NMR study of the human NCK2 SH3 domains structure determination, binding diversity, folding and amyloidogenesis 4

More specifically, the first transition is involved in the collapse of the random-coil-like polypeptide chain into a non-native helical intermediate state mainly specified by local inter

4.1 Introduction

Many proteins fold into unique three-dimensional scaffolds composed of spatially organized polypeptide fragments that possess different secondary structures, including the α-helix, β-sheet, reverse turns, and loops Some proteins were found to fold into their native states via intermediates, and in particular, the partially folded states termed

“molten globule” were studied extensively (Kim, P S., 1982; Kim, P S., 1990; Kuwajima, K., 1989; Ptitsyn, O B., 1995; Jennings, P A., 1993; Colón W, Roder H., 1996; Song, J., 1998; Song, J., 1999; Wei, Z., 2005; Daggett, V., 2003) Many protein-folding models emphasize the hierarchical formation of the native-like substructures during the folding process On the other hand, recent studies of several proteins, and particularly β-lactoglobulin, implied that the folding of β-proteins may follow a non-hierarchical mechanism in which two major transitions are essential to reach the final native β-structure More specifically, the first transition is involved in the collapse of the random-coil-like polypeptide chain into a non-native helical intermediate state mainly specified by local interactions, whereas the second transition is associated with transformation into the native β-structure, with the helical conformation disrupted by long-range interactions (Kuwajima, K., 1996; Hamada, D., 1996; Hamada, D., 1997; Arai, M., 1998; Fujiwara, K., 1999; Chikenji, G., 2000; Kuwata K, 2001) However, the conformational and dynamic properties of such non-native helical states still remain poorly understood Moreover, the sequence determinants for transformation from the helical intermediate into the native β-structure have not been identified So far, more than 4,000 SH3 modular domains have been identified in a variety of organisms The SH3

transmitting as well as integrating cellular signals, mainly by binding to proline-rich short motifs (Mayer, B J., 2001; Musacchio, A., 2002; Liu J, 2006) From structural point of view, all SH3 domains share a common β-barrel fold comprising five β-stands, which are organized into two β-sheets The first, the last and the first half of the second β-strand constitute the first β-sheet, whereas the second half of the second β-strand, together with the third and fourth β-strands constitutes the second β-sheet The second β-strand seems

to play a critical role in coordinating the two β-sheets together into a classical SH3 fold (Fig 4.1A) The 380-residue adaptor protein Nck2, which is composed of three SH3

domains and one C-terminal SH2 domain, in vivo, functions to coordinate the signalling

networks critical for the organization of the actin cytoskeleton, cell movement, or axon guidance, by connecting membrane receptors to the multiple intracellular signalling networks in a “Tyr(P)/SH2/SH3/effector” manner (Li, W., 2001; Ran, X., 2005) The NMR structure of the first Nck2 SH3 domain was previously determined (Park S, 2006), and the structures of the second and third SH3 domains as well as the SH2 domain were determined by our lab (Ran, X., 2005; Liu J, 2006) SH3 domain is an ideal model system for investigating the folding mechanism of β-proteins due to its small all-β fold and the absence of disulfide bridge Unexpectedly, we discovered, without any co-solvent or stabilizer, that the wild type form of the first Nck2 SH3 domain could be reversibly converted into a stable helical state of equilibrium at pH 2.0, as detected by circular dichroism CD spectroscopy To gain further insights, we performed extensive CD and NMR investigations, and the results not only presented us an NMR conformational and dynamic view of this non-native helical state, but also led to the further identification of a four-residue sequence that appears to play an important role in transforming the helical

state into the native SH3 fold Recently, a study was published demonstrating that a native helical intermediate was also populated in the kinetic refolding of the Src SH3 domain (Li, J.S., 2007) Another article from the same group showed that the Ala45-Gly mutant of the Src SH3 domain could also form a stable helical intermediate at equilibrium, which were studied by CD, fluorescence, and x-ray scattering (Li, J., 2007)

non-4.2 Materials and methods

4.2.1 Expression and purification

The mutant of L25A-W26A-L27A-L28A was obtained by the same method as wild type, with the middle two primers changed to the corresponding primers encoding the four Alanine residues The expression, 15N/(15N/13C) labelling and purification of the wild type were performed as described in Common Materials and Methods Chapter The refolding of the wild type SH3 was carried out by slowly titrating with a 5mM phosphate buffer until the pH value 6.5 An NMR sample was then concentrated to ~1mM for further NMR experiments The mutant was purified as described above and performed dialysis in a 1mM phosphate buffer The sample was then concentrated to ~1mM

4.2.2 CD and NMR experiments

Concentrations of the proteins were around 15 μM in a 5 mM phosphate buffer The

far-UV spectra of the mutant SH3-1 were acquired in the range of 190-260nm at different pH values, at intervals of pH 0.5 The wild type SH3 far-UV spectra were acquired at pH 2.0,

pH 4.5 and 6.5

NMR experiments included 1H NOESY, TOCSY (250ms and 350ms), 1H-15N HSQC,

1H-15N HSQC-TOCSY, 1H-15N HSQC-NOESY, HNCACB, CBCA(CO)NH, TOCSY and HCCH-NOESY for the SH3-1 wild type at pH 2.0 All these experiments

HCCH-except HCCH-TOCSY, HCCH-NOESY were also carried out for the mutant T 1 and T 2

experiments and a {1H}-15N heteronulear NOE experiment were also carried out for the wild type at pH 2.0 and the mutant at pH 6.5 To test whether SH3-1 WT at pH 2.0 is recoverable to a native state, the lyophilised powder was first dissolved in pure water (around 10ml) with a pH value ranging from 2.0 to 2.5 Subsequently, the solution was gradually titrated by a 5 mM phosphate buffer The titration was stopped when the pH value reached 6.5 The sample was finally concentrated to ~1 mM The prepared sample was further used to do HNCACB, CBCA(CO)NH and relaxation experiments

4.2.3 Structure modelling of wild type (pH 2.0) and 4AlaMut

Dihedral angles used for structure modelling were predicted by the TALOS program (Cornilescu, G., 1999), based on the chemical shifts of HA, CA and CB for the wild type

at pH2.0 and HA for the mutant at pH 6.5

4.2.4 Relaxation experiments

15N T 1 values for the wild type at pH 6.5 were obtained by fitting HSQC spectra recorded with different relaxation delays of 10, 400, 100, 300, 200, 350 and 250 ms 15N T 2 values were determined with relaxation delays of 10, 30, 45, 60, 75, 90 and 150ms for the wild type at pH 6.5; For the wild type at pH 2.0, relaxation delays for 15N T 1 are: 10, 700, 100,

600, 200, 500, 300, 400 ms; for 15N T 2: 10, 60, 100, 130, 160, 200, 230, 260 ms For 4AlaMut, delays for 15N T 1 are: 10, 350, 50, 300, 100, 400, 200 ms while for 15N T 2: 10, 30,

60, 90, 120, 140, 160 {1H}-15N steady-state NOEs were obtained by acquiring spectra with and without 1H presaturation with a duration of 3 s plus a relaxation delay of 6 s at

800 MHz

4.3 Result

4.3.1 Bioinformatics and CD characterization

Previously, a high-resolution NMR structure was reported for the first human Nck2 SH3 domain (Park S, 2006), which unambiguously showed that it adopts the classical all-β fold common to all SH3 domains On the other hand, as shown in Figure 4.1A and B, prediction of its secondary structures by several computational programs, including GOR4 (Garnier, J., 1996), PHD (Rost, B., 1994), Predator (Frishman, D., 1996), SHIMPA96 (Levin, J., 1997) and SOPMA (HC GeourjonH, 1995), suggested that a large portion of the SH3 domain had a high intrinsic propensity to form helical conformations

In particular, the results of GOR4 and Predator indicated even higher percentage of helical conformation than the β-sheet in the SH3-1 domain (Fig.4.1B) However, as empirical information including the tertiary structure from the Protein Data Bank was utilised in these programs to improve the prediction accuracy, we used Helix2, which implements the Lifson-Roig helix-coil transition theory and only uses the interactions between adjacent residues (Rohl, C A., 1996; Rohl, C A., 1998) The prediction of Helix2 also showed that the SH3-1 domain contained a 21% helix fraction However, the prediction by AGADIR (Munoz, V.1, 1995; Munoz, V.2, 1995) only yielded 2.2% and

1.1% helix fractions at pH 6.5 and 2.0, respectively In general, it is widely accepted that helix formation is mainly driven by local interactions (Rohl, C A., 1996; Munoz, V., 1995), whereas β-sheet formation largely depends on long-range interactions and thus is context-dependent (Minor D L Jr, 19941; Minor D L Jr2) The discrepancy between the NMR structure and secondary structure prediction results inspired us to speculate that at least for the first hNck2 SH3 domain, the formation of the all-β SH3 native fold might be extensively driven by specific long-range interactions, which also function to override the intrinsic helix-formation propensity Therefore, destabilization of the tertiary packing would trigger a conversion of the all-β SH3 fold into a helical state As shown in Fig4.2,

at pH 6.5, the far-UV CD spectrum of wild type SH3 domain demonstrates some unique characteristics (Venyaminov, S Y., 1996) The minimal ellipticity at 203 nm and maximal ellipticity at 227 nm suggest the presence of some polyproline II structure in the native state of the protein Although all SH3 domains adopt the same three-dimensional fold, they have diverse far-UV CD spectra and dynamic properties, as previously documented (Wales, T E., 2006; Liu J, 2006) The origin of these unique properties is not well-understood, and some might result from the existence of the very long (>10 residues) and unique loop in all SH3 domains, which doesn’t adopt regular secondary structure, but makes extensive tertiary contacts with other parts of the SH3 domains (Liu

J, 2006) By contrast, it appeared that at pH 2.0, the SH3 domain switched into a partially folded state containing some residual helical conformation Variable secondary-structure fractions are given in the analysis of CD spectra by different programs: 15% helix, 35% turn/strand, and 5% random coil by CONTINLL; 12% helix, 46% turn/strand, and 42% random coil by SELCON3; and 5% helix, 59% turn/strand, and 36% random coil by

Figure 4.1 Bioinformatics analysis of secondary structures A) Secondary structure prediction

of the first hNck2 SH3 domain by computational programs GOR4, PHD, Predator, SIMPA96 and SOPMA “E” stands for the β-strand, “H” stands for helix and “C” stands for the random coil The four residues “LWLL” identified here are highlighted in yellow The SH3 family alignment

is shown at the bottom “l” stands for aliphatic; “p” stands for polar B) Bar plots for the secondary structure contents predicted by five different programmes

Figure4.2 Far-UV CD characterization of the first hNck2 SH3 domain

Far-UV CD spectra for the wild type at pH 6.5 (black); wild type at pH 2.0 (gray); 4AlaMut at pH 6.5 (dotted line) and 4AlaMut at pH 2.0 (broken line) Inset: pH-induced conformational changes

of the wild type as monitored by ellipticity at 222 nm

CDSSTR (Sreerama, N., 2000) Furthermore, the fraction of helix was estimated at ~11%

by assuming the coexistence of the helix and random coil (Li, J.S., 2007) Detailed CD investigations also revealed that the pH-induced conformational change was reversible, with the transition occurring from pH 4 to 2, as monitored at 222 nm Further subsequent NMR studies were conducted extensively and these results confirmed (consistent with the

CD observations) that the first Nck2 SH3 domain switched into a highly populated helical state at pH 2.0 This result raised the question of whether a sequence region could

be identified in the first hNck2 SH3 domain that served as a switcher to transform the helical state into the native SH3 fold In other words, mutation of this region could be anticipated to keep the SH3 domain permanently in the helical state To address this question, multi-sequence alignment of large number of SH3 sequences selected from all subfamilies was performed, leading to the identification of a four-residue region, Leu25-Trp26-Leu27-Leu28, on the second β-strand (Figure4.1A) Three of these residues were highly conserved in the majority of sequences Further prediction of secondary structures

of a variety of SH3 sequences by different programs also showed that this four-residue region always formed a β-strand Therefore, we experimentally constructed a mutant with these four residues changed to Ala residues Surprisingly, as seen in Figure4.2, the 4AlaMut protein showed very similar CD spectra at pH values of 6.5 and 2.0, suggesting that it completely lost the ability to undergo the pH-induced conformational changes observed in the wild type SH3 domain Importantly, even at pH 6.5, 4AlaMut had a far-

UV CD spectrum similar to that of the wild type at pH 2.0, indicating that 4AlaMut were

in a partially folded state bearing a similar fraction of the helical conformation even at a neutral pH Far-UV CD spectra were also acquired at different protein concentrations in

the range from 5 to 100 mM for the wild type and 4AlaMut Superimposition of the spectra showed no obvious difference, indicating they were concentration-independent and no significant aggregation occurs under these concentrations

4.3.2 NMR characterization and modelling

Isotope-labelled recombinant wild type hNck2 SH3-1 and 4AlaMut SH3-1 were also prepared for a detailed characterization by NMR No significant changes in cross-peak width and position were detected in their HSQC spectra, acquired at different concentrations (50–1,000 mM), indicating that the proteins are monomeric under these experimental conditions Successful sequential assignments for the wild type SH3 domain at both pH 6.5 and 2.0 were done based on analysis of the triple-resonance NMR spectra (TableS.2) As shown in Figure4.3A, the HSQC spectrum of wild type SH3-1 domain shows it is a typical well-folded β-protein, with spectral dispersions of ~2.7 ppm for the amide 1H and ~22 ppm for the 15N dimensions By contrast, the narrow HSQC spectral dispersion (~0.9 ppm for the amide 1H, and ~18 ppm for 15N) for the wild type at

pH 2.0 (Figure4.3B) and for 4AlaMut at pH 6.5 (Figure4.3C), indicate that their tight tertiary packing was severely disrupted Thus, they are only partially folded or unfolded under these experimental conditions For comparison, we downloaded the chemical shifts (BMRB accession number: 6854) previously used for the structural determination of the first hNck2 SH3 domain (Park S, 2006) Although our prepared hNck2 SH3-1 protein is four residues shorter at the N-terminus and one residue longer at the C-terminus, the Cα and Cβ chemical shifts of the majority of residues at pH6.5 obtained here are almost identical to those previously deposited, indicating that at pH 6.5, the SH3-1 domain we

Figure 4.3 1 H- 15 N NMR HSQC spectra for three forms of SH3-1: A) Wild type at pH 2.0; B)

4AlaMut at pH 6.5; C) Wild type at pH 6.5 All spectra were acquired on an 800 MHz NMR spectrometer at 278 K The sequential assignments were labelled for all spectra.

studied here adopts the same solution structure as previously determined (Park S, 2006) Because NMR chemical-shift deviations from those expected for random coils are very sensitive indicators of protein secondary structure (Merutka G, 1995; Schwarzinger S, 2000; Wishart D S., 1998; Avbelj F., 2004), Cα and Hα chemical-shift deviations from the random-coil values previously reported were calculated (Merutka G, 1995; Schwarzinger S, 2000) As clearly indicated by the Cα deviation (Figure4.4A), the wild type SH3 domain adopts a β-like secondary structure at pH 6.5, whereas upon lowering the pH value to 2.0, it switched into a helical conformation More specifically, it appeared that this β-to-helix transition occurred in the region of the majority of the RT-loop, the second and third β-strands As shown in Figure4.3B, the large Hα deviations confirmed the formation of highly populated helical fragments with a similar pattern in both the wild type at pH 2.0 and 4AlaMut at pH 6.5 The NMR conformational shift was extensively used to estimate the population of the secondary structure in partially folded peptides and proteins (Ramirez-Alvarado, 1996; Jourdan M, 2000) Recently, it was shown that conformational shifts of amino acids were highly dependent on the extent of solvent exposure (Avbelj F., 2004) Statistically, the chemical-shift deviations of residues exposed to solvent are smaller than those of buried residues in the same secondary structure According to their narrow HSQC dispersions, the wild type SH3-1 domain at

pH 2.0 and 4AlaMut at pH 6.5 lacked tight tertiary packing, and as a result, the majority

of residues were anticipated to be exposed to the bulk solvent Therefore, it is reasonable

to use the average chemical shifts of the exposed helix residues (Avbelj F., 2006) as a reference for comparisons In this regard, by assuming a two-state (random coil and helix) folding model, the helix populations can be approximately represented by the ratio