Archives of Oral Biology (2005) 50, 141—145

www.intl.elsevierhealth.com/journals/arob

Studies on Pax9—Msx1 protein interactions

Takuya Ogawaa, Hitesh Kapadiaa, Bailiang Wangb, Rena N. D’Souzaa,*

aDepartment of Orthodontics, University of Texas Health Science Center at Houston, Dental Branch,Room 346, 6516 MD Anderson Boulevard, Houston, TX 77030, USAbDepartment of Breast Medical Oncology, U.T.M.D. Anderson Cancer Center, 1515 Holcombe Boulevard,Houston, TX 77030, USA

Accepted 22 September 2004

KEYWORDSPax9;Msx1;Protein—protein

interaction;Tooth agenesis;Co-immuno-

precipitation;GST interaction assay

Abbreviations: GST, glutathione S-tgalovirus

* Corresponding author. Tel.: +1 713fax: +1 713 500 4123.

E-mail address: rena.n.dsouza@uth

0003–9969/$ — see front matter # 20doi:10.1016/j.archoralbio.2004.09.011

Summary Pax9 belongs to the Pax family of transcriptional regulators that aredefined by a highly conserved DNA-binding region, the paired domain. Drosophila,mouse and human genetics have shown that Pax proteins play multiple roles in tissuepatterning and organogenesis by mediating their functions in a highly tissue-specificmanner. Members of the Pax family, Pax9 and Pax1, act synergistically duringvertebral formation. However, only Pax9 is essential for tooth formation. Further-more, mutations of PAX9 are associated with human tooth agenesis. The highly tooth-specific molecular functions of Pax9 suggest that its activity is tightly regulated. Mostlikely, this occurs through interactions with other protein factors. Among the reg-ulatory molecules that are expressed in dental mesenchyme, the Msx1 homeoproteinis of particular interest. The closely overlapping expression patterns of Pax9 and Msx1are consistent with a role in epithelial—mesenchymal interactions. To demonstratethat Pax9 interacts with Msx1 physiologically in vivo and in vitro, we performed co-immunoprecipitation and GST interaction assays. Our results indicate that there is aphysical association between the two proteins. Our biochemical data, coupled withhuman genetic studies and expression analysis in amousemodel, indicate a functionalrelationship between Pax9 and Msx1 during tooth development.# 2004 Elsevier Ltd. All rights reserved.

Introduction

Tooth agenesis or the developmental absence ofteeth is a common genetic disorder that involvesthe patterning of the human dentition. Our current

ransferase; CMV, cytome-

500 4218;

.tmc.edu (R.N. D’Souza).

04 Elsevier Ltd. All rights rese

understanding of the molecular mechanisms under-lying odontogenesis implicates a number of tran-scription factors and signaling molecules (i.e.growth factors and their receptors). Several linesof evidence indicate that synergistic and antagonis-tic interactions of signaling molecules are recur-sively utilized in tooth development. This leads tothe local activation or inhibition of transcriptionfactors in tooth epithelium and mesenchyme.1

Msx1 and Pax9 are among the best studied toothmesenchymal transcription factors that appear to

rved.

142 T. Ogawa et al.

have master regulatory functions in the early phasesof odontogenesis.1—3 Their parallel expression pat-terns are consistent with a role in epithelial—mesenchymal interactions. Pax9 (�/�) and Msx1(�/�) tooth organs each arrest at the bud stage.While Msx1 is reduced in Pax9 null mutants, Pax9expression is normal in Msx1 (�/�) mice implyingthat Pax9 is upstream to Msx1.4—6 While this issuggestive of a molecular relationship betweenthe two transcription factors, the precise mechan-ism remains unclear.

Human genetic studies of tooth agenesis furtherunderscore the importance of understanding possi-ble molecular interactions between Pax9 and Msx1.We and other laboratories have shown that novelmutations in human PAX9 and MSX1 each result inposterior tooth agenesis.7—17 While PAX9 mutationsare dominantly associated with molar agenesis,MSX1 mutations mainly affect premolars. Sufficientphenotypic variability exists in families with muta-tions in either gene to suggest modifying effects.Independent association studies have provided evi-dence of a statistically significant interactionbetween PAX9 and MSX1.18 These data add to themounting evidence from in vivo mouse and humangenetic studies that suggests a functional relation-ship between Pax9 and Msx1 in tooth development.

Based on recent advances in our understanding ofthe molecular events during tooth development, wehypothesize that Pax9 interacts with Msx1 at thepost-transcriptional level during tooth develop-ment. By clarifying the relationship between Pax9and Msx1 on the protein level, these studies willadvance our understanding of how transcriptionfactors work together to mediate changes duringtooth morphogenesis. In this report, we demon-strate that the Pax9 paired domain protein formsdimeric complexes with Msx1 homeoprotein in vivoand in vitro. Based on these biochemical data, wepropose that such knowledge is critical to our under-standing of how aberrant gene functions contributeto the pathogenesis of tooth agenesis, the mostcommon defect in the patterning of the humandentition.

Materials and methods

Plasmid constructs

The mammalian expression vector pCMV—Pax9 withcMyc epitope tag and GST-fused expression plasmidpGEX—Pax9 are described elsewhere.19 A murineMsx1 cDNA clone comprising the full-length codingsequence region was kindly provided by Dr. JohnRubenstein (University of California at San Fran-

cisco). Expression plasmids containing the CMV pro-moter linked to the full-coding sequence of Msx1was constructed in pCMV—Tag2b (Stratagene, CA).The Flag epitope is in frame with the N-terminus ofMsx1. To construct GST-fused expression plasmidpGEX—Msx1, we subcloned full-length Msx1 cDNAinto BamHI—SalI sites of the GST-fused expressionvector (Amersham Pharmacia Biotech, NJ). To con-struct Pax9 and Msx1, the full-length Pax9 and Msx1cDNAs were cloned into BamHI—SalI sites of theplasmids pGEM3Zf(+) (Promega, WI).

Protein preparations

Proteins were produced by in vitro transcription—translation using the TNT Quick coupled transcrip-tion/translation system (Promega, WI). Recombi-nant proteins were GST fusion proteins.Production of GST fusion proteins was describedpreviously.19 As noted previously, Escherichia coliBL21 competent cells were transformed with therecombinant plasmid and used to inoculate LBmedia. Cells were grown at 37 8C and proteinexpression induced by the addition of IPTG. Cellswere harvested by centrifugation and protein pur-ification performed on a glutathione column.

Co-immunoprecipitation

COS7 cells were maintained as described pre-viously19 and cotransfected with cMyc—Pax9 andeither FLAG—Msx1 using Fugene6 (Roche MolecularBiochemicals, IN) as provided by the manufacturer.After 24 h, cells were resuspended in lysis buffer(50 mM Tris, pH 8.0, 400 mM NaCl, 1% Triton-X-100)with protease inhibitor cocktail tablets (Roche) andincubated for 30 min on ice. Cell lysates were addedto 20 ml of anti-Flag M2 affinity gel (Sigma, MO) androtated at 4 8C overnight. The affinity gels werewashed with the lysis buffer five times and elutedwith Laemmli buffer, and analyzed by western blot-ting with 1:200 dilution of goat anti-Pax9 polyclonalantibody (Santa Cruz, CA) or 1:1000 dilution mouseanti-Flag M2 monoclonal antibody (Sigma—Aldrich,MO). The signals were detected by incubating themembranes with ECL reagents (Amersham, NJ)according to instructions from the manufacturer.

GST interaction assay

The GST interaction assay was performed by incu-bating approximately 0.1 mg of GST—Pax9 proteinbound to glutathione—sepharose resin with about100 ml of the 35S radiolabeled Msx1 and viceversa. Binding reactions were maintained at 25 8Cfor 10 min glutathione—sepharose resin-bound

Studies on Pax9—Msx1 protein interactions 143

Figure 1 Pax9 interacts with Msx1 in vivo. COS7 cellswere transfected with Myc—Pax9 and either FLAG—Msx1or FLAG tag expression vector alone. After 24 h, total celllysates were analysed before co-immunoprecipitation toverify expression of Myc—Pax9 and FLAG—Msx1 (lanes 3and 4). After co-immunoprecipitation with antibody to theFLAG tag, Pax9 was found only in the presence of FLAG—Msx1 (lanes 1 and 2).

Figure 2 Pax9 and Msx1 proteins interact in vitro. GSTinteraction assays were performed using GST or the indi-cated GST-fusion protein and the indicated 35S radiola-belled Pax9 (A) or Msx1 (B) proteins. Immobilized proteins(indicated by arrows) were resolved by SDS-PAGE andvisualized by autoradiography.

Pax9-protein complexes were washed four timeswith 400 ml of washing buffer for 10 min at 4 8C toremove nonspecifically bound protein complexes.Pax9-protein complexes were eluted from glu-tathione—sepharose resin with 25 ml of 0.9 M glu-tathione, pH 9.6 at 4 8C for 30 min. The elutes werecombinedwith SDS-PAGE sample buffer and resolvedon a 4% SDS-PAGE.

Results

Co-immunoprecipitation

To assess the formation of Pax9—Msx1 complex invivo, we transiently expressed full-length mousePax9 together with mouse full-length Msx1 inCOS7. Western blot analysis of immunoprecipitationshowed that anti-FLAG antibody was able to pre-cipitate Pax9 protein only in the presence of FLAG—Msx1. As a control, the anti-FLAG antibody did notprecipitate Pax9 in the absence of FLAG—Msx1(Fig. 1). These experiments suggest that Pax9 andMsx1 are physically associated in vivo.

GST interaction assay

To examine the direct interaction between Pax9 andMsx1 in vitro, we performed the GST interactionassay using recombinant GST—Pax9 or GST—Msx1and complementary 35S radiolabeled Msx1 or Pax9proteins obtained by in vitro translation. Theseinteraction assays showed that 35S radiolabeledMsx1 interacted specifically with GST—Pax9 and,conversely, that 35S radiolabeled Pax9 interactedwith GST—Msx1 (Fig. 2).

Discussion

These findings provide new insights into themechanisms regulating epithelial—mesenchymalinteractions during the early phases of odontogen-esis. We demonstrate that Pax9 is able to form aprotein complex with Msx1, both in vivo and in vitro.Based on our results and the cumulative data frommouse and human studies, we propose that Pax9 andMsx1 participate in key regulatory events throughprotein—protein interactions in the early phases ofodontogenesis.

The physiological significance of the molecularinteractions between Pax9 and Msx1 are furtherunderscored by human genetic studies of toothagenesis. We and others have shown that novelmutations in human PAX9 and MSX1 each result inposterior tooth agenesis. While PAX9 mutations aredominantly associated with molar agenesis, MSX1mutations mainly affect premolars. Sufficient phe-notypic variability exists in families with mutationsin either gene to suggest modifying effects. Ourrecent analysis of two unrelated families with toothagenesis due to novel mutations in PAX9 revealedthe presence of a non-synonymous SNP (C101G) inMSX1 (unpublished data). It is reasonable tohypothesize that the combinatorial effect of twoaltered alleles, i.e. PAX9 and MSX1, may influencethe phenotype of tooth agenesis.

The identification of protein—protein interac-tions between Pax9 and Msx1 also offers an explana-tion of the early events underlying odontogenesis.Since the capacity to direct tooth developmentshifts from the oral epithelium to the dentalmesenchyme at E11.5, coinciding with a shift ofBmp4 expression to the mesenchyme, it is likelythat the interactions between these two partnertranscription factors involves Bmp4. There is cur-rently little known about the mechanisms that reg-ulate the expression of Bmp4 at the initiation stage.

144 T. Ogawa et al.

However, its regulation at the bud stage is beginningto be understood. At this stage, Pax9 and Msx1 arecoexpressed in the mesenchyme in which the func-tion of both genes is required for the expression ofBmp4.4,6 Interestingly, after E11.5, Pax9 expressionnot only has become independent of activatingepithelial signals, but BMPs are also no longer ableto inhibit Pax9 expression in explants of the man-dibular arch mesenchyme. This suggests that theswitch of the odontogenic potential from theepithelium to the mesenchyme involves a switchof the hierarchy between Bmp4 and Pax9. On theother hand, experimental evidence indicates thatthe mesenchymal expressions of Bmp4 and Msx1 atthe bud stage act in a positive feedback loop. Theapplication of recombinant BMP4 to Msx1(�/�)mesenchyme can restore the expression of theHMG-box containing gene Lef1, whereas in thoseexperiments Bmp4 expression was not induced.Moreover, an Msx1-deficient tooth bud undergoesepithelial differentiation from the bud stage to theearly cap stage by application of exogenous BMP4.6

It is likely then that the functions of Pax9 and Msx1define a mesenchymal checkpoint of the crosstalkbetween the dental tissues and are essential for theestablishment of the odontogenic potential of themesenchyme.1 These data suggest that a key func-tion of Pax9 and Msx1 is the maintenance ofmesenchymal Bmp4 expression.

Recent studies have shown that Msx proteinsmay repress transcription through interactionswith other homeoproteins. Heterodimer formationbetween Msx1 and other homeoprotein containingproteins like Dlx2, Lhx2 and Pax3, which are them-selves transcriptional activators, forms transcrip-tionally inactive complexes that cannot bind toDNA.20—22 Interestingly, the protein complex formedby Msx1 and Pax3 inhibits DNA binding by Pax3through the paired domain of Pax3 and the home-odomain of Msx1.22 Thus, it can be inferred from ourresults that Pax9 proteins may bind Msx1 proteinsthrough the same interacting domains of Pax3and Msx1. However, further studies are requiredto confirm the interacting protein domains and todetermine if this interaction has any functionalsignificance for the activation/repression of down-stream target genes.

Despite the significant progress made throughmouse and human genetics, our knowledge aboutthe functions of Pax9 remains unclear. By clarifyingthe relationship between Pax9 and Msx1 on thepost transcriptional level, these studies willadvance our understanding of how transcriptionfactors work together to mediate changes duringtooth morphogenesis. Furthermore, unifying bio-chemical approaches with human genetics will offer

valuable insights into how PAX9 mutations contri-bute to the pathogenesis of tooth agenesis and willprovide new and compelling hypotheses to test therole of genetic modifiers that contribute to thetooth agenesis phenotype.

Acknowledgements

We thank Ms. Adriana Cavender for her helpfultechnical assistance. This work was supported inparts by grants from the National Institute for Dentaland Craniofacial Research, NIH (DE011663) to R.D.S.and NIH K08-DE14237 to H.K.

References

1. Peters H, Balling R. Teeth. Where and how to make them.Trends Genet 1999;15:59—65.

2. Thesleff I. Developmental biology and building a tooth.Quintessence Int 2003;34:613—20.

3. Jernvall J, Thesleff I. Reiterative signaling and patterningduring mammalian tooth morphogenesis. Mech Dev 2000;92:19—29.

4. Peters H, Neubuser A, Kratochwil K, Balling R. Pax9-deficientmice lack pharyngeal pouch derivatives and teeth and exhibitcraniofacial and limb abnormalities. Genes Dev 1998;12:2735—47.

5. Satokata I, Maas R. Msx1 deficient mice exhibited cleft palateand abnormalities of craniofacial and tooth development.Nat Genet 1994;6:348—55.

6. Chen Y, Bei M, Woo I, Satokata I, Maas R. Msx1 controlsinductive signaling in mammalian tooth morphogenesis.Development 1996;122:3035—44.

7. Stockton DW, Das P, Goldenberg M, D’Souza RN, Patel PI.Mutation of PAX9 is associated with oligodontia. Nat Genet2000;24:18—9.

8. Goldenberg M, Das P, Messersmith M, Stockton DW, Patel PI,D’Souza RN. Clinical, radiographic, and genetic evaluation ofa novel form of autosomal-dominant oligodontia. J Dent Res2000;79:1469—75.

9. Nieminen P, Arte S, Tanner D, Paulin L, Alaluusua S, Thesleff I,et al. Identification of a nonsense mutation in the PAX9 genein molar oligodontia. Eur J Hum Genet 2001;9:743—6.

10. Frazier-Bowers SA, Guo DC, Cavender A, Xue L, Evans B, KingT, et al. A novel mutation in human PAX9 causes molaroligodontia. J Dent Res 2002;81:129—33.

11. Das P, Stockton DW, Bauer C, Shaffer LG, D’Souza RN, WrightT, et al. Haploinsufficiency of PAX9 is associated with auto-somal dominant hypodontia. Hum Genet 2002;110:371—6.

12. Das P, Hai M, Elcock C, Leal SM, Brown DT, Brook AH, et al.Novel missense mutations and a 288-bp exonic insertion inPAX9 in families with autosomal dominant hypodontia. Am JMed Genet 2003;118A:35—42.

13. Mostowska A, Kobielak A, Biedziak B, Trzeciak WH. Novelmutation in the paired box sequence of PAX9 gene in asporadic form of oligodontia. Eur J Oral Sci 2003;11:272—6.

14. Vastardis H, Karimbux N, Guthua SW, Seidman JG, SeidmanCE. A human MSX1 homeodomain missense mutation causesselective tooth agenesis. Nat Genet 1996;13:417—21.

Studies on Pax9—Msx1 protein interactions 145

15. Van den Boogaard MJ, Dorland M, Beemer FA, van Amstel HK.MSX1 mutation is associated with orofacial clefting and toothagenesis in humans. Nat Genet 2000;24:342—3.

16. Jumlongras D, Bei M, Stimson JM, Wang WF, DePalma SR,Seidman CE, et al. A nonsense mutation in MSX1 causesWitkop syndrome. Am J Hum Genet 2001;69:67—74.

17. Lidral AC, Reising BC. The role of MSX1 in human toothagenesis. J Dent Res 2002;81:274—8.

18. Vieira AR, Meira R, Modesto A, Murray JC. MSX1, PAX9 andTGFA contribute to tooth agenesis in South America. J DentRes 2004;83:723—7.

19. Mensah JK, Ogawa T, Kapadia H, Cavender AC, D’Souza RN.Functional analysis of a mutation in PAX9 associated with

familial tooth agenesis in humans. J Biol Chem 2004;279:5924—33.

20. Zhang H, Hu G, Wang H, Sciavolino P, Iler N, Shen MM,et al. Heterodimerization of Msx and Dlx homeoproteinsresults in functional antagonism. Mol Cell Biol 1997;17:2920—32.

21. Bendall AJ, Rincon-Limas D, Botas J, Abate-Shen C. Proteincomplex formation between Msx1 and Lhx2 homeoproteins isincompatible with DNA binding activity. Differentiation1998;63:151—7.

22. Bendall AJ, Ding J, Hu G, Shen MM, Abate-Shen C. Msx1antagonizes the myogenic activity of Pax3 in migrating limbmuscle precursors. Development 1999;126:4965—76.