Cryptic binding properties of a transient folding intermediate in a PDZ tandem repeat

Abstract

PDZ domains are the most diffused protein–protein interaction modules of the human proteome and are often present in tandem repeats. An example is PDZD2, a protein characterized by the presence of six PDZ domains that undergoes a proteolytic cleavage producing sPDZD2, comprising a tandem of two PDZ domains, namely PDZ5 and PDZ6. Albeit the physiopathological importance of sPDZD2 is well‐established, the interaction with endogenous ligands has been poorly characterized. To understand the determinants of the stability and function of sPDZD2, we investigated its folding pathway. Our data highlights the presence of a complex scenario involving a transiently populated folding intermediate that may be accumulated from the concurrent denaturation of both PDZ5 and PDZ6 domains. Importantly, double jump kinetic experiments allowed us to pinpoint the ability of this transient intermediate to bind the physiological ligand of sPDZD2 with increased affinity compared to the native state. In summary, our results provide an interesting example of a functionally competent misfolded intermediate, which may exert a cryptic function that is not captured from the analysis of the native state only.

1. INTRODUCTION

The majority of the human proteome is organized in building blocks commonly referred to as protein domains. ¹ Among these, protein–protein interaction domains are particularly important as they enhance the function of proteins by governing their capability to reversibly bind other domains or smaller peptide stretches. ² , ³ , ⁴ , ⁵ PDZ domains are the most abundant protein–protein interaction domains in the human proteome, counting on 267 different domains that may be found in 150 proteins. ⁶ The function of PDZ domains lies in the specific recognition of short amino acid sequences typically located at the C‐terminus of their physiological partners. ⁷

In most cases, PDZ domains are organized in tandem repeats, whereby more than one adjacent copy is displayed in the same array. ⁸ The general significance of grouping PDZ domains in repeats is not known. In fact, in these cases often only one domain of the repeat interacts with a physiological partner and is directly implicated in the assembly of macromolecular complexes. There is some evidence suggesting the mutual interaction of the PDZ domains to regulate the binding to target ligands, either by stabilizing the overall structure or by fine‐tuning the architecture of the binding pocket of the active domain. ⁹ , ¹⁰ , ¹¹ , ¹² In other cases, however, isolated PDZ domains appear to function in a similar manner as compared to their behavior in the presence of the adjacent domains. ¹³ , ¹⁴

PDZD2 is multi‐PDZ‐containing proteins comprising six PDZ domains and displaying a high sequence identity as compared to interleukin‐16. ¹⁵ The proteolytic processing by caspase of PDZD2 produces a secreted form of PDZD2 (sPDZD2), which acts as a signaling molecule and consists of a tandem of two PDZ domains namely PDZ5 and PDZ6. ¹⁶ Several studies have shown that sPDZD2 is unregulated in several types of cancers. ¹⁵ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ Despite its importance, a specific ligand of sPDZD2 has not been directly identified; although previous studies in cellula have suggested an interaction with the C‐terminal tail of the D4 immunoglobulin‐like domain of the CD4 co‐receptor. ²² , ²³

To shed light on the stability and function of sPDZD2, we investigated its pathway of folding and unfolding. As detailed below, we demonstrate the presence of a folding intermediate, which occurs transiently as a consequence of the concurrent denaturation of both PDZ5 and PDZ6. This behavior is reminiscent of what previously observed in the protein whirlin, ²⁴ , ²⁵ , ²⁶ also consisting of a tandem of two PDZ domains, suggesting that transient misfolding of contiguous domains may be a general property of PDZ tandems. Remarkably, as detailed below, we unambiguously show that, in the case of sPDZD2, the folding intermediate is capable to bind the C‐terminal tail of the D4 domain of the CD4 co‐receptor, as opposed to what is observed for the native state, which binds the same ligand only very weakly. Overall, the data suggest a fascinating scenario whereby a folding intermediate exerts a cryptic function, suggesting a new paradigm to understand the existence of PDZ tandems, whereby the activity of non‐native states may directly modulate signaling pathways.

2. RESULTS

A powerful method to understand the folding and function of tandem repeats is to compare their properties to those observed for the individual isolated domains. Accordingly, in analogy to our previous work on whirlin, ²⁴ , ²⁵ , ²⁶ we resorted to recombinantly express and characterize PDZ5 and PDZ6 in isolation and then compare their behaviors with the full‐length sPDZD2. For the sake of clarity, we will first describe the folding experiments obtained for the individual domains and then detail the findings on the full‐length sPDZD2.

2.1. The folding mechanism of PDZ5 and PDZ6 in isolation

The GdnHCl‐induced equilibrium denaturations of PDZ5 and PDZ6 were followed both by fluorescence and by circular dichroism and are reported in Figure 1. Whilst in the case of PDZ6 we could observe a denaturation profile by following both near‐UV CD and fluorescence, in the case of PDZ5 there was not an appreciable change in fluorescence upon unfolding. Consequently, in the latter case, the denaturation could be followed only by CD. In all cases, the observed transitions were consistent with a two state behavior, with a single cooperative transition. The CD spectra of the native domains are reported in Figure S1. The calculated denaturation midpoint for PDZ5 and observed m _D−N values, which could only be measured by CD, were 0.86 ± 0.07 M and 2.4 ± 0.10 kcal mol⁻¹ M⁻¹, respectively. These values were similar to those obtained for PDZ6, being 0.92 ± 0.05 M and 2.8 ± 0.10 kcal mol⁻¹ M⁻¹, when fitting the fluorescence and CD data with shared thermodynamic parameters. This finding indicates that the two domains display a similar thermodynamic stability. Importantly, in the case of PDZ6, the very good agreement between the parameters obtained by CD and fluorescence, being 0.82 ± 0.08 M and 2.6 ± 0.15 kcal mol⁻¹ M⁻¹ for fluorescence and 0.94 ± 0.05 M and 2.9 ± 0.2 kcal mol⁻¹ M⁻¹ for CD, when fitting the two experiments without sharing the thermodynamic parameters, confirms the two state nature of the equilibrium transition observed.

FIGURE 1.

Equilibrium and kinetics folding properties of PDZ5 and PDZ6 from sPDZD2. (a) Equilibrium denaturation of PDZ6 by following florescence (white circle) and circular dichroism (white diamond). (b) Equilibrium denaturation of PDZ5 by following circular dichroism (gray diamond). In both cases, lines represent the best fit to a two‐state model. (c) Chevron plot of PDZ6 (in white) versus PDZ5 (in gray). All experiments were carried out at 25°C in buffer 50 mM TrisHCl pH 7.5 with 0.3 M NaCl

To infer the details of the folding mechanism of the individual domains, we resorted to perform kinetic experiments. Hence, both PDZ5 and PDZ6 were subjected to fluorescence monitored stopped‐flow experiments. In the case of PDZ5, since we could not observe any reliable change in fluorescence upon (un)folding, the kinetics were also measured for an engineered fluorescent mutant, namely F2670W. A semi‐logarithmic plot of the observed folding/unfolding rate constant versus denaturant concentration (chevron plot), measured for the different constructs, are reported in Figure 1c. Interestingly, whereas the logarithm of the observed refolding rate constant decreases linearly with increasing denaturant concentration, the observed unfolding rate constants present a downward curvature as a function of denaturant (roll‐over effect). This feature, which has been previously observed on other PDZ domains, ²⁷ , ²⁸ , ²⁹ has been already ascribed to a common folding mechanism for this class of proteins involving the presence of a high energy folding intermediate.

2.2. The folding behavior of full‐length sPDZD2

The GdnHCl‐induced equilibrium denaturations of full‐length sPDZD2 were followed both by fluorescence and by CD and are reported in Figure 2a. Interestingly, also the bi‐domain construct displayed an apparent two state transition with a single sigmoidal behavior and similar equilibrium parameters could be extracted from the analysis of the fluorescence and CD monitored experiments, being 1.01 ± 0.09 M and 2.48 ± 0.04 kcal mol⁻¹ M⁻¹. Importantly, the apparent m _D−N is consistent with the value expected from a single PDZ domain, ³⁰ indicating that the two domains to not unfold as a single cooperative unit. In fact, as exemplified in the case of the GK and SH3 domains from PSD‐95, ³¹ if that were the case, the apparent m _D−N value would have been substantially increased.

FIGURE 2.

Equilibrium and kinetics folding properties of sPDZD2. (a) Equilibrium denaturation of sPDZD2 by following florescence (dark circle) and circular dichroism (black diamond). Lines represent the best fit to a two‐state model. (b) Chevron plot of PDZ6 (in white) versus sPDZD2 (in gray). Black stars represent a slow kinetic phase of sPDZD2 occurring in refolding experiments at low final concentration of denaturant (between 0.1 and 0.7 M). All experiments were carried out at 25°C in buffer 50 mM TrisHCl pH 7.5 with 0.3 M NaCl

To interpret the data measured for full‐length sPDZD2, it is particularly instructive to consider the behavior observed for the individual domains (Figure 1). Since both PDZ5 and PDZ6 display a very similar stability, it is not surprising to observe a single transition by CD, indicating that the presence of the two domains in tandem affects their stability only very marginally. In fact, in this case, the full‐length sPDZD2 is expected to return the sum of the two individual domains that, being very similar, may appear as a single transition. This consideration is further supported by the analysis of the fluorescence data, where only PDZ6 can be monitored. Also in this case, the transition observed for the full‐length sPDZD2 is very similar to PDZ6, indicating that the presence of PDZ5 has only marginal effects on the stability of PDZ6.

The results described above allow concluding that the fluorescence monitored transitions measured on full‐length sPDZD2 are diagnostic of the folding on PDZ6 only. Hence, to test the effect of the presence of the contiguous domain of its folding and folding mechanism, we performed single and double jump (un)folding experiments. The chevron plot of sPDZD2 compared to that of PDZ6 is reported in Figure 2b. It is evident that, whilst the unfolding kinetics is only marginally affected by the presence of PDZ5, the refolding time courses display additional complexities. In fact, in the case of sPDZD2 we could monitor a double exponential behavior, diagnostic of the transient accumulation of an intermediate. A representative fluorescence monitored refolding time course of sPDZD2 in comparison to that observed in the case of isolated PDZ6 is reported in Figure S2. Notably, the fast phase parallels the refolding phase observed in the case of isolated PDZ6, with the emergence of a slow refolding accounting of circa 70% of the total amplitude that is indicative of a relevant population folding through a slower pathway.

We have previously shown that, in the case of whirlin, the emergence of a slow refolding event is due to the accumulation of a misfolded intermediate that may occur from the transient interaction between the two contiguous PDZ domains. ²⁴ , ²⁵ , ²⁶ Importantly, the interaction between PDZ1 and PDZ2 of whirlin may only occur when both PDZ domains were denatured. This hypothesis could be tested thanks to the fortuitous difference in stability of the two domains that could be selectively denatured at different denaturant concentrations. ²⁴ Testing the above hypothesis in the case of sPDZD2 is particularly challenging, due to the very similar stabilities of PDZ5 and PDZ6. Hence, we resorted to perform double‐jump interrupted unfolding experiments.

Double jump interrupted unfolding experiments were designed to test whether the presence of an additional slow refolding phase in sPDZD2 was dependent on the concurrent denaturation of both its PDZ domains. Consequently, native sPDZD2 was rapidly mixed in a first jump versus the unfolding buffer at a final concentration of 1.2 M GdnHCl; subsequently, after a controlled delay time, the reaction mix was challenged with refolding buffer, leading to a final concentration of GdnHCl 0.6 M, while monitoring the time evolution of the fluorescence. The relative amplitudes of the fast and slow phases as a function of the delay times are reported in Figure 3. It is evident that, at very short delay times, the observed reaction conformed to a single exponential decay (Figure 3a), with the relative amplitude of the slow phase being very small. Importantly, a quantitative analysis of the dependence of the relative amplitudes of the fast and slow phases from delay times, as obtained from double jump experiments, returns a rate constant of 6 ± 1 s⁻¹, which is very similar to the unfolding rate constant measured at the same experimental conditions for the F2670W variant of PDZ5 being 5 ± 0.5 s⁻¹. On the basis of this observations, we conclude that, in analogy to what previously observed in the case of whirlin, ²⁴ , ²⁵ , ²⁶ heterogeneity in the refolding of sPDZD2 is due to the accumulation of a misfolded intermediate that results from the transient interaction between the two contiguous denatured domains. Further support of this conclusion comes from the observation that PDZ6 in isolation is capable to fold via a single exponential process, which is essentially identical to the fast phase observed in the case of sPDZD2.

FIGURE 3.

Interrupted unfolding experiments on sPDZD2. (a) Fluorescence time course observed in double mixing interrupted refolding experiments of sPDZD2 protein at a delay time of 50 ms. The residuals to a single exponential fit are shown in the graph. It is evident that, whilst a minor deviation from single exponential could be detected, the observed transition appears consistent with the fast refolding phase only. (b) Relative amplitudes of slow and the fast phases, obtained from a double exponential fit, plotted against the delay times between the first and the second mix (ranging from 0.05 to 10 s)

2.3. The binding properties of sPDZD2: a cryptic binding activity of a folding intermediate

Despite intense research, ¹⁵ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ understanding the binding of PDZD2 has remained elusive. In fact, notwithstanding its pivotal role in several types of cancers, no clear‐cut evidence has been provided to univocally identify a physiological binding partner. Of additional interest, a structural analysis of all the PDZ domains in the human proteome revealed a peculiar feature of the PDZ6 from sPDZD2, namely the presence of a Trp residue in its binding pocket, ¹⁶ a feature that is shared only with the PDZ domain of interleukin‐16. ³² In pancreatic INS‐1E β cells, depletion of sPDZD2 by RNA interference suppressed the expression of β‐cell genes Ins1, Glut2 and MafA whereas treatment with recombinant sPDZD2 rescued the suppressive effect. ¹⁵ In particular, sPDZD2 stimulated intracellular cAMP levels, activated β‐cell gene expression in a PKA‐dependent manner. On the basis of these observations, it has been hypothesized that sPDZD2 exerts its growth and differentiation and T cells activation functions via a PKA dependent pathway, ²² which is ultimately mediated by the interaction with the D4 domain of the CD4 co‐receptor, following a similar pathway to that proposed for interleukin‐16. ³³ , ³⁴

To test the binding of sPDZD2 to D4 domain of the CD4 co‐receptor we carried out fluorescence and NMR experiments. In all experiments, in analogy to previous work on PDZ domains, we employed a peptide displaying the C‐terminal sequence of the D4 immunoglobulin‐like domain of CD4 glycoprotein as a model, to mimic the physiological partner (CD4*). The fluorescence and NMR monitored binding transition of sPDZD2 to CD4* is reported in Figure 4. In the case of fluorescence experiments, binding was monitored by measuring FRET between the Trp residues of sPDZD2 and a dansyl group covalently attached to the N‐terminus of CD4*. However, own to the poor binding capability of native sPDZD2, the binding to the peptide could not be measured unambiguously and only a qualitative estimate of the correspondent K _D may be assigned in the order of mM concentration. Figure S2 reports the HSQC spectra of the double labeled ¹⁵N–¹³C of sPDZD2 and PDZ6 in isolation. Own to the poor solubility of PDZ5, we were not able to record its NMR spectrum. To address the binding properties of sPDZD2, we also recorded the HSQC spectra of the double labeled ¹⁵N–¹³C protein in the presence of variable amounts of CD4*. The measured spectra, reported in Figure 4c, confirm the very weak binding properties of the native protein, as revealed by the perturbation of several reveals peaks upon increasing peptide concentration. Fluorescence monitored kinetic binding experiments using either sPDZD2, PDZ5, or PDZ6 did not return any reliable time course. This observation is not surprising, given the very high value of K _D estimated from equilibrium fluorescence and NMR experiments, being at the limit of the solubility of the peptide. Similarly, fluorescence monitored equilibrium experiments did not return any reliable change in signal. On the basis of these observations, we conclude that native sPDZD2, as well as the isolated PDZ5 and PDZ6, binds only very weakly to CD4*.

FIGURE 4.

The binding of native pPDZD2 to CD4*. (a) Fluorescence monitored transition measured at equilibrium by titrating a constant concentration of sPDZD2 with increasing concentration of a dansylated CD4* peptide. (b) Fraction of bound protein as obtained from NMR titration. The complete HSQC spectra and two close‐up views of the spectra obtained at different peptide concentrations are reported in panels c, d, and e, respectively

An interesting feature, characterizing whirlin, ²⁴ lies in its ability to maintain its binding properties even populating a transient misfolded conformation. To interrogate whether this property was also present in the case of sPDZD2, we performed double jump stopped flow experiments. In particular, denatured sPDZD2 was rapidly mixed with refolding buffer and, after a controlled delay time, was subsequently mixed against different concentrations of CD4*. Binding was monitored by measuring FRET between the Trp residues of sPDZD2 and a dansyl group covalently attached to the N‐terminus of CD4*. The rational of the experiment lies in allowing the protein to populate the misfolded intermediate and then mix it with the ligand, prior the accumulation of the native state. Surprisingly, at variance with what observed for the native state, in the case of the misfolded intermediate, we could measure reliable binding transitions occurring in the milliseconds time regime. The dependence of the amplitude of the observed binding phase as a function of the delay time between the first and the second mix are reported in Figure 5. It is evident that, with increasing delay times, the amplitude increases (intermediate accumulation during delay time) and then decreases (intermediate converting to native state). Remarkably, a fit to a double exponential decay returns the rate constant of 7 ± 1 s⁻¹ and 1 ± 0.3 s⁻¹; the latter rate constant being in agreement with the slow refolding rate constant measured in the chevron plot of sPDZD2 (Figure 2), which may be ascribed to the inter‐conversion of the intermediate to the native state.

FIGURE 5.

Binding kinetics of the misfolded intermediate. (a) Fluorescence time course obtained in double‐jump refolding and binding experiment at a delay time of 30 ms. The experimental set up is described in the main text. (b) The amplitude of the binding traces obtained by mixing CD4* (50 μΜ final) and sPDZD2 (2.5 μΜ final) plotted versus different delay times. (c) Pseudo‐first order binding kinetics between the intermediate of sPDZD2 and CD4*

A pseudo‐first‐order plot of the observed rate constant measured for the misfolded intermediate of sPDZD2 is reported in Figure 5c. The rate constants were obtained by repeating the double jump experiment summarized above at different concentrations of CD4* and at a constant delay time of 300 ms, where the binding reaction could be monitored clearly. A linear fit to the observed rate constants returns a k _on and k _off of 1.5 ± 0.3 μM⁻¹ s⁻¹ and 13 ± 2 s⁻¹, respectively, which allows estimating a K _D of 9 ± 2 μM. In this respect, it is also important to note that the observed dependence of the rate constants on peptide concentration represents an intrinsic validation of the bimolecular nature of the observed reaction, thereby unequivocally indicating that the observed kinetic transitions refer to genuine binding between the intermediate state and CD4*. Furthermore, since the amplitude of the binding traces decreases with increasing delay time it may be concluded that a direct comparison between the binding properties of the intermediate and native state indicates that the latter is less competent for binding. Interestingly, the value of K _D of obtained for the intermediate is in the same order of magnitude with the values typically observed for PDZ domains and their physiological ligands. ⁶ , ⁷ To further probe if the binding of the intermediate state to CD4* were specific, we resorted to perform to additional control experiments by using a C‐terminal blocked peptide (i.e., dansyl‐WQCLLSD‐NH2) and a peptide with the same degree of hydrophobicity but containing a scrambled sequence (i.e., dansyl‐QLCLWSD). Remarkably, in both cases, we could not observe any binding transition both by equilibrium and double jump experiments, confirming that both the native and intermediate states of sPDZD2 are not competent for binding to these two peptides and, thereby, reinforcing the importance of the interaction between the intermediate state and CD4*.

3. DISCUSSION

Whilst the folding pathway of single domain proteins has been addressed on numerous different systems, our current knowledge concerning multidomain proteins is substantially more limited. In fact, because of their complexity, these systems are typically very elusive to a rigorous experimental characterization and, therefore, general rules defining their properties are still relatively far to be elucidated.

The results presented in this work provide a clear‐cut demonstration of the presence of transient misfolding in the case of a PDZ tandem repeat. Of particular interest, the comparison between the folding of the tandem to that of the isolated domains, taken together with the double jump experiments, clearly reveals the emergence of a kinetic trap that competes with productive folding. Importantly, this intermediate may be accumulated only when the two contiguous domains are both denatured, indicating the transient interaction between the two denatured states. This finding parallels what recently observed in the case of the first two PDZ domains of whirlin, ²⁴ , ²⁵ , ²⁶ reinforcing the view that these events might involve transient domain swapping, ³⁵ , ³⁶ , ³⁷ whereby the two domains exchange a part of their structure to form an intertwined dimer.

The most interesting feature shared by the transient misfolded states found in whirlin and sPDZD2 lies in their ability to bind their physiological ligands. Remarkably, however, whilst in the case of whirlin this property was similar to that of the native state, ²⁴ in the case of sPDZD2 it can be observed that the binding capability is strongly enhanced in the misfolded conformation as compared to the native state, with a K _D decreasing by more than one order of magnitude. Hence, observed data seem to highlight a conundrum whereby a function is observed by a misfolded conformation as opposite to what found in the native state.

It is at this stage very difficult to elucidate if the misfolded conformation of sPDZD2 exerts any physiological functions in the cellular environment. Nevertheless, the binding properties of the intermediate, which exceeds those observed for the native conformation, are particularly surprising and demand some additional considerations. In this context, it is worth to consider that sPDZD2 undergoes secretion from the cytoplasm, ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ²² a phenomenon that has been commonly associated to folding defects. ³⁸ , ³⁹ Indeed, a growing number of human diseases are being associated to defects in secretory protein folding, and many of these products are targeted for a process known as endoplasmic reticulum‐associated degradation. ⁴⁰ , ⁴¹ , ⁴² Furthermore, it is important to note that, despite intense research and its critical role in modulating cellular proliferation and differentiation, no physiological ligand has been directly documented for native sPDZD2. On the basis of these observations, it is temptative to speculate that the transient folding intermediate exerts a cryptic activity that contributes substantially to the cellular functions of sPDZD2.

4. CONCLUSIONS

The general notion that protein misfolding is intimately linked to dysfunction and disease represents one of the milestones of modern chemistry and molecular biology. In fact, several devastating disorders, ranging from neurologic disorders to systemic amyloidosis, may be ascribed to aberrant incorrect structures that often culminate to extensive protein aggregation. ⁴³ , ⁴⁴ , ⁴⁵ The results exemplified by the case of sPDZD2 provide an interesting example of a functionally competent misfolded intermediate, which may exert a cryptic function that is not captured from the analysis of the native state only. Moreover, the data provide a new paradigm to understand the existence of PDZ tandems, which are highly abundant in the human proteome and whose significance is still to be fully understood. Future studies on other PDZ tandems will further clarify the general nature of these effects.

5. MATERIALS AND METHODS

5.1. Constructs and site‐directed mutagenesis

Genes encoding for sPDZD2 and PDZ6 in their cysteine in serine variants were synthesized and subcloned into a pET28b + vector. The construct encoding for PDZ5 was obtained inserting a stop codon in position 2724 by using the sPDZD2 construct as template and site directed mutagenesis, using QuikChange Lightning Mutagenesis Kit (Agilent) following manufacturer instructions.

The PDZ5* variant, corresponding to the F2670W mutation, was obtained by site‐directed mutagenesis QuikChange Lightning Mutagenesis Kit following manufacturer instructions. Primers oligos were purchased from Eurofins Genomics and all sequences were confirmed by DNA sequencing.

5.2. Protein expression and purification

sPDZD2, PDZ6, PDZ5, and variant PDZ5* were transformed into Escherichia coli BL21 (DE3) competent cells. The cells were grown in 1 L of Luria broth (LB) or in minimal M9 media supplemented with ¹⁵NH₄Cl (1 g L⁻¹) and ¹³C‐glucose (3 g L⁻¹) to label the protein for NMR experiments in presence of Kanamycin (30 μg ml⁻¹) until OD₆₀₀ of 0.6–0.7 and then induced with 1 mM isopropyl β‐d‐1‐thio‐galactopyranoside (IPTG). For the expression of sPDZD2 and PDZ6 the cells were incubated at 37°C overnight, for the expression of PDZ5 and PDZ5* where incubated at 37°C for 3 hr and then overnight at 25°C. Whereas sPDZD2 and PDZ6 was purified from the soluble fraction, PDZ5 and PDZ5* were recovered from the inclusion bodies.

The cells expressing sPDZD2 and PDZ6 were resuspended using a buffer containing 50 mM TrisHCl pH 7.5, 0.3 M NaCl, 10 mM imidazole and a protease inhibitor cocktail tablet (cOmplete, EDTA‐free, Roche). The cells were then sonicated and centrifuged, and the supernatant was loaded in a pre‐equilibrated HiTrap Chelating High‐Performance column (GE Healthcare). The elution was performed with a gradient of 0.01–1 M imidazole, and the imidazole was finally removed using an HiPrep desalting column (GE Healthcare), equilibrated with the buffer 50 mM TrisHCl pH 7.5, 0.3 M NaCl. The cells expressing PDZ5 and PDZ5* were resuspended using a buffer containing 50 mM TrisHCl pH 7.5, 0.3 M NaCl and a protease inhibitor cocktail tablet. The cells were sonicated and centrifuged; the pellet was then resuspended with buffer containing 50 mM TrisHCl pH 7.5 with 4 M Urea, and centrifuged. The supernatant was then loaded in a pre‐equilibrated HiTrap SP HP cation exchange chromatography column (GE Healthcare) and manually eluted with 1 M NaCl; the buffer exchange was carried out using a Amicon centrifugal filter MWCO 10KDa (Millipore). Protein purity was confirmed by SDS‐PAGE.

The aminoacidic sequences of the constructs used in this study were the following: sPDZD2 2622—FIVLNRKEGSGLGFSVAGGTDVEPKSITVHRVFSQGAAS QEGTMNRGDFLLSVNGASLAGLAHGNVLKVLHQAQLHKDALVVIKKGMDQPRPSARQEPPTANGKGLLSRKTIPLEPGIGRSVAVHDALSVEVLKTSAGLGLSLDGGKSSVTGDGPLVIKRVYKGGAAEQAGIIEAGDEILAINGKPLVGLMHFDAWNIMKSVPEGPVQLLIRKHRNSS—2839; PDZ5 2622—FIVLNRKEGSGLGFSVAGGTDVEPKSITVHRVFSQGAASQEGTMNRGDFLLSVNGASLAGLAHGNVLKVLHQAQLHKDALVVIKKGMDQPRPSARQEPPTAN—2723; PDZ6 2750 —SVEVLKTSAGLGLSLDGGKSSVTGDGPLVIKRVYK GGAAEQAGIIEAGDEILAINGKPLVGLMHFDAWNIMKSVPEGPVQLLIRKHRNSS—2839.

5.3. Equilibrium experiments

Equilibrium denaturation experiments were performed by monitoring both fluorescence and circular dichroism. All experiments were carried out at 25°C and using 50 mM TrisHCl pH 7.5 with 0.3 M NaCl as a buffer and guanidium hydrochloride as denaturant agent.

Fluorescence equilibrium (un)folding experiments were executed by excitation of the tryptophan residue 2,817 of PDZ6 and sPDZD2 at 280 nm, and the emission spectra were recorded in a range between 300 and 400 nm using a FluoroMax‐4 single photon counting spectrofluorometer (Horiba). The experiments were carried out using a 1 cm‐thick quartz cuvette and a protein concentration of 1 μM.

The far‐UV CD (un)folding spectra were collected between 200 and 250 nm utilizing a Jasco 810 dichrograph, which was flushed with N2 and equipped with a Peltier thermoregulation system. Each spectrum is the average of 8 acquisitions carried out at a scanning speed of 50 nm min⁻¹ with a data pitch of 0.5 nm in a 1 cm‐thick quartz cuvette and with a protein concentration of 5 μM for PDZ5, PDZ6, and sPDZD2.

5.4. Kinetics experiments

Kinetic (un)folding experiments were conducted monitoring the changes in fluorescence emission of the protein using a rapid mixing stopped‐flow apparatus (SX‐18 MV, Applied Photophysics), in buffer 50 mM TrisHCl pH 7.5 with 0.3 M NaCl and guanidium hydrochloride as a denaturant agent.

Fluorescence emissions of the tryptophan residues in position 2817 and 2670, for PDZ6 and sPDZD2, and PDZ5* respectively, were measured with a 320‐nm bandpass filter and an excitation wavelength of 280 nm. The experiments were carried out at 25°C, using 1 μM final protein concentration and a denaturant concentration ranging from 0.2 to 5.27 M. The unfolding and refolding traces of PDZ6 and PDZ5* were fitted with a single exponential equation while the refolding traces of sPDZD2 showing two kinetic phases were fitted with a double exponential equation. The traces are the average of three to six acquisitions.

Kinetic binding experiments were carried out using a stopped‐flow apparatus set up in single mixing mode. A fixed concentration of sPDZD2 protein (3 μM final) were mixed with increasing final concentration (ranging from 5 to 40 μM) of a N‐dansyl peptide (sequence WQCLLSD) mimicking D4 immunoglobulin‐like domain of CD4 glycoprotein. This modification allowed us to monitor the binding reaction by Förster Resonance Energy Transfer (FRET) using the naturally present tryptophan residue on PDZ6 as donor and dansyl group as acceptor. Samples were excited at 280 nm, and the emission fluorescence was recorded by using a band‐pass filter. The resulting k _obs were measured from the average of 3–6 traces by fitting the time course for binding using a single exponential equation and then plotted versus ligand concentration. In all conditions, the buffer contained 4 μM DTT.

Double‐jump kinetic experiments were performed using a MX‐17 stopped‐flow instrument with double‐jump capability at an excitation wavelength of 280 nm and a 320 nm band‐pass filter. Double‐jump interrupted unfolding experiments was carried out at 25°C. In the first mix, folded sPDZD2 (12 μM in buffer 50 mM TrisHCl pH 7.5) was first mixed against an unfolding buffer (2.4 M GdnHCl at 50 mM TrisHCl pH 7.5). After different unfolding delay time (ranging from 0.05 to 10 s), the unfolded protein was then mixed with a refolding buffer. Traces were fitted with Applied Photophysics software, and the resulting amplitudes of the fast and slow phases were plotted versus the delay time.

Double‐jump interrupted binding experiments were executed at 25°C monitoring the fluorescence emission of sPDZD2 upon binding with the peptide mimicking the D4 immunoglobulin‐like domain of CD4 glycoprotein (sequence WQCLLSD). Kinetics were measured by monitoring FRET between the tryptophan 2817 on PDZ6, acting as a donor, and a dansyl group covalently attached to the N‐terminus of the peptide, acting as an acceptor.

AUTHOR CONTRIBUTIONS

Francesca Malagrinò: Conceptualization (equal); data curation (equal); formal analysis (equal); investigation (equal); writing – review and editing (equal). Giuliana Fusco: Data curation (equal); formal analysis (equal); investigation (equal). Valeria Pennacchietti: Data curation (equal); investigation (equal). Angelo Toto: Data curation (equal); formal analysis (equal); investigation (equal); writing – review and editing (equal). Caterina Nardella: Data curation (equal); formal analysis (equal); investigation (equal). Livia Pagano: Data curation (equal); formal analysis (equal); investigation (equal). Alfonso De Simone: Conceptualization (equal); data curation (equal); formal analysis (equal); investigation (equal); writing – review and editing (equal). Stefano Gianni: Conceptualization (equal); formal analysis (equal); funding acquisition (equal); supervision (equal); writing – original draft (equal); writing – review and editing (equal).

AKNOWLEDGMENTS

Work partly supported by grants from the Italian Ministero dell'Istruzione dell'Università e della Ricerca (Progetto di Interesse “Invecchiamento” to Stefano Gianni), Sapienza University of Rome (RP11715C34AEAC9B and RM1181641C2C24B9, RM11916B414C897E, RG12017297FA7223 to Stefano Gianni, AR22117A3CED340A to Caterina Nardella), by an ACIP grant (ACIP 485‐21) from Institut Pasteur Paris to Stefano Gianni, the Associazione Ital‐iana per la Ricerca sul Cancro (Individual Grant—IG 24551 to Stefano Gianni), the Regione Lazio (Progetti Gruppi di Ricerca LazioInnova A0375‐2020‐36559 to Stefano Gianni), the Istituto Pasteur Italia, “Teresa Ariaudo Research Project” 2018, and “Research Program 2022‐2023 Under 45 Call 2020” (to Angelo Toto). Francesca Malagrinò was supported by a fellowship from the FIRC—Associazione Italiana per la Ricer‐ca sul Cancro (Filomena Todini fellowship).

CONFLICT OF INTEREST

All the authors declare no conflicts of interests.

Supporting information

Figure S1. CD spectra of native PDZ5, PDZ6, and sPDZD2

Figure S2. Comparison of refolding time course of PDZ6 and sPDZD2

Figure S3. Comparison of the HSQC spectra measured for native PDZ6 and sPDZD2

Malagrinò F, Fusco G, Pennacchietti V, Toto A, Nardella C, Pagano L, et al. Cryptic binding properties of a transient folding intermediate in a PDZ tandem repeat. Protein Science. 2022;31(9):e4396. 10.1002/pro.4396