*3.4. Cycloserine—Target: Peptidoglycan*

*Discovery.* In contrast to INH, EMB and ETH, which are synthetic compounds, cycloserine is a natural product produced by *Streptomyces* with broad-spectrum activity. Cycloserine (D-4-amino-3-isoxazolidone, also called oxamycin or seromycin) was isolated from *Streptomyces garyphalus* **[139]**, *Streptomyces lavendulae* **[140]**, *Streptomyces roseochromogenus* [141] and *Streptomyces orchidaceus* [142] in the early 1950s.

*Activity and toxicity.* Cycloserine is an oral drug, highly soluble in water (877 g/L). Cycloserine is a bacteriostatic second-line TB drug, with a MIC ranging from 50 to 250 μm against drug-susceptible *M. tuberculosis*. An analog of cycloserine, terizidone, a compound made of two molecules of cycloserine, is also used as a second-line TB drug to treat MDR TB. Cycloserine was shown to inhibit the growth of *M. tuberculosis* in vitro [143]. In vivo, cycloserine is rather inactive in a mouse or guinea pig model of *M. tuberculosis* infection [143–145] but effective in *M. tuberculosis*-infected monkeys and humans [146–148]. The differences in cycloserine activity in animal models were correlated with levels of D-alanine in sera [149] and rate of cycloserine excretion [146]. In a cohort study in China, TB patients with MDR TB had a better outcome when cycloserine was added to the treatment but that was not the case for TB patients infected with pre-XDR or XDR *M. tuberculosis* strains [150]. As an inhibitor of peptidoglycan biosynthesis, cycloserine has the distinctive feature of having a unique mechanism of action thus preventing any cross resistance with other first-line and second-line TB drugs. Cycloserine is, therefore, a useful addition to second-line TB drugs, although its use might be limited by its severe toxicity. Serious adverse effects were observed during cycloserine treatment such as neuropathy and behavioral changes [151]. Cycloserine is, therefore, contraindicated in patients suffering from severe depression, suicidal tendencies, kidney failure, epilepsy or seizures.

*Mode of action.* The mechanism of action of cycloserine was primarily deciphered in *Staphylococcus aureus.* Cycloserine is an analog of D-alanine and works as an antagonist of D-alanine [152]. Cycloserine inhibits the alanine racemase Alr (Rv3423c), which converts L-alanine to D-alanine, and the D-Ala:D-Ala ligase DdlA (Rv2981c) [153] preventing the integration of alanine into the pentapeptide core of the peptidoglycan. In *M. tuberculosis*, DdlA is thought to be the primary target of cycloserine [154–156].

*Resistance.* D-cycloserine-resistant *M. tuberculosis* mutants were isolated as early as 1957 in TB patients treated with cycloserine [157]. Cycloserine resistance has been associated with mutations in the genes encoding the alanine transporter CycA (Rv1704c), the L-alanine dehydrogenase Ald (Rv2780) and the alanine racemase Alr. A Gly122Ser mutation in *cycA* is present in the naturally cycloserine-resistant *M. bovis* BCG vaccine strain [158]. Complementation of BCG with a cosmid containing *M. tuberculosis cycA* renders BCG more susceptible to cycloserine than the parental strain leading the authors to conclude that *cycA* may be a factor in cycloserine resistance in BCG [158]. Desjardins and colleagues demonstrated that deletion of *ald* increased the resistance to cycloserine 2-fold in *M. tuberculosis*, while complementation of the *ald* mutant with a plasmid expressing *M. tuberculosis ald* only partially restored cycloserine susceptibility. Notably, complementation of BCG with the *ald* plasmid did not alter the strain resistance to cycloserine [159]. Mutations in *alr* (M319T, Y364D, R373L and c-8t in the promoter region) have been identified in XDR TB strains isolated from TB patients treated with cycloserine [160]. Further, the clinical isolate with the R373L mutation in *alr* also contained a deletion in *ald*.

In an in vitro experiment, 18 cycloserine-resistant mutants were obtained by culturing *M. tuberculosis* H37Rv on plates containing increasing concentrations of cycloserine (from 0.2 to 3 mM) and characterized using whole-genome sequencing [161]. Mutants were only obtained on plates containing less than 0.8 mM of cycloserine. A mutation was identified in *alr* (D344N), but no mutations in *ddlA***,** *ald* or *cycA* were found. Fifteen novel mutations were identified in genes involved in various pathways however no complementation or allelic exchange experiments were performed to confirm that these novel mutations were indeed involved in cycloserine resistance.

In a recent in vitro study [162], the authors showed that spontaneous cycloserine-resistant mutants emerged at a lower frequency (10<sup>−</sup>10–10−11) than rifampicin- (10−9) or isoniazid-resistant (10−8) mutants in *M. tuberculosis*. The authors characterized 11 independent cycloserine-resistant *M. tuberculosis* mutants by whole-genome sequencing and found mutations in *alr* (gene (D322N) and promoter region) but again not in *ddlA, ald* or *cycA*. None of these 11 cycloserine-resistant mutants showed any cross-resistance to other first-line and second-line TB drugs. The *alr* promoter mutation upregulated *alr* gene transcript, causing an increase in Alr protein level by up to 30-fold. The *alr* D322N mutation present in 8/11 mutants reduced the binding affinity for cycloserine by 240 fold while having limited effect on Alr enzymatic activity. Evangelopoulos and colleagues postulated that the *alr* mutations protected the enzymatic function of both cycloserine targets by decreasing the affinity of cycloserine to Alr and increasing the levels of D-alanine preventing the binding of cycloserine to DdlA. Their main conclusion was that DdlA was the "lethal target" for cycloserine.
