Ins_(1,4,5)P₃ was purchased from Sigma and Cayman Chemical. (dihexanoyl- or dioctonoyl-) PtdIns₍₄₎P₁, PtdIns₍₅₎ P₁ and PtdIns_(4,5)P₂ were purchased from Cayman Chemical. Phosphopeptides derived from β₃CT(bi-β₃: ⁷³⁶RAKWDTANNPL(pY)KEATSTFTNIT(pY)RGT⁷⁶² and c-β₃: ⁷⁵⁰ATSTFTNIT(pY)RGT⁷⁶²) were synthesized by NEO-Peptide.

The Shc PTB domain (17 - 207) containing the pET15b vector was transformed into a Rosetta (DE3) cell line to pursue optimal expression levels [4] . SH2 domain (380 - 473) was cloned into NdeI and XhoI sites in pET21b vector and then transformed into a BL21 (DE3) cell line. The vector of pET28a-full length Shc (1 - 473), generously provided by Dr. Ladbury (University of Texas MD Anderson Cancer Center, TX), was expressed in a BL21 (DE3) cell line as well. Typically, 10 ml overnight culture was used to inoculate 1 liter LB broth. Cells were grown at 37˚C until OD₆₀₀ reached 0.6. At this point the culture was cooled to room temperature and then the overnight overexpression was induced by 1 mM isopropyl-β-D-thiogalactopyranoside. Harvested cells were kept at −20˚C for future purification or resuspended in lysis buffer supplemented with protease inhibitors. Cells were further lysed by french press and the cell debris was removed by centrifugation at 18,000 rpm at 4˚C. The N-terminal His-tag fused PTB domain and full length Shc, and the C-terminal His-tag fused SH2 domain were first purified under native conditions with Ni-NTA resin (QIAGEN) using the standard protocol followed by a buffer exchange to 50 mM Tris, pH 7.5, 100 mM NaCl, 5 mM β-mercaptoethanol in a Superdex-75 column (GE Healthcare) where all three recombinant proteins showed a proper folding in monomeric state (Supplementary Figure S1). The overexpression of the target protein and the purified protein were confirmed by SDS-PAGE and the bands of the expected mass were observed. The determination of protein concentration by UV measurement was consistent with Bradford assay.

Calorimetric measurements of a full length Shc, PTB domain, and SH2 domain to β₃ phosphopeptides or PtdIns were performed on a low volume Nano ITC (TA Instruments). All experiments were carried out in a buffer of 50 mM Tris, pH 7.5, 100 mM NaCl, 5 mM β-mercaptoethanol (unless described otherwise) at 12˚C or 25˚C with a stirring speed of 250 rpm to minimize the precipitation of the protein. 300 second time intervals were set between injections. The concentration of protein samples was determined by UV measurements. The titrants (β₃ phosphopeptides or PtdIns) were prepared in the same buffer by the dilution of higher concentration stocks. The concentrations of the protein ranged from 30 μM to 150 μM. The concentration of protein was examined after each ITC experiment by UV to ensure the stability of proteins under the test condition. The analysis of the data was done with Nano Analyze Software (TA Instruments) suite using an “independent” model. In all cases, a stoichiometry of 1 ± 0.15 was revealed for the interaction between protein and β₃ phosphopeptides/PtdIns.

¹⁵N-labelled recombinant proteins were overexpressed in M9 medium supplemented with ¹⁵NH₄Cl as the sole nitrogen source. The same purification procedures as described above were followed. NMR samples of 100 - 200 µM were prepared in a buffer of 50 mM Tris, pH 7.5, 100 mM NaCl, 5 mM β-mercaptoethanol, 7% D₂O, 1 mM 4,4-dimethyl-4-silapentane-1-sulfonic acid as the internal standard. ¹⁵N-HSQC titration experiments were performed on a Varian/Agilent Inova 600 MHz spectrometer equipped with inverse-triple resonance cryo-probe at 35˚C. All spectra were processed with NMRPipe [22] and analyzed in CCPNmr Analysis software suite [23] .

We started our investigation by studying Shc binding to Ins_(1,4,5)P₃ using ITC. We were greatly surprised that we were not able to reproduce the data reported by George and co-workers, who have reported a tight dissociation constant (K_d of 0.7 µM) for the full length Shc [21] . No binding was found under any conditions tested (to have a better control on all possible variables, we have purchased Ins_(1,4,5)P₃ from the same supplier (Cayman Chemical) as reported in the paper). The conditions we tested include: i) Shc PTB domain at 12˚C and 25˚C in several different buffers (including exactly the same as the one described by George, which is 50 mM Tris, pH 8.0, 200 mM NaCl, 1 mM Dithiothreitol); ii) Shc full length at 12˚C and 25˚C; and iii) Shc SH2 domain at 12˚C and 25˚C (typical raw titration data is shown as an example in the Supplementary Figures S2(a)-(c)). In the attempt to interpret our negative findings, we hypothesized that the heat of the reaction was too small, actually below the detection range of our NanoITC calorimeter. Thus, we turned to NMR as the method of choice due to its endogenous ability to capture the interactions with minimal or no enthalpy changes. In this case, we performed ¹⁵N- HSQC chemical shifts mapping experiments by titrating the unlabeled Ins_(1,4,5)P₃ into the ¹⁵N-labled Shc PTB domain. Under the conditions tested (see Materials and Methods for details), we only observed minimal chemical shift perturbations for a handful of residues in the ¹⁵N-HSQC spectra even at the maximum of lipid to protein ratio (15 to 1, Supplementary Figure S2(d)). Uncertain about these findings, we decided to choose different lipid mimetics for further investigation.

With minimal success observing Shc interaction with Ins_(1,4,5)P₃, representing exclusively the polar head-groups of phosphatidylinositols, we wondered whether the aliphatic chains of the lipids were necessary for the measurability of the binding by ITC and NMR. Long lipid chains are not exactly soluble in aqueous solution. Thus, to perform the binding studies described below, we restricted the length of aliphatic chains to six carbons in dihexanoyl, or eight in dioctonoyl phosphatidylinositol derivatives and concentrated on PTB domain of Shc, previously shown to interact specifically with PtdIns₍₄₎P₁ and PtdIns_(4,5)P₂ [15] . Binding affinities for the apo PTB domain, originally examined in the unilamellar liposomes centrifugation assay, were reported to be 52 μM for PtdIns₍₄₎P₁ and 140 μM for PtdIns_(4,5)P₂. Binding affinity of the PTB domain complexed with TrkA peptide was reported to be even weaker, with a roughly estimated K_d of 450 μM for both lipids.

First, we employed ITC to determine the binding affinities of the PTB domain to three different phosphatidylinositol containing soluble probes (see the Materials and Methods section for details). We measured a K_d of 95 µM and 125 µM for PtdIns₍₄₎P₁ and PtdIns_(4,5)P₂ respectively (titration curves and isotherms are presented in Figure 1(a) and Figure 1(b)). ITC was not able to detect the binding PtdIns₍₅₎P₁ (titration curve is shown in the Supplementary Figure S3), thus confirming the specificity of interaction reported by Zhou and coworkers. The ITC measurements demonstrated similar binding affinity or PtdIns_(4,5)P₂ as compared to the affinity previously determined by centrifugation assay data analysis, though PtdIns₍₄₎P₁ appeared to bind about two times weaker than reported before [15] . Small amounts of heat are consumed upon binding (in addition to the relatively large heat of solvation), demonstrating endothermic interactions at 25˚C that are driven predominantly by entropy. Considering these data, it is not entirely surprising that we were only able to observe minimal Shc PTB binding with Ins_(1,4,5)P₃ alone in ITC. The complete thermodynamic analysis of the above interactions is presented in the Table 1.

To find out the molecular details of these weak Shc-PtdIns interactions, we next employed NMR and performed ¹⁵N-HSQC chemical shifts mapping experiments which are especially useful to define weak binding [24] . Two non-labeled PtdIns derivatives, PtdIns₍₄₎P₁ and PtdIns_(4,5)P₂, demonstrated to interact with Shc PTB by ITC experiments, were titrated into the samples containing ¹⁵N-labeled apo PTB domain. Chemical shift perturbations, associated with the probes binding, were monitored. Overall, the observed shifts in Shc PTB resonance frequencies were small. However, they were concentration dependent, reproducible, and similar among the two PtdIns tested. Furthermore, the same residues were affected as the ones in Ins_(1,4,5)P₃ titration described above (Supplementary Figure S2(d)), but the shifts were more pronounced in case of PtdIns_(4,5)P₂, even at 5 to 1 lipid to protein ratio as compared of 15 to 1 ratio for Ins_(1,4,5)P₃. The close-ups of ¹⁵N-HSQC spectra of PTB in the presence of PtdIns_(4,5)P₂, exemplifying the most shifted resonances, are shown in Figure 2. These chemical shifts perturbations occur at three major binding regions. The first one, represented by the residues Q¹⁴⁸, S¹⁴⁹ and A¹⁶⁸, overlaps with the classical phospho-tyrosine binding pocket of the protein-peptide binary complex. However, the canonical PTB fold cannot be formed without the stabilizing hydrophobic residue (pY-5) of the target peptide. Thus, in apo PTB domain Ins_(1,4,5)P₃ could only establish intermediate dynamic contacts with the amine functional groups of R⁶⁷ or Q¹⁴⁸ side chains. The second binding site, including the residues L⁶³, Q⁷⁶ and S¹⁰⁷, is formed at the interface of the second helix, the loop connecting the first beta strand with the second helix and the beginning of the second beta strand. This second PtdIns binding site is distinct from the second novel phospho- tyrosine binding site we previously described for Shc PTB-bi-β₃ complex [4] , although it includes several overlapping residues. The third PtdIns binding site is restricted to a single but the most perturbed residue of the spectra, F³⁰, which is located within the unstructured Shc PTB N-terminus. The chemical shift perturbations in ¹⁵N- HSQC spectrum of apo Shc PTB upon PtdIns_(4,5)P₂ titration, plotted against the residue number, are presented in Figure 2. The inset demonstrates affected residues mapped onto PTB surface, which is colored according to the absolute value of the corresponding chemical shift perturbations.

Two of the Shc three distinct domains, PTB and SH2, have the potential to bind phospho-tyrosines and/or phos- pholipids. However, in all the experiments performed we did not observe any indications of Shc SH2 domain interaction with either PtdIns or β₃ integrin derived tyrosine(s)-phosphorylated peptides. Thus, we focused our investigation on studying the PTB domain alone and comparing it to the full length Shc in order to understand the thermodynamic forces driving the interactions. We started with a bi-β₃ peptide, which was used previously for the structural characterization of the Shc PTB/integrin binary complex [4] . We found that at 25˚C the K_d of this interaction was 5.1 μM (ΔG = −30.17 kJ/mol) and it was predominantly driven by enthalpy (ΔH = −22.98 kJ/mol). A representative calorimetric isotherm and the corresponding titration curve, depicted in Figure 3(a), demonstrate that the interaction is exothermic at 25˚C, releasing the heat upon peptide binding with a stoichiometry about 1. The driving force of this complex formation is very different from the one defined for the TrkA peptide (enthalpy vs. entropy), and it shows a higher specificity, even though the binding affinity is about twenty five times weaker (5.14 µM vs. 0.19 µM). We have speculated that the secondary binding site for pY⁷⁴⁷ within the complex might be responsible, at least in parts, for the difference. To find out whether this was the case, we ran ITC experiments for the smaller c-β₃ peptide containing only one terminal pY⁷⁵⁹ (Figure 3(b)). Our reasoning appeared to be wrong. The shorter (13 vs. 27 residues) mono-phosphorylated peptide had an even higher binding affinity with the K_d of 0.34 µM (ΔG = −36.86 kJ/mol) and an even larger favorable enthalpy term (ΔH = −25.41 kJ/mol). The thermodynamic analysis of the above interactions is presented in Table 2.

Next, we decided to find out how the Shc solvent-accessible surface area was affected during the reaction. This was accomplished by investigating the change in constant pressure heat capacity (ΔCp). ΔCp is given by the slope of the linear regression analysis of ΔH plotted vs. temperature, and it is often independent of temperature within the narrow physiological range (reviewed in [25] ). Thus, two temperature points, 12˚C and 25˚C, would provide an estimate for the slope. For binding reactions, negative ΔCp is expected to demonstrate the reduction in protein surface area in contact with the solvent. The larger and more negative values reflect a larger surface area buried upon complex formation. Through this correlation, structural information of the macromolecular complex could be inferred from the thermodynamic parameters. As expected, the longer bi-β₃ peptide in complex with Shc PTB domain demonstrated a more negative ΔCp of −0.88 kJ/mol∙K as compared to the shorter c-β₃ peptide represented by ΔCp of −0.09 kJ/mol∙K (see Table 2 for details). This correlates well with the larger solvent-accessible surface area buried upon binding of the longer peptide. As we already established that the Shc SH2 domain does not bind to either of the above peptides, we decided to confirm that the full length Shc behaves similarly to PTB domain alone. To our great surprise that was not the case. We found that while bi-β₃ peptide in complex with full length Shc showed a quite similar ΔCp of −0.96 kJ/mol∙K, binding of c-β₃ peptide resulted in a positive ΔCp of 1.04 kJ/mol∙K. That means that upon complex formation with c-β₃ peptide more of the Shc solvent-accessible surface area was exposed. The only reasonable explanation of this phenomenon that we can imagine is the unraveling of the intramolecular interactions between the domains within Shc itself.

We were also curious about the relation between binding sites for PtdIns and β₃-peptides (bi-β₃ and c-β₃) on Shc PTB surface. If these sites do not overlap, the presence of β₃-peptide should not interfere with the interaction between PTB and PtdIns; thus, PtdIns should cause concentration-dependent perturbations in the spectra of the PTB-β₃-peptides complex. In case of a competitive binding between PtdIns and β₃-peptides to PTB, since Shc’s affinity to bi-β₃ and c-β₃ is much higher in comparison to its affinity to phospholipids, it is reasonable to speculate that integrin tyrosine(s)-phosphorylated CTs can easily replace PtdIns from the partially overlapping binding sites on the Shc surface. We have proved the latter scenario by using NMR. First, we prepared protein samples of the PTB domain with PtdIns added at protein-PtdIns ratios of 1:0, 1:3, 1:5 and 1:15. Then bi-β₃ or c-β₃ was added to the sample at a 1:2 protein-to-peptide ratio and ¹⁵N-HSQC experiments were performed. The ¹⁵N- HSQC spectra of both Shc PTB-bi-β₃ and Shc PTB-c-β₃ complexes in the absence and the presence of PtdIns₍₄₎P₁, PtdIns_{(4, 5)}P₂or Ins_{(1, 4, 5)}P₃overlap perfectly (an example of the overlapping spectra is presented Supplementary Figure S4). Therefore, we have confirmed that the binary interfaces between Shc PTB and phosphopeptides remained the same in the presence of phospholipids added in excess.Also, bound peptides did not allow PtdIns to occupy their partially accessible (according to the chemical shifts mapping experiments summarized in Figure 2) potential binding sites on the surface of PTB domain. The remaining question was whether PtdIns could affect the thermodynamic profiles of Shc interaction with phosphorylated peptides. To test this, we investigated the binding of full length Shc and Shc PTB domain to bi-β₃ and c-β₃ peptides in the presence of Ins_(1,4,5)P₃ by ITC. As expected, our data indicate that the protein samples over-saturated by Ins_(1,4,5)P₃ at 3 to 1 lipid to protein ratio were not much different in Shc binding to either of the peptides. The most noticeable difference, observed in a single case, was the perturbation in the thermodynamic profile of the full-length Shc titrated by bi-β₃, where the unfavorable 3 kJ/mol change in enthalpy was almost compensated by the favorable change in entropy term, resulting in a slight reduction of K_d value (see Table 3). The impact from Ins_(1,4,5)P₃ on the peptide interaction with Shc was also evidenced in this case by the appearance of small negative peaks ahead the dominating positive peaks in the titration curve (presented in Figure 4), when the reaction was close to saturation. In the absence of the lipid mimetic, the small negative peaks were not present. This phenomenon is quite specific, as no similar patterns were witnessed in any other ITC experiments performed for studying Shc-bi-β₃/c- β₃ binding. Lastly, to complete the investigation and to confirm the inability of PtdIns to replace the phosphopeptides, we performed titrations of PtdIns into Shc PTB complexed with phosphorylated peptides (bi-β₃ or c-β₃, at a 2:1 peptide to protein ratio). We then monitored potential changes in HSQC spectra by NMR or in calorimetric isotherms by ITC. As expected, no indications of binding were observed by either method.