Addressing the role of the α-helical extension in the folding of the third PDZ domain from PSD-95

PDZ domains are one of the most important protein-protein interaction domains in human. While presenting a conserved three dimensional structure, a substantial number of PDZ domains display structural extensions suggested to be involved in their folding and binding mechanisms. The C-terminal α-helix extension (α3) of the third PDZ domain from PSD-95 (PDZ3) has been reported to have a role in function of the domain as well as in the stabilization of the native fold. Here we report an evaluation of the effect of the truncation of this additional helix on the folding and unfolding kinetics of PDZ3. Fluorescent variants of full length and truncated PDZ3 were produced and stopped-flow fluorescence measurements were made under different experimental conditions (pH, ionic strength and temperature) to investigate the folding kinetics of the respective variant. The results show that folding of PDZ3 is robust and that the mechanism is only marginally affected by the truncation, which contributes to a destabilization of the native state, but otherwise do not change the overall observed kinetics. Furthermore, the increase in the unfolding rate constants, but not the folding rate constant upon deletion of α3 suggests that the α-helical extension is largely unstructured in the folding transition state.

The role of α3 in the molecular architecture of PDZ3 was also analyzed from a thermodynamic perspective. In fact, it was shown that the deletion of α3 resulted in an increased propensity for this protein to aggregate 19,22 , because of a destabilization of the native state as well as an overall expansion of the polypeptide chain 20 . Here, to obtain a better picture of the role of the C-terminal extension we continue these studies by investigating the folding pathway of the truncated variant of PDZ3 (PDZ3Δα3). A comparative kinetic characterization of the folding of PDZ3 and PDZ3Δα3 reveals that folding is surprisingly robust. The overall kinetic folding mechanism appears similar when the protein is challenged with acidic conditions, different ionic strengths and different temperatures.

Results and Discussion
In order to test the effect of the α-helical extension in the folding of PDZ3, in analogy to our previous work on this protein family [23][24][25][26][27][28] , we used a fluorescent pseudo-wild type variant of PDZ3, where a Phe residue is replaced with a Trp in position 337 (F337W). Thus, all experiments on PDZ3 and PDZ3Δα3 were performed using Trp337 as a probe to monitor fluorescence change upon (un)folding.
The folding and unfolding of PDZ3 and PDZ3Δα3 were investigated by stopped flow kinetics, by rapidly mixing the native protein with solutions containing different concentrations of urea (unfolding) and by diluting the urea denatured protein with buffer (refolding). Folding was then recorded by monitoring fluorescence emission >320 nm. In all cases, for both PDZ3 and PDZ3Δα3, observed time courses were consistent with a single exponential decay (Fig. 2), suggesting the absence of low-energy transiently populated intermediates 29 .
The semi-logarithmic plot of the observed (un)folding rate constants as a function of urea concentration (chevron plot) for PDZ3 and PDZ3Δα3, recorded in 50 mM sodium phosphate buffer pH 7.2 and at 25 °C, are reported in Fig. 2. It is evident that while the truncation of α3 does not affect the folding arm of the chevron, there is a detectable increase of the unfolding rate constant of about 2.5 fold, corresponding to a relatively marginal destabilization of the native state of about 0.54 kcal mol −1 . Since the folding rate constant is the same in PDZ3 and PDZ3Δα3, it follows that the deletion of α3 has a negligible effect on the structure of the transition state and that folding of α3 in PDZ3 occurs downhill the main barrier 30 .
Valuable parameters to obtain overall structural features of folding intermediates and transition states are the so-called kinetic m-values 31 , which represent the slopes of the folding (m F ) and unfolding (m U ) branches of the chevron plots. The m values are correlated with the changes in accessible surface area between the two ground states (native and denatured, respectively) and the transition state between them. The comparison between the chevron plots of PDZ3 and PDZ3Δα3 at neutral pH is reported in Fig. 2 and shows that truncation of the α-helical extension has little effect on the folding and unfolding m-values as the apparent branches of the chevron plot appear parallel. Thus, whilst it has been previously observed that PDZ3Δα3 is expanded compared to PDZ3 20 , such expansion cannot be detected from folding kinetics, which appear very similar for the two proteins.
To obtain a comprehensive comparison of the folding mechanisms of the two proteins, we investigated the folding at different pH conditions (ranging from pH 2.0 to 7.2), ionic strengths (from 20 mM to 1 M) and temperatures (from 15 °C to 42 °C). The resulting experiments are shown in Fig. 3. In a previously published series of pap ers 7,25,26,28,32,33 , we have shown how the folding of PDZ domains may vary from an apparent two-state (V-shaped chevron plot) to more complex scenarios (curved chevron plot), under different experimental conditions (e.g. different pH values or in the presence of a stabilizing salt) or with site-directed conservative mutants. To rationalize these apparently different mechanisms, we showed that a global quantitative analysis of all folding data can The conserved structure of PDZ domains is colored in grey, the additional C-terminal α-helix is highlighted in blue and the backbone of the peptide is colored in red. Residue number 337, which was mutated into Trp as a probe for fluorescence monitored folding studies, is highlighted in sticks.
be described in terms of a reaction scheme involving a single set of intermediates and transition states. In this context, the presence/absence of curvatures in the chevron plot would arise from selective (de)stabilization of the activation barriers and ground states, when varying solvent composition or protein sequence.
In the case of PDZ3, we observed that, whilst the chevron plot is essentially V-shaped at physiological conditions, it displays a detectable curvature when the protein is destabilized by lowering pH or by site-directed mutagenesis 32 . Analogously, in this work, we observed a clear curvature in the unfolding arms of the chevron plots of PDZ3Δα3, whilst in the case of PDZ3 a deviation from linearity could be clearly observed at acidic pH conditions. This finding suggests the presence of a high-energy folding intermediate between two different transition states, namely a denatured-like transition state TS1 and a native like transition state TS2. At increasing denaturant concentrations there is a switch in rate limiting transition state from TS1 to TS2. The folding of both PDZ3 and PDZ3Δα3 carries the signature of such an intermediate, indicating that the truncation of the C-terminal extension does not contribute to a detectable distortion of the overall folding mechanism.
By applying a three-state model 34 the data reported in Fig. 3 were used to calculate the folding rate constant k F and two different unfolding rate constants k U1 and k U2 associated with the two transition states TS1 and TS2, respectively. Furthermore, quantitative analysis of the chevron plots allows measuring the associated Tanford β-values, which reflect the relative positions of TS1 and TS2 along the reaction coordinate with regard to compactness. Tanford β-values for PDZ3Δα3 were 0.4 ± 0.1 for the denatured like transition state TS1 and 0.82 ± 0.03 for the native-like TS2. These values appear very similar to those of PDZ3 (0.5 ± 0.2 and 0.85 ± 0.07), showing that truncation of α3 does not distort the overall folding transition state structures. In experiments performed as a function of temperature and ionic strength, we could not observe a detectable curvature in the unfolding arm of the chevron plot for either PDZ3 or PDZ3Δα3. This observation indicates that, in both proteins, the transition state TS1 represents the rate limiting step at every concentration of urea experimentally accessible and no change in rate determining barrier can be detected. Accordingly, those data were fitted to a standard two-state chevron 29 .
The dependence of the microscopic folding and unfolding rate constants on ionic strength, temperature and pH are reported in Fig. 4. In all cases, the folding rate constant k F is the same within experimental error for both of PDZ3 and PDZ3Δα3 when varying the experimental conditions, indicating that the structure of the folding transition state is very robust. Furthermore, the overall behaviour of both unfolding rate constant k U1 and k U2 appear very similar, displaying the same dependence on temperature, ionic strength and pH. On the basis of these observations we conclude that the overall folding mechanism of PDZ3 is not substantially perturbed and, while destabilized, is not affected by the deletion of helix α3. A schematic energy diagram summarizing the effect of the deletion of helix α3 is presented in Fig. 5.

Conclusions
It has been previously suggested that the presence of an additional helix, the C-terminal α3 extension in PDZ3 is critical in modulating its function as well as aggregation properties 14,17,[19][20][21][22] . Furthermore, truncation of α3 has been shown to lead to an overall expansion of the polypeptide chain, while exhibiting no significant changes in the overall structure 20 . In this work, we completed these analyses by comparing the folding and unfolding of PDZ3 and PDZ3Δα3 under a variety of experimental conditions. Our data reveal that the mechanism by which the denatured state of PDZ3 achieves its native conformation is rather robust. In fact, whilst a destabilization of the native state could be observed, there is no effect in the folding rate constant at any of the experimental conditions explored, indicating that the C-terminal extension α3 forms downhill of the main barrier for folding. This finding appears to contrast, at least in part, the observed stabilization of an equilibrium intermediate in PDZ3Δα3, which leads to a more complex folding behaviour as well as to an increased propensity for the protein to aggregate 19,35 . It is important to notice, however, that the intermediate previously identified by Murciano-Calles et al. by differential scanning calorimetry, is most likely a distinct species from the meta-stable state responsible for the curvature of the chevron plot of PDZ3, which is a high-energy species that never accumulates. On the basis of our experimental work, we conclude that the main folding transition state is by and large unaffected by the deletion of α3 and the overall folding mechanism of PDZ3 appears robust. Future work based on protein engineering will further elucidate the mechanistic details of the role of helix α3 in the folding and function of PDZ3.

Methods
Protein Expression and Purification. All experiments were performed with a pseudo-wild type (pWT) PSD-95 PDZ3 containing a Trp at position 337. The constructs pWT-6His-PDZ3F337W and pWT-6His-PDZ3F337W-Δα3 coding for residues 309 to 402 and 309 to 394, respectively, of PSD-95 PDZ3 were used for protein expression and purification as previously described 27 Fig. 3. The folding rate constant k F is very similar for both PDZ3 and PDZ3Δα3 at all the experimental conditions explored. Data recorded at different pH conditions, left column, where fitted to the Henderson-Hasselbalch equation, returning a single transition with a robust apparent pK a of about 3.5. In all cases, an increase in the microscopic unfolding rate constant could be observed, consistent with a destabilization of the native state. The overall dependence of the unfolding rate constants were however unaffected by the truncation of α3.
concentrations of sodium phosphate buffer at pH 7.2 and addition of sodium chloride to obtain ionic strengths ranging from 20 mM to 1 M. Data Analysis. Kinetic traces were fitted with a single exponential decay using Applied Photophysics software to obtain the observed rate constant k obs . The logarithmic values of k obs were plotted versus urea concentration (chevron plot) and a equation describing a three-state folding reaction was used to estimate microscopic folding rate constants: where k F is the folding rate constant, K part is a partition constant between k U1 and k U2 , the two unfolding rate constants referring to the denatured-like and native-like transition states TS1 and TS2, respectively. k U2 , is equal to k U1 /K part . The dependence of the logarithm of k F , k U1 and K part was assumed to be linearly dependent on denaturant concentration, with a respective slope denoted as m F , m U1 and m part , respectively. The Tanford β-values were calculated as β Ti = m i /m D-N where m i is the m value at a given state i, and m D-N the total m value between the denatured and native state. Whilst the experiments performed as a function of pH allowed to infer the presence of both transition states for both PDZ3 and PDZ3Δα3, for data recorded as a function of temperature and increasing ionic strength we could not identify a major curvature in the chevron plots. Accordingly, in those cases, data were fitted to a standard two-state chevron 29 .

Figure 5.
Energy diagram for the folding of of PDZ3 (black) and PDZ3Δα3 (grey). Deletion of the of helix α3 affects slightly the stability of the native state without altering substantially the major folding pathway, which appears rather robust to the truncation.