The Role of the Hinge Loop in Domain Swapping

Bovine seminal ribonuclease (BS-RNase) is a covalent homodimeric enzyme homologous to pancreatic ribonuclease (RNase A), endowed with a number of special biological functions. It is isolated as an equilibrium mixture of swapped (MxM) and unswapped (M=M) dimers. The interchanged N termini are hinged on the main bodies through the peptide 16–22, which changes conformation in the two isomers. At variance with other proteins, domain swapping in BS-RNase involves two dimers having a similar and highly constrained quaternary association, mainly dictated by two interchain disulfide bonds. This provides the opportunity to study the intrinsic ability to swap as a function of the hinge sequence, without additional effects arising from dissociation or quaternary structure modifications. Two variants, having Pro19 or the whole sequence of the hinge replaced by the corresponding residues of RNase A, show equilibrium and kinetic parameters of the swapping similar to those of the parent protein. In comparison, the x-ray structures of MxM indicate, within a substantial constancy of the quaternary association, a greater mobility of the hinge residues. The relative insensitivity of the swapping tendency to the substitutions in the hinge region, and in particular to the replacement of Pro19 by Ala, contrasts with the results obtained for other swapped proteins and can be rationalized in terms of the unique features of the seminal enzyme. Moreover, the results indirectly lend credit to the hypothesis that the major role of Pro19 resides in directing the assembly of the non-covalent dimer, the species produced by selective reduction of the interchain disulfides and considered responsible for the special biological functions of BS-RNase.

Domain swapping, a process by which two or more protein molecules exchange identical structural elements to form dimers or higher oligomers (1), has been observed in an increasing number of proteins. More than 50 crystal structures of domain-swapped proteins have been deposited in PDB 1 since the first x-ray structure of a protein showing the interchange of N-terminal regions between the two polypeptide chains was reported (2). However, the possible physiological significance of this phenomenon is still unclear (3). It has been proposed that a large number of proteins may undergo this process under physiological or pathological conditions; thus it could represent a mechanism to regulate function, or even an evolutionary strategy to increase protein complexity.
In the panorama of the swapping proteins a special case is represented by bovine seminal ribonuclease (BS-RNase), a homodimeric protein in which the two subunits are covalently linked through two disulfide bridges between cysteines 31 and 32 of one subunit with cysteines 32 and 31 of the partner subunit, respectively (4). In this protein the swapping process involves two dimers, in which the two subunits change their tertiary structure within a basically invariant quaternary assembly imposed by the two interchain disulfides: in the dimerdubbed MxM the N-terminal arms (residues 1-15) are exchanged, or swapped, between the two subunits, whereas in the dimer indicated as MϭM no swapping occurs. Thus, in this particular system the swapping phenomenon does not depend on the overall concentration of the protein. Furthermore, in the quaternary structure of MxM, the acceptable values of the end-to-end distance, spanned by the hinge peptide, are almost as sharply restricted as they are within the tertiary structure of the unswapped dimer. This finding is at variance with what is usually observed in the swapping process, where a monomer to dimer (M/D) transition is commonly observed and the swapped dimer often presents a considerable degree of flexibility and, therefore, a certain degree of variability of the end-toend distance of the hinge peptide. In the latter case, the swapped state is expected to become statistically more favored as the rigidity of the hinge is increased. Indeed, this argument has been used to explain the elevated frequency of proline in the hinge peptide sequence of proteins prone to swap (5). Furthermore, in an M/D transformation other parameters may influence significantly the process, such as the nature and the extent of "O-interface," i.e. the additional interface formed in the dimer and exposed to the solvent in the monomer (1). In the MxM/MϭM equilibrium of BS-RNase, the quaternary struc-ture is highly preserved and the O-interface 2 varies only for the different conformation adopted by the hinge peptide in the two dimers (6). This feature offers the unique opportunity to isolate the effects of the hinge sequence on the equilibrium ratio between MxM and MϭM. In the native protein the relative amount of the two dimers is about 70:30 (7). Interestingly, the selective reduction of the interchain disulfides produces the species which could be involved into a M/D swapping transformation, as MxM gives rise to a non-covalent swapped dimer (NCD), whereas MϭM readily dissociates into monomers.
BS-RNase sequence is 81% identical to that of bovine pancreatic ribonuclease (RNase A), the first protein that was shown to dimerize via the N termini interchange between the two intervening chains, upon lyophilization in acetic acid (8). For this protein, however, more recent experiments have demonstrated that only a minor fraction of the dimers is swapped at the N terminus, whereas a major fraction is swapped at the C terminus (9 -11). In the former dimer the contacts at the socalled "C-interface," i.e. the interface between the swapped domain (residues 1-15) and the major domain (residues , are identical to those found in the covalent dimers of BS-RNase. The sequence alignment of the two proteins shows that four substitutions out of a total of 23 are located in the 16 -22-hinge region. In order to clarify the actual role of the hinge sequence in the swapping process of BS-RNase, a homologue-scanning mutagenesis approach has been followed, using as reference the sequence of RNase A. We have prepared two mutants, Ala 19 -BS-RNase and Ser 16 -Thr 17 -Ala 19 -Ala 20 -BS-RNase, in which either Pro 19 or all four residues of the BS-RNase sequence have been substituted with the corresponding ones of the pancreatic enzyme. Here we report the x-ray structures of the MxM form of the two mutants and discuss the results on the basis of the MxM/MϭM equilibrium data measured in solution for Ser 16 -Thr 17 -Ala 19 -Ala 20 -BS-RNase and those previously published for Ala 19 -BS-RNase and for the parent BS-RNase (12).
PCR amplification was performed with an Eppendorf Mastercycler amplifier as previously described (12). For the tetramutant the two mutagenic primers 5Ј-TAGCAGAGGTGCTGCTGTC-3Ј and 5Ј-AA-GAGCTACCAGCAGAG-3Ј were used in succession (nucleotides that represent mutations are underlined). The amplified, mutated genes were separated, excised, and purified from the agarose gel followed by cloning into the pET-22b(ϩ) plasmid between HindIII and NdeI sites. Mutations were confirmed by DNA sequencing.
To avoid heterogeneity (see "Results and Discussion"), the basic sequence of BS-RNase contains the substitution of Asn 67 with an aspartic residue. This modified sequence, together with the N-terminal Met, constitutes the parent protein (henceforth referred to as mBS for the monomeric species or BS-RNase for the dimers), whereas the unmodified one is referred to as native protein.
Recovery of Proteins-The proteins were expressed in Escherichia coli and purified in monomeric form, with cysteines 31 and 32 linked to two glutathione molecules, as previously described (14). Monomers with cysteines 31 and 32 in the reduced form were obtained by selective reduction of the mixed disulfide bridges with a 5:1 molar excess of dithiothreitol for 20 min at room temperature in 0.1 M Tris acetate buffer, pH 8.4. The samples were either carboxyamidomethylated with iodoacetamide (15), to obtain the monomeric proteins used for CD, or dialyzed against 0.1 M Tris acetate, pH 8.4 for 20 h at 4°C, to obtain dimers. The last step of the purification procedure was always a gel filtration on Sephadex G-75 to separate monomers from dimers. All dimerization steps were performed at 4°C.
Protein concentration was measured by UV spectrophotometry assuming ⑀ (0.1%, 278 nm, 1 cm) ϭ 0.5. Protein homogeneity was checked by SDS-PAGE and MALDI-TOF mass spectra, registered at "Sezione di Spettrometria di Massa" of the CIMCF, University of Naples Federico II. The correct folding of all the monomers was checked by CD. The enzymatic activity on yeast (16) was comparable to that of native mBS, so confirming the correct folding of the active site.
CD Spectra Measurements-CD spectra were recorded with a Jasco J-715 spectropolarimeter equipped with a Peltier type temperature control system (Model PTC-348WI). Molar ellipticity per mean residue, [] in deg cm 2 dmol Ϫ1 , was calculated from the equation: [] ϭ [] obs ⅐mrw/10⅐l⅐C, where [] obs is the ellipticity measured in degrees, mrw is the mean residue molecular mass, 117 Da (15), C is the protein concentration in g l Ϫ1 and l is the optical path length of the cell in cm. A 0.1-cm path length cell and a protein concentration of about 0.3 mg ml Ϫ1 in 10 mM sodium acetate buffer, pH 5.0, were used. CD spectra were recorded at 25°C with a time constant of 16 s, a 2-nm bandwidth, and a scan rate of 5 nm min Ϫ1 ; they were signal-averaged over five scans at least, and baseline corrected by subtracting the buffer spectrum.
Thermal unfolding curves were recorded in the temperature scan mode at 222 nm from 25 up to 85°C with a scan rate of 1.0 K min Ϫ1 .
NMR-NMR measurements were performed on a Bruker DRX500 spectrometer. All spectra were collected using the standard Bruker pulse sequence library. Protein concentration was 2 mM in 95% H 2 O, 5% D 2 O, pH 5.65. Spectra were processed with NMRPipe (17) and analyzed with NMRView (18) programs.
Kinetics of Interconversion of Dimeric Forms-To follow the interconversion kinetics, dimer samples were incubated at 37°C. At given times, aliquots were withdrawn, the interchain disulfide bridges were selectively reduced as described (7), and the mixture was chromatographed on an analytical Superdex 75 HR 10/30 column (Amersham Biosciences). The amount of MxM and MϭM was evaluated by integrating the peaks of dimer and monomer, respectively.
Extent of the N-terminal Swapping at Equilibrium-Cross-linking experiments were done using divinyl sulfone (DVS) as a 10% solution in ethanol. The dimers (20 g) in sodium acetate buffer (100 mM, pH 5, 100 l) and DVS (1 l of the 10% solution) were incubated at 30°C (19). This is approximately a 1,000-fold excess of sulfone to each subunit of the protein. Aliquots were withdrawn over a period of 96 h, quenched with 2-mercaptoethanol (final concentration 200 mM), incubated for 15-30 min at room temperature, and loaded on a gel for reducing SDS-PAGE. A qualitative estimation of the monomer to cross-linked dimer ratio was obtained by Coomassie Blue staining.
Crystallization and Data Collection-The swapped dimers of Ala 19 -BS-RNase and Ser 16 -Thr 17 -Ala 19 -Ala 20 -BS-RNase variants were crystallized using protein solutions containing the equilibrium mixture of the swapped and unswapped form without further purification. In detail, Ala 19 -BS-RNase and Ser 16 -Thr 17 -Ala 19 -Ala 20 -BS-RNase crystals were grown at room temperature by the vapor diffusion sitting drop method. Equal volumes of protein (27 mg/ml) and of a solution containing 30% (w/v) PEG 4000, 0.1 M Tris-HCl, pH 8.5 and 0.2 M sodium acetate were mixed and equilibrated against a 750-l reservoir containing the same precipitant solution. Single crystals of the two mutants were obtained after 1 day and present very similar morphology. They belong to the orthorhombic space group P2 1 2 1 2 1 and are isomorphous to the wild-type protein crystallized under very similar conditions (PDB code 1R5D) (20).
Diffraction data were collected on a ENRAF-NONIUS DIP area detector equipped with a FR591 rotating anode of the Istituto di Biostrutture e Bioimmagini (CNR, Naples). Ala 19 -BS-RNase data collection was carried out at room temperature up to 2.20-Å resolution. Because of the lower resolution of the diffraction data, a Ser 16 -Thr 17 -Ala 19 -Ala 20 -BS-RNase crystal was cryo-cooled at 100 K by using 18% (v/v) glycerol as cryo-protector, and the diffraction data were collected up to the same resolution (2.20 Å). The two data sets were processed and scaled with the HKL package (21). Crystal parameters and data collection statistics are given in Table II.
Structure Determination and Refinement-The crystal structures were solved by the molecular replacement method. The atomic coordinates of the native enzyme, stripped of all solvent molecules and hinge peptide residues, were used as starting model. Structure refinement, performed by using the CNS program (22), began with a cycle of rigid body refinement. In this step the starting model was divided in four groups formed by the two N-terminal helical segments (residues 1-15) and the two main bodies (residues 24 -124). Each mutant was then refined by alternating positional and temperature refinement with manual building by using O (23). The missing hinge peptide residues were built by fitting the electron density from difference Fourier maps. Water molecules were then added to the models through the automatic protocol of CNS together with the contribution of the disordered solvent.
Reflections with intensity greater than 2(I) in all resolution range (20.00 -2.20 Å for both mutants) were included in the minimization procedure. The final models of the two proteins have R factor /R free indexes of 0.171/0.202 and 0.208/0.259, respectively. The models have good geometry as evaluated with WHATCHECK (24). The full lists of refinement statistics are reported in Table II.
Superimposition of the mutants with the wild-type protein was achieved by using the C ␣ carbon atoms of residues 2-15 and 23-124 of the two chains. Because of the swapping, the quaternary structure was compared by first superimposing one structural unit (residues 2-15 of one chain and 23-124 of the second chain) and, successively, computing the angle needed to superimpose the second structural unit.

RESULTS AND DISCUSSION
Characterization of the Monomers-All the plasmids were coding for Asp at position 67 in order to avoid heterogeneity arising from the spontaneous deamidation of Asn 67 , which characterizes the native enzyme (25). Asn 67 is located in a disulfide-linked octapeptide loop (65-72) exposed to the solvent and on the most far site with respect to the swapping domain and to the O-interface. This substitution in RNase A has been shown not to influence the correct folding of the chain and its thermal stability (26). mBS shows a similar behavior, and, in addition, the extent of the swapping phenomenon in the dimeric species is not altered (12).
The first products of our purification procedure were about 15 mg/liter of recombinant BS-RNase or its variants in monomeric form, with cysteines 31 and 32 linked to two glutathione molecules. The thermal stability of the two mutant proteins was monitored by CD spectroscopy. Fig. 1 (27). For a more accurate evaluation of the effect of the mutations on the solution structure of the monomeric proteins we resorted to two-dimensional NMR spectra. The overlay of TOCSY spectra of Ser 16  most resonances were coincident, thus confirming that the two monomers have a very similar conformation. In the expanded regions reported in Fig. 2 it is evident the presence of three new NH-CH 3 connectivities in the spectrum of Ser 16 -Thr 17 -Ala 19 -Ala 20 -mBS (black), tentatively assigned to Thr 17 , Ala 19 , and Ala 20 .
The analysis of sequential contacts in the NOESY spectrum allowed the proton assignment of all the spin systems of the hinge residues, which is reported in Table I. The table also reports the proton resonances of the 16 -22 region for mBS and for RNase A (28) for comparison. Despite the sequence identity, the chemical shift values of Ser 16 -Thr 17 -Ala 19 -Ala 20 -mBS show a striking difference with those of RNase A, suggesting that the hinge loop has a different local environment in the two proteins.
Although the hinge peptide was found to be flexible in RNase A too (28), its greatly enhanced mobility in Ser 16 -Thr 17 -Ala 19 -Ala 20 -mBS clearly indicates that more substitutions external to the hinge region, such as Ser 80 replaced by Arg in BS-RNase, play a role in fixing this peptide in RNase A. In conclusion, the disorder, observed for the hinge region of the monomeric derivative of BS-RNase in the solid state (29) and in solution (28), appears to be also a feature of the mutants. Moreover, it seems reasonable to assume that similar disorder characterizes the corresponding unswapped dimers, where each subunit shares the global fold of the monomeric derivatives.
Extent of the Swapping-BS-RNase and its variants in their monomeric form, with cysteines 31 and 32 linked to two glutathione molecules, can be easily converted into dimers by selective reduction of the disulfide bridges followed by air oxidation of the exposed sulfhydryls. The characterization of the swapping process for Ala 19 -BS-RNase has been already reported (12). Two independent methods have been used to evaluate the  swapping tendency of the other mutant. The first protocol is based on the higher reactivity of the interchain disulfide bridges with respect to the intrachain ones (7). As in the case of the parent protein and of Ala 19 -BS-RNase, Ser 16 -Thr 17 -Ala 19 -Ala 20 -BS-RNase folds mainly in the unswapped form, and in freshly prepared samples the MxM content is about 15%. Fig. 3 displays the course of the MϭM to MxM conversion as a function of the incubation time at 37°C. The process in Ser 16 -Thr 17 -Ala 19 -Ala 20 -BS-RNase is very similar to that reported for BS-RNase, as it reaches the same 70:30 equilibrium between swapped and unswapped dimers, except for a slight increase in the conversion rate. Fig. 3 also reports the time course of the reverse process, namely the conversion of MxM into MϭM, which confirms the equilibrium ratio previously found. Both reactions were essentially complete within 2 days, whereas at 4°C the interconversion was effectively blocked (data not shown).
The extent of swapping on the equilibrium mixtures was also assessed by cross-linking experiments with divinyl sulfone (DVS), followed by SDS-PAGE analysis under reducing conditions. DVS joins covalently the two histidines of the active site (His 12 and His 119 ), which belong to the same subunit in MϭM and to different subunits in MxM. Under reducing and denaturing conditions, either a product of mass 27,000 Da (from MxM), or a product of mass 13,500 Da (from MϭM) is obtained on gel electrophoresis (19). The reaction requires more than 24 h to be completed, thus this method is not suitable to follow kinetically the interconversion. However, it gives for the equilibrium ratio between MxM and MϭM a semiquantitative indication, which does not depend on the reactivity of the disulfide bridges. Fig. 4 illustrates the time-dependent course of this reaction for Ser 16 -Thr 17 -Ala 19 -Ala 20 -BS-RNase in comparison with that of the parent BS-RNase. At any time, the two proteins show the same relative amount of exchanging and non-exchanging forms, thus indicating a very similar equilibrium composition.
Overall Structures-For comparison with the native MxM forms of the enzyme, data collection for both mutants was initially carried out at room temperature. Since crystals of Ser 16 -Thr 17 -Ala 19 -Ala 20 -BS-RNase diffracted at lower resolution (2.5 Å) and the quality of the diffraction data was considerably lower with respect to the other mutant, the data were recollected at 100 K. The final models of Ala 19 -BS-RNase and Ser 16 -Thr 17 -Ala 19 -Ala 20 -BS-RNase, containing 96 and 102 solvent molecules respectively, have been refined to R factor /R free values of 0.171/0.202 and of 0.208/0.259 using diffraction data up to 2.20-Å resolution. A full list of refinement statistics of the two proteins is reported in Table II. In both cases the quality of the electron density maps allowed a detailed description of nearly the whole molecule, with the exclusion of the hinge peptide of one chain of Ala 19 -BS-RNase.
The overall structure of the two mutants is very close to that of the MxM isomer of the wild-type enzyme (PDB code 1R5D). Fig. 5 shows the ribbon representation of Ser 16 -Thr 17 -Ala 19 -Ala 20 -BS-RNase. The two chains are related by a local 2-fold axis (the actual rotation is 177°in both cases); the root mean square deviation (RMSD) between corresponding C ␣ atoms is 0.33 Å for Ala 19 -BS-  With respect to the native enzyme the RMSD is 0.73 Å and 0.63 Å, respectively. When the comparison is limited to a single structural unit (residues 1-15 of one chain and 23-124 of the partner chain), the RMSD is halved as a result of a small difference in the relative orientation of the two structural units, which is less than 5°in both mutants. These results indicate that the substitutions at the hinge peptide only marginally perturb the quaternary assembly, which is highly constrained by the interchain disulfides and by the swapping. Indeed, even in the absence of the swapping, the quaternary association of the native enzyme, as shown by the structure of native MϭM (6) is hardly affected, and this is also expected to hold for the MϭM form of the mutants.
In both mutants, the substitution of Asn 67 of the native sequence with Asp does not modify the structure of the loop in which the residue is inserted, neither the hydrogen bonding network in which the asparagine side chain is involved. Even the nearby side chains of the catalytically important Asp 121 and His 119 are not significantly perturbed by the presence of the additional negative charge carried by Asp 67 .
Hinge Peptide-The electron density associated with the hinge region (residues 16 -22) of both mutants is significantly less well defined than in the remaining part of the molecules. In the native MxM forms, studied in two different crystal environments (2,20), and in the complexes with dinucleotides (20,30), the hinge peptide of the two chains adopts different structures (F and E), both characterized by the presence of a 1-4 ␤-bend with Pro 19 in the second position of the turn. The switch between the two structures occurs at Gly 16 , which adopts either a folded (F) or an extended (E) conformation (Fig.  6). In the F structure, Pro 19 side chain of one subunit is well placed in a niche lined by the side chains of Tyr 25 and Gln 101 of the other subunit. In the E structure the extended conformation of Gly 16 determines a different orientation of the ␤-bend and displaces the proline side chain out of the pocket. In this case the turn is somewhat distorted and the distance between the carbonyl of Ser 18 and the amide nitrogen of Ala 21 is slightly too long for a good hydrogen bonding interaction. In the crystal form of the wild-type enzyme, isomorphous to that of the present mutants (20), the F conformation is also stabilized by a number of hydrogen bonding interactions with a neighboring molecule, whereas the hinge peptide in the E conformation is relatively free from packing interactions.
As for the present mutants, the hinge peptide of Ala 19 -BS-RNase corresponding to the E conformation is fully disordered, whereas the second peptide can be confidently traced in the electron density map (Fig. 7a), and its structure approximately resembles that of the F conformation.
Because of the uncertainty in the orientation of the peptide groups, a comparison with the native structure is better grasped in terms of the virtual C ␣ i -C ␣ iϩ1 bond representation of the chain, which yields an almost unequivocal description of the backbone conformation. Indeed the torsion angles about the virtual bonds, given in Table III, clearly show the similarity of the peptide conformation with the F conformation of the native enzyme. For comparison, the values of angles for an unswapped conformation of the hinge peptide is also given; since this peptide is disordered in both MϭM (6) and in the monomeric derivative of BS-RNase (29), the conformational parameters presented in Table III are those of RNase A refined at atomic resolution at pH 5.9 (PDB code 1KF3) (31). The hinge peptide of this protein is also shown in Fig. 6 as a model of the unswapped conformation.
With respect to native MxM, the 1-4 ␤-bend, encompassing residues 18 -21, is partially disrupted, and the cavity, hosting the wild-type proline side chain, is occupied in the mutant by two water molecules, which bridge through hydrogen bonds the OG of Ser 18 to OG of Ser 82 and to OE1 of Gln-101 of the partner chain. The peptide forms a number of hydrogen bonds with a neighboring molecule, as for the native enzyme, although the specific pattern is slightly modified. The results, therefore, indicate that the replacement of Pro 19 by Ala considerably increases the flexibility of the hinge peptide, which is fully disordered in one chain and is more ordered in the partner chain, likely because of the stabilizing interactions due to crystal packing. Surprisingly, in the case of the tetramutant, both peptides present a better defined structure (Fig. 7b) with respect to Ala 19 -BS-RNase. It should be stressed, however, that in this case the diffraction data were collected at T ϭ 100 K, and it is conceivable that the low temperature may push each peptide to adopt a single conformation as it occurs in the native protein, despite its intrinsic greater flexibility. Also in this case, the conformation of the hinge peptide is different in the two chains and closely resemble the F and E conformations, respectively, of the native enzyme (Table III), showing that the replacement of Gly 16 by Ser does not prevent the peptide to assume the E conformation. CONCLUSION The data in this report on the MxM/MϭM mixtures of the two mutants clearly demonstrate that the substitutions do not significantly affect the equilibrium ratio in solution of the two isomers nor the kinetics of the swapping. Thus, the presence of a Pro in the hinge peptide of BS-RNase does not appear to be important for the swapping. However, it should be stressed that, at variance with the reported statistical analysis mostly performed on proteins involved into a M/D conversion process (3), the domain swapping in the seminal enzyme is an equilibrium process established between two dimers. They both have a well defined quaternary association, which remains practically unaffected not only by the swapping but also by the substitutions at the hinge site. Indeed, the present analysis shows that the MxM form of the two variants, apart from few minor adjustments, is very similar to that of native MxM. Moreover, given the similarity between MxM and MϭM of the native enzyme, it can be expected that the MϭM form of each mutant is also very similar to its MxM counterpart.
As for the hinge peptide, the x-ray data show that this region is more mobile in the MxM dimer of the mutants than in the native one. The peptide is also highly disordered in native MϭM (6) as well as in its monomeric derivative (29). Although no x-ray data are available for the MϭM forms of the mutants, NMR data of their monomeric derivatives indicate a high flexibility of the 16 -22 region, suggesting that the unswapped form of the peptide suffers a disorder similar to the one found in the native protein. Therefore, it is reasonable to conclude that the substitutions increase the flexibility of the swapped hinge peptide, thus reducing, with respect to the native protein, the entropy loss which accompanies the MϭM to MxM transformation. On the other hand, in the MxM form of the mutants, the disruption of good contacts between the hinge peptide and the protein matrix, and in particular of the stabilizing interactions of the proline side chain within the cavity formed by the side chains of Tyr 25 and Gln 101 , also reduces the enthalpic gain in the same transformation. Thus, the insensi-tivity of the swapping equilibrium to the substitutions in the hinge region appears to be the result of a subtle balance of enthalpic and entropic effects.
From an evolutionary point of view, the present results clearly contrast the general expectation that the substitutions in the hinge region of the seminal enzyme with respect to RNase A could be related to a greater efficiency of the swapping in the seminal enzyme. Vice versa, they indirectly lend new credit to a recently published hypothesis (32) regarding the role of these substitutions and in particular that of proline in position 19 of the seminal sequence. According to this hypothesis, their major role is associated to the stabilization of a quaternary structure of the swapped non-covalent dimer NCD, i.e. the dimer with the interchain disulfides broken, which is capable to evade the ribonuclease protein inhibitor and is considered responsible for the antitumor action of the enzyme.