The synthesis and the preliminary microbiological activity of some tryciclic 13-1actam derivatives designed combining the structural features of the natural carbapenems and the synthetic "tribactams" are described.The carbapenem thienamycin 1, isolated in 1976 from Streptomyces cattleya 1, was found to exhibit a broad spectrum antibacterial activity, including Pseudomonas aeruginosa, and remarkable stability towards bacterial [5-1actamases. Its discovery resulted in an increased synthetic effort devoted to the identification of new carbapenems 2 with an improved chemical stability and decreased susceptibility to renal dehydropeptidase-I (DHP-I). In this regard, the N-formimidoyl analogue of thienamycin, imipenem 2 (MK-0787), showed a greater chemical stability 3, but was susceptible to DHP-I, whereas the 1 p-methyl derivative, meropenem 3 (SM-7338) was found to be fairly insensitive to DHP-I, while retaining excellent antibacterial activity 4.Figure 1Some years ago researchers at Glaxo 5 designed the tricyclic derivatives (tribactams) of type 4 (Fig. 1), with the aim of further increasing the metabolic stability and to investigate the consequence of a more constrained ~ lactam derivative in enhancing the electtvphilicity of the [3-1actam carbonyl and the antibacterial activity.In particular, GG-326 4 (Fig. 1, R=OCH 3) is a clinically promising 13-1actam antibacterial agent, with a particularly broad-spectrum, resistance to 13-1actamases and stability to dihydropeptida~s.In view of the efficacy of these novel antibacterial agents, it was of interest to consider similar tricyclic derivatives of general structure 5 (Fig. 1), bearing a suitable heteroatom at the vinylic position, as in the case of some carbapenems and some synthetically available cephalosporins 6. Considering the highly strained nature of these derivatives, a potential problem could be their chemical instability. Therefore, betbre introducing suitable substituents to mimic the side chain of thienamycin, the unsubstituted derivative 5 (Fig. l, R=H), was synthesized to test the chemical feseability of this novel class of 4-heterosubstituted tfibactams.The first synthetic approach followed is showed in Scheme 1. Addition of two equivalents of the lithium enolate derived from the "~'--thiobutyrolactone 7 7 to the commercially available acetoxyazetidinone derivative 8 6, gave a 1:1 mixture of the 7~ and 7~ epimers in 91% yield. The mixture of 8 and 9 was smoothly transformed into the corresponding phosporane derivatives 10 and 11, according to the Woodward's three step procedure 9 and submitted to a Wittig-type cyclization, 10 without any attempt to separate the two isomers. The ring closure occurred in moderate yield by heating the mixture of phosphoranes in o-xylene at reflux for 5 hrs, 11 giving the highly strained tricyclic derivatives 12 and 1312, which were easily separable by MPLC.It is worth underlining that the more strained 713 isomer 13 was found to be unstable on silica gel and its isolation necessitated buffering the silica gel with AcOEt-TEA 95:5 before the purification of the reaction mixture.The removal of the silyl protecting group of the secondary alcohol was known to be problematic on carbapenems, due to the subsu-ate instability under several reaction conditions. Based on a reported procedure in the synthesis of thienamycin 13, the 7~ isomer 14 was isolated in 54% yield using an excess of TBAF in THF in the presence of AcOH at room temperature ftw 48 hrs. The final potassium salt 15 was obtained by a deallylation reaction catalyzed by Pd(PPh3) 4 in the presence of potassium 2-ethylhexanoate. 14 Despite the highly strained nature of this 13-1actam tricyclic system 15 was found to be completely stable in phosphate buffer at pH 7.4 at 37 o C.As far as the synthesis of the corresponding 713 analogue 16 is concerned, extensive decomposition occurred during the desilylation reaction of the intermediate 13, using the same reaction conditions for the deprotection of 12. However, it was possible to prepare 16 by first replacing the TBDMS protecting group with the more labile TMS (Scheme 2).The phosphoranes 10 and 11 were separated by MPLC and the TBDMS group easily removed using hydrolytic conditions. Treatment of 11 with 1:1:2 mixture of TFA-H20-AcOEt at room temperature for 24 hrs gave the con-esponding fi'ee hydroxy derivative 17 in 89% yield.Attempts to cyclize this deprotected phosphorane derivative in a variety of aprotic or protic solvents resulted in extensive decomposition and it was necessary to reprotect the secondary alcoholic function using TMSOTf and TEA at -78 o C, giving 18 in 97% yield. Cyclization, as reported above, by heating in o-xylene at reflux for 3 hrs, gave the tricyclic derivative 19 Jr, 47% yield. As expected, the next deprotection of the TMS group was found to be faster with respect to the corresponding TBMDS derivative. Instantaneous reaction occurred at -40°C using the same reagents (TBAF-AcOH), to give the unstable alcohol derivative 16 (extensive decomposition occurred just by storage overnight even at low temperature!), in moderate yield after chromatographic purification.a) i) 7 (2eq), LiHMDS (2eq), -78°(2, 30 min, ii) add 6, -78°C, lh, 91%; b) Allylglyoxylate, toluene, reflux (Dean-Stark), 3h; c) SOCI 2, 2,6-Lutidine, -40°(2, 30 min; d) PPh 3, 2,6-Lutidine, THF, 50°(2, 16 h, 64% (three steps); e) o-xylene, reflux, 5h, 25%; f) from 11 to 13, TBAF (5eq), AcOH (5eq), THF, r.t., 48 h, 54%; g) Pd(PPh3) 4, potassium 2-ethylhexanoate, PPh 3, r.t., 25 min, 87%.Deallylation of 16 gave the target compound 20 which was difficult to purity due to its instability of the in solid phase and in buffered solution (15% yield by HPLC).Considering the observed lability of 16 and 20 no further attempts to optimize the synthetic route were performed, trying to improve the yield of the removal of the trimethylsilyl protecting group under different reaction conditions. Therefore the initial objective of preparing suitably substituted analogues of 5 (Fig.l), as conformationally restricted analogues of thienamycin, was di~ontinued 15.To understand the reason for the reduced chemical stability of 20 with respect to the corresponding epimer 15, the two compounds were modelled with Sybyl 6.0316 on an Evans and Sutherland 3+, modifyinga) TFA, H20, AcOEt 1:1:2, r.t., 12 h, 89%; b) TMSOTf, TEA, THF, -78~, 20 min, 97%; c) o-xylene, reflux, 3h, 47%; d) TBAF (1 eq), AcOH (l.leq), THF, -40~C, 5 rain, 25%; e) Pd(PPh3) 4, potassium 2-ethylhexanoate, PPh 3, r.t., 30 min.the crystal structure of GG-326 4 (R=OCH3). After the definition of the ring A and B as an aggregate, the six membered ring of 4 was replaced with the 4-heterosubstituted five membered ring and the two epimers were minimized using a MAXMIN2 routine. Force ficld calculations showed that 15 lies 5.6 Kcal/mol below 20 (Fig. 2) and that the additional strain energy is mainly localized to the bending contributions (4 Kcal/mol and 1 Kcal/mol for the for the angle bending and out of plane bending, respectively). These results seem to be in agreement with the experimental observations of stability.The evaluation of preliminary antibacterial activity was performed only on the stable 7ct isomer 15, using a selection of Gram positive and Gram negative bacteria. Weak activity was observed against sensitive and resistent strains of S. aureus (2-32 lag/mL), S. epidermidis (2 lag/mL) and S. pneumoniae (2 lag/mL). Low level of antibacterial activity was obtained against E. coli 851, 1850-WT, 1852-PM and 1919-PM/TEM-1 (16 I.tg/mL). Moreover, as expected, 15 was found inactive against P. aeruginosa (>32 I.tg/mL). Concerning the metabolic stability towards renal DHP-I, 15 was found to be twice more sensitive to human kidney DHP-I than imipenem 17,18.