www.sciencesignaling.org/cgi/content/full/5/245/ra74/DC1

Supplementary Materials for

Specificity of Linear Motifs That Bind to a Common Mitogen-Activated Protein Kinase Docking Groove

Ágnes Garai, András Zeke, Gergı Gógl, Imre Törı, Ferenc Fördıs, Hagen Blankenburg,

Tünde Bárkai, János Varga, Anita Alexa, Dorothea Emig, Mario Albrecht, Attila Reményi*

*To whom correspondence should be addressed. E-mail: remenyi@elte.hu

Published 9 October 2012, Sci. Signal. 5, ra74 (2012)

DOI: 10.1126/scisignal.2003004 This PDF file includes:

Fig. S1. Binding affinity measurements of MAPK–docking peptide interactions. Fig. S2. Characterization of MKK6-mediated activation of p38α in vitro. Fig. S3. Crystal structures of MAPK–docking peptide complexes. Fig. S4. Manipulation of MAPK–D motif interaction specificity. Fig. S5. Modification and design of MAPK binding specificity. Fig. S6. Selected examples of sequence-specific, intrapeptide H bonds from various protein-peptide complex structures. Table S1 caption Table S2. Summary of the in silico D-motif search on the human proteome. Table S3. Location of functionally relevant and characterized S/TP phosphorylation sites in D motif–containing MAPK substrates. References (74) to (81)

Other Supplementary Material for this manuscript includes the following: (available at www.sciencesignaling.org/cgi/content/full/5/245/ra74/DC1)

Table S1 (Microsoft Excel format). Lists of MAPK binding consensus motif matching sequences in the human proteome.

Figure S1. Binding affinity measurements of MAPK–docking peptide interactions (A) Direct (left) and competitive (right) fluorescence polarization (FP) measurements for reporter peptides. Error bars indicate uncertainty of the fit to a direct or to a competitive binding equation. Y axis displays arbitrary fluorescence polarization (FP) units. (B) Competitive titrations with unlabeled peptides using the JNK1-[TAMRA-pepNFAT4], p38α-[TAMRA-pepMEF2A] and ERK2-[CF-pepHePTP] reporter systems. Error bars indicate uncertainty of the fit to a competition binding equation. Peptide titration experiments without fits to a competitive binding equation did not show complete competition with the labeled reporter peptides, which we took to indicate no specific binding. Each measurement is representative of at least two sets of independent experiments where Kd values were calculated from triplicate data points. Triplicates were independently prepared samples that were assayed at the same time.

Figure S2. Characterization of MKK6-mediated activation of p38α in vitro (A) Characterization of the MKK6-p38α interactions by surface plasmon resonance (SPR). Binding of p38α to GST-MKK6 and GST-pepMKK6 surfaces was analyzed by equilibrium titration experiment using a Biacore 3000 instrument (GE Healthcare). Anti-GST antibody immobilized by amine-coupling to a CM5 chip was used to capture GST fusion proteins, GST fusion peptides, or GST protein as a control (to correct for nonspecific binding of the analyte). Surface saturation curves were plotted based on GST control surface corrected response units (RU) measured after 3 min of analyte injection when equilibrium was reached (RUeq). Panels B and C below show the result of an SPR experiment with GST-MKK6 and GST-MKK6_pep0 surfaces, respectively. [The GST-MKK6_pep0 construct lacks the pepMKK6 sequence as indicated in the schematic figure; see (A).] (D, E) When 100 µM or 50 µM p38α is injected on these two surfaces, only the MKK6 surface with an intact D-motif shows detectable binding. (The kinetic profiles show GST control surface corrected sensorgrams.) Note that removal of the N-terminal D-motif does not interfere with the protein's folding because GST-MKK6_pep0 shows similar activity as GST-MKK6 on a model substrate [see (G)]. The results are shown from at least two experiments with a representative binding curve or sensorgrams. The binding affinity determined in the SPR equilibrium titration experiments indicates weaker binding for the pepMKK6-p38α interaction than that determined in FP based measurements (~30 µM or 9 µM binding affinities, respectively). However, the similar values for the SPR measurements with full-length or pepMKK6 constructs (30-40 µM) indicate that the D-motif determines this interaction quantitatively. (The differences in the Kd values determined by the two different methods are likely due to the limitations of SPR because interactions are evaluated on a surface rather than under real solution conditions.) (F) MAPK phosphorylation profile of MKK6 and MKK1. Constitutively active MKK6 and MKK1 (0.5 µM) proteins (MKK6EE and MKK1EE and docking motif lacking versions thereof) were recombinantly expressed and purified, and phosphorylation of ERK2 and p38α (5 µM) was tested in kinase assays using 32P autoradiography. MAPKs were inactive to eliminate autophosphorylation; MKK6 and MKK1 enzyme concentration was 0.1 µM and 0.5 µM, respectively. MKK6EE_pep0 and MKK1EE_pep0 proteins lacked their D-motifs. SDS-PAGE gels were subjected to phosphor imaging. Gels show results of a typical experiment (N=2). (G) Forms of MKK6 and MKK1 with N-terminal D-motif deletions do not show impaired catalytic capacity. 1 µM constitutively active version of MKK6 (MKK6EE: S207E, T211E) and MKK1 (MKK1EE: S218E, S222E) were incubated with 50 µM myelin basic protein (MBP) under standard kinase assay conditions. Gels show results of a typical experiment (N=2). (H) MKK6EE and MKK1EE phosphorylates p38α and ERK2 on their activation loop residues, respectively (right). Enzymes were incubated with wild-type or with MAPKs with mutations in their activation loop residues (p38α: T180V and Y182F; ERK2: S185A and Y187F). The lack of 32P incorporation into the activation loop mutants indicate that earlier experiments indeed monitored activating phophorylation events on these MAPKs. Gels show results of a typical experiment (N=2). (I) Importance of the p38α activation loop sequence on MKK6-mediated activation of p38α. 0.1 µM MKK6EE_pep0 protein was incubated with 2 µM p38α substrates: wild-type (p38α) or with an activation loop mutant [p38α TEY, in which the TGY activation segment was mutated to TEY, see (J)]. The bar graph shows initial phosphorylation rates and error bars show the standard deviations from mean value; N=3 experiments. (J) Activation loop sequences of ERK2 and p38α. (K) A schematic model of MAP2K-MAPK interactions involving both the MAP2K active site (a.) and the MAPK docking groove (b.) mediated interactions. Extensive sequence variations between the two MAPK activation loops in addition to the intervening residues located near the phosphorylated threonine or tyrosine likely also contribute to the catalytic incompatibility between non-cognate MAP2K-MAPK pairs (see underlined residues in (J)).

Figure S3. Crystal structures of MAPK–docking peptide complexes Electron-density of docking peptides contoured at 1σ (on sigmaA-weighted omit maps) on the left. The right panel shows close-ups of MAPK-docking peptide interfaces. The core motifs in the peptide sequence that could be located in the electron density maps are underlined. Amino acids for which only main-chain atoms could be located are shown in brackets. H bonds are indicated with dotted lines. The N-terminal region of the MKK6 D-motif cannot be seen on the electron density map of the pepMKK6-p38α complex. This suggests that the positively charged residues may not need to adopt a single conformation when bound in the negatively charged CD groove.

Figure S4. Manipulation of MAPK–D motif interaction specificity (A) Direct titration to determine Kds for ERK2 docking groove mutants (ERK2 m2, ERK2 m4, ERK m6). To simplify measurements and to carry out direct binding experiments with pepRSK1 and pepMK2, these peptides were also synthesized with an N-terminal CF group. (B) Direct titration for determining Kds for p38α docking groove mutants (p38α m2, p38α m4, and p38α m6) with labeled peptides. The impact of amino acid swaps in the p38α docking groove on pepRSK1 or pepMK2 binding is shown below and compared to wild-type ERK2 and p38α on the bar graph below. p38 m6 has an ERK2-like docking groove: key positions in the top, base and hinge were mutated to corresponding amino acids from the ERK2 docking groove (as shown in Fig. 4C.) Binding isotherms on the right of the vertical line in (A) and (B) show binding of ERK2 or p38α with docking groove mutations to reporter D-motif peptides. The binding curves indicate that docking groove swap mutations have the expected effect not only on revD, but also on classical D-motif binding. ERK2 m6 bound pepHePTP with somewhat decreased affinity compared to ERK2 whereas p38α m6 bound pepMEF2A with similar affinity compared both to ERK2 and p38α. pepHePTP and pepMEF2A did not efficiently discriminate the two MAPK docking grooves, whereas neither of the MAPK docking groove chimeras bound to the JNK-specific pepNFAT4 reporter peptide. (C) Direct or competitive titration experiments with pepRSK1-CF, pepMK2-CF reporter peptides or with alanine replacement versions. In the competitive binding experiments, CF-labeled reporter peptides (starting out from an approximately 60-70% pre-bound complex sample) were competed with unlabeled peptides. (D) Competitive titration experiments with RSK1c and RSK1c_S/A,Q/A. The C-terminal fragment of full-length RSK1 containing the CAMK-type kinase domain and the MAPK binding revD-motif was expressed as GST-fusion protein and purified (RSK1c). Corresponding mutations to pepRSK1_S/A, Q/A were introduced by PCR into RSK1c and the protein (RSK1c_S/A,Q/A) was expressed and purified in the same manner as wild-type RSK1c. The binding affinities of these constructs were measured in similar competition binding experiments as with peptides on (C). MAPK discrimination factors for peptide and longer protein constructs are compared in the bar graph on the right. This demonstrates that alanine replacements of H-bond staple forming residues have a similar impact on the specificity of peptides and longer protein constructs. (E,F) Direct titration experiments to asses the impact of artificial H-bond staples on ERK/p38 binding specificity. Panels on the right show potential H-bonding interactions for the modified residues of peptides when they are bound in the MAPK docking groove. Each measurement is representative of at least two sets of independent experiments where Kd values were calculated from triplicate data points. Triplicates were independently prepared samples that were assayed at the same time on all panels. Error bars, if indicated, are from standard deviations from the mean.

Figure S5. Modification and design of MAPK binding specificity (A) Competitive titration to determine Kds for MAPK-pepNFAT4 and MAPK-pepNFAT4m binding. (B) Competitive titration to determine Kds for MAPK-pepMKK4 and MAPK-pepMKK4m binding. (C) Competitive titration experiments to determine Kds for MAPK-[pepSyntH-D] binding. (D) Direct titration experiments to determine Kds for synthetic revD-motifs. To simplify measurements and to carry out direct binding experiments with pepSynth-revD, pepSynth-revD m1, and pepSynth-revD m2, peptides were synthesized with an N-terminal CF group. Each measurement in (A) to (D) is representative of at least two sets of independent experiments where Kd values were calculated from triplicate data points. Triplicates were independently prepared samples that were assayed at the same time. (E) Binding specificity of full-length JIP1 and the JIP1_pepSynth-revD m2 chimera. JIP1 protein was expressed as a GST fusion protein and its binding to recombinant expressed and purified JNK1 was tested in GST pull-down assays. JIP1_Dock- has alanine replacements in the D-motif of JIP1. JIP1 requires an intact D-motif to bind JNK1 (left). The GST lane, in which glutathione beads were only loaded with the GST fusion tag, was a negative control. The schematic of the JIP1_pepSynth-revD m2 chimera is shown on the right. pepSynth-revDm2 binds p38α with the highest specificity and replacing the JIP1 D-motif with this motif causes the originally JNK-specific JIP1 protein to become p38-specific. GST-RSK1c serves as a positive control for ERK2 binding. MAPK binding in these experiments were detected by anti-FLAG Western-blots because ERK2 and p38α were expressed with C-terminal FLAG tags in transiently transfected HEK293T cells. Cell lysates were incubated with GST, GST-RSK1c, GST-JIP1, GST-JIP1_Dock-, and GST-JIP1_pepSynth-revD m2 baits. Anti-GST Western-blots show the amounts of baits loaded for JIP1 constructs for both MAPK pull down experiments. The load lane contains about 1/8th of the total cell lysate used per binding reaction whereas 1/5th of the total retained MAPK sample was loaded in the GST pull down lanes. Western blots show the results of a typical experiment (N=2).

Figure S6. Selected examples of sequence-specific, intrapeptide H bonds from various protein-peptide complex structures. Evolutionarily conserved residues are involved in the H-stapling interaction in the structure of the ERK2-pepHePTP (Gln39), SMAD3-pepSARA (Asn775, Asn778, and Glu781) and RAD51-pepBRCA2 (Thr1526 and Ser1528) complexes (PDB IDs: 2GPH, 1MK2, and 1NOW). Mutation of Thr1526 to alanine in the last example eliminates binding of BRCA2 to RAD51 (40). These solvent-facing linear motif residues involved in H-bond stapling are all evolutionarily conserved. (Amino acids mediating sequence-specific H-staples are shown in yellow and H bonds with red dashed lines. Zoomed-in areas are indicated by squares on the insets below showing the full protein-peptide complex.)

Table S1. Lists of MAPK binding consensus motif matching sequences in the human proteome. Sheets 1 to 5 list proteins harboring a particular D-motif (pepJIP-type, pepNFAT4-type, pepMKK6-type, and pepHePTP-type) or revD-motif (RSK/MAPKAPK-type) and the biological processes that are significantly enriched among those proteins. A protein is described by different identifiers (columns “A” to “C”) and its sequence (columns “D” and “H”). Details on the putative motif are listed in columns “E” to “G”. The motif sequence and potential S/TP target sites are highlighted in column “’H” in bold red, disordered amino acids in italics. The logPSSM score is shown in column “I” (sheets 1-4) or “J” (sheet 5). Motif conservation in orthologous proteins of seven defined species is listed in column “J” (sheets 1 to 4) or “I” (sheet 5): If an ortholog harbors a conserved motif, the motif sequence is shown (additionally marked with a * if the sequence lies outside of the aligned human motif region); otherwise it is marked with a 0. N/A indicates that no ortholog was found in the particular species. Additional motif information is given in column “K” (sheets 1-4) or “N” (sheet 5). GO biological process terms that are significantly enriched (q-value < 0.05) among the proteins of a particular list are shown in its top section. Identifier and name of the term are listed in columns “A” and “H”. Columns “B” and “E” indicate the number of proteins that are annotated with a particular term in total (group 1: 1000-9999, group 2: 100-999, group 3: 10-99, group 4: 1-9). The observed and the expected number of proteins from the list annotated with a particular term is shown in columns “F” and “G”, the p- and q-values in columns “C” and “D”. Column “I” contains the UniProtKB accession numbers of annotated proteins from the list. Sheet 6 lists signaling pathways that are significantly enriched among the combined set of all proteins from sheets 1 to 5, in the same structure as the enrichment sections of sheets 1 to 5.

Table S2. Summary of the in silico D-motif search on the human proteome Motif type Motif pattern a Number of

motifs/proteins after motif scan

Number of motifs/proteins after applying filters b

Number of motifs/proteins after logPSSM cutoff c

Enriched GO Biological Process terms and PID annotations with q-value < 0.05 d

pepJIP1 (D) [Ψ][P]..[Ф].[Ф] 4512/3725 454/430 72/70 Positive regulation of GTPase activity (6, 6, 19.1), MAPKKK cascade (7, 4; 12.1)

pepNFAT4 (D) [Ψ]..[Ф].[Ф].[Ф] 19093/10608 1545/1330 115/115 Transcription (20, 16, 5.6)

pepMKK6 (D) [Ψ].(3,4)[Ф].[Ф].[Ф] 42224/15128 3738/2753 132/131 Positive regulation of cell differentiation (10, 5, 5.1)

pepHePTP (D) [Ф]..[Ψ][Ψ].....[Ф].[Ф].[Ф] 422/408 35/35 35/35 -

revD [Ф].[Ф].(1,2)[Ф].(4,6)[Ф]..[Ψ][Ψ] 2324/2106 296/287 296/287 Cellular response to stress (25, 20, 2.6)

In total 68575/16656 6068/3561 650/595 p38 MAPK signaling pathway (19, 7, 2.8); EGF receptor signaling network (32, 17, 1.7); JNK signaling in the CD4+ TCR pathway (6, 3, 3.9)

a Ψ, Ф and . denotes arginine or lysine, hydrophobic, or any residues, respectively. (Numbers in brackets indicate the possible length of intervening regions.) b Motifs had to be within intracellular protein fragments, and could not be part of domains or annotated secondary structural elements or interconnecting loops. D-motifs had to have an S/TP MAPK target sequence located within 10-50 amino acids from the end of the motif. Motifs also had to be located in disordered protein regions. (Note that the S/TP site filter was not used for revD-motifs as these bind MAPKs in a different N-C orientation compared to D-motifs.) c logPSSM cutoff was used only for D-motifs. d The first number in the brackets denotes the number of proteins associated with significantly enriched GO Biological Processes terms (for individual motif types) or Pathway Interaction Database (PID) annotations (for the last row) for selected examples. The second number denotes the number of proteins not formerly known to contain docking motifs and the third number shows fold enrichment (found or expected number of proteins) compared to random sampling. The table shows that individual motif class lists are enriched in proteins that regulate GTPase activity, MAPKKK cascades or transcription (pepJIP1 and pepNFAT4-type motifs). These lists also reflect the dual role of MAPKs in regulating developmental as well as extracellular stimulus mediated processes: The pepMKK6-type list was enriched in proteins regulating cell differentiation, and the revD-motif list was abundant in proteins associated with cellular response to stress. Furthermore, PID annotation analysis on the combined list showed that docking motif containing proteins are indeed enriched in p38, EGFR, or JNK signaling pathways.

Table S3. Location of functionally relevant and characterized S/TP phosphorylation sites in D motif–containing MAPK substrates

Protein name Phosphorylation site

Linker length (from D-motif)

Functional relevance

c-Jun (human) Ser63 20 Promotes transactivation (74)

c-Jun (human) Ser73 30 Promotes transactivation (74)

JunD (human) Ser90 35 Promotes transactivation (75)

JunD (human) Ser100 45 Promotes transactivation(75)

ATF2 (human) Thr69 13 Promotes transactivation(76)

ATF2 (human) Thr71 15 Promotes transactivation(76)

ATF7 (human) Thr51 13 Promotes proteolytic degradation of the short isoform ATF7-4

(77)

ATF7 (human) Thr53 15 Promotes proteolytic degradation of the short isoform ATF7-4

(77)

Elk-1 (human) Thr353 31 Promotes transactivation (59)

Elk-1 (human) Thr363 41 Promotes transactivation(59)

Elk-1 (human) Thr368 46 Promotes transactivation(59)

Elk-1 (human) Ser383 61 Promotes transactivation(59)

Elk-1 (human) Ser389 67 Promotes transactivation(59)

Elk-1 (human) Thr417 95 Promotes transactivation(59)

Elk-1 (human) Ser422 100 Promotes transactivation(59)

JIP1 (human) Thr103 -56 Facilitates DLK dissociation from JIP1

(60)

NFAT4 (human) Ser163 9 Inhibits nuclear accumulation of NFAT4 (63)

NFAT4 (human) Ser165 11 Inhibits nuclear accumulation of NFAT4 (63)

HSF1 (human) Ser303 89 Involved in repression of HSF1 activity under control temperature

(78)

HSF1 (human) Ser307 93 Involved in repression of HSF1 activity under control temperature

(78)

IRS1 (human) Ser307 -549 Inhibits insulin-dependent phsophorylation of IRS1 (64)

MEF2A (human) Thr312 34 Required for transactivation by MEF2A (79)

MEF2A (human) Thr319 41 Required for transactivation by MEF2A (79)

MEF2C (human) Ser293 33 Required for transactivation by MEF2C (80)

HePTP (human) Thr66 13 Promotes dissociation of HePTP from its substrate

(81)

HePTP (human) Ser93 40 Promotes dissociation of HePTP from its substrate

(81)

Sites shown in gray background are within 10-50 amino acid from D-motif ends.