The mycobacterial desaturase DesA2 is associated with mycolic acid biosynthesis

Mycolic acids are critical for the survival and virulence of Mycobacterium tuberculosis, the causative agent of tuberculosis. Double bond formation in the merochain of mycolic acids remains poorly understood, though we have previously shown desA1, encoding an aerobic desaturase, is involved in mycolic acid desaturation. Here we show that a second desaturase encoded by desA2 is also involved in mycolate biosynthesis. DesA2 is essential for growth of the fast-growing Mycobacterium smegmatis in laboratory media. Conditional depletion of DesA2 led to a decrease in mycolic acid biosynthesis and loss of mycobacterial viability. Additionally, DesA2-depleted cells also accumulated fatty acids of chain lengths C19-C24. The complete loss of mycolate biosynthesis following DesA2 depletion, and the absence of any monoenoic derivatives (found to accumulate on depletion of DesA1) suggests an early role for DesA2 in the mycolic acid biosynthesis machinery, highlighting its potential as a drug target.


Results
Homologues of desA2 are found in other mycolic acid-producing bacteria that utilise FAS-II. Homologues of desA2 are present in other mycobacterial species including M. smegmatis, several non-tuberculous mycobacterial species (NTMs) and the obligate intracellular pathogen Mycobacterium leprae suggesting a core mycobacterial function for desA2. To further query a bespoke role for desA2 in other mycolic acid-producing genera that synthesize long-chain mycolates using FAS-II, we used the M. tuberculosis desA2 amino acid sequence as a query to search genomes of other mycolic acid-producing bacteria (the mycolata). Homologues of desA2 were present in mycolata that produced long-chain mycolic acids, including species of Nocardia and Rhodococcus (Fig. 2). No homologues were found in the genomes of Corynebacterium species that produce shorter mycolic acid chains using FAS-I, suggesting desA2 is a putative desaturase gene found only in FAS-II encoding mycolata.
DesA2 is an essential gene in mycobacteria. We generated a recombinant phage designed to replace the native copy of M. smegmatis desA2 with a hygromycin resistance cassette (hyg). Repeated attempts to generate a null desA2 mutant in M. smegmatis by Specialized Transduction 16 were unsuccessful, suggesting that desA2 was essential for growth in laboratory media, an outcome often associated with mycolate biosynthesis genes. To demonstrate and validate the essentiality of desA2 in M. smegmatis, we then transduced a merodiploid strain of M. smegmatis (containing an acetamidase promoter-driven, second integrated copy of desA2) with the knockout phage. Hygromycin resistant transductants were obtained only on plates containing the inducer, acetamide. One such transductant was confirmed by whole-genome sequencing to be a conditional desA2 mutant, with the native copy of desA2 replaced by hyg. These results demonstrated that desA2 was an essential gene in M. smegmatis. A similar result was obtained with a M. smegmatis merodiploid containing acetamidase promoter-driven M. tuberculosis desA2, indicating that the M. tuberculosis homologue could compensate for the loss of function in M. smegmatis. The isolated conditional mutant termed ∆desA2 was unable to grow on plates (Fig. 3A) or in broth lacking acetamide (Fig. 3B,C) confirming the essentiality of desA2 in M. smegmatis. Saturation transposon mutagenesis screens in M. tuberculosis indicate that M. tuberculosis desA2 is also an essential gene 17 . Loss of desA2 impacts mycolic acid biosynthesis. To further probe the potential role of desA2 in mycolic acid biosynthesis, we analysed [ 14 C]-labelled mycolic acid methyl esters (MAMES) extracted from cultures of the ∆desA2 conditional mutant grown in the presence or absence of acetamide. Two dimensional (2D)argentation TLC is a useful tool for analysing heterogeneity in desaturation (double bond) levels in mycolic acid sub-classes. The second TLC dimension separates saturated and unsaturated species in an AgNO 3 -treated    13 . We now sought to assess the effects of depletion of DesA2 on mycolic acid biosynthesis and concurrently mycolic acid desaturation using this approach. Cultures of ∆desA2 grown in the presence or absence of acetamide were pulsed with [ 14 C]-acetic acid to label lipids. Mycolic acids were extracted from the cultures and converted to mycolic acid methyl esters (MAMEs) prior to separation by 2D-argentation TLC. 14 C-labelled MAMEs were visualised by autoradiography. We were not able to observe any MAMEs that migrated faster in the second dimension, but depletion seemed to result in an overall reduction in mycolic acid biosynthesis (Fig. 4A,B, Uncropped images available as Supplementary Information). The effects were specific to mycolic acids and mycolic acid containing lipids (TMM and TDM) as the biosynthesis of other lipids was not affected during the early stages of DesA2 depletion (Fig. 5), ruling out any non-specific effects on mycolate biosynthesis. A parallel ESI-MS analysis of MAMEs extracted from unlabelled cells did not reveal any qualita- www.nature.com/scientificreports/ tive differences between mycolic acids from cultures grown in the presence or absence of acetamide ( Fig. 4C) indicating the absence of any monoenoic intermediates like those observed for the conditional desA1 mutant 13 . These data showed that the loss of DesA2 specifically led to a loss of mycolic acids, confirming a role for DesA2 in mycolic acid biosynthesis. Surprisingly, the levels of other fatty acyl methyl esters (FAMEs) appeared elevated suggesting their accumulation in the DesA2-depleted cells (Fig. 4A).

Depletion of DesA2 leads to accumulation of fatty acids. We further probed the accumulation of
FAMEs in the TLCs shown in Fig. 4A by analysing the FAMEs/MAMEs extract by reverse phase TLC, which enabled the separation of (relatively) short-chain fatty acids. Interestingly, conditional depletion of DesA2 led to the accumulation of fatty acyl methyl esters corresponding to C 19 -C 24 (Fig. 6, uncropped image available as Supplementary Information). To determine if there were qualitative differences between the FAMEs extracted from  www.nature.com/scientificreports/ the two culture conditions, we further probed the nature of these accumulating FAMEs in the DesA2-depleted cultures using GC. There were slightly higher levels of C 19 -C 24 fatty acids in the ΔdesA2 conditional mutant grown in the absence of acetamide, compared to those seen in cultures grown in the presence of acetamide (Fig. 7). No unique species were identified in the acetamide-lacking cultures when compared to the GC spectra observed with extracts from the acetamide-containing cultures, suggesting that changes following depletion of DesA2 were a result of the accumulation of known mycobacterial fatty acids, rather than the accumulation of new classes of fatty acids.

Discussion
DesA1 and desA2 encode the two putative aerobic desaturases in mycobacterial genomes. Similar to desA1 13 , we show here that desA2 is also an essential gene in M. smegmatis and is associated with mycolic acid biosynthesis. However, the outcomes of desaturase depletion were very different for the conditional mutants of the two genes. While we detected a transient accumulation of minor mycolic acid-related species (likely monoenoic derivatives of α-mycolic acids) and free fatty acyl species of lengths ~ C 26 -C 48 in the conditional desA1 mutant, no such intermediates were observed on conditional depletion of DesA2. Instead, we saw the accumulation of C 19 -C 24 fatty acids which are likely end products of the bimodal FAS-I 18 . The accumulation of these intermediate chain fatty acids mimics what is observed following an early and complete shutdown of mycolic acid biosynthesis 19 core components of FAS-II 12,20 . There may also be an alternative explanation for the accumulation of C 19 -C 24 fatty acids in DesA2 depleted cells: they also represent the predicted mero chain lengths which would be synthesised prior to the addition of the first (proximal) double bond (Fig. 1). This, combined with the loss of mycolic acid biosynthesis and the absence of any monoenoic intermediates (unlike in the desA1 mutant) following DesA2 depletion in the M. smegmatis conditional mutant suggests an early role for DesA2, likely the addition of the proximal (first) double bond on the growing merochain. Our studies with the conditional desA2 mutant could not decipher further functional information about DesA2, but they prove its role in mycolic acid biosynthesis and highlight its potential as a new drug target within the mycolic acid biosynthesis pathway.

Methods
Chemicals. All solvents used were reagent grade from Sigma-Aldrich, Fisher Chemicals and Merk.
Bioinformatics. Protein homology among the bacterial species was determined with Blastp. Then, sequence alignments were performed using the T-coffee 21  The mc 2 155::pMVacRv1094 merodiploid strain was then subjected to specialized transduction as previously described 16 using the phage phΔMSMEG5248. Briefly, approximately 1 kb regions upstream and downstream of MSMEG5248 were amplified by PCR using the primers Ms5248LL (TTT TTT TTG CAT AAA TTG CCA CGA CCA CGC ACA AGT C), Ms5248LR (TTT TTT TTG CAT TTC TTG CCC CAC TGT TCT TCG AGG ATG), Ms5248RL  (TTT TTT TTC CAT AGA TTG GTC TTC TCC AAC CTC GTC AGC) and Ms5248RR (TTT TTT TTC CAT CTT TTG  GGG ATG TCC TTG CTC ACC AAC) and cloned on either side of a hygromycin resistance cassette in the p0004S vector to generate the plasmid pΔMSMEG5248. The insertion of pΔMSMEG5248 into the temperature-sensitive mycobacteriophage phAE159 16 produced the recombinant knockout phage phΔMSMEG5248. Transductants were selected on TSB-agar plates containing 150 μg/ml hygromycin and 0.02% acetamide at the non-permissive temperature of 37ºC. After confirmation of the gene replacement by PCR and whole-genome sequencing, one of the transductants was designated as ΔdesA2 and used for further analysis.
Conditional depletion of desA2 and determination of viable-cell counts. ΔdesA2 conditional mutant was grown in 3 ml TSB-T and acetamide up to an OD 600 of 1. Then, bacterial cells were washed with PBS to remove traces of acetamide and resuspended in their original volume with TSB-T. 30 µl of the cell suspension was used to grow 2 cultures of 3 ml TSB-T with and without acetamide at 37 °C ON. 1:10,000 dilution of the same cell suspension was used to inoculate the ΔdesA2 conditional mutant in TSB-agar plates with and without acetamide, which were grown at 37 °C for 3 days. For CFU determination, ΔdesA2 was grown in TSB-T with and without acetamide at 37 °C and dilutions of the aliquots taken at 0, 14, 18, 21, 24 and 45 h were spread on TSB-agar containing acetamide. Plates were incubated at 37 °C for 48-72 h before counting CFU.

Extraction of radiolabelled fatty acids and mycolic acids for Thin Layer Chromatography (TLC)
analysis. WT  www.nature.com/scientificreports/ with or without acetamide. In the case of ΔdesA2, 20 ml of the resuspended culture were transferred to 2 fresh flasks, adding acetamide to one of them. Then, 14 C-labelled acetic acid (PerkinElmer) was added to each culture at a concentration of 1 μCi/ml to be incubated at 37 °C ON. Bacterial cultures were harvested and washed with PBS to start the lipid extraction as described previously 28 . Briefly, apolar lipids from the pellets were extracted twice by adding 4 ml of petroleum ether (60-80 °C; Fisher Chemicals) to 2 ml of Methanol (Fisher Chemicals): 0.3% sodium chloride (Fisher Chemicals) (10:1), mixing on the rotator, centrifuging and transferring the upper layer containing apolar lipids to a new tube. For polar lipid extraction, 2.3 ml of chloroform (Fisher Chemicals): methanol: 0.3% sodium chloride (9:10:3) was added to the lower layer, mixed on the rotator and then centrifuged. The supernatant was transferred to a fresh tube and the pellets were extracted twice with 750 µl of chloroform:methanol: 0.3% sodium chloride (5:10:4). The pooled extracts were mixed with 1.3 ml of chloroform and 1.3 ml of 0.3% sodium chloride, centrifuged and the lower layer containing the polar lipids transferred to a new tube. Lipids were separated by 2D-TLC using solvent systems described by Dobson et al. 28 . The remaining delipidated cells and a polar lipids were exposed to alkaline hydrolysis 29 with 5% tetrabutylammonium hydroxide (TBAH; Sigma-Aldrich) at 100ºC ON, followed by adding 4 ml dichloromethane (Fisher Chemicals), 50 μl iodomethane (Merck), 2 ml water, and mixing for 30 min. The upper phase was discarded after centrifugation, and the lower phase was washed thrice with water, evaporated and dissolved in diethyl ether (Fisher Chemicals  ) and chloroform (Mobile phase B) were used for gradient elution. Initial conditions of the mobile phase at 90% A/10% B followed by a linear gradient to 10% A/90% B in 40 min. The Ion Spray voltage was maintained at 4.2 kV. The temperature was 600 °C for MAMEs. The column eluent was introduced into the Turbo Spray ion source of an ESI/MS system operated in the positive ion mode. The mass spectrum was acquired from m/z 700-1500 with a frequency of 1 scan/0.90 s for MAMEs. Typically, 10 μl of the sample was injected for analysis. Analyst 1.6.2 software (SCIEX Corp) was used for system control, data accumulation, and data analysis. FAME analyses were performed with a GC-2014 gas chromatograph (Shimadzu Corp. Ltd., Tokyo, Japan) equipped with a flame ionization detector and Equity-1 (30 m × 0.25 mm I.D., df 0.25 µm, capillary GC column; Supelco Inc., Bellefonte, Penn., USA) at a temperature ranging from 150 °C for 4 min to 250 °C at an increase of 4 °C/min, holding 3 min, and to 280 °C at an increase of 4 °C/min, holding 4 min. Each peak was identified by its retention time compared with those of a fatty acid methyl ester mixture (Supelco Inc.).