Tissue-specific Regulation of the Mouse αA-crystallin Gene in Lens via Recruitment of Pax6 and c-Maf to its Promoter

doi:10.1016/j.jmb.2005.05.072 J. Mol. Biol. (2005) 351, 453–469

Tissue-specific Regulation of the Mouse aA-crystallinGene in Lens via Recruitment of Pax6 and c-Maf toits Promoter

Ying Yang and Ales Cvekl*

The Departments ofOphthalmology and VisualSciences and MolecularGenetics, Albert EinsteinCollege of Medicine, BronxNY 10461, USA

0022-2836/$ - see front matter q 2005 E

Abbreviations used: qChIP, quanimmunoprecipitation; MARE, MafqRT-PCR, quantitative reverse transchain reaction; EMSA, electrophoreassay; PD, paired domain; HD, homE-mail address of the correspond

[email protected]

Pax6 is a lineage-restricted DNA-binding transcription factor regulatingthe formation of mammalian organs including brain, eye and pancreas.Pax6 plays key roles during the initial formation of lens lineage,proliferation of lens progenitor and precursor cells and their terminaldifferentiation. In addition to Pax6, lens fiber cell differentiation isregulated by c-Maf, Prox1 and Sox1. Crystallins are essential lens structuralproteins required for light refraction and transparency. Mouse aA-crystallin represents about 17% of all crystallins at the protein level andranks as one of the most abundant tissue-specific proteins. Lens-specificexpression of this gene is regulated at the level of transcription. A promoterfragment of K88 to C46 is capable of driving lens-specific expression intransgenic mouse. Here we provide data suggesting that this lens-specificpromoter fragment is comprised of multiple Pax6 and Maf-binding sites.Site-directed mutagenesis of regions within these sites resulted in partiallyor completely reduced promoter activities in lens cells. Co-transfectionsusing Pax6 and c-Maf alone revealed moderate and strong activations ofthis promoter, respectively. In contrast to synergistic activation of aB-crystallin by Pax6 and c-Maf, Pax6 has a neutral effect on c-Maf-mediatedaA-crystallin promoter activation. Chromatin immunoprecipitationsestablished in vivo interactions of Pax6 and c-Maf with the aA-crystallinpromoter in lens cells. Collectively, the present data support a molecularmodel in which tissue-specific expression of aA-crystallin is regulated byrecruitment of Pax6 and c-Maf, two proteins regulating multiple processesof lens differentiation, to its promoter. In addition, the data suggest amolecular model of temporal and spatial regulation of aB, aA andg-crystallin genes in mouse embryonic lens by using variants of the Pax6/Maf regulatory module.

q 2005 Elsevier Ltd. All rights reserved.

Keywords: lens; crystallins; differentiation; Pax6; c-Maf
*Corresponding author
Introduction

Pax6 is a lineage-restricted specific DNA-bindingtranscription factor regulating the formation ofmammalian organs including brain, eye andpancreas.1,2 It has a particularly critical role in thedevelopment of visual systems from cells of

lsevier Ltd. All rights reserve

titative chromatinresponsive element;criptase-polymerasetic mobility shifteodomain.ing author:

ectodermal, neuroectodermal, and mesenchymalorigin.3–5 Despite the central roles of Pax6 in theseprocesses, only about 20 tissue-specific and tissue-restricted genes have currently been shown asdirectly regulated by this factor in brain, cornea,lens, retina, and pancreas,1,6 It has been proposedthat Pax6 regulates expression of its direct targetgenes in concert with other lineage-restrictedtranscription factors including members of thelarge Maf family (MafA, MafB, c-Maf and NRL),Pax2, Prox1, Six3, Sox1 and Sox2.6–8 Expression ofcrystallin genes in lens provides an excellentopportunity to study the molecular mechanism ofgene activation and repression by Pax6.Lens lineage originates from a population of cells

d.

454 Regulation of aA-crystallin Gene Expression

comprising the head surface ectoderm by aconcerted action of transcription factors includingPax6, Otx2, and Sox2, and signaling pathways suchas BMPs, FGFs, and retinoic acid.9,10 An area of thesurface ectoderm in contact with the underlyingoptic vesicle thickens to form the lens placode andinitiates expression of additional lens lineage-restricted factors including Six3, Pitx3, FoxE3, andlater Prox1. In the mouse, aB-crystallin is the firstand only crystallin expressed in the lens placode.11,12

As the lens placode buckles inward, it produces arecognizable depression in the exterior surfacemarked by expression of c-Maf, and is followedby the onset of aA-crystallin expression.11 Theresulting structure, a spherical lens vesicle, ispolarized. Its anterior cells retain their epithelialmorphology whereas the posterior cells terminallydifferentiate, forming the primary lens fibers. Thisprocess is controlled by growth factors includingFGF2 originating from the prospective retina, inconjunction with Prox1’s critical nuclear role toinduce cell cycle arrest.13 Lens fiber cell differen-tiation is characterized by cellular elongation and adramatic increase in aA and aB-crystallin synthesisas well as other crystallins comprising the bg-crystallin subfamily.6,9,10 In parallel, expression ofPax6 is reduced but not abolished in differentiatinglens fiber cells and high level of c-Maf expression ismaintained throughout the embryonic stages oflens differentiation and growth.14,15 Since aA-crystallin is the most abundant crystallin, represent-ing about 17% of all crystallins16 in the newbornmouse lens, understanding the molecular basis ofits temporally and spatially regulated expression inthe lens is an important issue of cellular differen-tiation in general.

Lens-specific expression of the aA-crystallingene is regulated at the level of transcription.6

A promoter fragment of K88 to C46 is sufficient tosupport lens-specific expression of the chloramphe-nicol acetyl transferase (CAT) reporter gene.17

Transcription factors regulating in vivo expressionof this gene emerged from gene targeting studies oflens lineage-specifying genes described above.Inactivation of c-Maf,18,19 but not of Prox113 andSox2,20 resulted in a significant reduction of bothaA and aB-crystallin expression in c-Maf nulllenses. Although defective fiber cell differentiationis a major phenotype in c-Maf homozygous mouse,c-Maf plays important roles in organs not expres-sing the aA-crystallin.21–24 Conditional inactivationof Pax6 from the lens placode completely abrogatedlens development with no aA-crystallinexpression.25 A c-Maf binding site (MARE) inter-acting with recombinant c-Maf was shown betweennucleotides K110 and K98 of the extendedpromoter18 while a Pax6-binding site (K54/K35)was identified earlier.26 A DNaseI footprintingstudy of the aA-crystallin promoter revealedbinding of lens nuclear proteins to sites surround-ing this Pax6-binding site and an adjacent TATA-region;27 however, interactions with lens-lineagerestricted transcriptions were not further pursued.

Binding of ubiquitously expressed AP-1 factor c-Junfrom mouse lens epithelial cells to the downstreamregion (C27/C31) was also shown earlier.28 Here,we tested the hypothesis that a lens-specificpromoter fragment K88 to C46 of the mouse aA-crystallin promoter harbors novel binding sites forlens-lineage specific transcription factors since asingle Pax6-binding site described above is insuffi-cient to elicit lens-specific expression driven by theK88 to C46 promoter fragment.17,26,29 We indeedfound that this region contains three novel largeMAREs. There are four large Maf proteins; MafA,MafB, c-Maf, and NRL, all of which are expressed inthe lens.6,9 We show that each large Maf canactivate the mouse aA-crystallin promoter intransfected cells. We have previously showntranscriptional synergism between large Mafs andPax6 proteins in the mouse aB-crystallinpromoter30,31 that contrasted with transcriptionalco-repression by Pax6 and Six3 of the Maf/Sox-mediated activation of the gF-crystallin promoter.31

Since aA is expressed later than aB-crystallin butprior to the onset of g-crystallins, and aA is themostabundant of the 16 crystallins in mouse and humanlens, functional interactions between Pax6 andMafson this gene were probed. We also present evidencefor three Pax6-binding sites in the mouse aA-crystallin promoter. Using chromatin immuno-precipitations (ChIPs) we show for the first timethat the aA-crystallin promoter binds both Pax6 andc-Maf in vivo. Finally, the present data led us topropose a molecular model that explains sequentialactivation of aB, aA and gF-crystallins in thedeveloping mouse lens.

Results

Prediction of novel Maf and Pax6-binding sitesin the mouse aA-crystallin promoter

Tissue-specific gene regulation is based on acombinatorial deployment of multiple DNA-bind-ing transcription factors with lineage-restrictedexpression.32 To identify putative transcriptionfactor-binding sites residing in the lens-specificpromoter region K88 to C46, we first usedalgorithms to predict these sites as described inMaterials and Methods followed by a focusedsearch for possible binding sites of lens lineage-specific factors including Pax6, Pitx3, Prox1, Six3,Sox1, and Sox2.6 From the first search, generalhomeodomain-binding sites, Ets/GATA and Smad-binding sites emerged. EMSAs were performed toassess whether proteins from these families couldinteract in vitrowith the predicted sites and point tothe possible mechanism of lens-specificity of theK88 toC46 fragment shown in transgenic mice;17,29

however, no positive data were obtained (data notshown). Next, using a consensus binding sequence50-TGMTGANYNNGCA-30 for T-MARE33 and usinga consensus sequence 5 0-TGCTGANYCNG-3 0

obtained from an alignment of 13 well-established

Regulation of aA-crystallin Gene Expression 455

large Maf binding sites34 we found three novelputative Maf-binding sites as shown in Figure 1.These sites are calledMARE (K79), MARE(K6) andMARE(C28). Since c-Maf serves as an importantfactor regulating expression of mouse aA-crystallinin vivo,18,35 but its only known binding site (K110 toK98, DE1/CRE/MARE)18 is not essential forestablishing lens-specific expression,17,29 predictionof additional Maf-binding sites in the K88 to C46promoter fragment was considered significant forits transcriptional regulation. In addition, thisregion may contain two novel Pax6-binding sites,K71 to K51 (site B) and C15 to C34 (site C), asindicated by the sequence conservation betweenthese sequences and the optimal binding sites forthe Pax6 paired domain, P6CON, 5 0-ANNTTCACGCWTSANTKMNY-3 0 (Figure 1).36

Large Mafs activate the mouse aA-crystallinpromoter

To initially test our hypothesis, we assessed theactivity of the K111 to C46 promoter region andloss of the 5 0 and 3 0-MAREs, i.e. DE1/CRE/MAREand MARE(C28). Four fragments, K111 to C46,K88 to C46, K111 to C24, and K88 to C24, thatdrive the expression of luciferase in cultured lens

Figure 1. Diagrammatic representation of large Maf and(a) The mouse promoter region is shown from positions K11and Pax6-binding site A are shown in bold. Novel T-MARE(according to the nucleotide (underlined) in the center of the 1alignment with the consensus binding site for Pax6, ANNTTC(B and C, boxed). KZG/T; MZA/C; NZA/C/G/T; SZCcrystallin T-MAREwith a 13 (left) and an 11 -base-pair (right) cB and C with their consensus binding sites. Mismatched nuc

cells were tested as shown in Figure 2(a). The results(Figure 2(b)) revealed reduced activities of trun-cated promoters. Even the shortest fragment, K88to C26, with two predicted MAREs and Pax6-binding sites had still appreciable promoter activity.Western immunoblotting was performed to evalu-ate expression of Pax6, c-Maf, MafA and MafB inlens cultured cells, mouse aTN4-1 and rabbitN/N1003A, and in 293Tcells used for transfections.Expression of these proteins was detected in allinstances except for no expression of Pax6 in 293Tcells. Thus, both cultured lens cell epithelial lines,aTN4-1 and N/N1003A, express all important lenslineage specific transcription factors Pax6, MafA,MafB and c-Maf.To determine whether the wild-typeK111 toC46

promoter and its truncated fragments could beactivated by large Mafs, we performed a series oftransient cotransfections in 293T cells using indi-vidual expression plasmids encoding MafA, MafB,c-Maf, and NRL. The wild-type promoter wasstrongly activated by each Maf in the order ofc-MafOMafBOMafAONRL (see Figure 3).Deletion of DE1/CRE/MARE decreased largeMaf’s co-activations to 18, 3, 3 and 11% of normalactivity of MafA, MafB, c-Maf and NRL, respect-ively. In contrast, deletion of a region containing

Pax6-binding sites in the mouse aA-crystallin promoter.1 to C46. The previously established DE1/CRE/T-MAREK79), T-MARE(K6) and T-MARE(C28) were numbered3 bp consensus binding, TGMTGANYNNGCA. A similarACGCWTSANTKMNY, predicts two novel binding sites/G; WZA/T and YZC/T. (b) Alignments of each aA-onsensusMAREs. (c) Alignments of Pax6-binding sites A,leotides are shown in lower case letters.

Figure 2. Promoter activity of the mouse aA-crystallinfragment K111/C46 and its truncations in lens cells.(a) A diagrammatic representation of the wild-type andtruncated promoters cloned in the pGL2 vector with thefirefly luciferase gene. (b) Relative promoter activities ofK111/C46, K111/C26, K88/C46, and K88/C26promoter fragments in cultured lens cells normalizedagainst the activity of the K111/C46 promoter. Thetransfections were performed in hexaplets andnormalized using Renilla luciferase as an internal control.The results are shown as means with standard deviationsindicated by error bars. (c) Results of Western immuno-bloting of Pax6, c-Maf, MafA and MafB in lens cultured(aTN4-1 and N/N1003A) and 293T cells. An immunoblotwith b-actin is shown as a loading control.


downstream MARE(C28) had the least dramaticimpact with 37, 32, 74, and 33% of normal activitiesof co-tranfected MafA, MafB, c-Maf and NRL,respectively. The promoter fragment lacking bothDE1/CRE/MARE and MARE(C28), K88 to C26,was also activated by each largeMaf (see Figure 3(a)).Expression of ectopic Mafs in transfected 293T cellswas assessed by Western blots and M2 anti-flagantibody (Figure 3(b)). From these data, we con-cluded that the mouse aA-crystallin promoter

fragment, K111 to C46, possesses multiple MAREsand that c-Maf was the strongest activator of aA-crystallin promoter activity in these transfectionassays.

Identification of Maf and Pax6-binding sitesin vitro in the mouse aA-crystallin promoter

EMSAs and DNase I footprinting were used toexamine in vitro protein–DNA interactions of themouse aA-crystallin promoter. The lens nuclearproteins were incubated with a series of four probesK113/K91, K92/K69, K15/C20, and C18/C46harboring putative MAREs (see Figure 1) followedby gel electrophoresis. Each probe formed a broadband of overlapping specific complexes (Figure 4,lanes 2, 7, 13 and 19) that were abolished or reducedin the presence of oligonucleotides containingconsensus T-MARE (lanes 4, 10, 16 and 22) butless in the presence of mutated T-MARE sequences(lanes 5, 11, 17 and 23). The data suggested that eachprobe could form specific complexes correspondingto activities of Maf proteins as shown directly by asimilar probe K113/K91 elsewhere.18 Results ofspecific cross-competitions of this sequence withprobes K92/K69, K15/C20 and C18/C46 areshown in Figure 4, lanes 9, 15 and 21, respectively.

To identify the regions of the mouse aA-crystallinpromoter interacting with Pax6 proteins, we usedDNaseI footprinting and radioactively end-labeledpromoter fragments K166 to C46. We assessed theinteraction of both variants of Pax6: Pax6 containingthe 128 amino acid residue paired domain (PD), andPax6 (5a) containing the alternatively spliced 142amino acid residue extended paired domain(PD5a). The comparison of protected areas on bothlower (Figure 5(a)) and upper strand (Figure 5(b))using proteins containing only Pax6 PD (lanes 4 and5) or both the PD and the homeodomain (HD)confirmed that the HD restricts DNA-bindingproperties of both the PD and the PD(5a), inagreement with earlier studies.6 Strong binding ofPax6 proteins with both the PD and the HD(Figure 5(a), lanes 11 and 12, and Figure 5(b),lanes 11 and 12) overlapped the positions depictedin Figure 1. From these data we concluded that theaA-crystallin promoter possesses three Pax6-bind-ing sites, labeled Pax6-site A, B and C. In addition,binding of Pax6 was found in the region K138 toK110, labeled D (see Figure 5(a) and (b), lanes 11and 12).

To assess if Pax6 proteins present in lens nuclearextracts can bind to sites A, B and C, we performeda series of EMSAs. The Pax6-site A oligonucleotide(K59/K29) was incubated with rabbit and mouselens nuclear extracts. Three major specific andseveral minor complexes were detected (Figure 6(a),lanes 2 to 4). Two of them were reduced in thepresence of Pax6-specific antibody recognizing itsDNA-binding PD (compare lanes 4 and 5). Next, weexamined if an addition of an excess of anoligonucleotide containing “optimal” Pax6-bindingsite, P6CON, could prevent formation of these

Figure 3. Transcriptional activation of four mouse aA-crystallin promoter fragments by individual c-Maf, MafB, MafAand NRL. (a) Activities of a series of aA-crystallin promoter fragments in the presence of 200 ng of each individualcDNA encoding c-Maf, MafB, MafA and NRL in transiently co-transfected 293Tcells. The experiments were performedand evaluated as described in the legend to Figure 2(b). The “basal” normalized activity of K111/C46, K111/C26,K88/C46 and K88/C26 promoter fragments compared to the wild-type fragment K111/C46 in 293T cells were 1.0,1.43, 1.5 and 0.76, respectively. Activation potentials of individual Mafs are expressed as ratios in the presence of Mafdivided by the activity in the presence of identical amounts of empty vector, pKW10. (b) Western immunoblotting todemonstrate expression of ectopic MafA, MafB, c-Maf and NRL proteins in 293T cells. 293T cells were transfected with80 ng of expression plasmid and 20 and 40 mg of whole cell extracts were loaded in adjacent lanes. Expression of Pax6/Pax6 (5a) is also shown as this experiment relates to data shown in Figure 9. Expression of b-actin was used as a loadingcontrol.


complexes. Indeed, two complexes inhibited by theanti-Pax6 antibody were also significantlyabolished in the presence of P6CON (Figure 6(b),lane 8) but not in the presence of oligonucleotideswith PU.1 (lane 9) and GATA3 (lane 10) bindingsequences. We next incubated an oligonucleotideK88 to K56 containing Pax6-binding site B withmouse lens nuclear extracts and identified twospecific complexes (Figure 6(c), lane 2), which wereabolished in the presence of P6CON but not in the

presence of oligonucletides with GATA3 and NF-kBbinding sites (Figure 6(c), compare lanes 2 and 7).Note that the faster migrating Pax6-containingcomplex was more efficiently inhibited comparedto the slower Pax6-complex, likely containingadditional protein(s). Finally, we incubated anoligonucleotide probe C18/C46 with lens nuclearproteins; however this probe formed a number ofslow and closely migrating complexes that were notcompeted in the presence of P6CON (data not

Figure 4. In vitro DNA-protein interactions. EMSAs with four probes derived from the mouse aA-crystallin promoter(see Figure 1) incubated with lens nuclear proteins. (a) Interaction of lens nuclear proteins (aNE) with oligonucleotideprobeK113/K91, (b) probeK92/K69 (lanes 6 to 11), (c) probeK15/C20 (lanes 12 to 17) and (d) probeC18/C46 (lanes18 to 23). Specific oligonucleotide competitors were included in lanes 3K5, 8K11, 14K17, and 20K23 as indicated.Specific complexes containing Maf-like activities and other specific complexes are labeled by filled and open arrows,respectively.


shown). It also bound recombinant Pax6 proteins inEMSAs (data not shown). Earlier, binding of lensnuclear proteins to a region of C24 to C43 wasshown by DNase I footprinting.27 Thus, binding ofPax6 to site C was not conclusively shown.Together, these data show that the mouse aA-crystallin promoter contains at least two Pax6-binding sites A and B.

Site-directed mutagenesis of the mouseaA-crystallin promoter fragment K111 to C46

To demonstrate the functional roles of sequencesincluding MAREs and Pax6-binding sites of the aA-crystallin promoter, we performed site-directedmutagenesis of the promoter and tested corre-sponding promoter activities in cultured lens cells.

Figure 5. Interactions of recombinant Pax6 proteins with the mouse aA-crystallin promoter fragment using DNase Ifootprinting. Recombinant Pax6 PD, PD5a, PDCHD, and PD5aCHD proteins were incubated with K166 to C46 end-labeled probes and digested in the presence of DNase I. (a) Binding to the lower strand. (b) Binding to the upper strand.Areas of interaction for Pax6 PD (filled boxes), and Pax6 PD5a (shaded boxes). Four binding sites of Pax6 PDCHDproteins (brackets) were labeled A, B, C, and D (see Figure 1 for additional information).


Twelve mutants are shown in Figure 7(a) and theresults of transfections are given in Figure 7(b).Promoter activities were reduced but not abolishedfor all 12 tested mutations, which overlapped withthe binding sites of largeMaf and Pax6 proteins (seeFigure 7(a)). Note that mutation M12 affected bothMARE and Pax6-binding sequences. To testwhether these mutations indeed influencedbinding of c-Maf and Pax6, we tested mutationsM2 and M3, in MARE(K79) and M4 to M7, in Pax6-binding site B. Mutations M2 and M6 moststrongly reduced promoter activity. EMSAs showedthat M2 could not reduce formation of Maf-likecomplexes (Figure 7(c), compare lanes 2 and 5).Similarly, M6 could not inhibit formation of Pax6-containing complexes using Pax6-binding site B(Figure 7(d), compare lanes 8 and 12). In addition,mutation M3 (lane 6) retained some ability tocompete Maf-like complex as it caused only a 40%reduction of the aA-crystallin promoter activity(Figure 7(b)).

We then hypothesized that multiple mutations

would result in the additional loss of promoteractivities. To test this, we generated five doublemutants and two triple mutants (see Figure 8(a))and determined the promoter activities (seeFigure 8(b)) as described above. We focused onthose combinations including DE1/CRE/MAREand Pax6-binding site B as we could correlate theimpact of these mutations with their previousstudies in transgenic mice.29 Two double mutations(M15 and M16) and both triple mutations (M14 andM19) eliminated the promoter activity. M15 corre-sponded to the double mutation resulting in loss ofpromoter activity tested in transgenic mice.29 Thesedouble mutations had impaired MARE and eitherPax6-site B or A. Two double mutations, M13(affecting Pax6 sites A and B) and M17 (affectingPax6 sites B and C), resulted in 8 and 18% of activityof the wild-type promoter, respectively. Amutation,M19, affected both Pax6-binding sites A and Bestablished in Figures 5 and 6, and a site C,established only by use of recombinant Pax6proteins (see Figure 5).

Figure 6. EMSAs with two probes derived from the mouse aA-crystallin promoter (see Figure 1) incubated with lensnuclear proteins. (a) An oligonucleotideK59/K29 (site A) was incubated with increasing amounts of rabbit lens nuclearextracts from N/N1003A cells (lanes 2 to 4) and in the presence of a specific anti-Pax6 antibody recognizing Pax6 PD(lane5). (b) An oligonuclotide K59/K29 (site A) was incubated with increasing amounts of mouse lens nuclear extractsfrom aTN4-1 cells (lanes 2 and 3) and in the presence of approximately 50 molar excess of indicated specificoligonucleotide competitors (lanes 4 to 10). (c) An oligonucleotide K88/K56 (site B) was incubated with mouse lensnuclear extracts from aTN4-1 cells (lanes 2) and in the presence of approximately 50 molar excess of indicated specificoligonucleotide competitors (lanes 3 to 7). Specific complexes containing Pax6 proteins and other specific complexes arelabeled by filled and open arrows, respectively.


Activation of the mouse aA-crystallin promoterby transcription factors co-expressed in devel-oping lens epithelium and in lens fibers

We have recently shown that mouse aB-crystal-lin, expressed earlier than aA-crystallin, is syner-gistically activated by Pax6 and large Mafproteins.30,31 In contrast, the gF-crystallin gene,which is activated by a combination of Maf andSox1/2 proteins, after aA-crystallin, was repressedby Pax6.31,37 Here we cotransfected the wild-typeaA-crystallin promoter with Pax6, Pax6 (5a), Maf B,and c-Maf as the onset of aA-crystallin expressionin lens pit could be triggered by Pax6 and emergingMafB and c-Maf. Pax6 proteins alone weaklyactivated the mouse aA-crystallin promoter in293T cells (see Figure 9(a)). A similar result wasreported earlier in COP-8 and C33-A cells.6,26 Incontrast, MafB and c-Maf strongly activated thispromoter in a concentration-dependent manner(Figure 9(a)). When combinations of these factorswere tested, the results showed moderate repres-sion of MafB and MafB/c-Maf-stimulated promoteractivities and a marginal increase of the promoteractivity by Pax6 proteins in combination with c-Maf(see Figure 9(b)). BothMafA andNRL also activatedthe promoter in a dose-dependent manner (seeFigure 9(c)). Similar results, i.e. weak repression ofMafA and MafA/c-Maf/NRL-mediated transcrip-tional activation of the K111 to C46 promoter by

Pax6 proteins and insignificant gain of MafA/c-Maf-stimulated promoter activity by Pax6 pro-teins were found (Figure 9(d)). Next, we testedthese combinations in cultured lens epithelial cells.Neither factor alone or in combination was able toactivate this promoter (see Figure 1, SupplementaryData) likely due to the high basal activity of thispromoter in lens cells compared to 293Tcells due tothe expression of endogenous Pax6, MafA, Maf Band c-Maf proteins (see Figure 2(c)).

Chromatin immunoprecipitations oftranscription factors interacting with the mouseaA-crystallin promoter in vivo

Interactions of Pax6, c-Maf, and CREB with theendogenous aA-crystallin promoter in lens cellswere determined using quantitative ChIPs.38–40

Binding of general transcription factor TFIID/TBPto the promoter and transcription factor Eya1expressed in lens was assessed in parallel. We alsotested modification of histones using the anti-H3 K9antibody, which recognizes acetylated lysine K9 ofhistone3. Our results (see Figure 10(a)) revealedenrichment of Pax6, c-Maf, CREB, and TFIID/TBPbut not of Eya1 in the aA-crystallin promoter regionin vivo. In contrast, these factors did not exhibitenrichments in the region located approximately4 kb downstream from the promoter (Figure 10(b))or on erythroid-specific gene ALASE (Figure 10(c)).

Figure 7. Activities of a series ofmouse aA-crystallin promotermutants in cultured lens cells.(a) Site-directed mutagenesis ofthe promoter fragment K111 toC46. Twelve block mutations arelabeled M1 to M12. Four T-MARE(shaded boxes) and three Pax6(open boxes) binding sites arelabeled. Nucleotide sequencesused for mutagenesis are shownunder the promoter sequence.(b) Relative promoter activities ofM1 toM12 compared to theK111 toC46 promoter (WT). The resultswere calculated as described in thelegend to Figure 2. (c) EMSAs usingMARE(K79) (see Figures 1(a) and4(b)) and mouse lens nuclearextract (lane 2) in the presence ofabout 50-fold excess of indicatedoligonucleotide competitors har-boring mutations M2 and M3shown in (a) (lanes 5 and 6).(d) EMSAs using Pax6 site B (seeFigures 1(a) and 6(c)) and mouselens nuclear extract (lane 8) and inthe presence of about 50-fold excessof indicated oligonucleotide com-petitors harboring mutations M4,M5, M6 and M7 shown in (a) (lanes10 to 13).


Interestingly, no enrichments of MafA, MafB, orNRL were detected in these tests (data not shown).Based on these findings, we evaluated levels ofexpression of all largeMafs, MafA, MafB, c-Maf andNRL, by qRT-PCR as they are expressed in the lenswithdistinct temporal and spatial features.14,18,19,41–43

In one-day old mouse lens, we found significantexpression of c-Maf, but not of other Mafs (data not

shown). In contrast, expression of both MafA andNRL was detected in the retina and expression ofMafB was found in the mouse cornea (data notshown). Next, we determined expression of thesegenes in the microdissected two-day old rat lensepithelium and fibers (Figure 11). As expected,14,15

the lens epithelium expressed higher levels of Pax6compared to lens fibers (Figure 11), and, conversely,

Figure 8. Activities of a series ofmouse aA-crystallin promoterdouble and triple mutants in cul-tured lens cells. (a) Multiple site-directed mutagenesis of thepromoter fragment K111 to C46.Five double and two triple blockmutations are labeled M14 to M20.DE1/CRE/T-MARE (M), two Pax6(sites A and B), and overlappingPax6/MARE (site C) binding sitesare boxed. Nucleotide sequencesused for mutagenesis are shownunder the promoter sequence.(b) Relative promoter activities ofM14 to M20 compared to the K111to C46 promoter (WT). The resultswere calculated as described in thelegend to Figure 2.


higher levels of c-Maf were found in lens fiber cellscompared to lens epithelium (Figure 11). Expressionof aA-crystallin between lens epithelium and fibersis shown in parallel. Thus, these data show in vivoassociation of the aA-crystallin promoter with Pax6,c-Maf, CREB, and TFIID/TBP and they demon-strate significant levels of expression of both Pax6and c-Maf in one-day old mouse lens but not of anyother large Mafs.

Discussion

The present study had two goals. First, to identifybinding sites for lens lineage-restricted transcrip-tion factors in the lens-specific promoter fragmentK88 to C46 of the mouse aA-crystallin. Second, toinvestigate the molecular mechanism by whichactivation of this gene occurs by focusing on in vivoprotein-DNA interactions and establishing a test-able model for sequential activation of aB, aA andg-crystallin expression in the developing embryoniclens.

Previous studies have suggested that Pax6

regulates tissue-specific gene expression via specificDNA-mediated interactions with other lineage-restricted transcription factors. In lens, cooperativeinteractions between Pax6 and Sox2 were shown totrigger expression of chicken d1-crystallin in thelens precursor cells forming the lens placode.7,44 Inaddition, Pax6 in combination with MafB andretinoic acid receptors RARb/RXRb or Pax6 withc-Maf synergistically activated the mouse aB-crystallin promoter.30,31 Similar studies of pancreas-specific expression of insulin, glucagon, andsomatostatin revealed positive effects of Pax6 oneach gene45 and cooperativity between MafA andPax6 on the glucagon promoter.46 Moreover, thehighest activation potential of MafA was foundamong other insulin regulatory factors.47 It isimportant to note that other tissue-restricted genesincluding BETA-2/NeuroD, Cdx2/3, Nkx2.2 andPdx-1 are involved in the regulation of these genesin pancreas.48 In corneal epithelium, Pax6 function-ally interacts with AP-2a to regulate expression ofgelatinase B/MMP9.49 The present data show thatthe mouse aA-crystallin promoter is comprised ofan array of multiple Pax6 and Maf-binding sites.

Figure 9. The regulation of mouse aA-crystallin promoter activity by transcription factors expressed in lens epitheliumand in lens fibers. The promoter fragmentK111 toC46 was co-transfected with cDNAs encoding Pax6/Pax6(5a), MafB,c-Maf, MafA, and NRL as indicated in 293T cells. (a) Results showing concentration-dependent activation by Pax6/Pax6(5a) (100/12.5 and 200/25 ng), MafB (80 and 200 ng) and c-Maf (80 and 200 ng). (b) Results of co-transfections oftranscription factors, Pax6, MafB and c-Maf, co-expressed in lens epithelium. (c) Results showing concentration-dependent activation by MafA (80 and 200 ng) and NRL (80 and 200 ng). (d) Results of co-transfections of transcriptionfactors, Pax6, MafA, c-Maf, and NRL, co-expressed in lens fiber cells.


The finding of both Pax6 and Maf-binding sites invirtually all crystallin promoters and/or lens-specific enhancers indicates common mechanismsto achieve both high level and lens-specificexpression. Moreover, studies of invertebratecrystallin promoters suggest similar mechanismsevolutionarily conserved although vertebrateand invertebrate crystallins are structurallyunrelated.50,51

The present data suggest that the mouse aA-crystallin promoter fragmentK111 to C46 containsfour MAREs and at least two Pax6-binding sites.

Earlier studies have identified a distal DE1/CRE/MARE and an internal Pax6-binding site A.18,26

Present data show that the distal region actuallycontains a tandem of MAREs followed by a tandemof Pax6-binding sites (see Figure 12). Mutagenesisof each novel MARE and Pax6-binding site resultedin the loss of promoter activity in transfected lenscells. Disruption of the distal promoter regionharboring a tandem of MAREs was achieved byinserting an additional Pax6 consensus-binding sitebetween two MAREs. This insertion dramaticallychanged the expression pattern of the K366/C46

Figure 10. Pax6, c-Maf, CREB and TBP proteins interactwith the mouse aA-crystallin promoter in vivo. Quanti-tative ChIPs using antibodies recognizing Pax6, c-Maf,CREB, TBP, Eya1, and H3K9Ac. (a) Primers amplified theaA-crystallin promoter region and (b) the 4 kb down-stream sequence of the promoter in formaldehydecrosslinked aTN4-1 lens cells. (c) Interactions with thepromoter region of 5-aminolevulinate synthase (ALASE),are shown as “irrelevant” gene control. Values shown arenormalized on a standard curve.38–40 Relative input unitswere calculated as described in Materials and Methods.Dotted lines represent averaged background signal in thepresence of normal rabbit IgG for each primer set/site.Average “enrichments” are shown including standarddeviations. Each experiment was performed using twoindependent chromatin preparations.


promoter fragment in transgenic mice. The expan-sion of expression from lens fibers into lensepithelium52 provided in vivo evidence that theintegrity of distal MAREs is critical for promoterfunction. Transgenic constructs bearing mutationsin either DE1/CRE/T-MARE or Pax6-binding site Bof the K111/C46 promoter fragment only margin-ally influenced lens expression of the CAT reportergene in adult lenses.29 In contrast, simultaneousinactivation of both sites abolished expression ofCAT in transgenic lenses.29 Our results are consist-ent with these data, demonstrating reduction ofpromoter activity when individual sites weremutated and a complete loss of activity of thedouble mutation M15. In addition, similarmutations (M16) affecting a different pair of Pax6and Maf-binding sites also resulted in the abolitionof promoter activity. It remains to be tested whetherother combinations of double Maf and Pax6-binding site mutations are also effective in abolish-ing the aA-crystallin promoter activities.

Site-directed mutagenesis of each Pax6-bindingsite resulted in a reduction of aA-crystallinpromoter activities, indicating that Pax6-binding iscritical for its activity in lens epithelial cells. Todemonstrate direct interactions of Pax6 and c-Mafwith the endogenous aA-crystallin loci in culturedlens cells, we performed qChIPs. The resultsshowed that the aA-crystallin promoter wasoccupied by these factors in vivo and was alsoenriched for CREB and TFIID/TBP. Acetylation oflysine 9 on histone H3 indicates open chromatin inthe promoter. We propose that expression of Pax6prior to c-Maf serves dual functions: Pax6 directlyregulates c-Maf expression in the lens,53 andbinding of Pax6 to the aA-crystallin promotermight function to open its chromatin structure.

Here we show that c-Maf is a much strongertranscriptional activator of the mouse aA-crystallinpromoter compared to MafA and MafB, and itsactivation is not inhibited by Pax6. While c-Mafactivated this promoter 256-fold (Figure 9), themouse aB-crystallin promoter was activated onlytenfold in a similar assay.31 Analysis of c-Maf nullmouse lenses at E11.5 revealed 99% reduction of theaA-crystallin transcripts, directly showing thatMafA and MafB do not compensate for the loss ofc-Maf.18 No significant expression of MafA, MafB,and NRL by qRT-PCR was shown in one-day oldmouse lens. These results were confirmed byWestern immunoblot analysis of mouse lensnuclear proteins detecting c-Maf and Pax6 but notMafA and MafB (data not shown). Taken togetherthese results show that c-Maf is the key activator ofthe mouse aA-crystallin due to its specificity,18

high-level of expression (Figure 11) and abilityto most strongly activate its promoter (Figures 3and 9).

Previous studies of the chicken aA-crystallingene suggested that MafA/L-Maf is the keyactivator of this gene, acting downstream ofPax6.35,54,55 Genetic analysis in mouse providesevidence that loss of c-Maf, resulting in the

Figure 11. Quantitative RT-PCR analysis of aA-crystallin, Pax6 and c-Maf expression in microdissected two-day oldrat lens epithelia and lens fibers. Expression levels were normalized using standard curves for each amplicon and areexpressed relative to the CCNI gene.

Figure 12. Molecular model ofdevelopmentally controlledexpression of aB, aA, and g-crystallins in mouse embryoniclens. (a) At mouse E9.5, lensplacode (left panel) is formedfrom the surface ectoderm andexpresses the aB-crystallin. AtE10.5, in the invaginating lensplacode (middle panel), aA-crys-tallin expression is triggered. FromE11.5 to 13.5, primary lens fibercells are formed from the posteriorpart of lens vesicle (right panel).The process is accompanied bydramatic upregulation of aA, b,and g-crystallins in these cells.(b) Specific organization of Mafand Pax6-binding sites in crystallinpromoters may regulate their tem-poral pattern of expression. Regu-lation of the mouse aB-crystallingene (left panel). Synergistic inter-actions (double arrows) betweenPax6, retinoic acid activatedreceptors (RA), and earlyexpressed Maf proteins (c-Mafand MafB) interacting with over-

lapping sites in lens specific regions LSR1 and LSR2. Regulation of the mouse aA-crystallin gene (middle panel).Moderate expression of this gene in lens pit and vesicle requires both Pax6 and Mafs. Pax6 might play an epigenetic rolein opening the promoter chromatin structure. As expression of c-Maf increases in the differentiating primary lens fibers,two tandems of Maf-binding sites separated by a Pax6 tandem become fully occupied, resulting in dramaticupregulation of this gene from E11.5. Binding of CREB to the promoter agrees with earlier studies26 and ChIP datashown in Figure 10. Recruitment of CBP by c-Maf to this promoter has been shown earlier.64 Regulation of the mouseg-crystallin genes (right panel). Expression of g-crystallins is repressed by simultaneous action of Pax6 and Six3 proteinbound distally (about 140 bp between Pax6-binding site and MARE) from the Maf binding site until the onset of lensfiber cell differentiation. Repression is relieved as a result of reduced amounts of both Pax6 and Six3 proteins, andupregulation of c-Maf in primary lens fibers.



dramatic loss of aA-crystallin expression in c-Mafnull lenses, cannot be compensated by other Mafgenes. There could be additional differencesbetween the moderately conserved orthologousmouse and chicken aA-crystallin promoters56 andto date only a single T-MARE (K108/K97) has beenshown in the chicken aA-crystallin promoterfragment K162 to C44.54,55 Additional studies ofthe regulation of chicken aA-crystallin will berequired to clarify these issues.

The spatiotemporal expression patterns of 16crystallin genes can originate from specific place-ments of cis-regulatory sites interacting with asparse number of lens lineage-specific transcriptionfactors.6,7,57 Three crystallin promoters, aA, aB,and gF, differently responded to Pax6 and Mafs intransfection assays and these data correlate with theonset of their expression in the embryonic mouselens (see Figure 12). Mouse aB-crystallin is firstexpressed in the lens placode due to synergisticinteractions between Pax6, Mafs, and retinoic acidnuclear receptors.31 This synergism could resultfrom overlapping Pax6 and Maf-binding sites in theLSR1 region of the aB-promoter. Expression of aA-crystallin could be delayed due to the lack ofcooperativity between Pax6 and Mafs as theirbinding sites are arranged differently from themouse aB-crystallin gene (Figure 12). Levels ofaA-crystallin expression in lens epithelium and lensfibers seem to be proportional to the levels of c-Mafexpression.18,19 Expression of gF-crystallin isevident only later in primary lens fibers58 as Pax6serves as an inhibitor of its expression mediated bya functional synergism between Maf and Soxproteins.31,59 It is also of interest that Pax6 hasbeen established as a repressor of fiber-cell specificchicken bB1-crystallin competing with Maf proteinsfor the same region of DNA.15,60 In conclusion,different promoter architecture including mutualpositions of Maf and Pax6-binding sites, theirnumber, and their distances provide a number ofmechanisms to mediate tissue-specific geneexpression in the lens, pancreas, and likely inother tissues, during mammalian development.

† http://frodo.wi.mit.edu/cgi-bin/primer3/pri-mer3_www.cgi‡www.gene-regulation.com

Materials and Methods

Materials

The following oligonucleotides were synthesized byInvitrogen (Gaithersburg, MD) and were used as probesfor EMSA: K113/K91(5 0-AGCTGCTGACGGTGCAGCCTCTC) K92/K69(5 0-TCCCCCGAGCTGAGCATAGACATT); K15/C20(5 0-GAACGCTAGCTCACCACCGCACTGCCCAGAGGCTCCTG);C18/C46(5 0-CTCCTGTCTGACTCACTGCCAGCCTTCGG). The oligonucleotidesused as specific competitors: T-MARE (5 0-GCATTTGCTGACT CAGCATTTGGT); T-MARE-mutant (5 0-GCATTTcagGACTCctgATTTGGT). Mutated nucleotides areindicated in lower case, and the consensus sequence ofT-MARE is underlined.Antibodies used for Western blots: anti-b-actin (1:2500,

Abcam, Cambridge, UK), anti-Flag M2 monoclonal

(1:2500, Sigma, St. Louis, MO); secondary antibodieswere anti-mouse HRP (1:2000, Amersham, Piscataway,NJ). Antibodies used for ChIPs: anti-Pax6, anti-c-Maf,anti-TBP, and anti-Eya1 were purchased from Santa CruzBiotechnology (St. Cruz, CA); anti-CREB-1 from Active-Motif (Carlsbad, CA), anti-H3 K9 acetylation fromUpstate (Lake Placid, NY) and normal rabbit IgG fromOncogene Research Products (San Diego, CA).Primers were designed using Primer 3†. PCR products

were limited to 60–130 bp. Primers used for ChIPs were:aA-crystallin promoter (5 0-AGTCATGTCGGGAAGACCTG and 5 0-TGTGTGCTGGATGTGGTTCT;5 0-CCCGAGCTGAGCATAGACAT and 5 0-AGTCAGACAGGAGCCTCTGG);C4 kb downstream region of aA-crys-tallin promoter (5 0-AGCAAGAAGGCAAGAGCAAAand 5 0-GTCTAAGCTCCACCCCACAG); “negative” con-trol promoter of ALASE (mouse erythroid-specific5-aminolevulinate synthase) (5 0-ATCGAATGAGTTAGTTTCACTACC and 5 0-CCTTAAGGTATCTCAGGCCTCTGC).The following oligonucleotides were used as primers

for qRT-PCR analysis: Mouse c-Maf (forward 5 0-GTGGTGGTGATGGCTCTTTT and reverse 5 0-GTTACGGGGGAA TTCAGGTT); rat c-Maf (forward 5 0-CACACACACACACAGCAAGC and reverse 5 0-TACAGGGGAATTCAGGTTGG); mouse MafB (forward 5 0-GCCTCTTAGACT TGGGCAGA and reverse 5 0-CCTTCCAGCTTGGAGAAAAG); mouse MafA (forward 5 0-GCACCCGACTTCTTTCTGTG and reverse 5 0-GCCTGCGCAAACTTGTCC); mouse NRL (forward 5 0-CTCAGTCCCAGAATGGCTTT, and reverse 5 0-GAAGGCTCCCGCTTTATTTC); mouse and rat Pax6 (forward 5 0-GCACATGCAAACACACATGA and reverse 5 0-ACTTGGACGGGAACTGACAC); mouse and rat Prox1 (5 0-GCCCTCAACATGCACTACAA and reverse 5 0-GGCATTGAAAAACTCCCGTA); mouse and rat CCNI (forward5 0-CATTCCTGATTGGCTTC CTC and reverse 5 0-GTGGATCAACTGGGAGCTGT); mouse aA-crystallin(5 0-GAGATTCACG GCAAACACAA and 5 0-ACATTGGAAGGCAGACGGTA). Specificity of each primer setwas determined using authentic cDNAs encoding bothmouse and rat Mafs.

Database search to predict transcription factorbinding sites

Transcription factor sites were predicted using on-linetools from The Transcription Factor DataBase(TRANSFAC)‡.

Plasmids, transfections and reporter assays

Mouse aA-crystallin gene promoter sequences K111 toC46, and truncated fragments K88/C46, K111/C26and K88/C26 were inserted into pGL-2-basic (Promega,Madison, WI). Single, double and triple mutations withinthe aA-crystallin K111/C46 promoter were generatedusing the QuickChange site-directed mutagenesis kit(Stratagene, LaJolla, CA). Expression plasmids encodingPax6, Pax6(5a), MafA, MafB, c-Maf and NRL were alldriven by a CMV promoter as described earlier.30

Expression vectors containing 3x flag tagged Pax6,Pax6(5a), MafA, MafB, c-Maf and NRL were generatedin p3xFLAG-CMV-10 (Sigma, St. Louis, MO). 5 mg of

http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi

http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi

http://www.gene-regulation.com


plasmid containing the aA-crystallin promoterK111/C46 in pGL2 was co-transfected with Pax6 andPax6(5a), MafA, MafB, c-Maf and NRL in differentamounts into 293T cells as described in the Figurelegends.30 To assess the activity of the K111 to C46promoter region and the effect of deletions and specificmutations, 1 mg of aA-crystallin wild-type promoter andits specific mutants were transfected into aTN4-1 cellsusing Lipofectamine Plus Kit (Invitrogen). An internalcontrol plasmid, pCMV Renilla luciferase, was included inall transfections. Firefly luciferase activities werenormalized relative to the Renilla luciferase activity.Each transfection was conducted in triplicate at least twice.

Western blot analysis

The whole cell lysates of aTN4-1, N/N1003A and 293Tcells containing 50–80 mg of total protein were loadedonto 12% (w/v) SDS-PAGE precast gels (Bio-Rad,Richmond, CA) and transferred to nitrocellulose mem-branes (Bio-Rad) as described.31 The membranes wereincubated with the primary antibodies overnight at 4 8C.

Nuclear extracts, EMSA and DNase I footprinting

Four MARE-oligonucleotides K113/K91, K92/K69,K15/C20 and C18/C46 were radioactively end-labeledand incubated with the lens nuclear extracts preparedfrom the mouse lens epithelial cell line aTN4-1 asdescribed earlier. Three oligonucleotides with Pax6-binding sites were K59/K29 (site A), K88/K56 (site B)and C18/C46 (site C). EMSAs and antibody/EMSAswith rabbit antiPax6 polyclonal antibodies61 were per-formed as described.26,62 Specific oligonucleotide com-petitors (50 ng) were used at approximately 50-fold excessover the labeled probe. DNase I footprintings wereperformed as described.27

Chromatin immunoprecipitations (ChIPs)

70% confluent aTN4-1 cells were harvested and fixedby adding fresh formaldehyde to a final concentration of1% at room temperature for ten minutes. Crosslinkingwas quenched by the addition of 2.5 M glycine solution toa final concentration of 0.125 M. Cells were lysed andsonicated to prepare chromatin of an average length of500–600 bp. Antibody (2 mg) or rabbit IgG was incubatedwith blocked protein A and G beads (Sigma) overnight at4 8C. Precleared chromatin (100 ml) was incubated withantibody bound protein A and G beads for five hours at4 8C. Beads were washed three times and treated withProteinase K at 55 8C for two to three hours followed by65 8C overnight incubation to reverse crosslinking.Genomic DNA was finally eluted into 250 ml of waterusingQIAquickSpinGelPurificationKit (Qiagen,Valencia,CA). Quantitative PCR reactions were performed asdescribed below.38–40 The standard curve for each primerset was generated using 0.05, 0.2, 0.5 and 1% input DNAsamples. The enrichments, given in relative input units,were obtained from Ct values and standard curvesexpressed relative to the 1% input. Enrichments using thecontrol IgG were calculated separately. The experimentswere performed using two independent chromatin prep-arations and eachPCR reactionwas performed in triplicate.

Quantitative RT-PCR

Total RNA was extracted from 70% confluent aTN4-1

mouse lens epithelial cells, one-day and one-week oldCD-1 mouse lenses, micro-dissected two-day old rat lensepithelium and fiber cells using TRIzol (Invitrogen). RNAwas treated with DNase I (Roche, Indianapolis, IN) atroom temperature for 30 minutes followed by phenol-chloroform extraction to avoid genomic DNA contami-nation. P0 and P4 total RNA from retina was obtainedfrom Dr M. Dorrell (The Scripps Research Institute, LaJolla, CA). P22 total RNA from retina was obtained fromDr K. Mitton (Oakland Eye Institute, Rochester, MI). P14total RNA from corneal epithelium was obtained from DrT. Kays (National Eye Institute, Bethesda, MD). cDNAswere synthesized using SuperScriptIITM RNase H-RT(Invitrogen) according to the manufacturer’s instructionsand were diluted into a final volume of 250 ml with water.Diluted cDNA (2 ml) was mixedwith 2 mMprimers and 2xSYBG Mix (ABI, Foster City, CA) in a total volume of 8 ml.Samples were analyzed in a ABI 7900HT SequenceDetection System using a three-step protocol consistingof denaturation at 95 8C for 30 seconds, annealing at 60 8Cfor 30 seconds and extension at 72 8C for 30 seconds for40 cycles. Data were analyzed using SDS2.0 (ABI) andMicrosoft Excel (Redmond, WA). Series dilutions wereused to determine primer efficiencies. Relative foldchanges were then calculated using CCNI as an internalcontrol as described.63

Acknowledgements

The authors thank Dr T. Stopka for advice onqChIPs and help with quantitative analysis, andDrs S. Saule and L. Wolf for critical reading of thismanuscript. We are grateful to Drs M. Dorrell,T. Kays and K. Mitton for retinal and corneal RNAsamples. We thank Drs M. Busslinger, J. A. Epstein,A. Sharma, A. Swaroop and K. Yoshida forexpression clones or cDNAs used in this study.We thank Dr S. Saule for anti-Pax6 antibody andA. Cvekl, Jr. for his assistance in editing themanuscript. We also acknowledge AECOM DNAsequencing core facility and use of the ABI 7900HTequipment. The work was supported from NIHgrants, EY12200 and 14237.

Supplementary Data

Supplementary data associated with thisarticle can be found, in the online version, atdoi:10.1016/j.jmb.2005.05.072

References

1. Simpson, T. I. & Price, D. J. (2002). Pax6; a pleiotropicplayer in development. Bioessays, 24, 1041–1051.

2. van Heyningen, V. & Williamson, K. A. (2002). PAX6in sensory development. Hum. Mol. Genet. 11,1161–1167.

3. Cvekl, A. & Tamm, E. R. (2004). Anterior eyedevelopment and ocular mesenchyme: new insightsfrom mouse models and human diseases. Bioessays,26, 374–386.

http://dx.doi.org/doi:10.1016/j.jmb.2005.05.072


4. Donner, A. L. & Maas, R. L. (2004). Conservation andnon-conservation of genetic pathways in eye specifi-cation. Int. J. Dev. Biol. 48, 743–753.

5. Gould, D. B., Smith, R. S. & John, S. W. M. (2004).Anterior segment development relevant to glaucoma.Int. J. Dev. Biol. 48, 1015–1029.

6. Cvekl, A., Yang, Y., Chauhan, B. K. & Cveklova, K.(2004). Regulation of gene expression by Pax6 inocular cells: a case of tissue-preferred expression ofcrystallins in lens. Int. J. Dev. Biol. 48, 829–844.

7. Kondoh, H., Uchikawa, M. & Kamachi, Y. (2004).Interplay of Pax6 and SOX2 in lens development as aparadigm of genetic switch mechanisms for celldifferentiation. Int. J. Dev. Biol. 48, 819–827.

8. Reza, H. M. & Yasuda, K. (2004). Lens differentiationand crystallin regulation: a chick model. Int. J. Dev.Biol. 48, 805–817.

9. Graw, J. (2003). The genetic and molecular basis ofcongenital eye defects. Nature Rev. Genet. 4, 876–888.

10. Bhat, S. P. (2003). Crystallins, genes and cataract. Prog.Drug Res. 60, 205–262.

11. Robinson, M. L. & Overbeek, P. A. (1996). Differentialexpression of alpha A- and alpha B-crystallin duringmurine ocular development. Invest. Ophthalmol. Vis.Sci. 37, 2276–2284.

12. Haynes, J. I., 2nd, Duncan, M. K. & Piatigorsky, J.(1996). Spatial and temporal activity of the alphaB-crystallin/small heat shock protein gene promoterin transgenic mice. Dev. Dyn. 207, 75–88.

13. Wigle, J. T., Chowdhury, K., Gruss, P. & Oliver, G.(1999). Prox1 function is crucial for mouse lens-fibreelongation. Nature Genet. 21, 318–322.

14. Duncan, M. K., Xie, L., David, L. L., Robinson, M. L.,Taube, J. R., Cui, W. & Reneker, L. W. (2004). EctopicPax6 expression disturbs lens fiber cell differentiation.Invest. Ophthalmol. Vis. Sci. 45, 3589–3598.

15. Duncan, M. K., Haynes, J. I., 2nd, Cvekl, A. &Piatigorsky, J. (1998). Dual roles for Pax-6: a transcrip-tional repressor of lens fiber cell-specific beta-crystal-lin genes. Mol. Cell. Biol. 18, 5579–5586.

16. Ueda, Y., Duncan, M. K. & David, L. L. (2002). Lensproteomics: the accumulation of crystallin modifi-cations in the mouse lens with age. Invest. Ophthalmol.Vis. Sci. 43, 205–215.

17. Wawrousek, E. F., Chepelinsky, A. B., McDermott, J. B.& Piatigorsky, J. (1990). Regulation of the murinealpha A-crystallin promoter in transgenic mice. Dev.Biol. 137, 68–76.

18. Ring, B. Z., Cordes, S. P., Overbeek, P. A. & Barsh, G. S.(2000). Regulation of mouse lens fiber celldevelopment and differentiation by the Maf gene.Development, 127, 307–317.

19. Kawauchi, S., Takahashi, S., Nakajima, O., Ogino, H.,Morita, M., Nishizawa, M. et al. (1999). Regulation oflens fiber cell differentiation by transcription factorc-Maf. J. Biol. Chem. 274, 19254–19260.

20. Nishiguchi, S., Wood, H., Kondoh, H., Lovell-Badge,R. & Episkopou, V. (1998). Sox1 directly regulates thegamma-crystallin genes and is essential for lensdevelopment in mice. Genes. Dev. 12, 776–781.

21. Ho, I. C., Hodge, M. R., Rooney, J. W. & Glimcher, L. H.(1996). The proto-oncogene c-maf is responsible fortissue-specific expression of interleukin-4. Cell, 85,973–983.

22. Hedge, S. P., Kumar, A., Kurschner, C. & Shapiro, L. H.(1998). c-Maf interactswith c-Myb to regulate transcrip-tion of an early myeloid gene during differentiation.Mol. Cell. Biol. 18, 2729–2737.

23. Lyon,M. F., Jamieson, R. V., Perveen, R., Glenister, P. H.,

Griffiths, R., Boyd, Y. et al. (2003). A dominant mutationwithin the DNA-binding domain of the bZIP transcrip-tion factor Maf causes murine cataract and results inselective alteration in DNA binding. Hum. Mol. Genet.12, 585–594.

24. MacLean, H. E., Kim, J. I., Glimcher, M. J., Wang, J.,Kronenberg, H. M. & Glimcher, L. H. (2003). Absenceof transcription factor c-maf causes abnormalterminal differentiation of hypertrophic chondrocytesduring endochondral bone development. Dev. Biol.262, 51–63.

25. Ashery-Padan, R., Marquardt, T., Zhou, X. & Gruss, P.(2000). Pax6 activity in the lens primordium isrequired for lens formation and for correct placementof a single retina in the eye. Genes Dev. 14, 2701–2711.

26. Cvekl, A., Kashanchi, F., Sax, C. M., Brady, J. N. &Piatigorsky, J. (1995). Transcriptional regulation of themouse alpha A-crystallin gene: activation dependenton a cyclic AMP-responsive element (DE1/CRE) anda Pax-6-binding site. Mol. Cell. Biol. 15, 653–660.

27. Kantorow, M., Cvekl, A., Sax, C. M. & Piatigorsky, J.(1993). Protein–DNA interactions of the mouse alphaA-crystallin control regions. Differences betweenexpressing and non-expressing cells. J. Mol. Biol.230, 425–435.

28. Ilagan, J. G., Cvekl, A., Kantorow, M., Piatigorsky, J. &Sax, C. M. (1999). Regulation of alphaA-crystallingene expression. Lens specificity achieved throughthe differential placement of similar transcriptionalcontrol elements in mouse and chicken. J. Biol. Chem.274, 19973–19978.

29. Sax, C. M., Ilagan, J. G. & Piatigorsky, J. (1993).Functional redundancy of the DE-1 and alphaA-CRYBP1 regulatory sites of the mouse alphaA-crystallin promoter. Nucl. Acids Res. 21, 2633–2640.

30. Chauhan, B. K., Yang, Y., Cveklova, K. & Cvekl, A.(2004). Functional interactions between alternativelyspliced forms of Pax6 in crystallin gene regulationand in haploinsufficiency. Nucl. Acids Res. 32,1696–1709.

31. Yang, Y., Chauhan, B. K., Cveklova, K. & Cvekl, A.(2004). Transcriptional regulation of mouse alphaB-and gammaF-crystallin genes in lens: oppositepromoter-specific interactions between Pax6 andlarge Maf transcription factors. J. Mol. Biol. 344,351–368.

32. Semenza, G. L. (1998). Transcription Factors and HumanDisease Oxford Monographs on Medical Genetics No. 37,Oxford University Press, Oxford.

33. Blank, V. & Andrews, N. C. (1997). The Maftranscription factors: regulators of differentiation.Trends Biochem. Sci. 22, 437–441.

34. Yang, Y. & Cvekl, A. (2005). Large Maf transcriptionfactors: cousins of AP-1 proteins and importantregulators of cellular differentiation. Einstein J. Biol.Med. in the press.

35. Yoshida, T. & Yasuda, K. (2002). Characterization ofthe chicken L-Maf, MafB and c-Maf in crystallin generegulation and lens differentiation. Genes Cells, 7,693–706.

36. Epstein, J., Cai, J., Glaser, T., Jepeal, L. & Maas, R.(1994). Identification of a Pax paired domain recog-nition sequence and evidence for DNA-dependentconformational changes. J. Biol. Chem. 269, 8355–8361.

37. Lengler, J., Krausz, E., Tomarev, S., Prescott, A.,Quinlan, R. A. & Graw, J. (2001). Antagonistic actionof Six3 and Prox1 at the gamma-crystallin promoter.Nucl. Acids Res. 29, 515–526.

38. Rekhtman, N., Choe, K. S., Matushansky, I., Murray,


S., Stopka, T. & Skoultchi, A. I. (2003). PU.1 and pRBinteract and cooperate to repress GATA-1 and blockerythroid differentiation.Mol. Cell. Biol. 23, 7460–7474.

39. Irvine, R. A. & Hsieh, C. L. (2004). Q-PCR incombination with ChIP assays to detect changes inchromatin acetylation. Methods Mol. Biol. 287, 45–52.

40. Zavadil, J., Cermak, L., Soto-Nieves, N. & Bottinger,E. P. (2004). Integration of TGF-beta/Smad andJagged1/Notch signalling in epithelial-to-mesenchy-mal transition. EMBO J. 23, 1155–1165.

41. Liu, Q., Ji, X., Breitman, M. L., Hitchcock, P. F. &Swaroop, A. (1996). Expression of the bZIP transcrip-tion factor gene Nrl in the developing nervoussystem. Oncogene, 12, 207–211.

42. Sakai, M., Imaki, J., Yoshida, K., Ogata, A.,Matsushima-Hibaya, Y., Kuboki, Y., Nishizawa, M.& Nishi, S. (1997). Rat maf related genes: specificexpression in chondrocytes, lens and spinal cord.Oncogene, 14, 745–750.

43. Lecoin, L., Sii-Felice, K., Pouponnot, C., Eychene, A. &Felder-Schmittbuhl, M. P. (2004). Comparison of mafgene expression patterns during chick embryodevelopment. Gene Expr. Patterns, 4, 35–46.

44. Kamachi, Y., Uchikawa, M., Tanouchi, A., Sekido, R. &Kondoh, H. (2001). Pax6 and SOX2 form a co-DNA-binding partner complex that regulates initiation oflens development. Genes Dev. 15, 1272–1286.

45. Sander, M., Neubuser, A., Kalamaras, J., Ee, H. C.,Martin, G. R. & German, M. S. (1997). Genetic analysisreveals that PAX6 is required for normal transcriptionof pancreatic hormone genes and islet development.Genes Dev. 11, 1662–1673.

46. Planque, N., Leconte, L., Coquelle, F. M., Benkhelifa,S., Martin, P., Felder-Schmittbuhl, M. P. & Saule, S.(2001). Interaction of Maf transcription factors withPax-6 results in synergistic activation of the glucagonpromoter. J. Biol. Chem. 276, 35751–35760.

47. Kataoka, K., Shioda, S., Ando, K., Sakagami, K.,Handa, H. & Yasuda, K. (2004). Differentiallyexpressed Maf family transcription factors, c-Mafand MafA, activate glucagon and insulin geneexpression in pancreatic islet alpha- and beta-cells.J. Mol. Endocrinol. 32, 9–20.

48. Brink, C. (2003). Promoter elements in endocrinepancreas development and hormone regulation. Cell.Mol. Life Sci. 60, 1033–1048.

49. Sivak, J. M., West-Mays, J. A., Yee, A., Williams, T. &Fini, M. E. (2004). Transcription factors Pax6 and AP-2alpha interact to coordinate corneal epithelial repairby controlling expression of matrix metalloproteinasegelatinase B. Mol. Cell. Biol. 24, 245–257.

50. Kozmik, Z., Daube, M., Frei, E., Norman, B., Kos, L.,Dishaw, L. J. et al. (2003). Role of Pax genes in eyeevolution: a cnidarian PaxB gene uniting Pax2 andPax6 functions. Dev. Cell, 5, 773–785.

51. Carosa, E., Kozmik, Z., Rall, J. E. & Piatigorsky, J.(2002). Structure and expression of the scallopOmega-crystallin gene. Evidence for convergentevolution of promoter sequences. J. Biol. Chem. 277,656–664.

52. Zhao, H., Yang, Y., Rizo, C. M., Overbeek, P. A. &Robinson, M. L. (2004). Insertion of a Pax6 consensusbinding site into the alphaA-crystallin promoter actsas a lens epithelial cell enhancer in transgenic mice.Invest. Ophthalmol. Vis. Sci. 45, 1930–1939.

53. Sakai, M., Serria, M. S., Ikeda, H., Yoshida, K., Imaki, J.& Nishi, S. (2001). Regulation of c-maf geneexpression by Pax6 in cultured cells. Nucl. Acids Res.29, 1228–1237.

54. Ogino, H. & Yasuda, K. (1998). Induction of lensdifferentiation by activation of a bZIP transcriptionfactor, L-Maf. Science, 280, 115–118.

55. Reza, H. M., Ogino, H. & Yasuda, K. L-M. (2002). Maf,a downstream target of Pax6, is essential for chick lensdevelopment. Mech. Dev. 116, 61–73.

56. Duncan, M. K., Cvekl, A., Kantorow, M. &Piatigorsky, J. (2004). Lens crystallins. In Developmentof the Ocular Lens (Lovicu, F. J. & Robinson, M. L., eds),pp. 119–150, Cambridge University Press.

57. Reza, H. M. & Yasuda, K. (2004). Roles of Maf familyproteins in lens development. Dev. Dyn. 229, 440–448.

58. Goring, D. R., Rossant, J., Clapoff, S., Breitman, M. L.& Tsui, L. C. (1987). In situ detection of beta-galactosidase in lenses of transgenic mice with agamma-crystallin/lacZ gene. Science, 235, 456–458.

59. Rajaram, N. & Kerppola, T. K. (2004). Synergistictranscription activation by Maf and Sox and theirsubnuclear localization are disrupted by amutation inMaf that causes cataract.Mol. Cell. Biol. 24, 5694–5709.

60. Cui, W., Tomarev, S. I., Piatigorsky, J., Chepelinsky, A.B. & Duncan, M. K. (2004). Mafs, Prox1, and Pax6 canregulate chicken betaB1-crystallin gene expression.J. Biol. Chem. 279, 11088–11095.

61. Carriere, C., Plaza, S., Martin, P., Quatannens, B.,Bailly, M., Stehelin, D. & Saule, S. (1993). Character-ization of quail Pax-6 (Pax-QNR) proteins expressedin the neuroretina. Mol. Cell. Biol. 13, 7257–7266.

62. Chauhan, B. K., Yang, Y., Cveklova, K. & Cvekl, A.(2004). Functional properties of natural human PAX6and PAX6(5a) mutants. Invest. Ophthalmol. Vis. Sci. 45,385–392.

63. Livak, K. J. & Schmittgen, T. D. (2001). Analysis ofrelative gene expression data using real-time quanti-tative PCR and the 2(-Delta Delta C(T)) method.Methods, 25, 402–408.

64. Chen, Q., Dowhan, D. H., Liang, D., Moore, D. D. &Overbeek, P. A. (2002). CREB-binding protein/p300co-activation of crystallin gene expression. J. Biol.Chem. 277, 24081–24089.

Edited by M. Yaniv

(Received 4 February 2005; received in revised form 25 April 2005; accepted 25 May 2005)Available online 6 July 2005

Documents

Tissue-specific Regulation of the Mouse αA-crystallin Gene in Lens via Recruitment of Pax6 and c-Maf to its Promoter