Carbon 13 Nuclear Magnetic Resonance Spectroscopy of ... · THE JOURNAL OB BIOLOGICAL CHEMISTRY Vol. 248, No. 10, Issue (>f May 25, pp. 3721-3732, 1073 Printed in U.S.A. Carbon 13

THE JOURNAL OB BIOLOGICAL CHEMISTRY Vol. 248, No. 10, Issue (>f May 25, pp. 3721-3732, 1073

Printed in U.S.A.

Carbon 13 Nuclear Magnetic Resonance

Myoglobins Carboxymethylated with

Enriched [ 2”C]Bromoacetate*

Spectroscopy of

(Received for publication, 11ecember 7, 1972)

ALAN $I. NIGEN,~ PHILIP KEIM, ROBERT C. MARSHALL, JON S. %‘IORROW,§ ROBERT A. VIGNA,~ AND FRANK R. N. GURD

From the Department of Chemistry, Indiana University, BloomGgton, Indiana /t7,$01

SUMMARY

Cyanoferrimyoglobins of harbor seal and sperm whale were carboxymethylated with enriched [Z-13C] bromoacetate. The enriched adducts were readily observed by 13C nuclear magnetic resonance. Resonances were identified by comparison with enriched adducts to appropriate small mole- cules, and Tr values were determined under various conditions. In addition to the expected adducts to the NHz- terminal residues and to histidine and lysine, an alkali-labile glycolate ester product was observed. Denatured forms were obtained by carboxymethylation in 8 M urea which led to complete conversion of all histidine residues to the dicarboxymethyl form. By carrying out carboxymethylation in stages it was possible to limit the labeling to either the normally external or the normally internal set of histidine residues. Treatment under denaturing conditions led to carboxymethylation of both methionine residues. The TI values place some limits on the interpretation of the degrees of mobility of the adduct carbon nuclei.

The immediately preceding reports (1, 2) provide a detailed basis for the description of the results of i3C nuclear magnetic resonance studies of myoglobins carboxymethylated with enriched [2-Wlbromoacetate (3). The methylene carbons of the carboxymethyl adducts are readily visualized in the protein spectra, and reIaxation measurements can be followed over an appreciable range of concentration. Minor products that are easily overlooked in chemical analysis are detectable. Model studies have been made to facilitate assignment of all resonances attributable to the enriched carbon nuclei. The results indicate

* This work was supported by United States Public Health Ser- vice Research Grant HL-14680. This is the 49th naner in a series dealing with coordination complexes and catalytic-properties of proteins and related substances (see Ref. 1).

$ Supported in part by United States Public Health Service Biochemistry Training Grant TO1 GM 1046. Present address, The Rockefeller University, New York, New York 10021.

0 Supported in part by United States Public Health Service Biochemistry Training Grant TO1 GM 1046.

the potential of protein enrichment for W nuclear magnetic resonance techniques (4-9).

EXPERIMEKTAL PROCEDURE

Profein &fodi$cation-The carboxymethylation and analytical procedures have been described (I, 2). The NH2-terminal analysis was according to Stark and Smyth (10). Concentration procedures and conversion to the cyanoferrimyoglobin derivatives were achieved as described previously (1).

Peptide Modi$cation-The pentapeptides, glycylglycyl-n- IysyIglycylglycine and glycyIgIycyl-L-methionylglycylglycine were prepared as part of another study (11) .i a-N-Acetyl-n- histidine was obtained from Calbiochem. L-Valyl-L-leucyl-L- seryl-n-glutamylglycine was a preparation used previously (12).

To determine the chemical shifts of the NH2-terminal derivatives, the lysine-containing pentapeptide, 81 nmoles dissolved in 3 ml of deionized water was treated with a 4: 1 mole ratio of 60% enriched [2-i3C]bromoacetate at pH 7.6. The reaction n-as performed under a nitrogen atmosphere and controlled by the pH stat at 25” (2). After 12 hours the reaction u-as stopped by titrating the solution to pH 1.8. The excess bromoacetate aas removed by three extractions with equal volumes of ether. The aqueous layer was taken to dryness, dissolved in 2 ml of water, and adjusted to pH 5.96. The sample, following the addition of one drop of dioxane as an internal standard, was analyzed by the NMR technique (I). Following that, the reaction was continued at pH 10.5 to obtain an e-amino derivative and worked up as before for the NMR analysis. Another sample of the pentapeptide was treated with the bromoacetate at pH 9.5 for 24 hours, 40 nmoles, 4: 1 molar ratio of bromoacetate, in 3 ml of 0.1 M borate buffer.

The pentapeptide with NHz-terminal valine was carboxymethylated at pH 7.5 under the same conditions as first described above for the lysine-containing pentapeptide. Derivatives of histidine were obtained by treatment of ol-N-acetyl-L-histidine (160 nmoles) with a 2:l molar ratio of 75% enriched [2-*3(‘]- bromoacetate in 3 ml of 0.5 11 phosphate buffer, pH 6.8, for 25 hours. The reaction was stopped as described above.

The carboxymethylmethionine derivative was prepared by

1 V. Glushko, F. R. N. Gurd, P. Keim, P. J. Lawson, R. C. Mar- shall, A. M. Nigen, and R. A. Vigna, observations on model pep- tides, in preparation.

3725

treating the methionine-containing pentapeptide with a 4 : 1 molar proportion of 75% enriched [2-%]bromoacetate at p1-T 5.5 in 0.5 M phosphate for 24 hours.

Synthesis of Enriched [d-%‘]Bromoacetate-The procedure was modified from that of Shaw (13). One gram of 60% enriched [2J3C]acetic acid (Mallinckrodt Chemical Works) was pipetted into a 5-ml round bottom flask, containing a magnetic stirring bar, and fitted with a side arm. Following addition of 0.037 ml of acetic anhydride a reflux condenser and cold finger assembly fitted with a drying tube were placed above the flask. Through the side -arm 0.1 to 0.2 ml of dried and distilled bromine was added. After heating the mixture for about 30 min in an oil bath at 90” the remainder of the stoichiometric amount of bromine was added, and the reaction allowed to proceed for a few hours. When the deep red color had lightened, a further 0.2 to 0.3 ml of bromine was added and the reaction maintained for 2 more hours. Finally, excess bromine was carried off in a stream of dry air. The remaining solid was recrystallized from anhy- drous ethyl ether and petroleum ether at 4”. Yields were normally between 70 and 80%. The melting point was 49 to 50”.

in Fig. 1C would be expected to contain the majority of the resonances of the histidine derivatives, with each of the resonance peaks representing one of the two possible adducts on the imidazole ring. Close scrutiny of Fig. 1B shows that the twin resonances make their appearance on the upfield edge of the 01 carbon envelope. An expanded view of the region of interest, accu- mulatcd over a 50-ppm window is shown in Fig. 2. This spectrum confirms the presence of only four adduct bands and brings out more clearly the greater width of the resonance band centered at 143.3 ppm upfield from CS,. The protein background is barely recognizable.

Carboxymethyl Derivative of Sperm Whale Cyanoferrimyo-

globin-Sperm whale ferrimyoglobin was treated in the same manner with 60% enriched [2-‘%]bromoacetate for 8 days. The composition again was very close to that given previously (Ref. 2, Table I). The pertinent values, expressed as residues per molecule, are 6.0, 3.9, 1.5, and 0.8 for histidine, dicarboxymethylhistidine, e-carboxymethylhistidine and &carboxymethylhisti-

Calculated: C 17.29, H 2.17, Br 57.51, 0 23.02

Found : C 16.92, H 2.45, Br 57.24, 0 22.82

Mass spectrometry showed the enrichment wit,h respect to ‘“C to be 59yo. A similar preparation from nominally 90% enriched [2-‘Qacetic acid (IsoMet) yielded a confirmed 75% 13C enrichment in the bromoacetic acid.

Synthesis of y-Carboxymethyl Ester of AXutamic Acid-The procedure of Takahashi et al. (14) yielded a product heavily contaminated with starting materials that was partially purified by Dowex 50-X8 chromatography on a column (5 X 27 cm) de- veloped with 0.2 M pyridine acetate buffer at pH 3.1. The product finally used for the NMR studies was still contaminated with glycolic acid. The elution profile on the amino acid ana- lyzer corresponded to that obtained by Takahashi et al. (14).

RESULTS AND DISCUSSION

Carboxymethyl Derivative of Harbor Seal Cyano~errimyoglobin- Harbor seal ferrimyoglobin was treated in the usual manner (2) for 6 days, with 30% enriched [2-%]bromoacetate. The composition was very close to that given previously (Ref. 2, Table I). The pertinent values, expressed as residues per molecule, are 6.5. 4.7, 1.2, and 0.7 for histidine, dicarboxymethylhistidine, t-carboxymethylhistidine, and &carboxymethylhistidine, respectively, as well as 11.1 and 18.4 for glycine and lysine: respectively. The NHz-terminal determination of Stark and Smyth (10) gave a value of 0.1 residue of glycine, indicating extensive modification of this residue.

The sample was concentrated in the Amicon apparatus to 14

I / I I I / I I I 0 50 100 150

wm FIG. 1. Proton-decoupled Fourier transform 13C NMR spectra

of harbor seal cyanoferrimyoglobins in 0.1 M phosphate. Reso- nance positions are given in parts per million upfield from CS2. A, untreated myoglobin, pH 7.41, 14 mM, 16,384 accumulations, 1.360 s recycle time, 35”; B, myoglobin treated with bromoacetate at natural abundance with respect to isotopes (I), pH 7.08,12 mM, 60,000 accumulations, 1.360 s recycle time, 35”; C, myoglobin treated as described for B but with 30% enrichment of the methylene carbon of the bromoacetate (see text), pH 7.27, 14.5 mM, 8,192 accumulations, 0.555 s recycle time, 26”.

mnq made 0.1 M in phosphate, pH 7.0, and converted to the cyano form (1). Fig. 1 shows a comparison of the pulsed Fourier

I

transform 13C NMR spectra near pH 7 of unmodified Ij

:’ 1 ,,:”

myoglobin (Fig. IA), harbor seal myoglobin treated with bromo- ;:” acetate, at natural abundance with respect to isotopes (1) (Fig. lB), and the harbor seal myoglobin treated as described above

p ~$@&p~$p~;~~ 1 l,w~~~~~~~,~~~~,~~,~~,~~” ~~~+hP ‘9~.~~V,~~~~~~~~~~,~~~~ “$,q*y#yqj, ;‘h,l

(Fig. IC). In Fig. 1C four resonances stand out from the protein I I I I background, namely at 129.4, 134.4, 141.0, and 143.3 ppm up- IZO 130 140 150 160

field from CS2. The C’ resonance of lysine at 153.4 ppm (7) was QQm

used as a reference to locate the chemical shift positions of these FIG. 2. Expanded view of the region of the resonances of the

peaks. enriched carbon sites in Fig. 16, accumulated over a 50-ppm window. The conditions were unchanged except that 6640 accumula-

Simply from the standpoint of their size, the large twin peaks tions were obtained at a recycle time of 3.100 s.

3726

I I 1 I

130 180 w m

FIG. 3. Spectral region of 13C NiVIR containing resonances of the adducts derived from [2J3C]bromoacetate. A, carboxymethyl derivative of harbor seal cyanoferrimyoglobin, pH 7.27, 14.5 mM, 8,192 accumulations, 0.555 s recycle time, 26”, adduct enrichment BOcj, ; B, carboxymethyl derivative of sperm whale cyanoferrimyoglobin prepared with 60?/‘, adduct enrichment as described in the text, pH 6.94, 3.5 mM, 46,159 accumulations, 0.555 s recycle time, 28”.

dine, respectively, as well as 7.0 and 18.3 for valine and lysine, respectively. The NHz-terminal determination of Stark and Smyth (10) corresponded to the recovery of at most 5”i0 of the valine following the modification. In keeping with previous experience (15), a low recovery of about 70% was obtained with the untreated sperm whale myoglobin.

The region of the i3C Nh1R spectrum corresponding to that occupied by the adducts is shown in Fig. 3, where the seal myoglobin derivative is shown in A and the sperm whale myoglobin is shown in B. Similarities are great, although the resonance at 134.4 ppm upfield of CS2 is significantly absent from the spectrum of the sperm whale derivative. Various peptide derivatives wcrc then studied to permit identification of the adduct resonances.

Peptide Derivatives: NHz-terminal-The sample of glycylglycyl- n-lysylglycylglycine carboxymethylated with [2-%]bromoacetate at pH 7.6 was found by amino acid analysis to contain monocarboxymethylglycinc. The dicarboxymethylated product dots not give a positive test with ninhydrin, although its presence was inferred from the extent of loss of glycine in amino acid analysis. The lysine residue was not modified. Two resonances were observed in i3C NMR at 134.8 and 143.4 ppm in the solution at pH 5.96. Upon titration with NaOH to pH 10.42, the resonance initially at 134.8 ppm shifted to 134.0 ppm, while the other shifted to 140.5 ppm. Since the upfield peak disappeared following t)he second round of carboxymethylation at pH 10.5 and readjustment of the pH to 6.10, and the downfield resonance increased concomitantly, the assignments of the downfield and upfield peaks were taken as the dicarboxymethyl and monocarboxymethyl derivatives, respectively.

For the NH2 terminus of the sperm whale prot,ein a pcptide corresponding to the first 5 residues in the sequence (16) was available (12). The peptide yielded on carboxymethylation a product that had an additional resonance (12) at 143.0 ppm at pH 5.20. By amino acid analysis it was found to contain mono- carboxymethylvaline. This result matches the chemical shift of the corresponding monocarbox,ymet,hylglycine derivative. By

analogy, the dicarboxymethylvalinc would be expected to occur near 135 ppm at neutral pH. The absence of a new resonance near this position in Fig. 3B confirms that the NH.&erminal valine of the sperm whale protein was only monocarboxymethyl- ated. The restriction to this product, which would contribute to the largest peak in Fig. 3B, is consistent with the higher pK, for the terminal amino group in this protein than in the harbor seal protein, as well as with some evidence for a generally more hindered reactivity.3

Peptide Derivative: Lysine t-Amino Group-The sample of the peptide treated as described above at pH 10.5 showed a resonance at 135.6 ppm at pH 7.52 ascribable to the dicarboxymethyl derivative of the e-amino group of the lysine residue, in addition to the dicarboxymethylglycine resonance. The presence of this lysine derivative but not of the mono derivative was confirmed by amino acid analysis. A separate treatment at pH 9.5 for 24 hours produced both e-amino derivatives in addition to the mono- carboxymethylglycine. Spectral measurements at p1-I values of 7.52, 8.34, and 10.35 made it possible to recognize the resonance of the monocarboxymcthyllysine adduct, which appeared at 143.5, 143.4, and 141.4 ppm under these three conditions, respectively.

Peptide Derivative: Histidine Imidazole Group-The reaction product obtained from the treatment of oc-N-acctyl-L-histidine was found on amino acid analysis to contain the e-carboxymethyl, Lcarboxymethyl, and 6, e-dicarboxymethyl derivatives of histidine in the molar ratio of 3 : 1: 1. The i3C NMR analysis at pH 7.02 gave chemical shifts of 140.8 and 143.1 ppm which could be recognized from separate experiments as representing the E- and b-adducts in the dicarboxymethyl derivative. The most intense resonance, at 142.2 ppm, could be ascribed directly to the e-carboxymethyl adduct in the e-monocarboxymethyl derivative. The resonance at 144.0 ppm was taken to represent the b-cnr- boxymethyl adduct in the fi-monocarboxymethyl derivative. The areas under the assigned peaks corresponded to the ratios expected from amino acid analysis. This sample was then

treated further with a 15-fold molar ratio of unenriched bromoacetate at pH 7.2 for 36 hours. The most intense peak, still representing the e-adduct but now cont,ained in the dicarboxymethylhistidine form, was shifted from 142.2 ppm to 140.8 ppm at pH 7.02. Similarly, the original resonance at 144.0 ppm moved coincident with the resonance at 143.1 ppm, still rcpre- senting the d-adduct but now contained in the dicarboxymethylhistidine form.

Peptide Derivative: Methionine Thioether Group-Tile methionine sulfonium salt formed by carboxymethylation with the [2- %]bromoacetate of the pentapeptide glycylglycyl-L-methionylglycylglycine showed a single strong resonance at 145.7 ppm at pH 6.21.

Table I summarizes the chemical shift positions for the various carboxymethylated derivatives dealt with above.

Assignment of Recognized Derivatives in Protein Spectra-The resonances in Fig. 1C or Fig. 2 at 134.4, 141 .O, and 143.3 ppm can be assigned dominantly to the dicarboxymethyl derivative of glycine, the e-carboxymethyl adduct in dicarboxymethylhistidine, and the d-carboxymethyl adduct in dicarboxymethylhistidine, respectively. The presence of monocarboxymethyl derivatives of the terminal glycine or lysine side chains, as well as either of the monocarboxymethyl derivatives of histidine

2 The formation of this derivative was postulated previously on tentative grounds (17).

3 M. II. Garner, W. 1-I. Garner, and F. R. N. Curd, in prepara tion.

3727

TABLH I

Chemical sh,ift positio,x of various carboxymethyl clerivcctives

The amino acid residlles listed were alkylated with [2-lG]bromoacetate to obtain enriched carboxymethyl adducts whose characteristic resonance positions are expressed as parts per million upfield of CS,. Reaction conditions are given in the text. The 1sC NMR measllrements were made near 28”. The four sets of observations were made at pH 7.02, 7.52, 7.52, and 6.21, respectively.

Residue carbosymethylated

IIist,idine”

(ilycine”

Lysine”

Methioninec

- Derivative observed

Form

&Monocarboxymethyl E-Monocarhoxymethyl I)icarboxymethyl individllal

adducts &Carboxymethyl &:trboxymethyl

NHz-terminal addu& Monocarboxymethyl Uicarboxymethyl

t-NH, adducts Monocarboxymethyl Dicarboxymethyl

Carboxymethylsulfonium salt

Chemical shift

144.0 142.2

143.1 140.8

142.0 134.0

143.5 135.6 145.7

Q Initially or-N-acetyl-L-histidine. b Initially glycylglycyl-L-lysylglycylglycine. c Initially glycylglycyl-r,-methionylglycylglycine

would be obscured under the large resonance at 143.3 ppm. This band is relatively broad compared with the one at 141.0 ppm (Fig. 2), consistent with these conclusions. There is no evidence in these spectra of the dicarboxymethylated lysine derirat,ive or of the carbosymcthylsulfonium salt of methionine.

Similar conclusions apply to the sperm whale cyanoferrimyoglobin spectrum in Fig. 3B. The absence of a dicarboxymethyl- valine NHz-terminal adduct is apparent and has been discussed above. In this case the relative importance of the monocar- boxymethylhistidine forms compared with the dicarboxymethylhistidine is greater than for the seal myoglobin derivative shown in Fig. 3A. This fact may account for the relatively more striking dominance of the 143.3 ppm resonance band.

Identijication of Resonance at 129./, ppm-None of the model reactions investigated above gave aiy indication of the nature

of the resonance at 129.4 ppm upfield of C&. This resonance position falls between those characteristic of threonine CS and serine Co (7). During an experiment on carboxymethyl myoglobin at pH 10.0 this resonance began to disappear with time, suggesting that the associated adduct was either covalently bound to the protein by a bond labile to base, or bound by a non- covalent interaction that was disrupted at high pH. The results of exposure to pH 10.0 are shown in Fig. 4. The control spectrum at pH 7.06 of the carboxymethylated harbor seal myoglobin is shown in Fig. 4R. After equilibration for 24 hours at 26” at pH 10.04 the spectrum shown in Fig. 4B was obtained. Re- equilibration at pH 6.8 for 3 weeks led to the spectrum in Fig. 4C, taken at pH 7.08. The resonance at 129.4 ppm that was lost at pH 10 failed to reappear, but a barely noticeable narrow resonance made its appearance at 131 ppm.

The results point to the ester adduct reported by Takahashi et al. (14) who observed the formation of the glycolate ester of a glutamic acid residue in carboxymethylated ribonucleasc T,.

PH

I I I I

130 180 twm

FIG. 4. Spectra of carboxymethyl cyanoferrimyoglobin of harbor seal taken through a cycle to pH 10 to identify an adduct resonance labile to base. A, control spectrum before exposure to base of myoglobin treated with 30% enriched [2-‘3Clbromoacetate for 6 days, pH 7.06, 4.5 mM, 44,856 accumulations, 26”; B, same sample after 24 hours at pH 10.04 at 2G”, observed at pH 10.04, 22,457 accumulations, 28”; C, same sample re-equilibrated at pH 6.8 for 3 weeks, observed at pH 7.08, 15,172 accumulations, 27”.

The same type of reaction has been postulated in other cases (18, 19). The enriched methylene should have a chemical shift close to that observed (20, 21). As a free small molecule glycolate should undergo very rapid tumbling with a greatly length- ened relaxation time and the observed resonance should exhibit marked attenuation at the recycle time employed here (22). The y-carboxymethyl ester of L-glutamic acid was then prepared (14). Assignments could be made for each of the carbon nuclei in the molecule, with the alcohol methylene located at 129.4 ppm, matching the position obtained in the protein spectra. The presence of free glycolate in the base-treated sample also allowed confirmation of the resonance at 131 ppm as Co of glycolate, in keeping with a separate observation on an authentic glycolate sample.

The glycolate adduct remained with the polypeptide moiety during the short exposure to low pH as the heme was removed (23) from the carboxymethyl seal myoglobin. In an experiment to be described more fully later an unenriched carboxymethyl seal apomyoglobin preparation was further treated with [2- ‘VJbromoacetate in the presence of urea (1). This experiment showed an intensity of the resonance at 129.4 ppm corresponding to the addition of approximately two glycolatc ester groups coming on top of the one to two groups added in the first alkylation stage. Furthermore, this second st,age was carried out with a disrupted protein structure. Most probably the glycolate ester formation depends on no special aid from neighboring residues, in contrast to the relatively facilitated reaction described for ribonuclease T, (14). In the present case a discrimi-

3728

minal adduct (134 ppm) which is observable only in the seal protein derivative. The glycolate ester methylene (129 ppm) has nearly twice as large a Tl value, reflecting more freedom of motion in this adduct attached to a relatively flexible side chain (1, 7, 11, 22, 26-28). The value for the NH*-terminal adduct represents the average of the two carbon nuclei in this dicarboxymethyl derivative. Those for each of the histidine adducts represent averages over larger numbers of nuclei, about 6 or 7 modified residues. Even so, the relaxation of these adducts is charac- terized adequately for each resonance by a single T1 value. Furthermore, the standard deviations of the Tl values are of modest degree and no multiple components are observed in t.he relaxation spectra obtained near the null point.

As a first approximation each directly bonded hydrogen borne by a particular carbon will affect the relaxation of the carbon nucleus to the same degree, and hence comparison between carbon types can be made in terms of NTl, where N is the number of such directly bonded hydrogens (1, 22, 27). The NT1 value for the histidine and NHs-terminal adducts in most cases is about 76 ms compared to 28 ms for the Q: carbons of the protein (1). This result indicates that the enriched loci reflect some contributions from internal motion relative to the average motion of the a! carbons (7, 26-28). The NT1 values found here could be classified in the range of those side chain carbons to which relatively low mobility has been ascribed (7). Further com- parisons are postponed until the results on the denatured con- formations have been presented.

Carboxymethylated Apomyoglobin Spectra-l3efore turning to some enriched preparations corresponding to the highly car- boxymet.hylated preparation that was studied in great detail in the preceding paper (1) to show the behavior of a denatured myoglobin, it is instructive to observe the effects of pH on tile spectrum of the apomyoglobin prepared from a normally alkylated sample of the seal protein. The sample treated for 6 days with 30% enriched [2-Wlbromoacetate, shown in Fig. IC and presented in the first two entries in Table II, was freed of heme and studied under different conditions of pll. Apomyo- globin is known to undergo conformational changes at high pH (23)) and carboxymethylated myoglobins experience conformational changes near pH 5 and pH 11 (29).

The spectra in Fig. 5 show the carboxymethyl cganoferrimyo- globin from harbor seal from which the apomyoglobin was prepared (Fig. 5A), to be studied at pH 5.00 (Fig. 5B), 1pII 9.39 (Fig. 5C), and pH 12.24 (Fig. 513). The spectrum at pH 5.00 deserves attention representing as it does a pH region where aggregation of the protein is incipient but structural disruption is not yet observed (23, 29). The effects seen in Fig. 5B include line broadening and reduction of intensity, both apparent throughout the spectrum (B) compared to A or C, but most clearly seen with the major adduct resonances. The area under these rcsonanccs is in fact reduced to about one-half of its value in the unaggre- gated preparations. It is possible that aggregation has reached the point where signal strength is much attenuated. The de- crease in signal intensity may also reflect the effects of an increased rn on T2, reflected in line width, and on the nuclear Ovcrhauser enhancement, reflected in intensity of these proton- decoupled spectra (27). Such an argument puts certain limits on the value of the correlation times describing the motion of the enriched adducts (27). Furthermore, the possibility of exchange broadening cannot be ignored. If such an effect contributes to spinlattice relaxation it might also reduce the nuclear Over- hauser cnhancemcnt.

TABLE II

1; values for resona?lcea corresponding to enriched carboxymethyl adducts

Spin-lattice relaxation times, 2 1, in milliseconds, are given for the four prorninent resonances, identified by chemical shifts expressed in parts per million upfield of C&. The cyanoferrimyoglobin derivatives are grouped according to animal species, and the bromoacetate percentage enrichment used, pH, and concentration of measurement are listed. Methods of preparation and TX determination are referred to in the text. The probe tempera- tures were stable within 1” during each experiment. Dilutions were with 0.1 M phosphate buffer, pH 6.8. Computed standard deviations are given in parentheses. The average signal-to-noise ratio for the major resonances was 8:l.

Sample conditions Chemical shift of resonance

Carboxymelhylated seal cyanoferrimyoglobin

30 7.27 14.0 26.0” 72(8) 40(l) 38(3) 39(2) 30 7.06 4.5 28.0 @@I 38(2) 40(2) 38(2) 75 6.87 0.9 31.3 58(14) 43(3) 40(4) 35(3)

Carboxymelhylated sperm whale cyanoferrimyoglobin

nation between glutamic acid and aspartic acid reaction sites cannot be made. The lability of the product during acid hydrol- ysis has delayed its detection.

Relaxation Behavior of Adducts in Native Proteins-Table II lists the values of the spin-lattice relaxation time, T1, in milliseconds, for various concentrations of the carboxymethyl derivatives of the harbor seal and sperm whale cyanoferrimyoglobins. The resonances of each enriched adduct are represented, with standard deviations given in parentheses. With the exception of the resonance at 134 ppm, which is absent from the sperm whale myoglobin spectrum (Fig. 3B), the patterns are very similar between the two species. This is in agreement mith the nu- merous points of similarity seen in the natural abundance spectra (1) and the general similarity of the histidine reaction patterns (2). Considering the possibilities for heterogeneous distribution of sites of glycolate ester formation it is not surprising that some of the standard deviations are quite large for Tl for the glycolate resonance at 129 ppm. The relatively low signal-to-noise for this resonance also contributes to the magnitude of the standard deviations observed.

An important advantage of the W enrichment is the increased spectral sensitivity gained. This has made possible the study of a 15.fold concentration range covered in Table II. The absence of clear trends in TI with concentration, as well as of any indi- cations of line broadening or discontinuities in relative signal-to- noise due to aggregation (24), is evidence that the high concentrations required for the NMR do not bring with them intractable intermolecular interactions. The self-diffusion study of myoglobin by Riveros-1Moreno and Wittenberg (25) showed that below about 6 mM very little concentration dependence was evi- dent in translational diffusion.

The relaxation times for the methylene carbons of the histidine adducts (141 and 143 ppm) are similar to those of the NHz-ter-

3729

1

0 I

50 I

100 wm

I

150

FIG. 5. Effect of pH on spectra of carboxymethyl apomyoglobins. A, control spectrum of carboxymethyl cyanoferrimyoglobin, as in Fig. IC. This material was used as described in the text to prepare the apomyoglobin studied in Spectra B, C, and D. B, the apomyoglobin from A, pH 5.00, 5 mM, 20,624 accumulations, 0.555 s recycle time, 27’; C, same as in B, pH 9.39,20,001 accumulations; D, same as in B, pH 12.24, 16,384 accumulations; E, a separate preparation of apomyoglobin of harbor seal obtained by first treating with 75% enriched [2-i3Clbromoacetate for 6 days as the native protein, then removing the heme and treating the apomyoglobin in 8 M urea with unenriched bromoacetate as described in the text. The observations were at pH G.86,2 mM, 16,384 accumulations, 0.555 s recycle time, 32”.

Note that the spectrum in Fig. 5C at pH 9.39, 5.0 mM, rc- sembles quite closely that in Fig. 5A for the ferrimyoglobin. Structural changes on removal of the heme are not very striking on a time-average observational basis (23, 30, 31). The similarities extend to line widths, intensities, and, as shown below Tl values. However, in Fig. 50, at pH 12.24, the spectrum be- comes characteristic of a substantially altered state. There is an obvious narrowing of the resonance peaks, particularly of the adducts. The sharpening of the resonance at 134.4 ppm is very dramatic indeed. So far as the protein background is concerned in Fig. 50, many examples of sharpening may be seen, not least in the carbonyl region. The spectrum is reminiscent of that of the fully carboxymethylated derivative studied previously at natural abundance (1). The conversion back from pH 12.24 to the pH 9.39 range was found to yield the spectrum in Fig. 5C in a fully reversible fashion. The areas under the adduct resonances were comparable in the two spectra, confirming that the nuclear Overhauser enhancement was probably near its maximum in the native structure (Ref. 27, Fig. 4).

heme and then by a second, separate alkylation treatment in the presence of 8 RI urea as described previously (1). In both cases the final product corresponded to a preparation already reported (I) and was found on analysis to contain no unmodified histidine or methionine and to have suffered a loss of 2 or 3 lysine residues. The two preparations were treated with 75c/, enriched [2-WJbro- moacetate in one stage and with natural abundance bromoacetate in the other. They differed in the essential way that the first was enriched in the first stage and the second in the second stage. The reaction patterns were very similar. The initial product contained 5.8 and 5.7 unmodified histidinc residues per molecule in the two cases, respectively.

The preparation enriched in the first stage corresponds in its enrichment pattern to those groups modified in the native structure and is suitably included in Fig. 5 as the spectrum marked E. The comparison with Fig. 5A, C, or D shows first a comparable area under the main resonance bands in each case when allowance is made for differences in concentration, number of accumulations and vertical display scale. This observation is added support for the contention that the nuclear Overhauser enhancement is near maximum in all these cases (27). As expected for the disrupted structure obtained with the second alkylation stage (1) the spectrum in Fig. 5E is sharpened with relatively narrow resonance bands.

Denalured Carboxymethylated Apomyoglobin Spectra-Two Direct comparison between these two highly modified apomyo- preparations of harbor seal apomyoglobin were obtained in which globin preparations is made in Fig. 6. The first spectrum (A) is all histidine residues were converted to the dicarboxymethyl that already presented in Fig. 5E. The second (B) presented on form. In both cases a first treatment was made for 6 days under a different scale represents the preparation enriched in the second the standard alkylating conditions, followed by removal of the stage during which the protein structure was disrupted and the

ill

1 105

I I I 130 155 180

PPm

FIG. 6. Comparison of spectra of harbor seal apomyoglobin carboxymethylated in two stages, one with enrichment at the native stage and the other with enrichment at the denatured stage. A, same as Fig. 5E, with enrichment at the native stage; B, the 75% enriched 12-Wlbromoacetate was aunlied at the stage of treatment of the apoprotein in 8 M urea.

-_ The pH values were 6.86 and

6.84, respectively. Except that the second spectrum is displayed at a higher vertical magnification, the conditions were identical: 2 mM, 16,384 accumulations, 0.555 s recycle time, 32”. Spectrum A is labeled with respect to adducts to the normally reactive sites in the native protein structure and Spectrum B with respect to those normally internally masked.

3730

TABLE III

7’1 values for carboxymethyl seal apomyoglobins

The T1 values are in milliseconds, and chemical shift positions are upfield of CS?. Standard deviations are given in parentheses. Methods of preparation are given in the text. See Table II for further information. The signal-to-noise ratio for the last two preparations listed was at least 35: 1. Otherwise it was 8: 1 (Table II) or better.

m.w

Nat.ive, heme intact. 7.27 I 1

14.0 26.0” As above, heme removed 9.39 5.0 26.5 Sample from pH 9.39. 12.24 5.0 26.0 Two stage, “extelnal” label., 6.86 2.0 33.0 TWO stage, “internal” label.. 6.84 2.0 33.0

normally interual nucleophilic groups became emphasized by alkylation with the enriched [2-‘%]bromoacctate. Two new

resonances are present in Fi g. 6B at 145.5 and 135.0 ppm. On

the basis of the peptide model reactions already presented, these can be recognized as the methylene adduct carbons of the methi-

ouiue carboxymethylsulfouium salt and some dicarboxymethyl-

Iysiue, respectively. No further formation of the terminal adduct, dicarboxymethylglycine, is indicated. As discussed earlier,

the glycolate ester continued to be formed. The intensity of the

resonance of the methioniue adduct, representing two carbon nuclei per molecule, is of the expected relative magnitude. The

contrasting results for methionine labeling in the two parts of

Fig. 6 point up the specificity of the alkylation reaction with rc- spect to the &act protein structure more clearly than any other

single piece of evidence obtained with myoglobin. Relaxation Beha,vior of Carboxymeihyl Apomyoglobins-The

spin-lattice relaxation times, T,, of the various enriched, car-

boxymethylated apomyoglobins are collected in Table III. The values of T1 in milliseconds are given for the five resonances

discussed above in terms of the carboxymethyl combination with the carboxyl groups (129 ppm), the amino terminus (134 ppm), the histidine e-derivative of the dicarboxymethyl form (141 ppm), both histidine monocarboxymethyl derivatives and the d-derivative of the dicarboxymethyl form (143 ppm), and the methioninc side chaius (146 ppm). The values already presented in Table II for the carboxymethyl cyanoferrimyoglobin of the harbor seal are entered first for reference, followed by the corresponding apomyoglobin form at pH 9.39 and pH 12.24. Then the values for the highly carboxymethylated denatured apomyoglobin

preparations are presented, first with the labeling in the initial (native) stage and second with the labeling in the second (denatured) stage. For convenient reference, the normal Fourier transform spectra of these preparations are to be found, respectively, in Fig. 1C (also Fig. 2 and Fig. 3A), Fig. 5C, Fig. 50, Fig. 6A (also Fig. 5E), and, finally Fig. 6B.

Bs shown in Table III, the removal of the heme had little effect on the major histidine adduct T1 values. The low signal strength made necessary the omission of a value for the NH2- terminal adduct in some cases. Slow loss of the glycolate ester peak likewise eliminated any reliable T1 value for the 129.ppm resonance in alkaline solution. The material observed at pH 12.24 showed a significant increase in Tl for the NHz-terminal adduct and possibly significant increases for the histidine adducts. The latter resonances exhibited considerable heterogeneity in relaxation behavior under these conditions. The highly carboxymethylated, disrupted structure labeled in the first reaction

Protein

Sample conditions Chemical shift of resonance

-

pH COIlCell- tration Temperature 129 ppm 134 ppm 141 ppm 143 ppm 146 ppm

72(x) 40(l) 380) 39(2) 366) 35(5)

626) 47 (7) 43(7) 129(16) 87(2) 89(2) I 710) 188 (5) 93(4) 74(2) 157 (12)

stage, and therefore identical in mrichmcnt pattern to the preparations already listed, showed very substantially longer Tl values. Lastly, the preparation labeled in the second stage showed at least as long TI values for the histidine adducts as its converse preparation, as well as a louger value for the glycolate ester. The rather long T1 value for the mcthionine adducts is noteworthy but not subject to comparison within this study.

In these last two preparations the histidine adducts are present solely in t,he dicarboxymethyl form. Hence, with the cxxception of a small overlay of resonance in the 143 ppm position of lysine adduct, the two peaks at 141 and 143 ppm are nearly directly comparable. The e-adduct at 141 ppm shows a somewhat longer T1 than the d-adduct in both cases. This obscrvatiou fits with

the results found previously with the same (unenriched) chemical modification of the protein (I), in which the histidine C’ and C6 resonances showed T1 values of 86 and 78 ms, respectively. In that case the experimental error was rather too large for the C6 to support a firm comparison with the present results. It is interesting to note that where the dicnrboxymcthylglycine (NHS- terminal) resonance can be seen it shows a T1 value coircsponding

to that of the e-adduct on the histidine residues. Rafational Correlation Times--The rotational correlation time,

7f1,4 for the isotropic motion of the o( carbolls in the native, harbor

4A detailed description of the interpretation of T1, the spin- lattice relaxation time, and 7’2, the spin-spin relaxation time, in terms of rotational correlation times is available (27). In the present case where the 13C_1H dipolar relaxation mechanism is dominant (22), the T, of the Ca envelope, measured with complete proton decoupling, can be interpreted directly in terms of a single 7n, the correlation time for over-all reorientation, only if one characteristic isotropic motion is assumed. Although the assump- tion of a single 7~ for th Cm envelope is not an unreasonable one for the native myoglobin conformation (l), a more general description in terms of doff, the effective correlation time for over-all motion, allows for more than one defining correlation time to be included implicitly in the treatment of the denatured structure. Whenever the T, vallles for protonated carbon nuclei beyond the Ca backbone are considered, the use of 7eff is clearly warranted since several independent motions are likely to contribute to the description. For the special case where a specific 7R for a rigid backbone can be assumed, the internal motion of a directly attached protonated carbon can be described by an additional correlation time, 70 (27). In this case the 7eff for a peripheral carbon can be sepa- rated into two components, 7~ and 7~. This treatment can be shown to apply directly to p carbons with a single degree of freedom of internal motion attached to an isotropically reorienting, rigid 01 carbon system. Application to carbon nuclei further removed, i.e. 0, 0, and Cc, is strictly approximate in view of the potential sources of significant contributions from several degrees of motional freedom.

3731

seal myoglobins can be computed from the reported Tl values (1) as either 3.3 or 17 ns according to the solution of a function yielding two real values (7, 8). A similar calculation for the denatured harbor seal apomyoglobin, based on a value (Ref. 1, Table II) of 52 ms for T,, yields a solution for TR of approximately 1.0 ns that can be selected with confidence on the basis of line width (1). This value is very close to that previously reported for denatured ribonuclease A (7). The T, for the cr carbon nuclei in the denatured protein is dominated by the correlation time of segmental motion of the backbone (28).

The same treatment can be extended as a crude approximation to yield Tefr values (7,s) for the enriched carboxymethyl adducts of 1.5 ns for the native harbor seal protein. The companion Tefr value of 29 ns is excluded on the basis of careful observation of line width (Fig. 2) and signal intensity (27). In the denatured carboxymethyl protein the adducts on N6 and N’ gave Tefr values by this treatment of 0.33 and 0.26 ns, respectively. Ex- cept in those cases where a number of rotational correlation times are required to interpret T,, the value of Terf can be taken as a measure of relative motion (7, 8, 22, 24, 27, 28).

The above simple computations argue that the methylene carbons of the adducts undergo increased internal motion compared to the o( carbons when either state of the protein is considered. Making a comparison in terms of T, values for carbon nuclei in the unmodified myoglobins and the unenriched carboxymethyl derivatives (l), it can be seen that the adducts fall generally between values characteristic of p and y carbons. The methylene carbons of the adducts, however, are more the equivalent of 6 carbons in terms of their potential degrees of freedom.

The treatment of Doddrell et al. (27) is properly limited to the first degree of freedom of motion relative to a rigid backbone. It is suitable for the appropriate cases of extended rigid side chain groups (24, 28). The present case requires a more elaborate analysis such as that proposed by Levine et al. (32), for which it would bc necessary to have more detailed knowledge of the relaxation behavior of all carbon nuclei in the side chain.

1%~ comparison with an alkyl side chain, the bulky aromatic ring of the modified histidine side chain in the protein may be expected to bc more hindered in its internal rotational motion. This effect could have the observed consequence of putting the methylenc adduct into the Tl class of p or y carbon rather than of a 6 carbon, especially in view of the ranges of Teff involved (27). Indeed, for the aromatic ring of phenylalanine in these myoglobins there is evidence from the magnitudes of the T1

values that motion about the C~-C~ bond axis is not dominant. When such motion dominates, T1 values for CY and C’ may be expected to be distinctly greater than for Cc (33, 34). The basis for that difference would depend on the failure of such motion to modulate the 13C-111 dipole-dipole interaction (33). For phenylalanine residues in the native protein the TI values (Ref. 1, Table II) for C* and C’ are 40 ms, and for Cl, 46 ms; the corresponding values for the denatured protein are 101 and 107 ms, respectively. Clearly, other types of motion dominate that about the CD-0 bond axis in these cases.

On steric grounds the imidazole-bearing side chain of histidine may be expected to have comparable motional restraints to those of the phenylalanine. In terms of range of T1 values in the denatured case for the imidazole C* and Cc (Ref. 1, Table II) this prediction is borne out. Within the expected range of contributions to T,ff, if the motion about the CD-C? bond axis were dominant a clear distinction between @ values for C6 and Cf would be expected (35). Such a distinction is not observed. If the correlation time for rotation about the Cp-CY axis is suffi-

ciently long that it does not contribute importantly to the T1

values of the imidazole ring, then a calculation may be appropriate to interpret the Tl values for the imidazole ring carbons in terms of just two correlation times. These correlation times are the 7~ for the isotropic motion of the (Y carbons and the 7c for the internal motion about the C”-CY bond axis (27). From this computation 76 for the imidazole ring is approximately 0.5 ns for the denatured protein. This argument could be tested if Tl

values for the ,L? carbons were observable. Enough degrees of freedom have been shown to be in play in

the nanosecond range of correlation time that the motional behavior of the enriched adducts in the denatured protein must be described in terms of contributions involving all the correlation times. Provided that segmental motion of the backbone and internal rotation about the C”-Co bond axis are dominant, their correlation times should modulate to the same extent the T, values of the enriched adducts. The significant differences between Tl for the two enriched methylene carbon nuclei listed for the denatured cases in Table III probably depend, therefore, on differences in internal motion about the C-N bonds. On general steric grounds greater freedom of rotation is to be expected for the e-carboxymethyl adduct.

A corresponding analysis of the native case is hampered by the uncertainty in choice of 711 value and the poorer definition in the spectra. For example, detailed information about the ring motion is not available. Considering the elements of restricted motion characteristic of the native structure, the interpretation from the 7,ff values of increased internal motion for the adducts relative to the 01 carbons is probably justified. It is interesting that the T1 values for the enriched 6- and e-mcthylene carbon nuclei are indistinguishable in this case. More detailed specific enrichment with 13C is very desirable.

Acknowledgmelzts-The help of illr. A. Clousc and Mr. T. Roseberry with the magnetic resonance equipment is gratefully acknowledged. In addition to providing advice and encourage- ment, Professor A. Mlcrhand made available a computer pro- gram. Mr. J. Schneider kindly performed the analyses by mass spectrometry.

RFFERENCFS 2

1. NIGEN, A.M., KEIX,P., MARSHALL, 11. C., GLUSIIKO, V., L.iw- SON, 1'. J., AND GURD, F. 1:. N. (1973) J. Biol. Chem. 248, 3716-3723

2. NIGEN, A. M., AND GURD, F. R. N. (1973) J. Biol. Chem. 248, 3708-3715

3. NIGEX, A. M., KEIM, P., MARSH.\LL, R. C., MORROW, J. S., AND GURD, F. R. N. (1972) J. Biol. Chem. 247, 41OOG4102

4. FREEDMBN, M. H., COHEN, J. S., AND CHAIIU~N, I. M. (1971) Biochem. Biophys. Bes. Commun. 42, 1148-1155

5. HORSLEY, W. J., AND STERNLICIIT, II. (1968) J. Amer. Chem. sot. 90, 3738-3748

6. GIBBONS, W. A., SOGN, J. A., STERN, A., CRAIG, L. C., AND JOHNSON. L. F. (1970) Nature 227. 840-842

7. GLUSHKO, $., LA&oN,‘P. J., AND GURD, F. R. N. (1972) J. Biol. Chem. 247, 3176-3185

8. GURD, F. R. N., AND KFAM, P. Methods Enzymol., in press 9. SAUNDERS, D. J., AND OFFORD, R. IL (1972) Fed. Eur. Biochem.

Sot. Lelt. 26, 286-288 10. STaRI<, G. I:., AND SMYTH, D. G. (1963) J. Biol. Chem. 238,

214-226 11. GURD, F. R. N., KEIM, P., GLUSHI<O, V. G., LAWSON, P. J.,

MARSHALL, R.C., NIGEN, A.M., AND VIGNA, R. A. (1972) in Chemistry and Biology of Peplides (MEIENHOFER, J., ed) pp. 45-49 Ann Arbor Science Publishers, Ann Arbor, Mich- igan

12. GURD, F. R. N., LAWSON, P. J., COCHRAN, D. W., ~XD Wsx- ICERT, 15. (1971) J. Biol. Chem. 246, 3725-3730

3732

13. SHAW, 13. 1). (1923) J. Chm. Sot. 2233-2240 14. '~.\IMHMII, K., STMN, W. II., .\xD ~~OORr:, s. (1967) .I. &0l.

25. ALLF:RHIND, A., I)ODDRI:LL, I)., GLUSHII~, V., C~CHR.\N, n.

Chenz. 242, 468%4(i90 W., WENI<I~JRT, E., LEMON, P. J., .UYD GURD, F. Ii. N. (1971) .J. Amer. Chem. Sot. 93, 544454(i

15. HAPX.F:R, K. I>., 131t.\Ds1~.\w, I<. A., H.U~TZI.:LL, C. I<., .uY\‘I) GURD, F. 1~. N. (1968) J. Hiol. Chem. 243, G83-689

27. I)ODDRl:LL, I)., GLUSHIIO, V., ;SD ALIII~:RH.lND, A. (1971) J. Chem. Phys. 66, 3G83-3A89

16. 15m1umms, A. B. (1965) ,Valure 206, 883-887 17. H~GLI, T. 15. (1968) Ph.1). t,hcsis, Indiana University

28. ALLI~:RHMD, A., .\ND H.ULSTONI,:, R. I>. (1972) J. Chem. Ph{/s.

18. SZI~;WCZUI~, A., .YND COSNJSI,I,, (:. P:. (1965) Hiochim. Wiophys. 66, 3718-3720

Ada 106, 352-367 29. CL.~RI<, J. F., .\XD (GIRD, F. It. N. (1967) J. Biol. Chem. 242,

19. I?~,u~I~:I~, W. XI., .\ND BLMYI, K. (1972) J. 13ioZ. Chem. 247, 3257-3264

3123-3133 30. F12~~t~~~~~, 8. C., AND BLOUT, I<. 11. (1965) J. Hiol. Chem. 240,

20. IIAGEN, It., .\SD I~OIS~:RTS, J. 1). (1968) .I. Amer. Chenz. see. 91, 229-303

4504m45CA 31. BILI~:SLOW, IC., I~ISYCHOI~, S., 131\~~~h~, K. I)., .LND Gutm, F.

21. B.\RTUSIU, V. J., .\ND MACII~:L, G. E. (1971)J. Mag?~. IZesonance It. N. (1965) J. Biol. Chem. 240, 304-309

6, 211-219 3%. TI~:v~x~~:, Y. K., I'I\~~~~~~~~, I'., I<OHERTS, CT. C. K., BIRDS\LL,

22. ALLIXH~~ND, A., I>ODDRI~;LL, I)., .\ND KOMOROSIX, It. (1971) 3'. J. R/I., LI.:I,:, A. (:., .\ND MI.:TCALFJ‘:, J. C. (1972) Fed. Em.

J. Chem. Phys. 66, 1811-198 Kochem. Sot. Left. 23, 203-207.

23. BRISSLOW, E. (1964) J. Riol. Chem. 239, 48&4$X 33. I,I,:vY, G. C., WHITI,:, I). hr., .~ND ANICT, F. A. I,. (197%) .I. 24. KOMOROSI<I, It. A., MD ALLERH.\ND, A. (1972) I’TOC. Arat. Magn. Resonance 6, 453-455

Acad. Sci. U. S. A. 69. 180441808 34. LJ:VY, (:. C. (1972) J. fi4agn. I<esonance 8, 122-125 25. I~IVI~ROS-~~~oI~I~:No, V., .\i~ WITTXNP'.RCI, J. B. (1972) J. l?iol. 35. DON~HUI,:, J., ‘YND C:ARON~ A. (1064) Acta Crystallogr. 17, 117%

Chem. 247, 895-901 1180

Documents

Carbon 13 Nuclear Magnetic Resonance Spectroscopy of ... · THE JOURNAL OB BIOLOGICAL CHEMISTRY Vol. 248, No. 10, Issue (>f May 25, pp. 3721-3732, 1073 Printed in U.S.A. Carbon 13