The Solution Conformation of the Ferrichromes

SUMMARY The kinetics of hydrogen-deuterium exchange for the four individual protected amides of alumichrome, the A13+ analogue of ferrichrome, have been studied by proton magnetic resonance. range 7 the exchange rates are relatively invariant. A tighter binding as is raised results in a conformational stability gain, which compensates for base catalysis of amide within The analysis of the exchange kinetic data versus temperature within the framework of the “absolute reaction rate theory” yields the enthalpy entropy to of activation pD from -3 -7 amides proceed energy barrier significant exposure


SUMMARY
The kinetics of hydrogen-deuterium exchange for the four individual protected amides of alumichrome, the A13+ analogue of ferrichrome, have been studied by proton magnetic resonance.
In the range 3 < pD < 7 the exchange rates are relatively invariant.
A tighter binding to the metal as the pD is raised results in a conformational stability gain, which compensates for the known base catalysis of the intrinsic amide hydrogen exchange within this pD range.
The analysis of the exchange kinetic data versus temperature within the framework of the "absolute reaction rate theory" yields the enthalpy (AH?) and entropy (AS?) contributions to the free energy of activation (AFT).
Depending on the particular amide, AH? and ASt appear to vary over wider ranges than does AF t. On the average it is found that raising the pD from -3 to -7 increases AHt while decreasing AS?. It is proposed that while conformational fluctuations are of importance at low pD, at neutrality the exchange of certain amides might proceed through a higher energy barrier without significant exposure to the solvent.
Proton magnetic rcsonauce has recently proven to be an excellrrrt tool for conformational analysis of small cyclic peptides, depsil~eptides, and rnacrotetrolides (l-4). The increasing availability of stronger static magnetic fields has enabled resolution of many of the proton resonances of interest in these low molecular weight compounds.
In particular, the identification of single amide NIZ resonances makes this spectroscopy especially useful to monitor the hydrogendeuterium exchange at specific sites within the polyp&de.
Ferrichrome is a cyclohexapeptide of composition ( ~<:ly"-(~ly~-(~ly~-~rn~-0rn2-0m'3 ) --3 Fef3, _-~~ -._ * This work was supported by (irant AI04156 from the United States l'uhlic Health Service, (Grant GB5276X from the National Science Formdstion, and the United States Atomic I~Zncrgy Commission. This paper is part of the Ph.D. thesis of M.I,l., University of California, Berkeley, 1971. The preceding paper in this series is Rcfercncc 11. where 6) and the residues are labeled as previously described (7). The metal is coordinated by the three hydroxamic acid ligands provided by the acylated d-N-hydroxy ornithyl side chains. Ferrichrome acts as a growth factor for a number of microbes (6) and is presumably an iron carrier for Ustilago sphaerogena (8).
The conformation of ferrichrome and of related seryl-con taining peptides has been well characterized both by x-ray (9, 10) and P;\IRr (7,11) studies. The substitution of A13f or Ga3f for Fe3+ was necessary in the PMR work to eliminate line broadening by the paramagnetic ion. We have already discussed the conformational similarity between these complexes (7,II), while Emery has fouud that both the A13+ ("alumichrome") and Ga3f ("gallicllrorlle") nrrnlogucs of ferrichrome arc biologically active (8).
The cotlforlllatiotlnl model which has been proposed is depicted irr Fig. 1. Ferricllrome thus possesses a compact, globular structure.
The 3 consecutive substituted ornithyl residues have their side chains folded in a manner which optimizes octahedral coordination of the metal ion. The peptide backbone itself resembles an anti-parallel P-pleated sheet structure with the Orn3 and Gly3 residues paired by two carbonylarnide transannular hydrogen bonds, as in the Schwyzer model for cyclohexapeptides (12). As can be deduced from the x-ray and PMR data, these are weak hydrogen bonds with the Orrr3-NH.
. .O=C-G1.y3 bond (2.99 A according to the x-ray) being more stable than the conjugated Orrr3-C=O. . 'HN-Gly3 bond. The amide hy-drogens of glgcgls 1 and 2 are exposed and free to interact (H-bond) with the solvent, while those belonging to the remaining ornithyl residues are either involved in a short (2.80 A according to t.he x-ray), stable hydrogen bond directed to its own side chain N-O hydroxamate oxygen atom (Orn2), or buried in a pouch limited by the peptide backbone ring itself and the side chains of the ornithyl residues embracing the metal (Ornr). The model depicted in Fig. 1 thus shows amide hydrogen atoms with different degrees of intramolecular hydrogen bonding (On? > Orn3 > Gly3) and steric shielding which ranges from complete esposure (glgcyls I and 2) to significant occlusion iu a hydrophobic environment (Orni). Since ferrichrome is extremely soluble in water, no instrunrrntal sensitivity problem handicaps its PLMR study in aqueous solution.
Furthermore, the high stability of the metal complrs results in a very rigid structure which practically "freezes" the environment around each single proton, yielding excellent spectroscopic resolution of the six amide SH's whose resonances are spread over a range of -4 ppm. Fortunately, the water (or HDO) resonance is sufficiently shifted to higher fields that it does not interfere with the detection of any of the amide absorption peaks. Fig. 2 shows the amide NII resonance region for alumichrome in aqueous solution, and the already demonstrated assignment of the absorntions to the corresponding residues (11).
The couformational state of the pept,ide can be drastically affected by the binding of the metal (7,13). Emery (13) found that the bulk hydrogeii-tritium exchange of the chelates (ferrichrome and ferrichrome -2) was much slower than that of the deferri-peptides.
Even though his data supported the x-ray model for crystalline ferrichrome A, direct assignment of the exchanging hydrogens to the residues in the peptide sequence is not possible from the data obtained by the Englander two-column gel filtration technique.
FIN. 1. The conformtttionnl model for fenichrome (7,10). Chemical bonds along the peptide backbone ring are represented with henzrier lines, and the corresponding residnes, whether glycyl or ornithyl, are labeled by numbers i?z circles at theol-carbon atom. The metal is represented by :1/1 'and, for simplicity, only those hydrogen atoms belonging to the four slowly exchanging amides, namely, glycyl 3 and the three ornithyls, are shown. While the existence of the hydrogen bond bridging Gly3-NH.
.O==C-Orn3 is not clear from previous work (7, ll)? in this communication it is suggested that in alumichrome the stability of this amide towards hydrogen exchange is mainly due to steric rat,her than to hydrogen bonding protection.
A distinction is made between very weak (' .) and stronger (----) hydrogen bonds. In this paper our interest is limited to the use of the slow amide hydrogen exchange kinetics in alumichrome as conformational probes and models for retarded amide hydrogen exchange in proteins.
Hence no attention will be paid to the two relatively fast amides, namely, those of glycyls 1 and 2, since the knowledge that they are fast is sufficient information for these purposes. The temperature dependence of the exchange rate constants will be analyzed iu the coutext of Eyriug's "Absolute Reaction Rate Theory" so that ASt (the activation entropy) arid A/It (the activation enthalpy) will be estimated at different pl) levels. From the data for each alumichrome amide the relative steric and hydrogeli-bonding contributions to their hydrogen exchange retardation, and hence to the conformation of the molecule, will be discussed. Finally, the mechanistic implications for the hydrogen cschange of the amides will be analyzed within the contexts of Klotz's "direct exchange" (14) and the Linderstr(dm-Lang (15) hypotheses at various pD levels.

METHODS
The alumichrome sample was prepared as previously described (7). The pD levels were calculated by adding 0.4 to the pH reading from a glass electrode pH meter (16). The solutions at pD 7.22 aud 5.14 were buffered in 0.005 M sodium phosphate and 0.005 M sodium acetate, respectively; at lower pl) levels no buffer was judged necessary because the peptide itself, through its chelated hydroxamate side chains, provided enough buffering capacity at the concentrations used. All pD levels were adjusted with concentrated NaOD and DCI.
In all experiments the spectrometer probe was pre-equilibrated to the desired temperature for at least 1 hour. Then the field homogeneity was adjusted with a sample of composition identical with the one to bc studied.
The sharp lines of the free methyls in the hydroxamate acyl groups provided an excellent internal standard for rapid tuning.
The hgdrogeu exchange experiments were initiated by dissolving 70 mg of peptide, preweighed in the NXlR tube, in 0.8 ml of buffered l&O, or DC1 solution in D,O, to give an approximately 0.125 M solution at the desired pD. When the conditions resulted in fast exchange, the samy)le-contaiiling NMR tube was chilled during dissolution of the peptide and until its insertion in the spectrometer.
The field homogeneity was then quickly readjusted with the fine controls; during the few minutes required by this process the sample would equilibrate to the probe tenperature.
The exchange was followed in time by successive scannings of the amide proton resonance, the interval between scans depending on the exchange rate of the particular amide. Typically, an amide would be monitored for at least one exchange half- at .X..i". This spectrum was recorded at pH 5.14, hut, the chemical shifts are invariant, within the pH (pD) range studied in this work. I<ach resonance is assigned IO its corresponding residue according to the convention followed in Fig. 1 (11). The chemical shift scale is referred to internal ter-butyl alcohol and is expressed in parts per million (ppm) at t,he top and in Hertz (Hz) at the bottom.
time. The spectrometer gain was constant within the accuracy of our measurements.
The rate of exchange can be calculated either from the time dependence of the integrated amide proton resonance area or its peak amplitude.
The second method was chosen because of drifts in the spectrometer integrator. When monitoring the exchange kinetics of a particular set of amides under constant pD and temperature, the radiofrequency power, receiver gain, amplification, and noise filtering were kept constant so that the uncertainty introduced by electronic noise was practically the same for all points in a kinetic curve.

RESULTS
The semilogarithmic plots of amide NH peak amplitude versus time were linearly least squares fitted by giving the same weight to each point.
The slopes of the line yield k, the first order exchange rate constant.
The rate constants were then plotted semilogarithmically versus inverse temperature.
The Eyring plots were least squares fitted, the weight for each point being given by the inverse logarithmic standard error of the corresponding k. Hydrogen-deuterium exchange kinetics for the four slowly exchanging amides in alumichrome at pD 5.14 (O.OOS M ds-acetate) in D20 and at 24.2". The dots plot the logarithm of the experimentally measured proton resonance peak amplitudes .
. versus time m mmutee. The li?Les are least squares fits of the data points assuming a first order decay and, together with the data points, are normalized to 10 for zero time. The experimental points for each amide are marked according to the key included in the figure, which also shows, in parentheses, the times in minutes for half decay of the corresponding amide resonances. The indices identify the resonance, Q according to the convention followed in the text. The exchange rate constants determined from linear fits of this kind, together with their standard deviations, are given in Table  I for the slow amides of al\lmichrome under different pD and t,emperature conditions. 917 chrome at pD 5.14 and 24.2". This and similar plots yield the first order rate constants for hydrogen eschange for each of the slow amides studied under their particular pD and temperature conditions.
The values so determined, with their standard deviations, are given in Table I. Fig. 4 depicts Eyring plots for the rate constants given in Table I. The AH? (slope) and As? (intercept) values, together with their standard deviations, are given in Table II. These figures summarize all of the kinetic data provided by the first order exchange curves of the type exemplified in Fig. 3 which are not shown individually.
Except for the case of alumichrome at pD 3.23, where the relatively higher ionic strength of the solution resulted in a poorer balancing of the spectrometer probe, the standard errors of the slopes of the kinetic curves (k) are relatively small. However, the dispersions of the Eyring plots are larger.
Since the experimentally accessible temperatures are within the range 3.66 x 1OP > T-1 > 2.68 x 10ea 'K-r, it is to be expected that the standard deviations in the slope (AHI) would be smaller than in the intercept (As?), calculated at the rather removed point T-'l = 0 OK-'. Furthermore, the relative errors for A8t are larger than for AHt because the absolute values of As? are closer to zero.
Obviously, in the discussion that follows any conclusion based on the comparative values of the kinetic parameters will be valid within the accuracy of their determinations.
In most cases the experimental uncertainties are small enough for our purposes; cases where this is not so will be noted explicitly.

General
Kinetic Analysis-Hvidt and Nielsen have extended an early proposal of Linderstrem-Lang (17) and of Berger and Linderstrgm-Lang (I 8) to rationalize protein hydrogen exchange in general (15). The molecule is assumed to fluctuate between more or less folded conformations (N states) in which the labile hydrogens are unexchanged and buried, and more or less relaxed conformations (I states) in which the labile hydrogen is unexchanged but exposed to the bulk solvent.
It is assumed that the concentrations of the various protein conformations are stationary in the exchanging solution and that chemical equilibrium holds. h N h 2 I + exchange kz (1) In the I conformations hydrogen exchange can readily take place with a first order rate constant, ka, whose magnitude approximates the first order exchange rate constants for low molecular weight compounds under similar conditions.
Furthermore, it is assumed that the conformational drifts between the N and I states are characterized by first order rate constants, kl and kp.
Along these lines the unfolding process in the alumichromes can be thought of as a substitution of the trivalent metal, ionically coordinated to the three hydroxamate ligands, by one, two, or Uiree protons.
pK values for these equilibria have not been measured.
As judged from the acetyl hydroxamate stability constants, alumichrome should be a weaker complex than fer-richrome (19). For ferrichrome, Anderegg et al. (20) have deter- The first order rat,e constants (k) for the hydrogen-deuterium expected in ferrichrome, and even more in alumichrome. Our exchange of the slowly exchanging amides of alumichrome.
The observation of a rapid exchange of AP+ for Fe+ at pH 3 demonvalues are calculated for the individual amides from the slopes of the linear least squares fits of the experimental data points zs strates its existence (7). Furthermore, the PMR spectrum of in Fig. 3. The k values together with their standard deviations alumichrome at this low pD shows the presence of extra peaks in are t,a.bulated versus temperature (7') at each pD. The pD levels the amide region, which exchange with deuterium at relatively studied were 3.23 (a) 5.04 (b), 5.14 (c), and 7.22 (d). The units faster rates and which can be assigned to the metal-free peptide. in this table are minutes-l for k and "C for 2'. Each particular k The bulk hydrogen exchange kinetics of ferrichrome and reflects a linear fit of a number of points which typically varied from 5 to 10 depending on the particular amide, temperature, and alumichrome show that on lowering the pH (pD) from about ~1). Because the exchange of Gly3 at pI1 7.22 is too fast, it wis 7 to about 3 the number of slowly exchanging hydrogens innot monitored and the corresponding data are absent in d. creases for both compounds (Table III). The data show that at neutrality alumichrome exchanges slower than ferrichrome T Gly3 Or" Or"* OtX3 and that the kinetic difference tends to disappear (a reversal is hinted) as the pH is lowered. This behavior suggests that the (a) p0 3.23 rate constants for the ligand-metaP+ complex dissociation might be smaller for A13+ than for Fe3+ near pH (pD) 7 and that the stability of the A13f complex is more dependent upon H+ (D+) concentration than that of the ferric analogue. This is supported by our observation that at neutrality no exchange of A13+ for Fe3f is detected, while it is known that 5QFe3f readily exchanges with ferrichrome (about 8 min for half completion) at pH 6.3 and 37" (21). The differences in the hydrogen exchange kinetics should not be attributable to isotope effects since Emery verified that rates for hydrogen-tritium exchange do not differ significantly from those for deuterium-tritium exchange.
Furthermore, throughout our experiments we observed no changes in the positions of the NH resonances as deuteration was proceeding, indicating that the conformation of the molecule is insensitive to the hydrogen isotope at the amide. Since hydrogen exchange in the presence of excess Fe3+ did not affect the kinetics of the process, Emery suggested that local environmental factors rather than a conformational "breathing" process would dominate the observed exchange rates. However, a metal-free intermediate could be of negligible importance as a contribution to the conformational fluctuations responsible for the observed exchange.
Thus, random modulation of the distance between the metallic center and each side chain bidentate might result in short-lived di-and eventually monohydroxamate complexes which could account for the "unfolded," loosely structured intermediates responsible for most of the measured exchange.
If such were the case, the kinetics of the folding process would be practically independent of the excess metal ion concentration.
Furthermore, the effect of excess metal would be obscured by the formation of 1: 1 complexes, thus opening the molecule (22). _, equilibrium between the "native" (the hexadentate chelate) peptide and any and all of its "unfolded" (partially or totally nonchelated) forms can be represented as N and I states, respectively.
We propose that an increased proton concentration shifts the equilibrium to the right both by increasing kl and by decreasing kz. It should be obvious, however, that in general the k,:lcQ ratios as well as the conformational change contributions to the free energy of activation of the exchange reaction will be different for each of the amides in the molecule.
The rate of exchange for a free amide hydrogen is known to he at a minimum at about pH 3 (15, 23). However, upon going from pD 5.14 to 3.23 the rates of hydrogen exchange for all of the alumichrome ornithyl amides increase from exchange half-times of 416, 220, and 219 min to 12.3, 8.0, and 9.3 min for Ornl, Orn2, and Orn3, respectively (Table II).
In contrast, the exchange Alumichrome Inverse Temperature PK)-'x IO' half-time for Gly3 does not change significantly, from 6.2 to 7.5 min. A simple explanation can be given for these effects; on lowering the pD the intrinsic rate of amide hydrogen exchange drops, but the stability of the chelate is so reduced as to more than compensate for this effect and the over-all kinetics is accelerated.
For Gly3 an almost exact compensation results so that no major kinetic change is observed.
At about pD 3, stability differences due e.g. to intramolecular hydrogen-bonding are of little influence since the exchange proceeds mainly through the relatively abundant unfolded conformation.
At pD 5.14, the over-all ground state conformation is enforced and small stabilizing differences such as intramolecular hydrogen-bonding would show more pronounced relative effects. On raising the pD from 5.14 to 7.22 the stability of the chelate is increased further and these effects become even more characteristic for each amide. At pD 7.22 intrinsic rates of amide hydrogen exchange due to base catalysis are relatively large. il change in pD from 5.14 to 7.22 results in a 20-and 75-fold increase in hydrogen-deuterium exchange rates for Or2 and Orn3, respectively, while Gly3, exhibiting a half-life of 6.2 min at t.he lower pD, exchanges immeasurably rapidly at neutrality. This pD change, however, does not appear to affect significantly the exchange rate of Ornl, whose exchange half-time changes from 416 min (pD 5.14) to 362 min (pD 7.22). Thus, while at pD 3.23 the four slowly exchanging amides exhibit rate constants of the same order of magnitude, at pH 5.14 the observed k for Gly3 is about lo2 larger than for the ornithyl amides and, at neutrality, the orders of the observed rate constants differ one from the other by at least one order of magnitude (2.4 x 10-l, 7.0 X lOP, and 1.9 X lo+ min-' for Orn3, Orn2, and Ornl, respectively). half-times, Il,s(25). calculated for all of the measllred amides at 2.i" from the absolute reaction rate theory expression I,,:, = (0.6M/KT) esp( -APT/R?') are also inclltded. The reader is reminded, however, that the measured rate constants are given in Table I, together with their experimental uncertainties. As stated in Table I  amides as the pD is raised from 3.23 to 5.14 to 7.22 probably also reflects a change in the exchange mechanism with PD. ht low pD, the amide hydrogens show first order rate constants of hydrogen exchange which are comparable to those of the COW formationally loose poly(uL-alanine). The expression kDzO = 50(100.3-pD + 10 pD-6.3)100.OS' t-20lmin-l (where t = temperature in "C), proposed by Hvidt and Nielsen (15), yields Ic = 0.18 min-' for this polymer. As the pD is raised, the rate constants for the alumichrome amides depart quite dramatically from the values predicted for poly(m,-alanine), 6.1 and 741 minP1 at pD 5.14 and 7.22, respectively, and remain far below these values.

Components of the Activation
Barrier-Values for AH?, As?, and Apt (25') were calculated for N-methylacetamide on the basis of Eyring plots from the exchange data of Klotz and Frank (24)  These values will be modified when considering the exchange behavior of amides within a peptide or protein in its native state; H bond atld steric shielding effects will contribute both to AH? and As? since the transitioll state I might require H bond breakage and even partial unfolding of the structure.
Furthermore, in the discussion that follows, both the temperature dependences of the dissociation of D,O and of the hydroxamate binding constant have, perforce, been ignored.
Correcting for the enthalpy of ionization of water would shift by the same amount all of the tabulated AH? values for the amides while leaving unaffected their differential value. Furthermore, as discussed above, the dissociation of the hydroxamate complex is the main contribution to the peptide unfolding process and hence its temperature dependence does not need to be accounted for separately.
Thus, although the data represent the gross exchange process, they still enable useful conclusions to be made regarding the over-all trends of the kinetics and, in particular, the extent to which conformational factors participate in the eschangc mechanism.
On goitlg from pD 5.14 to 3.23, AIZ t and A&'1 decrease about 6 Cal and 13 e.u., rcsl)ectivrly, for the ornithines, while for Gly3 A1f t increases 4 Cal and ASt 13 e.u. This suggests a different nature for the exchange mechanism of these two types of amides. Raising the pD from 5.14 to 7.22 appears not to affect Ast appreciably either for Orn2 or for Orn3, the increased rate of eschangr for these resulting from a 1 to 2 Cal decrease in Alit. However, the same increment in basicity increases AZIt by about 4 Cal and A&'7 by about 13.5 e.u. for Orn'.
The AHi values show a parallel increase so that at 25" the three ornithyl amides have about the same free energies of activation (AFT = 23.3 to 23.7 Cal) while the G1y3 is about 2 Cal (As't = 21.2 Cal) below these values.
The kinetic data prove useful in pointing out conformational differences between Gly3 and Ori?. Since A11t for Gly3 is about 5.5 Cal below the value for Orn3 (= 20.9 Cal) and hydrogen bond energies are of the order of 3 to 8 Cal per mole (25), the enthalpy difference between these two amides may well account for the contributions that hydrogen bond stabilization would make. Leichtiilg and Klotz (26) have discussed the importance of inductive effects on the amide eschange stability.
At 25" the pK, values of ornithine are 1.94 and 8.65 and of glycine 2.34 and 9.60 (27). This indicates that any differential inductive effects would tend to increase the positive charge on the ornithyl relative to the glycyl amide NH and C=O and hence would result in a relative increase in the electrostatic contribution to the strength of the Orn3 cross-amide hydrogen bond while illcreasing its intrinsic rate of base-cakalyzed hydrogen exchanee. The values for the ferric complexes were measured at 30" by Emery (13), while those for alumichrome are calculated, for the same temperature, from the free energy of activation at 30" and on the basis of the kinetic data reported in Table II This reinforces the suspicion that while Orn3 is transannularly hydrogen-bonded, Gly3 might not be in this state. Why then the relative protection of the Gly3 NH? The answer is given from the AS? differences.
The Gly3 NH is about 11.3 e.u. (= 3.4 Cal at 25") more stable than the Or+ NH, suggesting that indeed steric effects are relatively more important for the glycyl than for its paired ornithyl.
At pD 3.23 and 25" the APt values are similar for all of the slow amides, ornithines, and glycine alike. The large negative ASt values found for the ornithyl residues probably indicate that their exchange is proceeding through a different conformation, favored by relatively lower AH+ values (16.3 Cal y AHt 7 14.5 Cal). At this lower pD the ornithyl amides provide a clearer picture of the molecular fluctuations than do the figures for the Gly3 exchange, not only because of a relatively better fit of the experimental points, but also because they are more directly sensitive to the pH-dependent metal chelation.
Mechanistic Implications-By calculating the exchange rate constant ka for a random polypeptide, namely, poly(nL-alanine), on the basis of Equation 2 and dividing the observed k by this number, a gross estimate can be obtained of the relative impedance to exchange for the different amides under the various pD conditions.
It is within the framework of the Linderstdm-Lang, Hvidt, and Nielsen treatment that the meaning of k/k2 becomes mechanistically meaningful (15). According to the theory, if the exchangeable amide hydrogens do not remain too long in the exposed state (ki < ICJ, two extreme limits are of interest, the unimolecular EX1 (ki < kz << k3) and the bimolecular EX2 (ki < k2 >> k~) mechanisms.
In the EX1 mechanism the rate-limiting step is the unfolding of the native structure and the observed k equals kl, while in the EX2 mechanism the leak-out ka process is weighted by the extent of unfolding of the peptide and the observed k equals (kJlCJk3. For alumichrome at pD 3.23 and 25", 0.3 "< lG/k~ 2 0.5 for the slow amides, which is too large for an EX, mechanism. By contrast, the EXr mechanism would imply k = ki << k~ which is contradictory since k3 2 0.2 mm-1 (poly(nL-alanine)) and the observed k 2 0.4 min-i.
It is then likely that at this pD the actual mechanism be intermediate between the extremes EXi and EX2, namely, ki < kz r k~.
From the experimental pD dependence of the hydrogen exchange rate constant above pH 4 for poly(nL-alanine) (Equation 2) it can be derived that: 7.22 178.18 5.55 1.58 By assuming kr and kz to be relatively pH independent, it can be shown that an EX2 mechanism should yield: =I (Hvidt and Nielsen (15)) This analysis was applied to alumichrome at pD 5.14 by studying the exchange kinetics at pD 5.04. The data at these two pD levels as shown in Table I and Fig. 4, may be compared. A A(pD) = -0.1 results in about a doubling of the hydrogen exchange rate of the ornithyl amides, while the Gly3 rate is halved, 2.e. : = 1.6 for Gly3 -1.1 for Orn Thus, while Gly3 seems to satisfy the above criterion and might exchange through an EX2 mechanism, the data are less clear for the ornithyl amide mechanism. This minor pD shift appears to affect the Gly3 and Orn values of AH? and ASt in different ways. In the case of Gly3 the decrease in the exchange rate is due to a 6.5 e.u. drop in AS? that overcomes the opposing I .5 Cal drop in AH t. The ornithyl amides, however, decrease their exchange rate due to slightly more favorable values of both AH t and AS'f. These changes for the ornithyl amides, although small, are in the direction that would be expected from an increase in the acid-catalyzed metal exchange as discussed when considering the exchange kinetic data for pD 3.23. By contrast, changes in the Gly3 amide kinetic parameters, in accordance with the k~ dependence on pD, are suggestive of a tighter coupling of this amide to a relatively more pD invariant peptide backbone conformation.
The k:ks ratios, however, although larger for Gly3 than for the ornithyl amides, are all of the right order of magnitude for an EX, exchange mechanism at this pD. In summary, it is proposed that the reason why the above differential criterion does not apply for the ornithyl amides is that their ki and kz values are so dependent on the acidity of the medium that they fall outside its range of applicability.
On raising the pD from 5.14 to 7.22, metal chelation-dependent kl decreases and kz increases. The ratios k:k3 now yield the values 0.36 x 10e5, 9.54 x 10d5, and 32.9 x lo-5 for Ornr, Orn*, and Orn3, respectively.
The range covered is about two orders of magnitude, suggesting now more differentiated exchange pathways within the applicability of an EX2 mechanism, if it applies to them at all. Indeed, at pD 7.22, Jc3 = 7.4 x lo2 and it is likely that an EX1 mechanism (kg < kz << ki) is responsible for the exchange.
In alumirhrotne the redured rate of amide lt~drogett eschattge, even at p1) 3.14 and 0", could still be accounted for in terms of some conformat.ional rigidity conferred to the pcptidc by the metal. Hence, we considered it convenient to observe the eschange kinetics of the metal-free peptide.
The exchange of the three resolved I'MR bands in tiefrrliferriclirome (7) was followed at pl> 3.0 and 3.5". The measured half-lives are: The \-alucs for l)ol~(t)L-alttttittc) wcrc ralculnted as brforr.
.11tltough tltr data suggest a more l)t'otrrtrd locatiott for the amides with rcsottattccs at higher firltls, thr I)rotrrtiott is ttot great. It is then quitr likely that in aqurous solution tltc l)lrferred cotiformation dors ttot have ititcrnnl attiidrs iti suffirictit relatiTr cottcetitratioti to br PMR dctrctablr. Sucll at1 idea has been advattccd in regard to the lack of lt~~3rogctt rscltattge kinetic differentiation between I'MR obsrrvable atliidc hydrogens itI synthetic cyclic, hesapeptidos (32, 33). It slmuld be noted, however, that, at higher pII levels t,he rschattge in gramiciditt S-21 is much faster than in aluniiclnonic.
The structural stabilization rottfrrred by the metal rhrlatr moiet\-clearly diffrretttiates the two compounds.
Conclusions--A merit of alun~icltrome as a model for amide h?-drogen exchange is that it shows the relative requirement of both the Klotz and the Liiidet,strZtn-Lattg approaches for esplainittg the observed kinetics.
Both local environmental effects and c!oiifo~tnational fluctuations are present, the relative con The diagram is cottstrttcted on the basis of the Linderstrdm-Lang scheme: Vertical scales are arbitrary. The intention here is to show the shifts of Ab',t relative to SF'31 as the pl) is varied. It shortld be not iced that dtto to the opposite pl) dcpendcnces which AFlt and AFat exhibit, the net observed Pt is not much affected b>-pL).
In the text arguments are given that support the predominance of this exchange mechanism at pl) 3 and even at pl) 6, butt, for certain amides, not at ttelttrality. tributiotts of rash drprttdittg ott tlto particular atnidr, its loratiott witltitt thr molerule, and thr i,I-I.
It 11:~s boett suggrstrd (31) tlutt thr relative ittsettsitivi@ of t,he frrrirlirotnr lt~~irogen errhangr towards ~1-1 changes might reflect an ittsrttsitivit~ of k1 in rspression (1) which would be rate-litniting in an ISX1 mechattiem. We believe, howevrr, that, if is the relatively strong depcndrttre of kt on pH that results in the relative $1 independence of the over-all exchange rates.
Our view of the l)I> effect on the mechanism for the alumichrome amide hydlogen-deute~ittIn cschange is summarized in Fig. 5, where AZ<','/ is the free energy barrier to reach the intermediate, unfolded state I and AF,i is the free energy barrier for the direct exchange of the unhindered amide. While AFtt has a relatively high entropic contribution, AFat is mainI?-enthalpic, and while OI)-cat&zes the srcond step, I>+ catalyzes the first. III Fig. 5 t,hc rrlativcl tret& of the AFIt and the Ar;,t barriers with 1111 arc depicted.
.1t pl) -3 the Al;'lt barrier is so low that the> over-all exchange rate is determined by the Aft,7 step, the, irrt<rinsic rate of amide hydrogell-deuterium eschangc beiug niillil0aL1. At pD -7, the structure becomes reinforced because of the high kinetic stability of the &elate moiety.
Eve11 though Afr',j iti now relatively low due to the base catalysis? the height of the Aplt barrier results in measurably low exchange rates. .it 111) -5 the situation is intermediat'e between the two previous casts. Even though base catalysis should result in a relatively fast intrinsic exchange, the structure is more stabilized by the Ala+ t<rihydroxarnate comples than at lower pD, and some cxchange retardation becomes apparent.
AUthough at this pD the EX, mechanism may be predominant, a kl-controlled exchange ma\-commence to contribute to the measured rates.
Olle would expect that arly conformational fluctuations present at pl> 7 will also be present at lower pD, in addition to those controlled by the metal. The A&'t values for Ornl, Orn2 and Ort13 at pD 7.22 show that these extra contributions to the conformational fluctuations are small. This suggests a lack of flexibility for the peptide backbone ring per se and that the whole molecule is tightly structured once the metal exchange rate is reduced.
Out of this investigation, a picture emerges that provides US with a molecular dynamics view of the ferrichrome solution conforniation.