twist1 dimer selection regulates cranial suture patterning...

RESEARCH ARTICLE

Twist1 Dimer Selection Regulates CranialSuture Patterning and FusionJeannette Connerney,† Viktoria Andreeva,† Yael Leshem,†‡ Christian Muentener,Miguel A. Mercado, and Douglas B. Spicer*

Saethre-Chotzen syndrome is associated with haploinsufficiency of the basic-helix–loop–helix (bHLH)transcription factor TWIST1 and is characterized by premature closure of the cranial sutures, termedcraniosynostosis; however, the mechanisms underlying this defect are unclear. Twist1 has been shown toplay both positive and negative roles in mesenchymal specification and differentiation, and here we showthat the activity of Twist1 is dependent on its dimer partner. Twist1 forms both homodimers (T/T) andheterodimers with E2A E proteins (T/E) and the relative level of Twist1 to the HLH inhibitor Id proteinsdetermines which dimer forms. On the basis of the expression patterns of Twist1 and Id1 within the cranialsutures, we hypothesized that Twist1 forms homodimers in the osteogenic fronts and T/E heterodimers inthe mid-sutures. In support of this hypothesis, we have found that genes regulated by T/T homodimers, suchas FGFR2 and periostin, are expressed in the osteogenic fronts, whereas genes regulated by T/Eheterodimers, such as thrombospondin-1, are expressed in the mid-sutures. The ratio between these dimersis altered in the sutures of Twist1�/� mice, favoring an increase in homodimers and an expansion of theosteogenic fronts. Of interest, the T/T to T/E ratio is greater in the coronal versus the sagittal suture, andthis finding may contribute to making the coronal suture more susceptible to fusion due to TWISThaploinsufficiency. Importantly, we were able to inhibit suture fusion in Twist1�/� mice by modulating thebalance between these dimers toward T/E formation, by either increasing the expression of E2A E12 or bydecreasing Id expression. Therefore, we have identified dimer partner selection as an important mediatorof Twist1 function and provide a mechanistic understanding of craniosynostosis due to TWISThaploinsufficiency. Developmental Dynamics 235:1345–1357, 2006. © 2006 Wiley-Liss, Inc.

Key words: Twist; bHLH; craniosynostosis; gene expression; Saethre-Chotzen syndrome; craniofacial development

Accepted 16 January 2006

INTRODUCTION

The growth of the cranium occursthrough a complex interplay betweenthe intervening suture mesenchyme,the osteogenic fronts, and the under-lying dura mater and brain and in-volves several different signaling

pathways and transcriptional regula-tors (Opperman, 2000; Jiang et al.,2002). Although there has been quitea bit of progress made toward theidentification of many of the genesthat are important in this process,much less is known about how thesepathways interact with each other to

mediate these events (Ishii et al.,2003; Rice et al., 2005). Here, we focuson the role of the transcription factorTwist1 in mediating cranial suturepatency.

Craniosynostosis is a fairly commondisorder occurring in approximately 1in 2,500 individuals. Nonsyndromic

Center for Molecular Medicine, Maine Medical Center Research Institute, Scarborough, MaineGrant support: NIH/NIDCR/NCRR; Grant number: P20 RR15555; Grant number: R01 DE015329; Grant sponsor: American Heart Associ-ation; Grant number: 0130393.†Drs. Connerney, Andreeva, and Leshem contributed equally to this work.‡Dr. Leshem’s present address is Department of Biological Chemistry, Institute of Life Sciences, The Hebrew University of Jerusalem,91904, Israel.*Correspondence to: Douglas B. Spicer, Center for Molecular Medicine, Maine Medical Center Research Institute, 81 Re-search Drive, Scarborough, ME 04074. E-mail: [email protected]

DOI 10.1002/dvdy.20717Published online 24 February 2006 in Wiley InterScience (www.interscience.wiley.com).

DEVELOPMENTAL DYNAMICS 235:1345–1357, 2006

© 2006 Wiley-Liss, Inc.

craniosynostosis is most common;however, 20% of all cases are associ-ated with mutations in the geneTWIST1 or one of the fibroblastgrowth factor receptor (FGFR) genes.TWIST1 haploinsufficiency is associ-ated with Saethre-Chotzen syndrome,which is the most common autosomaldominant disorder of craniosynostosis(el Ghouzzi et al., 1997; Howard et al.,1997). More than 50 different muta-tions have been identified in theTWIST1 gene and are predicted tocause loss of function (Gripp et al.,2000), which is also indicated byTwist1�/� mice presenting a similar

phenotype (el Ghouzzi et al., 1997;Bourgeois et al., 1998; Carver et al.,2002). Conversely, a family has beenidentified with trisomy at the TWIST1locus resulting in cranium bifidum,which is characterized by a persistentcalvarial foramen and open sutures(Stankiewicz et al., 2001). These twoopposing clinical phenotypes resultingfrom either half or one and a halftimes the normal amount of Twist1,respectively, illustrate the critical re-quirement for the tight regulation ofTwist1 expression.

We and others have shown thatTwist1 has both positive and negative

functions regulating mesenchymalcell specification and differentiation.Twist was originally identified in Dro-sophila as a gene required for gastru-lation and mesoderm formation butalso has a seemingly opposing but evo-lutionarily conserved function of in-hibiting the differentiation of mesen-chymal tissues, including muscle andbone (Castanon and Baylies, 2002;O’Rourke and Tam, 2002). The mech-anisms underlying these disparatefunctions are unclear but may be de-pendent on the dimer partner ofTwist.

Basic-helix–loop–helix (bHLH) tran-scription factors are classified into dif-ferent categories based on their tissuedistribution, partner choice, andDNA-binding and structural proper-ties. Twist falls into Class II, whichcontains tissue-specific bHLH pro-teins such as MyoD. These proteinsform heterodimers with Class I bHLHproteins, termed E proteins, which arewidely expressed in many tissues. Idproteins represent a third class ofHLH proteins that lack the basic do-main and, therefore, cannot bindDNA. Id proteins preferentiallydimerize with E proteins and disruptfunctional Class I/II bHLH het-erodimers from forming (Massari andMurre, 2000). The majority of Class IIbHLH proteins do not form stable ho-modimers (Vinals et al., 2004; Vinalsand Ventura, 2004); however, studiesin Drosophila suggest that ho-modimers of Twist mediate mesodermformation (Castanon et al., 2001).Therefore, Twist may uniquely formfunctional heterodimers (T/E) and ho-modimers (T/T) that may have differ-ent activities and may account for theopposing actions ascribed to Twist1.The ratio of T/T to T/E within a cell,therefore, would determine the func-tional output of Twist1 expression.

The cranial sutures are the growthcenters separating the bones of theskull and are composed of two oppos-ing osteogenic fronts and an interven-ing mesenchyme (Fig. 1). Twist1 is ex-pressed throughout the suturemesenchyme and osteogenic fronts,whereas Id1 expression is confined tothe osteogenic fronts (Johnson et al.,2000; Rice et al., 2000). E2A (E12 andE47) and HEB E proteins are ex-pressed throughout the suture and indifferentiating osteoblasts (Funato et

Fig. 1. Model of the regulation of cranial suture patency by Twist1. a: T/E heterodimers induceTSP-1 expression in the mid-suture mesenchyme, which then activates latent transforming growthfactor-beta (TGF�) and inhibits differentiation. Bone morphogenetic protein (BMP) and TGF�activity may coordinately restrict Id expression to the osteogenic fronts. Id expression promotesT/T homodimer formation, which enhances fibroblast growth factor receptor-2 (FGFR2) expression,further driving the formation of the osteogenic fronts. Changes in fibroblast growth factor (FGF),BMP, or TGF� levels would then alter the balance between T/E and T/T to either enhance or inhibitsuture closure. b: Proposed mechanism promoting craniosynostosis due to Twist1 haploinsuffi-ciency. Graph of the relative levels of Id and Twist1 across the suture shown in (a). In the osteogenicfronts where Id levels are higher than Twist1, T/T dimers form while T/E dimers form in themid-suture. When Twist1 levels are lowered due to haploinsufficiency, the area where Id is higherthan Twist1 expands toward the mid-suture. [Color figure can be viewed in the online issue, whichis available at www.interscience.wiley.com.]

1346 SPICER ET AL.

al., 2001). We hypothesized that Idcompetes with Twist1 for dimeriza-tion with E proteins in the osteogenicfronts, forcing Twist1 to form ho-modimers, while in the interveningsuture mesenchyme, where Id is ab-sent, Twist1 forms T/E heterodimers.These dimers then differentially regu-late gene expression and cell behaviorin these areas. Our data support thishypothesis and indicate that there isdynamic regulation of Twist1 dimerformation in the cranial sutures thatis altered in Twist1�/� mice. Twist1haploinsufficiency increases the ratioof T/T to T/E, and conditions that pro-mote T/E formation in the sutures pre-vent fusion in Twist1�/� mice.

RESULTS

Id Levels Can ModulateTwist1 Dimer Composition

Twist1 is expressed in both the osteo-genic fronts and in the mid-suture(Johnson et al., 2000; Rice et al., 2000;Oshima et al., 2002), whereas Id1 isonly in the osteogenic fronts (Rice etal., 2000), and, therefore, it was sug-gested that Id inhibits Twist1 activityin the osteogenic fronts. Given that Idpreferentially dimerizes with E pro-teins rather than class II HLH pro-teins like Twist1 (Benezra et al.,1990), we have tested an alternativehypothesis, that Id expression pro-motes Twist1 homodimer formation inthe osteogenic fronts. To determinewhether increasing Id levels woulddrive Twist1 to form T/T over T/Edimers, we developed a sodium dode-cyl sulfate-polyacrylamide gel electro-phoresis (SDS-PAGE) analysis wherewe could detect the Twist1 dimersthat formed. The 293T HEK cells weretransfected with constructs express-ing myc-tagged Twist1, myc-taggedId1, and E2A E12 in different combi-nations. Protein extracts were sub-jected to nonreducing SDS-PAGE gelsfollowed by Western blot analysiswithout heating the samples beforeloading the gels (Fig. 2a, middle andright gels). When either Twist1 orE2A E12 was transfected alone (lanes2 and 3, respectively), two bands weredetected in the nonreducing gels (mid-dle and right gels) corresponding tothe size of the monomers (T and E)and the respective homodimers (T/T

Fig. 2. Id levels alter the ratio of T/T and T/E dimers. a: HEK293 cells were transfected withTwist1- (T), E2A E12- (E), and Id1 (I) -expressing plasmids as indicated. Id1 levels were 0.25-, 0.5-,1-, and 2-fold the level of Twist1 and E2A E12 in lanes 5–8, respectively. For the two right gels,samples were prepared without �-mercaptoethanol and were not heated before loading but werethe same as the reducing gel on the left in all other respects. The Twist1 and the Id1 proteinscontained myc-epitope tags and were detected with 9E10 anti-myc monoclonal antibodies in theleft and right blots. E2A E12 proteins were detected with anti-E2a polyclonal antibody, whichrecognizes both E2A E12 and E47, in the center and right blots. The sizes for the T, E, and Idmonomers and the T/T, T/E, and E/E dimers are indicated on the sides. b: Diagram depicting thestructure of the “forced dimmers.” Two Twist1 monomers or Twist1 and E2A E12 monomers arelinked in frame by a 29 amino acid linker. c: In vitro translated proteins were used in an electro-phoretic mobility shift assay (EMSA) using a DNA probe containing the right E Box from the MCKgene. A total of 0.5-, 1-, and 2-fold amounts of Id1 protein were added to reactions with either theTE heterodimer (TE) or with Twist1 and E2A E12 (T/E) monomeric proteins where indicated. Lysatecontrol (L). d: EMSA analysis as in c, where 0.5-, 1-, and 2-fold amounts of E2A E12 protein wereadded to reactions with either the tethered homodimer (TT) or with the Twist1 (T) protein whereindicated.

REGULATION OF GENE EXPRESSION BY TWIST1 1347

and E/E). Cotransfection of Twist1and E2A E12 led to the formation ofa new band corresponding to the sizeof T/E heterodimers that was de-tected with both anti-myc and anti-E2A E12 antibodies, along with thedisappearance of the homodimerbands (lane 4, middle and right gels).Increasing amounts of Id1 proteincaused the T/E band to decrease anda band corresponding to T/T ho-modimers to appear (lane 4 – 8, mid-dle gel). No dimers were detectedwhen extracts were heated before

loading and run under reducing con-ditions (Fig. 2a, left gel). These datasupport previous work indicatingthat Id preferentially interacts withE proteins (Benezra et al., 1990) anddoes not efficiently interact withTwist1, which we have also con-firmed using GST fusion and in vitrotranslated proteins (data notshown). Therefore, Id levels deter-mine the amount of free E proteinthat is available to dimerize withTwist1 and, hence, can determinewhich Twist1 dimer is formed.

Forced Dimers of Twist1 AreResistant to Competition byother HLH Proteins

To determine whether T/T homo- andT/E heterodimers have different activi-ties, we constructed “forced” ho-modimers of Twist1 (TT) and het-erodimers of Twist1 and E2A E12 (TE)where the two monomers are linked bya flexible glycine–serine polylinker (Fig.2b). This strategy was originally usedwith MyoD and the E protein E2A E47,where it was shown that the tethereddimer was more resistant to inhibitionby Id than the separate monomers(Neuhold and Wold, 1993). The strategyhas now been used successfully withseveral different transcription factors(Sigvardsson et al., 1997; Castanon etal., 2001; Bakiri et al., 2002). Becausewe have found no difference in the ac-tivity of Twist1 heterodimers composedof E2A E12, E2A E47, or HEB E pro-teins (data not shown), we used E2AE12 for our studies here. To confirmthat the tethered dimers could bind to atarget DNA-binding sequence (E box)and were resistant to inhibition byother HLH proteins, we performed anelectrophoretic mobility shift assay(EMSA) comparing the “forced” dimersto the separated monomers in theirability to bind oligos containing an Ebox. Using in vitro translated proteins,we found that both tethered dimers (TTand TE) formed specific complexes withthe E box probe (Fig. 2c,d). These com-plexes were slightly smaller than thecomplexes formed by the separatedmonomers (T/T and T/E), and this sizedifference is consistent with each of theseparated monomers having mycepitope tags whereas only the amino-terminal partner of the tethered dimershas myc tags. As expected, the additionof increasing amounts of Id led to disso-ciation of the T/E monomer complex,whereas DNA binding by the TE teth-ered dimer was not affected by Id (Fig.2c). Similarly, addition of the E2A E12monomer quickly decreased the T/T ho-modimer complex with a commensurateincrease in a T/E complex bound toDNA. The TT tethered dimer complex,however, was more resistant to dissoci-ation by increasing levels of E2A E12(Fig. 2d). Therefore, both T/E het-erodimers and T/T homodimers formand bind to E box-containing DNA. Fur-thermore, TT and TE tethered dimers

Fig. 3. Twist1 dimers differentially regulate gene expression (a–e) 10T1/2 cells were infected withpWZL retroviral constructs to stably express Twist1 (T), the T/T homodimer (TT), or the T/Eheterodimer (TE). Control 10T1/2 cells were infected with insertless pWZL virus (WZL). a: TTenhanced periostin expression, whereas TE cells inhibited it. The 10T1/2 cell lines were grown in10% fetal bovine serum (FBS) without or with 100 ng/ml bone morphogenetic protein-7 (BMP7) for4 days and then assayed for periostin and �-actin expression by Western blot analysis. b: Id1 isup-regulated by BMP signaling. 10T1/2 cells were grown in 10% FBS without or with 100 ng/mlBMP7 for 4 days and then assayed for Id1, Id3, and glyceraldehydes-3-phosphate dehydrogenase(GAPDH) expression by reverse transcriptase-polymerase chain reaction (RT-PCR). c: TE inducedthrombospondin 1 (TSP-1) expression. 10T1/2 cell lines were grown in 10% FBS for 48 hr and thenassayed for TSP-1 expression by immunofluorescence. d: 10T1/2 cell lines were grown in 0.2%FBS for 48 hr, and then the conditioned medium was assayed for TSP-1 expression by Western blotanalysis after concentration on a heparin–sepharose column and elution with 1.5 M NaCl. e: TT andTE dimers differentially regulate fibroblast growth factor receptor-2 (FGFR2) expression. 10T1/2 celllines were grown without or with 100 ng/ml BMP7 for 4 days and were analyzed for FGFR1, FGFR2,and GAPDH expression by RT-PCR. [Color figure can be viewed in the online issue, which isavailable at www.interscience.wiley.com.]

1348 SPICER ET AL.

can also effectively bind these DNAprobes and their complexes with DNAare more stable than the separatemonomers when challenged by otherHLH proteins.

Twist1 Dimers DifferentiallyRegulate Gene Expression

While Twist1 has been implicated inthe regulation of cell specification anddifferentiation little is known aboutwhat genes Twist1 regulates to medi-ate these effects. We, therefore,wanted to determine whether T/T andT/E dimers regulated different sets ofgenes. We also reasoned that the ex-pression patterns of genes that are dif-ferentially regulated by the Twist1dimers could be used to identify re-gions in the sutures where each dimerresides. We previously had performedmicroarray analysis comparing con-trol C3H10T1/2 embryonal fibroblasts(10T1/2 cells) with ones overexpress-ing Twist1. We had used 10T1/2 cellsbecause they are a multipotentialmesenchymal cell line able to undergochondrogenesis and osteogenesis, butunlike primary calvarial cells, the en-dogenous levels of Twist1 are verylow, allowing us to better determinethe specific effects of the expression ofdifferent forms of Twist1. We foundthat periostin, a member of the fasci-clin I protein family, was significantlyrepressed by Twist1 and these datawere confirmed by RT-PCR (data notshown) and Western analysis (Fig.3a). Of interest, periostin was identi-fied recently as a gene significantlyinduced by Twist1 in SAOS2 osteosar-coma cells (Oshima et al., 2002). Wereasoned that our differing resultsmight be due to the different cultureconditions favoring the formation ofeither Twist1 homodimers or het-erodimers with endogenous E pro-teins, and these dimers may differen-tially affect periostin expression.Indeed we found that 10T1/2 cells ex-pressing TT enhanced periostin ex-pression, whereas expression of TE in-hibited it (Fig. 3a). Cells expressingthe Twist1 monomer more resembledTE-expressing cells and expressedless periostin. We next asked whetherincreasing Id levels in the Twist1-ex-pressing cells would promote more of aTT phenotype. Bone morphogeneticproteins (BMPs) induce Id1 expres-

sion in the calvarial sutures (Rice etal., 2000) as well as in many cell linesincluding 10T1/2 cells (Ogata et al.,1993). We confirmed that BMP7 in-duced Id1 expression in 10T1/2 cellsand found that Id3 expression wasalso slightly induced in these cells(Fig. 3b). Id2 and Id4 were not de-tected (data not shown). Consistentwith our hypothesis, in the presence ofBMP7 where Id levels were high,Twist1-induced periostin expression(Fig. 3a, compare lanes 2 and 6).

Another gene that we found to bedifferentially regulated by the Twist1dimers was thrombospondin 1 (TSP-1). Recently, TSP-1 was shown to beup-regulated in mesenchymal tissuesof Id1�/� mice (Volpert et al., 2002).Because Id does not directly regulategene expression, we hypothesized thatthere are more free E proteins avail-able in the Id1�/� mice, which maylead to increasing amounts of T/Edimers formed and these dimers maymediate the induction of TSP-1 ex-pression. Consistent with this inter-pretation, we found that TSP-1 ex-pression was significantly induced inthe 10T1/2 cells expressing TE (Fig.3c). There was also significantly moreTSP-1 in the conditioned media of TE-expressing cells than in control or TTcells and the Twist1-expressing cellshad an intermediate level of TSP-1(Fig. 3d). Therefore, TT dimers in-duced periostin expression, whereasTE dimers inhibited periostin and in-duced TSP-1 expression. Further-more, Twist1 cells behaved similar toTE cells under conditions where Idlevels were low but acted like TT cellswhen Id levels were increased.

Twist1 Dimers DifferentiallyRegulate FGFR2 Expression

FGFR2 is normally expressed in theosteogenic fronts, and its expressionexpands into the mid-suture ofTwist1�/� mice (Rice et al., 2000),possibly suggesting that Twist1 nor-mally represses FGFR2 expression.However, Twist1 has been shown re-cently to bind to the FGFR2 promoterand induce its expression (Guenou etal., 2005). Furthermore, Twist inducesFGFR expression during Drosophilagastrulation (Shishido et al., 1993),and this finding may be mediated byTwist homodimers (Castanon et al.,

2001). We, therefore, examined our10T1/2 cell lines to see if either of theTwist1 dimers affected the expressionof FGFR2. As can be seen in Figure 3e,TT-expressing 10T1/2 cells inducedFGFR2 expression while there was noexpression in TE and control cells. Un-like the case with periostin and TSP-1where the Twist1-expressing cells be-haved more similarly to the TE cells,FGFR2 was induced in the T cells sim-ilarly to the TT cells. Because there ismost likely a mix of T/T and T/Edimers in the T cells, this finding sug-gests that genes are differentially sen-sitive to the ratio or amount of the twoTwist1 dimers. FGFR1, whose expres-sion does not overlap with Twist1 inthe sutures (Johnson et al., 2000; Riceet al., 2000), was expressed equiva-lently in all of the cell lines. The addi-tion of BMP7 induced FGFR2 expres-sion in control 10T1/2 cells; however,TE expression inhibited this induc-tion. Therefore, T/T and T/E dimershave opposing effects on FGFR2 ex-pression.

T/E and T/T RegulatedGenes Are Expressed in thePredicted Domains Withinthe Cranial Sutures

We next asked whether the expressionpattern of these genes that are differen-tially regulated by the Twist1 dimerscould be predicted in the cranial suturesbased upon the expression patterns ofTwist1 and Id1. As outlined above, be-cause Twist1 is expressed in both theosteogenic fronts and in the mid-suturewhile Id1 is only in the osteogenic fronts(Johnson et al., 2000; Rice et al., 2000;Oshima et al., 2002)(Fig. 4n), we pre-dicted that T/T dimers would be formedin the osteogenic fronts whereas T/Edimers would be in the mid-sutures.Consistent with our hypothesis, we ob-served periostin expression in the osteo-genic fronts (Fig. 4a), while TSP-1 wasexpressed in the mid-suture area (Fig.4e). Of interest, TSP-1 was only signif-icantly expressed in the sagittal andfrontal sutures and not in the coronalsuture (Fig. 4e,f,m), while periostin wasexpressed in all the sutures (Fig.4a,b,m). Based upon our model (see Fig.1), this finding suggested that the coro-nal suture may have a higher ratio ofT/T to T/E. We, therefore, analyzed theexpression patterns of Twist1 and Id


proteins in the sagittal and coronal su-tures to see if they would suggest a dif-ference in dimer formation. Twist1 wassimilarly expressed in the osteogenicfronts and middle of both sutures (Fig.4n); however, Id proteins were more ex-tensively expressed in the coronal ver-sus the sagittal suture (Fig. 4n). In thesagittal suture, Id proteins were highlyexpressed in the osteogenic fronts, withlittle or low expression in the mid-su-ture while in the coronal suture, only afew cells in the mid-suture did not ex-press Id. Consistent with an increase inT/T formation in this area, there wasonly a small area in the mid-coronalsuture that did not express periostin(Fig. 4b). We propose that this differ-ence may be part of the reason that it isprimarily the coronal suture that fusesdue to TWIST1 haploinsufficiency (seebelow).

T/T and T/E RegulatedGenes Are Altered inTwist1�/� Mice

The mechanism that promotes cranio-synostosis due to TWIST1 haploinsuf-ficiency is still unclear, but we suggestthat the decrease in the level of Twist1alters the balance between T/T andT/E dimers in the sutures, resulting ina change of gene expression and cellbehavior. The expansion of FGFR2 ex-pression into the mid-suture (Rice etal., 2000) suggests that there is anincrease in T/T formation in the su-tures of Twist1�/� mice. We proposethat this is due to Id levels being rel-atively higher than Twist1 in a largerarea resulting in less free E proteinsbeing available for dimerization withTwist1 (see Fig. 1b). To test this hy-pothesis, we examined the expressionof T/T and T/E-regulated genes in thesutures of wild-type and Twist1�/�mice. As was identified previously(Rice et al., 2000), we found that inTwist1�/� mice the expression do-main of FGFR2 in the sagittal suturewas expanded into the mid-suture(Fig. 4, compare i with k). We alsofound that, whereas periostin expres-sion barely extended to the tip of thecalvaria bones in wild-type mice, itsexpression was extended both ven-trally and medially toward the mid-suture of Twist1�/� mice (Fig. 4,compare a with c). There was a simi-lar, but more subtle, effect in the coro-

nal suture. Although there seemed tobe a decrease in the level of periostinexpression in the coronal suture of

Twist1�/� mice, the expression do-main of both periostin and FGFR2 ex-panded to cover the entire suture (Fig.

Fig. 4. Periostin and TSP-1 are expressed in the predicted homodimer (T/T) and heterodimer (T/E)domains in the cranial sutures and are altered by changes in Twist1 expression. a–l: Sections throughthe sagittal (a,c,e,g,i,k) and coronal (b,d,f,h,j,l) sutures of the skulls of newborn (P1) wild-type andTwist1�/� mice were analyzed by in situ hybridization for periostin (a–d), TSP-1 (e–h), and fibroblastgrowth factor receptor-2 (FGFR2; i–l). Calvaria bones are outlined by dotted red lines. Note that, inwild-type sutures, periostin expression does not extend to the end of the cranial bones (a) but does inTwist1�/� sutures (c). Expression of FGFR2 is increased (compare I and k) and TSP-1 is decreased(compare e and g) in the mid-sagittal suture of Twist1�/� mice. Arrows point to gene expression. m:Skulls from wild-type P1 mice were analyzed for periostin and TSP-1 expression by whole mount in situhybridization. Note the decreased expression of TSP-1 in the coronal suture compared with the sagittaland frontal sutures. C, coronal; F, frontal; S, sagittal. n: Twist1 and Id protein expression in the sagittaland coronal sutures. Paraffin sections of sagittal and coronal sutures of wild-type P1 mice wereanalyzed by immunohistochemistry for Twist1 and Id protein expression. Note that the Id antibodyrecognizes all Id proteins (Id1-4). o: Wild-type and Twist1�/� P1 skulls were analyzed for TSP-1expression in the sagittal suture by whole-mount in situ hybridization. Note the decrease in TSP-1staining in the Twist1�/� skull. p: Increased Twist1 expression in the osteogenic fronts induces TSP-1expression. P1 skulls of CAGCAT-Twist1 and CAGCAT-Twist1/Wnt1-Cre mice were analyzed forTSP-1 expression by whole-mount in situ hybridization. Arrowheads point to the osteogenic fronts andarrows indicate expression of TSP-1 in the frontal and sagittal sutures.

1350 SPICER ET AL.

4, compare b with d, and j with l).More dramatically, there was a signif-icant decrease in TSP-1 expression inthe sagittal suture of Twist1�/� mice(Fig. 4, compare e with g, and Fig. 4o).This increase in FGFR2 and periostinand decrease in TSP-1 is consistentwith our hypothesis that, in the su-tures of Twist1�/� mice, there is anincrease in the ratio of T/T to T/E.

Because decreasing Twist1 levelsresulted in an increase in the T/T toT/E ratio, we next performed the con-verse experiment by increasingTwist1 expression in the cranial neu-ral crest, which gives rise to these su-tural cells (Jiang et al., 2002), to see if

Fig. 6.

Fig. 4. (Continued)

Fig. 6. Promotion of heterodimers (T/E) dimerformation prevents suture fusion in Twist1�/�mice. a,b: Skulls of wild-type (a) and Twist1�/�(b) newborn (P1) mice. c–f: Skull explants fromnewborn (P1) wild-type (c) and Twist1�/� (d–f)mice grown in culture for 5 weeks. Skulls wereprocessed with alizarin red and then paraffinsectioned. Sections were then treated with vonKossa stain to indicate calcified bone (blackstaining). Sections through the left (L) and right(R) coronal sutures are shown below skulls. Theright coronal suture of e and f was injected withadenovirus expressing either green fluorescentprotein (GFP; e) or E2A E12 (f) at the time ofexplant. Note the fusion of the sutures in theTwist1�/� skulls and the lack of fusion in theright coronal suture injected with Ad-E2A E12 (f).


this promoted T/E dimer formation.To achieve this objective, we usedWnt1-cre transgenic mice to activatean inducible transgene expressingTwist1. Consistent with our hypothe-sis, we found that TSP-1 expressionwas induced in the osteogenic fronts ofthese mice (Fig. 4p). TSP-1 expressionwas also lost from the mid-suture ofthese mice, suggesting a change indimer formation there as well. Thefull phenotype of these mice will bedescribed elsewhere. Therefore, de-

creased and increased levels of Twist1in the cranial sutures change the ex-pression of the genes regulated by theT/T and T/E dimers in a predictedmanner that supports our hypothesis.

Twist1 RequiresHeterodimerization toInhibit OsteoblastDifferentiation

Overexpression of Twist1 has beenshown to negatively regulate osteo-

blast differentiation (Lee et al., 1999;Funato et al., 2001), and osteoblastsisolated from individuals that werehaploinsufficient for TWIST1 differ-entiated faster than control cells(Yousfi et al., 2001). Furthermore,Twist1 has been implicated recentlyin directly inhibiting the activity ofthe transcription factor Runx2, whichpromotes osteoblast differentiation(Bialek et al., 2004). Runx2-depen-dent osteoblast differentiation, how-ever, initiates in the osteogenic frontswhere Twist1 is expressed. Primarycalvaria cells were infected with ade-novirus expressing Twist1 or E2A E12alone or in combination to determinewhether Twist1 required hetero-dimerization to inhibit osteoblast dif-ferentiation. Equal levels of myc-tagged Twist1 and E2A E12 wereexpressed after 2 weeks of differenti-ation (Fig. 5a). Surprisingly, we foundthat cells transduced with a Twist1-expressing adenovirus alone were notinhibited from differentiating asRunx2, �1(I) collagen, bone sialopro-tein (BSP), and osteocalcin (OC) tran-scripts were all induced (Fig. 5b). Thecombination of Twist1 and E2A E12,however, inhibited the later markersof osteoblast differentiation (BSP andOC) but did not affect the expressionof the early markers (Runx2 and �1(I)collagen). E2A E12 alone inhibited theexpression of OC, and this inhibitionmay have been due to it dimerizingwith the endogenous Twist1 proteinsthat were expressed at high levels atthat time (Fig. 4b). These results indi-cate that, similar to its inhibition ofmyogenesis (Spicer et al., 1996), theinhibitory effect of Twist1 on osteo-genic differentiation is mediated bythe T/E dimer.

Promotion of T/E FormationPrevents Suture Fusion inTwist1�/� Mice

Our data suggest that there is an in-crease in the ratio of T/T to T/E in thesutures of Twist1�/� mice thatdrives craniosynostosis. We, there-fore, asked whether we could preventsuture fusion in these mice by alteringE protein or Id levels to promote theformation of T/E dimers. To increase Eprotein expression, we infected the su-tures of Twist1�/� mice with an ad-enovirus expressing E2A E12. As in

Fig. 5. Twist1 requires heterodimerization to inhibit osteoblast differentiation. Primary calvariacells were infected with adenovirus expressing myc-Twist1 (T) and or myc-E2A E12 (E) as indicatedabove gels. Green fluorescent protein (GFP) -expressing adenovirus was used as a control whereno virus is indicated. a,b: Cells were placed in differentiation medium for the indicated times beforebeing harvested for (a) Western analysis for myc-Twist1 (T) and myc-E2A E12 (E) or (b) reversetranscriptase-polymerase chain reaction analysis for Runx2, �1(I) collagen (Col 1a1), bone sialo-protein (BSP), osteocalcin (OC), endogenous Twist1, and glyceraldehydes-3-phosphate dehydro-genase (GAPDH) transcripts.

1352 SPICER ET AL.

humans, Twist1 haploinsufficiency inmice promotes the fusion of primarilythe coronal suture; therefore, we fo-cused our analysis on this suture. Be-cause the coronal suture of live pups ishard to discern, we used explant cul-ture of the skull (Ogle, 2000). The su-tures of Twist1�/� mice do not fuseuntil 3–4 weeks after birth (Carver etal., 2002), so we altered the cultureconditions to try and maintain skullmorphology during 4–5 weeks in cul-ture. This was achieved by sectioningthe skull horizontally from the base ofthe nose to the occipital bone, leavingthe dura mater and brain intact (Fig.6). After 5 weeks in culture the skullsof wild-type mice looked remarkablynormal, with both the sagittal andcoronal sutures remaining patent(Fig. 6c). The coronal sutures ofTwist1�/� mice are patent and rela-tively normal at birth (Fig. 6b); how-ever, after 5 weeks in culture, distinctfusion of this suture was observed. Inthree of four uninjected Twist1�/�control explants and three of fourTwist1�/� explants injected in theright coronal suture with adenovirusexpressing green fluorescent protein(GFP), both the left and right coronalsutures had fused and the sagittal su-ture remained patent (Fig. 6d,e). How-ever, the right coronal suture failed tofuse in four of five Twist1�/� skullswhen injected with adenovirus ex-pressing E2A E12 at the time of ex-plant (Fig. 6f).

As another means to promote T/E for-mation, we lowered Id levels by cross-ing Twist1�/� mice with Id1�/�;Id3�/� mice. Id3 has a similarexpression pattern as Id1 in the cra-nial sutures (data not shown) and be-cause Id1�/�;Id3�/� mice are via-

ble (Lyden et al., 1999), we decided toremove from one to three alleles of Id1and Id3 genes on the Twist1�/�background. As can be seen in Table 1,there was a significant decrease in thepercentage of mice showing any coro-nal suture fusion as the number of Idalleles was removed. Approximately87% of Twist1�/� mice had cranio-synostosis 5 weeks after birth, andthis rate decreased to 0% when threeId alleles were removed. To achieve amore accurate assessment of the de-gree to which craniosynostosis wasrescued on the different genetic back-grounds, we used a craniosynostosisindex (CI) similar to one used to ana-lyze Twist1 and Snail genetic interac-tions (Oram and Gridley, 2005). Leftand right coronal sutures were as-sessed individually and assigned anumber between 0 (completely un-fused) to 3 (completely fused), and theCI given in Table 1 indicates the av-erage degree of craniosynostosis for acoronal suture on the indicated ge-netic background. As can be seen fromthe large standard deviation of the CIfor Twist1�/� mice, the degree of su-ture fusion was quite variable. Thisfinding was also true with Twist1�/�;Id1�/� mice; however, there was anoticeable decrease in the severity ofsuture fusion, but the difference wasnot significant. The removal of two Idalleles, either Twist1�/�;Id1�/� orTwist1�/�;Id1�/�;Id3�/�, how-ever, produced almost a full rescue,decreasing the CI from 1.64 forTwist1�/� to 0.07. We have only ob-tained 5 Twist1�/�;Id1�/�;Id3�/�mice so far, but all of these have com-pletely patent and normal looking su-tures. Therefore, conditions that pro-mote T/E formation, either by

increasing E proteins or decreasing Idlevels, result in the inhibition of su-ture fusion in Twist1�/� mice.

DISCUSSION

We and others have shown that Twisthas both positive and negative func-tions regulating mesenchymal cellspecification and differentiation (Cas-tanon and Baylies, 2002; O’Rourkeand Tam, 2002). The mechanisms un-derlying these disparate functions arestill unclear, but we suggest that theymay be dependent on the dimer part-ner of Twist. We previously found thatTwist1 heterodimers inhibited myo-genesis (Spicer et al., 1996); however,in Drosophila, Twist homodimers arethought to mediate mesoderm forma-tion (Castanon et al., 2001), indicatingthat Twist may be unique amongClass II bHLH proteins in formingfunctional heterodimers and ho-modimers. Therefore, we thought itessential to determine whether verte-brate Twist1 formed functional ho-modimers and if they had differentactivities from Twist1 heterodimers.We found that these two dimers doform and that the relative level ofTwist1 to Id determines which dimerforms. We have characterized the ac-tivities of T/T and T/E dimers andhave found that they differentiallyregulate cell behavior and gene ex-pression. Because the two dimers reg-ulate different sets of genes, it createsthe possibility for a more dynamic con-trol of gene expression by modulatingthe availability of E proteins fordimerization. This regulation can bequite complex as many different HLHproteins can compete with Twist1 fordimerization with E proteins. We

TABLE 1. Craniosynostosis in Progeny of the Twist11�/� X Id1�/�;Id3�/� Cross at 5 Weeksa

Genotype n CI � SEM Penetrance (%)

Wild-type 21 0.02 � 0.02 0.05Tw�/� 17 1.64 � 1.27 87.50Tw�/�; Id1�/� 29 0.85 � 1.46 75.86Tw�/�; Id1�/� or Tw �/�; Id1�/�; Id3�/� 21 0.07 � 0.07 4.75Tw �/�; Id1�/�; Id3�/� 5 0 0

aSkulls were assessed using a scoring system in which a suture was assigned a value from 0 to 3: 0 � completely unfused; 1 � 50%fused; 3 � 100% fused. Left and right coronal sutures were scored individually for each skull. For each genotype, the scoring mean �SEM was determined and was termed the craniosynostosis index (CI). Statistical analysis was performed by one-way analysis ofvariance. The difference between Tw�/� and Tw�/� mice, which lost two copies of Id, was highly significant (P � 10�15).Penetrance was calculated as the percentage of animals that exhibited any coronal suture fusion.


chose the cranial sutures to test ourhypothesis, because changes inTwist1 levels are known to affect thisregion and there are only a few HLHproteins that are known to be ex-pressed there.

On the basis of the expression pat-terns of HLH genes within the cranialsutures, we hypothesized that Twist1forms homodimers in the osteogenicfronts and Twist1/E protein het-erodimers in the mid-sutures. Wetested this hypothesis by determiningif genes that are differentially regu-lated by these dimers are expressed inthe predicted domains within the su-tures. Our results support this hy-pothesis, and importantly, we havefound that the Twist1 dimers differen-tially regulate the expression of medi-ators of the signaling pathways thatregulate suture patency (Nah, 2000;Wilkie et al., 2001). Twist1 ho-modimers induce the expression ofFGFR2, whereas T/E heterodimers in-duce the expression of TSP-1. Of inter-est, one of the functions of TSP-1 is toactivate latent transforming growthfactor-beta (TGF�; Annes et al., 2003).TGF� signaling plays an integral rolein the regulation of suture patency(Opperman, 2000); however, themechanism of latent TGF� activationin the sutures has not been investi-gated and TSP-1 may play this role.

Twist1 and Craniosynostosis

It has been paradoxical that the genefamilies that are associated with themajority of craniosynostosis syn-dromes (TWIST1 and FGFRs) areboth defined as inhibitors of in vitroosteoblast differentiation, yet cranio-synostosis occurs due to activation ofthe FGFRs and haploinsufficiency ofTWIST1 (Nah, 2000; Wilkie et al.,2001). The mechanisms underlyingcraniosynostosis are not well under-stood, and we propose a model thatexplains why TWIST1 haploinsuffi-ciency leads to craniosynostosis andlinks Twist1 expression with FGF sig-naling (Fig. 1a). Although Twist1 hasbeen shown to regulate FGFR expres-sion in Drosophila (Shishido et al.,1993), the relationship betweenTwist1 and FGFR in vertebrates hasbeen less clear. We have found therelationship to be quite complex, withT/T dimers inducing FGFR2 expres-

sion and T/E dimers inhibiting it. Thisantagonistic regulation between theTwist1 dimers may help set up a moredistinct boundary of gene expression.Due to increased competition by Id inTwist1�/� mice (Fig. 1b), this bound-ary is altered favoring T/T formationand FGFR2 expression is extendedinto the mid-sutural mesenchyme(Rice et al., 2000). Conversely, wefound a decrease in the T/E-regulatedgene TSP-1 in the mid-suture ofTwist1�/� mice (Fig. 4). Importantlywe were able to inhibit suture fusionin Twist1�/� mice by promoting T/Eformation by either increasing the ex-pression of E2A E12 or decreasing Idexpression (Fig. 6; Table 1).

Recently, Twist1 has been shown tointeract with and inhibit the bHLHprotein Hand2 (Firulli et al., 2005). Inaddition to craniosynostosis, Twist1haploinsufficiency is also associatedwith limb abnormalities. Of interest,crossing Twist1�/� mice withHand2�/� mice rescues these limbdefects but not craniosynostosis(Firulli et al., 2005), indicating thatthese two phenotypes may be medi-ated by different mechanisms. Consis-tent with this, we have not found thatthere is a significant rescue of theselimb defects in Twist1�/� mice withdiffering numbers of functional Id al-leles (data not shown).

Twist1 and OsteoblastDifferentiation

Twist1 has been shown recently to di-rectly inhibit the activity but not theexpression of the osteogenic transcrip-tion factor Runx2 (Bialek et al., 2004).Our data agree with this as we foundthat Twist1 inhibited the expressionof the later osteoblast differentiationmarkers BSP and OC without affect-ing the expression of Runx2. The inhi-bition of Runx2 by Twist1, however,was shown to only require the C-ter-minus and not the bHLH domain ofTwist1 (Bialek et al., 2004), indicatingthat Twist1 does not need to dimerizeor bind DNA to affect Runx2 function.Our data here do not support this con-clusion, as Twist1 required the pres-ence of E2A E12 to inhibit the differ-entiation of calvaria cells (Fig. 5).Whereas many TWIST1 mutationsfound in Saethre-Chotzen syndrome(SCS) patients result in the deletion of

the carboxy terminus, most of thesemutations also delete part or all of thebHLH domain. Furthermore, otherSCS mutations are point mutationswithin the bHLH region that disruptDNA binding and/or dimerization andleave the carboxy terminus intact(Gripp et al., 2000), suggesting thatthe transcriptional activity of Twist1is required for normal regulation ofsuture patency. We suggest that theregulation of osteoblast differentia-tion by Twist1 may be at least par-tially indirect through the regulationof other genes, such as FGFR2, peri-ostin, and TSP-1. Of interest, TSP-1has been shown to inhibit osteoblastdifferentiation (Canfield et al., 1996),whereas periostin enhances certainaspects of osteoblast recruitment anddifferentiation (Horiuchi et al., 1999;Litvin et al., 2004). Therefore, thetransition from T/E to T/T in the os-teogenic fronts may result in a switchfrom inhibition of differentiation to in-duction of at least the early phases ofosteoblast differentiation. It is stillunclear whether all of these genes aredirectly transactivated by Twist1 orwhat is the mechanism that differen-tiates transactivation by T/T vs. T/Edimmers; however, initial promoteranalysis with periostin (Oshima et al.,2002) and TSP-1 (data not shown)suggests that at least these two genesmay be directly regulated by Twist1.

Here, we have identified dimer part-ner selection as an important media-tor of Twist1 function. Small changesin the level of Twist1 expression alterthe ratio of T/T to T/E, which can havedramatic phenotypic consequencessuch as craniosynostosis. Twist1 playsimportant roles in other areas duringembryogenesis (Castanon and Bay-lies, 2002; O’Rourke and Tam, 2002)and has been implicated recently inpromoting tumor metastasis (Yang etal., 2004); we anticipate that the reg-ulation of Twist1 dimer formationplays critical roles in these processesas well.

EXPERIMENTALPROCEDURES

Plasmids and ViralConstructs

CS2-mTwist1 and CSA-E2A E12were described previously (Spicer et

1354 SPICER ET AL.

al., 1996). The tethered dimers wereconstructed as follows: For mTwist1,KpnI and EcoRI sites were gener-ated at the 3�-end removing the stopcodon. For E2A E12, an EcoRI sitewas generated in front of the initia-tion codon. A double-stranded oligo-nucleotide linker encoding thepolypeptide tether (GGT ACC GGCGGA GGC TCG AGC GGT GGG AGCGGC GGA GGG TCC GGC GGAGGC TCG AGC GGT GGG AGC GGCGGA GGG TCC GGC GAA TTC) wasinserted into KpnI–EcoRI-linearizedpCS2mycTwist1-E2A E12 or pCS2-mycTwist1-Twist1 vector. To con-struct the T, TT, and TE retrovirusvectors, the corresponding cDNA wereexcised and ligated into the pWZL-hygro vector (gift from Scott Lowe;Serrano et al., 1997). Twist1 and E2AE12 adenovirus were prepared by us-ing the Cre-lox recombination systemas described previously (Hardy et al.,1997) at a titer of 1012 viral parti-cles/ml.

Electrophoretic MobilityShift Assays

EMSA analysis was performed usingin vitro translated proteins and dou-ble-stranded oligos containing the Ebox-binding site from the muscle cre-atine kinase (MCK) enhancer as aprobe as previously described (Lassaret al., 1991).

Cell Culture and GeneTransfer

C3H10T1/2 cells (10T1/2) and 293THEK cells (ATCC) were cultured inDMEM supplemented with 10% fetalbovine serum and penicillin–strepto-mycin. The 293T HEK cells weretransfected with CS2-(myc)Twist1,CS2-(myc)Id1, and CSA-E2A E12 ex-pression constructs as previously de-scribed (Spicer et al., 1996). Retroviralsupernatants were prepared usingtransient transfection of pWZL (Ser-rano et al., 1997) Twist1 constructsinto the BOSC23 packaging cell lineand 10T1/2 cells were infected as pre-viously described (Novitch et al.,1999).

To prepare primary osteoblasts, cal-varia from P1 pups were cleaned fromassociated tissues and cut in half. Frag-ments were washed with 4 mM ethyl-

enediaminetetraacetic acid (EDTA)three times, followed by three washes inphosphate buffered saline, and then di-gested in collagenase solution (Worth-ington Collagenase Type 2 @ 317/mg), shaking for 10 min at 37°C.Digestions were repeated five timesand digestions 3, 4, and 5 were pooled.Cells were filtered through mesh toremove any pieces of bone and resus-pended in �-MEM 10% fetal calf se-rum (FCS). Calvaria cells were differ-entiated in the same mediumcontaining 10 mg/ml L-ascorbic acid,500 mM �-glycerol phosphate. For ad-enoviral transduction, cells were incu-bated in serum-free medium with 103

viral particles/cell in the presence ofpoly-D-lysine (Sigma) (5 � 103 mole-cules/viral particle) for 2 hr at 37°C,after which the medium was replacedwith �-MEM 10% FCS.

Western Blotting

Western blot analysis was performed aspreviously described (Leshem et al.,2000). Samples run in nonreducing con-ditions contained no �-mercaptoethanoland were not heated before loading.The following primary antibodieswere used: monoclonal anti-� actin (1:1,000 Sigma); monoclonal anti-TSP1(Ab-11; 1:750, NeoMarker); rabbitpolyclonal anti-Periostin (1:1,000; akind gift from Roger Markwald); andrabbit polyclonal anti-E2a (sc-349,1:1,000, Santa Cruz Biotechnology).

Immunostaining

10T1/2 cells were analyzed by immu-nofluorescence as previously de-scribed (Leshem et al., 2000) usingmonoclonal anti-TSP1 (Ab-1; 1:250,NeoMarker). Images were obtainedusing confocal fluorescence micros-copy. Paraffin sections of coronal andsagittal sutures of P1 pups wereprobed with anti-Twist1 rabbit poly-clonal Ab (Santa Cruz, H-81) and an-ti-Id rabbit polyclonal Ab (Santa Cruz,Z-8; note this antibody recognizes allfour Id proteins), followed by Tyra-mide signal amplification (TSA BiotinSystem, Perkin Elmer) and Vec-tastain Elite ABC kit (Vector Labora-tories).

Heparin–Sepharose AffinityChromatography

10T1/2 cell lines were plated at 5 �106 cells per 15-cm dish with 20 ml of0.2% fetal bovine serum (FBS) for 48hr. Affinity chromatography was per-formed using CL-6B Heparin–Sepha-rose columns (Amersham PharmaciaBiotech). The column was equilibratedwith 50 mM Tris-HCl buffer, pH 7.4,containing 10 mM EDTA. Conditionedmedium was applied to the column,and the bound material was elutedwith 1.5 M NaCl. Fractions were col-lected and concentrated using YM-10centricons (Millipore) then assayed forTSP1 by Western blot analysis.

Reverse Transcription-Polymerase Chain Reaction

Total cellular RNA was isolated from cul-tures using Tri-reagent (Sigma) asprevious described (Leshem et al., 2000).The following mouse primer sets (5�–3�)and annealing temperature were usedfor each particular complementaryDNA (cDNA) amplification: Id1 GGTG-GATCCACCATG AAGGTCGCCAGTG,TGGATCCGTCCATCTGGTCCCT-CAGTGC; Id3 AGGCGCTGAGCCCG-GTGC, CGGGAGGTGCCAGGACG(60°C); FGFR2 CTGTGCCGAATGA-AGAACACGACC; CAAAGTCTG-CTATCTTCATCAC; FGFR1 AAG-GACAAACCCAACCGTGTGACC;CAAAGTCTGCTATCTTCATCAC(55°C); GAPDH TGCGACTTCAA-CAGCAACTC; GATGGAAATTGT-GAGGGAGA (50°C).

Mice

Twist1�/� and Wnt1-Cre mice wereobtained from the Jackson Laboratory.A cre-responsive transgene CAGCAT-Twist1 was constructed by replacingthe lacZ region of CAG-CAT-Z (Araki etal., 1995) with the murine Twist1cDNA. This construct was used for mi-croinjection to establish a transgenicline. Neural crest-specific transgene ex-pression was achieved by crossing theCAGCAT-Twist1 mice with Wnt1-cremice. Id1�/�;Id3�/� mice (gift fromR. Benezra; Lyden et al., 1999) werecrossed with the Twist1�/� mice to ob-tain Twist1�/� mice with varyingnumbers of functional Id alleles.


In Situ Hybridization

Skulls from newborn mice (P1) werefixed in 4% paraformaldehyde, cryo-sectioned, and analyzed by in situ hy-bridization as in (Yoshida et al., 2005).In situ probes were against periostin(gift from C. Vary), TSP-1 (gift from P.Bornstein), and FGFR2 (gift fromC. Basilico) as indicated. For eachprobe, sections from wild-type andTwist1�/� skulls were processed inthe same container for the same pe-riod of time to be able to directly com-pare gene expression. Whole-mount insitu analysis was performed on P1skulls after removal of the skin andbrain. The in situ protocol was as de-scribed by Isaac et al. (2000) with themodification that the skulls were di-gested with proteinase K digestion for45 min.

Calvarial Explants

Heads of P1 pups were de-skinnedand sectioned horizontally below thenose. Calvaria were placed on Falconcell culture inserts (PET pore size0.4 m) in six-well dishes containingDMEM:Ham’s F-12K 50:50, 10 mg/mlL-ascorbic acid, 500 mM �-glycerolphosphate, 10% FBS, 1% penicillin–streptomycin. Explants were incu-bated in a CO2 incubator at 37°C withmedium changed daily.

ACKNOWLEDGMENTSWe thank the members of the Spicerlab and MMCRI for helpful commentsand discussions. D.B.S. was funded byNIH grants from the National Centerfor Research Resources and the Na-tional Institute of Dental and Cranio-facial Research and an AmericanHeart Association Scientist Develop-ment Grant.

REFERENCES

Annes JP, Munger JS, Rifkin DB. 2003.Making sense of latent TGFbeta activa-tion. J Cell Sci 116:217–224.

Araki K, Araki M, Miyazaki J, Vassalli P.1995. Site-specific recombination of atransgene in fertilized eggs by transientexpression of Cre recombinase. Proc NatlAcad Sci U S A 92:160–164.

Bakiri L, Matsuo K, Wisniewska M, Wag-ner EF, Yaniv M. 2002. Promoter speci-ficity and biological activity of tetheredAP-1 dimers. Mol Cell Biol 22:4952–4964.

Benezra R, Davis RL, Lockshon D, TurnerDL, Weintraub H. 1990. The protein Id:a negative regulator of helix-loop-helixDNA binding proteins. Cell 61:49–59.

Bialek P, Kern B, Yang X, Schrock M, SosicD, Hong N, Wu H, Yu K, Ornitz DM,Olson EN, Justice MJ, Karsenty G. 2004.A twist code determines the onset of os-teoblast differentiation. Dev Cell 6:423–435.

Bourgeois P, Bolcato-Bellemin AL, DanseJM, Bloch-Zupan A, Yoshiba K, StoetzelC, Perrin-Schmitt F. 1998. The variableexpressivity and incomplete penetranceof the twist-null heterozygous mousephenotype resemble those of humanSaethre-Chotzen syndrome. Hum MolGenet 7:945–957.

Canfield AE, Sutton AB, Hoyland JA,Schor AM. 1996. Association of throm-bospondin-1 with osteogenic differentia-tion of retinal pericytes in vitro. J CellSci 109(Pt 2):343–353.

Carver EA, Oram KF, Gridley T. 2002.Craniosynostosis in Twist heterozygousmice: a model for Saethre-Chotzen syn-drome. Anat Rec 268:90–92.

Castanon I, Baylies MK. 2002. A Twist infate: evolutionary comparison of Twiststructure and function. Gene 287:11–22.

Castanon I, Von Stetina S, Kass J, BayliesMK. 2001. Dimerization partners deter-mine the activity of the Twist bHLH pro-tein during Drosophila mesoderm devel-opment. Development 128:3145–3159.

el Ghouzzi V, Le Merrer M, Perrin-SchmittF, Lajeunie E, Benit P, Renier D, Bour-geois P, Bolcato-Bellemin AL, MunnichA, Bonaventure J. 1997. Mutations of theTWIST gene in the Saethre-Chotzen syn-drome. Nat Genet 15:42–46.

Firulli BA, Krawchuk D, Centonze VE,Vargesson N, Virshup DM, Conway SJ,Cserjesi P, Laufer E, Firulli AB. 2005.Altered Twist1 and Hand2 dimerizationis associated with Saethre-Chotzen syn-drome and limb abnormalities. NatGenet 37:373–381.

Funato N, Ohtani K, Ohyama K, Kuroda T,Nakamura M. 2001. Common regulationof growth arrest and differentiation ofosteoblasts by helix-loop-helix factors.Mol Cell Biol 21:7416–7428.

Gripp KW, Zackai EH, Stolle CA. 2000.Mutations in the human TWIST gene.Hum Mutat 15:479.

Guenou H, Kaabeche K, Mee SL, Marie PJ.2005. A role for fibroblast growth factorreceptor-2 in the altered osteoblast phe-notype induced by Twist haploinsuffi-ciency in the Saethre-Chotzen syndrome.Hum Mol Genet 14:1429–1439.

Hardy S, Kitamura M, Harris-Stansil T,Dai Y, Phipps ML. 1997. Construction ofadenovirus vectors through Cre-lox re-combination. J Virol 71:1842–1849.

Horiuchi K, Amizuka N, Takeshita S,Takamatsu H, Katsuura M, Ozawa H,Toyama Y, Bonewald LF, Kudo A. 1999.Identification and characterization of anovel protein, periostin, with restrictedexpression to periosteum and periodon-tal ligament and increased expression by

transforming growth factor beta. J BoneMiner Res 14:1239–1249.

Howard TD, Paznekas WA, Green ED,Chiang LC, Ma N, Ortiz de Luna RI,Garcia Delgado C, Gonzalez-Ramos M,Kline AD, Jabs EW. 1997. Mutations inTWIST, a basic helix-loop-helix tran-scription factor, in Saethre-Chotzen syn-drome. Nat Genet 15:36–41.

Isaac A, Cohn MJ, Ashby P, Ataliotis P,Spicer DB, Cooke J, Tickle C. 2000. FGFand genes encoding transcription factorsin early limb specification. Mech Dev 93:41–48.

Ishii M, Merrill AE, Chan YS, Gitelman I,Rice DP, Sucov HM, Maxson RE Jr.2003. Msx2 and Twist cooperatively con-trol the development of the neural crest-derived skeletogenic mesenchyme of themurine skull vault. Development 130:6131–6142.

Jiang X, Iseki S, Maxson RE, Sucov HM,Morriss-Kay GM. 2002. Tissue originsand interactions in the mammalian skullvault. Dev Biol 241:106–116.

Johnson D, Iseki S, Wilkie AO, Morriss-Kay GM. 2000. Expression patterns ofTwist and Fgfr1, -2 and -3 in the devel-oping mouse coronal suture suggest akey role for twist in suture initiation andbiogenesis. Mech Dev 91:341–345.

Lassar AB, Davis RD, Wright WE, KadeshT, Murre C, Voronova A, Baltimore D,Weintraub H. 1991. Functional Activityof Myogenic HLH Proteins Requires Het-ero-oligomerization with E12/E47-likeProteins in Vivo. Cell 66:305–315.

Lee MS, Lowe GN, Strong DD, WergedalJE, Glackin CA. 1999. TWIST, a basichelix-loop-helix transcription factor, canregulate the human osteogenic lineage.J Cell Biochem 75:566–577.

Leshem Y, Spicer DB, Gal-Levi R, HalevyO. 2000. Hepatocyte growth factor (HGF)inhibits skeletal muscle cell differentia-tion: a role for the bHLH protein twistand the cdk inhibitor p27. J Cell Physiol184:101–109.

Litvin J, Selim AH, Montgomery MO, Le-hmann K, Rico MC, Devlin H, BednarikDP, Safadi FF. 2004. Expression andfunction of periostin-isoforms in bone.J Cell Biochem 92:1044–1061.

Lyden D, Young A, Zagzag D, Yan W, Ger-ald W, O’Reilly R, Bader B, Hynes R,Zhuang Y, Manova K, Benezra R. 1999.Id1 and Id3 are required for neurogen-esis, angiogenesis and vascularization oftumour xenografts. Nature 401:670–677.

Massari ME, Murre C. 2000. Helix-loop-helix proteins: regulators of transcrip-tion in eucaryotic organisms. Mol CellBiol 20:429–440.

Nah H. 2000. Suture biology: lessons frommolecular genetics of craniosynostosissyndromes. Clin Orthod Res 3:37–45.

Neuhold LA, Wold B. 1993. HLH forceddimers: tethering MyoD to E47 gener-ates a dominant positive myogenic factorinsulated from negative regulation by Id.Cell 74:1033–1042.

Novitch BG, Spicer DB, Kim PS, CheungWL, Lassar AB. 1999. pRb is required for

1356 SPICER ET AL.

MEF2-dependent gene expression aswell as cell-cycle arrest during skeletalmuscle differentiation. Curr Biol 9:449–459.

Ogata T, Wozney JM, Benezra R, Noda M.1993. Bone morphogenetic protein 2transiently enhances expression of agene, Id (inhibitor of differentiation), en-coding a helix-loop-helix molecule in os-teoblast-like cells. Proc Natl Acad Sci US A 90:9219–9222.

Ogle RC. 2000. Craniofacial skeletal mor-phogenesis in vitro. Methods Mol Biol136:55–59.

Opperman LA. 2000. Cranial sutures asintramembranous bone growth sites.Dev Dyn 219:472–485.

Oram KF, Gridley T. 2005. Mutations insnail family genes enhance craniosynos-tosis of Twist1 haplo-insufficient mice:implications for Saethre-Chotzen Syn-drome. Genetics 170:971–974.

O’Rourke MP, Tam PP. 2002. Twist func-tions in mouse development. Int J DevBiol 46:401–413.

Oshima A, Tanabe H, Yan T, Lowe GN,Glackin CA, Kudo A. 2002. A novelmechanism for the regulation of osteo-blast differentiation: transcription ofperiostin, a member of the fasciclin Ifamily, is regulated by the bHLH tran-scription factor, twist. J Cell Biochem86:792–804.

Rice DP, Aberg T, Chan Y, Tang Z, Ket-tunen PJ, Pakarinen L, Maxson RE,Thesleff I. 2000. Integration of FGF and

TWIST in calvarial bone and suture de-velopment. Development 127:1845–1855.

Rice R, Rice DP, Thesleff I. 2005. Foxc1integrates Fgf and Bmp signalling inde-pendently of twist or noggin during cal-varial bone development. Dev Dyn 233:847–852.

Serrano M, Lin AW, McCurrach ME,Beach D, Lowe SW. 1997. Oncogenic rasprovokes premature cell senescence as-sociated with accumulation of p53 andp16INK4a. Cell 88:593–602.

Shishido E, Higashijima S, Emori Y, SaigoK. 1993. Two FGF-receptor homologuesof Drosophila: one is expressed in meso-dermal primordium in early embryos.Development 117:751–761.

Sigvardsson M, O’Riordan M, GrosschedlR. 1997. EBF and E47 collaborate to in-duce expression of the endogenous im-munoglobulin surrogate light chaingenes. Immunity 7:25–36.

Spicer DB, Rhee J, Cheung WL, LassarAB. 1996. Inhibition of myogenic bHLHand MEF2 transcription factors by thebHLH protein Twist. Science 272:1476–1480.

Stankiewicz P, Thiele H, Baldermann C,Kruger A, Giannakudis I, Dorr S,Werner N, Kunz J, Rappold GA, Hans-mann I. 2001. Phenotypic findings due totrisomy 7p15.3-pter including theTWIST locus. Am J Med Genet 103:56–62.

Vinals F, Ventura F. 2004. Myogenin pro-tein stability is decreased by BMP-2through a mechanism implicating Id1.J Biol Chem 279:45766–45772.

Vinals F, Reiriz J, Ambrosio S, Bartrons R,Rosa JL, Ventura F. 2004. BMP-2 de-creases Mash1 stability by increasingId1 expression. EMBO J 23:3527–3537.

Volpert OV, Pili R, Sikder HA, Nelius T,Zaichuk T, Morris C, Shiflett CB, DevlinMK, Conant K, Alani RM. 2002. Id1 reg-ulates angiogenesis through transcrip-tional repression of thrombospondin-1.Cancer Cell 2:473–483.

Wilkie AO, Oldridge M, Tang Z, MaxsonRE Jr. 2001. Craniosynostosis and re-lated limb anomalies. Novartis FoundSymp 232:122–133; discussion 133–143.

Yang J, Mani SA, Donaher JL, Ra-maswamy S, Itzykson RA, Come C, Sav-agner P, Gitelman I, Richardson A,Weinberg RA. 2004. Twist, a master reg-ulator of morphogenesis, plays an essen-tial role in tumor metastasis. Cell 117:927–939.

Yoshida T, Phylactou LA, Uney JB, Ish-ikawa I, Eto K, Iseki S. 2005. Twist isrequired for establishment of the mousecoronal suture. J Anat 206:437–444.

Yousfi M, Lasmoles F, Lomri A, DelannoyP, Marie PJ. 2001. Increased bone forma-tion and decreased osteocalcin expres-sion induced by reduced Twist dosage inSaethre-Chotzen syndrome. J Clin In-vest 107:1153–1161.


twist1 dimer selection regulates cranial suture patterning...

Documents