THEJOURNAL

OF

Vot. 259, No. 17, Issue of September 10, pp. 1080-10806,19& Printed in U.S. A.

BIOLOGICAL CHEMISTRY

0 1984 by The American Society of Biological Chemists, Inc.

A Revised Primary Structure for Neocarzinostatin Based on Fast Atom Bombardment and Gas Chromatographic-Mass Spectrometry* (Received for publication, January 16, 1984)

Bradford W. Gibsona,Walter C. Herlihy”‘, T. S. Anantha Samybp,Kyung-Soo Hahm‘, Hiroshi Maedad,Johannes Meienhofer‘, and Klaus Biemann“* From the “Departmentof Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, the *Comprehensive Cancer Center, University of Miami, School of Medicine, Miami, Florida 33101, the ‘Division of Biochemistry, College of Biochemistry, Yunsei University, Seoul, Korea, the dDepartment of Microbiology, Kumamoto University Medical School. Kumamoto. 860 JaDan, and the ‘Bio-Organic Chemistry Department, Roche Research Center, Hoffman-LaRoche, Nutley, New Jersey07110-

The amino acid sequence of the antitumor protein neocarzinostatin was revisedon the basis of mass spectrometric studies. Gas chromatographic mass spectrometry on the 0-trimethylsilyl polyaminoalcohol derivatives of peptide mixtures derived from tetra Scarboxymethyl-neocarzinostatin were used to partially sequence neocarzinostatin. In addition, fast atom bombardment-massspectrometric experiments on neocarzinostatin and its tryptic fragments gave the molecular weights of various peptides and, in some cases, partial sequence information. The revised sequence involved reordering of two chymotryptic peptides, the identification of a newdi- and tripeptide sequence (Ala-Asp and Ala-Ser-Thr), the repositioning of Trp at position 39, and the assignment of the remaining Asx residues. The revised structure for neocarzinostatin (Mr= 11,105) now shows considerable homology with the other antitumor antibiotic proteins macromomycin and actinoxanthin.

Neocarzinostatin is an antibiotic protein isolated from culture filtrates of Streptomyces carzinostaticus (1) which possesses antitumor activities (1-5). It is a single-chain polypeptide with a molecular weight of approximately 11,000 and exists as an apoprotein and an associated nonprotein chromophore. A primary structure of neocarzinostatin was first proposed in 1972 by Meienhofer et al. (6), with the completed study appearing 2 years later (7, 8). Since that time many investigators have studied the biological properties of neocarzinostatin at the molecular level (9). Clinical studies showed that it is a potent anticancer agent useful for the treatment of the cancer of the bladder (10) or, in the form of a new derivative, for the treatment of liver cancer (11).In addition, a partial structureof the chromophore has been proposed (12, 13). Evidence suggested that the chromophore is primarily responsible for the biological activity of neocarzinostatin and its interaction with DNA while the apoprotein is essential for its stability and transport (14). Recently we determined the primary structure of a related * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Present address, Repligen Corporation, Cambridge,MA 02142. Recipient of Grant CH232 from the American Cancer Society in support of this work. * Recipient of National Institutes of Health Grants RR00317 and GM05472 in support of this work.

’

protein macromomycin from Streptomyces macromomyceticus by a combination of automated Edman degradation, GCMS’ protein sequencing, and FABMS (15). Macromomycin, a low chromophore-containing preparation, is similar to neocarzinostatin and its chromophore (16-18). A comparison of the sequences of macromomycin and neocarzinostatin showed considerable homology in the NH2- and COOH-terminal regions but much less in the middle region. The relative positions of the four half-cystine residues in neocarzinostatin and macromomycin perhaps best demonstrate this point. While the first, third, and fourth half-cystines in macromomycin, positions 36, 88, and 93, and in neocarzinostatin, positions 37, 84, and 89, occur at similar positions, the second halfcystines that belong to thisnonhomologous middle region are located significantly further apart at positions 46 and 56, respectively. A similar comparison was made a few years ago between neocarzinostatin and the related antitumor protein actinoxanthin (19) that showed the same scattered pattern of homology whilemacromomycin and actinoxanthin had a more complete and consistent homology (15).Although the disparity in sequence homology between neocarzinostatin and its two relatives, macromomycin and actinoxanthin, could have resulted from their evolutionary divergence, we suspected instead that theprimary structure of neocarzinostatin needed to be revised. Therefore, we reinvestigated the previously proposed primary structure of neocarzinostatin by a combination of FABMS and GCMS protein sequencing. These studies revealed a substantially modified amino acid sequence for neocarzinostatin. Most significant were the presence of a di- and tripeptide sequence and a different sequence near the disulfide bonds. These changes showed neocarzinostatin to be more closely related to macromomycin and actinoxanthin than the previously proposed sequence indicated. MATERIALSANDMETHODS

Neocarzinostatin was obtained from Kayaku Antibiotics Research Laboratories, Tokyo, Japan, and purifiedaccording to previously published procedures (20). Trypsin (treated with ~-1-tosylamido-2phenylethyl chloromethyl ketone) and a-chymotrypsin werepurchased from Worthington. Thermolysin and elastase were from Calbiochem and Sephadex G-50 from Pharmacia. Iodoacetic acid was recrystallized from petroleum ether prior to use. All other chemicals were of reagent grade. Amino acid compositions of the tryptic peptides from tetra& The abbreviations used are: GCMS, gas chromatographic-mass spectrometry; FABMS, fast atom bombardment-mass spectrometry; HPLC, high-performance liquid chromatography, MH+,protonated molecular ion.

10801

10802

RevisedStructure Primary

carboxymethyl-neocarzinostatin were determined using a Beckman model 121 MB analyzer after hydrolysis of the peptide with 6 N HCI containing 5 mM phenol in evacuated ampules at 110 "C for 20 h. Gas chromatographic-mass spectrometry was used in the partial sequencing of tetra-S-carboxymethyl-neocarzinostatin and itstryptic fragments. In all cases 50-100 nmol of S-carboxymethyl-neocarzinostatin or the corresponding second tryptic fragment (T-2) were partially hydrolyzed to maximize the yield of di- to hexpapeptides with 6 N HCI (107 'C, 18 min), elastase (37 "C, 3 h), or thermolysin (55 "C, 2 h). The peptide mixtures were converted to the corresponding 0trimethylsilyl polyaminoalcohols and analyzed by GCMS (21). A Varian 3700 gas chromatograph equipped with an on-column injector and a 30-meter SE-30 (0.32 mm, inner diameter)fused silica capillary column (J & W Scientific) was interfaced via an open split coupling to a Finnegan MAT 212 mass spectrometer. A 2-5% aliquot of the derivatized peptides was injected a t 40 "C and the temperature programmed at 3 "C/min to 320 "C. Mass spectra were collected at a scan speed of 2 s/decade and processed on a Finnegan SS-200 data system. Fast atom bombardment-mass spectrometric experiments on tetraS-carboxymethyl-neocarzinostatinand its various peptide fragments were carried out using approximately 10-30 nmol of material. The tryptic peptides were either run intact (T-1 and T-4) or after thermolytic (T-2 and T-5) andchymotryptic (T-2) hydrolysis. From T-2 several chymotryptic peptides were isolated and subjected to FABMS separately. The peptides were dissolved in 10-20 pl of 2:l glycerolmethanol and acidified with oxalic acid or acetic acid to enhance the protonated molecular ion, MH'. In some cases NaCl was added to increase the abundance of the cationated molecular ions, MNa+, (M - H + 2Na)+,. . .. Approximately 1pl of the sample preparations was placed on a stainlesssteel fast atom bombardment probe and inserted into the ion source of a Varian MAT 731 mass spectrometer. A modified Ion Tech fast atombombardment source was operated at a current of approximately 25 pA at 6-7 kV with xenon (22). The enzymatic digests for the GCMS and FABMS experiments were carried out in 0.1-1.0mlof0.05 M NHlOAc (pH = 8.0-8.5) containing 0.5 mM CaC&and 0.2% NaN3. Enzyme to substrate ratios were 1:50 to 1:lOO (w/w). In all cases the progress of the enzymatic hydrolysis was monitored by HPLC (23). Tetra-S-carboxymethyl-neocarzinostatin-The two disulfide bonds in neocarzinostatin were reduced by first dissolving 53.5 mg of neocarzinostatin in 30 ml of 0.7 M Tris/HCl buffer (pH 8.7) containing 14.4 g of urea and 60 mgof Na2EDTA. The reaction mixture was deaerated with a stream of nitrogen and then kept a t 37 "C for 2 h. To this mixture 0.7 ml of 10 mM 2-mercaptoethanol was added and the mixture held at room temperature for 18 h. Reduced neocarzinostatin was alkylated by the addition of 2.05 g of recrystallized iodoacetic acid in 12 ml of 1 N NaOH and allowing the reaction to proceed in the dark. After 2 h, 0.2 ml of 2-mercaptoethanol was added to destroy excess iodoacetic acid. The solution was then extensively dialyzed against water at 4 "C in the dark. The dialyzed carboxymethylated protein was lyophilized yielding a total weight of 52.0 mg. Tryptic Hydrolysis of the Lyophilized Protein (52 mg)-This was carried out in 30 ml of 50 mM ammonium bicarbonate (pH 8.4) with the addition of250 plof a freshly prepared solution of trypsin containing 6 mg in 1 ml of 1.0 mM acetic acid and 0.1 mM CaC12.The solution was incubated at 37 "C for 18 h with an additional 125 pl of the trypsin solution added after 5 h. The hydrolysis was stopped by freezing and lyophilizing the solution. The lyophilized hydrolysate was then dissolved in 9 ml of 10 mM ammonium bicarbonate and applied to a Sephadex G-50 column. The peptides were eluted from the column and collected in 9.2-ml fractions. The absorbances a t 230 nm were recorded for each tube. In addition, a 300-pl aliquot from each tube was hydrolyzed with 20% NaOH and reacted with ninhydrin, and theUV absorbance was read at 570 nm (Fig. 1).

for Neocarzinostatin

Fracllon Number

FIG. 1. The tryptic peptides of S-carboxymethyl-neocarzinostatin as chromatographed on a Sephadex (3-50 column. Fractions were collected in 9.2-ml aliquots and the absorbance measured a t 230 nm (0). After reaction with ninhydrin, the absorbance was also recorded at 570 nm (0).

nonhomologous region when comparing neocarzinostatin with macromomycin. First, T-2was hydrolyzed with a-chymotrypsin and the total hydrolysate subjected to FABMS to obtain the molecular weights of the resultingpeptides. A second set of overlapping peptides was then produced by digestion of T2 with a complementary enzyme, thermolysin. The FABMS spectrum of the chymotrypticdigest indicated the presence of at least six peptides, five of which are apparent in the partial spectrum shown in Fig. 2. Each peptide appears as the corresponding protonated and sodium-cationated molecular ions, MH', MNa', and (M - H 2Na)'. Only three of the six expected chymotryptic peptides listed in Table I could be found in this spectrum; those with M, = 374, 794, and 1077. The peptide with M , = 602 did not fit any fragment, regardless of cleavage sites,and while the two remaining fragments could be matched with the published sequence on the basis of their molecular weights, their formation would require cleavages of peptide bonds not normally hydrolyzed by chymotrypsin? A similar situation existed for the FABMS of the thermolytic fragments (see Table I); three peptides fit primary cleavage sites, three did not conform to thermolysin specificity, and three could not be fitted to the sequence a t all. Thus, there were a number of peptidesfromboth digests with molecular weights that conformed to thesequence, indicating the correctnessof these regions, and some that did not, clearly showing the sequence to be in error. In order to determine the amino acid sequence of T-2, it was subjected to a partial acid hydrolysis and an exhaustive thermolytic digestion, respectively. The resulting peptide mixture was converted to the 0-trimethylsilyl polyaminoalcohol derivatives and analyzed by GCMS (see Fig. 3 for the gas chromatogram of the derivatives from the partial acid hydrolysate). From the short sequences determined by GCMS one can assemble three longer ones (Table11):two sequences of eight and Ser-Alaamino acids, Phe-Ser-Ser-Val-Thr-Ala-Asx-Ala Ser-Thr-Ser-Leu-Thr-Val, as well as a tetrapeptide Val-LeuAla-Cys. The latter requires that the chymotryptic peptide which ends in Val-Leu is followed by that which begins with Ala-Cys. Finally, these GCMS data also revealed the presence RESULTS of a previously missing sequence Ala-Ser-Thr that precedes Although the second tryptic fragment from neocarzinostathe fifth chymotryptic peptide. The single remaining unastin contained the only region that appeared in need of reinvestigation, the entiresequence was checked. * A computer program was used to generate all possible peptides Tryptic Peptide T-2 (21-66)"Two FABMSexperiments and their corresponding molecular weights from a given amino acid were carried out on the tryptic peptideVal-21-Arg-66, previ- sequence. In some cases, more than one possibility existed fora ously designated T-2 (6, 8). It contained nearly all of the particular value and enzyme specificity was usedto assign the peptide.

+

10803

Revised Primary Structure for Neocarzinostatin

c U J

795

w Ly

FIG. 2. A fast atom bombardment-mass spectrum of T-2that had been hydrolyzed with a-chymotrypsin. Theprotonated molecular ions, MH+,are labeled for each peptide. The ions with an asterisk correspond to the

600

650

700

750

BOO

E50

900

950

1000

M/Z

Na+ adducts. a

n J

1078

TABLE I Peptides identified from enzymatic hydrolysisof T-2 Uncorrected sequence (Refs. 6 and 8)

Corrected sequence

Chymotrypsin 374 375 603 602 795 836 835 1078 1077 1368 1367

Thr-64-Arg-66 None Ala-55-Phe-61 None Val-21-Tyr-32 None

Thr-68-Arg-70 Val-40-Leu-45 Ala-46-Phe-52 ASP-33-Trp-39 Val-21-Leu-32 Ser-53-Leu-67

Thermolysin 373 374 838 837 1081 1080 1181 1180 1232 1231 1282 1283 1540 1539 1553 1552

Val-21-Gly-25 Leu-26-Asp-33 None None None Leu-26-Cys-37 None None

Val-21-Gly-25 Leu-26-Asp-33 Val-55-Leu-67 Val-44-Ser-54 Val-40-Asn-51 Leu-26-Cys-37 Leu-26-Trp-39 Val-40-Ser-54

MH'

794

M,

signed peptide, Asp-Gly, identified in the GCMS experiment of the partial acid hydrolysate is placed at thenew positions 60-61, between the two other sequences deduced from these data.3 This allows matching of two more molecular weights determined by FABMS (thermolytic M , = 1080 and chymotryptic M , = 1367) which corresponded to Val-55-Set-66 and Ser-53-Leu-67, respectively. However, it was still not possible to correlateall the FABMS data (Table I) with this modified sequence. While The numbering system used from this point on refers to thefinal sequence (Fig. 6) that includes the reordering of the chymotryptic peptides and the additional amino acids.

three additional peptides could be placed, Val-44-Ser-54 and the two mentioned above, there were still five that did not match. Three of these peptides, one chymotryptic and two = 602,1231, and thermolytic, are only one mass unit less (M, 1552, respectively) than the molecular weights of peptides Val-40-Leu-45, Val-40-Asn-51, and Val-40-Ser-54. A difference of one mass unit can only result from a change of an acid to an amide. It follows that the only acidic amino acid present in all three of these, Asp-41, must be Asn instead. The placing of the Asp-Gly mentioned above was further confirmed by FABMS and amino acid analysis of the chymotryptic peptide Ser-53-Leu-67, which had been isolated by HPLC. FABMS showed an abundant MH+ of 1368 (Mr = 1367) which was in agreement with the amino acid analysis of this peptide (chl T-2, Table III), both of which indicated the presence of one Gly and two Asp. It will be noted that the sequence so far did not contain Trp, yet peptide T-2 is known to have a UV spectrum corresponding to the presence of this amino acid. In order to identify its location a second peak was isolated from the chymotryptic digest of T-2 by HPLC while monitoring both the absorbance at 215 and 280 nm. This peak, which had the strongest absorbance at 280 nm, gave an ion of mlz 836 by FABMS corresponding to M , = 835 (ch2 T-2), one of two remaining unassigned FABMS peptides (Table11). The amino acid analysis agreed with the composition of the region Asp33-Ser-39 except that there was no Ser (Table 111).Thus Trp must be in the place of this Ser. This substitution is in agreement with both the molecular weight for this peptide and the specificity of chymotrypsin for cleavage at Trp. This revised sequence also matches now the last unidentified thermolytic peptide Leu-26-Trp-39 with M , = 1539. The final amino acid sequence of T-2 is shown in Fig. 4.

T-2

Revised Primary Structure for Neocarzinostatin

10804 FIG. 3. A GCMS total ion chromatogram of the partial acid digest of T-2after conversion of the peptides to their corresponding trimethylsilyl 0-polyaminoalcohols. Peptides identified are listed in order of elution: VA, PA, GL, AS, SA, TA, VL, AD, DG, DA, C, SV VAG, SL, VD, EA, DV, SS, ST, TS, VLA, GVL, LE, AC, CA, FS, SAS, PAD, DF,SSV,GLE, TAD, LEA, TSL, SSV, CD, EC, SVTA, ACD, STSL, ECA, VLAC. Cysteine was in the form of its carboxymethyl derivative.

SCAN NURBER

TABLE I1 New amino acid sequences deduced from peptides identified in thermolytic and partial acid hydrolysates of NCS T-2 Phe-Ser-Ser (t)" Phe-Ser-Ser-Val (t) Ser-Ser-Val (35, t) Ser-Val-Thr-Ala (38) Phe-Ser-Ser-Val-Thr-Ala-Asp-Ala Thr-Ala-Asp (32) Asp-Ala (10, t)

I

4 c,

Val

Thr Ala Asp Ala Asp Gly Ser Ala S e t Thr Ser Leu'Thr '>is1 4r;

T (70)

7 " "

==s-

" 7 -

FIG. 4. The sequence of T-2showing the peptides sequences Ser-Ala (5) Ser-Ala-Ser (27) Ala-Ser (4, t) Ser-Thr-Ser (t) Ser-Thr-Ser-Leu (40) Thr-Ser-Leu (34, t ) Ser-Leu (14, t) Leu-Thr (t) Thr-Val (t)

identified by the two GCMS experiments: partial acid (M), thermolysin (U and ) both , (M and ), the peptides identified by FABMS (,"--4. Enzyme cleavage sites for chymotrypsin (c) and thermolysin ( t ) are indicated.

Ser-Ala-Ser-Thr-Ser-Leu-Thr-Val

Val-Leu-Ala (21, t) Val-Leu-Ala-Cvs (42)

Val-Leu-Ala-Css

"Numbers in parentheses refer to peptides identified in partial acid hydrolysis of T-2 (see Fig. 3) and t in parentheses refers to peptide observed in thermolytic hydrolysate.

TABLE I11 Amino acid composition of tryptic and chymotryptic peptides Peptides were hydrolyzed for 20 h, and no corrections were made for the partialdecomposition of Ser and Thr. Cys was determined as its carboxymethyl derivative, and Trp was determined qualitatively by measuring the absorbance at 280 nm (+). Numbers in parentheses correspond to the nominal values for the amino acids based on the corrected sequence. The molecular weights of T-1, T-4, Chl T-2, Ch2 T-2 were separately determined by FABMS. Chl

T-4

T-2

1857)

CYS Asx Thr Ser Glx Pro Giy Ala 3.0 Val 3.4 Ile Leu TYr Phe Lys His A% Trp

(M.=

T-5

1274)

1.5 (2) 1.1 (1) 5.5 (6) 3.7 (4) 4.3 (5) 2.5 (3) 4.0 (5) 2.0 (2) 1.9 (2) 1.0 (1) 2.2 (2) 5.8 (6) (3) 9.0 (10) (3) 6.3 (6) 1.0 (1)

thermolysin ( t ) .

TABLE IV Peotide identified from thermolvtic hvdrolvsisof T-5

Ch2 T-2

T-1

(M,=

FIG. 5 . The complete corrected sequence for neocarzinostatin showing the FABMS peptides (1-,) and GCMS peptides (-) that were used to correct and verify thisstructure. Enzymatic cleavage sitesaretrypsin ( T ) , chymotrypsin ( c ) , and

2.8 (3) 0.8 (1) 0.6 (1)

1.6 (2) 1.0 (1) 3.7 (4) 1.0 (1) 2.6 (3) 0.9 (1) 1.7 (2) 1.1 (1) 1.9 (2) 1.0 (1) 2.1 (2) 4.7 (5) 4.6 (5) 2.9 (3) 0.8 (1) 1.0 (1) 1.0 (1) 3.4 (3)

(M, =

(M,=

1367) 388

835)

1.9 (2) 1.0 (1) 1.7 (2) 4.3 (5)1526 1.0 (1) 1.0 (1) 2.9 (3) 1.0 (1)

1.0 (1) 1.1 (1) 1.0 (1)

1.0 (1)

0.8 (1)

1.0 (1) 0.7 (1) + (1)

1.0 (1)

+

(1)

+

M.

MH+ ~~

~

389 340 988 1158 1527

479 987 1157

Val-108-Ser-111 Ile-110-Asn-113 Leu-97-Gly-107 Leu-97-Ala-109 Tm-83-Glv-96

Tryptic Peptides T-1,T-4, and Td-The FABMS data of tryptic peptides T-1 and T-4 confirmed the correction of the published sequences. The M , of T-1 was found to be1857 and that of T-4 was M , 1274, and both spectraexhibited a number of sequence ions which were also in agreement with the structure. The COOH-terminal tryptic peptide, T-5, had an expected M , of 3125 or 3126, depending on the assignment of the Asx86 residue. This is beyond the present massrange of our mass spectrometer, andT-4 was, therefore, further hydrolyzed en-

FIG. 6. Amino acid sequences of macromomycin, actinoxanthin, and neocarzinostatin. Boxed regions represent sequence homologies.

_-

I 1"

lu-Cly-Ala-Ala-Cln Scr-Gly Ala-Ilc-ThrPh 100

Ser-Gly Leu-Aan-Leu-Cly-Hla-

zymatically. Since pepsin was used in the original sequence work on T-4, thermolysin was chosen to generate a different set of fragments thatwould confirm the ordering of the peptic fragments. Along withtheGCMSdata (Fig. 5 ) , thefour thermolytic peptides identified by FABMS from this experiment provided confirmatory evidence (Table IV) for the correctness of the previously published sequence (6), except that the amino acids 105 and 106 are Pro-Glu and not Glu-Pro. This reversal was first pointed out to usby A. Murai and K. H i r a ~ a m a . ~ was I t confirmed by GCMS data which revealed the derivativesof Pro-Glu andGly-Pro-Glu. It should be noted that the FABMS spectrum had ions one mass unit lower than both MH' = 998 and 1158 of almost equal abundance. This would indicate that one of the three acidic amino acidsin the region covered by these two peptides a t Asp-99, Asp-103, or Glu-106 may arise from partial deamidation.5 Lastly, the FABMS data settled the question of the assignment of Asx-86, which must be an Asp and not anAsn because the thermolytic peptidecovering this region had a M , of 1526 rather than 1525. The Ordering of the Tryptic Peptides-While the proper order of the tryptic fragments was not in doubt the overlap peptides were searched for in the GCMS dataof the elastase digest of neocarzinostatin. The GCMS peptide Lys-Val-Ala showed T-1 and T-2to be adjacent, but no arginine-containing peptides were observed because the hydrolysates had not been subjected to hydrazinolysis prior to derivatization and GCMS analysis. Thus, toverify the remainingcrucial overlap of T-2-T-3-T-4, where T - 3 is a single Arg, 10 nm of carboxymethyl-neocarzinostatin was digested with a-chymotrypsin expected to excise the peptide 68-Thr-Val-Arg-Ser-Phe-73 from the cleavages at thechymotryptic sitesLeu-67 and Phe73. The FABMS spectrumof this digest confirmed the presence of a peptide with M , = 764 which matched the value

' A. Murai and K. Hirayma, to be published. The most likely candidate for an amide in this region would be Asp-99whichwould make neocarzinostatin homologous with actinoxanthin at Asn-98.

7ft

Ala

1107

Val-Ala Leu Thr-Phe Gly

expected for this overlap sequence. This completes the determination of the primary structure of neocarzinostatin which is shown in Fig. 5. DISCUSSION

In the examination of the primary structure of neocarzinostatin we relied almost exclusively on the FABMS and GCMS sequencing techniques that proved invaluable in our earlier sequencing work on macromomycin (15). The combination of these two mass spectrometric techniques made it possible to rapidly identify regions in neocarzinostatin that needed to be reinvestigated and to arrive at the correct structure without having to resequence the entire protein. The strategy involved the initial use of FABMS to determine the molecular weights of numerous peptide fragments derived by enzymatic hydrolysis of neocarzinostatin. These fragments appeared primarily as the protonated and sodiumcationated molecular ions, MH+, MNa+, (M - H + 2Na)+, . . . in spectra of both pure peptides, as was the casefor the smaller tryptic fragments T-1 and T-4, and in complex mixtures such as the chymotryptic or thermolytic hydrolysates of the larger tryptic peptides T-2 and T-5. The molecular weight data obtained from these spectra were then compared with the expected values based on the published sequence (6, 8). For both T-1 and T-4 themolecular weights agreed with the predicted values, and in addition,sufficient fragmentation of the peptideoccurred during ionization to confirm a large part of the sequence. In the cases where the experimental determined weights did not match theexpected values, as in some regions of T-2, the sequence was checked by GCMS experiments on peptide mixturesproduced by extensive enzymatic or partial acid hydrolysis of T-2 and intactneocarzinostatin. The corresponding trimethylsilyl 0-polyaminoalcoholmass spectra gave di- to hexapeptide sequences which were used to confirm or correctregions throughout the protein. This information, combinedwith the FABMS molecular weight and partial sequence data, led to a revised structure of neocarzinostatin which differsfrom that proposedearlier by the

10806

Revised Neocarzinostatin Structure for Primary Bradner, W. T., and Hutchison, D. J. (1966) Cancer Chemother. Rep. 50, 79-87 Takahashi, M., Toriyama, K., Maeda, H., Kikuchi, M., Kumagai, K., and Ishida, N. (1969) Tohoku J. Ezp. Med. 98, 273-279 Kumagai, K., Maeda, H., and Ishida, N. (1967) Antimicrob. Agents Chemother. 1 9 6 6 , 546-550 Maeda, H., Kumagai, K., and Ishida, N. (1966) J. Antibiot. (Tokyo)Ser. A 19,253-259 Meienhofer, J., Maeda, H., Glaser, C. B., Czomboz, J., and Kuromizu, K. (1972) Science (Wash.D.C.) 178,875-876 Maeda, H., Glaser, C. B., Czombos, J., and Meienhofer, J. (1974) Arch. Biochem. Biophys. 164,369-378 Maeda, H., Glaser, C . B., Kuromizu, K.,and Meienhofer, J. (1974) Arch. Biochem. Biophys. 164,379-385 Goldberg, I. H., Hatayama, T., Kappen, L. S., Napier, M. A., and for Povirk, L. S. (1981) in Molecular ActionsandTargets Chemotherapeutic Agents (Sartorelli, A. C., Bertino, J . R., and Lazo, J . S., eds) pp. 163-191, Academic Press, New York Maeda, H. (1981) Anticancer Res. 1, 175-185 Konno, T., Maeda, H., Iwai, K., Tasiro, S., Maki, S., Morinaga, T., Mochnaga, T., Mochinaga, M., Hiroaka, T., and Yokoyama, I. (1983) Eur. J. Cancer Clin. Oncol. 19,1053-1065 Albers-Schoherg, G., Dewey, R. S., Hensens, 0. D., Liesch, J. M., Napier, M. A., and Goldberg, I. H. (1980) Biochem. Biophys. Res. Commun. 95,1351-1356 Napier, M. A., and Goldsherg, I. H. (1981) Biochem.Biophys. Res. Commun. 100,1703-1712 Kappen, L. S., Napier, M. A., and Goldberg, I. H. (1980) Proc. Natl. Acad. Sci. U. S. A. 77, 1970-1974 Samy, T. S. A., Hahm, K.-S., Modest, E. J., Lampman, G. W., Keutmann, H. T., Umezawa, H., Herlihy, W. C., Gibson, B. W., Carr, S. A., and Biemann, K. (1983) J. Biol. Chem. 258, 183-191 Kappen, L. S., Goldberg, I.H., andSamy, T. S. A. (1979) Biochemistry 18,5123-5127 Napier, M. A,, Holmquist, B., Strydom, D. J., and Goldberg, I. H. (1979) Biochem. Biophys. Res. Commun. 89,635-642 Kappen, L. S., Napier, M. A., Goldberg, I. H., and Samy, T. S. A. (1980) Biochemistry 19,4780-4785 Khokhlov, A. S., Reshetov, P. D., Chupova, L. A., Cherches, B. Z., Zhigis, L. S., and Stoyachenko,I. A. (1976) J. Antibiot. (Tokyo)2 9 , 1026-1034 Samy, T. S. A., Hu, G.-M., Meienhofer, J., Nazareth, P., and Johnson, R. K. (1977) J. Natl. Cancer Inst. 58, 1765-1768 Carr, S. A., Herlihy, W.C., and Biemann, K. (1981) Biomed. Mass Spectrom. 8,51-56 Martin. S. A.. Costello. C. E.. and Biemann,K. (1982) Anal. Chem. 54, 2362-2368’ Fullmer. C. S.. and Wasserman. R. H. (1979) . , J. Biol. Chern. 254, 7208-7212 ’ Pletnev, V. Z., Kuzin, A. P., Trakhanvov, S. D., Kostetsky, P. V., Popovich, V. A,, and Tsigannik,M. M. (1981) Biopolymers 2 0 , 679-694 Samy, T. S. A., Atreyi, M., Maeda, H., and Meienhofer, J. (1974) Biochemistry 13, 1007-1014 26. Biemann, K. (1982) Int. J. Mass Spectrom. Ion Phys. 45, 183194 27. Webster, T. A,, Gibson, B. W., Keng, T., Biemann, K., and Schimmel, P. (1983) J. Biol. Chem. 258, 10637-10641

2. correction of misidentified or misplaced amino acids (Asn-47, Ala-52, Asp-58, and Trp-46), the additional di- and tripeptide 3. sequences (Ala-59-Asp-60and Ala-63-Ser-Thr-65),a reordering of two of the chymotryptic peptides from T-2 (Ser-534. Leu-67 and Ala-46-Phe-52), and the assignments of the re5. maining Asx residues (Asp-48 andAsp-87). The modified primary structure of neocarzinostatin dis6. plays a high degree of homology (Fig. 6) with macromomycin 7. and with the recently proposed structure for actinoxanthin (17), also an anti-antitumor protein. In particular, the posi8. tions of all fourhalf-cystine residues are now conserved. Since the disulfide linkages have been determined for actinoxanthin 9. (17), one can reasonably assume that they are the same for the similarly placed half-cystines in neocarzinostatin. This would link cysteines at positions 37 and 47, and 88 and 93, 10. producing two loops of 11 and 6 amino acids, respectively. In 11. addition, the tyrosine at position 32 in neocarzinostatin is largely conserved among these three proteins and is known to be “buried” in the interior of actinoxanthin by crystallo- 12. graphic studies (24).If this is also the case in neocarzinostatin, it would explain why Tyr-32 is inaccessible to oxidation by 13. N-bromosuccinimide (25). With the primary structurefor neocarzinostatin now avail- 14. able,it will be of considerable interesttodeterminethe 15. complete structure of the associated chromophore as well as those for actinoxanthin and auromomycin. This information should contribute to the understanding of the mechanism by which these apoproteins stabilize the chromophore and pro- 16. mote their interaction with DNA. 17. It is conceivable that some of the differences between the 18. structuredeterminedearlier by classical methods usinga batch of neocarzinostatin produced in the late 1960s and the 19. sequence derivedin this paper may be due to mutations rather than errors in the experimental procedure. Mutations could cause heterogeneity, as in thecase with taka-amylase, or the 20. selection of a more potent strain over the years. However, the 21. ease and speed by which the need for revision of the earlier preliminary structure became evident andwas then corrected 22. indicates the utility of FABMS in such cases. The approach has broad applicability whenever a proposed or hypothetical 23. protein structure needs tobe verified or corrected. Examples 24. are the case of neocarzinostatin discussed in this paper and the verification of amino acid sequences of proteins deduced from the basesequence of their corresponding genes (26, 27). 25. ~~

REFERENCES 1. Ishida, N., Miyazaki, K., Kumagai, K., and Rikimara, M. (1965) J. Antibiot. (Tokyo) l 8 , 6 8 - 7 6

~~

~