Fragmentation mechanisms of protonated actinomycins and their use in structural determination of unknown analogues

JOURNAL OF MASS SPECTROMETRY, VOL. 30, 1111-1125 (1995)

Fragment ation Mechanisms of Prot onated Actinomycins and Their Use in Structural Determination of Unknown Analoguest

Darren Thomas,$ Michael Morris,$ Jonathan M. Curtis and Robert K. Boyd Institute for Marine Biosciences, National Research Council, 141 1 Oxford Street, Halifax, Nova Scotia, B3H 321, Canada

A combination of mass shifts arising from known structural variations of neutral precursors, and accurate m a s measurements of key fragment ions, has led to a proposed systematization of the lowenergy fragmentation mechanisms for MH' ions of actinomycins. These rationalizations are used in a predictive manner to interpret fragment ion spectra, of unknown impurities in actinomycin standards, in terms of structure. Both the strengths and limitations of this approach are emphasized. This work has uncovered actinomycin variants in which one of the threonine residues is substituted by serine, a structural variation which does not appear to have been reported previously.

INTRODUCTION

The actinomycins are brightly coloured antibiotic metabolites produced by various species of Strep- tomyces. Detailed reviews of the origins, chemistry and biological action of the actinomycins have been published.'-3 Although the actinomycins are amongst the most effective anti-tumour agents known, their clini- cal use has been limited owing to their extreme toxicity. Considerable work has been devoted to production of analogues of the naturally occurring compounds, with the hope of improving the chemotherapeutic index (ratio of maximum tolerable dose to minimum curative dose). Chemically the actinomycins consist of a chromophore moiety (3-amino-1,8-dimethyl-2-phenoxazone-4, 5-dicarboxylic acid) which carries, as amides at its two carboxyl groups, two pentapeptides each of which forms a cyclodepsipeptide ring structure (Fig. 1). The amino acids linked by amide bonds to the carboxyl groups of the chromophore moiety are almost always L-threonine in those compounds characterized thus far.'-3 The p- hydroxyl group of each Thr residue is always lactonized with the carboxyl group of the fifth amino acid in the same pentapeptide side-chain. A listing of 33 documented actinomycin variants was published in 198K4 Some of these variants correspond to substitutions on the aromatic ring in the chromophore moiety, but most arise from substitution of one or more amino acids in some suitable reference compound.

In the present work it will be convenient to use actinomycin D (Fig. 1) as the reference structure, since this compound has two identical pentapeptide residues, viz. L-Thr-D-Val-L-Pro-Sar-L-MeVal (in this work, methylated residues are taken to imply N-methyl deriv-

t NRCC No. 38088. $ Present address: Hazelton Europe, Harrogate, North Yorkshire

5 Present address: Fisons Instruments, Biotech MS, Altrincham, HG3 IPY, UK.

Cheshire WA14 5R2, UK.

atives only). In the comDounds documented." the Sar residues in position 4 are'almost invariant although one case is known in which Gly replaces Sar. The Thr residues are also almost invariant but exceptions, in which the methyl substituent on the P-carbon of Thr is replaced by hydroxymethyl, are known (the actinomycin Z family4). Other common substitutions on the actinomycin structure are D-allo-isoleucine (DaIle) for D-Val in position 2, ~ - H y p (y-hydroxyproline) or L- Oxopro (y-ketoproline) or L-Pip (pipecolic acid, i.e. 2- piperidinecarboxylic acid) or Sar for L-Pro in position 3, and L-MeIle for L-MeVal in position 5. Variants which contain substituents in the aromatic ring (other than the so-called a-peptide) of the chromophore are synthesized products, while the quinoid ring (carries the P-peptide) has been invariant thus far.

Mass spectrometry played only a minor role in the early work on structure elucidation of the actinomycins, using ionization by electron impact,' field desorption,6 plasma desorption' and desorption chemical ionization.' However, the introduction of fast atom bombard- ment (FAB) ionization has permitted mass spectrometry to play a much more important role, and two examples of the application of FAB to the actinomycins were p ~ b l i s h e d ~ . ' ~ in 1988. These investigations used techniques of tandem mass spectrometry to investigate the fragmentation pathways of protonated actinomycins. More recently, the introduction of electrospray ionization (ESI) has greatly facilitated the on-line coupling of high-performance liquid chromatography (HPLC) to mass spectrometry, and the present investigation exploits the new HPLC/ESI-MS technology for investigations of the actinomycins.

The mass spectrometry of other cyclodepsipeptides, i.e. cyclic peptides in which the backbone contains ester (lactone) linkages in addition to amides, is well established, and in fact some of the earliest applications of mass spectrometry to peptides involved compounds of this type."~lZ These early investigations employing electron impact ionization provided useful information, but the extent of fragmentation of the molecular ions

CCC 1076-5 1 74/95/08 1 1 1 1 - 15 0 Crown copyright (Canada)

Received 7 March I994 Accepted 5 April 1995

1112 D. THOMAS ET AL.

Actinomycin I Actinomycin C, Actinomycin C,

Figure 1. Structures of actinomycin standa~rds.~ All amino acid residues are the L-forms unless noted otherwise (o-Ale, D-dlo-isoleucine, is one of the diastereoisomers of He). Structure I refers to protonated actinomycin D.

was high and the discrimination amongst possible fragmentation routes was low. The use d softer ionization techniques for depsipeptide analysis was introduced in 1979 by Das et who employed chemical ionization (both methane and isobutane) to study some simple monocyclic depsipeptides. This work will be referred to in the ensuing discussion, and was innportant because it established that the ester linkage(s) in a depsipeptide is (are) those preferentially cleaved in protonated molecules MH'. This general conclusion was later confirmedi4 for MH' ions produced from simple cyclodepsipeptides by FAB ionization. A comparison of high-energy collision-induced dissociation (CID) spectra of each of the M", MH' and [M + Me]+ ions (Me = Na, K) of the cyclodepsipeptide valinomycin, formed by various ionization techniques, has been published re~ent1y.I~

The purpose of the present work was to employ ESI-MS and MS/MS, coupled to HPLC when necessary, to elucidate further the fragmentation pathways of MH + ions of actinomycin standards. The understand- ing thus gained is then used to identify minor impurities (other actinomycins) observed in the standard preparations. In this regard, it is emphasized here that these investigations of unknowns do not claim to be de nouo structural determinations. For example, all such investigations of cyclic peptides by mass spectrometry are plagued by ambiguities surrounding the choice between a proposed sequence and its refro sequence. This problem was first pointed out by Tomer et ~ l . , ' ~ and was recently emphasized in an excellent review by

Eckart.I7 The present work avoids this problem by using an approach first described by Barber et af.I8 for cyclic peptides. This approach is based on an assumption that the unknown compounds are related to one or more standard compounds of known structures via substitutions of amino acid residues and has been used, for example, to identify 18 previously unknown tyrocidines in the tyrothricin complex from Bacillus breuis.Ig In the present work this general principle is adapted to the case of the actinomycins.

~~

EXPERIMENTAL

The actinomycin samples were purchased from Sigma Chemical (St Louis, MO, USA) and were used as received. The structures of these compounds are shown in Fig. 1, but it should be noted that actinomycins C2 and C, were supplied as a mixture. Standard solutions of these commercial preparations were prepared in 50 : 50 acetonitrile-water, at concentrations of 1 mg ml-' and acidified with acetic acid.

Characterization of these solutions by HPLC with UV detection (254 nm) was achieved using a HP 1090 Series I1 liquid chromatography system, equipped with a diode-array detector and HP 3365 ChemStation soft- ware installed on a HP 486/33U computer (Hewlett- Packard, Palo Alto, CA, USA). A 5 pl volume of each standard solution was injected on to a Vydac 218TP52 CI8 reversed phase column (25 cmx2.1 mm i.d., 5 pm

ACTINOMYCINS ANALOGUES 1113

particle size; Keystone Scientific, Bellefonte, PA, USA), and eluted by an isocratic mobile phase composed of 50 : 50 acetonitrile-water containing 0.1 % trifluoroacetic acid, at a flow rate of 200 p1 min-'.

LC/MS experiments, and MS/MS experiments employing flow injection of samples bypassing the column, used a SCIEX API III+ triple quadrupole mass spectrometer (SCIEX, Thornhill, Ontario, Canada) equipped with an IonSpray atmospheric pres- sure ionization source operated in the positive-ion mode. Medical grade air was used as nebulizer gas, and the spraying needle was maintained at + 5 kV from ground. For LC/MS experiments using the same HPLC conditions as those described above, 10 pl of sample solution were injected on the column and approximately 15 pl min-' of the column eluate was split off and directed to the ion source. For MS/MS experiments, 2 p1 of sample solution were injected into 15 pl min-' of the same mobile phase, which was transferred without splitting to the ion source. For the MS/MS experiments, argon collision gas was used at a target thickness of 3.5 x lOI5 atoms cm-'. The (laboratory frame) collision energy (nominal) was 30 eV for the singly charged ions studied here, although at the target gas thicknesses employed the ions emerge from the collision cell with extremely low energies." The quadrupole used to analyse the fragment ions was operated at unit-mass resolution in most cases (unless low abundances required otherwise), and was scanned over the range m/z 100-1300 (sometimes 1400) with a step size of 0.1 m/z unit and a dwell time per step of 1 ms. Data were accumulated in multi-channel analyser (continuum) mode, and all spectra shown here represent accumulations of data acquired from 2-4 injections. For LC/MS experiments (MH' ions only), a limited scan range (m/z 1100-1500) was employed.

Accurate mass measurements of MH' ions, and of the more abundant fragments formed by CID in the skimmer region of the source interface, were conducted using a VG-ZAB-EQ double-focusing instrument (Fisons Instruments, Manchester, UK) equipped with an electrospray ion source (Fisons Instruments). The ion accelerating potential was 4 kV and the mass resolving power was 7000 (10% valley definition). Exter- nal mass calibration, using polyethylene glycol (PEG) doped with sodium iodide as mass standard, was accomplished by acquiring spectra of the calibrant by voltage scanning in the continuum acquisition mode. A range of about 55 m/z units was used in order to include two PEG reference peaks (which are 44 u apart). By repeatedly alternating flow-injections of PEG and the sample, and by using the averages of two adjacent PEG injections to calibrate the linear mass scale for the sample injection made between them, a number of inde- pendent mass measurements were made for each ion of interest. The standard deviations of the mass values mainly reflected the absolute abundances of the mass peaks thus measured.

RESULTS AND DISCUSSION

The fragment ion spectrum of the MH' ion of actinomycin D (Fig. 1) will be described first in some detail,

since this compound will provide the standard structure to which all other substitutions will be referred, in the manner introduced by Barber et ~ 1 . ' ~ The other standard compounds (Fig. 1) correspond to either substitutions in the aromatic ring or of amino acid residues in one or both of the depsipeptide rings, and these two types will be discussed separately. Finally, the under- standing of the fragmentation pathways thus obtained will be applied to deduce structures of unknown minor contaminants in the standard preparations.

Standard compounds

Fragmentations of MH' ions of actinomycin D. The fragment ion spectrum of MH' from actinomycin D (m/z 1255.6 for monoisotopic species) is shown in Fig. 2. A detailed listing of the fragment ions observed in this spectrum is given in Table 1. The interpretations of these fragments, also summarized in Table 1, are discussed below.

In order to rationalize the observed fragmentations it is necessary to assume that the cyclodepsipeptide rings are first opened, most probably at the ester (lactone) f~nctionality. '~. '~ In the case of the MH' ions of the actinomycins, it was proposed by Roboz et ul." that this ring opening proceeded via a McLafferty rearrange- ment mechanism, but the scheme illustrating their proposed mechanism (Fig. 3 of Ref. 10) involves the neutral molecule rather than the MH' ion. The most stable form of the MH+ ion probably involves localization of the ionizing proton on the primary amine substituent on the quinoid ring, as shown in structure I. Activation

loo 1

100 200 300 400

500 100,

600 700

900 1000 I100 1200 1300

mlz

Figure 2. Fragment ion spectrum of MH+ ion from actinomycin D (m/z 1255.6), obtained using a triple quadrupole instrument (conditions are described in the Experimental section).

1114

I Sar

I Sar

I Pro

I Pro I I

D. THOMAS ET AL.

Table 1. Fragmenit ions observed for MH+ ions of actinomycin D (m/z 1255.6)’

mil Interpretation mlz Interpretation

1227.7 (w) 1142.4 (w) 974.4 (m) 956.4 (i) 928.3 (m) 875.1 (m) 857.4 (i) 829.4 (w) 744.0 (w) 657.0 (m) 629.4 (w) 576.2 (m) 558.1 (i) 530.2 (m) 477.0 (m) 459.0 (i) 430.9 (m) 399.1 (i) 394.1 (m)

[MH + - CO] [MH+ - MeVal] [MH+ - (Pro-Sar-MeVal)] b2 a2 [MH + - (Val-Prc6ar-MeVal)l b l a1 VI I [V‘ + Val] [V + Val - CO] [V’II - (Pro-Sar)] V [V‘ - CO] [V’ll - (Val-Pro-Sar)] VI [V‘l - CO] v4

Vlll

386.7 (w) 380.9 (w) 376.0 (w) 371.2 (w) 365.9 (w) 354.0 (w) 347.9 (w) 320.0 (w) 300.1 (i) 281.9 (i) 267.9 (m) 202.9 (m) 168.9 (m) 131.8 (w)

? [v4 - H201 [Vl l l - H,O] [v, - COI [VIII - CO] [ Y ~ - C O - N H ~ I [VIII - H20 - CO] [VIII - H,O - 2CO] v3

[H-Val-Pro-Sar] +

v2

[H-Pro-Sar] +

v1

[v3 - H,Ol

a Relative abundances are indicated as i = intense, m = medium and w = weak. Data were obtained usin!g a triple quadrupole instrument (see Experimental).

of I, to reactive forms which can yield fragment ions, corresponds to transfer of this proton to the less basic lactone group. A possible representation is shown as structure 11, which can then undergo charge-site initiated ring opening to form structure 111 (Scheme 1). This results in a pentapeptide substituent including a dehy- drothreonine residue, designated dhThr for convenience, on the chromophore. In the postulated structure 111 (Scheme 1) the ionizing proton is localized on the C-terminal carboxyl of the MeVal re:sidue, from which it is readily mobilized to amide 1ink.ages in the same peptide side-chain or even to the lactone functionality of the second, as yet intact, cyclodepsipeptide. Thus it is possible to conceive of fragmentations occurring from structure 111 (one depsipeptide ring intact) or from a related structure with both rings opened to form two linear peptides attached to the chro:mophore through the dhThr residues. In the case of actinomycin D (I) the two cyclodepsipeptide moieties are idlentical, and since no fragments involving cleavage of the chromophore structure were observed using low-energy CID, it was not possible in the present work to distinguish between the depsipeptide rings attached to the different halves of the chromophore.

0-2-MeVal

I

CH----CO-DVal I

NH I

NH I I

CH3 CH3

I1

NH NH I I

I Some of the fragmentations listed in Table 1 corre-

spond to well known reactions of protonated linear peptides. Thus, the ions at lower m/z values, identified

HO;\C,OH 0-2-MeVal

d

- CH3 CH3

111

Scheme 1. Proposed charge-site initiated mechanism for opening of a protonated lactone ring in actinomycin D, assumed to be a manda- tory prerequisite for all subsequent fragmentations.

1115 ACTINOMYCINS ANALOGUES

as y, (1 < n < 4), are the protonated peptides" gener- ated conceptually by progressive losses of the N- terminal residues from the protonated peptide [(H-D- Val-PreSar-MeVal-OH) + H']. (Of course, the present CID experiments can not distinguish between optical isomers.) In the present example of actinomycin D, which carries two identical depsipeptide rings, only one set of y, ions is observed, together with smaller fragments derived from them. The ions at higher m/z values described (Table 1) as [MH+ - (X)], where X represents one or more residues from the C-terminus of the ring-opened peptide structure I11 (Scheme 1), are the products of expulsion of C-terminal residues as first described for alkali-metal-cationized pep tide^''-'^ and later extended to their protonated counterparts.25926 These latter ions contain the intact chromophore, as do the higher mass ions labelled b, (n = 1, 2) which are the acylium ions2' resulting from expulsion of neutral fragments which are formally neutral amino acids (not residues) from the ring-opened peptides. Structure IV is that proposed for the b, ion, for the special case in which both depsipeptide rings have been opened and in which the fi-peptide has fragmented. Such b, ions readily lose CO to yield the corresponding a, ions, as noted here. The absence of observable b, ions for n > 2 is an example of the well known proline effect, a pro- pensity for facile cleavage on the N-terminal side of Pro residues.2 7 ~ 2 8

OH I

MeVal

n,

The remaining ions, observed (Table 1) in the middle m/z range, correspond to fragments in which neutral moieties have been lost from both depsipeptide rings, here assumed to have been opened to yield linear peptide chains linked to the chromophore via dhThr residues. Thus, the cyclic structure V for the ion at m/z 558.1 can be rationalized as resulting from the b, ion (structure IV), by attack of the acylium functionality of the dhThr residue of the fi-chain on the a-carbon of the D-Val residue (Scheme 2). Concurrent transfer of the hydrogen atom, originally on the D-Val a-carbon, to the N-terminus of the Pro residue yields a neutral fragment (H-Pro-Sar-MeVal-OH). Structure VI (m/z 459.0) is similar to V, except that in this case the acylium group in structure IV has attacked the a-carbon of the dhThr residue in the other peptide chain. Both structures V and VI are expected to lose CO readily, as observed (Table 1). Structure VII (m/z 744.0) is simply a case in which one ring-opened peptide chain has formed the bl structure, while the other has lost the C-terminal MeVal residue. Finally, structure VIII is proposed to account for the ion at m/z 394.1. The mechanism by which such an ion, in which only one Thr residue survives attached to the chromophore, is unclear. However, such a structure can rationalize the the ready formation of this fragment ion, and also its subsequent losses of H,O and CO neutrals (Table 1).

Thus far the structural and mechanistic interpreta-

( H-Pro-Sar-MeVal-OH )

+

V

Scheme 2. Proposed mechanism for expulsion of an (H-Pro-Sar-MeVal-OH) neutral fragment from a b, ion via electrophilic attack of the acylium group on the a-carbon of the o-Val residue in the other depsipeptide side-chain. The hydrogen atom originally bonded to this bonded atom is transferred to the N-terminus of the Pro residue.

VI VI1 VIlI


75 .

50 .

tions have been based solely on chemical intuition. Experimental evidence supporting these assignments were of two types, viz. mass shifts of fragment ions observed for other standard compounds of known structures, and accurate mass measurements of fragments formed in the skimmer region of an electrospray ionization source interfaced to a dou ble-focusing mass spectrometer. This evidence is summarized below.

Fragmentations of MH' ions of compounds related to actinomycin D by substitution in the benzene ring. TWO such compounds, 7-nitro-and 7-aminoactinoimycin D, were available (Fig. 1). If the interpretation summarized in Table 1 is correct, all of the fragment ions containing the chromophore should be shifted1 up in mass by appropriate amounts (45 and 15 u for the nitro and amino compounds, respectively), while those not including the chromophore (the yn ions and peptidic fragments derived from them) should not change. The fragment spectrum of the MH' ion of 7- nitroactinomycin D (m/z 1300.6, Fig. 3), obeys this pre- diction (only the more abundant ions are annotated in Fig. 3). The same result was obtained in the case of 7- aminoactinomycin D (not shown). These results confirm that none of the observed fragments result from cleavage of the chromophore ring system.

-

25 . '

Fragmentations of MH' ions of compouinds related to actinomycin D by amino acid substitution(s) in the depsipeptide rings. The simplest standard of this type, available for

3LUl

159 0 **1, ''59 1121 1619 w30 26.79 I

100,

75 1 1020

I99 2

:$1 I I 1 100 200 300 400 500

pm4

I

600 700 800 900 100 , l l W 8

' loo0 1 100 1200 1300

mlz

Figure 3. Fragment ion spectrum of MH+ ion from 7- nitroactinomycin D (m/z 1300.6), obtained using a triple quadrupole instrument (conditions are described in the Experimental section).

the present work, was actinomycin I (Fig. l), which is related to actinomycin D by a single amino acid substitution in the /?-depsipeptide ring (that linked to the quinoid ring). This substitution, hydroxyproline (Hyp) for Pro, corresponds to a 16 u upward shift at this position in the /?-chain. It is easy to predict, on the basis of the rationalizations summarized in Table 1, that all fragments supposed to contain both Pro residues in the case of actinomycin D should now be shifted upwards by 16 u, those fragments containing one Pro residue should now be doublets with one member shifted up by 16 u, and those fragments interpreted as containing no Pro residues should remain the same as for actinomycin D, e.g. the y4 and y3 fragments should now be doublets separated by 16 u while the y2 and y, ions should be the same singlets as those observed for actinomycin D (Table 1). Figure 4 shows the fragment ion spectrum of the MH' ions of actinomycin I. The above predictions are amply confirmed by this spectrum, although not all of the less abundant ions (Table 1) were observed in this case. In addition, in cases where doublets separated by 16 u are predicted for cleavages on the N-terminal side of Pro/Hyp, the ions corresponding to fragmentation at the Pro residue were invariably more abundant than those from the competing fragmentation at Hyp in the other peptide moiety. Table 2 summarizes these findings for actinomycin I.

The actinomycin C2 and C3 standards were supplied as a mixture. These compounds (Fig. 1) correspond to actinomycin D (I) in which DaIle substitutes for D-Val in one (C2) or both (C3) depsipeptide rings, giving rise

900 1000 1 100 1200 1300

m/Z

Figure 4. Fragment ion spectrum of MH+ ion from actinomycin I (m/z 1271.6), obtained using a triple quadrupole instrument (conditions are decribed in the Experimental section).


Table 2. Fragment ions observed for MH' ions of actinomycin I (m/z 1271.6)'

m h

1243.6 990.4 974.4 972.5 956.6

944.6 928.4 873.5 857.4

760.5 744.1 657.2

629.3

576.3 558.3

530.2

477.2

459.1

Interpretation

[MH+ - CO] [MH + - (Pro/HypSar-MeVal)]

bz

a2

b,

VII

[V + Val]

[V + Val - CO]

[VII - (Pro/Hyp-Sar)] V

[V - CO]

[VII - (Val-Pro/Hyp-Sar)]

VI

No. of Pro

2 1

1

1

1

1

0

0

0 0

0

0

0

mlz

431 .O 41 5.3 399.0 394.2

387.0

376.2 366.0

348.0

320.1 31 6.0 300.1

298.0 282.1 284.0 267.9

203.0

185.0 168.9

132.1

Interpretation No. of Pro

[VI - CO] Y4

0 1

Vlll 0

7 07

[VIII - H,O] 0 [VIII - CO] 0

[VIII - H,O - CO] 0

[VIII - H,O - 2CO] Yo 1

0

[Y3 - H,Ol 1

[ti-Val-Pro/HypSar] + 1

Yz 0

[H-Pro/HypSar] + 1

Yl 0

a The numbers of Pro residues are those in the corresponding ions from actinomycin D (Table 1 ), and in the present case should lead to a single peak at the same m/z for no Pro residues and at 16 u higher for 2, whereas for one Pro residue Table 1 predicts a doublet with 16 u spacing. Data were obtained using a triple quadrupole instrument (see Experimental).

to upward mass shifts of 14 or 28 u, respectively. Figure 5 summarizes the results of an LC/MS analysis of this mixture, which shows that the mixture contains a sig- nificant impurity with MH' at m/z 1255.6. The reten- tion time and fragment ion spectrum of this impurity were indistinguishable from those of actinomycin D. No chromatographically resolvable isomers of either of the two main constituents were observed using the HPLC conditions employed here.

The fragment ion spectrum of the MH' ions of actinomycin C2 can be predicted from that for actinomycin D, in terms of the present interpretation (Table l), in a manner similar to that described above for actinomycin I. In the case of actinomycin C2, however, the mass shifts are 14 u and the crucial single substitution is DaIle for D-Val. The resulting predictions of mass shifts and/or fragment doublets were again wholly ful- filled by the experimental spectrum (not shown). The corresponding spectrum for the MH' ion of actinomycin C3 (both D-Val residues substituted by DaIle) also showed the predicted 28 u shifts and predicted unchanged fragment masses, and also showed only singlets as required by the present interpretation.

Accurate mass measurements of fragment ions. Accurate mass measurement (to within accuracy and precision of a few parts per million) of intact protonated sample molecules, using double-focusing analysers equipped with electrospray ionization sources, was first reported by Larsen and and by Cody et ~ 1 . ~ '

However, extension of such measurements to fragment ions formed by CID in the skimmer region of the electrospray interface was demonstrated later by Starrett and D i D ~ n a t o . ~ ' The techniques used here were identical in all essential aspects to those described pre- v i o ~ s l y , ~ ~ except that sodiated polyethylene glycol ions were used as calibrants.

Table 3 summarizes the results thus obtained for some key fragment ions derived from actinomycin D. The precisions of these data are comparable to those reported p r e v i o ~ s l y , ~ ~ and in all cases the accuracy (deviation of measured from the predicted mass) was no worse than 5 ppm (Table 3). In no case was the precision sufficient to uniquely specify the atomic composi- tion of the ion, if this were regarded as a completely unknown quantity. The value of these measurements lies in the additional credibility they provide to the present interpretation of the fragmentations, derived from observations of the (nominal mass) shifts resulting from known variations in the molecular structure.

It is also pertinent to note that the relative abundances of fragments observed in the skimmer-cone CID spectra, obtained using the double-focusing instrument, did not match those observed in the tandem mass spectra obtained using the triple-quadrupole instrument. Therefore, it does not necessarily follow that it should always be possible to achieve accurate mass measurements of fragment ions observed at high abundance in low-energy CID experiments. Another observation of practical value concerns the use of sodiated


loo 3 TIC n

- p 25 ::uL 0

loo RIC miz 1283.6 ,-. - ; I : : I , , , , , , JL,, , , 2 25

0 0 5 10 15 20 25 30 35

Time (minutes)

Figure 5. Results of LC/MS analysis of i i commercial sample stated to contain actinomycins C2 (MH+ at m/z 1269.6) and C3 (MH+ at m/z 1283.6). The mass chroma1:ograms were recon- structed from the data obtained by scanning from m/z 1400-1 000. The LC/MS conditions are described in the Experimental section.

polyethylene glycol ions as mass standards. These ions were found to survive the variations in CID conditions produced by changes in skimmer-come potential, necessary to produce different fragment ions at usable abundances, much better than ammoniated ions of e.g. polypropylene glycols. This difference was important, since it was found that best accurac:y and precision of mass measurement demanded that the same skimmer- cone potential be used for acquisition of the peaks for the unknown ion and for the standard ions.

Attempts to measure the masses of the MH+ ions of the contaminants, discussed below, were unsuccessful owing to the low abundances of these ions.

Deduction of molecular structures of minor contaminants

In this section, the low-energy CID spectra of MH' ions of contaminants in the standard samples will be interpreted in terms of molecular structure. These deductions will depend on the assumption of the valid- ity of the present interpretation of the fragmentations of the actinomycin standards. In addition, in each case it will be assumed that the impurities are related to the main constituent via substitutions of amino acid residues, as described previou~ly. '~~ '~ A few of the spectra to be interpreted below were obtained using LC/MS/ MS techniques, but most of the MS/MS experiments employed direct flow injection of the entire sample. In view of the relative simplicity of each mixture in the present case, no attempt was made to preconcentrate the unknown contaminants by semi-preparative HPLC as was done previously for the much more complicated tyrocidine complex.

Impurity in the sample of actinomycin D. The most abundant impurity in the sample of actinomycin D exhibited an MH+ ion at m/z 1241.8, a downward mass shift of 14 u relative to the main component. On the assumption that this shift corresponds to a single substitution of an amino acid residue in structure I by a lower homologue, the following possibilities arise: MeVal + Val or MeAbu or MeAib; Sar-Gly; Val+Abu or Aib or MeAla; Thr - Ser; where Abu is 2-aminobutyric acid, Aib is 2-aminoisobutyric acid and the Me substitutions are N-methyl substitutions in each case. There is no known lower homologue of proline.

The fragment ion spectrum of the impurity MH+ ion at m/z 1241.8 is shown in Fig. 6, and is listed in Table 4. Not surprisingly, the signal-to-noise ratio was suficient- ly low that fewer fragment ion peaks could be reliably identified and mass-measured compared with those for actinomycin D (Table 1). Nonetheless, considerable progress can be made towards assigning a structure to the impurity by a comparison of Tables 1 and 4. The

Table 3. Accurate mass measurements of fragment ions formed ionization of actinomycin D'

Fragment Ion Attornic composltion Calcd. mjz Measured mlz

b2 a2 bl V v-co VI Y4

Vlll Y3

c'48H62N9012

C:4,H62N901 1

c'43H53N8011

C:2C3H28N507

"2EH ZEN 5O6

C:24H19N406

C'19H35N405

C20H16N306

('1 4H26N304

956.451 8 928.4568 857.3833 558.1 988 530.2039 459.1 304 399.2607 394.1 039 300.1 923

956.4526 f 0.001 1 928.4561 f 0.0041

558.1 997 f 0.0008 530.2040 f 0.0031

399.2627 f 0.0048 394.1 029 * 0.0020 300.1 934 f 0.0031

857.3856 * 0.0027

459.1 286 i 0.001 9

by electrospray

n A (pwOb

6 +1 5 -1 4 +3 7 + 2 7 +1 5 -4 7 +5 7 -3 4 +4

a The reported values are combinations of measurements made on two different days, using a Fisons Instrument ZAB-EQ double-focusing mass spectrometer equipped with a Fisons electroslpray ion source. b A = measured - calculated mass.

ACTINOMYCINS ANALOGUES

100 - h

75. ,x 2 .e

4 50 - P) .- +a 5 25 - 2

1 MS/MS of mlz 1241.8

MS/MS of m/z 1277.6

9048

401 1 10688 414 6 7261 7961

6912 161 1 879 2

0 1 ' . ' \

1119

1241 8

I h

% 843.6 x +a .- 2 8 e 50 -

300 2 942.8 c1

399A 4594 544.4 25 . 282.4

most striking feature of Table 4 is that the abundant b, and bl ions (and also the a2 and a, ions) are shifted down by 14 u relative to those for actinomycin D, whereas their y, and y2 complements are not mass- shifted. This observation suggests that the impurity corresponds to actinomycin D with one Thr residue replaced by Ser. However, this cannot be the entire explanation, since each of the b, and b2 ions has an appreciably less abundant homologue 14 u higher (at the m/z values appropriate for the b fragments from actinomycin D). This observation is inconsistent with a Thr -+ Ser substitution, since the b2 and b, ions contain both of the Thr/Ser residues. This conclusion is sup-

ported by the fact that each of the structure V and VJ ions in Table 4 consist of a homologous pair, and these ions also contain both Thr/Ser residues. In particular the structure V1 ions (no residues other than Thr/Ser) strongly suggest that some of these impurities contain a Thr -+ Ser substitution, while some contain both Thr residues intact.

It appears, therefore, that there must be two impurities, chromatographically inseparable under the present HPLC conditions, giving rise to MH + ions at m/z 1241.8. The more abundant of these impurities corresponds to a Ser residue substituting for one of the Thr residues in actinomycin D, giving rise to the mass-

Table 4. Fragment ions observed for MH' ions at m/z 1241.8, arising from an impurity in the sample of actinomycin D'

m h Interpretation mi2 Interpretation

1214.6 (m) [MH+ - CO] 459.4 (m) V I

1128.8 (w) [MH+ - MeVal] 445.4 (m)

399.4 (m) y4

300.2 (i) y3 960.6 (w) [MH+ - (Pro-Sar-MeVal)]

956.6 (w) b, 942.8 (i) 282.4 (m) [y3 - H,O] 928.3 (w) a, 268.2 (w) [H-Val-Pr*Sar]+ 914.4 (m)

857.2 (w) b, 843.6 (i) 829.4 (w) a, 815.8 (w) 643.2 (w) [V+Val] 558.6 (w) V 544.4 (m)

202.8 (w) y,

a Relative abundances are indicated as i = intense, m =medium and w = weak. Data were obtained using a triple quadrupole instrument (see Experimental).

I120 D. THOMAS ET AL.

413.1

shifted b ions and the invariant y ions. The less abundant impurity must contain both Thr residues, and a lower homologue substituting for oine of the other amino acid residues. The y2, y3 and y4 ions should provide some information on this point, but unfor- tunately no clearly defined homolog,ous mass peaks were observed, within the available signal-to-noise ratio, in any of these cases (Fig. 6). Hence this ultra-low-level impurity still remains unidentified. It could possibly correspond to actinomycin Do, previously identified3, as corresponding to one Sar replaced by a Gly residue. If this were the correct interpretation, each of the y2, y, and y4 ions should have shown a low-abundance homologue 14 u lower, but these could nod be observed as mentioned above. This example well exemplifies both the power and the limitations of the pr'esent approach.

Another ion in the electrospray mass spectrum of the actinomycin D sample was observed at m/z 1277.6, i.e. 22 u higher than the MH' ion. This was almost certain- ly the [M + Na]' ion of actinomycin D, and its fragment ion spectrum (Fig. 6), although of excellent signal-to-noise ratio, is relatively sparse and does not contain much interpretable structural information. The fragment at m/z 1164.5 (Fig. 6) corresponds to the p r e d i ~ t e d ~ ~ - ~ " [M + Na - MeVal]' ion, while most of the other fragment ions are probab1:y the products of charge-remote fragmentation^^^.^^^"^ rather than charge-site initiated cleavages of the peptide backbone.

lVlS.5

12865 9'65 Irnl I

Impurities in the sample of 7-nitroactinomycin D. In this case two homologous MH' ions were observed, one 14 u higher and the other 14 u lower than the MH' ion of the main constituent at m/z 1300.6. The fragment ion spectra of these two impurity ions are shown in Fig. 7. The less abundant lower homologue (m/z 1286.5) yielded a simple spectrum containing only fragments 14 u lower than those observed for 7-nitro-actinomycin D

1 MS/MS of mli: 12863

399 1

itself (Fig. 3). Since each of the b,, b2 , y, , y4, structure V and structure VI ions had m/z values 14 u lower than their counterparts from 7-nitroactinomycin D itself, with no sign of homology within the spectrum, the only structure for this impurity consistent with the available evidence again corresponds to a replacement of one of the Thr residues by serine. The only potential direct evidence for this, from spectra of the present kind, would come from a lack of homology in structure VI ions appearing only at the lower homologue m/z value (445.3), but these are not usually of high abundance and could not be reliably assigned in this case. However, the observation of a Thr -+ Ser substitution in actinomycin D (see the previous section) lends some credibility to this assignment in 7-nitroactinomycin D.

The fragment ion spectrum of the more abundant homologue impurity (MH' at m/z 1314.7) shows that all of the bl, b2 , y3 and y4 ions appear at the same m/z values as for 7-nitroactinomycin D itself, but accompa- nied by homologous ions of comparable abundances 14 u higher. On the other hand, the ions at m/z 602.7 and 504.3 (Fig. 7), corresponding to nitro-substituted structures V and VI, respectively, show no similar homology. These observations require that the amino acid substitution be of one of the Pro, Sar or MeVal residues by a higher homologue: Pro -+ Pip; Sar -+ Abu or Aib; MeVal 4 MeLeu or MeIle.

In principle, the lower mass fragments, e.g. y2 and yl, should have pinpointed the substituted residue, but these ions could not be observed under the CID conditions used.

Impurities in the sample of 7-aminoactinomycin D. In this case also the electrospray mass spectrum contained MH' ions separated by 14 u from that of 7- aminoactinomycin D (m/z 1270.7), both higher and lower homologues. The fragment ion spectrum of the

8H8 4

1 9871

12865

1268 7

0 200 400 12w

200 400 600 8CQ loo0 1200

m / z

Figure 7. Fragment ion spectra of the NIH' ions at m/z 1286.5 and 1314.7, from impurities in the sample of 7-nitroactinomycin D, obtained using a triple quadrupole instrument (conditions are described in the Experimental section).


830.6

higher homologue (m/z 1284.7, Fig. 8) is very character- istic. Each of the complementary b, and y4 fragments appear as homologous doublets of comparable abundances, comprising ions at m/z values appropriate to these fragments for 7-aminoactinomycin itself (m/z 872.4 and 399.2, respectively), together with the higher homologues (m/z 886.4 and 413.2). This is not true, however, of the complementary b, and y3 fragments, which appear only at the higher homologue value (m/z 985.5) and unshifted value (300.1). (A very low-abundance ion at m/z 971.2, Fig. 8, is discussed below). This evidence strongly suggests that it is one of the D-Val residues which is substituted by a higher homologue (Leu, Ile or possibly norleucine) in this case. As is well known,,' low-energy CID cannot distinguish between such isomeric residues. However, identification of the substituted residue as one of the D-Val residues is confirmed by the fragment observed at m/z 686.1 (structure V + Val), which appears at the higher homologue m/z value only, compared with the homologous doublet at m/z 587.4 and 573.1 (structure V contains just one of the Val or homologue residues). Similarly, the structure VI ion (contains no Val) appears only at the value appropriate to 7-aminoactinomycin D itself. Other less abundant ions in the spectrum support this interpretation. However, the appearance of a low-abundance but well defined b2 fragment at m/z 971.2 (Fig. 8) suggests that once again this impurity is itself contaminated with an extremely low-level isobaric compound. This trace impurity must contain both D-Val residues intact, so the homologous substitution must occur for one of the Pro, Sar or MeVal residues.

The fragment ion spectrum of the lower mass impurity (MH' at m/z 1256.7) is consistent with a single Thr + Ser substitution, as for the corresponding impurity in the actinomycin D sample (see earlier). Here again, however, the appearance of a low-abundance bl

812b 915 3

l W ] MS/MS of mlz 1256.7

5 75

J h

75 -

g 5 0 -

E 0 .-

n

0 > 414 0 .- c

ion at m/z 872.6 (Fig. 8), accompanying the much more abundant b, ion at m/z 858.3 (which in turn is consistent with the Thr +Ser hypothesis), suggests that a minor proportion of these impurity MH' ions contain both Thr residues plus a lower homologue substitution for one of the other residues.

1284 7

886 4

98s 5

Impurities in the sample of actioomycin I. The electrospray mass spectrum of the actinomycin I sample indicated three impurities, with MH' ions at m/z 1241.8, 1255.7 and 1257.7. The (A + 2) isotopomer of the MH' ions at m/z 1255.7 contributed significantly to the abundance at m/z 1257.7, leading to some difficulties in interpretation as described below.

The fragment spectrum of the impurity MH+ ion at m/z 1241.8 is summarized in Table 5. Comparison with Table 4, which summarizes the fragment spectrum of the impurity with the same nominal mass in the actinomycin D sample, shows that the two isobaric impurities are different from one another. In fact the fragment spectrum summarized in Table 5 is entirely consistent with that predicted for actinomycin Do ,33 corresponding to actinomycin D in which one of the Sar residues has been replaced by Gly. All predictions of mass shifts and of homologue doublets, based on this assumed structure together with the fragmentation pathways elu- cidated in the present work, are observed in the data presented in Table 5. However, as in all of the present work, it is not possible to determine which of the two depsipeptide rings has been thus substituted. Such a determination would require observation of fragment ions corresponding to cleavage of the chromophore moiety while maintaining the peptide side-chains intact, an unlikely combination of events given the relative sta- bilities of the two substructures.

The fragment spectra of the two impurity MH' ions at m/z 1255.7 and 1257.7 are shown in Figs 9 and 10,

0 > .- c 2 25 -

' 2 O f

858.3 LlSb 7

1284 7

414.0

I2566 300.3 399.2 413.2 573.0

, I L I # L, 1824 \ / 5873 686.1 901.1 y71 1 m 3

h

75 -

g 5 0 -

E 0 .-

n

loo 1 MS/MS of mlr 1284.7 J

872 4

I

2 25 - ' 2

O f

I2566 3003 3992 4132 573 0

, I L I # L, 1824 \ / 5873 686 1 1 m 3

9011 y712

I 98S5

mlz

Figure 8. Fragment ion spectra of the MH+ ions at m/z 1256.7 and 1284.7, from impurities in the sample of 7-aminoactinomycin D. obtained using a trip\e quadrupole instrument (conditions are described in the Experimental section).

1122

15 ~

50 .

D. THOMAS ET AL.

Table 5. Fragmeint ions observed for MH' ions at m/z 1241.8, arising from an impurity in the sample of actinomycin I'

mlz

1213.8 (w)

1128.5 (w)

974.6 (w)

956.6 (m) 942.5 (m) 928.7 (w) 914.7 (w) 857.4 (i) 843.6 (i) 744.3 (w) 730.6 (w) 657.0 (w)

576.2 (m)

558.2 (m) 530.2 (m) 477.3 (m) 459.2 (i) 431.3 (w)

Interpretation

[MH+- CO]

[MH+ - MeVal]

[MH+ - (Pro-Sar-MeVal)]

b2

a2

b,

VII

[V + Val]

[VII - (Pro-Sar/Gly)]

V [V - CO]

[VII - (Val-Pro-Sar/Gly)] VI

[VI - CO]

mlz

399.2 (m) 385.3 (m)

394.0 (m)

366.2 (w)

348.0 (w) 300.1 (i) 286.1 (m) 282.1 (m) 268.2 (m) 203.1 (m) 189.0 (w) 168.8 (w) 155.8 (w)

132.0 (w)

Interpretation

Y4

Vlll

[Vlll - CO]

[VIII - CO - H,O] Y3

[YB - HZOI

Yz

[H-Pro-Sar/Gly] +

Y1

a Relative abundances are indicated as i = intense, m = medium and w = weak. Data were obtained using a triple quadrupole instrument (see Experimental).

juxtaposed for convenience of comparison. The top spectrum in each case is for the MH" precursor at m/z 1255.7, and is indistinguishable from that obtained (Fig. 2 and Table 1) for actinomycin D although some of the low-abundance fragments are not well defined in Figs 9 and 10. The identification of this impurity is thus straightforward. The fragment spectrum for the precur-

sor at m/z 1257.7, however, represents a mixture of ions derived from the MH' ion of that m/z value together with those from the (A + 2) isotopomer (mostly 13C,) of the MH+ ion at m/z 1255.7. The abundance of this (A + 2) ion is calculated to be 31% of that of its A ion at m/z 1255.7. Based on the relative abundances of m/z 1255.7 and 1277.7 in the mass spectrum of the actino-

1 MS/MS of m/z 1255.7

x * .- !2 c

3w.3

459.2

25 1 I 3992 I 558.4 ~~

282.0 5303

0

100 MS/MS ofml:! 1257.7

h

P) 300.2

I

445 1 459 I 544 4 25

643 3

0 200 300 400 500 600

t?dZ

Figure 9. Partial fragment ion spectra (m/z 150-650) of the MH+ ions at m/z 1255.7 and 1257.7, from impurities in the sample of actinomycin I, obtained using a triple quadrupole instrument (conditions are described in the Experimental section).


100

75

50

25

0

851.6 1255 7

M S N S of m1z 1255.7

956 8

928 b 1M 7 1

M S N S of m/z 1257.1 1

L 885.8 814.9

12577

1243 2

700 800 900 loo0 1100 1200 13M)

mlZ

Figure 10. Partial fragment ion spectra (m/z 650-1300) of the MH+ ions at m/z 1255.7 and 1257.7, from impurities in the sample of actinomycin I , obtained using a triple quadrupole instrument (conditions are described in the Experimental section).

mycin I sample (not shown), it is thus estimated that approximately 50% of the observed abundance at m/z 1257.7 is due to this (A + 2) ion.

The identification of those peaks in the fragment spectrum of the precursor at m/z 1257.7 which arose only from the MH' ions at that m/z value, distinct from those arising from the actinomycin D isotopomer, was possible only because strict unit mass resolving power was maintained for both mass filters in the triple quadrupole instrument. A good example of this is provided by the region m/z 840-960 in Fig. 10 (the b, and b2 region). In the spectrum of the m/z 1257.7 precursor, the ions at m/z 843.6 and 942.6 are well defined and clearly arise only from the MH' ion since no ions at or near these m/z values are observed in the fragment spectrum of the MH' precursor at m/z 1255.7. The problem arises with the two triplets at m/z 857.7, 858.7, 859.7 and at m/z 956.6, 957.6, 958.6. The peaks within each triplet are all clearly resolved, which would not have been the case if the fragment analyser had been tuned with a mass window of e.g. 3 u. In that case, an entirely mis- leading single averaged fragment mass would have been assigned. The b, and b2 ions from the m/z 1255.7 precursor are at m/z 857.6 and 956.8, so for the (A + 2) isotopomer of this precursor the corresponding fragments are at m/z 857.6, 858.6, 859.6, etc. and m/z 956.8, 957.8, 958.8, etc., respectively, with decreasing relative abundances within each group. This is what is observed in the fragment spectrum of the m/z 1257.7 precursor, with the important exception that the third member in each case (m/z 859.7 and 958.6) is the most intense peak within its group. Clearly, then, these intense peaks must arise from the MH+ ion at m/z 1257.7, while the less intense peaks at lower m/z most likely correspond to the 13C0 and 13C1 variants of the fragments from the (A + 2) isotopomer of the MH+ species whose A ion is at m/z 1255.7. (Minor contributions to the abundances

at m/z 859.7 and 958.6 are made by the I3C2 versions of the appropriate fragments of this (A + 2) precursor ion.)

In this way it was possible to sort out which of the fragment ions of the m/z 1257.7 precursor arose from the MH+ ion, and which from the (A + 2) version of the MH' ion at m/z 1255.7. Interpretation of the corrected spectrum then followed a process similar to that described above. The impurity of interest has a molecular mass 14 u lower than that of actinomycin I, and therefore is assumed to arise from replacement of one residue by its lower homologue. Since actinomycin I already has different depsipeptide moieties (Fig. l), dis- tinguished by the presence of Pro in one and Hyp in the other, the substitution question is now not only one of which residue is replaced by its lower homologue, but is complicated by the question as to which of the depsipeptide moieties is thus substituted.

A first approach to this question can be made via the y3 and y4 ions. There are two such versions of each of the y3 and y4 fragments (arising from the a and fl peptide chains), one of which contains a Pro and the other a Hyp, residue. If the residue which is substituted by its lower homologue, and thus accounts for the 14 u downward mass shift of the MH+ species, is denoted as R, the following combinations are possible for the predicted doublets on the assumption that the R residue is indeed one of those in the y3 and/or y4 fragments:

y3: [Pro + R] and [Hyp + (R - 1411: m/z 300 and 302.

[Pro + (R - 14)] and [Hyp + R]: m/z 286 and 316.

y4: [Pro + R] and [Hyp + (R - 1411: m/z 399 and 4 0 1 .

[Pro + (R - 14)] and [Hyp + R]: m/z 385 and 415.

In fact, the only possibilities actually observed (Fig. 9) are those containing the unmodified R fragment, viz. m/z 300, 316, 399 and 415. This observation strongly


suggests that R is in fact one of the two Thr residues. The only alternative explanation would appear to require that presence of the (R - 14) homologue somehow inhibits the fragmentation pathways which lead to the y3 and y4 ions.

The evidence afforded by the bl and b2 fragments leads to a similar conclusion. In each case, after correc- tion for the contributions from the (A + 2) isotopomer of m/z 1255.7, as described above, two ions separated by 16 u are observed (Fig. 10). These ions are exactly those predicted for a structure in which no substitution occurs in the (-D-Val-Pro/HypSar-MeVal) moieties, with comparable probabilities for expulsion of the neutral fragments containing Pro or Hyp to form the b, and b, pairs. Again, this evidence points to a Thr -+ Ser substitution as accounting for the observed hiomology.

This interpretation is further supported by comparison of the structure V ions. Those objserved (Fig. 9) in the fragment spectrum of the m/z 1257.7 precursor include the predicted isotope cluster at m/z 558.4-561.4, arising from the (A + 2) version of the MH' ion of actinomycin D, together with a single peak (no isotopic variants) at m/z 544.4. The latter is exactly that predicted as the structure V fragment containing one Thr and one Ser residue. The only other residue in structure V is one of the D-Val residues, which can be ruled out as candidates for the substituted R residue by examination of the structure VI ions. These includle (Fig. 9) the isotopic cluster at m/z 459.1-461.1 arising from the (A + 2) version of the MH' ion of actinomycin D, plus a single peak at m/z 445.1 exactly as predicted for a single Thr -, Ser substitution. Other less abundant fragments in Figs 9 and 10 also support this conclusion.

CONCLUSIONS

The interpretation of the fragmentati.ons of protonated actinomycins resulting from low-energy CID, proposed here, appears to systematize successfudly all of the available information, including mass shifts resulting from known structural variations and accurate mass mea-

surements of some key fragment ions. Of course, the details of structures I-VIII are largely hypothetical, but even these are consistent with the experimental information and seem chemically reasonable.

Exploitation of these mechanistic interpretations in a more predictive manner, to deduce structural information about actinomycin analogues of unknown structures, met with mixed success. The key assumption, that the unknown compounds are related to standards by simple substitutions of amino acid residues,' appeared to be reasonable in the present work to the extent that the substitutions thus deduced all involved homologous residues. In those cases where interpretation to this point was not possible, the problem was attributable to poor signal-to-noise ratio arising from low abundances of these impurities. However, the present approach has serious intrinsic limitations. No distinction amongst possible optical isomers is possible, and the low-energy CID technique cannot distinguish amongst structurally isomeric amino acid residues such as Leu, Ile and Nle. Further, since the chromophore sub-structure of the actinomycins is considerably more stable than the depsipeptide moieties, it is not possible to produce fragment ions in which the chromophore is dissociated while the depsipeptide substituents remain intact (or largely so). For this reason, it is not possible in general, using the present techniques, to distinguish between the a and fl depsipeptide rings.

The strength of this MS/MS approach lies in the speed with which it can provide a useful degree of structural information on components of mixtures of actinomycin analogues, even when these are present in low abundance (both relative and absolute). This work has uncovered examples where it appears that one of the Thr residues is substituted by Ser, a variation which does not appear in the most extensive listing known to US.^ Confirmation of this and other findings of this work, and determination of other structural features such as the optical isomerism and whether the substitution occurred in the a or /? depsipeptide ring, would require isolation of sufficient purified material for analysis by NMR spectroscopy and other techniques.

REFERENCES

1. U. Hollstein, Chem. Rev. 74, 625 (1 974). 2. W. A. Rerners, Actinomycin, Chemistry of Antitumour Anti-

bodies, Vol. 1. Wiley, New York (1 979). 3. A. 8. Mauger, The Actinornycins, Topics in Antibiotic Chem-

istry, Vol. 5, edited by P. Sammes, p. 2i!5. Ellis Horwood, Chi- Chester (1 980).

4. B. W. Bycroft (Editor), Dictionary of Antibiotics and Related Substances. Chapman and Hall, London (1 988).

5. A. Mauger and E. Katz, Arch. Biochem. Biophys. 176, 181 (1976). H. Schulten, in Soft Ionization Biological Mass Spectrometry, edited by H. Morris, p. 27. Heyden, London (1981). A. Mauger, 0. Stuart, J. Ferretti and J. Silverton, J. Am. Chem. SOC. 107,7154 (1985). J. Greaves, M. McCarnish and J. Roboz, in Proceedings of the 32nd ASMS Conference in Mass Spectrometry and Allied Topics. San Antonio, TX, 7984, p. 447. M. Barber, D. Bell, M . Morris, L. Tetler, M . Woods, B. W. Bycroft, J. J. Monaghan, W. E. Morden and B. N. Green, Talanta 35,605 (1 988).

10. J. Roboz, E. Nieves, J. F. Holland, M. McCarnish and C. Smith, Biomed. Environ. Mass Spectrom. 16,67 (1 988).

11. D. W. Russell, 3. Chem. SOC. 753 (1 962). 12. C. G. Macdonald, J. S. Shannon and A. Taylor, Tetrahedron

Lett. 31, 2087 (1 964). 13. B. C. Das, P. Varenne and A. Taylor, J. Antibiot. 32, 569

(1979). 14. H. A. Gillis, D. W. Russell, A. Taylor and J. A. Walter, Can. J.

Chem. 68,19 (1 990). 15. J. M. Curtis, C. D. Bradley, P. J. Derrick and M. M. Sheil, Org.

Mass Spectrom. 27,502 (1 992). 16. K. B. Torner, F. W. Crow, M. L. Gross and K. D. Kopple, Anal.

Chem. 56,880 (1 984). 17. K. Eckart, Mass Spectrom. Rev. 13,23 (1994). 18. M. Barber, L. Tetler, D. Bell, B. Bycroft, J. Monaghan, B.

Morden and 0. N. Green, in Proceedings of the 35th ASMS Conference on Mass Spectrometry and Allied Topics, Denver, Go, 1987, p. 3.

19. X.-J. Tang, P. Thibault and R. K. Boyd, Int. J . Mass Spectrom. /on Processes 1 22,153 (1 992).


20. M. Morris, P. Thibault and R. K. Boyd, J. Am. SOC. Mass

21. K. Biemann, Methods Enzymol. 193, 886 (1 990). 22. 0. Renner and G. Spiteller, 8iom8d. Environ. Mass Specrrom.

15,75 (1988). 23. X. Tang, W. Ens, K. G. Standing and J. B. Westmore. Anal.

ch8m. 60,1791 (1 988). 24. R. R. Crese, R. L. Cerny and M. L. Gross, J. Am. Chem. SOC.

111,2835 (1989). 25. G. C. Thorne and S. J. Gaskell, Rapid Commun. Mass

Spectrom. 3,217 (1 989). 26. A. J. Alexander, P. Thibault, R. K. Boyd, J. M. Curtis and K. L.

Rinehart, lnr. J. Mass Spectrom. Ion Processes 98, 107 (1990).

27. K. Biemann and S. A. Martin, Mass Spectrom. Rev. 6, 1 (1987).

Specrrom. 5,1042 (1 994). 28. D. F. Hunt, J. R. Yates, 111. J. Shabanowitz, S. Winston and

C. R. Hauer, Proc. Narl. Acad. Sci. USA 83. 6233 (1986). 29. B. S. Larsen and C. N. McEwen. J. Am. SOC. Mass Spectrom.

2, 205 (1991). 30. C. N. McEwen and B. S. Larsen, Rapid Commun. Mass

Specrrom. 6, 1 73 (1 992). 31. R. B. Cody. J. Tamura and B. D. Musselman, Anal. Chem. 04,

1561 (1992). 32. A. M. Starrett and G. C. DiDonato, Rapid Commun. Mass

Specrrom. 7,12 (1 993). 33. M. L. Devan, T. 1. Orlova and A. B. Silaev, Anribioriki

(Moscow) 19,107 (1 974) ; Chem. Abstr. 81,36401 t ( 1 974). 34. R. P. Grese and M. L. Gross, J. Am. Chem. SOC. 112, 5098

35. K. B. Tomer, L. J. Deterding and C. Guenat, Biol. Mass (1 990).

Specrrom. 20,121 (1 991 ).

Documents

Fragmentation mechanisms of protonated actinomycins and their use in structural determination of unknown analogues