1362 Journal of Medicinal Chemistry, 1977, Vol. 20, No. 11
function of the bis-mustards;cf. ref 9 and T. J. Bardos, N. Datta-Gupta, and P. Hebborn,J. Med. Chem., 9,221 (1966). (8) “Instruction 14,Screening Data Summary Interpretation”, Drug Evaluation Branch, National Cancer Institute, Bethesda, Md., 1963, p 16. (9) R. M. Peck, A. P. O’Connell, and H. J. Creech, J . Med. Chem., 9,217 (1966). We have adopted their original method with the stated modifications. (10) “Protocols for Screening Agents and Natural Products against Animal Tumors and Other Biological Systems”,Drug
Acton, Ryan, Luetzow
Evaluation Branch, National Cancer Institute,Silver Spring, Md., 1971. (11) R. E. Billingham and P. B. Medawar,J. Exp. Biol., 28, 385 (1951). (12) R. L. Sneath, J. E. Wright, A. H. Soloway, S.M. O’Keefe, A. S.Dey, and W. D. Smolnycki,J . Med. Chem., 19, 1290 (1976). (13) 0. Grummitt et al., Org. Prep. Proced., 2, 5 (1970). (14) A. Luettringhaus and R. Schneider,Justus Liebigs Ann. Chem., 679, 123 (1964).
Antitumor Septacidin Analogues Edward M. Acton,* Kenneth J. Ryan, and Arthur E. Luetzow Life Sciences Division, Stanford Research Institute, Menlo Park, California 94025. Received June 2, 1977
In the first approach by total synthesis to the structure of the antitumor antibiotic septacidin, analogues have been obtained which show similar inhibition of RNA-DNA synthesis in cultured leukemia L1210 cells and similar activity against transplanted leukemia P388 in mice. In these analogues, the natural aminoheptose moiety is replaced by 4-amino-4-deoxy-and 4-amino-4,6-dideoxy-~-glucose, to retain the natural configuration of the pyranose ring. Also retained is the lipophilic fatty acid-amino acid side chain attached to the 4-amino group and glycosylation at the 6-NH2of adenine. If the fatty acid chain was shortened from c16 to c6, if the fatty chain was shifted to the glycine unit, or if the glycine unit was omitted,activity was completely lost. However, activity was retained if the c16 chain was shortened only to C12or if the glycine unit was extended to &alanine. Both active and inactive analogues were nonbinding to DNA and nonmutagenic to Salmonella strains. The synthetic approach was to start with a suitably protected sugar (L-fucose and L-galactose), construct the adenine moiety at C-1, introduce a 4-amino group, and finally attach the preformed side chain. Activity vs. CA 755 was retained in a number of analogues, showing that certain structure changes are acceptable. The synthesis and evaluation of any such structures have received surprisingly little attention. Only a handful of cases are r e p ~ r t e d where ~ - ~ a sugar is attached to the 6-amino group of adenine. Even the isomer of adenosine, 6-(~-~-ribofuranosylamino)purine, was only recently claimed.’O Nor has 4-amino-4-deoxy-~-glucose previously been synthesized, a key degradation product in the structure proof. Comparison with the D enantiomer had to be relied ~ p o n . ~This , ~ ’ report describes the first totally synthetic approach to the septacidin structure, with the synthesis and biological evaluation of septacidin analogues 2-12 (Table I), of which all but 2 are completely new substances. Chemistry. The general approach was to start with a sugar, construct on to it the adenine moiety, and attach the fatty acid-amino acid side chain last. In choosing the initial targets, the L-glycero-L-glucoheptose unit of 1 was replaced by hexoses (Table I, R’ = CHzOH and CHJ that were more accessible yet retained the L-gluco configuration of the pyranose ring. To introduce a 4-aminodeoxy function in the L-gluco configuration logically required 4-O-methanesulfonyl-~-galactoprecursors (Scheme I, 22 1 1, R = HOCH 23). This would be as in the enantiomeric series in the I synthesis of 4-aminodeoxy-~-glucosederivative^.'^^'^ The CH,OH most readily available L-galacto sugar as starting material was L-fucose (14a, 6-deoxy-~-galactose).Also, L-galactose To explore the potential of 1 as a lead for cancer drug (14b) was obtained by reduction of ~-galactono-1,4-lactne development, further studies required the flexibility of (13).14 The septacidin analogues 2-12 have been syntotal synthesis for systematic structure variation. So far, thesized starting from these sugars. the structure determination of septacidin rests entirely on Selective functionalization of the galactopyranose ring elegant chemical degradation studies and spectral analat the axial 4 position is readily attained.’2~15~16 In the y~es.’,~Partial degradation permitted preparation of a fucose series (Scheme I, series a) the 2,3-di-O-benzoate number of septacidin analogues, through cleavage of the 4-0-mesylate 16a is formed in high yield.16 Treatment of fatty acid-glycine side chain and reattachment of modified 16a with hydrogen chloride in the presence of titanium side chains and by chain shortening of the sugar unit to a hexopyranose or lengthening i t to an oct~pyranose.~ chloride in benzene solution afforded a crystalline chloro
The antitumor antibiotic ~eptacidinl-~ (1) is structurally unique, combining fatty acid, glycine, amino sugar, and adenine units. The amino sugar is a heptose, 4-amino4-deoxy-~-glycero-~-glucoheptopyranose, not encountered elsewhere. It is linked to the adenine unit by glycosylation at the 6-amino group rather than a t N-9 or other ring nitrogen as in the purine nucleosides. In published data,’a3 septacidin was cytotoxic against Earle’s L cells in culture; it was active against adenocarcinoma CA 755 in mice but was inactive against Walker carcinosarcoma 256 and L1210 lymphoid leukemia, two experimental mouse tumors used extensively by the National Cancer Institute for screening new agent^.^ These data could suggest either that the septacidin structure has no potential for cancer treatment or that it has an unusual spectrum of antitumor activity.
-
Journal of Medicinal Chemistry, 1977, Vol. 20, No. 11 1363
Antitumor Septacidin Analogues Table I. Septacidin and Synthetic Analogues R'NH
HO
R2 (CH,),CH(CH,);,CONHCH,CO
R' =
NH
I
HOCH I CH,OH
R' = CH,OH 2
1 (septacidin)
isopalmi toylgly cy1
R' = CH,
3 (active)
(active)
CH,(CH,),,CONHCH,CO
4
palmitoylgly cy1
(active) 5 (active)
CH,(CH,),,CONHCH,CO lauroylglycyl
6 (active)
CH,(CH,),CONHCH,CO
7
caproylglycy1
(inactive)
CH,CH,OCH,CH,CONHCH,CO
9
8 (inactive) 10 (active)
(ethoxypropiony1)glycyl
(CH,),CH(CH,),,CONHCH~CH,CO isopalmitoyl-p-alanyl
(inactive)
CH,CONHCH[CH,(CH,), IC0
11 (inactive) 12 (inactive)
a -acetamidostearoyl
(CH,),CH(CH,)l,co isopalmitoyl Scheme I
0-
@"-
Ms
ti0
MsOQx-
BzO
BzO
HO
14,R'= H 15, R' = Me
BzO
16,X = OMe
18,X = N, 19,X = NH,
17,X = C1
Q -Q
-
MsO
MsO
BzO
20,
NH
x =0
21,X = H
BzO
NH
22
I R'O
I I
t?i
23,X = N,; R' = Bz 24,X = N,;R' = H 25,X = NH,;R' = H
temperature. Catalytic hydrogenation afforded the weakly basic glycosylamine 19a, which was not isolated but was condensed7 immediately with 4-amino-6-chloro-5-nitropyrimidine to give 20a. Hydrogenation and crystallization afforded the 4,5-diaminopyrimidine 21a as the chromatographically homogeneous @-anomerin 74 % yield. Diamine 21a was unexpectedly accompanied by about 2% of the less soluble a-anomer 26, identified by comparison of the circular dichroism spectrum with that of 21a. Cotton effects at wavelengths associated with the heterocyclic chromophore showed a pseudoenantiomeric relationship, with molar ellipticities [e] of -8520O at 291 nm for 26 and +12000° at 285 nm for 21a, as could be expected for a pair of a- and @-anomem. No reference compounds with measured circular dichroism were available for comparison with 26 and 21a. Assignment of absolute configuration as a-Land @-L, respectively, was confirmed by the fact that the major isomer 21a ultimately gave septacidin analogues (vide infra) having circular dichroism closely resembling that of septacidin, of established @-L configuration. (Anomeric pairs of adenine nucleosides, for example, have shown similar pseudoenantiomeric relationships, but of reversed ~ i g n . ' ~ J ~ ) Stronger Cotton effects in 21a and 26 at 225-230 and 239-240 nm were identical in sign and similar in intensity and were associated with the sugar benzoate chromophores. Compounds 21a and 26 were nearly identical in infrared and mass spectra, and elemental analyses and 'H NMR spectra clearly indicated they were isomeric. Since the
series a, R = CH, series b, R = CH,OH (14,15)or CH,OBz (16-25)
sugar as the pure a-anomer 17a. This method,l' though little used in the preparation of chloro sugars, is convenient in forming a relatively stable glycosyl halide directly from a methyl glycoside. The 0-mesyl group was stable in this step and also during the subsequent construction of the adenine moiety through the gly~osylamine~~' 19a. The crystalline @-1-azide18a was obtained by selective displacement with azide ion in dimethyl sulfoxide at room
?&
N\
6"
MsBq bNI NQ
MsO BzO
26,R = CH,
/N
"2
27,
R = CH,
1364 Journal of Medicinal Chemistry, 1977, Vol. 20, No. i l
Scheme I1
Acton, Ryan, Luetzow Scheme 111
(CH,),CHCH,CH=O
+ Ph,k2H(CH2),COOMe
28
29
25a /---
-+
(CH,j,GH(CH,),,CONHCH,COOR- -33
-
--.?E-+
(CH,),CHCH,CH=CH(CH,),COOR 30, R = Me
31, R
=
CH,(CH,), ,CONHCH,COOR 34
H
25b _ I _ _
33@
(CH,),CH(CH,),,COOH" 32
\
precursor 18a was obtained as a single anomer, it must be assumed that minor anomerization occurred in a following step. Ring closure of 21a to form the 6-N-substituted adenine 22a was carried out with triethyl orthoformate-acetic anhydride6,20 to avoid or at least minimize formation of any N-9-substituted isomers, e.g., 27. A nearly quantitative yield of 22a was obtained, but it contained two minor impurities according to thin-layer chromatographyon silica gel. One, just slightly less polar than 22a was suspected of being the isomer 27. This by-product showed an ultraviolet spectrum like adenosine (Amm 257 nm in acid, 258 nm in base) when a sample was separated from a condensation using diethoxymethyl acetate. This reagent was tried on the likelihoodmathat it is the true reactant when ethyl orthoformate-acetic anhydride is used. Surprisingly, 22a was then formed with an equal amount of the byproduct assumed to be 27. Ethyl orthoformate-acetic anhydride was clearly the reagent of choice. The desired product 22a was identified by an ultraviolet spectrum like that of septacidin.2 A pure sample could be obtained in 25% yield, but for preparation purposes it was efficient to use this intermediate without complete purification. Introduction of the azide group with configurational inversion at C-4 to form 23a required 18 h of heating with sodium azide in dimethyl sulfoxide at 100 OC. Saponification of the benzoates afforded the azide 24a as a homogeneous crystalline solid. The overall yield for eight steps from 16a was 18%. Hydrogenation with 10% palladium on carbon gave amine 25a as an amorphous, hygroscopic precipitate in 61% yield. According to thin-layer chromatography there was a contaminant that was negative to ninhydrin detection. A completely pure sample was not obtained by column chromatography on silica gel, and 25a did not form a crystalline HC1 salt or trifluoroacetamide. By high-pressure liquid chromatographic analysis (HPLC) the amine was 95% pure. Acylation of 25a to attach fatty acid-amino acid side chains yielded readily purified septacidin analogues. N-Acylglycine p-nitrophenyl esters were used as acylating agents, as was done in the modification of natural septacidin following its partial degradati~n.~Scheme I11 shows all the synthetic septacidin analogues formed by coupling various side chains with 25a or with the amine 25b when it was obtained from L-galactose. The fatty acid found in septacidin is the rare isopalmitic acid (32, 14methylpentadecanoic acid) and quantities of it had to be obtained by synthesis. Isovaleraldehyde (28) was coupled with the Wittig reagent21,2229 from methyl ll-iodoundecanoate, and the product was reduced and saponified to yield 32 (Scheme 11). For the N-acylation of glycine by the mixed anhydride procedure,23the triethylamine salt of 32 in tetrahydrofuran was treated with ethyl chloroformate followed by aqueous sodium glycinate to yield 33c (Scheme 111). Treatment with p-nitrophenol and di-
CH,( CH,), ,CONHCH,COOR 35 CH ,(CH, ),CONHCH,COOR 36
--
-
______--
-,3 2
- 4
--
25a ,- 6
25b
.5 7
2%-
EtOCH,CH,CONHCH,COORg$= - < - 37
(CH,),CH(CH,),,CONH(CH,),COOR 38 CH,CONHCHCOOR I
CHACH,), 39
8
4
-25a
lo
--2 6a
11
i
(CH ) ,CH( CII, ) , CO 0R
25a +
12
40
series c , R = H series d , R = p-NO,C,H,
cyclohexylcarbodiimide24 gave the ester 33d. Acylation of amine 25a by 33d was carried out in dimethylformamide solution. Septacidin analogue 3 was isolated by water precipitation as a solid that could be recrystallized to chromatographic homogeneity and analytical purity. It showed the distinctive ultraviolet spect,rum of septacidin,2 A,, 273 nm in acid, 264 nm at neutral pH, and 272 nm in base. The synthesis of 25b (Scheme I) was completely analogous to the synthesis of 25a, except that L-galactose first had to be prepared. Subsequent intermediates to 25b, even though enantiomeric to known sugars in the D series, were obtained in the L series for the first time. The 1-azide 18b, like 18a, was a readily purified crystalline solid. No attempts were made to isolate the glycosylamine 19b, which was generated in situ and used immediately. Only small samples of intermediates 20b, 21b, 22b, and 23b were obtained completely pure. It was more efficient to accept purity of 75-80% (according to HPLC) and make use of higher yields of these intermediates. The several impurities were not isolated or identified, and it is possible that one of the impurities in 21b might have been the a-L-anomer 26 (R = CH20H). Circular dichroism of the pure sample of 21b showed a positive Cotton effect at 287 nm, like that for 21a though weaker ( [ e ] +5940°), and confirmed that 21b and 21a had the same anomeric configuration. Partial purification occurred during isolation of the 4-azide 24b by ion-exchange chromatography, but the maximum purity achieved was 88%, and three impurities were still present. Reduction gave amine 25b in nearly quantitative yield and of 85% purity. It was still most efficient to attach the side chains and complete the purification on the target septacidin analogues. The product of coupling 25a and 25b with a p-nitrophenyl ester was in every case a solid that could be recrystallized or reprecipitated. Crude yields were excellent, but purification usually involved considerable losses in these unoptimized experiments. Each target was obtained pure by elemental analysis and homogeneous by thin-layer
Journal of Medicinal Chemistry, 1977, Vol. 20, No. 11 1365
Antitumor Septacidin Analogues Table V. Biological Test Results
Vs. cultured L1210 leukemia cellsb --
270538
Inhibition of DNA, ED.,. ..",uM . 0.40 (0.30) 1.6 1.8 2.5 20 29 >lo00 >lo00 >lo00 100 > 1000 >lo00
123127
1.5
Compd no.
NSCa
Id
65104 266219 268251 278178 267216 266218 271946 269447 269446 278179
2 3 4
5 6 7 8 9 10f 11 12
Adriamycing
no.
synthesis RNA, ED,,. _", .uM
1.0 (0.77) 0.69 1.1 1.6 5.5 18 > 1000 > 1000 > 1000 93 > 1000 >loo0 0.67
Vs. P388 leukemia in miceC
Cytotoxicity, IC". UM ~ " -
Optimum dose, qd 1-9, mdkg
Antitumor efficacy T/C, %
>5 > 10 > 20 > 100 3 > 10 250 > 500 > 500
0.50 (0.38) 0.50 1.0 0.50 1.0 1.0 (0.50- 16)e (3.1-50)" (3.1-50)" 1.6
154 149 160 136 133 137
> 100
( 0 .50-€qe
0.032
~~
1.0
Binding to isolated helical DNA in soln,b AT,,"C 111
1.7 0.6 3.6 0.5 0.4
156
1.9 0.5
193
13.6
Assay described in ref 4 , Protocol Assay described in ref 25. a Accession number of the National Cancer Institute. 1.200. Other doses were higher or lower by a factor of 2. Each compound had two to four active doses. Except for 7-9, compounds were of borderline solubility, and some doses were injected as suspensions. Water, water plus Tween, or saline plus Tween were used as vehicles. By definition, a compound is active if T/C 2 125. The T/C for 1 is the average from 13 tests; for 2, 3, 5, and 6 from two tests; for 4, 7-10, and 1 2 from single tests. Optimum dose is the one producing the highest T/C. Purity of 1 was estimated to be 75%, based on elemental analysis for N; corrected doses assuming inInsoluble compound, the active impurities are shown in parentheses. " Inactive doses; for 7-9 and 12, T/C = 88-116. Data from only one tested in suspension in vitro as well as in vivo. Cytotoxicity could not be determined for this reason. ref 25. The ATm of 13.6 "C was obtained by the modified procedure.
chromatography. Subsequently, purity of the targets was checked by HPLC. Most were from 98.5 to 99.6% pure, and 10;all from amine 25b) were found but some (4,5,8, to contain 6-12% of a single impurity that was undetectable by other means. Compound 2 was previously reported3 from the modification of 1, but there were no data that would permit comparison with the totally synthetic product. Ultraviolet extinctions of the synthetic septacidin analogues (Table 11) were consistently much higher than those in the previous work.3 Circular dichroism spectra (Table 111)for 1-12 were qualitatively nearly identical, differing only in the magnitude of their Cotton effects near 239 nm (positive) and 267 and 273 nm (negative). This confirms that the new synthetic analogues and septacidin (1) all had the same type of attachment of the chromophore to a chiral structure and, thus, share the 0-L-anomeric configuration. Table IV shows good correspondence of the lH NMR spectra for 1 and 2-12, with the sugar H-1 consistently at 6 5.4-5.6 as a doublet with Jl,2 = 7-8 Hz. (See paragraph a t end of paper regarding supplementary material for Tables 11-IV.) The side chains were chosen first of all to duplicate the side chain of 1 (as in 2 and 3) and then to explore the effect on biological properties of an unbranched fatty acid (4), of shorter chain fatty acids (5-7), of inserting ether 0 in a short chain (8 and 9), of extending the natural side chain by inserting one -CH2- in the amino acid (lo), of moving the fatty chain to the amino acid (1 l), and of omitting the amino acid (12). Biological Tests. The septacidin analogues were tested first in cultured lymphoid leukemia L1210 cells for inhibition of nucleic acid synthesis and for cytotoxicity, by previously reported procedure^.^^ A sample of natural septacidin (I) was obtained from the National Cancer Institutez6and included for comparison. The analogues were then tested, along with 1 as parent, against lymphocytic leukemia P388 implanted in mice, under the auspices of the NCI and according to ita protocols4 which use the increased survival time of treated animals compared to controls as the measure of antitumor efficacy.
Results are given in Table V. The mouse tests clearly show that synthetic analogues 2-6 and 10 retain the high antitumor potency of 1, with "confirmed activity" according to NCI criteria for 2-6 and 10 at doses of 0.5-1.6 mg/kg. In contrast, 7-9 and 12 were completely inactive at all doses tested. The in vitro tests for inhibition of nucleic acid synthesis showed the same clean break between active and inactive compounds but gave a wider range of response for the active compounds. Analogues 2-4 showed essentially identical responses; they were nearly the same as 1 in the RNA test but not as potent as 1 in the DNA test. Analogues 5 and 6 were noticeably less potent; 10 appeared to be only weakly active, but this may be due to its insolubility such that 10 had to be tested as a suspension. These results indicate that small changes at the terminal carbon of the sugar (CH3 or CHzOH, in place of CHOH-CHzOH in 1) have no effect on activity but that the nature of the fatty acid-amino acid side chain is critical. When the fatty acid was shortened from 16 carbons (2-4)to 12 carbons (5 and 6),there was decreased activity in vitro and a slight loss of efficacy in vivo. When shortened to six carbons (7) or the equivalent (8 and 91, loss of activity was total. Thus lipophilic character is important. The near identity of 2 and 4 suggests that chain branching is not required and that palmitic acid can be substituted in synthetic structures for the difficultly accessible isopalmitic. On the other hand, the lipophilic chain cannot be shifted from the N-acyl group to the a-carbon on the amino acid moiety as in 11 (in vitro test only), nor can the amino acid moiety simply be omitted as in 12. Apparently the side chain can be extended by inserting a carbon into the amino acid, judging from the in vivo test of 10. Both in vivo and in vitro data show that 2-6 and 10 are comparable in potency to the antibiotic adriamycin (Table V), chosen for comparison because of the great current interest in its anticancer properties. The in vivo efficacy of adriamycin is much higher, but there are other differences. Adriamycin is highly cytotoxic, while cytotoxic effects of 2-4,6-9,and 11 could not be observed at soluble
1366 Journal of Medicinal Chemistry, 1977, Vol. 20, No. 11
Acton, Ryan, Luetzow
rmm temperature for 18 h protected from moisture. It was washed with two 700-mL portions of ice water, 700 mL of saturated NaHCO, solution, and another 700 mL of water. The benzene solution was dried and evaporated to a syrup (-150 mL), which was dissolved in 100 mL of C1CH2CH2Cl. Petroleum ether (bp 60-110 "C) was added to the cloud point (-250 mL required), and the mixture was seeded and chilled a t 3 "C (seed crystals were obtained by scratching the syrup on a watch glass with ClCH&H&l-petroleum ether). After 18 h, an additional 100 mL of petroleum ether was added, and chilling was continued for 3 days. The white solid was collected by filtration to yield 50.6 g (84%): mp 124.5-126 "C; Rf 0.7 in solvent C (61); IR (Nujol) 5.77 (C=O), 7.33 (br),and 8.46 1.1 (OSO,); 'H NMR (CDC13)6 6.49 (d, H - l ) , 5.82 (quintet, H-2 and H-3 as ABMX, H-2-H-3 = 6 5.79-5.85), 5.28 (m, H-4), 4.62 (4, H-5), 3.04 (s, OSOzCH3),1.42 = 3.0, J2,3= 10.7, J3,4 = 2.0, J5,6 = 6.6 Hz). Anal. (d, CCH3) (J1,z Experimental Section (Cz3HziC108) C, H. 2,3-Di- 0-benzoyl-6-deoxy-4-0-methanesulfonyl-8-LMelting points were taken on a Fisher-Johns hot stage and are galactopyranosyl Azide (Ma). The chloro sugar 17a (54.6 g, uncorrected. Organic solutions were dried over MgS04 and 0.116 mol) was added to 500 mL of Me2S0, followed by 55 g (0.85 filtered. Evaporations or concentrations were carried out in vacuo mol) of NaN3, and the mixture was stirred at room temperature on a spin evaporator. Stoichiometry of reactions was based on for 18 h. The chloro sugar dissolved rapidly, and NaCl slowly the assumption that all intermediates were completely pure, even precipitated. The mixture was poured into 3.2 L of water, and though actual purity of some by HPLC was only 75%. Target stirring was continued for 1h. The white solid precipitate was compounds 2-12 were dried a t 140 "C (1 mm). collected on a filter and, while still wet, was dissolved in 2.8 L Spectra. IR spectra were determined routinely in Nujol mull of boiling MeOH. The solution was cooled in ice for 1h, and the on a Beckmann IR-4 spectrometer and were consistent with white precipitate was collected to yield 42 g (76%): mp 152.5-154 chemical structure. Benzoate bands (C=O) were near 5.8 1.1, NOz *C; Rf 0.50 in solvent C (1:l); IR (Nujol) 4.69 and 4.72 (N3),5.75 bands in 20a and 20b were observed a t 6.6 b, and strong N3 bands and 5.81 p (C=O); 'H NMR (CDC1,) 8 4.83 (d broadened, H-1, were always at 4.71 & 0.01 b, whether from 4-azidodeoxy sugars J1,z = 8 Hz), 3.04 (9, OSOZCHJ, 1.45 (d, CCH3, J =7.5 Hz). Anal. or from glycosyl azides. In the methanesulfonates, OSOz bands were hard to distinguish from other absorption. In the pyrimidines (C21HziN308) C, H, N. A second crop was obtained, 10.5 g (18%), mp 138-148 O C , and purines, heterocyclic absorption appeared as one or two strong identical except for a minor contaminant [R,0.55; 'H NMR 6 1.43 bands a t 6.14.3 p. Spectra of 2-12 were all nearly identical, with (d, CCHJ; I 5 % ] . predominant amide bands at 3.0,6.1, and 6.45 1.1. Proton magnetic 4-Amino-6-(2,3-di- 0-benzoyl-6-deoxy-4- 0- m e t h a n e resonance ('H NMR) was determined on Varian A-60A and, where sulfonyl-~-~-galactopyranosylamino)-5-nitropyrimidine indicated, XL-100 spectrometers. Solutions in CDC13or MeaO-d6 (20a). A solution of 10.5 g (0.022 mol) of 1-azide 18a in 100 mL were with Me4% (6 = 0.0) as internal reference; solutions in DzO of anhydrous benzene was treated with 1.1g of Pd black and were with Mel% external reference except as indicated for 15b shaken under H2at 2-3 atm for 7 days. The catalyst was removed and its p-anomer. Signals are designated as s (singlet),d (doublet), by fdtration and washed with 20 mL of benzene, and the combined t (triplet), q (quartet), or m (multiplet). Heating to 70 "C, as noted, filtrate containing amine 19a, R, 0.6 (ninhydrin-positive spot) in was often required to improve resolution, especially of H-1'. UV solvent C (1:2) compared to Rf 0.9 for 18a, was used immediately spectra were determined on Cary 11 or Cary 14 recording without exposure to moisture. Attempts to isolate the amine gave spectrophotometers. Circular dichroism was determined on a decomposition and a ninhydrin-negative by-product. Jasco ORD/UV-5 spectropolarimeter equipped with a Sproul 4-Amino-6-chloro-5-nitropyrirnidine (Aldrich; 4.0 g, 0.023 mol) Scientific SS-20 CD modification and a programmed 15-A slitwas dissolved in 100 mL of refluxing anhydrous benzene, 10 mL width control. Mass spectra were recorded on an LKB Model of benzene was removed by azeotropic distillation to ensure 9000 spectrometer at 12 eV. dryness, and the hot solution (sometimes containing a little Chromatography. Liquid column chromatography was done precipitate) was treated with the above solution of amine 19a using Bio-Sil A 2W325 mesh silica gel or Woelm neutral alumina. followed by 2.3 g (0.023mol) of Et3N. Another 10 mL of benzene TLC was done on 2 X 8 in. plates with 250-p layers of silica gel was distilled off and the mixture protected from moisture was GF (Uniplate) or aluminum oxide GF. Rjs are in solvent systems refluxed for 18 h. The mixture was cooled, treated with 1 mL A, CHC1,; B, CHC13-MeOH; and C, ClCHzCH2C1-EtOAc (ratios of Et3N and 1 mL of H20, and stirred overnight to convert any of mixed solvents given in parentheses). Spots were visualized unreacted 6-chloropyrimidine to the less soluble 6-hydroxy under UV light, except where otherwise noted. Analytical HPLC analogue. The latter and Et3N.HC1 were removed by filtration, was done on a Chromatronix Model 3500 liquid chromatograph and the filtrate was evaporated to dryness. The residue was with a Schoeffel SF770 Spectroflow variable wavelength UV dissolved in 300 mL of boiling 95% EtOH. The solution was detector. System X was a Spherisorb alumina column A5Y, 23 allowed to cool very slowly to room temperature, while much of X 0.64 cm o.d., with CH2Cl2-MeOH(99:l) as eluent. System Y the product separated as crystals. The mixture was kept at room was a 1.1 Bondapak/carbohydrate column, 30 X 0.64 cm, supplied temperature for 2 days and filtered to yield 9.1 g (70%) that was by Waters Associates and used with spectrograde CH3CN-Hz0 used in the next step: mp 158-165 "C; 'H NMR (CDC13)6 9.58 (4:l) as a reverse-phase eluent. System Y was used with targets (d, glycosyl NH, J = 8 Hz; exchanged with DzWD3COOD), 8.02 2-12 and with 24b, 25a, and 25b, and all showed A,, 264 nm for (s, pyrimidine H-2), 6.1-6.5 (m, H-1, H-2, H-3), 5.22 (br s, H-4), the HPLC peak; the peak for single impurities detected in 4, 5, 4.20 (9, H-5), 3.08 (9, OSOZCH3), 1.39 (d, CCHs, J5,6 = 6.2 Hz); and 7-10 had & 267-269 nm. System X was used on compounds TLC in solvent C (1:l)revealed traces of 19a of Rf 0.69 and with the sugar moiety blocked. 4-amino-6-chloro-5-nitropyrimidine of R, 0.75 in addition to 20a 2,3-Di- 0-benzoyl-6-deoxy-4-0-methanesulfonyl-a-Lof Rf 0.57. A sample for analysis was recrystallized from galactopyranosyl Chloride (17a). A solution of 60.0 g (0.129 C, mol) of methyl 2,3-di-0-benzoyl-6-deoxy-4-O-methanesulfonyl- EtOH-ClCH2CH2Cl mp 188-189 "C. Anal. (C25H24Nb010S) H, N. cu-L-galactopyranoside'6 (16a) [mp 161-163 C; 85% yield; 'H NMR 4,5-Diamino-6-(2,3-di-O-benzoyl-6-deoxy-4- O-methane(CDC13) 5.68 (q X d, H-2 and H-3 as ABXY, H-2-H-3 = 6 sulfonyl-B-L-galactopyranosy1amino)pyrimidine (21a) a n d 5.60-5.77), 5.20 (m, H-4), 5.12 (d, H-1), 4.30 (4, H-5), 3.43 (s, t h e a-Anomer 26. A solution of 10.1 g (0.017 mol) of nitro OCH3), 3.03 (s, OSOZCH,), 1.38 (d, CCHJ, J 1 , 2 = 3.3, J2.3 = 10.7, compound 20a in 200 mL of anhydrous pyridine was treated with J3,4= 2.3, J5,6 = 6.5 Hz (lit.16 mp 163-163.5 "C)] in 1.5 L of 20 mL of a pyridine slurry of Raney active nickel catalyst (Grace) anhydrous benzene at room temperature was treated with 10 mL (prewashed with pyridine to remove water). The mixture was of titanium tetrachloride and then a slow stream of anhydrous shaken under 2-3 atm of H2 for 2 h and filtered. The collected HCl gas for 30 min. The orange-yellow solution was stored at
levels. Adriamycin binds strongly to isolated helical DNA in solution (as measured25by its effect on the helix-coil transition temperature of the DNA, or ATrn)whereas 2-5, 8, 10, and 12 bind only weakly. This suggests that the potent inhibition of nucleic acid synthesis by 2-4 may be an indirect effect. Additional evidence that the active septacidin analogues do not attack nucleic acids directly is the absence of in vitro mutagenicity, observed when 2 and 5 were tested (along with the inactives 7 and 12) in Salmonella strains at levels up to 1 pmol per plate.*' Cancer drugs that are not mutagenic are of some irnportance, and further studies on the mechanism of action of septacidin analogues will be of considerable interest.
Antitumor Septacidin Analogues
Journal of Medicinal Chemistry, 1977, Vol. 20, No. 11 1367
anhydrous MeOH was protected from moisture and stirred at nickel was washed with 200 mL of pyridine, and the combined room temperature for 2 h and then treated with -500 g of 100-200 filtrates were evaporated to dryness. The residue was dissolved mesh Dowex 50 (H) to neutralize the base and absorb the purine. in 250 mL of CHC13. The solution was washed with 250 mL of After 1 h more of stirring, the resin was collected., washed with saturated aqueous NaC1, dried, and evaporated to dryness. The 200 mL of MeOH (to remove MeOBz), and added to a chrosolid residue was triturated with 150 mL of boiling benzene and matographic column, and the product was eluted with 8 L of -8 filtered while hot. The insoluble portion was washed with 25 mL N ammonium hydroxide. The eluate was concentrated with of hot benzene. The combined filtrate was chilled a t 3-5 "C for heating until a thick precipitate formed. The mixture was cooled 3 days to yield 7.0 g (73%)of an amorphous brown precipitate, and filtered to yield 44 g (42%). The filtrate was evaporated to mp 195-197 "C, that was homogeneous by TLC on alumina: R, dryness, and the residue was triturated with 50 mL of hot water 0.78 in solvent B (19:l); [ a ] " ~- 51.4" (c 1, MezSO); IR 3.0-3.2 on a steam bath. The suspension was filtered, the filtrate was (NH), 5.80 (C=O), 6.15 and 6.27 p (heterocycle), with absence chilled at 5 "C for 3 days, and the solid was collected to give an of absorption in the NOz region a t 5.5-5.6; 'H NMR (CDC13) 6 additional 11 g. The combined yield was 52% of homogeneous 7.97 (s, pyrimidine H-2), 6.42 (d, H-1), 6.0-5.5 (H-2, H-3), 5.2 (br 24a: mp 254-262 "C dec; Rf 0.7 in solvent B (9:l);IR 4.47 (wk) 8 , H-4), 4.22 (4, H-5), 3.05 (9, OSOZCH3), 1.40 (d, CCH3, J 5 , 6 = and 4.70 (str) (N3),6.1-6.2 p (purine), with absence of absorption 6.5 Hz); UV (in 95% EtOH) A, 216 nm (c 44 loo), 275 sh, 282 for C=O; 'H NMR (MezSO-d6, exchanged with DzO and 275 sh, 281 (12 900); UV (12 400); UV (with NaOH added) ,A, CD3COOD) 6 8.38 ( 8 ) and 8.28 (5) (H-2 and H-8),5.5 (d, H-l', Jy,r (with HCI added) ,A, 226 (40700), 284 (UOOO),312 (14200); CD = 8 f 1 Hz), 1.25 (d, 6'-CH3, J5t,6'= 5.5 Hz). A sample for analysis (in 95% EtOH) A 225 nm [[e] 52800 (deg L)/(cm mol)], 231 (01, was twice recrystallized from HzO: mp 261-264 OC dec. Anal. 239 (-52 800), 251 (0), 260 (5280), 285 (12 200); MS m/e 557 (M', ( C ~ ~ H ~ ~ N ~ O ~ *C, O .H, ~H N.Z O ) 4.7), 461 (M - CHaSOSH, 1.8), 356 (M - CH3SO3H - PhCO, 2.0), 6-(4-Amino-4,6-dideoxy-~-~-glucopyranosylamino)-9 H340 (M - CH3S03H - PhC02, 3.6). Anal. (CZ5Hz7N5O8S) C, H, purine (25a). A solution of 4.1 g (0.013 mol) of azide 24a in 40 N. mL of MeOH-HzO (1:l) was treated with 0.40 g of 10% Pd/C The benzene-insolublesolid above (0.4 g, 4%) showed a second and shaken under Hz a t 2-3 atm for 3 days. The catalyst was component, Rf 0.64, equal in intensity to 21 at Rf 0.78. Trituration collected on a filter and washed with 25 mL of HzO, and the with hot CHC13 removed 21, leaving the insoluble a-anomer 26 combined filtrate was evaporated to dryness. A solution of the (0.2 g, 2%): homogeneous on alumina, Rf 0.64 in B (19:l);[a]"D residue (4.0 g) in 50 mL of hot MeOH was clarified by filtration -191" (c 1, M e s o ) ; IR and MS nearly identical with those of 21a; and cooled. The resultant precipitate was collected to yield 1.66 'H NMR (CDC13)6 7.98 (5, pyrimidine H-2), 6.65 (d, H-l), 6.2-5.5 g (44%) of amorphous solid: free of IR bands for N3; 'H NMR (H-2, H-3), 5.1 (m H-4), 4.43 (q, H-5), 3.09 (s, OS02CHJ, 1.34 = (DzO) 6 8.11 (s) and 7.97 (s) (H-2 and H-B), 5.37 (d, H-l', J1,,2, (d, 6-CH3, 5 5 , 6 = 6.5 Hz); UV (in 95% EtOH) ,A, 217 nm (t 9 Hz), 1.24 (d, CCH3, J5',6' = 5.6 Hz). The filtrate was poured 42600), 275 sh, 283 (11600); UV (with NaOH'added) A,, 275 into 500 mL of ethyl ether to precipitate an additional 2.3 g (total sh, 281 (13 700); UV (with HC1 added) A,, 227 (41 200), 283 yield 61%) that was identical by IR and TLC, Rf 0.5 (UV ab(l0300), 314 (13000); CD (in 95% EtOH) X 225 nm [[e] 0 (deg sorbing and ninhydrin positive), with contaminants of Rf 0.1 and L)/(cm mol)], 229 (16200), 232 (O), 240 (-62700), 291 (-8580); 0.9 (UV absorbing only) in solvent B (1:l). MS as the per(Me3Si) MS m/e 557 (M', 2.0), 461 (M - CH3S03H, 2.7), 356 (M derivative suggested there was complete pyrolysis at the glycosidic CH3S03H - PhCO, 2.8), 340 (M - CH3SO3H - PhC02,3.6) Anal. bond of 25a in the injector at 300 "C: m/e 361 [dideoxy ( C ~ ~ H Z N ~ O ~ S *C,HH, ~ ON.) 6-(2,3-Di-0-benzoyl-4-O-methanesulfonyl-6-deoxy-~-~- Me3SiNH - bis(0Me3Si)hexosene],279 [bis(NMe3Si)adenine]. Purity by HPLC system Y was 95%,with three impurities (2.5 galactopyranosylamino)-9H-purine(22a). A mixture of 138 and 0.1%). g (0.247 mol) of diamine 21a, 250 mL of triethyl orthoformate, Methyl a-L-Galactopyranoside (15b). ~-Galactono-1,4and 250 mL of acetic anhydride was heated to reflux (slight lactone (13; Pfanstiehl) was reduced to L-galactose (14) with exotherm). The resultant solution was stirred and refluxed for sodium amalgam by modification of previous procedure^.'^ A 2 h, cooled, and evaporated in vacuo. The oily residue was solution of 300 g (1.68 mol) of 13 in 4 L of HzO was chilled to 0 dissolved in 2.5 L of hot benzene, and the solution was diluted OC and acidified to pH 2.5-3.0 with 3.6 N HzSO4. With vigorous to the cloud point with cyclohexane (-300 mL), clarified by stirring, 4.1 kg of 2.5-3.0% sodium amalgam was added in heating, allowed to cool slowly to room temperature, and stored 200-300-g portions during 1.5 h, while pH 2.5-3.0 was maintained for 24 h. The precipitate was collected by filtration to yield 105 by further additions of 3.6 N H2S04(2 L was required). Stirring g of dark brown solid, mp 155-167 "C. The filtrate was conwas continued for 1 h and the temperature rose to 15 "C. The centrated to 500 mL, diluted to the cloud point with more cysolution was decanted from the mercury, treated with decolorizing clohexane, chilled a t 5 "C for 8 days, and filtered to yield 35 g of a second crop, mp 152-165 "C (total yield 140 g, 99%). The charcoal, filtered, neutralized with 10 N NaOH to pH 6.7, and evaporated. The semisolid residue was stirred with 2 L of hot product, Rf 0.47 in solvent B ( 1 9 0 , was acceptable for use in the MeOH, the mixture was filtered, and the filtrate was evaporated. next step; it was free of 21a but showed minor impurities of Rf 0.53 (estimated 4%, perhaps the 9-isomer of 22a, formed in the A solution of the residue in 1 L of HzO was added to a column (1.9 X 46 cm) of RG-501-X8 (H, OH) mixed-bed resin, 20-50 mesh. reaction) and Rf 0.93. Reprecipitation by cooling a benzene solution afforded a homogeneous sample for analysis (25%yield): The column was eluted with HzO (4 L) until the eluate showed mp 156-178 "C; 'H NMR (CDC13,100 MHz) 6 8.54 (s) and 8.18 little or no 14 when spotted on paper and sprayed with AgN03-NaOH. The eluate was evaporated, the residue was (s) (H-2 and H-8), 6.44 (br s, H-l'), 5.9-5.6 (H-2', H-3'), 4.31 (q, dissolved in 600 mL of MeOH, and the solution was chilled a t H-59, 3.10 (s, OS02CH,), 1.48 (d, CCH3, J5',6' = 6.5 Hz); UV (in 95% EtOH) ,A, 232 nm (t 28500), 264 (19600); UV (with HC1 0 "C for 18 h. The solid was collected to yield 168 g (56%). The added) A, 231 (28 400), 278 (19 700); UV (with NaOH added) filtrate gave a second crop upon concentration to 200 mL and chilling, 32 g (total yield, 66%). The product was identified as A,, 278 (17000). Anal. (Cz6Hz5N508S)C, H, N. 6-(4-Azido-2,3-di-0 -benzoyl-4,6-dideoxy-~-~-gluco- L-galactose (14), containing 5 1 0 % galactitol, by gas chromatographic comparison with D-galaCtOSe.28 pyranosy1amino)-9 H-purine (23a). A mixture of 14.0 g (24.4 mmol) of 22a and 7.2 g (110 mmol) of NaN3 in 72 mL of MezSO A solution of 50 g of 14 in 500 mL of MeOH was treated with was stirred and heated a t 100 "C for 18 h and poured into 1 L -75 mL of Dowex 50 (H) resin, refluxed for 3 days, and filtered. of ice and water. After 30 min of stirring, the precipitate was The filtrate waa evaporated to dryness. Processing as in an elegant collected and dissolved in 200 mL of CHC13. The solution was studym of the D series is recommended over the earlier methods. washed with 100 mL of saturated brine, dried, and evaporated The a-anomer 15b was obtained as the monohydrate: mp 95-105 to yield 13.8 g (109%)of residual foamed glass; TLC, Rf 0.50 in "C; 'H NMR (DzO, internal Me3SiCHzCHzCHzS03Na)6 4.86 solvent B (19:1), showed absence of 22a but revealed a minor (unresolved d, H-l), 3.41 (s, OCH3). Drying a t 63 "C (1 mm) for impurity of R, 0.53 that could not be removed by preparative TLC 3 days gave the anhydrous form analytically pure: mp 90-98 "C; or attempts a t crystallization. [aIz0~ -181" (c 1, HzO). Anal. (C7H1406) C, H. For the D6-(4-Azido-4,6-dideoxy-~-~-glucopyranosylamino)-9 Hmonohydrate, l k m mp was 90-103 OC and [ a ] = D was +178.1° (0.1 purine (24a). A solution of 175 g (0.340 mol) of 23a and 27 g M, H20);'H NMR30 was a t 6 4.83 and 3.39. For the anhydrous D form, lit.29mp was 107-112 or 116-116.5 OC. (0.50 mol) of NaOMe (Aldrich, anhydrous powder) in 1.2 L of
1368 Journal of Medicinal Chemistry, 1977, Vol. 20, No. 11 From the mother liquors, methyl /3+galactopyranoside was obtained in low yield mp 177-181 "C; 'H NMR (DzO, internal Me3SiCHzCH2CHzS03Na) 6 4.32 (4, H-l), 3.60 (s, OCH,); [ N ] +0.2" ( c 1, HzO). Anal. (C7H1406"20) C, H. For the D enantiomer, lit." mp was 177-180 "C and [ N ] ' ~was ~ --0.3 to -0.5' (0.1 M, H2O); 'H NMR3' was at 6 4.30 and 3.56. Methyl 2,3,6-Tri-0-benzoyl-a-L-galactopyranoside. By a p r ~ c e d u r efor ' ~ the D enantiomer, the yield was 27%: mp 136-138 "c; [.InD -118" (c 1,CHCl,); identical with that of the D form;" 'H NMR (CDC13)6 5.83 (s, H-2, H-3, half-width at 4 Hz), 5.28 (s, H-1, half-width a t 4 Hz), 4.8-4.3 (H-4, H-5, H2-6), 3.47 (s, OCH,); for the D form, mp 139-14015 and 135.5-137.0 'C," +123" (CHC13)'5 and +120' (CHClJ." Anal. (Cz&605) C, H. Methyl 2,3,6-Tri-0-benzoyl-4-0-methanesulfonyl-a-r.galactopyranoside (16b). procedure^'^*'^ for the D enantiomer were modified by the use of 1.7 molar equiv of methanesulfonyl chloride, keeping the temperature below 10 "C during addition. The reaction mixture was stirred overnight and then heated to 60 "C for 30 min. Hydrolysis in ice water gave a precipitate that was collected on a filter but melted a t room temperature. The resultant syrup was dissolved in CHC13 (10 mL/g) with heating and processed. The product was recrystallized from benzenepetroleum ether (bp 30-60 "C) to give a 73% yield: mp 142-144 "c;[.I2'D -100' (c 1,CHCl3); identical with that of the D form;" 'H NMR (CDC13)6 5.9-5.65 (m, H-2 and H-3), 5.51 (m, H-4), 5.23 (d, H-1, J = 3 Hz), 3.44 (s, OCH,), 3.06 (s, OS02CH3);for the D form, mp 143-14415 and 142-143 "C," DI.[ +102" (CHCl,). Anal. (CZ&Z~OIIS) C, H. 2,3,6-Tri-0-benzoyl-4-0-methanesulfonyl-a-1,-galactopyranosyl Chloride (17b). A stirred solution of 270 g (0.462 mol) of 16b in 2 L of anhydrous benzene protected from moisture was treated with 30 mL of titanium tetrachloride and then a stream of anhydrous HC1 gas for 1 h. After 20 h at room temperature (without stirring) the HCl was reintroduced for 30 min and again after 4 days. As measured by 'H NMR, conversion to the chloro sugar was 71% complete after 3 days, but 6 days were required for 92% completion. The solution was poured into 2 L of ice water. As for 17a,benzene extracts were washed with NaHC03 solution and water, dried, and evaporated. The residual syrup, 290 g (106%of theory), contained a little benzene: 'H NMR (CDC13)6 6.55 (d, H-1, J1,2 = 3.2 Hz), 5.9 (rough t, H-2 and H-3j, 5.58 (rough d, H-4, J = 2.5 Hz), 3.00 (s, OS02CH3);there was no evidence of a signal for H-1 of the P-anomer (estimated