Titration Behavior and Tautomeric States of Individual Histidine Residues of Myoglobins

The titration behavior of individual histidine residues of myoglobins has been studied by observing the pH dependence of the chemical shifts of the nonprotonated aromatic carbon resonances in natural abundance 13C Fourier transform NMR spectra (at 15.18 MHz and 37”),of horse ferrimy- oglobin, horse cyanoferrimyoglobin, and red kangaroo cyanoferrimyoglobin. In the case of the cyanoferrimyoglobins, all nonprotonated side chain carbons of aromatic amino acid residues yield detectable resonances, but only 24 of the 28 carbons of this type yield detectable resonances in the case of horse ferrimyoglobin. Eight of the 11 histidine residues of horse cyanoferrimyoglobin (and 7 of the 10 histi- dines of the kangaroo protein) exhibit titration behavior (pK values in the range 4.4 to 6.6). The imidazole form of each titrating histidine is predominantly (or entirely) in the NfZ-H tautomeric state. Two of the titrating resonances of the cyanoferrimyoglobins (with pK values of 5.3 and about 4.5) do not yield detectable signals in spectra of horse ferrimyoglobin. These two resonances are assigned to 0’ of His-64 and His-97

The titration behavior of individual histidine residues of myoglobins has been studied by observing the pH dependence of the chemical shifts of the nonprotonated aromatic carbon resonances in natural abundance 13C Fourier transform NMR spectra (at 15.18 MHz and 37"),of horse ferrimyoglobin, horse cyanoferrimyoglobin, and red kangaroo cyanoferrimyoglobin.
In the case of the cyanoferrimyoglobins, all nonprotonated side chain carbons of aromatic amino acid residues yield detectable resonances, but only 24 of the 28 carbons of this type yield detectable resonances in the case of horse ferrimyoglobin.
Eight of the 11 histidine residues of horse cyanoferrimyoglobin (and 7 of the 10 histidines of the kangaroo protein) exhibit titration behavior (pK values in the range 4.4 to 6.6). The imidazole form of each titrating histidine is predominantly (or entirely) in the NfZ-H tautomeric state. Two of the titrating resonances of the cyanoferrimyoglobins (with pK values of 5.3 and about 4.5) do not yield detectable signals in spectra of horse ferrimyoglobin. These two resonances are assigned to 0' of His-64 and His-97 (not on a one-to-one basis). Five of the six titrating resonances of horse ferrimyoglobin have pK values (5.5, 5.7, 6.5, 6.6, and 6.6) which are consistent with those of five of the six reported pK values that were obtained from proton NMR spectra. The sixth pK (<5), observed in our 13C NMR spectra of horse ferrimyoglobin and the cyanoferrimyoglobins, does not have a detected counterpart in the reported proton NMR data. Also, the "high" pK (about 7.4 to 8.0) reported in the proton NMR studies of ferrimyoglobins has no counterpart in our 13C NMR spectra of horse ferrimyoglobin and the cyanoferrimyoglobins from horse, kangaroo, and sperm whale.
One of the three nontitrating histidine CY resonances of the cyanoferrimyoglobins from horse and kangaroo (not observed in spectra of horse ferrimyoglobin) is assigned to the coordinated His-93. Our results indicate that the two uncoordinated nontitrating histidine residues are either in the imidazolium or in the NSl-H imidazole state (or a mixture of the two states). The crystal structure of myoglobin suggests that these are His-24 and His-36. We also identify the resonances of CY, C62, and @ of the two tryptophan * This work was supported by the National Science Foundation (Grants MPS 73-04969 and CHE 76-17572) and by the United States Public Health Service (Grants NS 10977-03 and GM 22620-01).
residues, and the resonance of CL of the nontitrating Tyr-146.
Proton NMR spectroscopy has been the method of choice for observing the environment and protonation state of titratable histidine residues of proteins, because the resonances of He1 (and sometimes Hsz) of these residues ( Fig. 1) can be usually observed as resolved single hydrogen peaks with pa-dependent chemical shifts (for a review, see Ref. 1). However, natural abundance 'T NMR spectra of small native proteins yield numerous narrow single carbon resonances of nonprotonated aromatic carbons (2)(3)(4), and the titration behavior of the resonances of CY of histidine residues can be readily observed (4). We believe that, when sensitivity limitations are overcome by the use of large amounts of protein (51, 13C Fourier transform NMR is an attractive alternative (or complement) to 'H NMR for the study of histidine residues, especially when dealing with a protein which contains many histidines. It is noteworthy that 13C NMR can yield information about the tautomeric state of the imidazole form of each histidine residue (6)(7)(8). If the imidazolium form ( Fig. lA) of a histidine residue deprotonates at NSl to yield the WZ-H imidazole tautomer (Fig. lB), then the CY resonance should undergo a downfield shift of about 6 ppm (6-8); if deprotonation yields the N6j-H tautomer (Fig. lC), then the CY resonance should move about 2 ppm upfield (6)(7)(8). In contrast, the proton resonance of HE1 moves upfield when the imidazolium form of the residue is deprotonated either at NSf or at N'* (compare the 'T NMR study of azurin in Ref. 8 with the 'H NMR results of Ref. 9).
In this report, we study the effect of pH on the chemical shiRs of the nonprotonated aromatic carbon resonances of horse ferri-and cyanoferrimyoglobin and of kangaroo cyanoferrimyoglobin.
We assign some of these resonances to specific residues in the sequence, and we present information about the ionization behavior (and tautomeric state) of the histidine residues. Most (but not all) of our results are consistent with reported interpretations of the pH dependence of proton NMR spectra of myoglobins (10-13).

EXPERIMENTAL PROCEDURES
Most materials and methods have been described (14). Some of the spectra obtained with the use of commercial horse myoglobin  (4). However, the 16 nonprotonated aromatic carbons of the heme in a paramagnetic protein have not yet yielded clearly identified resonances in the aromatic region (4). Presumably, these carbons yield resonances that are significantly broadened or shifled (or both) by the paramagnetic center (16). Carbon 13 NMR studies of cytochrome c (4,171 suggest that the resonances of only a few of the nonprotonated aromatic carbons of amino acid residues of myoglobins in paramagnetic states should be significantly affected by the paramagnetic center. In Fig. 2, we show the regions of aromatic carbons (and Cc of arginine residues) in the convolution-difference proton-decoupled 13C NMR spectra of horse ferrimyoglobin (at pH 7.7, Fig Fig. 2, B and C, in our peak numbering system, because it appears that this is a heme carbon resonance (see below). Peak 28 of horse cyanoferrimyoglobin is relatively broad and not clearly detectable in Fig. 2B, but it is observed in spectra with somewhat higher signal-to-noise ratios (Fig. 7).
There are seven types of nonprotonated aromatic carbons of amino acid residues: C' of phenylalanines, Cy and Cc of tyrosines, CT of histidines, and C?, C?, and CY of tryptophans. For convenience, we shall include Cc of arginine residues in our discussion of nonprotonated aromatic carbons of amino acid residues. In the case of horse myoglobin, the 2 arginines, 2 tyrosines, 7 phenylalanines, 11 histidines, and 2 tryptophans (18) contribute a total of 30 nonprotonated side chain carbons.
Only 26 of these carbons yield detectable resonances in spectra of horse ferrimyoglobin (Figs. 2A and 3), but all 30 can be observed in spectra of horse cyanoferrimyoglobin (Figs. 2B, 4, and 7). When going from horse to red kangaroo myoglobin, Tyr-103 and Leu-149 become phenylalanines, His-113 and His-116 become glutamines, and Asn-140 becomes a histidine (18). The arginines and aromatic residues of kangaroo myoglobin contribute 29 nonprotonated side chain carbons, all of which can be detected in spectra of the cyanoferrimyoglobin (Figs. 2C and 5). When going from horse to sperm whale myoglobin, Phe-151 becomes a tyrosine, Asn-12 becomes a histidine, and the number of arginines increases to 4 (18). As a result, the arginines and aromatic residues contribute 34 nonprotonated side chain carbons, 33 of which yield detectable resonances in spectra of the cyanoferrimyoglobin. The exception is Cy of the coordinated His-93 (see below).
Temperature Dependence of Chemical Shifts -Implicit in the discussion of the previous subsection was the assumption that all the numbered peaks of Figs. 2 and 7, but not Peaks a, b, and c of the cyanoferrimyoglobins (see Figs. 7 and 8), arise from nonprotonated side chain carbons of aromatic amino acid residues (and CY of arginine residues). In order to rule out the possibility that some of the peaks under consideration are resonances shifted into the aromatic region by the paramag- netic iron, we examined the temperature dependence of the chemical shifts. As a first approximation, the contact and pseudocontact shifts should be proportional to the reciprocal of the absolute temperature (16). In the case of Peaks 1 to 26 of horse ferrimyoglobin (Fig. 2.4), extrapolation to l/T = 0 yields chemical shifts that differ by no more than about 5 ppm from the corresponding values in Fig. 2A. We conclude that Peaks 1 to 26 of Figs. 2A and 3 arise from 26 of the 30 nonprotonated side chain carbons of aromatic and arginine residues of horse ferrimyoglobin.
The four undetected resonances must be broadened beyond detection (or shifted outside the range of chemical shifts of Fig. 2 The remaining resonances (Peaks 1 to 27, 29,30, a, b, and c) arise from a total of 32 carbons, 29 of which must be nonprotonated side chain carbons of aromatic and arginine residues. The chemical shifts (at 39") and linewidths of Peaks a, b, and c suggest that these resonances arise from nonpromnated porphyrin carbons. The chemical shifts of Peaks a, b, and c extrapolated to l/T = 0 (16 mM horse cyanoferrimyoglobin in DzO, pH meter reading 8.2) are about 140, 98, and 94 ppm, respectively. The small temperature dependence of these chemical shifts is surprising. The peculiar effect of pH on the chemical shifts of these resonances is shown in Fig. 8 (Peak a is not detected in our spectra of sperm whale cyanoferrimyoglobin).
We did not examine the temperature dependence of the chemical shiRs of the cyanoferrimyoglobins from red kangaroo and sperm whale. By analogy with the horse cyanoferrimyoglobin, it is reasonable to conclude that Peaks 1 to 29 of can be observed in our spectra, but one of the 34 carbons of this type in sperm whale cyanoferrimyoglobin is not detected. We show below that the "missing" resonance is probably that of Cy of the coordinated His-93 residue.  Fig. 2) contains only the resonances of C' of tyrosine and arginine residues, and that the upfield region (about 109 to 112 ppm in Fig. 2) contains only the resonances of Cy of the two tryptophan residues. The central region (about 124 to 141 ppm in Fig. 2) contains the remaining resonances of nonprotonated aromatic carbons (4). The pH dependence of only the central region is shown in Figs. 3 to 6. The effect of pH on the chemical shifts of Ci of the tyrosine and arginine residues has been presented elsewhere (14). The chemical shifts of Cy of the two tryptophan residues are essentially independent of PH.
The effect of pH on chemical shifts can be used to identify the resonances of CY and C? of titratable tyrosine residues (14) and those of CY of titratable histidine residues (4). In a previous publication (14), we showed that Tyr-103 (horse and sperm whale) and Tyr-151 (sperm whale) show titration behavior at pH 3 9, and that Tyr-146 (horse, kangaroo, and sperm whale) does not titrate. We also presented specific assignments for the resonances of Cc and 0 of the titratable tyrosine residues (14) (14).
At pH 6 8, six resonances of horse ferrimyoglobin show titration behavior (Fig. 3). The values of the chemical shifts of these resonances at high pH, and the direction of the titration shifts as the pH is lowered, are characteristic of y carbons of titrating histidine residues whose imidazole form is predominantly (or entirely) in the N'Z-H tautomeric state (6)(7)(8). This tautomeric state is the predominant one of histidine residues in small peptides (6), but the Nsl-H tautomeric state has been encountered in some titratable histidine residues of proteins (8, 21). Application of the usual Henderson-Hasselbalch treatment to the chemical shifts of Fig. 3 yields the pK values, the chemical shifts of the imidazole state @a), and the titration shifts (As*) given in Table I. best-fit (single pK) theoretical titration curves.
In the case of horse cyanoferrimyoglobin, we find 8 titrating hi&dine residues, all of which have the imidazole form predominantly (or entirely) in the N'?-H tautomeric state. Fig. 4 yields the values of pK, 8s, and ABA given in Table I. It is apparent from Table I (and by visual inspection of Figs. 3 and 4) that each of the 6 observed titrating histidines of horse ferrimyoglobin (Fig. 3) has a readily identifiable counterpart in the spectrum of the cyanoferrimyoglobin (Fig. 4). The additional 2 observed titrating histidines of the cyanoferrimyoglobin are those which give rise to Peaks 11 and 20 of Figs. 2B and 4. We conclude that these peaks arise from titrating histidine Titration Behavior of Histidines of Myoglobins residues which are close to the iron and, therefore, yield resonances that are too broad for detection in spectra of horse ferrimyoglobin.
We estimate that the Cy resonances of His-93, His-97, His-64, and the other 8 histidine residues (of the ferrimyoglobin) should undergo a dipolar paramagnetic broadening of about 800, 200, 60, and $2 Hz, respectively (see below). On this basis, we assign Peaks 11 and 20 of horse cyanoferrimyoglobin to CY of His-64 and His-97 (not on a one-to-one basis).
Spectra of kangaroo cyanoferrimyoglobin indicate 7 titrating histidine residues, all of which have the imidazole form mainly (or entirely) in the NEP-H tautomeric state. Fig. 5 yields the values of pK, 6s, and AaBA given in Table I. The pH dependence of the chemical shifts of sperm whale cyanoferrimyoglobin was investigated only at pH 2 7.5 (Fig. 6).
It is apparent from Table I that five of the eight titrating histidine resonances of horse cyanoferrimyoglobin have exact counterparts (same pK, as, and A,& in the spectra of kangaroo cyanoferrimyoglobin, and are, therefore, assigned to histidine residues common to both species. Consider now the other There is no resonance in the spectrum of the kangaroo protein that could possibly correspond to Peak 19 of the horse protein (A,, = 3.6 ppm, 8a = 133.6 ppm, pK = 5.51. Thus, Peak 19 of horse cyanoferrimyoglobin must arise from His-113 or His-116. However, Botelho (13) has unambiguously established by means of proton NMR titration studies of human myoglobin (which has Gln-113 and Gln-1161, California sea lion myoglobin (which has His-113 andGln-1161, andother myoglobins (with His-113 andHis-1161, that His-113 andHis-116 have pK values of about 5.5 and 6.5, respectively. Therefore, we assign Peak 19 of horse cyanoferrimyoglobin to His-113. Peaks 14 and 21 of horse cyanoferrimyoglobin (both with pK values of about 6.4) have 3s and AsA values similar to (but not identical with) those of Peaks 15 and 21, respectively, of the kangaroo protein (  (14). Selective proton decoupling can be used to distinguish the resonances of Cc of all tyrosine residues from those of Cc of arginine residues, as follows (4). Single frequency 'H decoupling is applied at or near the resonance frequencies of the arginine CH, protons (about 3.2 ppm downfield from the 'H resonance of Me&, with low enough power to prevent decoupling of the aromatic protons of the tyrosines (6.8 to 7.2 ppm). The arginine Cc resonances remain sharp, while the tyrosine C5 resonances broaden considerably (4). In order to avoid the possible complicating effect of observable scalar coupling of the arginine < carbons to slowly exchanging NH protons of the guanidinium groups, these selective decoupling experiments are usually done on protein solutions in D,O. Peaks 2 and 3 of horse ferriand cyanoferrimyoglobin,  selective proton decoupling as when full proton decoupling is applied. Thus, these peaks are assigned to CY of the 2 arginine residues of these proteins. Peaks 1 and 4 of horse ferri-and cyanoferrimyoglobin and Peak 3 of kangaroo cyanoferrimyoglobin, which undergo broadening when selective decoupling is used, are assigned to C? of tyrosine residues. Thus, Peak 3 of kangaroo cyanoferrimyoglobin must arise from Tyr-146. Since Peak 4 of horse cyanoferrimyoglobin and Peak 1 of horse ferrimyoglobin have been assigned to the titratable Tyr-103 (14), Peak 1 of the cyanoferrimyoglobin and Peak 4 of the ferrimyoglobin must be assigned to Tyr-146. Peak 4 of horse ferrimyoglobin ( Fig. 2A) is relatively broad and has a slightly temperature-dependent chemical shift, 155.4 ppm at 33" and about 158 ppm at l/T = 0 (at pH 6.5). Both effects are consistent with the proximity of C? of Tyr-146 to the iron. With the use of the crystal coordinates of sperm whale myoglobin (19), we calculate that the distances from the iron to the 5 carbons of Tyr-146, Tyr-103, and the arginines are about 10, 12, and more than 15 A, respectively. Note that dipolar paramagnetic broadening is proportional to the inverse sixth power of the distance to the paramagnetic center (16). We can use the measured paramagnetic contribution to the linewidth of the C? resonance of Tyr-146 of horse ferrimyoglobin to estimate the expected paramagnetic (dipolar) broadening of other resonances. Peak 4 of horse ferrimyoglobin ( Fig.  2A) has a paramagnetic broadening of about 5 Hz (measured from spectra obtained without the use of the convolutiondifference method). From this result, we estimate that the paramagnetic (dipolar) broadening should be less than 2 Hz for all remaining nonprotonated side chain carbons of aromatic (and arginine) residues of horse ferrimyoglobin, except the -y carbons of the following residues (numbers in brackets indicate the estimated paramagnetic dipolar broadening, in hertz): His-93 [760], His-97 [1601, His-64 [611, Phe-43 [291, Tyr-103 [4],  Therefore, all resonances except those of CY of His-93, His-97, His-64, and Phe-43 should be readily detectable in our spectra of horse ferrimyoglobin. (' It is difficult to determine from our data if Peaks 10 and 11 of horse ferrimyoglobin ( Fig. 2A) do or do not cross at pH 8 (see Fig. 3). We have assumed that crossover takes place. If this is not the case, the best fit values of pK, &,, and ABBA are 6.5, 136.3,, and 4.6, 2 (4), and CT of nontitrating histidine residues in the imidazolium (4) or NSZ-H imidazole state (6)(7)(8); the nontitrating resonances in the range 134 to 141 ppm can arise from CY of phenylalanine residues (4), CY of tryptophan residues (4), and CY of nontitrating histidine residues in the NL1-H imidazole form (4,(6)(7)(8).
Theoretical and experimental results presented elsewhere (3,4) indicate that the resonances of Ca2 (and @ if D,O is the solvent) of tryptophan residues can be identified on the basis of their long spin-lattice relaxation times (T,), relative to the T, values of y carbons of tyrosine, histidine, and phenylalanine residues (3). Whenever two classes of carbons have different T, values, PRF"l' spectra can be used to distinguish their resonances (4,221. We have applied the PRFT method to ferrimyoglobin and cyanoferrimyoglobin from horse. In the spectrum of the cyanoferrimyoglobin, two components of Peak 9-11 arise from cL2, and Peak 25-26 arises from Cs2 of the tryptophan residues. Because of limited signal-to-noise ratios, our PRFT spectrum of the ferrimyoglobin did not yield the assignment of 01 of the tryptophan residues. Two components of Peaks 22 to 24 ( Figs. 2A and 3) arise from C" of these residues.
By elimination, Peaks 19, 21, and one component of Peaks 22 to 24 of horse ferrimyoglobin ( Figs. 2A and 3) must arise from CY of Tyr-146 and 2 nontitrating histidine residues (but not the coordinated His-93). Also by elimination, the counterparts of these resonances in the spectrum of horse cyanoferrimyoglobin are Peaks 22,23, and 24 (Fig. 4). In the spectrum of kangaroo cyanoferrimyoglobin (Fig. 51, Peaks 22 to 26 have practically the same chemical shifts as Peaks 22 to 26 of horse cyanoferrimyoglobin (Fig. 4). Thus, the assignments presented above for Peaks 22 to 26 of the horse protein are probably applicable to the corresponding peaks of the kangaroo protein. By elimination, Peak 27 of kangaroo cyanoferrimyoglobin (Fig. 5)  (not necessarily on a one-to-one basis). The relatively broad Peak 28 of the horse protein (Fig. 7), assigned above to Cy of His-93, does not have a clearly detectable counterpart in our spectra of sperm whale cyanoferrimyoglobin.  (13) have reported proton NMR studies of ferrimyoglobins from various species. They observed 6 titrating histidine residues in the case of horse ferrimyoglobin.
In Table III, we compare the reported pK values with the ones presented here. The 'H and 13C NMR  7.6 7.6 6.6 6.6 7.0 6.6 6.5 6.9 6.5 6.4 6.6 5.7 5.7 6.0 5.5 5.4 5.7 5.0 a At 36" in H 2 0. b At 16" in DZO, with "pH" taken as uncorrected pH meter reading. c At 20" in D,O, with "pH" taken as uncorrected pH meter reading.
studies are in agreement with respect to the presence of 5 titrating histidines with pK values in the range 5.5 to 6.6. However, our results indicate the presence of a histidine residue with a pK G 5, not detected by 'H NMR. Furthermore, when going to horse cyanoferrimyoglobin, we observe 2 additional titrating histidine residues, with pK values of 5.3 and about 4.4 (see Table I), which were not detected in 'H NMR spectra of sperm whale cyanoferrimyoglobin (13). Most peculiar is the fact that we do not observe the histidine with a high pK (7.4 to 8.0) that has been reported in the 'H NMR studies of ferrimyoglobins (12,13). This high pK has been assigned first to 24) and then to 13). One way to reconcile the 'H and 13C NMR results is to invoke the presence of a tautomeric equilibrium between the N'z-H and N*I-H imidazole forms, which are expected to have Cy chemical shifts about 6 ppm downfield and about 2 ppm upfield, respectively, from the Cy chemical shiR of the imidazolium form of the residue (6-S). If there is fast exchange between appropriate proportions of the two imidazole tautomers, then it is possible for the imidazole form of the residue to have a CY chemical shift that is very similar to that of the imidazolium form. In such a case a titrating histidine would yield a "nontitrating' Cy resonance, but a titrating behavior in the 'H NMR spec- trum (see introductory section). However, the 'H NMR spectra of carbon monoxide myoglobin from horse (10) and sperm whale (13) and of oxymyoglobin and cyanoferrimyoglobin from sperm whale (13) do not yield evidence for the presence of a histidine with the high pK. In the csse of sperm whale szidoferrimyoglobin, Hayes et al. (12) report&d the presence of the high pK, while Botelho (13) did not observe it. Additional work seems in order.
Our results indicate the presence of only 2 nontitratiug histidine residues (other than His-93) in cyanoferrimyoglobins (see Table II). These histidines have chemical shifts consistent with assignments to imidszolium or NS1-H imidazole states (8). Strictly speaking, we cannot rule out the possibility that one or both of these histidines titrate, but with the imidazole form being a mixture of the N"-H and N*l-H mummers, in proportions that yield a chemical shift essentially indistinguishable from that of the imidazolium form (see above). The crystal structure of myoglobin (19) indicates that His-24 and His-36 are likely candidates for the NS1-H tautomeric state.