Side chain modifications in lankacidin group antibiotics.

Novel N-acyl analogs of lankacidin may be prepared from 3-isocyanatolankone diformate [7,13-bis(formyloxy)-2-isocyanato-1,4,10,19-tetramethyl-16- oxabicyclo[13.2.2.]nonadeca-3,5,9,11-tetraen-17,18-dione]. Of seven such analogs evaluated in vitro only homolankacidin diformate showed significant activity. However, in a cell-free system two of the inactive analogs inhibited polypeptide synthesis as well as did lankacidin itself or erythromycin. Antibacterial activity, therefore, is a function of the ability of a congener to penetrate the bacterial cell membrane in addition to its intrinsic activity. Similarly, lankacidinol is as potent as lankacidin or erythromycin as an inhibitor of bacterial polypeptide synthesis in a cell-free system. This intrinsic activity is expressed as potent antibacterial activity against growing gram-positive cultures in O(2')-acyl derivatives with the proper lipophilicity.

Novel N-acyl analogs of lankacidin may be prepared from 3-isocyanatolankone diformate [7,13-bis(formyloxy) -2 -isocyanato -1,4,10,19 -tetramethyl -16 -oxabicyclo [13.2.2]nonadeca -3,5,9,11 -tetraen -17, 18dione]. Of seven such analogs evaluated in vitro only homolankacidin diformate showed significant activity. However, in a cell-free system two of the inactive analogs inhibited polypeptide synthesis as well as did lankacidin itself or erythromycin. Antibacterial activity, therefore, is a function of the ability of a congener to penetrate the bacterial cell membrane in addition to its intrinsic activity. Similarly, lankacidinol is as potent as lankacidin or erythromycin as an inhibitor of bacterial polypeptide synthesis in a cell-free system. This intrinsic activity is expressed as potent antibacterial activity against growing gram-positive cultures in 0(2')-acyl derivatives with the proper lipophilicity.
An ideal antibiotic for treating common respiratory tract infections would be active against pathogenic strains of the following species, including those resistant to commercial agents such as ampicillin and erythromycin: Staphylococcus aureus, Streptococcus pyogenes, and Streptococcus pneumoniae (gram positive) and Haemophilus influenzae (gram negative). In addition, oral activity is required. Lankacidin, a 17-membered ring macrolide that has been known since 1960 (5,7,9,11,16,19), possesses these properties to various degrees, and although not of commercial caliber itself, it affords an opportunity for semisynthesis with the aim of discovering a derivative worthy of clinical development. Earlier work with lankacidin involved its degradation, oxidation, and reduction products (6,8), all showing greatly reduced activity. Some esters have been described (10), but at best these represent only pro-drug forms of the parent. Chemical progress in this area is difficult owing to the lability of lankacidin group antibiotics to even mildly acidic or basic conditions. Lankacidinol (3), a reduction product of lankacidin, was first reported in 1969 (7) and is greatly inferior to lankacidin in both in vitro and in vivo evaluations. However, as will be shown in the present work, both lankacidin and lankacidinol are at least equal to erythromycin as inhibitors of polypeptide synthesis in a cell-free system. Hence, the therapeutic potential of 17-membered ring macrolides has not yet been realized.
Although we are not now able to report the discovery of a derivative for clinical study, the present work discloses some chemical procedures for modifying the side chain of lankacidin in an original way and describes derivatives of lankacidinol which are highly potent in in vitro tests.

MATERIALS AND METHODS
General. All thin-layer and column chromatographies were performed with silica gel as the absorbent. Unless otherwise indicated methanol was the solvent used in UV absorption studies, chloroform was used for infrared ( Therefore, samples not giving UV absorption intensities of this degree or showing absorption at longer wavelengths are easily recognized as being impure. The compounds described below were purified to single-spot materials (TLC analysis) with satisfactory UV spectra. In no case did impurities interfere with the assignment of the chemical structure by spectrographic methods.
Structure determination. All new compounds have been assigned structures consistent with their IR, UV, and 1H and l3C NMR spectra. Reference 1H and 13C NMR spectra are available in the literature (6,8,18). Only those features in the spectra most critical to the structural assignments of new compounds will be noted below.
(i) Lankacidin (1). Lankacidin (1)   (ii) Lankacidin diformate (2). A solution of 4.59 g (0.01 mol) of lankacidin and 50 ml of pyridine was cooled to 0 to 5°C. With magnetic stirring, 10 ml of acetoformic anhydride was added cautiously. The reaction solution was allowed to warm to room temperature. After 1 h, the solution was poured into 600 ml of water to produce a colorless solid which was filtered and washed with water. The solid was taken up in 100 ml of CH2Cl2 and washed successively with 50 ml of water, 100 ml of water at pH 4 (adjusted with 1 N HCl), 50 ml of water, and 50 ml of saturated aqueous NaCl.
The organic phase was dried over anhydrous Na2SO4, filtered, and evaporated. The residue was triturated with diethyl ether (Et2O), and the Et2O-insoluble material was filtered and discarded. (iii) Lankacidin oxime diformate (3). A solution of 1.00 g (2.13 mmol) of 2, 167 mg (2.4 mmol) of HONH2 * HCl, 1.0 ml of pyridine, and 20 ml of methanol (MeOH) was stirred at room temperature. After 20 min, a mixture of the syn-and antiisomers of the oxime precipitated: yield of 430 mg (42%). The more polar and less polar isomers were readily separated from each other and from 2 by TLC (eluant, benzene-EtOAc-diethylamine [10:10:1]). Column chromatography (benzene-EtOAc [3:1]) of a 2.0-g sample afforded 1.14 g of the pure crystalline less polar isomer and 0.32 g of the more polar product containing a small amount of its isomer. Subsequent experiments with these samples showed that the isomers and their mixture were equally effective as substrates for preparing 3-isocyanatolankone diformate (see below). Therefore, in all further work the mixture of isomers was used to prepare the isocyanate. (iv) 3-Isocyanatolankone diformate (4). An oil bath was heated to 110°C. In a 50-ml single-neck, round-bottom flask equipped with a reflux condenser and a boiling chip, a solution of 1.00 g (1.9 mmol) of 3,0.75 ml of pyridine, and 20 ml of freshly prepared ethanol-free chloroform was treated with 0.20 ml (2.7 mmol) of SOCl2. The flask was placed immediately in the oil bath, and the solution was allowed to heat under reflux for 3 min. The flask was then cooled rapidly in an ice bath, and the contents were washed once with water and once with saturated aqueous NaCl. The organic phase was dried over anhydrous Na2SO4, filtered, and evaporated. The residue was stirred with 100 ml of Et2O for 1 h. Insoluble material was filtered, and the ether was evaporated to furnish 4: yield of 460 mg (51%); IR 2247 (NCO), 1755, 1722 cm-'; UV max 228 nm (log e 4.59); 1H NMR (lack of COCOCH3); 13C NMR (lack of NHCOC-(=NOH)CH3).

v) 3-[(3,5-Di-tert-butyl-4-hydroxybenzyl)oxycarbonylami
(vii) 3-Formamidolankone diformate (7, R = H). A solution of the isocyanate (690 mg, 1.46 mmol) 4 in 25 ml of ethanolfree CHCl3 was treated with 0.1 ml of formic acid and 0.3 ml of acetoformic anhydride and allowed to stand at room temperature for 5 h. The solution was then washed successively with water, saturated aqueous NaHCO3, water, and saturated aqueous NaCl. The organic phase was dried over anhydrous Na2SO4, filtered, and evaporated to furnish a yellow solid which was chromatographed (EtOAc-hexane (viii) 3-(1,2-Dioxobutylamino)lankone diformate (7, R = COCH2CH3; homolankacidin diformate). In a manner similar to the preparation described above, the title com- (ix) 3-Acetamidolankone diformate (7, R = CH3). The carbamate 5 was prepared in situ from a solution of 1.03 g (2.2 mmol) of the isocyanate 4, 568 mg (2.4 mmol) of 3,5-ditert-butyl-4-hydroxybenzyl alcohol, and 0.1 ml of dibutyltin dilaurate in 40 ml of CH2Cl2. After 3 days at room temperature when the isocyanate was no longer evident (TLC or IR), the solution was treated successively with 0.2 ml of acetyl chloride, 0.6 ml of AC2O, and 0.8 ml of diisopropylethylamine. After 2 days at room temperature, the volatile components were evaporated under reduced pressure, and the residue was chromatographed (EtOAc) to furnish a light-  (xiii) Lankacidinol diformate (9). A magnetically stirred solution of 1.0 g (1.9 mmol) of 2, 30 ml of MeOH, and 30 ml of THF was cooled to 5°C and treated with 18 mg (0.47 mmol) of NaBH4. After 30 min, the solution was evaporated under reduced pressure. The residue was taken up in CH2Cl2 and washed with water and saturated aqueous NaCl. The organic phase was dried over anhydrous Na2SO4, filtered, and evaporated to afford 0.6 g of pale-yellow foam. This material was chromatographed (eluant, EtOAc) to furnish the less polar epimer of lankacidinol diformate (9A; 140 mg, 4 OH Fe(III) or 0 Sn(IV) OH 14% yield) and the more polar epimer (9B; 204 mg, 20% yield). There was also an intermediate fraction containing both epimers (10%).
(xiv) Lankacidinol (10). In a manner similar to that for preparing 9, 1 was reduced by NaBH4 to afford 10 as a mixture of less polar and more polar epimers. To visualize these epimers on TLC it was necessary to elute the plate three times with EtOAc. To obtain the pure epimers it was expedient to hydrolyze the pure epimeric diformate derivatives 9A and 9B.
(xv) 10A. A solution of 180 mg of 9A, 5 ml of MeOH, 5 ml of THF, 1 ml of water, and 15 drops of 1 N K2CO3 was allowed to stand at room temperature for 15 min. The solution was acidified with 3 drops of 6 N HCl and evaporated under reduced pressure to furnish solids in an aqueous suspension. The mixture was filtered, washed thoroughly with water, and allowed to dry. The pure less polar epimer of lankacidinol (1OA) was obtained: yield of 110 mg (69%); mp 184 to 185°C (mp 178 to 179°C [3]).
(xvii) 0(2')-acyl lankacidinol diformates (11)  VOL. 25,1984 allowed to stand at room temperature for 24 h. The solution was poured into 100 ml of water, and the organic matter was extracted from the aqueous suspension with CH2Cl2. The organic phase was washed successively with 2 N HCl, water, and saturated aqueous NaCl. It then was dried over anhydrous Na2SO4, filtered and evaporated to furnish nearly pure 11. Chromatography (EtOAc-hexane) afforded an epimeric mixture of 11 (yield of 50 to 65%).
(xviii) 0(2')-acyl lankacidinol diformates (11) from acid chlorides. A solution of 1.5 g of 9 in 15 ml of pyridine and 15 ml of CH2Cl2 was cooled to 5°C and treated with 2 to 4 ml of the acid chloride. After 2 to 3 h, the reaction solution was poured into 100 ml of water, 2 N HCl was added to bring the mixture to pH 4, and the mixture was then extracted with CH2Cl2. After washing, drying, filtering, evaporating, and column chromatography, 11 as an epimeric mixture was obtained (yield of ca. 50%).
(xix) 0(2')-acyl lankacidinols (12). A solution of 1.0 g of 11, 10 ml of THF, 10 ml of MeOH, 5 ml of water, and 2 ml of 1 N K2CO3 was allowed to stand at room temperature for 20 min. The solution changed from pH 10.8 to 9.8 during this time, was adjusted to pH 4.5 by the addition of 6 N HCl, and was concentrated under reduced pressure to furnish an aqueous suspension of organic solids. More water was added, and the mixture was extracted with CH2Cl2. The organic phase was dried over anhydrous Na2SO4, filtered, and evaporated. The residue was chromatographed to afford an epimeric mixture of 12 (yield of 42 to 82%). MICs were determined by standard agar plate dilution techniques (15).
Studies on the ability of macrolide antibiotics to inhibit cell-free MS2 viral mRNA-directed polypeptide synthesis were conducted as described by English and co-workers (2). Cell extract (S 30) was prepared from Escherichia coli MRE 600 RNase I as described elsewhere (14). A 3H-amino acid mixture was used as a source of radioactivity.

RESULTS AND DISCUSSION
In general, there are severe limitations to chemically modifying lankacidin because of its sensitivity to even mildly acidic or basic conditions. Outside of the range pH 3 to 12, the compound decomposes rapidly; at pH 4 or 11 it can be manipulated at room temperature for up to 20 min without  significant degradation; within the range pH 5 to 8 the compound is stable for many hours at room temperature. Formyl satisfactorily protects the hydroxy groups. Not only is 2 easily prepared in good yield, but at later stages in the semisynthesis sequence the formyl groups can be removed under mild chemical conditions. In contrast, acetate esters, although easily formed, cannot be hydrolyzed chemically under conditions compatible with the stability of the lankacidin system. The earlier literature reports that such esters can be hydrolyzed by enzymatic methods (4,13), but that is much less convenient. The diformate esters of lankacidin group compounds may also be tested in vitro directly with only the loss of one or two levels in the dilution sequence; diacetate esters are generally much less active (Table 1 and reference 10).
One of the more attractive approaches to side chain modification would be to remove the pyruvoyl group to produce the free amine, 3-aminolankone diformate, (Fig. 1,  number 6). Such an intermediate could then be acylated with a variety of agents to produce a broad range of novel derivatives. Our initial attempts to remove pyruvoyl by classical methods demonstrated the extreme resistance of the amide group to electrophilic agents. As a definitive experiment, lankacidin diacetate in CH2Cl2 was treated with triethyloxonium tetrafluoroborate. The solution changed from colorless to light purple (12 min) to dark purple (32 min). Samples were taken at 2, 12, and 32 min and evaluated by IR spectroscopy. Although many of the absorption bands changed during this obvious decomposition, the bands at 3390 and 1680 cm-', those associated with the amide function, remained unaltered in position and intensity. Clearly, imidate ester formation did not take place.
These results discouraged further attempts to directly break the N-CO bond and caused us to consider an indirect method. Swiss workers (17) reported removing the side chain of nocardicin A by means of a second order Beckmann rearrangement. Although the conditions employed there are much too severe for lankacidin, we followed a similar course with success. Figure 1 outlines the methods used to prepare methanolic solution. Alcohols reacted more slowly to produce carbamates which are more stable than the ureas but nevertheless also decomposed in polar solvents. It has long been known that isocyanates react directly with carboxylic acids to produce amides. Babusiaux and co-workers (1) published a definitive study of this reaction, and their ideas are incorporated in the latter half of Fig. 1 to illustrate how two compounds in the present work were prepared. Formic acid and the isocyanate 4 reacted within 5 h to produce the formamide 7, R = H in a 33% yield. The  reaction with 2-oxobutyric acid to produce 7, R = COCH2CH3 in an 8% yield took 3 days. Under these conditions, acetic and propionic acids worked even more poorly.
To improve the acylation results a novel scheme was used to produce 3-aminolankone diformate (6) in situ under mild conditions. Advantage was taken of the fact that carbamates can be formed cleanly and relatively quickly when iron or tin salts are used as catalysts. Further, a method was needed to release the amine from the carbamate under mild conditions; the one eventually used (Fig. 2) was modelled after the work of Kemp and Hoyng (12). Starting with the isocyanate, the amides are prepared in a three-step, one-pot reaction. The 3acylaminolankone diformates (7) were hydrolyzed in aqueous THF-MeOH starting at pH 10.8 and dropping naturally to pH 9.8 over a 45-min period to prepare 3-acylaminolankones. The low yields were a result of sacrificing some product to ensure complete removal of the formyl groups. Table 2 summarizes some in vitro test results for the new compounds prepared. MICs against Staphylococcus aureus 01A005 (an erythromycin-susceptible organism) and Staphylococcus aureus 01A400 (an erythromycin-resistant organism) are typical for a wide variety of gram-positive pathogenic isolates. With the exception of 7, R = COCH2CH3 the MIC data suggest that structural modifications of lankacidin greatly reduce potency. This may be due to the inability of the derivatives to penetrate the bacterial cell membrane rather than the loss of intrinsic activity per se. Studies of antibacterial action in a cell-free system should shed some light on this question. 8, R = CH3 and 8, R = CH2CH3 inhibited viral mRNA-directed polypeptide synthesis in a cell-free system as well as did lankacidin itself or erythromycin ( Table 3). The data also show that analogs with the C-14 hydroxy group protected do not express intrinsic activity. Apparently, during MIC incubations, the formyl group is hydrolyzed much better than is acetyl.
This observation of potent intrinsic activity in lankacidin derivatives that otherwise would be considered inactive raised the question as to whether lankacidinol was truly inactive at the site of action (the ribosome) or whether lankacidinol was merely another case of an intrinsically active antibiotic not being able to penetrate the bacterial cell membrane. Experiments demonstrate that lankacidinol is a highly potent inhibitor of cell-free polypeptide synthesis, and it makes no difference whether the more polar or less polar epimer is evaluated (Table 3). Thus, a new possibility for a clinically effective member of the lankacidin group antibiotics emerged. Because lankacidinol is more hydrophilic than its parent, lankacidin, it seemed probable that derivatives of lankacidinol with greater lipophilic character would exhibit the desired in vitro activity, that is, show significant MICs.
Simple esters of lankacidinol should meet this requirement. However, esterification of the C-14 hydroxyl group results in the loss of both in vitro and intrinsic activity among lankacidin derivatives (10 and this work). Esterification of the C-8 hydroxyl group is compatible with activity (10 and this work), but selective esterification at C-8 is limited in scope and requires enzymatic methods (4,13) or tedious chromatographic separation of isomers from partially esterified compounds.
Until now a convenient method to prepare esters at the 2' position selectively has not been available. A straightforward procedure is outlined in Fig. 3. Following this approach, we prepared 0(2')-acyl derivatives of 11 and 12 and evaluated them in vitro as antibacterial agents (Tables 4 and  5). As in the lankacidin series, the diformates could be tested directly in vitro with the expectation of detecting activity, although with some loss of potency. It is evident from the data in Table 5 that optimum potency was achieved with 12 [R = (CH2)7-H]. This derivative was not only equal to lankacidin in activity against Staphylococcus aureus strains but was dramatically superior in potency against Streptococcus pyogenes.