Receptor recognition sites reside in both lobes of human serum transferrin

The binding of iron by transferrin leads to a significant conformational change in each lobe of the protein. Numerous studies have shown that the transferrin receptor discriminates between iron-saturated and iron-free transferrin and that it modulates the release of iron. Given these observations, it seems likely that there is contact between each lobe of transferrin and the receptor. This is the case with chicken transferrin, in which it has been demonstrated unambiguously that both lobes are required for binding and iron donation to occur [Brown-Mason and Woodworth (1984) J. Biol. Chem. 259, 1866-1873]. Further support to this contention is added by the ability of both N- and C-domain-specific monoclonal antibodies to block the binding of a solution containing both lobes [Mason, Brown and Church (1987) J. Biol. Chem. 262, 9011-9015]. In the present study a similar conclusion is reached for the binding of human serum transferrin to the transferrin receptor. With the use of recombinant N- and C-lobes of human transferrin produced in a mammalian expression system, we show that both lobes are required to achieve full binding. (Production of recombinant C-lobe in the baby hamster kidney cell system is reported here for the first time.) Each lobe is able to donate iron to transferrin receptors on HeLa S3 cells in the presence of the contralateral lobe. The results are not identical with the chicken system, because the C-lobe alone shows a limited ability to bind to receptors and to donate iron. Further complications arise from the relatively weak re-association between the two lobes of human transferrin compared with the re-association of the ovotransferrin lobes. However, domain-specific monoclonal antibodies to either lobe block the binding of N- and C-lobe mixtures in the human system, thus substantiating the need for both.

A new ultrasensitive scanning calorimeter

A new ultrasensitive differential scanning calorimeter is described, having a number of novel features arising from integration between hardware and software. It is capable of high performance in either a scanning or isothermal mode of operation. Upscanning is carried out adiabatically while downscanning is nonadiabatic. By using software-controlled signals sent continuously to appropriate hardware devices, it is possible to improve adiabaticity and constancy of scan rate through use of empirical prerun information stored in memory rather than by using feedback systems which respond in real time and generate thermal noise. Also, instrument response time is software-selectable, maximizing performance for both slow- and fast-transient systems. While these and other sophisticated functionalities have been introduced into the instrument to improve performance and data analysis, they are virtually invisible and add no additional complexities into operation of the instrument. Noise and baseline repeatability are an order of magnitude better than published raw data from other instruments so that high-quality results can be obtained on protein solutions, for example, using as little as 50 microg of protein in the sample cell.

Association of the two lobes of ovotransferrin is a prerequisite for receptor recognition. Studies with recombinant ovotransferrins

Different recombinant N-lobes of chicken ovotransferrin (oTF/2N) have been isolated from the tissue-culture medium of baby hamster kidney cells transfected with the plasmid pNUT containing the relevant DNA coding sequence. Levels of up to 40, 55 and 30 mg/1 oTF/2N were obtained for constructs defining residues 1-319, 1-332 and 1-337-(Ala)3 respectively. In addition, a full-length non-glycosylated oTF was expressed at a maximum of 80 mg/1 and a foreshortened oTF consisting of residues 1-682 was expressed at a level of 95 mg/l. These preparations were then used to produce, proteolytically, two different C-lobes (oTF/2C) comprising residues 342-686 and 342-682. The purified recombinant N-lobes (oTF/2N) are similar to the proteolytically derived half-molecule with regard to immunoreactivity and spectral properties; they show some interesting differences in thermal stability. A sequence analysis of the cDNA revealed six changes at the nucleotide level that led to six differences in the amino acid sequence compared with that reported by Jeltsch and Chambon [(1982) Eur. J. Biochem. 122, 291-295]. Electrospray mass spectrometry gives results consistent with these six changes. Interaction between the various N- and C-lobes was measured by titration calorimetry. Studies show that only those lobes that associate in solution are able to bind to the receptors on chick embryo red blood cells. These findings do not support a previous report by Oratore et al.

Production and isolation of the recombinant N-lobe of human serum transferrin from the methylotrophic yeast Pichia pastoris

The N-lobe of human serum transferrin has been expressed in the methylotrophic yeast Pichia pastoris by placing the hTF/2N cDNA under the control of the methanol-inducible alcohol oxidase promoter. Following induction with methanol, the N-lobe was efficiently secreted into a basal salt medium in shake flasks at a level of 150-240 mg/liter. As judged by mobility on SDS-PAGE, immunoreactivity with two domain-specific monoclonal antibodies, and both thermal stability and spectral properties (indictative of correct folding and ability to bind iron), the recombinant N-lobe produced by the yeast cells appears to be identical to that produced in a mammalian expression system. Electrospray-mass spectrometry and a third domain specific antibody, however, show that approximately 80% of the protein from the yeast cells contains one or two hexose residues.

Effects of C-terminal deletions on the conformational state and denaturation of phosphoglycerate kinase

Phosphoglycerate kinase (PGK) contains two domains of approximately equal size, both of the alpha/beta type. An alpha-helix consisting of the middle section of the 415-amino acid polypeptide chain, and the N- and C-termini reside in the interdomain hinge region [Watson, H. C., et al. (1982) EMBO J. 1, 1635-1640]. The C-terminal end is an integral part of the N-terminal domain. The consequences of the deletion of fifteen and three C-terminal amino acids on the conformational state and on the guanidine hydrochloride-induced and thermal unfolding of PGK were investigated by using near- and far-UV CD, tryptophan fluorescence, 1-anilinonaphthalene-8-sulfonic acid binding, accessibility to chemical modification, and differential scanning calorimetry. The results of these studies indicate that the conformations of both domains and of the interdomain region were altered by these deletions. In the absence of the 15-amino acid C-terminal peptide [delta(401-415)], the N-terminal domain exhibits several characteristics of a molten globule state, whereas the C-terminal domain retains native-like, although distinctly different, tertiary structure. Deletion of three C-terminal amino acids [delta(413-415)] also globally affects PGK conformation, although to a much lesser extent. Both C-terminal deletions resulted in a significant decrease in protein stability, as demonstrated by their increased susceptibility to guanidine-induced and thermal denaturation. These results suggest that the formation of a native tertiary fold of PGK requires the presence of a complete polypeptide chain.

The serine receptor of bacterial chemotaxis exhibits half-site saturation for serine binding

Ligand binding to the serine receptor of Escherichia coli has been studied using isothermal titration calorimetry. Bacterial inner membranes enriched in the serine receptor (Tsr) were titrated as sonicated membrane samples and after solubilization in octyl beta-D-glucopyranoside (OG) to determine the number of moles of ligand bound per mole of receptor (n), the binding constant (Ka), and the enthalpy of binding (delta H) of serine to the receptor. The n value for serine binding to OG-solubilized Tsr protein (n = 0.5) was consistent with one molecule of serine binding to a receptor dimer, but in sonicated inner membrane samples, the n value was smaller (n approximately equal to 0.25), indicating that not all of the binding sites were accessible to added serine. At 7 and 27 degrees C, the values for Ka and delta H were equivalent for the membrane and OG-solubilized samples and were found to be 4.7 x 10(4) M-1 and -15 kcal/mol, and 3.6 x 10(4) M-1 and -18 kcal/mol, respectively. The influence of covalent modification at the sites of methylation on the affinity of the receptor for serine was also investigated, and found to have only a modest effect. The property of half-site saturation is suggestive of models for transmembrane signaling where the receptor subunit interactions are modulated by ligand binding.

Calorimetric studies of serum transferrin and ovotransferrin. Estimates of domain interactions, and study of the kinetic complexities of ferric ion binding

Human serum transferrin and hen ovotransferrin have been studied by differential scanning calorimetry (DSC), in an effort to quantitatively estimate the free energy of interaction of the N- and C-domains in each protein and to further understand their interaction with chelated ferric ions. In the case of serum transferrin, separate DSC transitions are observed for the two domains while only a single, coupled transition is seen for ovotransferrin. Although domain interactions are somewhat larger for ovotransferrin (-4100 cal/mol) than for serum transferrin (-3100 cal/mol), the major cause of separated transitions for serum transferrin is that the difference in intrinsic folding stability of the N- and C-domains is about 4-fold larger than for ovotransferrin. Chelated ferric ions bind strongly to each site in both proteins and produce changes in Tm by as much as 30 degrees C. When apparent binding constants are estimated from DSC results, these appear to be substantially larger than those estimated previously from equilibrium methods at low temperatures, where very long equilibrium times must be used because of slow ligand release. Although second DSC upscans on each protein show good "reversibility", downscans on serum transferrin revealed that liganded forms of the protein are in fact not in true equilibrium during upscanning, which causes Tm values during upscans to be higher than the true reversible Tm values. The likely reason for this kinetic control over unfolding is the slow release of bound ferric ions and those effects, for technical reasons, cannot be totally eliminated by lowering the scan rate.

Calorimetric studies of the binding of ferric ions to human serum transferrin

The binding of ferric ions, chelated with nitrilotriacetate, to human serum transferrin (hTF) has been studied using ultrasensitive titration calorimetry. Studies were done in both the presence and the absence of the synergistic bicarbonate anion. It was found that the C-site of hTF is capable of weakly binding bicarbonate (K of 250 M-1, delta H of -8 kcal) at the binding site even before ferric ion is added, although this does not happen to the same extent at the N-site. When preinsertion of the bicarbonate ion occurs, then ferric ion can subsequently bind very quickly to the C-site. Although the chelated ferric ion can bind weakly to the N-site in a fast reaction, the insertion of the bicarbonate ion occurs subsequently in a slow endothermic reaction. Binding of ferric ion to both sites is quickly reversible in the absence of bicarbonate but becomes kinetically controlled for long periods of time once bicarbonate has inserted into the metal-binding site due to the long time required for release of ferric ion. Estimates of the heats of binding to each site, apparent binding constants, and heat capacities of binding are made for different sets of solution conditions. Results from this study are compared to earlier results with ovotransferrin (Lin, L.-N., Mason, A. B., Woodworth, R. C., & Brandts, J. F. (1991) Biochemistry 30, 11660-11669), with major differences and some similarities noted.

Calorimetric studies of the N-terminal half-molecule of transferrin and mutant forms modified near the Fe(3+)-binding site

The effects of single amino acid substitution on the thermal stability of the N-terminal half-molecule of human transferrin and its iron-binding affinity have been studied by high-sensitivity scanning calorimetry. All site-directed mutations are located on the surface of the binding cleft, and they are D63-->S, D63-->C, G65-->R, H207-->E and K206-->Q. Differential scanning calorimetry results show that the mutations do not significantly alter the conformational stability of the apo-forms of the proteins. The changes in free energy of unfolding relative to the wild-type protein range from 0.83 to -2.4 kJ/mol. The D63-->S, G54-->R and H207-->E mutations slightly destabilize the apo-protein, while the D63-->C and K206-->Q mutations increase its stability by a small amount. However, there are large compensating enthalpy-entropy changes caused by all mutations. All mutants bind ferric ion, but with different affinities. Replacement of Asp-63 by either Ser or Cys decreases the apparent binding constant by 5-6 orders of magnitude. The G65-->R mutation also decreases the apparent binding constant by 5 orders of magnitude. The K206-->Q mutation increases the apparent binding constant by 20-fold, while the H207-->E mutation does not significantly change the apparent iron-binding affinity of the half-molecule.

Calorimetric studies of the binding of ferric ions to ovotransferrin and interactions between binding sites

Transferrins are two-domain proteins with a very strong site for iron binding located in each domain. Using ultrasensitive titration calorimetry, the binding of ferric ion (chelated with a 2-fold molar excess of nitrilotriacetate) to the two sites of ovotransferrin was studied in detail as well as the binding to the single site in the N- and C-terminal half-molecules. In the presence of excess bicarbonate ion, the binding occurs in two kinetic steps. The fast process of contact binding is instantaneous with respect to instrument response time, is strongly exothermic for the N site and less so for the C site, and corresponds to binding of the chelated ferric ion. The slower process of bicarbonate insertion with concomitant release of nitrilotriacetate occurs on a time scale of 2-20 min over the temperature range 7-37 degrees C and is endothermic for the N site and exothermic for the C site, with rates being significantly slower for insertion at the C site. The delta H of binding is strongly temperature-dependent for both sites, arising from a large negative delta Cp of binding which probably indicates removal of hydrophobic groups from contact with water. When bicarbonate ion is absent, only the fast process of contact binding is seen. Each site within a half-molecule is qualitatively similar to the same site in intact ovotransferrin, although quantitative differences were detected. It was shown that contact binding to ovotransferrin occurs reversibly with free exchange of Fe+3 between N and C sites, while the attachment to either site becomes essentially irreversible after bicarbonate insertion. The strong preference for the first ferric ion to bind to the N site is shown to be due to its larger contact binding constant and the faster rate of bicarbonate insertion, relative to the C site, and is not due to stronger thermodynamic binding after bicarbonate insertion. True equilibrium is achieved only over much longer periods of time. In another series of experiments, direct binding studies were carried out between the two half-molecules under different states of ligation with Fe+3 in the presence of bicarbonate. The results indicate that the two binding sites in ovotransferrin, separated by ca. 40 A, are not independent of one another but communicate as a result of ligand-dependent changes in the heats and free energies of domain-domain interactions.(ABSTRACT TRUNCATED AT 400 WORDS)

Study of strong to ultratight protein interactions using differential scanning calorimetry

Data from differential scanning calorimetry (DSC) may be used to estimate very large binding constants that cannot be conveniently measured by more conventional equilibrium techniques. Thermodynamic models have been formulated to describe interacting systems that involve either one thermal transition (protein-ligand) or two thermal transitions (protein-protein) and either 1:1 or higher binding stoichiometry. Methods are described for obtaining binding constants and heats of binding by two different methods: calculation or simulation fitting of data. Extensive DSC data on 2'CMP binding to RNase are presented and analyzed by the two methods. It is found that the methods agree when binding sites are completely saturated, but substantial errors arise in the calculation method when site saturation is incomplete and the transition of liganded molecules overlaps that of unliganded molecules. This arises primarily from an inability to determine TM (i.e., the temperature where concentrations of folded and unfolded protein are equal) under weak-binding conditions. Results from simulation show that the binding constants and heats of binding from the DSC method agree quantitatively with corresponding estimates obtained from equilibrium methods when extrapolated to the same temperature. It was also found from the DSC data that the binding constant decreases with increasing concentration of ligand, which might arise from nonideality effects associated with dimerization of 2'CMP. Simulations show that the DSC method is capable of estimating binding constants for ultratight interactions up to perhaps 10(40) M-1 or higher, while most equilibrium methods fail well below 10(10) M-1. DSC data from the literature on a number of interacting systems (trypsin-soybean trypsin inhibitor, trypsin-ovomucoid, trypsin-pancreatic trypsin inhibitor, chymotrypsin-subtilisin inhibitor, subtilisin BPN-subtilisin inhibitor, RNase S protein-RNase S peptide, avidin-biotin, ovotransferrin-Fe3+, superoxide dismutase-Zn2+, alkaline phosphatase-Zn2+, and assembly of regulatory and catalytic subunits of aspartate transcarbamoylase) were analyzed by simulation fitting or by calculation. Apparent single-site binding constants ranged from ca. 10(5) to 10(20) M-1, while the interaction constant for assembly of aspartate transcarbamoylase was estimated as 10(37) in molarity units. For most of these systems, the DSC interaction constants compared favorably with other literature estimates, for some it did not for reasons unknown, while for still others this represented the first estimate. Simulations show that for proteins having two binding sites for the same ligand within a single cooperative unit, ligand rearrangement will occur spontaneously during a DSC scan as the transition temperature of the unliganded protein is approached.(ABSTRACT TRUNCATED AT 400 WORDS)

Substitution of a proline for alanine 183 in the hinge region of phosphoglycerate kinase: effects on catalysis, activation by sulfate, and thermal stability

A "hinge-bending" domain movement has been postulated as an important part of the catalytic mechanism of phosphoglycerate kinase (PGK) (Banks et al., 1979). In order to test the role of the flexibility of a putative interdomain hinge in the substrate- and sulfate-induced conformational transitions, alanine-183 was replaced by proline using site-directed mutagenesis. The maximal velocity of the Ala 183----Pro mutant, measured at saturating concentrations of ATP and phosphoglycerate (5 mM and 10 mM, respectively) and in the absence of sulfate ions, is increased approximately 21% in comparison to the wild type PGK. The Km values for both substrates are essentially unchanged. The effect of sulfate on the specific activity of the Ala 183----Pro mutant and the wild type PGK was measured in the presence of 1 mM ATP and 2 mM 3-phosphoglycerate (3-PG). A maximum activation of 70% was observed at 20 mM sulfate for the mutant enzyme, as compared to 130% activation at 30 mM sulfate for the wild type PGK. These results demonstrate that the increased rigidity of the putative hinge, introduced by the Ala----Pro mutation, does not impair catalytic efficiency of phosphoglycerate kinase, while it appears to decrease the sulfate-dependent activation. The differential scanning calorimetry (DSC) studies demonstrate an increased susceptibility of the Ala 183----Pro mutant to thermal denaturation. In contrast to one asymmetric transition observed in the DSC scan for the wild type PGK, with Tm near 54 degrees C, two transitions are evident for the mutant enzyme with Tm values of about 45 and 54 degrees C.(ABSTRACT TRUNCATED AT 250 WORDS)

A simple model for proteins with interacting domains. Applications to scanning calorimetry data

A simple thermodynamic model is formulated for the purpose of interpreting scanning calorimetry data on proteins that have interacting domains. Interactions are quantified by inclusion of an interface free energy, delta GAB, in the thermodynamics of unfolding for multidomain proteins. The assumption is made that delta GAB goes to zero with the unfolding of either domain involved in pairwise interaction, so the interaction term appears to stabilize only the domain with the lower TM. Application of the model to calorimetric data leads to an estimate of -25,000 cal/mol for interactions between the regulatory and catalytic subunits of native aspartate transcarbamoylase and to a value of 0 for delta GAB between the transmembrane and cytoplasmic domains of band 3 of the human erythrocyte membrane. Estimates of changes in delta GAB are also obtained for mutant forms of yeast phosphoglycerate kinase that have been altered in the hinge region between amino-terminal and carboxy-terminal domains. The model is also applied to ligand binding to proteins having domains that communicate through pairwise interaction. It is shown that whenever the delta GAB term is ligand-dependent, then attachment of the ligand to the binding domain will be partially controlled by the other (regulatory) domain. This situation can sometimes be recognized and quantified when calorimetric scans are carried out at varying ligand concentrations. According to the model, the binding of MgATP to the carboxy-terminal domain of phosphoglycerate kinase is strongly stabilized (ca. 20% of the unitary free energy of binding) by participation of the amino-terminal domain, which acts to increase the binding constant 25-fold.(ABSTRACT TRUNCATED AT 250 WORDS)

Rapid measurement of binding constants and heats of binding using a new titration calorimeter

A new titration calorimeter is described and results are presented for the binding of cytidine 2'-monophosphate (2'CMP) to the active site of ribonuclease A. The instrument characteristics include very high sensitivity, rapid calorimetric response, and fast thermal equilibration. Convenient software is available for instrument operation, data collection, data reduction, and deconvolution to obtain least-squares estimates of binding parameters n, delta H degree, delta S degree, and the binding constant K. Sample through-put for the instrument is high, and under favorable conditions binding constants as large as 10(8) M-1 can be measured. The bovine ribonuclease A (RNase)/2'CMP system was studied over a 50-fold range of RNase concentration and at two different temperatures. The binding constants were in the 10(5) to 10(6) M-1 range, depending on conditions, and heats of binding ca. -15,000 cal/mol. Repeat determinations suggested errors of only a few percent in n, delta H degree, and K values over the most favorable concentration range.

Binding of plasma membrane glycoproteins to the cytoskeleton during patching and capping is consistent with an entropy-enhancement model

Concentrations of concanavalin A that induced patching and capping of cell surface receptors on Dictyostelium discoideum also induce binding of the receptors to the cortical cytoskeleton, which was isolated by density-gradient centrifugation. The receptors were solubilized by deoxycholate, purified by affinity chromatography, and used to determine whether the receptors bound directly to the cytoskeletal protein, actin. As the concentration of actin was increased, many of the receptors became bound to purified filamentous rabbit muscle actin, even in the absence of concanavalin A. As in the ligation-induced binding of receptors to the cortical cytoskeleton in cells, concanavalin A induced much stronger binding of the purified receptors to filamentous actin. The results were consistent with a previously stated hypothesis that induction of receptor binding to the cytoskeleton during their patching and capping is driven by clustering the receptors, which reduces their translational entropy and by doing so enhances their avidity for the cytoskeleton.

Separation of the nativelike intermediate from unfolded forms during refolding of ribonuclease A

In an effort to determine structural properties of the nativelike intermediate (i.e., IN) which forms during the refolding of RNase A, refolding samples were subjected to rapid HPLC gel filtration which allowed us to separate IN from unfolded forms of RNase. The comparison of these samples, enriched in IN and depleted of unfolded forms, with unseparated control samples at the same stage of refolding allowed certain conclusions to be drawn concerning the properties of IN. First, the results show that the transition from IN to native RNase occurs with only small changes in fluorescence. This means that the major fluorescence changes seen during normal refolding experiments must be associated with changes in proline isomerization of unfolded species and/or with the refolding step itself but not with the IN----N step. Second, the fluorescence assay for isomerization of proline-93 shows that IN exists with proline-93 in a state of isomerization identical with or very similar to native RNase; i.e., proline-93 is cis in IN and not trans as suggested by others. All results are semiquantitatively consistent with our earlier refolding model and not nearly so consistent with alternative models which assume that most or all of the slow-refolding forms of RNase have proline-93 in the incorrect trans state.

Catalysis of proline isomerization during protein-folding reactions

The enzyme peptidylprolyl cis-trans isomerase (PPI) is known to catalyze proline isomerization in short proline-containing peptides. If PPI can be shown to generally catalyze isomerization of proline residues in proteins, then it would be a valuable diagnostic reagent for recognition of isomerization, which has proven to be extremely difficult to characterize by other methods. In this study, the catalytic effect of PPI on the slow refolding reactions of seven different proteins has been studied, and in only two cases (RNase T1 and cytochrome c) could significant catalysis be seen. PPI also caused no enhancement in the rate for the 'subtle' conformational changes of native concanavalin A or native Fragment I of prothrombin, which have been suggested to be rate-limited by proline isomerization. There was a small effect of PPI observed for the generation of native RNAase A from the fully-reduced form when the glutathione concentration was low. The conclusion from these studies is that PPI can weakly catalyze some protein processes which are rate-limited by proline isomerization, but probably exhibits no measureable catalysis toward others. This somewhat limits the usefulness of PPI as a diagnostic reagent for proline isomerization.

Evidence for the existence of three or more slow phases in the refolding of ribonuclease A and some characteristics of the phases

The slow refolding kinetics of RNase A have been analyzed, by using a nonlinear least-squares program for deconvoluting the kinetic phases and applying statistical tests for quality of fit. It is found that a minimum of three slow phases are required to fit the kinetic data properly, and this is true whether the method of detection is absorbance of fluorescence. Since the number of phases and the relaxation times for each phase are independent of the method of detection, it is concluded that the same three rate-limiting processes are seen by absorbance and fluorescence. These phases correspond to the XY, CT, and ct phases described in our earlier studies. The fact that fluorescence-detected kinetics are somewhat slower than absorbance-detected kinetics is a trivial effect due not to differences in relaxation times but to the fact that the amplitude of the CT phase is enhanced in fluorescence measurements, at the expense of the faster XY phase, because of intrinsic fluorescence changes associated with the isomerization of proline-93. By use of a new double-jump technique [Schmid, F.X., Grafl, R., Wrba, A., & Beintema, J.J. (1986) Proc. Natl. Acad. Sci. U.S.A. 83, 872], it is shown that proline-93 isomerizes as the rate-limiting step in only one of the three phases, the CT phase, and that this phase involves only 25-30% of the RNase molecules. There is still no indication as to the molecular events that occur in the large, ammonium sulfate dependent XY phase, which is the pathway for formation of the nativelike intermediate.

Refolding of ribonuclease in the presence and absence of ammonium sulfate pulses. Comparison between experiments and simulations

Experiments have been carried out on ribonuclease A in which refolding in high concentrations of guanidine hydrochloride is either preceded or not preceded by a short ammonium sulfate pulse. Application of the pulse causes the rapid formation of the nativelike intermediate, and the effect of this pulse was determined by using three different methods for monitoring the subsequent refolding reaction: direct absorbance, direct fluorescence, and a double-jump fluorescence unfolding assay which is specific for the isomerization of proline-93. The effect of the pulse is quite different depending on the method of detection. With absorbance detection, the pulse causes a large reduction in the refolding amplitude with no change in the kinetics of the decay curve, while with the fluorescence unfolding assay, the pulse causes no change in the refolding amplitude but produces a large acceleration in the decay kinetics. The results with direct fluorescence are intermediate with some reduction seen in the refolding amplitude and some acceleration in the decay kinetics. The results of these experiments are simulated by using the simple model of Lin and Brandts (1984) [Lin, L.-N., & Brandts, J. F. (1984) Biochemistry 23, 5713] in which proline-93 must be in the correct cis configuration before folding to the native or nativelike state can occur. In all cases, the simulations accurately predict the experimental results for all three methods of detection, without any adjustment of parameter values from those published earlier.(ABSTRACT TRUNCATED AT 250 WORDS)

Isomer-specific proteolysis of model substrates: influence that the location of the proline residue exerts on cis/trans specificity

In an effort to further develop the technique of isomer-specific proteolysis, a number of proline-containing substrates were subjected to hydrolysis in the presence of chymotrypsin, trypsin, or prolidase. The objective was to determine whether direct hydrolysis of the cis form of the substrate could occur and, if so, the extent to which it is slower than the hydrolysis of the equivalent trans form. It is shown that for both peptide and amide substrates, which contain proline at the P2 position, the cis form can be hydrolyzed directly by either chymotrypsin or trypsin, in contrast to earlier suggestions in the literature. For similar amide substrates, it was found that chymotrypsin has a lower catalytic efficiency for the cis form, relative to the trans form, by a factor of 20 000 while, for trypsin and its substrate, the cis form was cleaved about 2000 times less efficiently. Results for a trypsin substrate with proline at the P2' position, rather than the P2 position, were quite different however, since there was no indication that the cis form could be directly cleaved even at the highest enzyme concentration. There was also no indication that prolidase could cleave the dipeptide Phe-Pro when the active bond itself is in the cis form. These collective results suggest that the ability of proteases to cleave a substrate with a cis peptide bond depends strongly on the location of the cis bond relative to the active bond that is being cleaved.

Involvement of prolines-114 and -117 in the slow refolding phase of ribonuclease A as determined by isomer-specific proteolysis

Using the method of isomer-specific proteolysis (ISP), the cis-trans nature of the peptide bonds involving prolines-114 and -117 in ribonuclease (RNase) has been investigated. These studies involve the pretreatment of RNase first with either a short pepsin pulse or a short mercaptoethanol pulse to irreversibly unfold the protein and then with a short chymotrypsin pulse to quickly cleave the Tyr115-Val116 bond so that the chain is suitably trimmed for the subsequent stereospecific cleavage either by aminopeptidase P, to investigate proline-117, or by a proline-specific endopeptidase, to investigate proline-114. The most reasonable interpretation of our results suggests that proline-117 is essentially 100% trans in both the native and unfolded states, so it apparently makes no direct contribution to the slow refolding kinetics of RNase. It is also determined that proline-114 is 100% cis in native RNase and ca. 95% cis in reversibly unfolded RNase so only 5% of the unfolded RNase can be rate limited by trans to cis isomerization of proline-114 during refolding. Careful spectroscopic studies of refolding show that the smallest and slowest of the refolding phases, the ct phase, has the proper amplitude (5%), relaxation time (400 s at 10 degrees C), and activation energy (17 kcal) for a phase that is rate limited by the trans to cis isomerization of proline-114. Measurements of the kinetics of binding of cytidine 2'-monophosphate during refolding further show that RNase does not become active until proline-114 has isomerized to the native cis configuration. It is concluded that none of the three prolines thus far examined (i.e., prolines-93, -114, and -117) by the ISP method is involved in the formation of a fully active, nativelike intermediate which has "incorrect" proline isomers. The specific structural process which is responsible for the largest of the three slow refolding phases, the XY phase, is still undetermined. Although ISP results on proline-42 are not yet available, it seems possible that this slow phase may be rate limited by a process other than proline isomerization. In unrelated studies, results from chymotrypsin hydrolyses of several short peptides containing the sequence -X-Y-Pro- show that cleavage of an active X-Y bond is very slow when it is immediately adjacent on the amino side of a proline peptide bond. Thus, chymotrypsin cleavage may not be generally useful as the analytical step in isomer-specific proteolysis.

Evidence showing that a proline-specific endopeptidase has an absolute requirement for a trans peptide bond immediately preceding the active bond

The proline-specific endopeptidase (EC 3.4.21.26) from Flavobacterium meningosepticum is specific for the cleavage of peptide bonds on the C-terminal side of prolyl residues. Such bonds will normally exist in the all-trans configuration. However, the preceding peptide bond in the sequence (i.e., on the N-terminal side of the prolyl residue) will exist as a mixture of cis and trans forms in solution. In this study, the activity of the proline-specific endopeptidase toward the substrates N-Cbz-Gly-Pro-MCA (where MCA = 4-methylcoumarinyl-7-amine) and N-Cbz-Gly-Pro-Leu-Gly has been examined. At a high ratio of enzyme activity/substrate concentration, the hydrolysis pattern for each substrate shows two well-separated kinetic phases. It is concluded that the fast kinetic phase, whose velocity depends on enzyme concentration, results from the direct hydrolysis of the active substrate bond (i.e., either the Pro-MCA or Pro-Leu bond, respectively) in molecules where the preceding Gly-Pro bond is trans. The slow phase, whose velocity is independent of enzyme concentration, is rate-limited by the cis-to-trans isomerization of those substrate molecules which initially have the preceding Gly-Pro bond in the cis configuration. That is, substrate molecules having the cis form of the Gly-Pro bond which precedes the active bond cannot be hydrolyzed directly but must first isomerize to the trans form before cleavage can occur. The amplitude, relaxation time, and activation energy for the slow phase are consistent with this interpretation. Thus, the proline-specific endopeptidase from Flavobacterium has an absolute requirement for a trans peptide bond at the position immediately preceding the active bond.

Determination of cis-trans proline isomerization by trypsin proteolysis. Application to a model pentapeptide and to oxidized ribonuclease A

It is shown, by examination of a model pentapeptide, that trypsin will only cleave substrate bonds in a polypeptide chain when the peptide bond following the active bond is in the trans isomeric state. The cis form must isomerize to trans before it can be cleaved. Taking advantage of this isomeric specificity, the sequence-Lys91-Tyr92-Pro93- is examined in oxidized RNase A. It is shown that the Tyr-Pro bond exists 33% in the cis form at equilibrium and that the cis-to-trans relaxation time for isomerization is 5.0 min at 10 degrees C. The fragment 92-98 has about the same cis content (35%) as does oxidized RNase A but has a much slower relaxation time (11 min). This suggests that overall chain dynamics may exert some effect on the kinetics of isomerization.

Isomerization of proline-93 during the unfolding and refolding of ribonuclease A

Using the method of isomer-specific proteolysis, the isomerization of proline-93 has been monitored directly during the time course of the unfolding and refolding reactions of RNase A. It has been found that proline-93 is 100% cis in the native protein and 70% cis in the reversibly unfolded protein. During the unfolding reaction, the change from 100% to 70% cis occurs as a first-order process with a relaxation time of 140 s in 8.5 M urea, 10 degrees C. For refolding, the change from 70% to 100% cis also occurs as a first-order process, with a relaxation time (10 degrees C) of 90 s in 0.3 M urea, 130 s in 1.0 M urea, and 310 s in 2.0 M urea. Parallel experiments which measured the recovery of enzyme activity during refolding were also conducted. These show that 30% of the activity recovers in a slow phase with a first-order relaxation time (10 degrees C) of 100 s in 0.3 M urea. Because of the excellent agreement of both the amplitude and relaxation time for trans-to-cis isomerization and for activity recovery, it is concluded that the slowest phase in the recovery of enzyme activity is rate limited by the isomerization of proline-93. These results demonstrate that proline-93 must be cis before refolding to the active form can take place, in contrast to previous suggestions, and argue against the existence of a nativelike intermediate form on the refolding pathway which contains proline-93 in the incorrect trans configuration.

Structural domains of vesicular stomatitis virus. A study by differential scanning calorimetry, thermal gel analysis, and thermal electron microscopy

Differential scanning calorimetry has been used in combination with thermal gel analysis and electron microscopy to identify and study the structural domains of the membrane-enclosed vesicular stomatitis virus and its isolated internal components. Three major endothermic transitions centered at approximately 52, 76, and 80 degrees C and at least two minor transitions are observed at pH 7.0 for the intact virion. Thermal gel analysis suggests the possibility that specific proteins of vesicular stomatitis virus are involved in two or more of the calorimetric transitions. The effect of heating to defined temperatures on the morphology of the virion was studied by negative stain electron microscopy. The results of these "thermal EM" studies show discrete irreversible morphological changes in the virion which seem to coincide with the three major calorimetric transitions.

Protein involvement in structural transition of erythrocyte ghosts. Use of thermal gel analysis to detect protein aggregation

In this study, it is shown that systematic temperature-induced protein aggregation occurs on the erythrocyte membrane by intermolecular disulfide bond formation. Specific protein bands disappear from acrylamide gel profiles over rather narrow temperature regions. The aggregation appears to be the result of irreversible structural transitions of the membrane, which can be seen in a sensitive scanning calorimeter. When this method of thermal gel analysis is used, the results suggest that spectrin is a participant in the A transition, that bands 2.1, 4.1, and 4.2 and the cytoplasma portion of 3 are involved in the B transition, and that the transmembrane portion of band 3 may undergo changes in the C transition, previously shown to occur in the anion transport domain of the membrane. The aggregation of specific proteins in the narrow temperature region of these transitions persists as the transitions are moved around on the temperature axis by varying solution conditions. The assignment of particular proteins to specific transitions is reinforced by selective extraction of membrane proteins. Large variations in both the calorimetry and the aggregation pattern occur as salt concentration is increased from 77 mosm to 310 mosm, which is manifested in the splitting of the B transition into two separate transitions, B1 and B2. It is speculated that this occurs as the result of a structural change which may involve components of the cytoskeletal network.

A relationship between anion transport and a structural transition of the human erythrocyte membrane

Scanning microcalorimetry was employed as an aid in examining some structural features of the anion transport system in red blood cell vesicles. Two structural transitions were previously shown to be sensitive to several covalent and non-covalent inhibitors of anion transport in red cells. In this study, these transitions were selectively removed, either thermally or enzymatically, and the subsequent effect on 35SO2- 4 efflux in red cell vesicles was determined. It is shown that removal of one of these transitions (B2) has a negligible inhibitory effect on anion transport. Cytoplasmic, intermolecular disulfide linkages between band 3 dimers are known to form during the B2 transition. The integrity of the 4,4-diisothiocyanostilbene-2,2-disulfonate-sensitive C transition, on the other hand, is shown to be a requirement for anion transport. The localized region of the membrane giving rise to this transition contains the transmembrane segment of band 3, as well as membrane phospholipids. The calorimetric results suggest a structure of band 3 which involves independent structural domains, and are consistent with the transmembrane segment playing a direct role in the transport process.

Evidence suggesting that some proteolytic enzymes may cleave only the trans form of the peptide bond

The rates of hydrolysis of glycy-L-proline and L-phenylalanyl-L-proline, catalyzed by prolidase, have been measured at several temperatures under conditions where a high ratio of prolidase activity to substrate concentration existed. Two well-separated kinetic phases, which can be adequately treated as two first-order reactions, were observed for the hydrolysis. The relative amplitudes of the two phases are nearly independent of temperature, but strongly dependent on the initial state of protonation of the dipeptides. It was found that the amplitude of the slow phase is strictly proportional to the known amount of cis isomer, while the amplitude of the fast phase correlates with the amount of the trans isomer. Furthermore, the relaxation time and activation energy of the slow phase of hydrolysis are in good agreement with the same parameters determined for cis-trans isomerization of the dipeptides, as measured by a pH-jump method for samples not being hydrolyzed. These results lead us to the conclusion that the slow phase seen for hydrolysis is rate limited by cis-trans isomerization of the X-pro peptide bond. Thus, this proline-specific protease appears to have an absolute requirement for the trans form of the peptide bond and appears not to cleave the cis form or to cleave it extremely slowly.

The effects of anion transport inhibitors on structural transitions in erythrocyte membranes

Red blood cell membranes have been labeled with several covalent and noncovalent inhibitors of anion transport and their heat capacity profiles determined as a function of temperature. Covalent inhibitors include the amino reactive agents 4,4'-diisothiocyanostilbene-2,2'-disulfonic acid, 4-acetamido-4'-isothiocyanostilbene-2,2'-disulfonic acid, pyridoxal phosphate and 1-fluoro-2,4-dinitro benzene. The non-covalent inhibitors include several well known local anesthetics. The study was underataken in order to identify regions of the membrane involved in anion transport. Covalent modification in all case resulted in a large upward shift of the C transition, which is beleived to involve a localized phospholipid region. Evidence is presented which indicates that Band III protein and this phospholipid region are in close physical proximity on the membrane. Addition of non-covalent inhibitors affects the membrane in either or both of two ways. In some cases, a lowering and broadening of the C transition occurs; in others the B1 and B2 transitions are altered. These latter transitions are beleived to involve both phospholipid and protein, including Band III. These results may indicate that the non-covalent inhibitors produce their inhibitory effect on anion transport at least in part by interacting with membrane phospholipid.

Calorimetric studies of the structural transitions of the human erythrocyte membrane. Studies of the B and C transitions

Differential scanning calorimetry has been used to study several structural transitions of the human erythrocyte membrane. Earlier studies have shown that one of these transitions (the A transition) is due to the thermal unfolding of spectrin on the membrane. In this paper, it is shown that two of the other transitions (B and C) exhibit a high sensitivity to a local anesthetic, benzyl alcohol. Increasing the ionic strength of the suspending medium results in a splitting of the B transition into two indepent transitions (B1 and B2). It is found that one of these (B2) is associated with titrating groups, since the midpoint for the transitions shifts by about 20 degrees C, with an apparent pK near 7.5 Extensive bilateral proteolysis by papain causes a drastic decrease in the size of all transitions except the C transition, which remains unaltered. On the other hand, treatment with phospholipase by A2 largely affects the C transition, causing its disappearance. Because of the lack of sensitivity to proteolysis and the high sensitivity to phospholipase, it appears that the C transition has a large extent of 'lipid involvement'. It might result from the melting of a small fraction of phospholipid which exists in a crystalline state under physiological conditions. Alternatively, the C transition could arise from changes in protein-lipid interactions or from lipid-dependent changes in protein-protein interactions, providing one assumes that only protease-resistant portions of membrane proteins are participating.

Unfolding and refolding occur much faster for a proline-free proteins than for most proline-containing proteins

The kinetics for unfolding and refolding of a parvalbumin (band 5) have been examined as a function of pH near the transition region, using stopped-flow techniques. This protein is rather unusual in that it has no proline residues, and therefore serves as a good example to test the hypothesis that the rate-limiting step seen in denaturation reactions is due to the cis-trans isomerization of proline peptide bonds in the denatured state. The kinetics for parvalbumin unfolding and refolding are complex, with the data being resolvable into two fast phases at 25 degrees. The slower of the two phases seen for the parvalbumin is about 100 to 500 times faster than the slow phase seen for proline-containing proteins under the same conditions! These results argue strongly in support of the proline isomerization hypothesis. It is also suggested that the slower phase seen for parvalbumin and the second-slowest phase seen for proline-containing proteins might be due to the cis-trans isomerization of peptide bonds of non-proline residues.

Consideration of the Possibility that the slow step in protein denaturation reactions is due to cis-trans isomerism of proline residues

A model is proposed to account for the observation that the denaturation of small proteins apparently occurs in two kinetic phases. It is suggested that only one of these phases--the fast one--is actually an unfolding process. The slow phase is assumed to arise from the cis-trans isomerism of proline residues in the denaturated protein. From model compound data, it is shown that the expected rate for isomerism is in satisfactory agreement with the rates actually observed for protein folding. It is also shown that a simple model of protein unfolding based on the isomerism concept is very successful in accounting for many known experimental characteristics of the kinetics and thermodynamic of protein denaturation. Thus, the model is able to predict that two kinetic phases will be seen in the transition region while none are seen in the base-line regions, that both the fast and slow refolding phases lead to the native protein as the product, that the fast phase becomes the only observable phase for jumps ending far in the denatured base-line region, that most or all small proteins show a limiting low-temperature activation energy of ca. 20,000 cal, and that the relaxtion time for the slow phase seen in cytochrome c denaturation is much shorter than for all other small proteins. By utilizing "double-jump" experiments, it is shown directly that the slow phase is not part of the unfolding process but that it corresponds to a transition among two or more denatured forms which have identical spectroscopic (286.5 nm) properties. Thus, the slow relaxation is "invisible" except in the transition region where it couples to the fast unfolding equilibrium. Finally, since the present model assumes that only one of the major kinetic phases seen in denaturation reactions is concerned with the denaturation process per se, it is in agreement with numerous thermodynamic studies which show consistency with the two-state model for unfolding.