The molecular basis of talin2's high affinity toward β1-integrin

1 Scien i ic RepoR s | 7:41989 | DOI: 10.1038/s ep41989 www.na u e.com/scien i ic epo s The molecula basis o alin2’s high a ini y owa d β1-in eg in Yaxia Yuan1,*, Liqing Li2,*, Yanyan Zhu1,*, Lei Qi2, La i eh Azizi3, Vesa P. Hy önen3, Chang-Guo Zhan1 & Cai Huang2,4 Talin in e ac s wi h β-in eg in ails and ac in o con ol in eg in ac i a ion, hus egula ing ocal adhesion dynamics and cell mig a ion. The e a e wo alin genes, Tln1 and Tln2, which encode alin1 and alin2, and i is gene ally belie ed ha alin2 unc ions edundan ly wi h alin1. Howe e , we show he e ha alin2 has a highe a ini y o β1-in eg in ails han alin1. Mu a ion o alin2 S339 o leucine, which can cause Fi h Finge Camp odac yly, a human gene ic disease, comple ely dis up ed i s binding o β–in eg in ails. Also, subs i u ion o alin1 C336 wi h Se enhanced he a ini y o alin1, whe eas subs i u ion o alin2 S339 wi h Cys diminished ha o alin2. Fu he compu a ional modeling analysis shows ha alin2 S339 o med a hyd ogen bond wi h E353, which is c i ical o inducing key hyd ogen bonds be ween alin2 N326 and β1-in eg in R760, and be ween alin2 K327 and β1-in eg in D759. Mu a ion a any o hese esidues signi ican ly diminished he in e ac ion o alin2 wi h β1- in eg in ails. These hyd ogen bonds we e no obse ed in alin1/β1-in eg in, bu did exis in alin1C336S/β1- in eg in complex. These esul s sugges ha alin2 S339 o ms a hyd ogen bond wi h E353 o media e i s high a ini y o β1-in eg in. In eg ins a e a amily o ansmemb ane adhesion ecep o s ha media e cell-ma ix and cell-cell adhesion1. In eg ins a e he e odime s, comp ising α (alpha) and β (be a) subuni s. Talin, a la ge ocal adhesion p o ein, binds o he be a subuni , consequen ly ac i a ing in eg in, which in u n egula es cell mig a ion2,3, in asion4,5, g ow h6,7, di e en ia ion6,8, and apop osis9. As a esul , in eg in ac i a ion modula es a a ie y o physiologi- cal and pa hological p ocesses, such as de elopmen , immuni y, in lamma ion, and umo me as asis. Thus, he alin-in eg in in e ac ion is one o he mos impo an p o ein-p o ein in e ac ions. Talin con ains an amino- e minal globula head domain and a ca boxy- e minal od domain10. The alin head domain con ains a FERM (band ou -poin -one, ez in, adixin, moesin homology) domain, which com- p ises h ee subdomains, F1, F2 and F3. The F2 domain is en i ely α -helical wi h a sho linked egion, and he F3 domain is a sandwich o wo o hogonal an ipa allel β -shee s ollowed by an α -helix11. The FERM domain is esponsible o he binding o alin o β -in eg in ails12, ype I phospha idylinosi ol 4-phospha e 5-kinase γ (PIPKIγ )13,14, and ocal adhesion kinase (FAK)15. The od domain has se e al inculin-binding si es, and wo ac in-binding si es16. The binding o alin o β in eg in ails is essen ial o in eg in ac i a ion12,17, which in u n egula es ocal adhesion (FA) dynamics, a key s ep in cell mig a ion18,19. Talin also media es calpain-induced FA disassembly19. We and ou collabo a o s ha e shown ha alin phospho yla ion by Cdk5 egula es FA dynamics, cell mig a ion and in asion20,21. Talin in e ac s wi h PIPKIγ , which p oduces PIP2 o egula e FA dynamics, cell mig a ion and in asion13,14,22–28. The e a e wo alin genes, Tln1 and Tln2, encoding alin1 and alin2, espec i ely. Talin1 has been well s udied, while he biological unc ion o alin2 is less clea . I was p esumed ha alin2 unc ions edundan ly wi h alin1. Howe e , ecen e idence indica es ha alin2 is unc ionally di e en om alin1. P e ious s udy shows ha alin2 egula es ocal adhesion assembly and ocal adhesion kinase (FAK) signaling in he absence o alin129. Talin2 is usually localized a la ge FAs and ib illa adhesions, whe eas alin1 is usually ound a smalle FAs in he pe iphe al egion30,31. T as uzumab, a HER2- a ge ing an ibody d ug o cance he apy, inhibi s cell mig a ion and in asion, mos likely h ough down- egula ing alin232. Recen ly, we epo ed ha alin2 binds o β in eg in 1Molecula Modeling and Biopha maceu ical Cen e , College o Pha macy, Uni e si y o Ken ucky, Lexing on, KY 40506, USA. 2Ma key Cance Cen e , Uni e si y o Ken ucky, Lexing on, KY 40506, USA. 3BioMediTech, Uni e si y o Tampe e, 33520 Tampe e, Finland and Fimlab Labo a o ies, 33520 Tampe e, Finland. 4Depa men o Pha macology and nu i ional sciences, Uni e si y o Ken ucky, Lexing on, KY 40506, USA. *These au ho s con ibu ed equally o his wo k. Co espondence and eques s o ma e ials should be add essed o C.G.Z. (email: [email protected]) o C.H. (email: [email protected]) ecei ed: 27 Oc obe 2016 Accep ed: 04 Janua y 2017 Published: 03 Feb ua y 2017 OPEN www.na u e.com/scien i ic epo s/ 2 Scien i ic RepoR s | 7:41989 | DOI: 10.1038/s ep41989 ails mo e s ongly han alin1, and a s ong in e ac ion o alin2 wi h β in eg ins is equi ed o he gene a ion o ac ion o ce, which in u n d i es in adopodium o ma ion and cell in asion33. Ne e heless, he unde lying molecula basis o he di e ence be ween alin1 and alin2 emains o be elucida ed. In he p esen s udy, we demons a ed ha alin2 had a highe a ini y o β 1-in eg in ails han alin1, alin2 Se 339 is la gely esponsible o his a ini y di e ence, and mu a ion a Se 339 educed i s binding o β 1-in eg in ails. Coinciden ly, a new s udy shows ha Fi h Finge Camp odac yly, a human gene ic disease, is caused by a Leucine mu a ion a alin2 Se 33934. We de ined he molecula basis o alin2 binding o β 1-in eg in using com- pu a ional modeling me hods, pe o med expe imen s o e i y he compu a ional model, and examined he ole o alin2 S339 in ocal adhesion assembly. Resul s Talin2 has a highe a ini y o β1-in eg in ails han alin1. Recen ly, we demons a ed ha alin2 had a s onge binding o β -in eg in ails han alin133. The e a e se e al esidues ha a e di e en be ween he in eg in-binding sequences o alin1 and alin2 (Fig.1A), bu Cys336 o alin1 and he co esponding Se 339 on alin2 a e la gely esponsible o hei binding di e ence. Subs i u ion o Cys336 wi h Se enhanced alin1’s bind- ing o β 1A-in eg in ails, whe eas subs i u ion o Se 339 wi h Cys o Leu signi ican ly a enua ed alin2 binding, sugges ing he c i ical ole o Se 339 in media ing alin2 binding o β -in eg ins. To s udy he in e ac ion o alin and mu an s wi h β 1-in eg in ails, we pu i ied His agged alin11–446WT, alin11–446C336S, alin21–449WT, and alin21–449S339C o e exp essed in E. coli. To examine he quali y o he p o eins, pu i ied p o eins we e analyzed by gel il a ion ch oma og aphy combined wi h s a ic ligh sca e ing de ec o . As shown in Supplemen a yFig.S1 and TableS1, all p o eins we e mainly monome ic in solu ion (judged by s a ic ligh sca e ing), wi h molecula weigh s app oxima ely 59 kDa, which is consis en wi h heo e ical molecula weigh s (~53 kDa). Also, he elu ed p o ein peak was a he symme ic in he case o all he p o ein o ms, sug- ges ing ha hese p o eins a e soluble and olded. To de e mine he a ini y o alin1 o β 1-in eg in ails, di e en concen a ions o His- agged- alin11–446 we e incuba ed wi h glu a hione aga ose beads ha we e p eloaded wi h GST-β 1A-in eg in ails. Unbound p o ein was washed away and bound p o ein was sepa a ed by SDS-PAGE, s ained wi h Coomassie blue, and quan i a ed wi h s anda ds un on he same gel. The β 1A-in eg in ails bound o alin11–446 mode a ely, bu GST did no (Fig.1B). The dissocia ion cons an (Kd) be ween alin11–446 and β 1A-in eg in ails is app ox. 0.88 ± 0.07 μ M (n = 5 ) (Fig.1C,D). We employed he same me hod o de e mine he Kd be ween alin21–449 and β 1 A in eg in ails. The Kd be ween alin21–449 and β 1A-in eg in ails is app ox. 0.35 ± 0.09 μ M (n = 5) (Fig.1B–1D). Subs i u ion o alin1 C336 wi h Se caused a sligh inc ease in i s a ini y, whe eas subs i u ion o alin2 S339 wi h Cys diminished i s a ini y (Fig.1C,D). Su p isingly, subs i u ion o alin2 S339 wi h Leu comple ely abolished i s in e ac ion wi h β 1 A in eg in ails (Fig.1E). These esul s indica e ha S339 is c i ical o alin2’s high a ini y o β in eg ins. Talin2 o ms close con ac s and mo e hyd ogen bonds wi h β1 in eg in ails han alin1. Fi s , we compa ed he ene gy minimiza ion and molecula dynamics (MD) in alin2WT/in eg in, alin1WT/in eg in, and alin1C336S/in eg in complexes. Depic ed in Fig.2 a e he s uc u al dynamics acked as posi ional oo -mean squa e de ia ion (RMSD) o Cα a oms along he 40 ns MD simula ions in he h ee alin/in eg in complex s uc- u es. A e a ime pe iod o ~10 ns, all he RMSD cu es become la , indica ing he complex s uc u e in he h ee MD sys ems a e elaxed and equilib a ed. The sequences o mouse alin2 (mTalin2), human alin2 (hTalin2), human alin1 (hTalin1) and human β 1D-in eg in ail a e shown in Fig.3A. To unde s and he s uc u al basis o alin1 and alin2 binding o β 1-in eg in ail, we pe o med molecula modeling and MD simula ions o explo e he s uc u al di e ences in alin1 and alin2. B ie ly, he alin1/in eg in ail complex s uc u e is cons uc ed based on alin2/in eg in com- plex s uc u e11 wi h an homology modeling me hod implemen ed by he MODELLER module o Disco e y S udio 2.5 as desc ibed p e iously35,36. As shown in Fig.3B1, he β 1-in eg in ail mos ly con ac s wi h hese wo β -shee s o F3 domain. Based on he ene gy minimized complex s uc u es om 40 ns MD simula ion, we ound ha he β 1-in eg in ail binds o alin1WT wi h a e y di e en o ien a ion compa ing wi h i s binding o alin2WT (Fig.3B2), while he single mu a ion C336S o alin1 swi ches he β 1-in eg in in Talin1WT/in eg in complex o a simila o ien a ion seen in alin2WT/in eg in complex (Fig.3B3,B4). Fu he mo e, he a e age s uc u es o he alin2WT/in eg in, alin1WT/in eg in, and alin1C336S/in eg in complexes o 40 ns simula ion also con i m he sys ema ic di e ence be ween alin2WT/in eg in and alin1WT/in eg in and he sys ema ic simila i y be ween al- in2WT/in eg in and alin1C336S/in eg in (Supplemen a yFig.S2). In e es ingly, he β -shee 1, especially he loop egion be ween β s and 1 (β 1) and β s and 2 (β 2) in β -shee 1 adop s signi ican ly di e en con o ma ions in alin2WT/in eg in and alin1WT/in eg in complex (Fig.3B5), while he C336S mu an o alin1 elimina es he di - e ence (Fig.3B6,3B7). This is consis en wi h expe imen al esul s ha alin2 has a highe a ini y o β 1-in eg in ails han alin1, and ha he subs i u ion o alin1 C336 wi h Se enhances he a ini y o alin1 (Fig.1). Role o S339 in alin/in eg in in e ac ion. As he β 1-β 2 loop o alin con ac s di ec ly wi h β 1-in eg in, con o ma ion o β -shee 1 should be one o he decisi e ac o s o he o ien a ion o β 1-in eg in binding wi h alin. Howe e , β -shee 1 o alin1 and alin2 a e ully conse ed excep in wo esidues, D338 and S339 in al- in2WT co esponding o E335 and C336 in alin1WT, espec i ely (Fig.3A). As he side chains o aspa ic acid and glu amic acid ha e simila chemical p ope y, S339 o alin2WT and C336 o alin1WT could be conside ed as he majo di e ence in β -shee 1 o he wo iso o ms. As a esul , we expec ha he C336S mu a ion o alin1WT would adap he con o ma ion o β -shee 1 simila o ha in alin2WT, hus alin1C336S and alin2WT would bind o β 1-in eg in wi h simila binding o ien a ion. Fig.4A depic s he acked dis ances and ypical snapsho s o he hyd ogen bond be ween S336/339 and E350/353 in alin/in eg in complex om he MD ajec o ies ( esidue numbe wi h slash indica es esidue in alin1/ alin2, espec i ely). The a e age dis ance o his hyd ogen bond www.na u e.com/scien i ic epo s/ 3 Scien i ic RepoR s | 7:41989 | DOI: 10.1038/s ep41989 (in he las 10 ns o simula ion) is 1.75 ± 0.18 Å ( equency o hyd ogen bond occu ence: 99.3%, wi h 2.5 Å H-O dis ance h eshold) and 1.78 ± 0.24 Å ( equency: 98.2%) in alin1C336S/in eg in and alin2WT/in eg in com- plex espec i ely, while a much la ge and mo e luc ua ing a e age dis ance o 4.39 ± 1.92 Å ( equency: 27.0%) 41 82 163 320 621 1169 Inpu (nM) Inpu (nM) -His- alin11-446WT -His- alin21-449WT -GST-β1A ail -GST-β1A ail A S anda d S anda d 1.06 0.53 0.27 0.13 0.07 µg 41 82 163 320 621 1169 B S anda d -His- alin21-449S339L -GST-β1A ail 78 157 313 627 1253 2507 Inpu (nM) 0 0.2 0.4 0.6 0.8 1 1.2 Tln1 WT Tln1 C336S Tln2 WT Tln2 S339C Kd (nM) P=0.002 P=0.04 P=0.02 D -GST -His- alin11-446WT β1A pulldown GST pulldown β1A pulldown -His- alin21-449WT -GST GST pulldown 1.06 0.53 0.27 0.13 0.07 µg C β1A pulldown GST pulldown -His- alin21-449S339L -GST E 1.06 0.53 0.27 0.13 0.07 µg 55- 35- 55- 25- 55- 35- 55- 25- 55- 55- 35- 25- -GST Figu e 1. A ini ies o alin11–446, alin11–446C336S, alin21–449, and alin21–449S339C o β1-in eg in ails. The binding o pu i ied His- agged p o eins o GST-β 1-in eg in ails was de e mined using GST pulldown assays. Bound p o ein was sepa a ed using SDS-PAGE, s ained wi h Coomassie blue, and calib a ed wi h s anda ds un on he same gels. The binding cu es and he Kd we e analyzed wi h SigmaPlo . (A) Sequence alignmen o he in eg in-binding egion o he F3 domains o alin1 and alin2. Unma ched esidues a e shown in Red. (B) GST-β 1 ails-bound alin11–446 and alin21–449 we e compa ed a e SDS-PAGE and Coomassie blue s aining. C opped gel images a e shown. The ull-leng h gel images a e included in Supplemen a yFig.S3. (C) The speci ic binding cu es o alin11–446, alin11–446C336S, alin21–449, and alin21–449S339C o β 1-in eg in ails. Please no e ha p o ein concen a ions used in binding assays we e di e en . (D) The dissocia ion cons an s o alin11–446, alin11–446C336S, alin21–449, and alin21–449S339C binding o β 1A-in eg in ails. Da a a e p esen ed as mean ± SEM. P alues om s uden ’s - es a e shown. (E) GST-β 1 ails-bound alin21–449S339L we e de ec ed by Coomassie blue s aining. C opped gel images a e shown. The ull-leng h gel images a e included in Supplemen a yFig.S3. www.na u e.com/scien i ic epo s/ 4 Scien i ic RepoR s | 7:41989 | DOI: 10.1038/s ep41989 Figu e 2. T acked posi ional RMSD alues o he wild ype Talin1/In eg in, C336S mu an Talin1/ In eg in and wild ype Talin2/In eg in complex base on 40 ns MD simula ion. Figu e 3. Supe imposi ion o alin/in eg in complexes. (A) Sequence alignmen o Mus musculus alin2 (mTalin2), Homo sapiens alin2 (hTalin2), and Homo sapiens alin1 (hTalin1) by PROMALS3D se e . β s ands o alin (β 1-β 7), α helix o β 1-in eg in as well as he name o he subuni a e ma ked by black lines. β -shee 1 and β -shee 2 a e colo ed in g een and yellow espec i ely. The key in e ac ion pai o Tln-S339/336 and Tln-E353/350 a e colo ed in ed, he key in e ac ion pai o Tln-K324/327 and In -D759 a e colo ed in pu ple, and he Tln- N323/326 and In -R760 in e ac ion pai a e colo ed in cyan. (B) (1) Typical alin2WT/in eg in complex s uc u e is de i ed om he las snapsho o he 40 ns MD ajec o y by ene gy minimiza ion. Talin2 and β 1-in eg in a e ep esen ed as cyan and blue ibbons, espec i ely. Tln-S339, Tln-E353, Tln-K327, Tln-N326, In -R760, and In -D759 a e showed in s ick-ball model. (2) Supe imposi ion o he alin2WT/in eg in and alin1WT/in eg in complexes. Fo alin2WT/in eg in complex, alin2WT and β 1-in eg in a e colo ed in cyan and blue, espec i ely. Fo alin1WT/in eg in complex, alin1WT and β 1-in eg in a e colo ed in golden and ed, espec i ely. (3) Only β -shee 1 o alin and β 1-in e g in a e shown. Al hough o he pa s o alin could be supe imposed well, β -shee 1 a ies a lo in alin2WT and alin1WT. (4) Supe imposi ion o he alin2WT/in eg in and alin1C336S/in eg in complex. Talin1C336S and β 1-in eg in a e colo ed in g een and pu ple, espec i ely. (5) Only β -shee 1 o alin and β 1-in eg in a e shown, and hey bo h ha e simila con o ma ion in alin2WT/in eg in and alin1C336S/in eg in complexes. (6) Supe imposi ion o all h ee alin/in eg in complexes. (7) Only β -shee 1 o alin and β 1-in e gin a e shown. www.na u e.com/scien i ic epo s/ 5 Scien i ic RepoR s | 7:41989 | DOI: 10.1038/s ep41989 be ween e minus hyd ogen o C336 and oxygen o ca boxyl g oup o E350 is obse ed in alin1WT/in eg in complex. Thus, he hyd ogen bond be ween S339 and E353 in alin2WT does no exis in alin1WT due o he weak hyd ogen bond capaci y o co esponding esidue C336 in alin1WT compa ing wi h S339 in alin2WT (Fig.4D,E). The e o e, C336S mu a ion o Talin1 is expec ed o es o e his hyd ogen bond (Fig.4F). In addi ion, as esidues S336/339 and E350/353 could o m hyd ogen bond be ween he β 3 and β 4 s and o β -shee 1, loss o his s ong in e ac ion in alin1WT/in eg in complex would in luence he s e ic alignmen o β s ands in β -shee 1, which is also obse ed in he ypical snapsho s om MD simula ion (Fig.3B5–7). The hyd ogen bonds be ween alin and β1-in eg in con ibu e o alin’s a ini y o β1-in eg ins. Since C336 in alin1WT al e s he con o ma ion o β -shee 1, he o ien a ion o β 1-in eg in ail binding o alin1WT is signi ican ly shi ed om i s binding s a us in alin2WT/in eg in complex. The e o e, he in e ac ion pai s in ol ed in he binding in e ace o alin1WT/in eg in and alin2WT/in eg in complex a e much di e en . Besides, ib a ion o β 1-in eg in ail in hese complex s uc u es is also obse ed du ing 40 ns simula ions, which implies ha he s a ic compa ison o in e ac ion pai s would be biased. We calcula ed he s a is ical hyd ogen bond p o ile o he binding in e ace o e eal he di e ence among he h ee binding com- plexes. Figu e5 depic s is he hea map ha shows he change in he hyd ogen bond p o ile o alin1WT/in eg- in and alin1C336S/in eg in compa ed wi h alin2WT/in eg in. Mo e hyd ogen bonds a e o med by Talin2WT/ in eg in complex han alin1WT/in eg in complex, suppo ing he s a emen ha β 1-in eg in binds o alin2WT wi h highe a ini y han binding o alin1WT (Fig.5A). Mo eo e , subs i u ion o alin1 C336 wi h Se p omo es hyd ogen bond o ma ion be ween alin1 and β 1-in eg in (Fig.5B). This is consis en wi h he expe imen esul ha subs i u ion o alin1 C336 wi h Se enhances he a ini y o alin1 o β 1-in eg in ails. Based on he MD simula ion, wo hyd ogen bonds be ween β 1- β 2 loop o alin and β 1-in eg in ails, i.e., hyd ogen bonds be ween Tln-K324/327 and In -D759, and be ween Tln-N323/326 and In -R760 (Tln e e s o alin, In o β 1-in eg in ails), a e obse ed in ypical snapsho s o alin1C336S/in eg in and alin2WT/in eg in complexes (Fig.4G,I), bu a e missing in alin1WT/in eg in complex (Fig.4H),. Acco ding o he acked dis ance o his hyd ogen bond depic ed in Fig.4B, he a e age dis ance (in he las 10 ns o simula ion) o hyd ogen bond be ween Tln-K324/327 and In -D759 is 2.02 ± 0.52 Å ( equency o hyd ogen bond occu ence: 95.1%, wi h 2.5 Å H-O dis ance Figu e 4. The hyd ogen bonds be ween Tln-S339 and Tln-E353, Tln-K327 and In -D759, and Tln-N326 and In -R760 con ibu e o hei a ini y o β1-in eg ins. (A) T acked dis ance o he in a-molecula hyd ogen bond o med by esidue S336/339 and E350/353 o alin. Black, blue, and ed acks ep esen he dis ance (Uni in Å) be ween he Hɣ a om o S339/C336/S336 and he Oε a om o E353/350 in alin2WT/ in eg in, alin1WT/in eg in, and alin1C336S/in eg in complexes, espec i ely. (B) T acked dis ance o he in e - molecula hyd ogen bond o med be ween he posi i e-cha ged side chain o K324/327 in alin and he nega i e cha ged side chain o D759 in β 1-in eg in. Black, blue, and ed acks ep esen he dis ance be ween he Hζ a om o K327/324 o alin and he Oδ a om a D759 o β 1-in eg in in alin2WT/in eg in, alin1WT/in eg in, alin1C336S/in eg in complexes, espec i ely. (C) T acked dis ance o he in e -molecula hyd ogen bond o med be ween he side chain o N323/326 in alin and side chain o R760 in β 1-in eg in. Black, blue, and ed acks ep esen he dis ance be ween he Oδ a om o N326/323 o alin and he Hη a om o R760 o β 1-in eg in in alin2WT/in eg in, alin1WT/in eg in, and alin1C336S/in eg in complexes, espec i ely. (D) Hyd ogen bond be ween S339 and E353 o alin wi h labeled dis ance in alin2WT/in eg in complex. Talin is ep esen as cyan ibbons. S339 and E353 a e shown in golden s ick-ball s yle. (E) Side chains o C336 and E350 a e a away om each o he in alin1WT/in eg in complex. (F) Hyd ogen bond be ween S336 and E350 wi h labeled dis ance in alin1C336S/in eg in complex. (G) Hyd ogen bond be ween Tln-K327 and In -D759, and Tln-N326 and In -R760 wi h labeled dis ance in alin2WT/in eg in complex. Talin and β 1-in eg in a e p esen ed as cyan ibbons and g een ibbons espec i ely. (H) Side chains o Tln-K324, Tln-N323, In -D759, and In -R760 a e a away om each o he in alin1WT/in eg in complex. (I) Hyd ogen bond be ween Tln-K324 and In -D759, and Tln-N323 and In -R760 wi h labeled dis ance in alin1C336S/in eg in complex. www.na u e.com/scien i ic epo s/ 6 Scien i ic RepoR s | 7:41989 | DOI: 10.1038/s ep41989 h eshold) and 2.00 ± 0.53 Å ( equency: 90.6%) in alin1C336S/in eg in and alin2WT/in eg in complex, espec- i ely, while his dis ance inc eases o 5.07 ± 1.07 Å ( equency: 0.1%) in alin1WT/in eg in complex. Simila ly, he a e age dis ance (in he las 10 ns o simula ion) o he hyd ogen bond be ween Tln-N323/326 and In -R760 is 2.39 ± 0.46 Å ( equency: 76.2%) and 2.51 ± 0.79 Å ( equency: 63.2%) in alin1C336S/in eg in and alin2WT/in e- g in complex, espec i ely, while his dis ance inc eases o 7.40 ± 1.47 Å ( equency: 0.0%) in alin1WT/in eg in complex (Fig.4C). Al hough he e a e many o he di e ences among he h ee complexes, s a us o hese wo hyd ogen bonds could be conside ed as key indica o s o alin-in eg in binding s a e. I should be no ed ha bo h Tln-K324/327 and Tln-N323/326 a e loca ed in he β 1- β 2 loop (Fig.4G), so he con o ma ion o he β 1-β 2 loop is expec ed o in luence he in e ac ion be ween alin and in e g in. In conclusion, esidue C336 o alin1WT dis u bs he con o ma ion o β -shee 1 as well as β 1-β 2 loop, hus elimina ing bo h hyd ogen bonds be ween Tln-K324 and In -D759, and Tln-N323 and In -R760, which esul s in an o -swi ch in he o ien a ion o β 1-in eg in binding and a dec ease in he a ini y o alin1WT o β 1-in eg in ails. On he con a y, alin1C336S adap s a con o ma ion o he β 1-β 2 loop seen in alin2/in eg in complex, hus es o ing he hyd ogen bonds be ween Tln-K324 and In -D759, and Tln-N323 and In -R760, which is expec ed o enhance he a ini y o alin1C336S o β 1-in eg in a ils. The c i ical ole o he hyd ogen bonds be ween Tln-S339 and Tln-E353, Tln-K327 and In -D759, and Tln-N326 and In -R760 o alin2’s high a ini y o β1-in eg in ails. Nex , we se ou o e i y ou compu a ional model. The phospho y osine binding (PTB)-like domain is esponsible o alin1 binding o β 1-in eg in ails. To p o e whe he his is ue o alin2, R361, W362, and S365 wi hin he alin2 PTB-like domain we e subs i u ed wi h Alanine, and he alin2 mu an s and he WT we e ansien ly ans ec ed in o CHO-K1 cells. The binding o hese p o eins o β 1-in eg in ails was de e mined using GST pulldown assays. As shown in Fig.6A, mu a ion a any o hese esidues caused d ama ic educ ion in alin2 binding o β 1-in eg in ails, indi- ca ing ha alin2 and alin1 employ he same mo i o bind β 1-in eg in ails. To con i m he c i ical ole o S339 in modula ing alin2 binding o β 1-in eg in ails, S339 and he adjacen esidues D338 and V340 we e mu a ed o Ala. These mu an s and he WT we e ans ec ed o CHO-K1 cells and hei binding o β 1 A ails was examined. Subs i u ion o ei he S339 o V340 wi h Ala signi ican ly diminished alin2 binding o β 1-in eg in ails; Subs i u ion o D338 wi h Ala also sligh ly dec eased alin2 binding (Fig.6B). Based on ou compu a ional model, alin2 S339 and E353 o m a hyd ogen bond be ween he β 3 and β 4 s and o β -shee 1, which p omo es alin2 binding o β 1-in eg in ails. To e i y he ole o E353 in media ing he alin2 β 1-in eg in in e ac ion, alin2 E353 was subs i u ed wi h Gly and Lys, espec i ely, and he in e ac ion o hese mu an s wi h β 1-in eg in ails was de e mined. Subs i u ion o alin2 E353 wi h ei he Gly o Lys signi ican ly diminished he binding o alin2 o β 1-in eg in ails (Fig.6C), sugges ing ha he hyd ogen bond be ween S339 and E353 is equi ed o he high a ini y o alin2 o β 1-in eg in ails. To know whe he he hyd ogen bond be ween alin2 K327 and β 1-in eg in D759 is c i ical o he alin2-in eg in in e ac ion, alin2 K327 was mu a ed o Glu and Ala, espec i ely, and he binding o hese mu an s o β 1-in eg in ails was examined. Subs i u ion o K327 wi h Glu abolished he in e ac ion o alin2 wi h β 1-in eg in ails (Fig.6C), and subs i u ion o K327 wi h Ala also educed he binding o alin2 (Fig.6D). In Figu e 5. The s a iscal hyd ogen bond p o ile among alin2WT/in eg in, alin1WT/in eg in and alin1C336S/ in eg in complexes. (A) The di e ence o a e age in e -molecula hyd ogen bond numbe be ween alin2WT/ in eg in and alin1WT/in eg in. As he di e ence is dependen on he h eshold o hyd ogen bond de inia ion, he p o ile o a ious hyd ogen bond dis ances and angle cu o s is shown. He e he hyd ogen bond dis ance is de ined as he dis ance be ween hyd ogen bond dono and accep o , while he hyd ogen bond angle is de ined as he angle be ween dono , hyd ogen, and accep o . Fo a ypical hyd ogen bond dis ance cu o o 3.0 Å and angle cu o o 120°, alin2WT/in eg in complex has abou 1 mo e in e -molecula hyd ogen bond han alin1WT/ in eg in complex. (B) The di e ence o he a e age in e -molecula hyd ogen bond numbe be ween alin2WT/ in eg in and alin1C336S/in eg in. I could be obse ed ha hese wo complexes o m an in e -molecula hyd ogen bond in a simila le el. www.na u e.com/scien i ic epo s/ 7 Scien i ic RepoR s | 7:41989 | DOI: 10.1038/s ep41989 addi ion, subs i u ion o N326 wi h Ala signi ican ly diminished he in e ac ion o alin2 wi h β 1-in eg in ails (Fig.6D), which also suppo s he p edic ed hyd ogen bond be ween N326 o alin2 and R760 o β 1-in eg in. Taken oge he , he hyd ogen bonds be ween alin2 S339 and E353 and be ween alin2 K327 and β 1-in eg in D759 a e c i ical o alin2 o main ain i s high a ini y o β 1-in eg ins. Figu e 6. Expe imen al examina ion o he compu a ional model. (A) Binding o EGFP- alin21–449WT, - alin21–449R361A, - alin21–449W362A, and - alin21–449S365A o β 1-in eg in ails by GST pulldown assays. The EGFP usion p o eins we e ansien ly exp essed in CHO-K1 cells. (B) EGFP- alin21–449WT, - alin21–449D338A, - alin21– 449S339A, and - alin21–449V340A we e ansien ly ans ec ed in o CHO-K1 cells. The binding o hese p o eins o β 1-in eg in ails was de e mined by GST pulldown assays. (C) Binding o EGFP- alin21–449WT, - alin21–449K327E, - alin21–449E353G, and - alin21–449E353K o β 1-in eg in ails by GST pulldown assays. (D) Binding o EGFP- alin21–449WT, - alin21–449K324A, - alin21–449N326A, and - alin21–449K327A o β 1-in eg in ails by GST pulldown assays. C opped blo /gel images a e shown. The ull-leng h blo /gel images a e included in Supplemen a yFig.S4. www.na u e.com/scien i ic epo s/ 8 Scien i ic RepoR s | 7:41989 | DOI: 10.1038/s ep41989 Talin2 is a ocal adhesion p o ein. To know whe he alin2 S339 is c i ical o ocal adhesion o ma ion, alin2-null human os eosa coma cells U2 OS we e ans ec ed wi h EGFP- alin2WT and – alin2S339C, espec i ely. Cells we e pla ed on glass-bo om dishes ha we e p e-coa ed wi h ib onec in (5 μ g/ml), ixed, and s ained wi h an an i-phospho-FAK an ibody. The images o EGFP and phospho-FAK we e eco ded wi h a TIRF mic oscope. EGFP- alin2WT was co-localized wi h phospho-FAK, whe eas EGFP- alin2S339C was de icien in ocal adhesion o ma ion and had a diminished co-localiza ion wi h phospho-FAK (Fig.7A,B), sugges ing ha a s ong in e ac- ion o alin2 wi h β 1-in eg ins is essen ial o e icien ocal adhesion assembly. Discussion We ound ha alin1 and alin2 in e ac ed wi h β 1A-in eg in ails wi h Kd o 0.88 and 0.35 μ M, esp ec i ely. Subs i u ion o S339 wi h Leucine comple ely dis up ed alin2 binding o β 1-in eg ins, and mu a ion o S339 o Cys also diminished i s binding, indica ing ha S339 is c i ical o alin2 binding o β 1-in eg ins. Ou indings sugges ha he incapaci y o S339L mu an binding o β -in eg ins could be he pa hological cause o i h inge camp odac yly. Subs i u ion o alin1 C336 wi h Se caused only a small inc ease in alin1’s a ini y (Fig.1), bu C336S mu an bound o β in eg in ails mo e s ongly han alin1WT when alin1 and C336S we e exp essed in mammalian cells33. Simila esul was obse ed as compa ing he binding o alin2WT and alin2S339C. These esul s sugges ha alin and mu an s om mammalian cells may ha e pos - ansla ional modi ica ions ha in luence hei binding o β –in eg in ails. Howe e , he possibili y ha p o eins om cell lysa es ha e supe io olding compa ed o hose om bac e ia canno be uled ou . P e ious s udy epo ed ha he F3 domains o alin1 and alin2 bound o β 1 A ails wi h Kd o 491 and 652 μ M, espec i ely, as measu ed by NMR11. Ou indings a e di e en om he esul s in his epo . As we discussed p e iously, he di e ence could be caused by using di e en agmen s o he alin head domains and di e en binding bu e s and blocking eagen s in he assays33. The olding o alin head domains pu i ied om bac e ia could also cause his di e ence. Al hough ou gel il a ion ch oma og aphy analysis indica es ha hese p o eins a e olded (Supplemen a yFig.S1), his me hod does no p o e ha he p o eins a e olded and pos - ansla ionally modi ied exac ly as in hei na u al hos . Ou unpublished da a show ha alin head domains induced by IPTG a 37 °C had highe β 1 A in eg in binding capaci ies han hose induced a 19 °C, p obably because he exp ession o chape ones ha con ol p o ein olding a e inhibi ed a a low empe a u e37,38. We used alin head domains ha we e induced by IPTG a 37 °C in ou a ini y assays. We do no ha e e idence ha he p o eins s udied he e would ha e been olded be e han in o he s udies. Howe e , ou a ini y assays wi h pu i ied ecombinan p o eins om E. coli a e consis en wi h he pulldown assay esul s conduc ed using mam- malian cell lysa es. Talin head domain pu i ied om CHO cells will be use ul o u he cla i y his di e ence. Ou compu a ional modeling esul s show ha alin2 S339 o ms hyd ogen bonds wi h E353, which is c i ical o key hyd ogen bonds be ween alin2 K327 and β 1-in eg in D759, and be ween alin2 N326 and β 1-in eg in R760 (Fig.4D,G). These hyd ogen bonds we e no obse ed in he alin1/in eg in complex (Fig.4E,H). Mu a ion a any o hese esidues signi ican ly diminished he binding o alin2 wi h β 1-in eg in ails (Fig.6). Howe e , EGFP An i-pFAK Me ge EGFP- alin2WT EGFP- alin2S339C 0 50 100 150 200 250 300 350 400 GFP-TLN2 WT GFP-TLN2 s339c FA Numbe A ea (µm2) A B Figu e 7. Subs i u ion o alin2 S339 wi h Cys diminished ocal adhesion o ma ion. Talin2-null U2 OS cells we e ans ec ed wi h EGFP- alin2WT and - alin2S339C, espec i ely, pla ed on ib onec in (5 μ g/ml), ixed wi h pa a o maldehyde, and s ained wi h an i-phospho-FAK[pY397] an ibody. Focal adhesions we e examined wi h a TIRF mic oscopy. Scale ba , 20 μ m. (A) The dis ibu ion o alin2, alin2S339C, and phospho-FAK in U2 OS cells. (B) A ea dis ibu ion o phospho-FAK s aining in U2 OS cells. Da a a e mean ± SEM o 3 expe imen s. In each g oup, FAs om 20 cells we e analyzed and plo ed. www.na u e.com/scien i ic epo s/ 9 Scien i ic RepoR s | 7:41989 | DOI: 10.1038/s ep41989 subs i u ion o alin1 C336 wi h Se p omo ed he o ma ion o hese hyd ogen bonds (Fig.4F,I), accompanying an inc ease in i s a ini y owa d β 1-in eg in ails. In e es ingly, mu a ion o alin2 S339 o Leu caused a signi ican educ ion in i s binding o β 1-in eg in ails (unpublished da a) and Fi h Finge Camp odac yly, sugges ing ha he educ ion in he alin2-in eg in in e ac ion could be he cause o he gene ic disease. These esul s sugges ha he hyd ogen bonds be ween alin2 S339 and E353 play a c i ical ole in main aining alin2’s high a ini y o β 1-in eg in. Because hese esidues a e loca ed ou side o he well-cha ac e ized β 1-in eg in-binding mo i , i is possible o design a small molecule o diminish alin2 binding o β 1-in eg ins wi hou a ec ing alin1. Besides media ing β -in eg in in e ac ion, alin2 K327 may also be in ol ed in o he unc ions o alin2. P e ious s udies show alin1 K324, which is aligned wi h K327 on alin2, o be a key esidue o alin1 binding o PIP2 and is essen ial o in eg in clus e ing39. This esidue also media es he head-and- od au o-inhibi ion o alin140–42. S udies ha e also shown ha PIP2 binds o alin1 and dis up s he head-and- od in e ac ion, hus causing alin1 ac i a ion40,43. I is likely ha alin2 K327 is a key esidue o he alin2-PIP2 in e ac ion and is also in ol ed in he head-and- od au o-inhibi ion o alin2. Howe e , i is unknown whe he binding o PIP2 dis up s o enhances alin2-β 1-in eg in in e ac ion. Talin2 N236 is ano he key esidue ha media es alin2-β 1-in eg in in e ac ion. In ac , subs i u ion o alin2 N326 wi h Ala had mo e signi ican e ec on he binding o alin2 o β 1-in eg in ails han ha o K327 wi h Ala (Fig.6D). Al hough subs i u ion o alin2 K327 wi h Glu dis up ed he in e ac ion o alin2 wi h β 1-in eg in ails (Fig.6C), his could be caused by he pe u ba ion o he nega i e cha ge o he Lys esidue on alin2 con o ma- ion. Since alin2 K327 may in ol e in PIP2 in e ac ion, he hyd ogen bond be ween Tln-N326 and In -R760 could play a key ole in he alin2-β 1 in eg in in e ac ion, especially in he p esence o PIP2. Ma e ials and Me hods Reagen s. An i-FAK[pY397] was om BD Biosciences. An i- ubulin an ibody was om Sigma. DyLigh 549 conjuga ed goa an i-mouse IgG (H + L) was om The mo Scien i ic. Fib onec in and ecombinan human EGF we e om Ak on Bio ech; G ow h ac o educed Ma igel was om BD Bioscience. P u Ul a was om Agilen Technologies. Cold Fusion Cloning Ki was om Sys em Biosciences (Palo Al o, CA). An i-GFP monoclonal an ibody and Sa ec ine RU50 ans ec ion ki we e pu chased om Syd Labs (Malden, MA). DNA p ime s we e syn hesized by Sigma-Ald ia. Plasmid cons uc ion. The ull-leng h pEGFP- alin2 WT was subcloned by he ollowing s eps: 1) DNA agmen s encoding esidues 1–1159 o human alin2 we e ampli ied by P u Ul a-based PCR using human alin2 cDNA clone as empla e and 5′ -a g cac cg agc a gg ggc cc g c c aaa ga g -3′ /5′ -ac gag g a ccg c cga gca gaa c aac a g gca -3′ as p ime s and subcloned in o pEGFP-C1 ia Xho1/Kpn1 si es; 2) agmen s encoding esidues 1160–2543 o alin2 we e ampli ied using human cDNA om U2 OS cells and 5′ -ggc gc a c gac aac cga ccc c-3′ /5′ - a a c a ga ag ccc ca c cc c c agc c-3′ and subcloned in o he esul ed plasmid in s ep 1 ia No 1/Xba1 si es. pEGFP- alin21–449 was gene a ed by ampli ying DNA agmen s encoding esidues 1–449 using 5′ -a g cac cg agc a gg ggc cc g c c aaa ga g -3′ /5′ -ggg ccc g c gac a gag ccg gc c gcc c cc-3′ as p ime s and subcloning in o pEGFP-C1 ec o ia Xho1/Sal1 si es. pEGFP- alin21–449S339C was c ea ed by p u Ul a-based PCR using pEGFP- alin21–449 as empla e and 5′ -gga ca cca aag ac g g ga gc gcg gg-3′ /5′ -cca cgc gca ca cac ag c gg ga cc-3′ as p ime s. pEGFP- alin21–449R361A, - alin21–449W362A, and - alin21–449S365A was gene a ed by p u Ul a-based PCR using pEGFP- alin21–449 as empla e and 5′ -c c acc accg c aag gcc gg gca gcc ca ccc-3′ /5′ -ggg ga ggc gc cca ggc c gac gg gg gag-3′ , 5′ -acc acc g c aag cgc gcg gca gcc ca ccc aag-3′ /5′ -c ggg ga ggc gc cgc gcg c gac gg gg -3′ , and 5′ -aag cgc gg gca gcc gca ccc aag agc c aca-3′ /5′ - g gaa gc c ggg gc ggc gc cca gcg c -3′ as p ime s, espec i ely. pEGFP- alin21–449D338A, - alin21–449S339A, and - alin21–449V340A was c ea ed by PCR using 5′ -c g ggg a cacc aaa gcc cg g g a g cgc g g-3′ /5′ -cac gcg ca cac cga ggc gg ga ccc cag − 3′ , 5′ -ggg a c acc aaa gac gcg g g a g cgc g g ga -3′ /5′ -a c cac gcg ca cac cgc g c gg ga ccc-3′ , and 5′ -a c acc aaa gac cg ggg a g cgc g g ga gag − 3′ /5′ -c c a c cac gcg ca ccc cga g c gg ga -3′ as p ime s, espec i ely. pEGFP- alin21–449K325A, - alin21–449N326A, and - alin21–449K327A was c ea ed by PCR using 5′ -aag gag aag a g aaa ggc gcg aac aag c g g g cc cg-3′ /5′ -cga ggc acc agc g c gcg cc c a c c cc -3′ , 5′ -aga aga ga aag gca agg cca agc gg gc c c gcc-3′ /5′ -ggc gag gca cca gc gg cc gc c ca c c -3′ , and 5′ -ga gaa agg caa gaa cgc gc gg gcc cg cc g-3′ /5′ -cag gcg agg cac cag cgc g c gcc ca c-3′ as p ime s, espec i ely. pEGFP- alin21–449E353G, - alin21–449E353K, and - alin21–449K327E was c ea ed by PCR using 5′ -agg aag gc gc agg gg ggc ccc ca cca c-3′ /5′ -g g g g agg ggc cac ccc gc agc ac cc − 3′ , 5′ -agg aag gc gc aga ag ggc ccc ca cca c − 3′ /5′ -g g g g agg ggc cac c gc agc ac cc − 3′ , and 5′ -a g aaa ggc aag aac gag c g g g cc cgc c − 3′ /5′ -ggc gag gca cca gc cg c gc c ca − 3′ as p ime s, espec i ely. The ull-leng h pEGFP- alin2S339C was c ea ed by diges ing ull-leng h pEGFP- alin2 wi h Bs G1/EcoRV and liga ing he esul ing la ge agmen wi h he smalle agmen s om pEGFP- alin21–449S339C. pQE- alin11–446 was gene - a ed by ampli ying he DNA agmen s using human alin1 cDNA as empla e and 5′ -ggg ccc gag c c a g g gca c ca c g aag a c ag-3′ /5′ -ggg ccc g c gac a aga gcc a g c c cac ccc c-3′ as p ime s and subcloning in o pQE-30 ec o ia Sac1/Sal1. pQE- alin11–446C336S was c ea ed by PCR using pQE- alin11–446 as empla e and 5′ -ca cac caa gga gag g ga gcg ag-3′ /5′ -c c gca ca cac c cc gg ga g − 3 ′ as p ime s. pQE- alin21–449 and - alin21–449S339C we e gene a ed by ampli ying he DNA agmen s using 5′ -a g cag aa cca gg gg ccc g cc aa aga g -3′ /5′ -ggg ccc g c gac a gag ccg gc c gcc c cc-3′ as p ime s, pEGFP- alin21–449 and pEGFP- alin21–449S339C as empla es, and subcloning in o pQE-30 ec o ia Sac1/Sal1 and BamH1/Sal1, espec i ely. All plasmids we e sequenced by Eu o ins MWG Ope on (Hun s ille, AL). P o ein in e ac ion assays. CHO-K1 cells we e ans ec ed wi h pEGFP- alin11–433, - alin11–446, - alin11– 449, o hei mu an s. A 28 h pos - ans ec ion, he cells we e ha es ed in lysis bu e A (50 mM T is-HCl pH 7.4, 1% NP-40, 150 mM NaCl, 1 mM EDTA and a p o ease inhibi o cock ail). Cell lysa es we e clea ed by cen i u- ga ion and incuba ed wi h glu a hione–Sepha ose beads loaded wi h GST o GST-β 1-in eg in ails a 4 °C o 2 h.