ADP-Ribosylation of Elongation Factor 2 by Diphtheria Toxin

THEJOURNAL

OF

BIOLOGICAL CHEMISTRY

Vol. 255, No. 22. Issue of November 25, pp. 10710-10716. 1980 Printed in U.S.A.

ADP-Ribosylation of Elongation Factor2 by Diphtheria Toxin NMR SPECTRA AND PROPOSED STRUCTURES OF RIBOSYL-DIPHTHAMIDE AND ITS HYDROLYSIS PRODUCTS* (Received for publication, April 10,19801

Brian G. Van Ness$, James Bryant Howard, and James W.Bo&y From the Departmentof Biochemistry, University of Minnesota, Minneapolis, Minnesota 55455

7 7

NMR spectralanalysis of thenovel amino acid, CH,-CH(NH,)-COOH diphthamide, in elongation factor 2 which is ADP-ribosylated by diphtheria toxin suggests that it is 2-[3carboxyamido-3-(trimethylammonio)propyl]histidine. Ribosyl-diphthamide was prepared by enzymatic hydrolysis of ADP-ribosyl-elongation factor2 and three compounds were produced by its chemical hydrolysis I 1 J. W. (1980) (Van Ness, B. G., Howard, J. B., and Bodley, C(H)+N-CH, J. BioL Chem 266,10717-10720). ProtonNMR spectrosI I copy in ‘HzO of diphthine demonstrated the elements 0.c CH, I of histidine minus the carbon 2 proton plus 14 addiNH, tional nonexchangeable protons. These protons were FIG. 1. Proposed structure of diphthamide, 2-[3-carboxyattributed to an extensive modificationat carbon 2 of the imidazole ring. Proton NMR spectroscopy in ‘HzO amido-3-(trimethyIaonio)propyI]histidine. of ribosyl-diphthamide showed only those protons seen of all four of these compounds and the naturalabundance I3Cin diphthine plus those expected of a ribofuranosylmoiety. Chemical shift dependenceon pH was demon- NMR spectrum of ribosyl-diphthamide.These data in comstratedforthehistidine-derivedprotons as well as bination with the previously described properties of these several protons of the modifying side chain. Natural compounds (1) lead us to propose that diphthamide possesses abundancecarbon13 NMR spectroscopyofribosylthe structure shown in Fig. 1. diphthamide also showed the elements of histidine and to the ribose plus7 additional carbon atoms attributed EXPERIMENTALPROCEDURES on the NMR modification of the imidazole ring. Based Preparation of Samples for NMR Spectroscopy-Ribosylspectral properties of the anomeric proton of ribosyl- diphthamide and its three hydrolysis products were isolated as dediphthamide we propose that the ribose is linked to one scribed in the preceding paper (1).Samples for NMR spectroscopy of the nitrogens of the histidine imidazole ring. were dried from the SP-Sephadex elution buffer (l),then twice from H20, once from 50 mM HCl, and twice from ‘Hz0 (100.0% Aldrich Chemical Co.). Dried samples were dissolved in 0.35 ml of 100.0% ‘HzO and transferred to 5-mm glass NMR tubes. The pH of the In the preceding paper (1) we described the isolation and samples was adjusted as needed using ‘HC1 or NaO’H. The pH values in ‘H20 are uncorrected for deuterium effects. chemical properties of a novel ribosyl-amino acid, ribosylTris (tris(hydroxymethy1)aminomethane)in amounts not detectadiphthamide, obtained by enzymaticdigestion of EF-2’ followble on amino acid analysis was occasionally encountered as a contaming its ADP-ribosylationby diphtheria toxin. We also isolated inant of ribosyl-diphthamide preparations. Tris, which was seen as a and characterized the final product of acidhydrolysis of singlet at 63.7 in the proton NMR spectrum, arose either as a carryribosyl-diphthamide,previously known as amino acid X (3), over from the earlier peptjde purification or was contained in the for which we proposed the trivial name, diphthine. Partial enzymes used for peptide digestion. Unfortunately, Tris, like ribosylhydrolysis of ribosyl-diphthamide yielded two new com- diphthamide, was retained by dihydroxyboryl-substituted cellulose pounds, ribosyl-diphthine and diphthamide, which we also and the two compounds could not be completely separated by gel filtration or chromatography on the amino acid analyzer. These isolated and characterized. We concluded that thelatter com- contaminated samples were therefore converted to diphthine which pound is the form of the amino acid in EF-2 which is acted was readily separated from Tris on the amino acid analyzer. upon by diphtheria toxin. NMR Parameters-Proton NMR spectra were obtained a t 360 In the present report we describe the proton NMR spectra M H z on a Nicolet 360 NMR spectrometeror a t 270 MHz on a Bruker HX-270 NMR spectrometer equipped with a Nicolet Technologies This work was supported inpart by National Institutes of Health 1180 computer/Fourier transform system and computer-controlled Grants GM-21359, GM23276,and GM26832 to J.W.B. and HL-24505 homonuclear decoupling accessory. Spectra obtained on the Nicolet to J.B.H. Support for the 360 MHz NMR spectrometer was provided 360 MHz spectrometer were accumulated at 23°C (64scans) using an by National Institutes of Health Grant RR01077 to John Markley. acquisition time of 2.0 s, a 45O deflection angle, and a 16K Fourier This is Paper XXVII in the series “Studies on Translocation.” The transform. Spectra obtained on the Bruker 270 MHz spectrometer preceding paper in the series is Ref. 1. A preliminary account of this were accumulated a t 23°C ( 5 0 0 to loo0 scans) using an acquisition work has been presented (2). The costs of publication of this article time of 1.4 s, a deflection angle of 45’, and a 16K Fourier transform. were defrayed in part by the payment of page charges. This article Sodium 3-trimethylsilylpropionate in a coaxial capillary was used as must therefore be hereby marked “aduertisement” in accordance an external standard. Decoupling experiments were performed by irradiating at the center of the desired resonance with sufficient power with 18 U.S.C. Section 1734 solely to indicate this fact. $ Present address, Institute for Cancer Research, Fox Chase, Phil- to eliminate it from the spectrum while minimizinginterference with neighboring resonances. adelphia, PA 19111. ’ The abbreviation used is: EF-2, elongation factor 2. Natural abundance I3C NMR spectra were obtained a t 25.2 MHz

10710

NMR Spectra and Proposed Structure of Ribosyl-diphthamide

10711

on a Varian XL-100 spectrometer. The sample concentration was about 50 mM in 100 pl of 'H20, pH 2. Spectra were accumulated at ambient temperature, with an acquisition timeof 0.73 s and a total of 80,000 to 100,000 transients. Sodium 3-trimethykilylpropionate in a as an external standard. coaxial capillary was used

single, apparently unique structure, which is consistent with the 13C NMR spectrum of this compound (described below) and its otherknown properties. The spectrum in the upper right of Fig. 2 shows the entire spectrum expanded so that all resonances are on scale. The RESULTS spectrum in the lowerpartof Fig. 2 is expanded so aa to reveal Criteria of NMR Spectral Purity-The purification of ri- its detail. The identification of the resonances in this spectrum bosyl-diphthamide described inthe accompanying paper (l), corresponds to thenumbering in the structure. The proposed was followed byproton NMR spectroscopyuntil the spectrum structure in the upper left ofFig. 2 was deduced from this was unchanged by further purification. These further purifi- spectrum and others as described below. The Proton NMR Spectrum of Dbhthine-The proton cation steps included paper chromatography, Bio-GelP-2 gel permeation chromatography, and chromatography on the NMR spectrum of ribosyl-diphthamide integrated to approxamino acid analyzer (4). In addition, the three compounds imately 24 nonexchangeable protons. Deternlination of the produced by hydrolysis of ribosyl-diphthamide were further proton NMR spectrum of diphthine, which contained only purifiedon the basis of -theirchromatographic properties about 18 nonexchangeable protons, simplified the structural is whichdifferedfrom those of the parent compound. Any problem. The 270 MHz spectrum of diphthine in 2Hz0 contaminants whichco-purifiedwithribosyl-diphthamide shown inthe lower panel of Fig. 3. Two conspicuousfeatures all samples. These are thesinglet should have been removed inthis way and none were found. of this spectrum were seen in Finally, one must be able to account for those carbon and seen at approximately 67.4 and the intense singlet at 63.3. The proton resonances seen in all four compounds in terms of a intense resonance at 63.3, within experimental uncertainty, set of related structures. These structures should account for invariably integrated to an area 9 times that of the resonance both the chemical properties and interconversions of the at 67.4. The spectrum also contained additional upfield resocompounds which were documented in the preceding com- nances which integrated to approximately 8 protons relative munication (1). to theresonance at 67.4. The resonance at 67.4 is in a positioncharacteristic of that The Proton NMR Spectrum of Ribosyl-diphthamide-The 360 MHz proton NMR spectrum of ribosyl-diphthamide in of the carbon 5 proton of the amino acid histidine.No other 'Hz0 and its proposed structure shown in Fig. 2 satisfy the amino acid, amongthe 20 common ones, exhibits a resonance criteria outlined above. Further purification of the compound in this region. These facts prompted us to compare the specdid not remove any of the resonances seen here. The reso- trum of diphthine with that of histidine. The spectrum of nances of the ribose protons have been identifiedas described histidine and its structure are seen in the upper portion of Fig. 3. below and only these protons are deleted upon hydrolysis. Histidine contains 5 nonexchangeable protons and 4 of these Moreover, all of these resonances are accommodated by a

-COOH

H-5

R- 1

1

"

"

l

"

"

l

8.0 6.5 7.0 7.5

"

"

l

"

'

I

'

"

'

I

'

6.0 4.5 5.0 5.5

"

'

I

'

'

"

i l

"

"

c-2

2, l

'

'

"

4.0

~

'

'

"

3.0 3.5

l

'

'

'

'

~

'

"

'

2.5

1

2.0

8 (ppm) FIG. 2. Proposed structure and proton NMR spectrum of ribosyl-diphthamide. The N M R spectrum was obtained in 'Hz0 at 360 MHz with a sample pH of 2. The proton resonances proposedto derive from histidine are designated withan H,those from ribose with an R, and those fromthe imidazole modification witha C. The inset (upper right) shows the same spectrum at a lower amplification.

NMR Spectra and

10712

Proposed Structure of Ribosyl-diphthamide the substitution(s) of the imidazole ring. The H-a and H-P protons of the two compoundscorresponded closely at all pH values. Thus, 4 protons seen in the NMR spectra of both diphthine and ribosyl-diphthamide have splitting and coupling patterns as well as titration behavior which closely correlate with 4 of the 5 protons in histidine. The 5thnonexchangeable proton of histidine, that on carbon 2, is missing from these spectra. Identification of the Ribose Protons in the Ribosyldiphthamide Spectrum-Careful correlation of the NMR spectra of numerous preparations revealed that, except as noted below, the spectrum of ribosyl-diphthamide differed from that of diphthine only by the presence of 6 additional protons. This is the number expected for the ribose residue. This observation leads to the conclusion that the only nonexchangeable protons removed bycomplete acid hydrolysis of ribosyl-diphthamide are those of ribose. The assignment of these six resonances to theribose protons was based on spin decoupling experiments. These experiments beganwith the assumption that the doublet seen at 66.23 (Table I) was derived from the anomeric proton. This assumption was based on both the chemical shift and multiplicTABLEI Chemical shift, integration, and multiplicity of proton resonances observed in 'H NMR of ribosyl-diphthamide 6"

,

9.0

1

1

8.0

1

I

7.0

I

I

6.0

I

I

5.0

I

I

4.0

,

I

3.0

1

,

2.0

6 ( ppm)

R-2 R-3 (upFIG. 3. The 270"Hz proton NMR spectra of histidine R-4 per) and diphthine (lower) in aHzO at pH. 2.3. The structure of histidine is shownin the upper left. The spectral assignments for histidine were those of Ref. 9. The small singletseen in the diphthine spectrum at 63.8 (immediately downfield ofthe doublet of doublets) is a contaminant, possiblyTris.

generate resonances which have counterparts in the diphthine spectrum. Absent from the diphthine spectrum is a resonance at 88.8 which would correspond to the resonance from the proton on carbon 2 of histidine. While the resonance at 67.4 was observed in all preparations of ribosyl-diphthamide and compounds derived by its hydrolysis, no resonance was ever observed at 88.8 in any of these preparations. These observations suggest that both diphthine and ribosyldiphthamide are derivatives of histidine in whicha substituent has replaced the proton on carbon 2. Several additional lines of evidence support these conclusions. SpectralCharacteristics of theHistidine-derived Protons-The integration and multiplicities of the histidine-derived protons observed in the ribosyl-diphthamide spectrum are recorded in Table I. The spin-decoupling experiments recorded in Table I1 show that the protons designated H-a: and H-fi are coupled in a manner consistent with an ABX system. The data in Fig. 4 compare the pH dependence of the chemical shifts of these protons with their counterparts in histidine. The symbols in Fig. 4 represent the observed shifts of the protons in ribosyl-diphthamide while the solid lines represent the chemical shifts of the protons of histidine observed in a separate experiment. A t no pH value was there a resonance observed in ribosyl-diphthamide which corresponded to the carbon 2 proton of histidine. The downfield resonance, however, did correspond closely with that of the histidine carbon 5 proton. The slight downfield shift of this proton compared to thatin histidine could have resulted from

7.55 4.19 3.37 R6.23 4.67 4.33 4.51 3.80 3.25 3.09, 2.59 4.08 3.30

Assignmentb

Integration'

Multiplicityd

H-5 H-a

1

S

1

H-P

2

1

1

t d d t t 9 dq m, m m dd

1

1

1 2 1, 1 2 1

R-5

c-1

c-2 c-3 C-7,8, 9

9

S

The chemical shift represents the center of the resonance. Assignments are those enumerated in Fig. 2. Values shown are relative integrations roundedto nearest whole number. dThe m indicates that the multiplicity was too complex to enumerate. a

1 I

i

I

pH

FIG. 4. Plot of chemical shifb uersuu pH meter reading for histidine and ribosyl-diphthamide. The lines represent the titration of the resonances from histidine as indicated. The data points represent the values obtained for H-a (O),H-@(a)and , H-5(A) of ribosyl-diphthamide.

10713

R

8.0

25

70

6.5

6.0

5.5

5.0

4.5

40

3.0

3.5

2.5

2.0

6 (ppm) FIG. 5. 360-MHz proton NMR spectra of ribosyl-diphthamide (lower) with spin decoupling of ribose protons R-1(middle)and R-2 (upper).The bur indicates the region decoupled; the arrow indicates the affected signal(s).

ity of this resonance. The resonance from this proton would be a doublet, through splitting by the single proton on carbon 2, if the glycosyl linkage was with an atom that did not possess a proton. Moreover, the chemical shift of this resonance is comparable to that of the anomeric proton of adenosine, for example (5). A representative set of decoupling experiments is shown in Fig. 5. In this experiment the only spectral change observed by decoupling the resonance at 66.23 is the collapse of the triplet, designated by the arrow in the middle spectrum, to a doublet. This observation allowed the assignment of this resonance to proton R-2. Irradiation of the R-2 proton resonance (Fig. 5, upper) collapsed the doublet at 66.23 as well as triplet at 64.33. The latterresonance therefore, is from R-4 proton R-3. In this manner all 6 ribose protons were unambiguously assigned (Table 11). Assignment of Additional Protons-A total of 10 protons in the spectrum of ribosyl-diphthamide were assigned as derived from either histidine or ribose. Fourteen remain to be assigned, 9 of which occur as a singlet at 63.30. All of these resonances were seen in the diphthine spectrum. The large singlet at 63.30 has almost exactly the same chemical shift as the resonance of the 9 protons of the quaternary trimethylammonio group of choline (6). On this basis we propose that the modifying side chain in both ribosyldiphthamide and diphthine contains a trimethylammonio function. The positive charge on this group together with that of the imidazole ring wouldaccount for the charge characteristics previously observed for ribosyl-diphthamide (1). Sequential spin decoupling (shownin Fig. 6 and summarized in Table 11) was used to define the coupling pattern of the remaining 5 protons. These resonances represent a single

TABLEI1 Sequential spin decoupling of the protons of ribosyl-diphthamide Spindecoupling was performedon the Nicolet 360 MHzNMR spectrometer. decoupled F'roton(s) Protons affected 6"

Ass' ment @;

k:z

Multiplicity'

H-P t+s H-a d+s R-2 t+d R-1 d+s R-3 t+d 4.33 R-3 R-2 t+ d R-4 q+t 4.51 R-3 t+ d R-5 dq+ q 3.80 R-5 R-4 q+ d 3.09, 3.25 c-1 c-2 m + m' 2.59 c-2 c-1 m + m' 4.08 c-3 dd+ s 4.08 c-3 2.59 c-2 m- m' The chemical shift represents the center of the resonance. Assignments are those enumerated in Fig. 2. e Multiplicity effect is shown as multiplicity before and after proton decoupling.The m indicates that the multiplicity was too complex to enumerate. The m' indicates reduced multiplicity which remained too complex to enumerate. 4.19 3.37 6.23 4.67

H-a H-P R- 1 R-2

6"

3.37 4.19 4.67 6.23 4.33 4.67 4.51 4.33 3.80 4.51 2.59 3.09, 3.25

linkage group since they are not altered by irradiation of any of the previously assignedresonances. The simplest relationship wasbetween the resonances at 64.08 and 62.59. The doublet of doublets (64.08) was collapsed to a singlet by spin decoupling the resonance at 62.59. Conversely, decouplingthe doublet of doublets (84.08) altered the multiplet at 62.59 but

10714

NMR Spectra and Proposed Structure of Ribosyl-diphthamide

methylammonio group, C-3 has been assigned to the site of attachment of these two functions. Thus, the aliphatic side chain is a 3-carboxyamido-3-(trimethy1ammonio)propylgroup attached via carbon 1 to carbon 2 of the imidazole ring. Spectra of the Hydrolysis Intermediates-Proton NMR spectra (datanot shown) were obtained for the acid (diphthamide) and alkaline (ribosyl-diphthine) hydrolysis intermediates. In accord with the chemical analysis of these compounds (l),the ribose protons were observed in the spectrum of ribosyl-diphthine but were absent from the spectrum of diphthamide. In all but one respect the spectrum of ribosyl-diphthine was identical to thatof ribosyl-diphthamide (Fig. 2)and the spectrum of diphthamide was identical to that of diphthine (Fig. 3). In ribosyl-diphthamide and diphthamide the chemical shift of the resonance designated C-3 was independent of pH (Fig. 7). In contrast, the C-3 resonance in the spectra of ribosyldiphthine and diphthine shifted upfield at pH values above 2 (Fig. 7). These observations are in accord with the view that the C-3 proton is not adjacent to anionizable group in ribosyldiphthamide and diphthamide but that in ribosyl-diphthine and diphthine the proton is adjacent to an ionizable group with a low pK,. Such behavior is predicted by the proposed structure and the hydrolysis of the amide and is consistent with the inferred charge characteristics of the compounds (1). The Nature of the Ribosyl-linkage-Previously, we have shown (1) that the sugar is not linked via the carboxyamide nitrogen. Given the proposed structure of diphthamide, the most likely site of attachment is to one of the imidazole ring nitrogens as shown in Fig. 2. The chemical shift of the reso1 " ~ ~ 1 1 1 1 ' 1 ~ 1 " 1 1 ' nance of the anomeric proton in ribosyl-diphthamide (R-1) is 4.0 3.5 30 2.5 in accord with this possibility since it is very near that seen 6 (ppm) FIG. 6. 360-MHz proton NMR spectra of ribosyl-diphthamide for the anomeric proton of adenosine (5) and considerably (A) with spin decoupling of protons C-3 (B), C-2 (0,and C-1 downfield of that seen for most other N - and 0-glycosides. (0).The bar indicates the region decoupled; the arrow indicates the Furthermore, the chemical shift of the R-1 resonance changes affected signal(s). Decoupling of the entire C-1 resonance proved in the region of the ionization of the imidazole ring (Fig. 7). impractical. Results comparable to those shown in Panel D were On the basis of these observations and because attachment to obtained when the downfield portion of C-1 was decoupled (data not the imidazoleN-3 is sterically slightly less favorable, we shown). propose that the glycosyl linkage is to imidazole N-1. The coupling constant (5.1 Hz) for the anomeric proton (R-1) is did not resolve it into an identifiable pattern. Therefore, the comparable to thatof the corresponding proton in adenosine single proton at 64.08 (designated C-3 in Fig. 2) is split by two which is of the p configuration. Yet, the alternative a confignonidentical protons both with resonances centered at 62.59. uration in ribosyl-diphthamide cannot be ruled out. Decoupling of the 62.59 resonance (designated C-2 in Fig. 2) Carbon 13 NMR Spectrum of Ribosyl-diphthamide-We also simplifiedthe multiplet at 83.09 and 63.25 (designated C- were unwilling to commit a large fraction of our supply of 1 in Fig. 2). Therefore, these 5 protons constitute a single ribosyl-diphthamide to hydrolysis. Therefore, the limited linkage group in the sequence C-1(2 protons), C-2 (2 protons), amounts of diphthine and the hydrolysis intermediates availC-3 (1proton). able to us were insufficient for 13CNMR studies. Indeed, the Because the C-1, C-2, and C-3 protons are characteristic of amount of pure ribosyl-diphthamide which remained after our an aliphatic chain which is presumably linked to theimidazole carbon 2 of histidine, the resonances of the proton(s) on the carbon adjoining the ring should shift concomitant with ionization of the ring. Fig. 7 shows the pH titration of the c-1 resonance. This multiplet shifted at a pH characteristic of an imidazole ionization.2The resonances designated C-2 and C-3 were invariant in this region. The resonance of the proton designated C-3 had a chemical shift characteristic of the aproton of an a-amino acid at low pH (cf Figs. 4 and 7). Because we established in the preceeding paper (1) that diphthamide probably contains a carboxyamido function and from the NMR analysis that the structure contains a triThese titration curves were obtained on the 270-MHz spectrometer. This instrument did not resolve the C-1 resonances to the degree seen with the 360-MHz instrument (e& Figs. 2 and 3). As a consequence, the pH-dependent shift was measured at theapparent center of the C-1 multiplet. At high pH values it appeared that both resonances in this multiplet had moved.

2 ~ 1 2 3 4 5 6 7 8 9 1 0 1 1 OH

FIG. 7. Plot of chemical shift versus pH meter reading forR1 (0)and C-1 (0) in ribosyl-diphthamide, and C-3 in diphthine

(m), ribosyl-diphthamide (El), diphthamide (X), and ribosyldiphthine (A).

-

NMR Swectra and Proposed Structure of Ribosvl-diwhthamide c-0 ?-I

10715

imidazole

FIG. 8. The 26.2 MHz "C NMR spectrum of ribosyl-diphthamide. The assignments shown above are based on comparison with standards.

other analyses was only minimallysufficient for such studies. As a result of this and the cost of preparing even larger amounts of material, I3C NMR studies were limited to a test of the structural predictions which were based on the proton NMR spectral studies. The proposed structure shown in Fig.2 contains 18 carbon atoms which, because of the equivalence of the three trimethylammonio carbons, should give 16different carbon resonances. Six of these resonances should have counterparts in the histidine spectrum and five should have counterparts in ribofuranose. Five more should result from the proposed side chain. The observed I 3 C NMR spectrum contained 16 resonances and is shown in Fig. 8. Five resonances were in the location (670-90) of a ribofuranose. Six resonances were also seen in the positions of the predicted histidine constituents. The carboxyl group of the histidine moiety was presumed to generate one of the two resonances seen downfield (6180-190). The a-carbon is presumed to generate one of the two resonances in the 860-70 region. Three upfield resonances were seen in the region expected of methylene carbons (620-30) and one of these is presumed to emanate from the histidine P-carbon. Three resonances were observed in the region expected of aromatic carbons (6120-150). Of these, the two upfield resonances correspond approximately to thechemical shift of the resonances of carbon 4 and carbon 5 of the imidazole ringof histidine. The third signal farther downfield is deshielded and reduced in intensity compared to thecarbon 2 of histidine. This is the expected result of a substitution at this position. The predicted five resonances remain. Two of these in the methylene region are presumed to correspond to carbon 1 and carbon 2 of the modifying side chain. One of the two resonances at 660-70 and one of the two resonances at 6170-180 presumably correspond to carbon 3 and carbon 4 of the side chain. Finally, the intense resonance at 653 is between the positions seen for the trimethylammonio carbons of choline and acetyl choline ( 7 ) . Thus, the 13C NMR spectrum of ribosyl-diphthamide is entirely consistent with the proposed structure.

DISCUSSION

The structure which we have proposed for ribosyldiphthamide (Fig. 2) apparently represents a unique explanation for the properties which we have observed of the compound and its hydrolysis products. The proposed structure derives primarily from proton NMR spectral analysis and appears to uniquely account for the nonexchangeable protons observed in this spectrum. Also essential to this proposal were interpreted as the observations (1)which could be most easily resulting from the presence of a carboxyamido group in the compound. The presence of imidazole and ammonio groups in conjunction with the base-labile carboxyamide accounts for the ionic properties of ribosyl-diphthamide and its hydrolysis products (1).Finally, the essential elements of this proposal were supported by I3C NMR spectral analysis. Attempts to employ several other potential methods of testing this structural proposal have thus far been unsuccessful. For example, to date we have been unable to prepare derivatives of any of these compounds with sufficient volatility for mass spectral analysis, nor have we been able to produce a crystal suitable for x-ray crystallography. Also, while elemental analysis would be useful in testing the structure proposed here, we have not been willing to commit the large percentage of our material which would be necessaryfor this destructive method of analysis. Ultimately, confirmation of this structure will require chemical synthesis. There arenumerous unusual amino acids known to arise by post-translational mechanisms (8).Most of these result from rather simple modifications of the 20 amino acids specitled by the amino acid code. Diphthamide, which occurs in a unique position in EF-2, is the most complex exampleof site-specific post-translational modification known to date. Almost certainly histidine is the ribosomal precursor of diphthamide. Several possible postribosomal routes exist for the modifying side chain on the carbon 2 of the imidazole ring with the amino acid methionine or perhaps glutamic acid as the most likely. The trimethylammonio group probably arises from Sadenosylmethionine. Including the amidation of the carboxyl

10716

NMR Spectra and Proposed Structure of Ribosyl-diphtharnide

group it seems likely that at least three enzymatic reactions would be required to affect this modification. The present observations concerning the chemistry of the diphtheria toxin modificationsite in EF-2 fit very nicely with recent frndings of Moehring et al. (10) concerning the biology of diptheria toxin resistance. These workers have identified two types of toxin-resistant mammalian cell mutants which exhibit in vitro resistance. One type has properties which suggest that itbears a mutation in the codon of the EF-2 gene which specifies the amino acid, presumably histidine, which is post-translationally modified to generate the site of toxin action. The second type of mutanthas properties which suggest that it bears a defect in an enzyme which carries out the post-translational modification of the original amino acid in EF-2 to generate the toxin modification site. This latter type of mutant has been subdivided into three complementation groups suggesting that three or morepost-translational enzymes are involved. Moehring et al. (10) have also shown that this modification to toxin sensitivity can take place in soluble cell extracts thus offering the promise of elucidating the enzymatic reactions which are involved in diphthamide synthesis. On the surface it would seem that diphthamide must perform an essential role in the function of EF-2. First of all, it appears to be present in EF-2 of all eukaryotes (11). This is particularly significant because its biosynthesis is likely, as noted above, to be complex and the residue is not known to occur in other proteins. In addition, the amino acid sequence in EF-2 surrounding this residue is highly conserved indivergent species (11).Finally, the modification of this residue by toxin inactivates EF-2, although the specific roleof the residue in the function of the protein remains obscure. In view of this of Moehring it is somewhat surprising to find that the mutants et al. (lo), which appear to lack this residue, do not seem defective in protein synthesis. However, recent experiments in our laboratory3indicate that toxin does not ADP-ribosylate the tryptic peptide from EF-2 which contains diphthamide so that it is possible that toxin resistance is more complicated than the simple absence of diphthamide. It wil be of interest to determine by chemical analysis what residue(s) replaces diphthamide in the EF-2 of these toxin-resistant mutants. Although a number of proteins are known to serve as acceptors of ADP-ribose in reactions comparable to that catalyzed by diphtheria toxin, in no prior case has the specific amino acid acceptor been identified. Ribosyl-imidazole linkages have not been previously observed in proteins (12), but an interesting precedent for the ribosyl-imidazole linkage proposed here is provided by the work of Alvisatos and coworkers (13-16). In studying a bovine spleen NAD+ glycohydrolase these investigators observed that it could not only transfer the ADP-ribosyl-moietyto water but also to a variety of imidazole derivatives (13, 14) including free histidine (15). Although the metabolic signfkance of this reaction, referred B. G . Van Ness and J. W. Bodley, manuscript in preparation.

to as imidazolysis (16),is unclear, its various products do provide a precedent for this type of linkage. The ribosyllinkage in these structures was seen to be very resistant to acid hydrolysis (14).The ribosyl-linkage in ribosyl-diphthamide was resistant to both acid and alkaline hydrolysis (I),but because the rate of hydrolysis of this type of compound is known to be markedly influenced by the nature of the imidazole substituents (15), detailed comparisons are not possible. Further work w l l ibe required to determine how diphthamide arises in EF-2, what role this unusual residue plays in the function of the protein, and whether it or related compounds occur in other proteins. Acknowledgments-We are extremely grateful to Barbara Barrowclough for her skill and diligence in preparing the large amount of ribosyl-diphthamide required for this study. We thank Robert Thrift for his assistance in operating the 260-MHz spectrometer at the Freshwater Biology Institute, John Wood for making this instrument available to us, and Robert Riddle of the Department of Chemistry for his assistance in obtaining the 13C NMR spectra. We are also grateful to John Markley for making the Purdue University Biochemical Magnetic Resonance Laboratory 360-MHz spectrometer available to us and to him and Milo Westler for assistance in its operation. Gary Gray and Harry Hogenkamp provided valuable assistance in interpreting the NMR spectra. We are indebted to Waldo E. Cohn for valuable suggestions regarding systematic nomenclature for the proposed structures and Thomas Moehring for making his experimental results available to us prior to publication. REFERENCES 1. Van Ness, B. G., Howard, J. B., and Bodley, J. W. (1980) J. Biol. Chem. 255,10717-10720 2. Bodley, J. W., Van Ness, B.G., and Howard, J. B. (1979) in Nouel ADP-ribosylations of Enzymes and Proteins (Smulson, M., and Sugimura, T., eds) pp. 413-423, Elsevier-North Holland, New York in press 3. Robinson, E. A,, Henriksen, O.,and Maxwell, E.S. (1974) J. Biol. Chem. 249,5088-5093 4. Simpson, R. J., Neuberger, M. R., and Lin, T.-Y. (1976) J.Bwl. Chem. 251,1936-1940 5. Pouchent, C. J., and Campbell, J. R. (1974) The AZdrich Library of NMR Spectra 8,119C 6. Pouchent, C. J., and Campbell, J. R. (1974) The AZdrich Library of

NMR Spectra 2,86C.

7. Simeral, L., and Maciel, G. E. (1974) Org. Magnetic Res. 6,226232 8. Uy, R.,and Wold, F. (1977) Science 198,890-896 9. McDonald, C.C., and Phillips, W. D. (1963) J. Am. Chem. SOC. 85,37363742 10. Moehring, J . M., Moehring, T. J., and Danley, D. E. (1980) Proc. Natl. Acad.Sci. U. S.A. 77,1010-1014 11. Brown, B.A., and Bodley, J. W. (1979) FEBS Lett. 103,253-255 12. Spiro, R. G . (1970) Annu. Rev. Bwchem. 39,599-638 13. Alivisatos, S . G . A., and Woolley, D. W. (1956) J. Bwl. Chem. 221,651-663 A., Ungar, F., Lukacs, L., and LaMantia, L. (1960) 14. Alivisatos, S. G. J. Bwl. Chem. 235,1742-1750 15. Alivisatos, S . G. A,, and LaMantia, L. (1960) Biochem. Biophys. Res. Commun. 2,164-168 16. Alivisatos, S. G. A., LaMantia, L., Ungar, F., and Savich, B. (1958) Biochim. Bwphys. Acta 30,660-661