CYP5122A1 encodes an essential sterol C4-methyl oxidase in Leishmania donovani and determines the antileishmanial activity of antifungal azoles

Visceral leishmaniasis, caused by Leishmania donovani, is a life-threatening parasitic disease, but current antileishmanial drugs are limited and have severe drawbacks. There have been efforts to repurpose antifungal azole drugs for the treatment of Leishmania infection. Antifungal azoles are known to potently inhibit the activity of cytochrome P450 (CYP) 51 enzymes which are responsible for removing the C14α-methyl group of lanosterol, a key step in ergosterol biosynthesis in Leishmania. However, they exhibit varying degrees of antileishmanial activities in culture, suggesting the existence of unrecognized molecular targets for these compounds. Our previous study reveals that, in Leishmania, lanosterol undergoes parallel C4- and C14-demethylation reactions to form 4α,14α-dimethylzymosterol and T-MAS, respectively. In the current study, CYP5122A1 is identified as a sterol C4-methyl oxidase that catalyzes the sequential oxidation of lanosterol to form C4-oxidation metabolites. CYP5122A1 is essential for both L. donovani promastigotes in culture and intracellular amastigotes in infected mice. Overexpression of CYP5122A1 results in growth delay, differentiation defects, increased tolerance to stress, and altered expression of lipophosphoglycan and proteophosphoglycan. CYP5122A1 also helps to determine the antileishmanial effect of antifungal azoles in vitro. Dual inhibitors of CYP51 and CYP5122A1, e.g., clotrimazole and posaconazole, possess superior antileishmanial activity against L. donovani promastigotes whereas CYP51-selective inhibitors, e.g., fluconazole and voriconazole, have little effect on promastigote growth. Our findings uncover the critical biochemical and biological role of CYP5122A1 in L. donovani and provide an important foundation for developing new antileishmanial drugs by targeting both CYP enzymes.

To date, there is no human vaccine for VL [6]. Chemotherapeutic drugs are limited and have 13 drawbacks such as toxicity, long treatment regimens, emerging resistance, and/or high cost [7]. 14 Therefore, there is still an unmet medical need for safe and effective antileishmanial drugs. The 15 pentamidine isethionate, paromomycin sulfate, and amphotericin B deoxycholate which were 1 dissolved in water. Briefly, a CYP enzyme (50 nM) was incubated with the fluorogenic substrate 2 7-benzyloxy-4-trifluoromethylcourmarin (BFC; 50 µM) and various concentrations of test 3 compounds in 100 mM phosphate buffer (pH 7.4) and 3.3 mM MgCl2. Compound solvents 4 (DMSO and water) were used as the negative control. The reaction was initiated with the addition 5 of cumene hydroperoxide (100 µM), incubated at 37 °C, and monitored at an emission wavelength 6 of 538 nm and an excitation wavelength of 410 nm on a Tecan Infinite ® M200 Pro microplate 7 reader. For test compounds, the percentage of inhibition at each concentration was calculated as 8 daily using a Beckman Z2 Cell Counter. Log phase promastigotes refer to replicative parasites at 18 densities <1.0 × 10 7 cells/ml, and stationary phase promastigotes refer to non-replicative parasites 19 at densities >2.0 × 10 7 cells/ml. 20 21 13 To delete chromosomal CYP5122A1 alleles, the upstream and downstream flanking sequences 1 of CYP5122A1 (∼1 Kb each) were amplified by PCR and cloned in the pUC18 vector. Genes 2 conferring resistance to blasticidin (BSD) and puromycin (PAC) were cloned between the upstream 3 and downstream flanking sequences to generate pUC18-KO-CYP5122A1:BSD and pUC18-KO-4 CYP5122A1:PAC, respectively. To generate the CYP5122A1+/heterozygotes 5 (∆CYP5122A1::BSD/CYP5122A1), wild-type (WT) L. donovani promastigotes were transfected 6 with linearized BSD knockout fragment (derived from pUC18-KO-CYP5122A1:BSD) by 7 electroporation and transfectants showing resistance to blasticidin were selected and later 8 confirmed to be CYP5122A1+/by Southern blot as previously described [32]. To delete the 9 second chromosomal allele of CYP5122A1, we used an episome-assisted approach as previously 10 described for other genes [33]. First, the CYP5122A1 open reading frame (ORF) was cloned into 11 the pXNG4 vector to generate pXNG4-22A1 and introduced into CYP5122A1+/parasites. The 12 resulting CYP5122A1+/-+pXNG4-22A1 cell lines were then transfected with 13 linearized PAC knockout fragment (derived from pUC18-KO-CYP5122A1:PAC) and selected 14 with 15 μg/ml of blasticidin, 15 μg/ml of puromycin and 150 μg/ml of nourseothricin. The 15 resulting CYP5122A1 chromosomal null mutants with pXNG4-522A1 16 (∆CYP5122A1::BSD/∆CYP5122A1::PAC + pXNG4-CYP5122A1 or Cyp5122A1¯+ pXNG4-22A1) 17 were validated by Southern blot as previously described for other genes [32]. 18 19

Determination of in vitro antileishmanial activities of antifungal azole compounds in 20
CYP5122A1 mutants of L. donovani strain 1S2D 21 Stock solutions for test compounds were prepared at 10 mM in DMSO. To measure the 22 antileishmanial activity of these inhibitors, log phase promastigotes were inoculated in complete 23 post infection, one-half of the mice received GCV at 7.5 mg/kg/day for 14 consecutive days (0.5 ml 1 each, intraperitoneal injection), while the other half (control group) received an equivalent volume 2 of sterile PBS. At 4-or 7-weeks post infection, mice were euthanized through a controlled flow of 3 CO2 asphyxiation and infected spleens were isolated and homogenized. Parasite numbers in spleen 4 homogenates were determined by limiting dilution assay [34] or qPCR as described below. 5 6 Quantitative PCR (qPCR) 7 To determine parasite loads in infected mice, genomic DNA was extracted from spleen 8 homogenate and qPCR reactions were run in triplicates using primers targeting the 28S rRNA gene 9 of L. donovani [32]. Cycle threshold (Ct) values were determined from melt curve analysis. A 10 standard curve of Ct values was generated using serially diluted genomic DNA samples from L. 11 donovani promastigotes (from 0.1 cell/reaction to 10 5 cells/reaction) and Ct values >30 were 12 considered negative. Parasite numbers in spleen samples were calculated based on their Ct values 13 using the standard curve. Control reactions included sterile water and DNA extracted from the 14 uninfected mouse spleen. 15

16
To determine pXNG4-22A1 plasmid levels in promastigotes and amastigotes, a similar standard 17 curve was generated using serially diluting pXNG4-22A1 plasmid DNA (from 0.1 copy/reaction 18 to 10 5 copies/reaction) and primers targeting the GFP region. qPCR was performed with the same 19 set of primers on DNA samples from promastigotes or spleen and the average plasmid copy 20 number per cell was determined by dividing the total plasmid copy number by the total parasite 21 number based on Ct values. 22 To determine the transcript levels of SHERP (small hydrophilic endoplasmic reticulum-associated 1 protein), total RNA was extracted from promastigotes and converted into cDNA using a high-2 capacity reverse transcription kit (Bio-Rad), followed by qPCR using primers targeting SHERP or 3 28S rRNA genes. The relative expression level of SHERP was normalized to that of 28s rRNA 4 using the 2 −ΔΔ(Ct) method [35]. Control reactions were carried out without leishmanial RNA and 5 without reverse transcriptase. 6 7 Leishmania donovani 1S2D stress response assays 8 L. donovani promastigotes were cultivated in complete M199 media (pH 7.4) at 27 °C until they 9 reach the stationary phase. For heat tolerance, promastigotes were incubated at 37 °C. To test their 10 sensitivity to acidic pH, promastigotes were transferred to a pH 5.0 medium (same as the complete 11 M199 medium except that the pH was adjusted to 5.0 using hydrochloric acid). For starvation 12 response, promastigotes were transferred to PBS (pH 7.4). For resistance to oxidative or nitrosative 13 stress, parasites were incubated in various concentrations of H2O2 or S-nitroso-N-14 acetylpenicillamine (SNAP). Cell viability was determined at the indicated times by flow 15 cytometry after staining with 5 μg/ml of propidium iodide. Parasite growth was monitored using a 16 infected hamster spleens were cultured in Schneider's Drosophila medium (Gibco) containing 25% 23 heat-inactivated fetal bovine serum (Sigma-Aldrich) and penicillin-streptomycin (Gibco). Growth 1 assays were performed after three-day incubation of promastigotes at 26 °C with or without azole 2 drugs in the above medium containing 50 U/mL penicillin and 50 μg/mL streptomycin using 3-3 expression and chromatographic purification, recombinant CYP5122A1 and CYP51 were 10 obtained at high purity as indicated by the SDS-PAGE result ( Fig. 1A-B). Subsequent Western 11 blot analysis of CYP5122A1 ( Supplementary Fig. 2) showed that the recombinant protein had 12 slightly lower molecular weight compared with the wild-type proteins detected in Leishmania spp. 13 because of the deletion of the transmembrane domain. 14 15 CYP enzymes have an iron-containing heme as a prosthetic group whose different oxidation states 16 can be characterized using UV-Vis spectroscopy. Results of spectral analysis demonstrated that 17 the two proteins possessed properties typical of a CYP enzyme. Purified CYP5122A1 and CYP51 18 were in the oxidized state and displayed the Soret peak at 420 nm and 419 nm, respectively ( Fig.  19 1C-D). Upon reduction with dithionite and binding to CO, both enzymes exhibited the Soret peak 20 at around 450 nm and almost no absorption at 420 nm in the presence of lanosterol (insets of Fig.  21 1C-D). This indicated that they existed in the catalytically active P450 form rather than the inactive 22 P420 form. 23 19 1 In the natural context, a CYP enzyme requires two electrons during a catalytic cycle. The electrons 2 are donated by the cofactor NADPH and transferred to the CYP sequentially by its redox partner, 3 CPR. However, three putative CPRs (XP_003862234.1, XP_003864839.1, and XP_003864409.1; 4 protein sequence identity ranges from 21% to 27% between them) have been reported in the L. 5 donovani genome and none has been studied or characterized to our knowledge. Previously, a 6 recombinant CPR from another trypanosomatid T. brucei (TbCPR) has been characterized for its 7 P450 reduction activity and used as a surrogate CPR to reconstitute catalytic activities of CYP 8 enzymes from Leishmania and T. cruzi [22,38,39]. Here, in the presence of NADPH, TbCPR was 9 able to transfer the first electron to leishmanial CYP5122A1 and CYP51, and the resulting reduced 10 CO difference spectra ( Supplementary Fig. 3) were similar to the ones reduced by sodium 11 dithionite. Therefore, TbCPR may be used as a surrogate redox partner in place of the endogenous 12 leishmanial CPR for studying the catalytic activities of the two leishmanial CYP enzymes, 13 although exact biochemical roles of the three putative leishmanial CPRs warrant future 14 investigations. 15 16 CYP5122A1 had binding specificity to C4-methylated sterols. 17 When a substrate binds to the CYP enzyme, it displaces the distal H2O ligand of the heme iron, 18 converting the heme iron from the low-spin state to the high-spin state. Such binding mode is 19 termed "type I binding" and will yield a UV-Vis difference spectrum with a peak at 390 nm and a 20 trough at 420 nm [30]. This unique feature was employed for identifying potential substrates of 21 CYP5122A1. Five intermediate sterols within the ergosterol biosynthetic pathway (lanosterol, 22 4,14-DMZ, FF-MAS, T-MAS, and zymosterol; Scheme 1) were tested due to their same tetracyclic 23 ring structure, only differing in the number of methyl groups at C4 and/or C14 positions. Among 1 them, sterols with at least one C4-methyl group (lanosterol, 4,14-DMZ, FF-MAS, and T-MAS) 2 showed type I binding to CYP5122A1 ( Fig. 2A), regardless of whether the C14-methyl group is 3 present (lanosterol and 4,14-DMZ) or absent (FF-MAS and T-MAS). Zymosterol, which has no 4 C4-or C14-methyl group, did not induce any appreciable change in the difference binding 5 spectrum ( Fig. 2A). These results suggest that C4-methylated sterols like lanosterol, 4,14-DMZ, 6 FF-MAS, and T-MAS may serve as substrates of CYP5122A1. In comparison, CYP51, which is 7 known to act as a C14-demethylase, exhibited a binding specificity distinct from CYP5122A1 ( Fig.  8 2B). Only sterols with a C14-methyl group (lanosterol and 4,14-DMZ) showed typical type I 9 binding to CYP51, consistent with a previous report [22]. The C14-demethylated sterols (FF-MAS 10 and T-MAS) did not elicit any change in their difference binding spectra. Interestingly, zymosterol, 11 which lacks both C4-and C14-methyl groups, produced an atypical difference binding spectrum 12 with a peak at 411 nm and a trough at 432 nm (Fig. 2B), which differs from the typical binding 13 spectral changes, i.e., type I, II and reverse type I (also called pseudo or modified type II) [40]. 14 Similar atypical spectral changes were also observed for the binding of racemic and S-bicalutamide 15 to CYP46A1 through the bridging water ligand [41]. This suggests that zymosterol can enter into 16 the binding pocket of CYP51, albeit unable to dislodge the hexacoordinated water molecule. 17 18 CYP5122A1 acted to oxidize the C4-methyl group of sterols in Leishmania. 19 Using TbCPR as the redox partner, the catalytic activities of CYP5122A1 and CYP51 were 20 reconstituted in vitro with the addition of NADPH. Interestingly, the optimal reaction pH was pH 21 6.2-6.6 for both enzymes, rather than physiological pH ( Supplementary Fig. 4). respectively. In addition, zymosterol was oxidized by CYP51 to form a hydroxylated metabolite, 12 which was detected in the MRM transition of 383>95 (Fig. 3E). This is consistent with the atypical 13 difference binding spectrum of zymosterol and CYP51 observed above, although the exact location 14 of hydroxylation remains unknown, presumably on or near C14. 15 16 When compared to CYP51, CYP5122A1 clearly showed a different substrate specificity and 17 formed oxidation metabolites of different identities. Sterols that exhibited type I binding spectral 18 change with CYP5122A1 (i.e., lanosterol, 4,14-DMZ, FF-MAS, and T-MAS) were also 19 metabolized by this enzyme, yielding a mixture of alcohol, aldehyde, and/or 20 carboxylate/formyloxy metabolites of the corresponding sterol substrate (Fig. 3A-D), whereas 21 none of the zymosterol oxidation metabolites was detected (Fig. 3E). Importantly, the oxidation 22 metabolites of lanosterol formed by CYP5122A1 had different LC retention times from those 23 formed by CYP51. For example, the alcohol metabolite formed by CYP5122A1 (peak 7 in Fig.  1 3A) was eluted at 9.6 min, whereas the alcohol metabolite formed by CYP51 (peak 3 in Fig. 3A) 2 was eluted at 7.1 min. This indicates that CYP5122A1 did not catalyze the same C14 methyl 3 oxidation as CYP51 did. In contrast to CYP51, CYP5122A1 converted FF-MAS and T-MAS (both 4 are C4-dimethylated and C14-demethylated sterols) to their corresponding alcohol, aldehyde, and 5 carboxylate/formyloxy metabolites ( Fig. 3C and 3D). Additionally, it was observed that 6 CYP5122A1 converted 4,14-DMZ (a C4a-monomethylated and C14-methylated sterol) to an 7 alcohol metabolite, but no aldehyde or carboxylate/formyloxy metabolite was detected (Fig. 3B). 8 This result, along with the weaker binding of 4,14-DMZ to CYP5122A1 ( Fig. 2A), suggests that 9 CYP5122A1 may not be the optimal enzyme to catalyze the second C4-demethylation reaction 10 and other enzymes like leishmanial Erg25 may still play a role, which warrants future 11 investigations. No oxidation metabolite formation was seen for zymosterol with CYP5122A1. As 12 expected, NADPH was required for the catalytic activity of both CYP51 and CYP5122A1 (Fig.  13 3). Taken together, these findings prompted us to propose CYP5122A1 as a sterol C4-methyl 14 oxidase or a sterol C4-demethylase, assuming that the carboxylate metabolites formed by 15 CYP5122A1 will be decarboxylated and reduced by a dehydrogenase/decarboxylase and a 3-16 ketosterol reductase, respectively. 17

18
To fully elucidate the biochemical role of CYP5122A1, its reaction with lanosterol was scaled up 19 from 0.1 mL to 385 mL so that sufficient amounts of oxidation metabolites could be isolated, 20 purified, and analyzed by NMR spectroscopy for structural identification. The 1D and 2D NMR 21 spectra of the aldehyde metabolite (2 mg; referred to as "unknown") were successfully collected. 22 The 1 H and 13 C NMR assignments of lanosterol were used as a reference for the structural 23 assignment of the unknown lanosterol metabolite. The singlet at 9.4 ppm in the 1 H spectrum and 1 the signal appearing at 207 ppm in the 13 C spectrum indicate the presence of the aldehyde group 2 in the molecule (Fig. 4A the stereochemistry of the unknown compound at the C-4 position could not be ascertained due to 14 the low signal intensities in the NOESY spectrum (data not shown). In addition to C-4, the 15 assignment of C-3 and C-5 was confirmed by the HSQC correlations between C-3 and C-5 with 16 their corresponding protons H-3 and H-5 ( Fig. 4G-H). However, it should be noted that not all of 17 the observed signals in the NMR spectra were contributed by the aldehyde metabolite. In the 1 H 18 NMR spectrum (Fig. 4C), the broad peak at 1.26 ppm and the multiplet at 0.86 ppm (indicated by 19 asterisks) were caused by the contamination of grease. The broad peak at 1.56 ppm corresponded 20 to water. The singlet at 1.51 ppm was due to an unknown impurity present in the sample and was 21 not related to the molecule of interest. Taken together, these NMR spectroscopy and LC-MS/MS 22 results clearly indicate that CYP5122A1 catalyzes the sequential oxidation reaction of the 23 lanosterol C4 methyl groups, leading to C4-demethylation during ergosterol biosynthesis in 1 Leishmania. To determine if CYP5122A1 is required for promastigotes, Ld22A1+/-+pXNG4-22A1 and 21 Ld22A1 -+pXNG4-22A1 (clone #1 and #2) were cultivated in the absence or presence of selective 22 drugs (nourseothricin as the positive selection drug to retain pXNG4-22A1 and GCV as the 23 25 negative selection drug to expel plasmid) and analyzed for GFP expression in each passage as a 1 readout for plasmid retention (Fig. 5). When growing in media containing nourseothricin (+SAT), 2 all cell lines maintained high levels of GFP fluorescence (> 95% GFP-high) through multiple 3 passages as expected (Fig. 5A-C). Without nourseothricin or GCV, the heterozygous mutant and 4 clone #1 of the chromosomal-null mutant retained high levels of GFP, whereas the percentage of 5 GFP-high population in clone #2 of the chromosomal-null mutant gradually decreased to ~30% 6 after 17 passages. These findings allude to clonal variations among the CYP5122A1 mutants in 7 their ability to retain pXNG4-22A1 in the absence of selective pressure. When cultivated in the 8 presence of GCV and absence of nourseothricin, cells would favor eliminating the plasmid to avoid 9 toxicity if the plasmid did not contain any essential genes. Conversely, if there was an essential 10 gene on the plasmid, cells would retain the plasmid despite the associated cost [43]. As shown in 11 clones isolated from the GFP-low population (#2-1 and #2-2) quickly regained GFP expression 20 after two rounds of amplification (+GCV, -SAT) (Fig. 5E-F). Southern blot analysis verified that 21 clones from both GFP-high (#2-15 and #2-16) and GFP-low (#2-1 and #2-2) populations had 22 retained the pXNG4-22A1 plasmid (Fig. 5G). Finally, western blots revealed robust expression of 23 CYP5122A1 in these sorted clones at levels similar to Ld22A1+/-+pXNG4-22A1 and Ld22A1 -1 +pXNG4-22A1 grown in the presence of SAT. Together, these results supported that CYP5122A1 2 is essential for L. donovani in the promastigote stage. It is possible that the effect of GCV (administered during the first two weeks) had worn down by 18 week 7. Importantly, amastigotes of Ld22A1 -+pXNG4-22A1 with either PBS or GCV treatment 19 maintained much higher levels of the plasmid (average of 8-28 copies per cell) than Ld22A1+/-20 +pXNG4-22A1 (average of < 2 copies per cell) (Fig. 6C). Together, these results demonstrated 21 that CYP5122A1 is indispensable for L. donovani amastigotes in mice. 22 Genetic manipulation of CYP5122A1 altered sterol composition and affected the expression 1 of surface glycoconjugates and promastigote stress responses. 2 The effects of genetic manipulation of CYP5122A1 on sterol synthesis were analyzed by LC-3 MS/MS (Table 1). Several 4,14-methylated sterols (lanosterol, 4CH2OH-LS, and 4,14-DMZ) were 4 significantly accumulated in Ld22A1+/during log and stationary phases, while CYP5122A1 5 overexpression caused downregulation of these sterols. In addition, neither FF-MAS nor T-MAS 6 (formed by lanosterol C14-demethylation by CYP51) was detected in the Ld1s WT, Ld22A1+/-, 7 Ld22A1+/-+pXNG4-22A1, or Ld22A1 -+pXNG4-22A1 promastigotes. This suggested that 8 lanosterol C4-demethylation, rather than C14-demethylation, appears to be the dominant reaction 9 during ergosterol biosynthesis in these parasites. [11]. Considering that we observed the altered sterol profiles by genetic manipulation of 18 CYP5122A1, we investigated if the expression of LPG and PPG was also impacted. Western blot 19 analysis (Fig. 7) showed that CYP5122A1 overexpression led to more cellular PPG but less 20 secreted PPG in both log and stationary phases. In addition, the expression of cellular LPG was 21 significantly decreased. Although it was unclear if these changes would have any implications on 22 the parasite infectivity, the results provided more evidence of the association between the sterol 1 composition and the synthesis of important virulence factors in Leishmania. 2 3 Next, we examined if genetic manipulation of CYP5122A1 affected the expression of the 4 metacyclic stage-specific marker SHERP (small hydrophilic endoplasmic reticulum-associated 5 protein) [47] (Fig. 8). As expected, SHERP mRNA levels were low in Ld1S WT and CYP5122A1 6 mutants during the log phase. Upon entering the stationary phase, their SHERP expression levels 7 were significantly increased. However, the degree of SHERP induction in stationary phase was 8 less pronounced as WT (Fig. 8A). After normalizing the expression levels of SHERP in mutants 9 to those in WT, we observed that CYP5122A1 half knockout and overexpression reduced the 10 SHERP expression in both log and stationary phases (Fig. 8B). overexpressors but reached over 60% in Ld1S WT and Ld22A1+/- (Fig. 9B). The increased 20 tolerance of Ld22A1+/-+pXNG4-22A1 and Ld22A1 -+pXNG4-22A1 to stress was also evident in 21 their responses to starvation. After incubation in PBS in the absence of nutrients for 36 h, they had 22 ~40% of dead cells whereas almost 90% of the WT and Ld22A1+/cells did not survive (Fig. 9C). 23 When the incubation temperature was set at 37 °C to mimic the mammalian body temperature, the 1 percentage of dead cells increased slower in CYP5122A1 mutants than in Ld1Ss WT although the 2 difference was not statistically significant (Fig. 9D). Apart from physical stresses, Leishmania 3 parasites would encounter reactive oxygen and nitrogen species which are produced by 4 macrophages as part of the host defense responses [48]. To evaluate their resistance to oxidative 5 or nitrosative stress (a condition they would encounter as part of the host defense [48]), parasites 6 were incubated in various concentrations of H2O2 or SNAP (Fig. 9E-F). CYP5122A1 7 overexpressors had superior viability than WT and Ld22A1+/at H2O2 concentrations up to 600 8 µM (p < 0.001) but exhibited no difference in response to SNAP. Taken together, CYP5122A1 9 overexpression rendered the cells more resistant to hostile environment. 10 11 Dual inhibition of CYP5122A1 and CYP51 was required for optimal antileishmanial 12 activities by antifungal azoles. Two selective CYP51 inhibitors (voriconazole and fluconazole) and two dual CYP51/CYP5122A1 23 inhibitors (clotrimazole and posaconazole) were further assessed for their binding modes with the 1 two CYP enzymes by UV-Vis spectrophotometric analysis (Supplementary Fig. 7). Clotrimazole 2 exhibited type II binding to both CYP51 and CYP5122A1, indicating that the nitrogen in its 3 imidazole group was directly coordinated with the heme iron at the active site. Posaconazole 4 exhibited type II binding to CYP51, but a type II-like binding to CYP5122A1 (the trough peak at 5 around 410 nm is absent). Fluconazole and voriconazole elicited type II binding with CYP51 but 6 no specific responses with CYP5122A1, consistent with the observation that they were CYP51-7 selective inhibitors. Moreover, we evaluated the effect of azole antifungal drugs on the 8 proliferation of L. donovani LV82 promastigotes (Table 2 and Supplementary Fig. 8). Of these 20 9 compounds tested, twelve displayed EC50 values <10 µM, while another six azoles exhibited EC50 10 values between 10 µM and 25 µM. Interestingly, voriconazole and fluconazole, the weakest 11 inhibitors of CYP5122A1, were the least effective at inhibiting parasite growth with EC50 12 values >100 µM. These results suggested that dual inhibitors of CYP51 and CYP5122A1 were 13 more likely to inhibit L. donovani growth. 14 15 Sterol analysis of LV82 promastigotes showed that, as expected, both clotrimazole (dual CYP51 16 and CYP5122A1 inhibitor) and voriconazole (selective CYP51 inhibitor) treatments resulted in 17 the depletion of 4,14-demethylated and 5-dehydro sterols in the downstream pathway of ergosterol 18 biosynthesis, and the accumulation of some 4-and/or 14-methylated sterols (Table 3)

Sensitivity of CYP5122A1 overexpressors to CYP5122A1-selective and dual inhibitors 11
To test whether CYP5122A1 overexpression affected the susceptibility to the CYP5122A1-12 selective and dual inhibitors, we measured the growth of Ld1S WT and CYP5122A1 mutants in 13 the presence or absence of different inhibitors (Fig. 10) reach EC50, while more than 12 µM was needed to inhibit growth further (Fig. 10B). In contrast, 19 CYP5122A1 overexpressors were much more resistant to posaconazole (EC50 = ~25 µM), 20 indicating that increased production of this enzyme can protect parasites against dual inhibitors. In this study, we identified leishmanial CYP5122A1 as a sterol C4-methyl oxidase and unveiled a 22 previously unrecognized important difference in the sterol C4-demethylation process between 23 Leishmania and other species. Leishmanial CYP5122A1 was found to be responsible for oxidizing 1 the lanosterol C4-methyl group into first a hydroxyl, then an aldehyde, and finally a 2 carboxylate/formyloxy metabolite. Although the stereochemistry of these metabolites at the C-4 3 position was not directly ascertained from the NOESY spectrum of the isolated metabolite due to 4 low signal intensity, it can be inferred as a sterol C4-methyl oxidase due to several lines of evidence. 5 First, CYP5122A1 oxidized 4,14-DMZ, which has an α-configuration at both C4-and C14-methyl 6 groups [16], and elicited a type-I binding with 4,14-DMZ (Fig. 2B). It also showed a type I binding 7 specificity to C4-methylated sterols, whereas CYP51 preferred C14-methylated sterols ( Fig. 2A). 8 Second, CYP5122A1 converted 4,14-DMZ into a hydroxylated product (peak 15 in Fig. 3B) that 9 is different from the C14-hydroxylmethyl product formed by CYP51 (peak 11 in Fig. 3B), 10 indicating that CYP5122A1 likely acted on the C4-methyl group, rather than the C14-methyl group 11 of 4,14-DMZ. This is further supported by our NMR structural determination of the C4-aldehyde 12 metabolite of lanosterol formed by CYP5122A1 (Fig. 4) these counterparts in Trypanosoma. It is still unknown whether trypanosomal CYP5122A1 23 enzymes possess the same biochemical function or exhibit the same inhibition profile as the 1 leishmanial CYP5122A1 enzyme. Moreover, enzymes that are involved in the C4-demethylation 2 process after C4-oxidation in Leishmania (i.e., dehydrogenase/decarboxylase and reductase) 3 remain uncharacterized, let alone other enzymes responsible for the multi-step conversion of 4 lanosterol to ergosterol in these parasites. An ortholog of yeast Erg25 has been found in the 5 Leishmania genome but its function remains to be characterized [53]. As such, the sterol 6 biosynthetic pathway in protozoa warrants future dedicated investigation to fully understand the 7 unique sterol biology of these single cell parasites. As demonstrated in this study, it is premature 8 to assume these protozoan parasites behave like fungi or mammalian species. 9 10 The essentiality of CYP5122A1 was first investigated in L. donovani promastigotes by using the 11 homologous replacement approach [25]. Deleting both alleles of CYP5122A1 was found to be 12 lethal, which suggested that the gene is required for parasite survival. However, failure to obtain 13 the null mutant in this way is often considered weak evidence for gene essentiality [54]. In the 14 present study, the more rigorous forced plasmid shuffle approach was applied to examine the gene 15 essentiality. We were able to generate the null mutant transfected with the plasmid containing 16 CYP5122A1 and a negative selection mark (TK). The null mutant without the episomal 17 CYP5122A1 could not survive. In the presence of negative selection, the parasites lacking in the 18 plasmid would be favored, yet retention of the plasmid was observed in parasites after multiple 19 passages. This provided compelling evidence supporting the essentiality of CYP5122A1 in L. 20 donovani promastigotes. Moreover, a more comprehensive assessment should also take into 21 account the amastigote stage of the parasite life cycle. Unlike promastigotes, intracellular 22 amastigotes live in a semi-quiescent state characterized by a negligible proliferation rate, a lower 23 biosynthetic capacity, a reduced bio-energetic level, and a stringent metabolic response [55,56]. 1 In particular, amastigotes acquire most of the sterols by salvage from the host although they retain 2 the capacity of de novo synthesis [57]. It was observed that cholesterol was much more abundant 3 than ergostane-based sterols in amastigotes [11]. In view of such different sterol profiles during 4 the Leishmania life cycle, we investigated whether CYP5122A1 was still essential during the 5 amastigote stage using a mouse model of visceral leishmaniasis infection. Under the pressure of 6 negative selection, the null mutant with the episomal complementation of CYP5122A1 was able to 7 proliferate in mice and the plasmid was highly retained. These results demonstrate the essentiality 8 of CYP5122A1 in L. donovani amastigotes, which is critical for the genetic validation of 9 CYP5122A1 as a drug target. 10 11 Antifungal azole drugs have been assessed for their potential to be used as antileishmanial agents. 12 Studies showed that they were potent inhibitors of leishmanial CYP51 that can block ergosterol 13 biosynthesis, but it was not fully understood why they had vastly different antileishmanial effects 14 in vitro [13,14,58]. In this study, we screened a panel of twenty marketed antifungal azole drugs 15 for the inhibition of CYP51 and CYP5122A1 using a fluorescence-based inhibition assay. Based 16 on the selectivity towards the CYP enzymes, they were grouped into CYP51-selective inhibitors 17 (e.g., fluconazole and voriconazole) and dual inhibitors of both CYP enzymes (e.g., posaconazole 18 and clotrimazole). Overall, dual inhibitors exhibited higher antileishmanial activities against LV82 19 promastigotes than CYP51-selective inhibitors. This supports the critical role of CYP5122A1 in 20 determining the azole activities against the parasites. When designing new compounds that target 21 the sterol biosynthesis in Leishmania, inhibiting both CYP51 and CYP5122A1 may be a better 22 option than inhibiting CYP51 alone. A previous study on a potent antileishmanial arylimidamide 23 DB766 implicated CYP5122A1 in the antileishmanial action of the compound [26], which could 1 be supported by our finding that DB766 is a strong CYP5122A1 inhibitor with IC50 of 1.0 ± 0.1 2 µM, although additional mechanisms are likely involved to account for the potent antileishmanial 3 activity of the compound with IC50 values ranging from 0.014 to 0.5 µM against multiple 4 Leishmania species [49] and against wild type L. donovani strain 1S2D (0.004 µM; Fig. 10A). 5 6 When analyzing the sterol profiles of azole-treated promastigotes along with our previous findings 7 [16], we found that CYP51-selective inhibitors and dual inhibitors resulted in distinct patterns of 8 sterol accumulation. The underlying cause could be the branched ergosterol biosynthesis pathway 9 that we previously proposed [16]. Lanosterol, the substrate of both CYP51 and CYP5122A1, can 10 undergo either a C4-or C14-demethylation reaction. Dual inhibitors caused lanosterol 11 accumulation by inhibiting both reactions whereas CYP51-selective inhibitors would allow C4-12 demethylation to proceed. Furthermore, sterol analysis of genetic mutants showed accumulation 13 of 4,14-methylated sterols (lanosterol, 4CH2OH-LS, and 4,14-DMZ) in Ld22A1+/and 14 downregulation in CYP5122A1 overexpressors (Ld22A1+/-+pXNG4-22A1 and 15 Ld22A1¯+pXNG4-22A1), suggesting that C4-demethylation, rather than C14-demethylation, is 16 the dominant reaction for lanosterol metabolism in Leishmania. This is further supported by the 17 lack of detection of FF-MAS and T-MAS in the Leishmania parasites (Tables 1 and 3 In summary, our study elucidated that CYP5122A1 is a sterol C4-methyl oxidase involved in the 5 sterol C4-demethylation process in Leishmania. The essentiality of CYP5122A1 was verified in 6 both L. donovani promastigotes and intracellular amastigotes. Overexpression of CYP5122A1 7 resulted in growth delay, differentiation defects, increased tolerance to environmental stresses, and 8 altered expression of surface glycoconjugates. Antifungal azoles acting as dual inhibitors of 9 CYP51 and CYP5122A1 had higher antileishmanial activity against L. donovani promastigotes 10 than CYP51-selective inhibitors, suggesting a new strategy to develop therapeutic agents to target 11 the sterol biosynthetic pathway in Leishmania.   The insets show the CO difference spectra of sodium dithionite-reduced CYPs which exhibited 7 the characteristic peak at around 450 nm.   carboxylate/formyloxy metabolites (purple), and 14-demethylated products (red) were indicated in the chromatograms. Peaks with asterisk were detected in both incubations with or without NADPH and hence considered as background signals.  whole cell lysates from log phase promastigotes before and after sorting were analyzed by western blot using anti-LdCYP5122A1 (top) or anti-a-tubulin antibodies.  CYP5122A1 overexpression alters the expression of LPG and PPG. Log phase (A12 D) and stationary phase (E-H) promastigotes were subjected to Western blot analyses using the monoclonal antibody CA7AE (for LPG and PPG) or an anti-a-tubulin antibody (as loading control). A and E: whole cell lysates. B and F: culture supernatants. Relative expression levels of LPG (C and G) and PPG (D and H) were determined with Ld1S WT as 1.0. Error bars represent standard deviations from three repeats (*: p < 0.05, **: p < 0.01, ***: p < 0.001).

Figure 8
CYP5122A1 half knockout and overexpression reduces the expression of SHERP. Total RNA from log phase and stationary phase promastigotes were subjected to RT-qPCR analyses using primers for SHERP and 28S rDNA (as loading control). (A) Expression levels of SHERP in each parasite line from log phase to stationary phase using log phase levels as 1.0. (B) Expression levels of SHERP relative to Ld1s WT (as 1.0). Error bars represent standard deviations from three repeats (*: p < 0.05, **: p < 0.01, ***: p < 0.001).

Figure 9
Figure 10 CYP5122A1 overexpression reduces sensitivity to CYP5122A1 inhibitors. Log phase promastigotes were inoculated in various concentrations of posaconazole (A) or DB766 (B) and culture densities were determined after 48 hours. Percentages of growth were indicated relative to control cells grown in the absence of inhibitors. Error bars represent standard deviations from three repeats (*: p < 0.05, ***: p < 0.001).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.