Supporting InformationOhnishi et al. 10.1073/pnas.0812956106Materials and MethodsSample Preparation. The portion of the human MyD88 geneencoding the death domain, including the internal domain(residues 18–141), was cloned into the vector pGEX-5X-3 (GEHealthcare). This vector was transformed into E. coli BL-21(DE3) (Novagen). The MyD88-DD�ID, which was expressed asa GST (GST) fusion protein, was first purified by glutathioneSepharose 4B FF (GE Healthcare) affinity chromatography, andthe GST-tag was removed by digestion with Factor Xa (GEHealthcare). Subsequently, the MyD88-DD�ID was purified bygel filtration (Sephacryl S-100 HR 26/60 column; GE Health-care) and anion-exchange chromatography (Mono-Q column;GE Healthcare).

Analytical Gel Filtration. Molecular masses of the purified MyD88-TIR (residues 148–296) and MyD88-DD�ID (residues 18–141)proteins were evaluated by size exclusion chromatography. Gelfiltration analysis was performed using a Superdex-75HR 10/30column attached to a FPLC system (GE Healthcare) at 4 °C. Thecolumn was equilibrated with 20 mM potassium phosphatebuffer (pH 6.0), 100 mM KCl, 0.1 mM EDTA, and 10 mM DTT.The column was calibrated using a low molecular weight cali-bration kit (GE Healthcare). A total of 300 �l of MyD88-TIR orMyD88-DD�ID protein was applied to the gel filtration column.Protein elution was monitored by UV absorption at 280 nm. Themolecular masses of these proteins were estimated using acalibration curve.

NMR Titration. One 15-�l aliquot of 100 �M non-labeled Mal-TIRwas added to 150 �l of 20 �M 15N-labeled MyD88-TIR or itsalanine substituted mutants up to 1.5 molar equivalent of the 15NMyD88-TIR. At each titration point, 1D 1H-{15N} and 2D1H-15N SOFAST-HMQC spectra were recorded.

NMR signals in 1D 1H-{15N} SOFAST-HMQC spectra, be-tween 6–11 ppm, were integrated to reveal the covered areas,which were then normalized with values obtained from thespectra of 15N MyD88-TIR, or its derivatives recorded in the

absence of Mal-TIR. The simple model shown below wasassumed to analyze the NMR signal attenuation data, where onlyfree MyD88-TIR was ‘‘NMR visible,’’ and the generated com-plex in any form was ‘‘NMR invisible.’’ 15N MyD88-TIR (NMRActive) � Non-Labeled Mal-TIR 3 complex (NMR inactive).Kd

app values were obtained for MyD88-TIR and its derivatives byfitting the attenuation data to this model (Fig. 2E). The fittingcalculation was performed using the software OriginPro ver.8(Origin Lab Corp.). It should be noted that the exact stoichi-ometry for the interaction was not known. Therefore, theobtained Kd

app values should not be overinterpreted. The valuesshould serve as a comparison of mutation data to estimate theeffect of alanine substitutions.

Docking Studies Between MyD88 and Mal. Structure modeling of theMal TIR domain was performed using the MyD88-TIR structureas a template on MOE software (1, 2). The docking simulationwas performed on AutoDock without any specific restraintsbetween the molecules (3). The partial structures of MyD88-TIR, which included Site II or Site III residues (namely residues195–204 or 279–296), were used as a ligand. Mal-TIR was usedas a target, similar to previous reports (4, 5). A Lamarckiangenetic search algorithm was chosen for all dockings (3). TheAutoDock run parameters were almost identical to those in thedefault setting, except for minor modifications in grid size,maximum number of energy evaluations, and maximum numberof generations. The grid dimensions for the first screening were72 Å*72 Å*72 Å for x, y, and z axis, respectively, and grid centerswere placed to cover the entire surface of the Mal-TIR structure.In the analyses, 100 dockings were performed, and the best-clustered regions were chosen for the second analysis. The griddimensions in the second refinement step were 40 Å*40 Å*30 Å,and the grid centers were placed on the top of the surface. Thecomplex Mal-TIR/MyD88-TIR/Mal-TIR model was constructedby superimposing Mal-TIR on MyD88-TIR via the dockedfragments and then removing the docked fragments (4). Themodeled structure was further energy minimized.

1. Levitt M (1992) Accurate modeling of protein conformation by automatic segmentmatching. J Mol Biol 226:507–533.

2. Fechteler T, Dengler U, Schomburg D (1995) Prediction of protein three-dimensionalstructures in insertion and deletion regions: a procedure for searching data bases ofrepresentative protein fragments using geometric scoring criteria. J Mol Biol 253:114–131.

3. Morris GM, Goodsell DS, Huey R, Olson AJ (1996) Distributed automated docking offlexible ligands to proteins: parallel applications of AutoDock 2.4. J Comput Aided MolDes 10:293–304.

4. Kato Z, et al. (2008) Positioning of autoimmune TCR-Ob. 2F3 and TCR-Ob. 3D1 on theMBP85–99/HLA-DR2 complex. Proc Natl Acad Sci USA 105:15523–15528.

5. Stoddard BL, Koshland DE, Jr (1992) Prediction of the structure of a receptor-proteincomplex using a binary docking method. Nature 358: 774–776.

Ohnishi et al. www.pnas.org/cgi/content/short/0812956106 1 of 5

BB loop

MyD88 M P E R F D A F I C Y C P - - - - - - - - - - - - S D I Q F V Q E M I R Q L E Q T N Y R L K L C V S D R D V L P G -

Mal W S K D Y D V C V C H S E - - - - - - - - - - - - E D L V A A Q D L V S Y L E G S T A S L R C F L Q L R D A T P G G

TLR1 N L Q F H A F I S Y S G H - - - - - - - - - - - D S F W V K N E L L P N L E - - K E G M Q I C L H E R N F V P G -

TLR2 N I C Y D A F V S Y S E R - - - - - - - - - - - D A Y W V E N L M V Q E L E N F N P P F K L C L H K R D F I P G -

TLR4 G E N I Y D A F V I Y S S Q - - - - - - - - - - - D E D W V R N E L V K N L E E G V P P F Q L C L H Y R D F I P G -

IL-1RAPL G D N K D Y D A Y L S Y T K V D P D Q W N Q E T G E E E R F A L E I L P D M L E - K H Y G Y K L F I P D R D L I P T -

MyD88 T C V W S I A S E L I E K R C R R M V V V V S D D Y L Q - S K E C D F Q T K F A L S L S P G A H Q K R L I P I K Y K

Mal A I V S E L C Q A L S S S H C R V L L I T P G F L Q - D P W C K Y Q M L Q A L T E A P G A E G C T I P L L S G

TLR1 K S I V E N I I T C I E K S Y K S I F V L S P N F V Q - S E W C H Y E L Y F A H H N L F H E G S N S L I L I L L E

TLR2 K W I I D N I I D S I E K S H K T V F V L S E N F V K - - S E W C K Y E L D F S H F R L F E E N N D A A I L I L L E

TLR4 V A I A A N I I H E G F H K S R K V I V V V S Q H F I Q - S R W C I F E Y E I A Q T W Q F L S S R A G I I F I V L Q

IL-1RAPL G T Y I E D V A R C V D - Q S K R L I I V M T P N Y V V R R G W S I F E L E T R L R N M L V T G E I K V I L I E C S

DD-DE long loop

MyD88 A M K K E - - - - - F P S I L R - - - - F I T V C D Y - T N P C T K S - - - - - - W F W T R L A K A L S L P - - -

Mal L S R A A - - - - - Y P P E L R - - - - F M Y Y V D G R G P D G G F R Q V K E A V M R C K L L Q E G E G E R D S A

TLR1 P I P Q Y S I P S S Y H K L K S L M A R R T Y L E W P K E K S K R G - - - - - - L F W - A N L R A A I N I K L T

TLR2 P I E K K A I P Q R F C K L R K I M N T K T Y L E W P M D E A Q R E - - - - - - G F W - V N L R A A I K S - - -

TLR4 K V E K T L L R Q Q V E L Y R L L S R N T Y L E W E D S V L G R H - - - - - - I F W - R R L R K A L L D G K S

IL-1RAPL E L R G I M N Y Q E V E A L K H T I K L L T V I K W H G P K C N K L N S - - - - K F W - K R L Q Y E M P F K R I

βA αA βB

αB βC αC αC'

190

230

box 3

box 2box 1

αE αE'βE

βD

240

270260

200

250

290280

160 170 180

210 220

Fig. S1. Sequence alignment of TIR domains of human MyD88, Mal, TLRs, and IL-1Rs. Green and red bars indicate the secondary structure of the MyD88 TIRdomain. Blue bars indicate conserved motifs of TIR domains. Closed triangles indicate the functionally important amino acid residues, as determined in this study.

Ohnishi et al. www.pnas.org/cgi/content/short/0812956106 2 of 5

Fig. S2. The evaluation of homo-dimerization of MyD88. (A and B) Results of analytical gel filtration. The TIR domain of MyD88 and the death domain withinternal domain (DD�ID) were analyzed by Superdex 75 gel filtration chromatography and were calibrated with molecular weight standards. (C) SDS/PAGEanalysis of MyD88-TIR and DD�ID. MyD88-TIR protein was identified as a monomeric protein (estimated molecular weight was 16100), while MyD88-DD�IDprotein was identified as a dimer protein (estimated molecular weight was 31500). These results suggest that MyD88 forms a dimer, which is dependent onself-interaction of the portion of DD�ID. However, the TIR domain of MyD88 is the free monomer in solution for interactions with other TIR proteins.

Ohnishi et al. www.pnas.org/cgi/content/short/0812956106 3 of 5

A691Repyt dliW

A882R,A691RA882R

020

4060

8010

0

050

100

150

200

050

100

150

010

2030

40

Sig

nal

Inte

nsi

ty (

arb

itra

ry u

nit

s)S

ign

al In

ten

sity

(ar

bit

rary

un

its)

Sig

nal

Inte

nsi

ty (

arb

itra

ry u

nit

s)S

ign

al In

ten

sity

(ar

bit

rary

un

its)

10 9 8 7 6

10 9 8 7 610 9 8 7 6

10 9 8 7 6

1H (ppm)1H (ppm)

1H (ppm)1H (ppm)

10 9 8 71H (ppm)

120

115

110

15 N

(p

pm)

10 9 8 71H (ppm)

10 9 8 7

120

115

110

120

115

110

120

115

110

15 N

(p

pm)

1H (ppm) 1H (ppm)

10 9 8 7

0 eq 0.5 eq

1.5 eq1.0 eq

B

A

Fig. S3. The results of the NMR titration study. (A) The 2D 1H-{15N} SOFAST-HMQC spectra of MyD88-TIR (wild type) in the presence of 0, 0.5, 1.0, and 1.5 molarequivalents of non-labeled Mal-TIR. (B) The 1D 1H-{15N} SOFAST-HMQC spectra of MyD88-TIR (wild type), MyD88-TIR (R196A), MyD88-TIR (R288A), and MyD88-TIR(R196A, R288A) in the presence of 0 (purple), 0.5 (green), 1.0 (red), and 1.5 (blue) molar equivalents of nonlabeled Mal-TIR. Plots of relative integrated area ofNMR signals, which were derived from spectra in a volume function of Mal-TIR added to 15N MyD88-TIR solution, are in Fig. 2E. As shown above, the 1D 1H-{15N}SOFAST-HMQC spectra were heavily perturbed following addition of Mal-TIR. More specifically, the NMR signals were attenuated as the amount of Mal-TIRincreased. The 2D spectra recorded at each titration point demonstrated that those remaining NMR signals originated exclusively from free 15N MyD88-TIR. Nonew NMR signals from the complex between MyD88-TIR and Mal-TIR emerged. These results undoubtedly indicate that MyD88-TIR interacted with Mal-TIR.However, results showed a dissimilarity to typical NMR titration experiments, where one should observe emerging NMR signals from the complex forms (in theslow exchange regime) or a position shift of original NMR signals (in the fast exchange regime) as the titrant concentration increases. This observation may beexplained by of Mal-TIR characteristics; Mal-TIR is susceptible to self-oligomerization (1, 2). Therefore, added Mal-TIR molecules interacted with MyD88-TIRmolecules. Simultaneously, self-oligomerization of Mal-TIR initiated transformation of 15N-labeled MyD88-TIR into high molecular weight hetero-TIR oligomers,so that NMR signals from the 15N-labeled proteins became undetectable. Thus, attenuation of NMR signals reflected reduction of free 15N MyD88-TIR. The degreeof signal attenuation was partially suppressed when Arg-196 or Arg-288 of 15N MyD88-TIR was substituted with alanine, which suggested that the interactionwas weakened by such substitutions. A more prominent suppression was observed if both arginine residues were doubly substituted with alanine (B and Fig.2E). These results were consistent with data from the GST pull-down assays (Fig. 2C). To obtain an estimation regarding the effect of alanine substitutions, theNMR signal attenuation curve was fit to a simple 1:1 binding model to evaluate apparent Kd

app values, although the exact stoichiometry was not known. Theapparent Kd

app values allowed for comparison of alanine substitution effects in the 15N MyD88-TIR. Results showed that contributions from both Arg-196 andArg-288 were comparable with one another. This suggested that contributions from Site II and Site III to the interaction were also comparable to each other.

1. Dunne A, Ejdeback M, Ludidi PL, O’Neill LA, Gay NJ (2003) Structural complementarity of Toll/interleukin-1 receptor domains in Toll-like receptors and the adaptors Mal and MyD88.J Biol Chem 278:41443–41451.

2. Fitzgerald KA, et al. (2001) Mal (MyD88-adapter-like) is required for Toll-like receptor-4 signal transduction. Nature 413:78–83.

Ohnishi et al. www.pnas.org/cgi/content/short/0812956106 4 of 5

Table S1. NMR structural statistics

NOE distance constraints

Total 3,119Intra-residue 773

Inter-residueSequential (�i � j� � 1) 937Medium-range (�i � j� � 4) 663Long-range (�i � j� � 5) 746Hydrogen bonds 66Dihedral angle restraints 172� 84� 88

Ramachandran Plot a (residues 157–246)Most favored region 72.2%Additionally allowed region 26.2%Generously allowed region 1.3%Disallowed region 0.3%

Rmsd from ideal geometryBonds (Å) 0.0018 � 0.00004Angles (°) 0.3436 � 0.0039Impropers (°) 0.2116 � 0.0055

Rmsd from mean structure (Å)Residues 157–296

Heavy 1.34Backbone 0.71

Residues 157–185, 188–194, 203–295Heavy 1.02Backbone 0.45

The 20 structures with lowest energy selected from 200 calculated final structures by CNS v1.1 (1) contain no distance violations greater than 0.3 Å and nodihedral angle violations greater than 3°. aData obtained by using the program PROCHECK-NMR (2).

1. Brunger AT, et al. (1998) Crystallography & NMR system: A new software suite for macromolecular structure determination. Acta Crystallogr D 54:905–921.2. Laskowski RA, Rullmann JAC, MacArthur MW, Kaptein R, Thornton JM (1996) AQUA and PROCHECK-NMR: Programs for checking the quality of protein structures solved by NMR.

Journal of Biomolecular Nmr 8:477–486.

Ohnishi et al. www.pnas.org/cgi/content/short/0812956106 5 of 5