Proline substitutions in a Mip-like peptidyl-prolyl cis-trans isomerase severely affect its structure, stability, shape and activity.

FKBP22, an Escherichia coli-specific peptidyl-prolyl cis-trans isomerase, shows substantial homology with the Mip-like virulence factors. Mip-like proteins are homodimeric and possess a V-shaped conformation. Their N-terminal domains form dimers, whereas their C-terminal domains bind protein/peptide substrates and distinct inhibitors such as rapamycin and FK506. Interestingly, the two domains of the Mip-like proteins are separated by a lengthy, protease-susceptible α-helix. To delineate the structural requirement of this domain-connecting region in Mip-like proteins, we have investigated a recombinant FKBP22 (rFKBP22) and its three point mutants I65P, V72P and A82P using different probes. Each mutant harbors a Pro substitution mutation at a distinct location in the hinge region. We report that the three mutants are not only different from each other but also different from rFKBP22 in structure and activity. Unlike rFKBP22, the three mutants were unfolded by a non-two state mechanism in the presence of urea. In addition, the stabilities of the mutants, particularly I65P and V72P, differed considerably from that of rFKBP22. Conversely, the rapamycin binding affinity of no mutant was different from that of rFKBP22. Of the mutants, I65P showed the highest levels of structural/functional loss and dissociated partly in solution. Our computational study indicated a severe collapse of the V-shape in I65P due to the anomalous movement of its C-terminal domains. The α-helical nature of the domain-connecting region is, therefore, critical for the Mip-like proteins.

Introduction

A nascent polypeptide, synthesized by a living cell, needs to fold properly prior to performing any biological function. Two types of proteins that usually convert a linear polypeptide chain into its appropriate three-dimensional form are the chaperones and isomerases [1]. Of the isomerases, peptidyl-prolyl cis-trans isomerase (PPIase) catalyzes the slow isomerization of peptide bond preceding the Pro residue. PPIases, expressed by all living organisms, belong primarily to three structurally dissimilar families: cyclophilins, FK506-binding proteins (FKBPs) and parvulins. While parvulins are inhibited by juglone, cyclophilins and FKBPs are inhibited by cyclosporin A and FK506/rapamycin, respectively. Interestingly, the complex formed between cyclosporin A or FK506/rapamycin and the cognate PPIase blocks T-cell activation by obstructing the specific signal transduction step [1], [2]. Therefore, the cyclophilin and FKBP inhibitors are employed extensively in immunosuppressive therapy [1], [2]. On the contrary, juglone is effective for treatment of different microbial infections and inflammatory diseases [3].

FKBP22, a PPIase (EC 5.1.2.8) originally purified from Escherichia coli [4], harbors 206 amino acid residues, forms dimers in solution and shares significant identity with several other PPIases including Mip (macrophage infectivity potentiator)-like PPIases [5], [6], [7], [8], [9], [10], [11], [12], [13], [14]. In many pathogens, the Mip-like PPIases also act as virulence factors. The isomerase activity of E. coli FKBP22 and related proteins are inhibited by FK506 and rapamycin but not by juglone or cyclosporin A. Structural investigations revealed that these enzymes possess a V-like shape by assembling two monomers [13], [14], [15], [16]. Each monomer is composed of an N-terminal domain (NTD), a C-terminal domain (CTD) and a domain-connecting flexible region. Dimerization of such enzymes occurs when the NTDs of two monomers interact [14], [17]. In contrast, the CTD contains both the substrate-binding and the inhibitor-binding sites [14], [18], [19]. The V-shaped conformation appears crucial for its enzymatic activity, particularly with a protein substrate [20]. Further studies indicate that the CTD is comparatively less stable than the NTD or the intact isomerase [13], [14]. Unfolding of a recombinant E. coli FKBP22 in the presence of urea and GdnCl occurs by a two-state and a three-state mechanism, respectively [14].

The domain-connecting hinges in many proteins appears to be crucial for preserving their conformation, stability and function [21], [22], [23], [24], [25]. The domain-connecting flexible regions in the E. coli FKBP22 and the orthologous proteins are largely structured with a lengthy, protease-sensitive helix designated α3 [14], [15], [16], [17], [18]. Helix α3 of FKBP22 is apparently formed by the amino acid residues 55–92 [14], [18]. It was previously shown that shortening or enlarging α3 severely affected the structure, function and stability of FKBP22 [26]. Even the isolated domains lacking a bit of α3 became inactive or unstable [18], [20]. Collectively, the presence of a helix in between the two domains of the Mip-like proteins could be critical for protecting their structural and functional integrities. To date, no systematic study has been performed to prove the necessity of a helix in between the two domains of any Mip-like PPIase, a promising drug target [27].

Proline, unlike other protein-forming amino acid residues, is a secondary amine and cannot contribute to the hydrogen bond generation. This residue, therefore, remains mostly missing within the α-helix and β-sheet as these protein structures are stabilized by hydrogen bonds. Previously, the indispensable role of α-helix in many proteins was demonstrated by introducing Pro into this structure [28], [29], [30], [31]. To precisely determine the contribution of α3 to the structure, function and stability of the Mip-like PPIases, we generated three mutants of E. coli FKBP22 by replacing three non-polar amino acid residues in its α3 with the helix-destabilizing Pro residue. Our biochemical, biophysical and computational studies on the mutants indicate that the presence of a helix between the domains of FKBP22 is essential for maintaining the structure, protein folding ability, shape, and stability of this enzyme.

Materials and methods

Materials

Plasmid isolation kit and oligonucleotides were purchased from Genetix Biotech Asia Pvt. Ltd. Enzymes (RNase T1, Phusion DNA polymerase, DpnI, NcoI, XhoI, etc.), DNA and protein markers were from Hysel India Pvt. Ltd. Rapamycin was purchased from BioVision, Ni-NTA resin from Qiagen and antibodies (anti-His and alkaline phosphatase-tagged goat anti-mouse antibody) from Santa cruz Biotechnology Inc. Acrylamide, bis-acrylamide, glutaraldehyde were purchased from Merck; isopropyl β-d-1-thiogalactopyranoside (IPTG) from Fermentas; phenylmethane sulfonylfluoride (PMSF) from Sigma; urea from Sisco Research Laboratories. Plasmid pET28a and E. coli BL21 (DE3) were obtained as the gifts from late Dr. Pradosh Roy, Bose Institute. E. coli TOP10 was donated by Dr. Pradeep Parrack, Bose Institute. Plasmids, bacterial strains and oligonucleotides used in the present investigation are listed in Supplementary Tables 1 and 2

Basic DNA and protein tools

All basic molecular methods such as agarose gel electrophoresis, SDS-PAGE, staining of gels, Western blotting, plasmid DNA purification, DNA/protein estimation, polymerase chain reaction (PCR), cleavage of DNA by restriction endonuclease, sequencing of DNA inserts, and DNA transformation were performed as previously described [26], [32], [33], [34].

Purification of different proteins

rFKBP22 was purified as described previously [26]. To express I65P, V72P, and A82P (all rFKBP22 variants harboring a specific Pro substitution mutation in helix α3), p1354, p1355, and p1356 DNAs were constructed, respectively, essentially by a standard procedure [35]. Primers I65P1 and I65P2 were used to create p1354. The p1355 DNA was generated using primers V72P1 and V72P2. The p1356 DNA was made using primers A82P1 and A82P2. To amplify all of the above plasmids, p1289 DNA [26] was used as the template. Strains SAU1354, SAU1355, and SAU1356 were produced by transforming plasmids p1354, p1355, and p1356 to E. coli BL21 (DE3), respectively. Proteins I65P, V72P, and A82P were purified from SAU1354, SAU1355, and SAU1356 respectively, as described [14]. Using the molecular masses of the monomeric forms of rFKBP22 and its variants, their molar concentrations were estimated.

Size, shape and structure of proteins

Glutaraldehyde-mediated chemical crosslinking and analytical gel filtration chromatography [14] were performed to gain an understanding of the oligomeric status and the molecular mass/shape of the proteins in question.

To obtain clues about the structures of rFKBP22 and its three mutants, circular dichroism (CD) spectra (200–260 and 250–310 nm) and intrinsic tryptophan fluorescence spectra (λem = 300–400 nm and λex = 295 nm) were recorded as described [14], [36], [37]. The amounts of secondary structures in proteins were determined by analyzing their far-UV CD spectra with CDNN software [38]. To study the solvent accessibility of tryptophan residues of the above proteins, acrylamide quenching of their tryptophan fluorescence and the Stern–Volmer constant Ksv were determined as stated earlier [14], [39].

Function of proteins

To determine the enzymatic activity (kcat/Km) of protein (60 nM), RNase T1 (ribonuclease T1) refolding assay was performed as stated earlier [1], [13], [14], [26].

The equilibrium dissociation constant (Kd) of rapamycin - protein interaction was determined by a standard procedure [14], [26] with minor modifications. Briefly, 5 µM protein was equilibrated with 0–20 µM rapamycin in 12 steps followed by the recording of their tryptophan fluorescence spectra using a fluorescence spectrophotometer (Photon Technology International, USA, model QM-4CW) equipped with a Peltier system for temperature control.

Studies on the urea-exposed proteins

To understand the urea-induced unfolding of rFKBP22 and its mutants, 10 µM of each protein was exposed to 0–7 M urea for 16–18 h at 4 °C followed by the recording of their far-UV CD and intrinsic tryptophan fluorescence spectra as described [14]. Protein unfolding was also investigated by the transverse urea-gradient gel electrophoresis (TUGE) [14], [40]. Refolding of urea-denatured proteins was investigated both by TUGE and tryptophan fluorescence spectroscopy [14], [26].

Thermodynamic parameters of protein unfolding

Considering that protein unfolding in the presence of urea followed a two-state mechanism (N ↔ U), different parameters like fu, the fraction of denatured protein molecules, ΔGW, free energy change at 0 M urea, m, cooperativity parameter of unfolding, Cm, urea concentration at the midpoint of unfolding transition (i.e. urea concentration at which ΔG = 0), and ΔΔG, the difference of free energy change between rFKBP22 and its mutant, were determined using standard equations [41].

Computational studies, simulation and principal component analysis

The amino acid residues at positions 65, 72 and 82 in the model structure of FKBP22 [26] were separately replaced with a Pro residue using Coot [42]. To compare the above mutated structures with that of the wild-type FKBP22, molecular dynamics (MD) simulations were performed in explicit solvent at 300 K for 10 ns using the sander module of AMBER 10 [43]. The protein was solvated using TIP3 Box with the FF99SB force field parameters. Before MD simulation, the molecules were minimized using the 500 cycles of steepest descent followed by the 20,000 cycles of conjugate gradient. The system was heated to 300 K within 40 ps and equilibrated following minimization. Bonds involving hydrogen were constrained with the help of the SHAKE algorithm. The production run was performed for 10 ns using constant pressure periodic boundary conditions. A non-bonded cut-off distance of 12 Å and the integration time of 2 fs were used for all simulations. The RMSD (root mean square deviation) and the RMSF (root mean square fluctuation) values for the backbone atoms (C, Cα, O, N) of all proteins were determined by analyzing the resulting MD trajectories with ptraj module of AMBER 10.

To extract the dominant motions in the MD simulations of FKBP22 and its mutants, principal component analysis (PCA) was performed by a standard procedure [44], [45], [46] using Bio3D package of R [47]. The related figures and movies were generated using PyMol [48] and VMD [49]. Each dimeric protein was considered as a single system during the above computational analyses.

Basic statistical analysis

All results enclosed here include the means and standard deviations of at least three separate observations. Mean, standard deviation, and p values were determined as described [26]. Two results were deemed significant if the analogous p value was <0.05.

Results

Purification of helix α3 mutants

To determine whether the presence of α3 in between the two domains of Mip-like PPIases is required, we have generated three E. coli FKBP22 mutants I65P, V72P, and A82P by substituting three hydrophobic amino acid residues Ile 65, Val 72 and Ala 82 in this helix with a Pro residue (Fig. 1). As described in the Legionella Mip structure [15], the side chains of the selected residues in FKBP22 should neither interact with those of the domain-forming residues nor contribute to the stability of helix α3. Both Ile 65 and Ala 82 of E. coli FKBP22 align with either identical or other hydrophobic residues of the homologous proteins [13]. Moreover, these three residues are well separated and also located nearly ten residues away from the putative domains of FKBP22. Pro residue was introduced as it frequently disrupts the α-helices in proteins [50]. To carry out structural and functional investigation of the FKBP22 mutants, the polyhistidine-tagged forms of these mutants were purified to near homogeneity (Supplementary Fig. 1).

Fig. 1

Sequences and structures of proteins. (A) A tertiary model structure of E. coli FKBP22. The V-shaped structure of FKBP22, developed by a standard method [26], is composed of its two monomers that interact using their N-terminal domains. NTD, CTD, and α3 indicate N-terminal domain, C-terminal domain, and the domain-connecting α-helix, respectively. The asterisk denotes the putative rapamycin binding site in FKBP22. (B) Alignment of the helix α3 sequences of rFKBP22 and its Pro substitution mutants. The ‘red colored’ P indicates the substituted Pro residue in the helix α3 of each indicated mutant.

Shapes and sizes of the proline-inserted helix α3 mutants

Pro substitution in a protein may affect its shape, size, structure, stability and function. To determine if such a substitution in α3 has altered the dimeric status of FKBP22, varying concentrations of rFKBP22, I65P, V72P, and A82P were analyzed by gel filtration chromatography individually. All of the proteins except I65P produced a single peak at both 0.5 and 20 µM (Fig. 2A). I65P resulted in a single peak at 20 µM but yielded two partly fused peaks at 0.5 µM concentration. At 20 µM, the retention volume (∼74.76 mL) of V72P was nearly similar to that of rFKBP22. Their elution volumes were changed very little (∼0.12 mL) upon forty fold dilution. Those of I65P and A82P were ∼1.25–1.75 mL less than those of rFKBP22 or V72P at 20 µM concentration. Two peaks of I65P at 0.5 µM corresponded to the elution volumes of ∼78.63 and 87.25 mL, respectively. Comparing the elution profiles of some monomeric proteins (ovalbumin, conalbumin, ribonuclease A, and carbonic anhydrase; data not shown) with those of the above proteins, we determined their apparent molecular masses. The apparent molecular mass of rFKBP22 or V72P at 0.5/20 µM is ∼54.5 kDa. In contrast, the apparent masses of I65P and A82P are ∼2–7 kDa higher than that of rFKBP22 or V72P at 20 µM. In addition, two partly combined peaks of I65P correspond to apparent molecular masses of ∼41.42 and 22.46 kDa, respectively. Calculated molecular masses of rFKBP22, I65P, V72P and A82P monomers are nearly 23 kDa. Taken together, we suggest that rFKBP22, V72P, and A82P at 0.5/20 µM exist primarily as dimers in solution. These molecules are eluted relatively early, possibly due to their non-globular conformation in solution [13]. I65P, like rFKBP22 and other mutants, might also be dimeric in solution at 20 µM. Our chemical cross-linking data confirm that all of the rFKBP22 mutants formed dimers in solution at higher concentrations (Fig. 2B).

Fig. 2

Determination of shape and size of rFKBP22 and its helix α3 substitution mutants. (A) Analysis of the proteins by analytical gel filtration chromatography. Elution profile of each indicated protein (20 or 0.5 µM) was generated by passing it through a Superdex S-200 column. (B) Analysis of the glutaraldehyde-exposed (+) and unexposed (−) proteins as labeled, by SDS-13.5% PAGE. D and M indicate dimer and monomer, respectively.

The formation of two conjoint peaks by the dimeric I65P at 0.5 µM might be due to its partial dissociation (to monomer) at this concentration. At 20 µM, relatively early elution of dimeric I65P and A82P indicates that these molecules possess larger size compared to rFKBP22 and/or V72P. Thus introduction of I65P and A82P mutations into α3 affect both the shape and size of FKBP22 significantly.

Structures of the proline containing helix α3 mutants

To further determine the effect of these point mutations, the structures of rFKBP22 and its mutant variants were studied by both CD and intrinsic tryptophan fluorescence spectroscopy. Fig. 3A shows the far-UV CD spectrum of each protein at 200–260 nm. Spectra of the proteins did not overlap though all of them have two peaks of large negative ellipticity at 208 and 222 nm. This indicates that these molecules are composed of different amounts of α helices. Additional analyses of the spectra by CDNN [38] showed that rFKBP22, I65P, V72P and A82P contain 37%, 27.8%, 32.9%, and 35.4% α-helix, respectively (Supplementary Table 3). Like the far-UV CD spectra, the near-UV CD spectra of I65P and V72P were completely different from that of rFKBP22 at 260–320 nm (Fig. 3B). Near UV-CD spectrum of A82P, however, appeared to partially overlap with that of rFKBP22.

Fig. 3

Far-UV CD (A), near-UV CD (B) and intrinsic Trp fluorescence (C) spectra of rFKBP22 and its variants. (D) Stern–Volmer plots show acrylamide-mediated quenching of Trp fluorescence of rFKBP22 and its derivatives.

The intrinsic tryptophan fluorescence spectra of I65P and V72P also differed completely from that of rFKBP22 (Fig. 3C). Compared to the tryptophan fluorescence spectrum of rFKBP22, tryptophan fluorescence spectrum of I65P yielded a less fluorescence intensity, whereas, that of V72P produced a higher intensity. The emission λmax values of the associated tryptophan fluorescence spectra of rFKBP22, I65P, V72P and A82P are 327 nm, 322 nm, 329 nm and 325 nm respectively (Fig. 3C). The spectrum of A82P partly overlaps with that of rFKBP22. The λmax value of V72P was red shifted compared to rFKBP22 indicating that its Trp residues may be more exposed. In contrast, Trp residues of I65P and A82P are relatively buried as the λmax value of these proteins were blue shifted. Collectively, substitution of the hydrophobic amino acid residues with a Pro residue in the helix α3 affects both the secondary and the tertiary structures of FKBP22 drastically.

Stern–Volmer plots [14] produced from the acrylamide quenching data in rFKBP22 and its mutants are primarily linear (Fig. 3D). The Stern–Volmer constant Ksv values, calculated from the slopes of the above plots, vary from 2.04 ± 0.08 to 2.49 ± 0.09 M−1 (Table 1), suggesting that the environments of Trp residues in the mutants were not significantly different (in comparison with that of rFKBP22; all p > 0.05) though they possess altered conformations.

Table 1

Ksv, Kcat/Km, and the Kd values.

Proteins	Ksv (M−1)a	Kd (µM)b	kcat/Km (µM−1 s−1)c
rFKBP22	2.32 ± 0.07	5.43 ± 0.57	0.818 ± 0.096
I65P	2.04 ± 0.08	5.06 ± 0.38	0.046 ± 0.010
V72P	2.49 ± 0.09	5.51 ± 0.26	0.491 ± 0.005
A82P	2.18 ± 0.07	5.41 ± 0.31	0.338 ± 0.026

Stern–Volmer constants (Ksv) were obtained from the slopes of the plots in Fig. 3D.

Kd values were estimated from Fig. 4A.

kcat/Km values were calculated from Fig. 4B.

Activities of the proline carrying helix α3 mutants

Structural alterations as apparent from Fig. 3 may lead to changes in the drug binding affinity and/or the PPIase activity of the mutants. To determine if the drug binding affinities of FKBP22 and its mutants are different, tryptophan fluorescence quenching of these proteins was studied in the presence of 0–20 µM rapamycin (Fig. 4A). The Kd values for the rapamycin-protein interactions (Table 1) do not vary notably, indicating that substitution of the non-polar amino acid residues with a Pro in α3 did not affect the rapamycin binding affinity of FKBP22.

Fig. 4

Rapamycin binding and RNase T1 refolding activities of rFKBP22 and its mutants. (A) Plots show the quenching of Trp fluorescence intensity of the indicated proteins (each 5 µM) in the presence of varying concentrations of rapamycin. (B) Plots exhibit the increase of Trp fluorescence during refolding of denatured RNase T1 in the presence/absence of the indicated proteins at 10 °C.

To determine the PPIase activities of rFKBP22 and its mutants, an RNase T1 refolding assay was performed in the presence and absence of these proteins independently. Fig. 4B shows that enzymatic activities of mutants differ from that of rFKBP22. The kcat/Km (enzymatic activity) values for all proteins were determined (from Fig. 4B) and presented in Table 1. All of the mutants possess less PPIase activity than rFKBP22. The loss of catalytic activity was also different among the different mutants. While the loss of PPIase activity in I65P was about 94%, those in V72P, and A82P were ∼40% and ∼59%, respectively. As the enzymatic activities of the mutants were significantly less than that of rFKBP22 (all p < 0.05), we suggest that the replacement of hydrophobic amino acid residues with a Pro residue in α3 severely affected the catalytic activity of rFKBP22.

Urea-induced unfolding of proline carrying helix α3 mutants

To compare the unfolding mechanism of the α3 mutants with that of rFKBP22 [26], we recorded their far-UV CD and intrinsic tryptophan fluorescence spectra in the presence of 0–7 M urea (Supplementary Fig. S2). The unfolding curves produced using the CD data of the proteins at 222 nm are given in Fig. 5A. I65P shows a biphasic curve, whereas the rest are monophasic (Fig. 5A). Conversely, all the tryptophan fluorescence intensity curves are monophasic in nature at 0–7 M urea (Fig. 5B). Upon raising the urea concentration, λmax values and intensity are both found to increase for tryptophan fluorescence, λmax going to ∼350 nm at the fluorescence intensity maximum (Supplementary Fig. S2). Interestingly, the curves for A82P and rFKBP22, generated by using either the Trp fluorescence or the CD data, partly overlap with the rFKBP22-specific curve, particularly, at the transition region. Further analyses by TUGE show that migration of rFKBP22 [26] and its mutants across the 0–8 M urea gradient produce nearly the sigmoidal curve-shaped bands (Fig. 6). Both spectroscopic and TUGE data indicate that the urea concentrations needed to initiate and terminate the unfolding of I65P and V72P are different from those of either rFKBP22 or A82P.

Fig. 5

Urea-induced unfolding of rFKBP22 and its mutants. (A) The plots describe the change of θ222 (ellipticity at 222 nm) values of the indicated proteins at 0–7 M urea. The θ222 values, derived from the CD spectra of the proteins (Fig. S2), were normalized as described [14]. (B) Plots show the change of the Trp fluorescence intensity values (at 326 nm) of the indicated proteins at 0–7 M urea. The Trp intensity values, obtained from the Trp fluorescence spectra of the proteins (Fig. S2), were normalized as described above. All lines through the spectroscopic signals indicate the best-fit curves.

Fig. 6

Migration of rFKBP22 and its derivatives across the transverse urea gradient polyacrylamide gel.

To test the reversibility in unfolding of rFKBP22 and its mutants at 0–7 M urea, they were denatured followed by their analyses using TUGE. All the proteins, like the folded counterparts (Fig. 6), show monophasic curve-shaped bands (Supplementary Fig. 3A). In addition, tryptophan fluorescence spectra of all the refolded proteins also completely superimpose with those of the corresponding native proteins (Supplementary Fig. 3B). Taken together, we conclude that urea-induced unfolding of rFKBP22 and its mutants are reversible in nature.

Previously, alteration of the length of α3 in rFKBP22 was shown to affect the molecule's stability [26]. To determine whether the helix-breaking mutations in α3 also affect the stability of rFKBP22, values of different thermodynamic parameters, such as ΔGW, Cm, m, and ΔΔG, were determined by fitting the sigmoidal curves (shown in Fig. 5B) to a two-state model [41]. Table 2 shows that the values of Cm (the most trustworthy thermodynamic parameter linked to the unfolding study) for rFKBP22, I65P, V72P and A82P are 3.5 ± 0.06 M, 4.03 ± 0.06 M, 2.69 ± 0.02 M and 3.49 ± 0.08 M, respectively. The ΔGW value of I65P, similar to its Cm value, was also significantly higher, whereas, that of V72P was considerably small in comparison with other proteins (all p < 0.05). The free energy change ΔΔG between rFKBP22 and I65P or V72P were greater than 1 kcal mol−1 (Table 2). However, the ΔGW or Cm values of rFKBP22 and A82P appear close to each other. Cm values of I65P and V72P determined from TUGE (Fig. 6) also support those estimated from spectroscopic studies (data not shown). In sum, the results suggest that V72P is the least stable and I65P is the most stable of the four proteins to urea denaturation.

Table 2

Thermodynamic parameters from the urea-induced unfolding curves.a

Protein	Cm (M)	M	ΔGW	ΔΔG
		(kcal mol−1 M−1)	(kcal mol−1)	(kcal mol−1)
rFKBP22	3.50 ± 0.06	1.83 ± 0.06	6.39 ± 0.34
I65P	4.03 ± 0.06	2.47 ± 0.09	9.95 ± 0.23	1.13 ± 0.01
V72P	2.69 ± 0.02	1.93 ± 0.02	5.19 ± 0.12	1.52 ± 0.11
A82P	3.49 ± 0.08	1.77 ± 0.02	6.19 ± 0.07	0.03 ± 0.01

To determine the thermodynamic parameters, the urea-induced unfolding curves, generated from the Trp fluorescence spectroscopic data of the indicated proteins, were analyzed as described in Materials and methods.

The unfolding curves of a protein, prepared using different spectroscopic data, will not overlap if there is synthesis of intermediates during its denaturation [51], [52], [53]. To understand whether the urea-induced unfolding of rFKBP22 and its variants occurs with the generation of intermediates, the fraction of unfolded proteins were determined (using all of the spectroscopic results) and plotted against the corresponding urea concentrations. The far-UV CD data of I65P yields a biphasic curve as expected (data not shown). In contrast, all others have sigmoidal ones (Fig. 7A). Like the I65P curves, the V72P or A82P ones also do not coincide when comparing CD and fluorescence, whereas those of rFKBP22 do [14], [26]. Hence, unlike rFKBP22, in the presence of 0–7 M urea, I65P, V72P and A82P may be unfolded via the synthesis of intermediate(s). To validate the above proposition, I320 (the Trp fluorescence intensity at 320 nm) values of these proteins were plotted against their I365 (Trp fluorescence intensity at 365 nm) values. Such plots are usually generated to recognize the hidden unfolding/refolding intermediates of proteins [54], [55], [56]. The plots yielded by V72P and A82P diverge from the linearity at urea concentrations of above 3 M (Fig. 7B), further indicating the synthesis of intermediates during the denaturation of these mutants. Conversely, the I320 versus I365 plot (Fig. 7B) does not clearly show the synthesis of an intermediate in I65P.

Fig. 7

Unfolding mechanism of rFKBP22 and its variants. (A) Plots of fractions of unfolded proteins versus urea concentrations. The fractions of unfolded protein molecules were determined by a standard procedure [41] using the CD and intrinsic Trp fluorescence spectra (Fig. S2) of the denoted proteins. (B) Phase diagrams show the urea-induced unfolding mechanism of I65P, V72P and A82P. I320 and I365 indicate the Trp fluorescence intensity at 320 and 365 nm, respectively.

Computational and statistical analyses of FKBP22 and its variants

MD simulation, a computational probe, has long been employed to understand the conformational changes in proteins with/without mutations [46], [57], [58], [59], [60]. To study the effects of α3 mutations on the conformation of FKBP22, we performed MD simulation with energy-minimized structures of both the wild-type and mutant FKBP22 proteins. Fig. 8A shows the RMSD profiles of all of the proteins during 10 ns simulation. Structures of different proteins appeared to diverge from each other roughly after 1 ns. After ∼7 ns, the order of structural deviation is I65P > A82P > V72P > FKBP22. At 10 ns, the RMSD values for the wild-type, A82P, V72P and I65P are 10.7, 13.3, 10.1, 13.6 Å respectively.

Fig. 8

MD simulation of rFKBP22 and its α3 substitution mutants. (A) RMSD profiles of the backbone atoms of the indicated proteins. (B) RMSF profiles of the amino acid residues of the denoted (dimeric) proteins. (C) Structural alteration of rFKBP22 and I65P during 10 ns MD simulation.

The RMSF profiles also showed a similar trend with the relatively higher fluctuations in the residues of mutant structures (Fig. 8B). For reasons not clearly known, a reasonably higher flexibility was observed in the A subunit (residues 1–206) of each mutant FKBP22 protein compared to that in the B subunit (residues 207–412).

The effects of mutations on the V-shape of the FKBP22 structure were also investigated by the MD simulation probe. As expected, the V-shape of the I65P structure was severely distorted, whereas those of FKBP22, V72P and A82P were largely retained at 10 ns (Fig. 8C and Supplementary Fig. 4).

To better understand the mechanism of V-shape loss in I65P, we performed PCA using the MD simulations of FKBP22 and its mutants. In all of the proteins, the first mode corresponded to the eigenvalue of >2 Å2 (Supplementary Movies 1–4), indicating that it would be sufficient to explain their principal motions reasonably [44]. The porcupine plots, generated with the first modes, showed the ways the motions of the mutants, particularly I65P, differed from that of the wild-type FKBP22 (Fig. 9). While the two α3-linked C-terminal domains in either FKBP22 or A82P (Fig. 9A or D) were moving in the reverse direction, those in I65P were moving towards each other (Fig. 9B). The domain movement in V72P appeared to be slightly similar to that of I65P (Fig. 9C). Taken together, the severe loss of the V-shape of I65P might be due to the abnormal movement of its domains.

Fig. 9

Porcupine plots demonstrating dominant motions of rFKBP22 (A), I65P (B), V72P (C), and A82P (D).

Supplementary video related to this article can be found at http://dx.doi.org/10.1016/j.biopen.2015.07.001.

The following are the supplementary video related to this article:

Supplementary Movie 1

Motion of rFKBP22 in first mode.

Supplementary Movie 2

Motion of I65P in first mode.

Supplementary Movie 3

Motion of V72P in first mode.

Supplementary Movie 4

Motion of A82P in first mode.

Discussion

Our studies demonstrated that the Pro substitution mutations in the helix α3 severely affected the shape, structure, protein folding ability and the unfolding mechanism of rFKBP22 (Fig. 2, Fig. 3, Fig. 4, Fig. 5, Fig. 6, Fig. 7 and Table 1). Mutations (particularly at positions 65 and 72) also altered the stability of rFKBP22 notably (Table 2). In contrast, these mutants did not lose the rapamycin binding affinity though the conformations of their C-terminal domains were altered to some extent (Fig. 3C). Previously, some rFKBP22 mutants harboring a deletion or an insertion mutation in the helix α3 also exhibited similar properties [26]. Taken together, we suggest that both the presence and the length of helix α3 are crucial for preserving the structure, protein folding ability and stability of FKBP22.

Of the substitution mutants, I65P was not only associated with the highest level of structural loss (including the substantial loss of its V-shape) (Fig. 3, Fig. 8) but also showed partial dissociation, particularly, at nanomolar concentrations (Fig. 2). Previously, an rFKBP22 mutant lacking I65 and the neighboring residues also showed partial dissociation at nanomolar concentrations and possessed an anomalous conformation [26]. In comparison with the wild-type FKBP22, this deletion mutant retained <1% PPIase activity. Various regions (such as C-terminal domain, the N-terminal domain, the V-shaped structure, and the length of helix α3) of FKBP22 and the related proteins were reported to be critical for their protein folding activity with a larger (i.e. protein) substrate [14], [17], [18], [19], [20], [26], [61]. The C-terminal domains of these proteins are, however, sufficient for their PPIase activity with a smaller (i.e. peptide) substrate. Although the CTD of I65P possessed an altered conformation (as evident from the Trp fluorescence spectroscopy), it retained fairly normal rapamycin binding affinity (Fig. 4A), indicating the presence of an intact catalytic center in this domain. Our quenching study (Fig. 3D) definitely supported this hypothesis as the accessibilities of two Trp residues (possibly forming the catalytic center) in the CTD of I65P were not changed notably. The CTD of I65P, therefore, may not be responsible for its decreased RNase T1 refolding ability (Fig. 4B). A previous MD simulation study indicated that a helix α3 region of Legionella Mip (encompassing the residues ∼73–83) is one of the dynamic regions in this enzyme [62]. The dynamic nature of the above helix α3 region possibly facilitates the Mip and the related proteins to hold the bigger (protein) substrate within their V-shaped gaps [20], [62]. The crystallographic B-values of Legionella Mip residues 55 to 70 (equivalent to residues 55–70 of E. coli FKBP22) were found higher than its mean B-value, indicating that the N-terminal end of helix α3 of this protein is relatively flexible in nature [15]. In I65P, destabilization of α3 due to the loss of one main chain hydrogen bond possibly made its N-terminal end more flexible, which eventually altered the V-shape of this mutant protein drastically. Taken together, we suggest that the severe loss of PPIase activity of I65P might be due to its altered V-shape and/or its partial dissociation. The conformational changes of two other substitution mutants were less drastic and hence accompanied by a moderate loss of their enzymatic activity. Collectively, the N-terminal part of helix α3 formed by residues ∼55–70 contributes relatively more to keep the structure, shape, stability and function of FKBP22 intact.

Protein structures are usually stabilized by various non-covalent bonds such as hydrogen bonds, hydrophobic interactions, ionic bonds, etc. [63], [64], [65]. Several hydrogen and ionic bonds in the helix α3 were also reported to be critical for the stability of Legionella Mip [15]. The disruption of helix α3 by Pro substitution at positions 65, 72 and 82 had different effects on the overall conformation of FKBP22, which indicated that FKBP22 and its substitution mutants could be stabilized by a different number of non-covalent bonds. Our MD simulation study indicated that FKBP22 and its substitution mutants are composed of dissimilar numbers/extents of hydrogen bonds, ionic bonds and surface area (data not shown). All of these stabilizing factors and the structural alterations together somehow contributed to the highest or lowest stability of I65P or V72P. The reason as to why A82P and rFKBP22 are equally stable is unknown.

Conclusion

E. coli FKBP22 and the related proteins (including Mip-like virulence factors) are the two-domain PPIase enzymes those dimerize and adopt a V-shape through the interaction of their N-terminal domains. Using three Pro substitution mutations in the domain-connecting helix (α3) of E. coli FKBP22, we have demonstrated that the presence of helix α3 (particularly its N-terminal end) is critical for maintaining the V-shape, secondary structure, tertiary structure, dimeric status, stability, and the protein folding ability of this enzyme. The disruption of helix α3, however, did not alter its drug binding affinity significantly.

Conflict of interest

The authors declare no competing financial interest.