Molecular Biology Reports 25: 1–12, 1998. 1c 1998 Kluwer Academic Publishers. Printed in Belgium.

The integrated activities of IRF-2 (HiNF-M), CDP/cut (HiNF-D) andH4TF-2 (HiNF-P) regulate transcription of a cell cycle controlled humanhistone H4 gene: mechanistic differences between distinct H4 genes

Farah Aziz1, Andre J. van Wijnen1, Patricia S. Vaughan1, Shujian Wu2, A. Rauf Shakoori1,Jane B. Lian1, Kenneth J. Soprano1;3, Janet L. Stein1 & Gary S. Stein1;�

1Department of Cell Biology, 55 Lake Avenue North, University of Massachusetts Medical Center, Worcester, MA01655, USA; 2Department of Microbiology and Immunology, 3Fels Institute for Cancer Research and MolecularBiology, Temple University School of Medicine, 3400 Broad Street, Philadelphia, PA 19140, USA; �Author forcorrespondence

Received 18 November 1996; accepted 8 January 1997

Key words: histone H4, cell cycle, interferon regulatory factor, homeodomain protein, transcription

Abstract

Maximal transcription of a prototypical cell cycle controlled histone H4 gene requires a proliferation-specificin vivo genomic protein/DNA interaction element, Site II. Three sequence-specific transcription factors interactwith overlapping recognition motifs within Site II: interferon regulatory factor IRF-2 (HiNF-M), the putative H4subtype-specific protein H4TF-2 (HiNF-P), and HiNF-D which represents a complex of the homeodomain proteinCDP/cut, CDC2, cyclin A and pRB. However, natural sequence variation in the Site II sequences of differenthuman H4 genes abolishes binding of specific trans-acting factors; the functional consequences of these variationshave not been investigated. To address the precise contribution of H4 promoter factors to the level of H4 genetranscription, we performed a systematic mutational analysis of Site II transcriptional motifs. These mutants weretested for ability to bind each of the Site II cognate proteins, and subsequently evaluated for ability to confer H4transcriptional activity using chimeric H4 promoter/CAT fusion constructs in different cell types. We also analyzedthe effect of over-expressing IRF-2 on CAT reporter gene expression driven by mutant H4 promoters and assessedH4 transcriptional control in cells nullizygous for IRF-1 and IRF-2. Our results show that the recognition sequencefor IRF-2 (HiNF-M) is the dominant component of Site II and modulates H4 gene transcription levels by 3 fold.However, the overlapping recognition sequences for IRF-2 (HiNF-M), H4TF-2 (HiNF-P) and CDP/cut (HiNF-D)together modulate H4 gene transcription levels by at least an order of magnitude. Thus, maximal activation of H4gene transcription during the cell cycle in vivo requires the integrated activities of multiple transcription factorsat Site II. We postulate that the composite organization of Site II supports responsiveness to multiple signallingpathways modulating the activities of H4 gene transcription factors during the cell cycle. Variations in Site IIsequences among different H4 genes may accomodate differential regulation of H4 gene expression in cells andtissues with unique phenotypic properties.

Introduction

The coordinate regulation of multiple histone genesin conjunction with DNA synthesis is required forthe ordered packaging of newly replicating DNA intonucleosomes during S phase [1–4]. This organizedassembly of chromatin is necesssary for maintaininggenome structure during successive cell cycle stages,

as well as for supporting cell- and tissue-specific tran-scriptional requirements. Histone genes represent aparadigm for the study of proliferation- and tissue-specific gene-regulatory mechanisms operative in anumber of diverse vertebrate species [5–22]. HistoneH4 represents the most highly conserved nucleosomalprotein and is encoded by a multi-gene family in mosteukaryotes [1, 23–25]. Human histone H4 gene tran-

2

scription has been most extensively examined with ahighly expressed H4 gene designated FO108 [22, 25].This gene has been shown to be transcribed in a cellcycle controlled manner [26] and is regulated by a com-plex array of distal and proximal cis-acting elements[27–30].

The proximal promoter of this H4 gene containstwo in vivo genomic protein/DNA interaction domains,Sites I and II [31], each having intricate and overlap-ping arrangements of recognition elements for interac-tions with multiple nuclear factors in vitro [32, 33].Site II is required and sufficient for proliferation-specific transcriptional activity. At least three distinctclasses of transcription factors (HiNF-M, -P and -D)whose DNA binding activities are independently reg-ulated [34, 35] interact with Site II. HiNF-M whichrecently has been identified as the oncoprotein IRF-2[36], and the related tumor suppressor protein IRF-1[37] each recognize the same element (M-box) locatedin the most distal segment of H4-Site II [34, 35]. H4TF-2 (HiNF-P) is a putative H4 gene specific factor [32,38]; the binding sites for IRF-2 (HiNF-M) and H4TF-2(HiNF-P) are distinct but overlap [33]. HiNF-D repres-ents a multi-component protein complex which con-tains the homeodomain protein CDP/cut as the DNAbinding subunit, as well as several key cell cycle reg-ulators, including the cyclin dependent kinase CDC2,cyclin A and pRB [39, 40]. CDP/cut (HiNF-D) inter-acts with the entire Site II, including the M- and P-boxes, as well as the proximally located TATA box[32, 33].

Because of the molecular complexity of Site II,one key question that remains to be resolved is: Whatare the precise functional contributions of individu-al recognition motifs and cognate factors in determ-ining the levels of H4 gene transcription in multiplecell types? In particular, we have observed naturalsequence variations in the Site II sequences of differ-ent H4 genes, which abolish binding of specific tran-scription factors [33]. This raises the question: Doesthe absence of distinct transcription factor recognitionsites have functional consequences for H4 gene tran-scription, and could this imply heterogeneity in tran-scriptional regulation of different H4 genes? In thisstudy, we address these questions by systematicallyanalyzing the effects of multiple distinct mutations ineach of the Site II recognition motifs on H4 transcrip-tion in vivo in several cell types.

Materials and methods

Mutagenesis of reporter gene constructs

Reporter gene constructs containing mutations in H4-Site II were prepared by replacing wild type sequenceswith mutant oligonucleotide cassettes using constructpFP-1 [32], which contains the wildtype H4 proxim-al promoter spanning nt�240 to �38 (relative to theATG start codon; mRNA cap site at nt �30). As avestige of the original design [33], most mutant oli-gonucleotides also contain a four nucleotide 50 linkersegment (50dGATC), which when incorporated into theH4 promoter will result in the substitution of three non-essential residues at the 50 border of Site II. For properdirect comparison, the resulting constructs were ana-lyzed relative to the wild type oligonucleotide (TM-3) containing only these linker-derived substitutions.Mutant test-oligonucleotides spanning nt �93 to �53of the sense-strand (Table 1) were gel purified [41] andeach annealed with a universal adaptor (ADP-3) span-ning nt�67 to�38 (plus 12 nt of PstI linker sequence)of the anti-sense strand. Single-stranded overhangs ofthe DNA hybrids were filled in using Klenow poly-merase and phosphorylatedusing T4 kinase. The linkersegment was cleaved with PstI and the resulting mutantH4-Site II fragments were unidirectionally cloned intothe NaeI/PstI sites of pFP-1. The 1.7 kB PstI/PstI frag-ment spanning the chloramphenicol acetyl transferase(CAT) gene derived from pPHCAT [26] was then inser-ted in each of the resulting mutant H4 promoter con-structs to yield chimeric H4 promoter/CAT reportergene constructs. Sequence analysis of both the sense-and anti-sense strands confirmed the presence of thedesigned mutations in each of the constructs.

Protein/DNA interaction experiments

Gel shift assays were performed as described previ-ously [32, 33] with the following modifications. Probesspanned the MunI/HindIII fragments of pFP-1 andmutant H4 promoter plasmids (nt �113 to �38) andwere produced by 50 end-labeling at the HindIII siteusing -32P-ATP and T4 polynucleotide kinase. Bind-ing reactions for HiNF-D, as well as for HiNF-M and -P contained different non-specific competitor DNAs tooptimize detection of the individual factors. Each reac-tion contained 10 fmole probe (0.5 nm), approximately5�g HeLa nuclear protein, as well as 100 ng/�l salmonsperm DNA [for detection of HiNF-M and H4TF-2] ora mixture of 100 ng/�l poly G/C DNA and 10 ng/�l

3

Table 1. Mutational analysis of H4 Site II recognition elements. The wild type (WT) sequences of the conserved region in Site II (nt�93 to �67)of the H4 (FO108) gene is shown in the first row. The left column indicates the name of the deletion-mutant. The nucleotides that differ from thewild type Site II sequences are indicated by small caps and deletions are indicated by closed triangles. The right column shows the effects of themutations on Site II binding of H4TF-2 (HiNF-P), IRF-2 (HiNF-M) and CDP/cut (HiNF-D), which are abbreviated, respectively, as M, P and D

poly I/C DNA [for detection of HiNF-D]. Electro-phoresis of the HiNF-D protein/DNA complex withH4-Site II was performed in 4% (80:1) native polyac-rylamide gels using 0.5 X Tris-borate/EDTA [41] asthe running buffer, whereas HiNF-M and HiNF-P pro-tein/DNA complexes were resolved in 4% (20:1) nativepolyacrylamidegels using 1 X Tris-glycine/EDTA buf-fer [41]. Competition analyses were performed with apanel of wild type and mutant oligonucleotides that arediagnostic for each of the H4-Site II binding proteins[33]. In each case, oligonucleotides were added to afinal concentration of 50 nm (approximately 100 foldmolar excess).

Transient expression analysis

Wild type and mutant reporter gene constructs weretransiently transfected into human HeLa S3 cervicalcarcinoma cells, normal diploid rat calvarial osteo-blasts (ROB), and rat ROS 17/2.8 osteosarcoma cells.HeLa S3 cells were cultured in DMEM (Gibco/BRL),

whereas ROB and ROS 17/2.8 cells were cultured inF12 medium. Each cell line was maintained with com-plete media under an atmosphere of 5% CO2 at 37 �C.HeLa S3 cells were seeded at a density of 1 � 106

cells per 100 mm plate, and transfected using theN,N-bis (2-hydroxyethyl)-2 aminoethanesulfonic acid(BES) supported calcium phosphate method [41] with20 �g purified plasmid DNA. Cells were harvested48 hr after transfection and refeeding. ROB and ROS17/2.8 cells were seeded at a density of 0:45 � 106

cells per 100 mm plate and were transfected with theDEAE-Dextran method using 20 �g purified plasmidDNA. Cells were harvested after 72 h.

Three independent transfection experiments wereperformed for each cell line with each series of report-er gene constructs, and each construct was analyzedin triplicate per transfection. CAT activity was assayedusing conditions within the linear range of the CATenzyme. CAT activity was determined by autoradio-graphy and by direct counting using a BetaScope 603analyzer (Betagen, MA). The level of expression for

4

each mutant reporter gene construct was defined aspercent conversion relative to the wild type controlconstruct.

Statistical evaluation of the data was performed byDr Stephen Baker (Academic Computing Department,University of Massachusetts Medical Center) usinganalysis of variance for repeated measures (ANOVA).CAT activity levels were transformed using natural log-arithms to improve compliance of the error to the nor-mal distribution. Statiscal significance is defined asdifferences having a probability under the null hypo-thesis of P < 0:05.

Co-transfection experiments

Experiments in which IRF-1 or IRF-2 was co-transfected with H4 promoter/CAT reporter gene con-structs were performed in hamster ts13 cells. Cellswere grown in DMEM containing 10% fetal calf ser-um and plated in 60-mm tissue-culture dishes at a con-centration of 5 � 106 cells/dish, and 24 h later cellswere transfected with the calcium phosphate method.Transfections were done with a total of 17 �g plas-mid DNA. This amount represents a mixture of theH4/CAT reporter gene construct (5 �g), the pSV-�galconstruct (2 �g) which functions as the internal stand-ard for transfection efficiency, and either pCMV-IRF2or pCMV-IRF1 (up to 10 �g) for expressing IRF pro-teins as described previously [36]. The amount of DNAin each calcium phosphate precipitate was kept con-stant by supplementing the transfection mixture withthe empty expression vector (pRC/CMV). The precip-itates were removed 24 h after transfection, and thecells were refed with fresh medium. Cell lysates wereprepared 24 h later, and assayed for �-galactosidaseactivity and CAT activity. CAT values were correctedfor transfection efficiency using �-galactosidase activ-ity. Experiments with cells nullizygous for IRF-2 andIRF-1 were performed as described previously [36].

Results

The distal segment of Site II is essential for high levelhistone H4 gene transcription

The sequences spanning the distal segment of Site IIlocated immediately upstream of the TATA-box in theH4 gene are highly conserved among vertebrate speciesand are also located in analogous regions in the pro-moters of several other human histone H4 genes [33].

Figure 1. Optimal detection of three distinct sequence-specificfactors binding to Site II requires different experimental conditions.Different Site II complexes are observed with the Site II probe (nt�113 to �38) depending on the non-specific competitor DNA inthe binding reaction (e.g., presence of random DNA, left, or polyG/C DNA, right) or electrophoretic conditions (e.g., differences inelectrophoretic running buffer and acrylamide:bisacrylamide cross-linking ratios)(see Materials and methods). Identity of specific HiNFcomplexes is established here by a diagnostic set of competitor oli-gonucleotides with mutations that specifically abolish (minus signs)or do not affect (plus signs) the binding of different factors [33, 36].The complexes mediated by H4TF-2 (HiNF-P), IRF-2 (HiNF-M)and CDP/cut (HiNF-D) are abbreviated, respectively, as M, P andD. All complexes not indicated by arrows in the left or right panelrepresent non-specific complexes. The CDP/cut (HiNF-D) complexis not detected in the presence of random DNA (from salmon sperm),and the H4TF-2 (HiNF-P) complex is not detected in the presenceof poly G/C DNA.

Probes spanning the wild type Site II sequences medi-ate formation of at least three distinct protein/DNAcomplexes in gel shift assays, which are formed byCDP/cut (HiNF-D), IRF-2 (HiNF-M) and H4TF-2(HiNF-P), but detection of these complexes stronglydepends on the binding conditions and electrophoreticparameters (Figure 1).

Deletion analysis has shown that Site II alone cansupport a low but significant level of histone H4 [27,28] or CAT reporter [26, 29] gene transcription in the

5

Figure 2. Schematic representation of transcription factor binding sites in Site II (nt �97 to �47) of the H4 gene promoter. The top of thediagram shows the relative location of elements for transcription factors (ovals and rounded boxes) in the H4 promoter [22]. Also shown aretwo in vivo genomic protein/DNA interaction domains Sites I and II (boxes), which were charactarized by DNaseI footprinting on sense andanti-sense strands (lines) or DMS fingerprinting (ovals) in the intact cell [31]. The location of sequence motifs (M-, C- and P-boxes) involvedin the binding of Site II binding proteins is indicated below the sequence. Minimal binding domains for CDP/cut (HiNF-D), IRF-2 (HiNF-M)and H4TF-2 (HiNF-P) are depicted by thin rounded boxes, and ovals symbolize methylation interference contacts for each factor.

absence of Site I, which when present strongly aug-ments transcription via interactions with Sp-1 and ATFproteins [29, 30]. To obtain an estimate of the maximalextent to which the conserved element in Site II con-tributes to H4 gene transcription within the native con-text of the entire proximal promoter, we introduced 7distinct nucleotide substitution mutations (mutant MC-7) into Site II. These substitutions are located at keyconserved nucleotides between nt �97 and �77, andencompass a series of in vivo G residue protein/DNAcontacts [31]. Most of these mutations correspond withmethylation interference contacts for Site II bindingproteins (Table 1 and Figure 2), and incorporation ofthese mutations into a Site II oligonucleotide (MC-7), abolishes binding of HiNF-M and HiNF-P, whilebinding of HiNF-D to Site II is severely reduced (Fig-

ure 3). Hence, the MC-7 oligonucleotide representsa triple mutant with respect to the binding of theseSite II factors. To assess directly the effects of thesemutations on H4 gene transcription, the MC-7 oligo-nucleotide was incorporated into an H4/CAT vector(see Materials and methods). The wild type TM-3 oli-gonucleotide was incorporated into the same parentvector for comparison. The TM-3 probe mediates theexpected binding of HiNF-M, HiNF-P and HiNF-Din gel shift assays (Figure 3). Upon transient trans-fection into proliferating HeLa S3 cells, the mutantMC-7/CAT construct displays up to 10 fold reducedpromoter activity relative to the wild type constructTM-3/CAT (Figure 4). Thus, the recognition motifsin Site II together are capable of modulating H4 genetranscription by an order of magnitude.

6

Figure 3. Effect of Site II mutations on the binding of IRF-2 (HiNF-M), H4TF-2 (HiNF-P) and CDP/cut (HiNF-D). The mutant H4 pro-moters GT-9, INS-10, SUB-11 and MC-7 were radio-labelled andused as probes in gel shift assays under different experimental condi-tions as described in Figure 1. Unlabelled specific mutant competitorDNA oligonucleotides were added to confirm the identity of each ofthe Site II binding proteins (indicated by arrowheads; same abbrevi-ations as in Figure 1).

The HiNF-P/H4TF-2 binding motif in the histone H4gene is dispensable for promoter activity in vivo

Located in the center of H4-Site II is a NGGTCCGNNmotif (nt �75 to �67) (Figure 2) that represents themost highly conserved sequence in vertebrate histoneH4 genes [33]. For example, this conserved elementis located in both the human H4-FO108 gene [33],and an analogous human H4 gene (H4.A) character-ized by Heintz and colleagues [38]. Cross-competitionassays have shown that both genes interact with HiNF-P, but only the H4-FO108 gene interacts additionallywith HiNF-M and HiNF-D. The NGGTCCGNN-motifcomprises the main cluster of methylation interferencecontacts for HiNF-P, and the integrity of this motif isessential for HiNF-P binding [33, 38].

Elimination of the NGGTCCGNN-motif by sub-stitution of all 9 nucleotides (mutant GT-9) abolishesHiNF-P binding, but has no effect on the interactionsof HiNF-M or HiNF-D with H4-Site II (Figure 3).Functional activity of the mutant GT-9 promoter in

Figure 4. Functional analysis of mutant H4 gene promoters bytransfection of H4/CAT reporter gene constructs in three differentcell types. Panel A: CAT activity of mutant H4 promoters (listedbelow the diagram) was normalized relative to the wild type con-struct. Results for human HeLa S3 cervical carcinoma cells (blackbars), ROS 17/2 osteosarcoma cells (white bars), or normal diploidrat osteoblasts (ROB; hatched bars). Error bars represent the stand-ard deviation for each set of samples, and the symbols above thebars (star, open caret and closet caret) indicate statistical signific-ance relative to the corresponding control. The graph is based on atleast three independent experiments with the full set of constructsusing triplicate samples. The ability of each mutant promoter to bindH4-Site II transcription factors (P, M, D) is listed below the diagramfor reference. Panel B: representative CAT assay with cell lysatesderived from cells transfected with the constructs in panel A.

transcription assays is comparable to that of the wildtype TM-3 promoter (Figure 4). Thus, integrity of thehighly conserved NGGTCCGNN motif, and recogni-tion by its cognate factor HiNF-P, is dispensable forhigh level transcription of the H4-FO108 gene whenother transcriptional motifs in Site II are unperturbed.

7

Specific variations in the natural sequences of humanhistone H4 genes alter transcriptional activity

To assess the extent to which natural sequence vari-ation in the promoters of human H4-FO108 and H4.Agenes may affect transcription, we designed two dis-tinct mutations designated INS-10 and SUB-11. Theconserved region of Site II in the H4-FO108 gene dif-fers from the H4.A gene by two insertion mutations(contained in INS-10) and six substitution mutations(represented in mutant SUB-11) [33]. One insertionmutation in the INS-10 promoter is located at nt�86in the center of the HiNF-M recognition site (M-box),whereas the second insertion is immediately down-stream of the NGGTCCGNN-motif at nt �67. Thesubstitution mutations of the SUB-11 promoter areclustered between the positions of the insertions (nuc-leotide substitutions at nt �82, �78, �76, �75, �68and �67; see Table 1). Gel shift analysis shows thatthe INS-10 promoter fragment interacts with HiNF-P and HiNF-D, but due to its mutation of the M-boxhas lost the ability to bind HiNF-M (Figure 3). Incontrast, mutations in the SUB-11 promoter fragmentaffect HiNF-D binding, but allow binding of HiNF-Mand HiNF-P. These results establish that the INS-10and SUB-11 promoters represent unique and specificmutants for HiNF-M and HiNF-D, respectively.

In vivo analysis of histone promoter activity showsthat mutant INS-10 has 3 fold reduced transcriptionalactivity. In contrast, transcription mediated by mutantSUB-11 is comparable to that of wild type TM-3(Figure 4). The result with mutant INS-10 indicatesthat natural sequence variation has functional con-sequences for the extent to which H4 promoters areactive. Because INS-10 represents a specific HiNF-M recognition site mutant, the observed decrease inreporter gene expression with the INS-10/CAT con-struct suggests that HiNF-M is important for high levelH4 gene transcription. However, the effect observedwith the INS-10 promoter is several-fold less than thatobtained with the MC-7 promoter. This indicates thatadditional Site II factors contribute to full activity of theH4 promoter. We note that the two insertions in Site IIembodied by mutant INS-10 potentially may affectthe spatial alignment of other histone gene promoterfactors interacting with Sites I and II. Additional M-box mutants were assayed to exclude this confoundingfactor for a definitive conclusion (see below). Absenceof a statistically significant effect on transcription formutant SUB-11 suggest that the role of HiNF-D, sim-ilar to HiNF-P, in mediating high levels of histone

H4 gene transcription is limited. The results obtainedwith the INS-10 and SUB-11 promoter mutants sug-gests that distinct types of natural sequence variationbetween human histone H4 genes have different effectson H4 gene transcription.

Similarities in mutational effects on H4 genetranscription in distinct proliferating cell typesdisplaying different cell growth and tissue-specificphenotypic properties

The MC-7, GT-9, INS-10 and SUB-11 mutant CATconstructs were also analyzed in ROS 17/2.8 osteo-sarcoma cells and normal diploid calvarial osteoblasts(ROB) to assess the mutational effects on H4 genetranscription in different cell types. Similar to HeLaS3 cervical carcinoma cells, these cell types containdetectable levels of HiNF-M, -P and -D when activelyproliferating [33–35]. The results show that the mag-nitude of mutational effects for each of the mutantsis very similar in each cell type (Figure 4). Thus,the relative importance of transcriptional recognitionmotifs and cognate factors in these three proliferatingcell types is quantitatively similar, regardless of dif-ferences in the expression of tissue-specific and cellgrowth related phenotypic properties.

Different mutations affecting the same specificH4-Site II recognition motifs have similar effects onH4 promoter activity

We tested a set of three additional H4-Site II promotermutants (TCN-12, FAM-14 and FAM-15) for effectson factor binding and transcription. Mutant TCN-12contains three nucleotide substitutions at nt�78,�76and�75), which do not affect binding of the three H4-Site II binding proteins (Figure 5). However, FAM-14 contains an additional precise deletion of nt �79,which eliminates HiNF-P binding but has no effect oneither HiNF-M or HiNF-D (Figure 5). FAM-15 con-tains a deletion in the distal terminus of H4-Site IIwhich abrogates HiNF-M binding, but does not affectHiNF-P or HiNF-D (Figure 5). Functional analysisof the FAM-14 promoter in three different cell typesshows that transcriptional activity of the FAM-14/CATconstruct is approximately equivalent to that of wildtype constructs TM-3/CAT and TCN-12/CAT. This res-ult indicates that abrogation of the HiNF-P interactionwith H4-Site II together with the concomittant altera-tion of the spacing between the M-box and the TATA-box, does not have a significant effect on H4 promoter

8

Figure 5. Effect of Site II mutations on the binding of IRF-2 (HiNF-M), H4TF-2 (HiNF-P) and CDP/cut (HiNF-D). The mutant H4 pro-moters TM-3, TCN-12, FAM-14 and FAM-15 were analyzed forbinding of the three Site II binding proteins (abbreviations as inFigure 2).

activity (Figure 6). The similarities in the results forthe HiNF-P mutant promoters GT-9 and FAM-14 (Fig-ures 4 and 6) corroborate the conclusion that HiNF-Pdoes not contribute significantly to the level of H4 genetranscription when other Site II factors remain capableof binding. Moreover, the difference in spacing of theM- and TATA-box between these constructs, whichcould alter possible interactions of HiNF-M with TATAbinding protein TFII-D, does not significantly affectthe outcome of the results.

In contrast, the FAM-15/CAT construct isexpressed at a 3 fold reduced level relative to the wildtype TM-3/CAT construct. The three fold reductionin reporter gene expression obtained with both INS-10/CAT and FAM-15/CAT (Figures 4 and 6), whicheach represent HiNF-M mutant promoters, supportsour model in which the HiNF-M interaction with H4-Site II plays a key role in determining the level of H4transcription. In addition, the effects observed with theINS-10 and FAM-15 promoters are quantitively simil-ar, although these constructs differ in the spacing with-in Site II (INS-10) or between Sites I and II (FAM-15)which could change spatial interactions between pro-moter factors. Taken together, the similarities in resultsfor GT-9 versus FAM-14 and INS-10 versus FAM-15 inthree different cell types are consistent with the conceptthat differences in the spatial alignments of histone H4

Figure 6. Functional analysis of mutant H4 gene promoters bytransfection of H4/CAT reporter gene constructs in three differentcell types. Panel A: CAT activity of each of the mutant H4 promoters(listed below the diagram) was normalized relative to the wild typeconstruct (see Figure 4 for explanation of symbols). Panel B showsthe experimental results of a representaive set of samples.

gene transcription factors binding to mutant promotersdo not have dominant effects on determining the levelof reporter gene transcription.

Targeted mutation of the HiNF-M/IRF-2 binding sitereduces H4 promoter activity

The mutations tested thus far suggest a key role forHiNF-M in H4 gene transcription. However, the INS-10 mutant which selectively abolishes HiNF-M bind-ing also may cause spatial differences in the alignmentof Site II binding proteins. Therefore, we introduceda precise dinucleotide substitution (mutant MSP-16)

9

Figure 7. Effect of Site II mutations on the binding of IRF-2 (HiNF-M), H4TF-2 (HiNF-P) and CDP/cut (HiNF-D). The wild type (108)and mutant H4 promoters (MSP-16, MPM-17 & IGM-18) were ana-lyzed for binding of the three Site II binding proteins (abbreviationsas in Figure 2).

in the region of the M-box that does not overlapwith the HiNF-P binding site. This mutation preventsbinding of HiNF-M but not HiNF-P or HiNF-D toSite II based on gel shift analysis (Figure 7). Thismutation and additional substitution of three nucle-otides in the NGGTCCGNN motif (mutant MPM-17)(Table 1) abolishes interactions of both HiNF-M andHiNF-P with Site II, while reducing binding of HiNF-D to Site II (Figure 7). Substitution of two guanineresidues (mutant IGM-18) representing methylationinterference contacts for both HiNF-M and HiNF-P(see Figure 1) abolishes HiNF-P and reduces bindingof HiNF-M and HiNF-D to Site II (Figure 7).

Comparison of reporter gene expression with theseconstructs relative to the wild type H4 promoter revealsthat the HiNF-M mutant MSP-16 construct displays 2-to 3-fold reduced promoter activity (Figure 8). Thequantitative resemblance of this mutational effect withthe 3-fold reduction observed with the other two HiNF-M mutant constructs INS-10/CAT and FAM-15/CAT,but in the absence of the potentially confounding para-meter of binding site spacing, indicates that the HiNF-M recognition element is essential for high level H4transcription.

Mutant MPM-17 expresses the CAT reporter geneat a more reduced level than MSP-16 (6 to 8 foldreduction) (Figure 8), similar to mutant MC-7 (see Fig-

Figure 8. Functional analysis of mutant H4 gene promoters bytransfection of H4/CAT reporter gene constructs in three differentcell types. CAT activity of each of the mutant H4 promoters (listedbelow the diagram) was normalized relative to the wild type con-struct (see Figure 4 for explanation of symbols). Panel B shows arepresentative experiment.

ure 4). Thus, while mutation of the HiNF-P or HiNF-Dbinding sites each by itself do not affect H4 promoteractivity (mutants GT-9 and SUB-11; see Figure 4), itappears that these factors play important auxiliary rolesin determining H4 gene transcription.

Mutations in the IGM-18/CAT construct decreasereporter gene expression approximately 3 fold relat-ive to the wild type H4 promoter, but expression withIGM-18/CAT remains significantly higher than thatobserved for MPM-17/CAT. As both of these con-structs do not bind HiNF-P and have similar reducedaffinity for HiNF-D (Figure 7), these mutants differonly in the relative strength of HiNF-M binding toSite II. Thus, it appears that quantitative differences

10

Figure 9. Analysis of mutant H4 gene promoters by co-transfectionexperiments with IRF-2 (panel A) and IRF-1 (panel B) expressionvectors. CAT activity (indicated on the vertical axis of each reportergene construct (listed below the graph) was evaluated in the presenceof different amounts of pCMV-IRF2 and pCMV-IRF1 expressionvectors (as indicated horizontally). CAT activity was normalizedrelative to �-gal activity and is expressed as fold induction relativeto activity observed in the absence of IRF’s (i.e., 0 �g expressionconstruct). The graph is based on a representative experiment withtriplicate samples for each titration point. Experimental variation foreach set of samples is below 25%.

in occupancy of Site II by IRF-2 (HiNF-M) in vitromay reflect corresponding quantitative differences inH4 promoter activity in vivo.

To corroborate these findings, we evaluated theextent to which modulation of the level of IRF-2

Figure 10. Analysis of SUB-11 (HiNF-D) and GT-9 (HiNF-P)mutant H4 promoters in DKO cells which do not contain IRF-2 andIRF-1. The graph shows CAT activity results of two independentexperiments with triplicate samples using either the SUB-11/CAT orGT-9/CAT mutant promoter constructs. The error-bars reflect stand-ard deviations for each set of samples.

(HiNF-M) influences transcriptional activity of MPM-17/CAT and IGM-18/CAT. We titrated the levels ofthis factor in vivo by co-expressing different amountsof IRF-2 and IRF-1 expression vectors with a fixedamount of each reporter gene plasmid (Figure 9).The results show that the transcription level of IGM-18/CAT, but not MPM-17, remains responsive to IRF-2but not IRF-1 as reflected by a significant enhancementof promoter activity upon increasing the amount ofIRF-2 expression plasmid. Thus, elevating the cellularlevels of IRF-2 compensates at least in part for reducedaffinity of IRF-2 (HiNF-M) for the IGM-18/CAT con-struct caused by mutation of the M-box. The observa-tion that H4 promoter activity is influenced both by amutation (IGM-18) that reduces the affinity of IRF-2(HiNF-M) (Figure 8) and by modulating the level ofthis factor (Figure 9), is consistent with the concept thatIRF-2 (HiNF-M) is an important rate-limiting factorfor Site II mediated H4 gene transcription.

Mutation of either the HiNF-D or HiNF-P motifsreduces H4 promoter activity in cells nullizygous forIRF-2 and IRF-1

Mutation of the HiNF-D or HiNF-P elements by itselfdoes not result in reduction of H4 promoter activity, butfunctional involvement of both factors is is suggestedby mutants in which the HiNF-D and HiNF-P motifsare mutated ‘in cis’ in conjunction with the HiNF-Mbinding site (Figs. 4 and 8). To corroborate these find-ings, we analyzed the effect of mutating the HiNF-D or

11

HiNF-P binding sites in double knock out (DKO) cells,which are nullizygous for IRF-2 and IRF-1. Theseexperiments eliminate the contributions of HiNF-M"in trans", and evaluate the extent to which HiNF-D orHiNF-P becomes rate-limiting when HiNF-M is absentin the cell. The data show that both the SUB-11 (HiNF-D) and GT-9 (HiNF-P) mutant promoters each displaya 2 to 4 fold reduction in H4 promoter activity whenanalyzed in DKO cells (Figure 10). Hence, HiNF-Dand HiNF-P each contribute to histone H4 gene tran-scription when the HiNF-M binding site is mutated orwhen this factor is absent in the cell.

Discussion

This study was aimed at dissecting the complexmolecular organization of the histone H4 gene pro-moter Site II sequences which are instrumental inmediating proliferation-specific transcriptional com-petency [26–28) and encode crucial information for H4gene cell cycle regulation [26]. Our results unequivoc-ally establish that the multiple overlapping recognitionsequences for cognate transcription factors in Site IItogether modulate H4 gene transcription levels by atleast an order of magnitude. We have also shown thatthe recognition sequence of IRF-2 (HiNF-M), whichcoincides with the H4 cell cycle regulatory element[26], is the dominant component and modulates H4gene transcription levels by 2 to 3 fold. Our data sug-gest that IRF-2 (HiNF-M) performs a dual functionby determining both basal levels and the cell cycleenhanced transcription at the G1/S transition observedin a previous study [26]. Thus, it appears that IRF-2(HiNF-M) is analogous to OCT-1 (OTF-1) which per-forms a similar dual function in cell cycle regulationof human histone H2B transcription [42].

Our results also represent the first direct indica-tion that natural sequence variation in the 50 regions ofdistinct H4 genes has functional consequences for theactivity of human H4 promoters. This observation isrelevant to the overall expression of histone H4 genes,because histone H4 proteins are translated from mul-tiple mRNAs encoding very similar or identical pro-teins [1, 23–25], and functional H4 mRNAs are tran-scribed from distinct H4 genes displaying considerablevariation in the organization of 50 flanking sequences.This promoter heterogeneity is reflected by differencesin the presence, number of copies, relative location,spacing, and orientation of consensus transcriptionalelements. For example, the Site II region of the H4.A

gene [38] and the H4 (FO108) gene (used in this study)differ at several key nucleotides resulting in absence ofbinding sites for IRF-2 (HiNF-M) and CDP/cut (HiNF-D) in the H4.A gene. Direct incorporation of one setof nucleotide variations (mutant INS-10, see Figure 4)in the promoter of the H4 (FO108) gene results in a3-fold decrease in the level of transcription. The pos-sibility arises that different H4 promoter organizationsmay have evolved to accomodate developmental andhomeostatic responsiveness to a broad spectrum of sig-nalling pathways that mediate competency for prolif-eration and cell cycle progression.

Our mutational analyses also show that the recogni-tion elements for H4TF-2 (HiNF-P), CDP/cut (HiNF-D) or both, are only detectable as rate-limiting forbasal transcription in the absence of a functional IRF-2(HiNF-M) binding site. Consistent with this finding,mutation of sites for either the H4TF-2 (HiNF-P) orCDP/cut (HiNF-D) binding site strongly reduces H4promoter activity when assayed in cells nullizygousfor both IRF-1 and IRF-2. It is possible that each ofthese factors operates via a simple independent andperhaps mutually exclusive transcriptional mechan-ism at Site II, with each providing only a quantitat-ively modest contribution (10% to 30%) to the over-all level of H4 gene transcription. However, equallyimportant is the possibility that Site II proteins func-tion in concert and that the relative contributions ofthese transcriptional activities is strongly cell cyclestage or cell type dependent. Discrimination betweenthese and other refinements of models for histone H4gene transcription requires further experimentation.However, our current results clearly suggest that thefundamental basis of H4 gene regulation by Site IIcannot be attributed to the action of one particular his-tone subtype-specific factor at a singular element, ashas been proposed for several other human histonegenes [21, 38, 42]. Rather, the integrated activitiesof multiple transcription factors at a composite regu-latory domain together elevate H4 gene transcriptionin different proliferating cells, perhaps in response todistinct cell growth- and tissue-related signals.

Acknowledgments

These studies were supported by NIH grants(GM32010 and AR39588). We thank Mark Birnbaumand Thomas Last for many stimulating discussions, aswell as Shirwin Pockwinse for assistance with photo-

12

graphy. We also thank Rosa Mastrorotaro, Jack Greenand Elizabeth Buffone for excellent technical support.

References

1. Stein GS, Stein JL & Marzluff, WF (eds) (1984), HistoneGenes, John Wiley & Sons, New York

2. Osley MA (1991) Annu. Rev. Biochem. 60: 827–8613. Wolfe SA & Grimes SR (1993) J. Cell Biochem. 53: 156–1604. Doenecke D, Albig W, Bouterfa H & Drabent B (1994) J. Cell.

Biochem. 54: 423–4315. Bouterfa HL, Piedrafita, FJ, Doenecke D & Pfahl M (1995)

DNA Cell Biol. 14: 909–9196. Eilers A, Bouterfa H, Triebe S & Doenecke D (1994) Eur. J.

Biochem. 223: 567–5747. Wolfe SA, van Wert JM & Grimes, SR (1995) Biochem. 34:

12461–124698. Duncliffe KN, Rondahl ME & Wells JR (1995) Gene 163:

227–2329. Bauer–Hofmann R & Alonso A (1995) Nucleic Acids Res. 23:

5034–504010. Sun JM, Ferraiuolo R & Davie JR (1996) Chromosoma 104:

504–51011. Sun JM, Penner CG & Davie JR (1993) FEBS Lett. 331: 141–

14412. Kaludov NK, Pabon–Pena L & Hurt MM (1996) Nucleic Acids

Res. 24: 523–53113. Bowman TL & Hurt MM (1995) Nucleic Acids Res. 23: 3083–

309214. Ivanova VS, Hatch CL & Bonner WM (1994) J. Biol. Chem.

269: 24189–2419415. Hatch CL & Bonner WM (1995) DNA Cell Biol. 14: 257–26616. Naeve GS, Zhou Y & Lee AS (1995) Nucleic Acids Res. 23:

475–48417. Takami Y & Nakayama T (1995) Biochim. Biophys. Acta

1264: 29–3418. Oswald F, Dobner T & Lipp M (1996) Mol. Cell Biol. 16:

1889–189519. el–Hodiri HM & Perry M (1995) Mol. Cell Biol. 15: 3587–

359620. Hinkley C & Perry M (1992) Mol. Cell Biol. 12: 4400–441121. Martinelli R & Heintz N (1994) Mol. Cell Biol. 14: 8322–833222. Stein GS, Stein JL, van Wijnen AJ & Lian JB (1994) J. Cell

Biochem. 54: 393–40423. Albig W, Kardalinou E, Drabent B, Zimmer A & Doenecke D

(1991) Genomics 10: 940–948

24. Drabent B, Kardalinou E, Bode C & Doenecke D (1995) DNACell Biol. 14: 591–597

25. Plumb M, Stein J & Stein G (1983) Nucleic Acids Res. 11:2391–2410

26. Ramsey–Ewing AL, van Wijnen AJ, Stein GS & Stein JL(1994) Proc. Natl. Acad. Sci. USA 91: 4475–4479

27. Kroeger P, Stewart C, Schaap T, van Wijnen A, Hirshman J,Helms S, Stein G & Stein J (1987) Proc. Natl. Acad. Sci. USA84: 3982–3986

28. Kroeger PE, van Wijnen AJ, Pauli U, Wright KL, Stein GS &Stein JL (1994) J. Cell Biochem. 57: 191–207

29. Wright KL, Birnbaum MJ, van Wijnen AJ, Stein GS & SteinJL (1995) J. Cell Biochem. 58: 372–379

30. Birnbaum MJ, Wright KL, van Wijnen AJ, Ramsey-Ewing AL,Bourke MT, Last TJ, Aziz F, Frenkel B, Rao BR, Aronin N,Stein GS & Stein JL (1994) Biochem. 34: 7648–7658

31. Pauli U, Chrysogelos S, Stein G, Stein J & Nick H (1987)Science 236: 1308–1311

32. van Wijnen AJ, Ramsey–Ewing AL, Bortell R, Owen TA, LianJB, Stein JL & Stein GS (1991) J. Cell Biochem. 46: 174–189

33. van Wijnen AJ, van den Ent FMI, Lian JB, Stein JL & SteinGS (1992) Mol. Cell Biol. 12: 3273–3287

34. Shakoori R, van Wijnen AJ, Cooper C, Aziz F, Birnbaum M,Reddy GPV, DeLuca A, Grana X, Giordano A, Lian JB, SteinJL, Quesenberry P & Stein GS (1995) J. Cell Biochem. 59:291–302

35. van den Ent FMI, van Wijnen AJ, Lian JB, Stein JL & SteinGS (1993) Cancer Res. 53: 2399–2409

36. Vaughan PS, Aziz F, van Wijnen AJ, Wu S, Harada H, Tanigu-chi T, Soprano KJ, Stein JL & Stein GS (1995) Nature 377:362–365

37. Tanaka N, Kawakami T & Taniguchi T (1993) Mol. Cell Biol.13: 4531–4538

38. Dailey L, Roberts SB & Heintz N (1988) Genes Dev. 2: 1700–1712

39. van Wijnen AJ, Aziz F, Grana X, DeLuca A, Desai RK, Jaars-veld K, Last TJ, Soprano K, Giordano A, Lian JB, Stein JL &Stein GS (1994) Proc. Natl. Acad. Sci. USA 91: 12882–12886

40. van Wijnen AJ, van Gurp MF, de Ridder M, Tufarelli C, LastTJ, Birnbaum M, Vaughan PS, Giordano A, Krek W, NeufeldEJ, Stein JL & Stein GS (1996) Proc. Natl. Acad. Sci. USA93: 11516–11521

41. Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JG,Smith JA, Struhl K (eds)(1987): Current Protocols in Molecu-lar Biology New York: John Wiley & Sons, Inc

42. LaBella F, Sive HL, Roeder HG & Heintz N (1988) Genes Dev.2: 32–39