP quantitation methods are not comparable mathematically, although relative differences can be compared accurately.
|
|
CHART-PCR was completed as detailed previously (25). Briefly, nuclei from 1–2 x 106 cells were isolated (26), then treated for 1 h at 37°C with restriction endonucleases recognizing accessible target sequences in the 3' (Bsu 36i or NcoI, 10 or 30 U, respectively) or µ (PvuII or PstI, 10 or 20 U, respectively) enhancers proximal to the relevant PU.1-binding element. Partially digested DNA was purified using the Qiagen Blood Purification kit (Qiagen) then quantitated with PicoGreen (22). Equivalent amounts (25 ng) of genomic DNA were quantitatively amplified using primers that flank the endonuclease recognition palindrome (see sequences below). Standard curves generated by amplification of 1–50 ng of uncut genomic DNA allowed quantitation of DNA in each enzyme treated sample. DNA accessibility was calculated as follows: (25 – X)/25 x 100, where X = initial DNA quantity according to the standard curve. All cell types, except BAL-17, gave overlapping standard curves upon amplification with a given primer set. For this reason, only BAL-17 samples were quantified based on a BAL-17 genomic DNA standard curve.
Primer sequences
ChIP: µ forward, 5'-AGCAGCTGGCAGGAAGCAG-3', and reverse, 5'-TCAAAACCACTTCTTCAAACCACAG-3'; forward, 5'-TAGCTACCGTCACACTGCTTTGATC-3', and reverse, 5'-GCTGTTGGAGGAGGATGGGAG-3'; 
-globin forward, 5'-GCCTTGCCTGTTCCTGCTC-3', and reverse, 5'-CAGACCATAAACTGTATTTTTCTTATTCAGCCC-3'; and CHART-PCR: µ forward, 5'-TGGCCGATCAGAACCAGAA-3', and reverse, 5'-CCAAATAGCCTTGCCACATGA-3'; /Bsu36i: used ChIP primers; /NcoI forward, 5'-ACTGGCCTGAGAAGATTAAAAAAAGTAATG-3', and reverse, 5'-CAGTGTGACGGTAGCTATGACAGTTG-3'.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
3' enhancer chromatin is accessible in pro-B cells
Although both characterized enhancers associate with proteins before recombination (1, 2), the protein footprinting pattern over only the 3' enhancer changes upon comparison between pro- and pre-B cells (2). Genetic deletion of the 3' enhancer additionally demonstrates that this enhancer plays an important role in rearrangement and expression in vivo (27). To further understand the changes occurring at the 3' enhancer as the locus is activated during B cell development, we measured accessibility of the enhancer to endonucleases as an indication of chromatin structure. CHART-PCR (25) was used to measure quantitatively 3' enhancer chromatin structure in a developmental series of B cell lines using non-B cells as controls. For these assays, two restriction endonucleases that recognize sequences 82 bp apart in the 3' enhancer, Bsu 36i and NcoI (Fig. 1A), were used to probe chromatin accessibility. Analysis with two enzymes eliminates the possibility of drawing conclusions from a single chromatin region packaged into a nonregulated accessible (or inaccessible) structure. Such a region might not provide an accurate representation of 3' chromatin structure. Chromatin accessibility profiles for the 3' enhancer demonstrated three discernible levels of accessibility on a population basis. As expected, nonlymphoid cells such as NIH 3T3 fibroblasts or RAW 264.7 macrophages were virtually inaccessible to Bsu 36i (Fig. 1B, 
). At the opposite end of the spectrum, immature (Wehi 231) and mature (BAL-17) B cells had highly accessible loci, approximating 60% cleavage (Fig. 1B,
locus is actively transcribed in these cells, a highly accessible locus suggests an expected lack of obstruction by chromatin packaging. Interestingly, the locus was packaged into chromatin of intermediate accessibility in pro-B cells, as represented by RAG–/– 63-12 and 38B9 pro-B cells (Fig. 1B, 
). Specifically, accessibility was 
40% in pro-B cells, consistently 20% lower than accessibility measured in more mature cell types. This data extends previously published nonquantitative analysis of other regions in 63-12 cells (6). Chromatin accessibility to NcoI (Fig. 1C) was qualitatively similar to results using Bsu 36i in all unstimulated cell lines. Specifically, 3T3 fibroblast DNA was not cut by NcoI (Fig. 1C, ), whereas pro-B cells were moderately accessible (40% cut, Fig. 1C, ). Cells representing more mature B cell stages were highly accessible, approaching 80% cutting. This correspondence of results with the two enzymes emphasizes the likelihood that neither recognizes an aberrant chromatin structure within the 3' region.
|
activation correlates with a more accessible 3' region, we took advantage of the demonstration that LPS treatment of 38B9 pro-B cells (which are normally D-J recombined at the µ locus and germline at ) activates the locus. locus activation by LPS in these cells can be measured by production of 0 sterile transcripts (15), which we have independently verified (14). Therefore, stimulated cells represent the pre-B stage of development. Interestingly, LPS-treated 38B9 cells also package the 3' enhancer in a highly accessible chromatin structure when probed with Bsu 36i (Fig. 1B,
accessibility measured for more mature B lineage members. In contrast, is only moderately accessible (approximating 40%) to NcoI (Fig. 1C). Because neither enzyme appears to cleave DNA at an unrepresentative structure in five other cell types as discussed above, we discount this possible explanation. Instead, different accessibility to the two enzymes may represent a transient structure formed during the demonstrated prolonged opening of 3' enhancer chromatin. Overall, the locus is packaged into at least three distinct, stable chromatin structures as measured by endonuclease accessibility in CHART-PCR: closed, moderately accessible, and highly accessible. This analysis is consistent with DNase I hypersensitivity site analyses showing hypersensitive sites in the 3' enhancer of both 38B9 pro-B and BALB mature B cells (1). However, the 20–30% differences in chromatin accessibility indicated by the CHART-PCR analysis would be difficult to detect on Southern blots, thus the moderately accessible chromatin structure packaging the 3' enhancer has not been described previously. More importantly, these data show that the first steps of locus activation occur well before the developmental transition from pro- to pre-B cell. Differences in protein occupancy shown by the in vivo footprint assays (1, 2) thereby correlate with notable changes in chromatin structure, eventually forming a highly accessible nucleoprotein structure concomitant with locus activation. The µ enhancer (Eµ) is packaged into an accessible chromatin structure at all stages of B cell development
To determine whether Ig enhancer accessibility and protein association generally increases during B cell development, we analyzed the Igµ intronic enhancer (Eµ). Eµ is activated early in B cell development, as evidenced by production of sterile Iµ transcripts in the earliest identified B cell precursor cell (28). The Eµ is also packaged by hyperacetylated histone H3 in pro- and pre-B cells (29). Chromatin structure of Eµ during maintenance of Igµ transcription in mature B cells has not been directly analyzed. We used CHART-PCR to document putative changes in µ chromatin structure during B cell development, taking advantage of PvuII or PstI recognition elements within the core enhancer (30) to probe accessibility in the same cells used for 3' analysis in Fig. 1. Eµ was in a highly accessible chromatin structure in all B cells tested, as indicated by cleavage of 80% of enhancers in the cellular population by PvuII (Fig. 2A) or PstI (Fig. 2B). Primary splenic B2 cells also package Eµ into a highly accessible chromatin structure as demonstrated by analysis with both enzymes (Fig. 2, A and B,
). These data demonstrate that Igµ enhancer accessibility does not generally change as B cells transition from pro-B to surface Ig-positive effector cells. Hence, the intermediate accessibility demonstrated for the 3' enhancer is a novel characteristic of the locus.
|
3' enhancer/transcription factor association increases during B cell development
PU.1 is an important transcriptional activator for both the 3' enhancer and Eµ (9, 31). PU.1 acts as a chromatin accessibility determinant for µ but not , as indicated by the ability of ectopically expressed PU.1 to increase µ (but not ) chromatin accessibility (14, 18). We questioned whether PU.1 association with the 3' enhancer, suggested by changes in footprinting patterns over the 3' PU.1 site in primary B cells, changes with increased chromatin accessibility of an activating locus. We used PU.1-specific Abs in ChIP assays to demonstrate PU.1/3' enhancer association in primary pro- and pre-B cells sorted from murine bone marrow-based CD23, B220, and CD43 surface expression (Fig. 3, A and B). PU.1 specifically associated with the 3' enhancer in both pro- and pre-B cells (Fig. 3C, or
3' association is additionally enriched to 93.7-fold over 2017 levels in primary splenic B2 cells as compared with pre-B cells (Fig. 3C,
3' association is indicated by the inability of an unrelated -histidine tag-specific Ab to precipitate the 3' enhancer (Fig. 3C, right bars). -PU.1 Ab specificity is demonstrated by the lack of PU.1 association with the inactive -globin promoter in all B cell subpopulations (Fig. 3D). Overall, this result is consistent with the interpretation that PU.1 exploits increased chromatin accessibility (Fig. 1) at the 3' enhancer as the cells activate the locus at the pre-B stage. The ChIP results from primary cells mirror data from B cell line analyses using the same set of cells used for accessibility analyses (D. McDevit and B. Nikolajczyk, unpublished observation).
Additionally, to investigate the relationship between 3' recombinational activation at the pre-B cell stage and PU.1 association with the 3' enhancer, we compared the ability of an epitope-specific -PU.1 Ab to precipitate PU.1 in unstimulated and LPS-stimulated 38B9 cells. This Ab associates with the C-terminal PU.1 peptide. PU.1-specific ChIP demonstrated increased PU.1 association with the 3' enhancer poststimulation (Fig. 3E). Specifically, PU.1 associated 5.1-fold over input genomic DNA in 38B9 cells (Fig. 3E, ) compared with 9.1-fold over input DNA in 38B9S cells (Fig. 3E,
-his/3' association by ChIP (Fig. 3E, right bars) or -PU.1/-globin promoter association (data not shown). This data corroborates data from primary bone marrow B cells showing that PU.1/ association is increased in pre- vs pro-B cells.
To test whether PU.1/DNA association also changes during development at another PU.1-regulated locus characterized by uniform chromatin accessibility, we measured PU.1/Eµ association by ChIP. Consistent with the 3.2-fold increase in apparent PU.1/3' enhancer association at the pro- to pre-B transition, PU.1/Eµ association was 4.2-fold stronger in bone marrow pre-B cells (8.4-fold over input; Fig. 4,
). Although PU.1/Eµ association was somewhat less robust in mature splenic B2 cells (6.8-fold over input; Fig. 4,
-histidine association with the Eµ (Fig. 4, right bars) and lack of PU.1/-globin association (Fig. 3D). Note PU.1/Eµ association (Fig. 4) and PU.1/3' association (Fig. 3C) were measured from the same pool of precipitated B cell DNA in parallel PCR reactions.
|
3' enhancer in an extrachromosomal context (13). If PU.1/3' enhancer association is limited by an intermediately accessible 3' enhancer chromatin structure in pro-B cells, we would expect IRF-4/3' enhancer association to also be limited before locus recombinational activation. To test this prediction, we completed ChIP assays using an -IRF-4 Ab on B cell lines with demonstrated intermediate or high accessibility. These experiments showed IRF-4 associated with the 3' enhancer in all B cells tested, but a significantly higher proportion of 3' enhancers were associated with IRF-4 in Wehi 231 immature B cells as compared with 38B9 pro-B cells, i.e., IRF-4 association increased after locus activation (Fig. 5A). Specifically, IRF-4 enriched 3' DNA 1.4-, 3.7-, or 5.7-fold over input DNA in pro-, late pro-, or immature B cells, respectively. Preliminary analysis shows IRF-4/ association in primary splenic B2 B cells quantitatively approximates Wehi 231 association, predicting IRF-4 association is maximum on the recombined locus. The -IRF-4-specific Ab did not specifically precipitate the inactive -globin promoter (Fig. 5B). Because our previous analyses ruled out the possibility that the PU.1/IRF-4 combination establishes an accessible locus (at least in 3T3 fibroblasts and pro-T cells; Ref. 14), the ChIP data are more likely explained by increased chromatin accessibility in activated pro-B cells. Specifically, the data support the model that increased PU.1/3' association (Fig. 3C) results in increased IRF-4 recruitment to the 3' enhancer and hence increased IRF-4/3' association by ChIP in pre- vs pro-B cells.
|
3' enhancer association
Our preliminary work demonstrated that PU.1 expression levels may, at least in part, explain increased PU.1/3' enhancer association detected in more mature members of the B lineage (D. McDevit and B. Nikolajczyk, unpublished observation). We analyzed PU.1 levels in additional members of the B lineage to additionally define the mechanisms by which PU.1 activates the 3' enhancer. Western blot analyses using an -PU.1-specific Ab demonstrated PU.1 levels increase substantially as cells begin recombining the Igµ locus at the pro-B stage (Fig. 6, top panel, lane 3). Approximately equivalent protein loading is indicated by actin levels detected in each sample (Fig. 6, lower panel). These data show that PU.1 levels in the immature Wehi 231 B cells are low relative to 38B9 (compare Fig. 6, lanes 3 and 4), despite demonstrating that the 3' enhancer is in a highly accessible structure in Wehi 231 but not 38B9 cells (Fig. 1). Clearly PU.1 levels do not strictly correlate with 3' enhancer accessibility or PU.1/3' association by ChIP.
|
3' enhancer has been demonstrated by transient transfection and EMSA (9, 13). Consistent with these data are our ChIP data (Figs. 3C and 5A) that demonstrated that both PU.1 and IRF-4 increasingly associate with the 3' enhancer in parallel in pro- and pre-B cells. We reasoned that if PU.1 recruitment, at least in part, determines IRF-4 association with the endogenous 3' enhancer, IRF-4 protein expression levels may not correlate with changes in 3' chromatin accessibility during B cell development. Western blot analyses of B cell nuclear extracts using an -IRF-4-specific Ab showed that IRF-4 protein is expressed at relatively consistent levels throughout B cell development (Fig. 6, middle panel). The one exception to this paradigm is the very low IRF-4 protein expression in 63-12 cells. Interestingly, a second RAG–/– line, AH-7, expresses Wehi- and 38B9-like IRF-4 levels, additionally suggesting IRF-4 protein levels in pro-B cells approximate levels later in development. Note that although 38B9 (Fig. 6, lane 3) and Wehi 231 (Fig. 6, lane 4) cells express comparable levels of IRF-4 protein, these levels do not necessarily correlate with levels of 3'-associated IRF-4 (Fig. 5A). Regardless of details of protein levels, it is clear that IRF-4 protein expression correlates with neither IRF-4/3' association or 3' accessibility, and hence, higher levels of IRF-4 expressed in more mature B cells are not required to drive IRF-4 onto the intermediately accessible 3' enhancer by simple equilibrium mechanisms. Instead, concomitant increases in PU.1/3' and IRF-4/3' association by ChIP, largely independent of PU.1 or IRF-4 protein expression levels, support the model that PU.1 recruits IRF-4 to the 3' locus as allowed by an increasingly accessible 3' chromatin structure.
Acetylation of 3' enhancer-packaging histone H3 increases upon recombination
The interpretation that association of PU.1 and IRF-4 with the 3' enhancer follows rather than initiates the demonstrated increases in chromatin accessibility does not explain how PU.1 and IRF-4/3' binding is more robust in mature B cells as compared with pre-B cells (Figs. 3C and 5A). Analysis of a third set of staged B cells, 3-1 pre-B and S194 plasmacytoma cells, confirmed that IRF-4 binding to the 3' enhancer was more robust in later cell stages (Fig. 7, A and B, lanes 5 and 6 or lanes 3 and 4, respectively). Because increased histone tail acetylation is widely associated with gene activation (32), we questioned whether this alternate measure of chromatin accessibility would correlate with changes in chromatin structure in later B cells not apparent in restriction endonuclease assays (Fig. 1). We addressed this possibility using ChIP assays to precipitate 3' enhancer DNA associated with -acetylated histone H3 or -acetylated histone H4 in 3-1 vs S194 cells. Acetylated H3 was more commonly associated with the 3' enhancer in S194 plasma cells as compared with 3-1 pre-B cells (Fig. 7A, lanes 7 and 8). Expressing the data relative to levels of DNA precipitated in 3-1 cells (Fig. 7B, lanes 5 and 6) showed a 4.1-fold increase in acetylated H3 associating with 3' in the more mature S194 cell. Similarly, both methods of analyses demonstrated that acetylation of histone H4 packaging the 3' enhancer increased 2.3-fold in S194 cells as compared with 3-1 pre-B cells (Fig. 7B, lanes 7 and 8). This finding extends previous H4 acetylation analyses in 63-12 pro-B and HC8 pre-B cells, which demonstrated that H4 packaging the J region is probably more highly acetylated in the more mature cells (6). H3 acetylation analyses were not included in these related studies. Overall, our data suggest that increased histone acetylation, especially H3 acetylation, correlates with increased transcription factor binding at the 3' enhancer and is consistent with the possibility that the transcription factors are taking advantage of increased chromatin accessibility even at later stages of B cell development.
Differential acetylation of histones packaging the 3' enhancer may indicate that such changes hold biological relevance to locus activation during B cell development. To test this possibility, we took advantage of the demonstration that the locus can be activated by LPS in 3-1 cells, as measured by 0 transcript production (16). ChIP specific for acetylated H3/3' association demonstrated that LPS treatment increases association between acetylated H3 and the enhancer 4.3-fold (Fig. 8, lanes 4 and 5). This increase is comparable to the 4.1-fold difference in acetylated H3/3' association measured in unstimulated 3.1 pre-B cells vs S194 plasma cells (Fig. 7B, lanes 5 and 6). Similarly, association between acetylated histone H4 and the 3' enhancer increases 2.8-fold upon LPS stimulation (Fig. 8, lanes 7 and 8), replicating the 2.3-fold enhancement of acetylated histone H4 association in S194 vs 3-1 cells (Fig. 7B). PU.1/3' association does not change significantly upon LPS treatment (1.1-fold increase post-LPS; Fig. 8, lanes 1 and 2), demonstrating specificity of increased acetylated histone/enhancer association upon experimental activation in pre-B cells.
We speculated that the change in acetylation status of the 3' enhancer during B cell development may be explained either in selective expression of a histone deacetylase (HDAC) in pre-B cells or the selective expression of a histone acetylase in more mature B lineage cells. To differentiate between these possibilities, we treated 3-1 pre-B cells with TSA, a HDAC inhibitor, then measured association of acetylated histones with the 3' enhancer by ChIP. Acetylated H3 precipitates the 3' enhancer 20-fold over control (untreated) 3-1 cells (Fig. 8, lanes 4 and 6). In contrast, association of acetylated histone H4 with the enhancer increases only marginally (2-fold) upon TSA treatment (Fig. 8, lanes 7 and 9). Specificity of these increases is demonstrated by the finding that PU.1/3' enhancer association changes insignificantly upon TSA treatment (0.6-fold; Fig. 8, lanes 1 and 3). Overall, these data suggest that pre-B cells express an inhibitable HDAC that prevents acetylation of 3'-associated histone H3 (and less likely, H4) but does not affect PU.1/3' enhancer binding. Inhibition of this putative HDAC has a particularly pronounced affect on acetylation of histone H3 at the 3' enhancer.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
3' enhancer has largely focused on demonstrating activating protein associations and chromatin structures during the early pre-B cell stage during which, by definition, recombination ensues. Although analyses on the J and V regions have been published (6), we analyzed the region most likely involved in initiating locus activation by selective protein/DNA interaction, the 3' enhancer. A second potential regulatory region, the intronic enhancer, does not impart stage- or lineage-specific expression, at least in some experimental scenarios (7). Furthermore, because intronic enhancer occupancy does not change at the pro- to pre-B transition (2), we did not include it in our studies. The 3' chromatin structure and protein association analyses herein clearly demonstrate activation begins in early pro-B cells before the initiation of Igµ recombination. Specifically, 3' chromatin is moderately accessible in pro-B cells, and intermediate levels of the transcriptional activators PU.1 and IRF-4 associate with this incompletely accessible structure. We propose these events are priming the locus for the more dramatic and functionally important event, V-J recombination. Taken together with the demonstration that RAGs cannot cleave the locus in primary or 63-12 pro-B cells (33, 34), locus accessibility is necessary but not sufficient for recombination, even in the presence of RAG protein in 38B9 cells. Hence, accessibility is probably mechanistically distinct from recombination. Overall, our work demonstrates that locus activation is a relatively lengthy, multistep process, regulated at several points, as opposed to a process initiated only after a signal is generated by surface pre-BCR. Clearly the data are consistent with the demonstration that can recombine in the absence of µ recombination during normal B cell development (35).
Because our analyses measure accessibility to small molecules rather than large complexes such as the transcription or recombinase machinery, the data suggest that the locus is activated by multiple regulated changes in chromatin accessibility. This model can be incorporated into either proposed explanation of activation by large RAG- or polymerase-containing complexes. Both of these models were developed initially to explain allelic exclusion, but due to the nature of the questions asked, these models incorporate explanations of how an accessible chromatin structure is initially established. The first model is based on the demonstration that a single allele is preferentially replicated (36), demethylated (37, 38, 39), and recombined during B cell development. Because this allele is highly likely to be allelically included in mature B cells, it follows that early biochemical changes in that allele predispose the DNA to biologically relevant activation (i.e., recombination). The accessibility data herein are consistent with this model because it is likely that the partial accessibility to endonucleases in Fig. 1 is measuring relatively high accessibility at one allele and low accessibility at the second (excluded) allele. Toward this end, we speculate that the demonstrated 40% accessibility of the enhancer is indicating a high degree of accessibility at one allele in each individual pro-B cell and that 60–80% accessibility (to Bsu 36i or NcoI, respectively) is indicating a relatively uniform activation of the second and more likely excluded allele upon transition to the pre-B cell stage. This interpretation is consistent with the demonstration that the 3' (but not intronic) enhancer plays an important role in allelic exclusion (40) and hence must likely be activated on both alleles in pre-B cells when exclusion is important. The demonstration that sterile transcripts are produced from both alleles in µ+/RAG-null (pre-B) and mature B cells (41) is consistent with the nearly complete access measured from both alleles of pre-B and mature B cells. Taken together, the data argue that establishing access occurs before successful µ recombination and that chromatin accessibility is mechanistically distinct from recombination. The second model describes activation as a rare stochastic event occurring at the pre-B cell stage but not in pro-B cells (42); our data likely detects less dramatic but more prevalent changes in chromatin that are putative harbingers for subsequent accessibility to large complexes such as the polymerase complex measured Liang et al. (42). Although our analyses do not support one model of allelic activation over the other, the data are consistent with either model based on the likely structural limitations put on access to small endonucleases vs large multisubunit complexes.
The new data are consistent with our previous observation that PU.1 is unlikely to serve as a chromatin accessibility factor for the Ig locus (14) and the likelihood that other factors such as Pax-5 establish 3' enhancer accessibility (43), at least to the intermediate level shown in Fig. 1. Although our data do not directly address the suggestion that Pax-5 is the critical accessibility factor (43) responsible for opening a closed structure at the appropriate point in development, the interpretation that PU.1 and IRF-4 take advantage of pre-established accessibility is consistent with this possibility. 3' structure in stimulated 38B9 cells shows that activation results in a structure moderately accessible to NcoI yet highly accessible to Bsu 36i. Because both enzymes yield similar accessibility profiles in all other cell types, it is unlikely that we are assaying nonrepresentative chromatin regions. Because the enzymes cleave DNA at 82 bp or approximately half a nucleosome’s worth of DNA apart, it is instead possible that we have captured incomplete establishment of a mature 3' chromatin structure. The possibility of an intermediate chromatin structure is consistent with the demonstration that association of hyperacetylated histones with the 3' enhancer increases upon stimulation of pre-B cells (Fig. 8). Therefore, we interpret these results as highlighting the continuum of locus opening en route to the highly accessible locus in immature/mature B cells.
Previous analyses on the effect of TSA on activation agree with our suggestion that a pre-B cell HDAC limits the extent of acetylation at histones packaging the 3' enhancer. In these studies, treating p815 mast cells with TSA led to enhanced bulk histone acetylation and specifically increased acetylation of histone H4 packaging the J region (6). However, TSA-mediated histone hyperacetylation in mast cells did not successfully establish the B cell-specific micrococcal nuclease pattern over locus, suggesting that changes in histone acetylation alone cannot establish a B cell-specific chromatin structure. Our new data complement these findings by emphasizing the likelihood that changes in histone acetylation status following establishment of an accessible locus (or at least 3' region) regulates activation of the locus in pre-B cells, introducing the concept that histone acetylation may act as a regulatory rheostat for activation even after initial events establish chromatin accessibility.
Data herein demonstrate that activation of signaling cascades by LPS leads to a new step in the continuum toward full chromatin accessibility and maximal PU.1/3' and IRF-4/3' enhancer association. An alternative interpretation, that the transcription factors are becoming better targets for immunoprecipitation, is unlikely, at least for PU.1, due to our demonstration that the -PU.1 antiserum detects a largely available epitope (or epitopes) in DNA-associated protein (D. McDevit and B. Nikolajczyk, unpublished observation). In vivo footprinting (1, 2) and our new data focus on the likelihood that the 3' enhancer rather than the intronic enhancer mediates LPS-induced activation. One attractive hypothesis stems from the demonstration that LPS treatment activates the cellular kinase CK2, which phosphorylates PU.1, thereby altering PU.1 function as a transcriptional activator (13, 44). Although both LPS (and TSA) have multiple ill-defined effects on cells, the data have presented interesting, testable hypotheses on mechanisms regulating 3' activation. Whether PU.1 phosphorylation alone could lead to increased PU.1/3' association or increased chromatin accessibility remains to be tested. Preliminary data showing no change in PU.1/3' association in EMSA following in vitro PU.1 phosphorylation by CK2 may or may not be relevant to effects on the endogenous chromatinized locus. Of critical importance will be development of a phospho-PU.1-specific ChIP competent Ab for ascertaining phospho-PU.1/3' association in vivo.
It is important to consider that the data shown, as with the published footprint data (1, 2), are averages of a population of cells and that each 3' allele is either cut (or not) in the CHART-PCR endonuclease step and bound (or not) by protein detected in ChIP assays. Hence, our data show the proportion of cells within the population with accessible chromatin and/or protein-associated increases as the cells display maturity characteristics, such as inducible 0 transcription in 38B9 cells. Analysis of a population of cells using a gain-of-signal assay such as ChIP may also explain how our PU.1/IRF-4/ association data corroborate the in vivo footprinting data of Shaffer et al. (2). In the latter analyses, footprint patterns over the 3' enhancer are indistinguishable in naked vs pro-B cell DNA, consistent with meager PU.1/IRF-4/3' association measured by ChIP. Similarly, the substantial changes in footprint pattern detected at the PU.1/IRF composite element in pre- vs pro-B 3' enhancer likely reflect the substantial increases in PU.1 and IRF-4 association with the enhancer in pre- vs pro-B cells detected by the more quantitative ChIP method. Hence, our data provide a more mechanistic and perhaps more quantitative explanation of changes previously demonstrated from the standpoint of DNA protection/hypersensitivity to dimethyl sulfate modification.
The Western blot data for IRF-4 (Fig. 6), although in agreement with relatively invariable IRF-4 mRNA levels before terminal B cell differentiation (45), do not completely agree with IRF-4 protein analysis in a separate developmental series of B cells, wherein IRF-4 levels increase at the pre-B to immature stage of B cell development (46). Our demonstration that IRF-4 levels can vary dramatically between representatives of the same B cell stage (i.e., RAG-null 63-12 and AH-7 pro-B cells) suggests that additional analysis on primary cells will be needed to determine how IRF-4 levels change during B cell development. Regardless of the results of this analysis, it is difficult to argue that IRF-4 protein levels dictate IRF-4/3' association when multiple pieces of evidence, including our ChIP data, are more consistent with this association being determined by PU.1-dependent IRF-4 recruitment to 3' DNA (13, 45). In fact, IRF-4/DNA association in the absence of PU.1 allows IRF-4 to suppress, rather than activate, gene expression in some cases (46). Hence, our interpretation that IRF-4 levels do not dictate increased IRF-4/3' association demonstrated in pre-B vs pro-B cells follows mechanistic analyses from independent investigations.
PU.1 and IRF-4 association can, in the simplest interpretation, serve as markers for highlighting changes in chromatin structure that precede Ig recombination. However, the likelihood that these 3' activators play more important roles in establishing a functional locus is significant based on the rich literature demonstrating the role PU.1 and IRF-4 play in activating the 3' enhancer in a relatively accessible (i.e., transfected plasmid) context. Although these proteins do not appear to dictate chromatin accessibility, they are clearly able to take advantage of access mediated by other (perhaps Pax-5 dominated) mechanisms. Hence, PU.1 and IRF-4 potentially play important roles in the demonstrated progressive changes in at the pro-B cell stage and beyond.
|
Disclosures |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
|
Acknowledgments |
|---|
|
Footnotes |
|---|
1 This work was funded by R01 AI54611 (to B.S.N.) and GM42415 (to M.L.A.). 
2 Address correspondence and reprint requests to Dr. Barbara S. Nikolajczyk, Department of Medicine, Boston Medical Center, 650 Albany Street X-438, Boston, MA 02118. E-mail address: bnikol{at}bumc.bu.edu
3 Abbreviations used in this paper: IRF-4, IFN regulatory factor-4; ChIP, chromatin immunoprecipitation; TSA, trichostatin A; CHART-PCR, chromatin accessibility by real-time PCR; Eµ, µ enhancer; HDAC, histone deacetylase.
Received for publication September 28, 2004. Accepted for publication December 14, 2004.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
3' enhancer. Mol. Cell. Biol. 16:3138.[Abstract]
light chain enhancers in primary pro- and pre-B cells: a model for locus activation. Immunity 6:131.[Medline]
light chain intronic enhancer and 3' enhancer in rearrangement and demethylation. Nat. Immunol. 3:463.[Medline]
locus involves both positive and negative sequence elements in the 3' enhancer that affect synergy with the intron enhancer. J. Biol. Chem. 274:3285. gene 3' enhancer: functional importance of three motifs that demonstrate B-cell-specific in vivo footprints. Mol. Cell. Biol. 12:5206. light chain gene promoter and enhancer are not responsible for B-cell restricted gene rearrangement. Nucleic Acids Res. 17:7403. 3' enhancer: regulation by Id, a repressor of helix-loop-helix transcription factors. Mol. Cell. Biol. 11:1040. gene enhancers synergistically activate gene expression but independently determine chromatin
