A conditional mutant of the fatty acid synthase unveils unexpected cross talks in mycobacterial lipid metabolism

Unlike most bacteria, mycobacteria rely on the multi-domain enzyme eukaryote-like fatty acid synthase I (FAS I) to make fatty acids de novo. These metabolites are precursors of the biosynthesis of most of the lipids present both in the complex mycobacteria cell wall and in the storage lipids inside the cell. In order to study the role of the type I FAS system in Mycobacterium lipid metabolism in vivo, we constructed a conditional mutant in the fas-acpS operon of Mycobacterium smegmatis and analysed in detail the impact of reduced de novo fatty acid biosynthesis on the global architecture of the cell envelope. As expected, the mutant exhibited growth defect in the non-permissive condition that correlated well with the lower expression of fas-acpS and the concomitant reduction of FAS I, confirming that FAS I is essential for survival. The reduction observed in FAS I provoked an accumulation of its substrates, acetyl-CoA and malonyl-CoA, and a strong reduction of C12 to C18 acyl-CoAs, but not of long-chain acyl-CoAs (C19 to C24). The most intriguing result was the ability of the mutant to keep synthesizing mycolic acids when fatty acid biosynthesis was impaired. A detailed comparative lipidomic analysis showed that although reduced FAS I levels had a strong impact on fatty acid and phospholipid biosynthesis, mycolic acids were still being synthesized in the mutant, although with a different relative species distribution. However, when triacylglycerol degradation was inhibited, mycolic acid biosynthesis was significantly reduced, suggesting that storage lipids could be an intracellular reservoir of fatty acids for the biosynthesis of complex lipids in mycobacteria. Understanding the interaction between FAS I and the metabolic pathways that rely on FAS I products is a key step to better understand how lipid homeostasis is regulated in this microorganism and how this regulation could play a role during infection in pathogenic mycobacteria.

Unlike most bacteria, mycobacteria rely on the multi-domain enzyme eukaryote-like fatty acid synthase I (FAS I) to make fatty acids de novo. These metabolites are precursors of the biosynthesis of most of the lipids present both in the complex mycobacteria cell wall and in the storage lipids inside the cell. In order to study the role of the type I FAS system in Mycobacterium lipid metabolism in vivo, we constructed a conditional mutant in the fas-acpS operon of Mycobacterium smegmatis and analysed in detail the impact of reduced de novo fatty acid biosynthesis on the global architecture of the cell envelope. As expected, the mutant exhibited growth defect in the non-permissive condition that correlated well with the lower expression of fas-acpS and the concomitant reduction of FAS I, confirming that FAS I is essential for survival. The reduction observed in FAS I provoked an accumulation of its substrates, acetyl-CoA and malonyl-CoA, and a strong reduction of C 12 to C 18 acyl-CoAs, but not of long-chain acyl-CoAs (C 19 to C 24 ). The most intriguing result was the ability of the mutant to keep synthesizing mycolic acids when fatty acid biosynthesis was impaired. A detailed comparative lipidomic analysis showed that although reduced FAS I levels had a strong impact on fatty acid and phospholipid biosynthesis, mycolic acids were still being synthesized in the mutant, although with a different relative species distribution. However, when triacylglycerol degradation was inhibited, mycolic acid biosynthesis was significantly reduced, suggesting that storage lipids could be an intracellular reservoir of fatty acids for the biosynthesis of complex lipids in mycobacteria. Understanding the interaction between FAS I and the metabolic pathways that rely on FAS I products is a key step to better understand how lipid homeostasis is regulated in this microorganism and how this regulation could play a role during infection in pathogenic mycobacteria.

Background
All organisms that produce fatty acids (FA) do so via a repeated cycle of reactions. The first committed step in FA biosynthesis is the carboxylation of acetyl-CoA by acetyl-CoA carboxylase (ACC), to form the common extender unit malonyl-CoA. This building block is subsequently condensed and reduced in an iterative fashion until the FA chain matures for use by the cell. In eukaryotes, these reactions are catalysed by a type I fatty acid synthase (FAS I), a large multifunctional protein to which the growing chain is covalently attached [1]. By contrast, most bacteria contain a type II system (FAS II) in which each reaction is catalysed by a discrete protein [2]. Remarkable exceptions to this rule are members of the Corynebacterineae, which possess a eukaryotic-like type I FAS [3]. In mycobacteria, FA biosynthesis is initiated by the multifunctional FAS I enzyme, which catalyses the de novo synthesis of medium and long-chain acyl-CoAs from acetyl-CoA, using malonyl-CoA as extender unit. These acyl-CoAs are subsequently used for the synthesis of membrane phospholipids (PL) or they are shuffled into the FAS II system for the synthesis of mycolic acids. In mycobacteria, the type II FAS system is responsible for the elongation of these acyl-CoAs through iterative condensation reactions with malonyl-ACP, leading to the synthesis of the very longchain meromycolyl-ACPs [4][5][6]. A remarkable characteristic of the mycobacterial FAS I system is the bimodal behaviour that was demonstrated in vitro. Besides producing medium chain acyl-CoAs, the FAS I system releases C 24:0 -CoA in Mycobacterium smegmatis [7,8] or C 26:0 -CoA in Mycobacterium tuberculosis [9] that becomes carboxylated by a dedicated long chain acyl-CoA carboxylase (LCC) complex to yield an a-carboxy-C 24/26 -CoA. These carboxy-acyl-CoAs are then condensated with the meromycolyl-AMP in a reaction catalysed by the polyketide synthase Pks13 (electronic supplementary material, figure S1), to produce mycolic acids (MA) and their glyco-derivatives, lipids of extreme importance for the maintenance of M. tuberculosis membrane properties and immunopathogenicity [10].
To add a higher level of complexity to M. tuberculosis lipid metabolism, acyl-CoAs synthesized by FAS I are not only incorporated into PL and/or into the storage lipids triacylglycerides (TAG), but also used as important biosynthetic precursors to produce virulence associated polyketide lipids like phtiocerol-dimycoseroic acid (PDIM), polyacylated trehalose (PATS) and sulfolipids (SL) [10,11] (electronic supplementary material, figure S1).
The necessary crosstalk between the FAS I and FAS II systems, the different polyketide synthases (PKS) and the TAG biosynthesis pathways expands the complexity of the putative regulatory network that maintain lipid homeostasis in M. tuberculosis. It also highlights the relevance of studying the different components and molecular mechanisms that control the homeostasis of lipid metabolism in order to better understand one of the key metabolisms of this pathogen and its relationship with its virulence. Only fragmented information is currently available about the mechanisms involved in the regulation of FA and MA metabolism in response to environmental conditions. Until now, several pieces of evidence have suggested that post-translational modifications using a Ser/Thr protein kinase (STPK)-mediated phosphorylation conduct a tight regulation of MA biosynthetic enzyme activity [12][13][14][15]. Additionally, a few transcriptional regulators of MA metabolism have been identified, such as FasR that controls fas expression [16], and FadR and MabR that are regulators of the fasII operon expression [17,18]. The fact that both MabR and FasR are essential for M. smegmatis viability indicates that keeping a tight balance between the activities of the two FAS systems is crucial for mycobacteria metabolism.
In order to study the role of the de novo FA biosynthesis (i.e. the role of the type I FAS system in Mycobacterium lipid metabolism in vivo) we constructed a conditional mutant in the fas gene and analysed in detail the impact of reduced de novo FA biosynthesis on MA biosynthesis and on the global architecture of the cell envelope. Our results demonstrate the essentiality of FAS I for Mycobacterium viability and the relevance of this system as a central player in the maintenance of the correct balance between the different mycobacterial lipids.

DNA manipulation, plasmid construction and mutant generation
Isolation of plasmid DNA, restriction enzyme digestion and agarose gel electrophoresis were carried out by conventional methods [20]. Genomic DNA of M. smegmatis was obtained as described previously [22]. For the construction of the M. smegmatis mutant allele Ptr:fas MS , the 5 0 region of the M. smegmatis fas gene (Msmeg_4757) was amplified with the primers FasMS-FNsiI (5 0 -TGGATGCATGTGACGATCTACGAACACGA-3 0 ) and FasMS-RXbaI (5 0 -GGTCTAGACTGCAGCTTCCACAGGTA GG-3 0 ). The 1517 bp PCR product was cloned in pCR BluntII TOPO (Invitrogen) and digested with NsiI/XbaI. The fragment was inserted in a pMP349 derivative plasmid that contains Ptr from pFRA50 cloned into EcoRI/PvuII sites, obtaining pFR42 plasmid. This plasmid has been sequenced in order to see that no errors were introduced during amplification. Finally, pFR42 was digested with SpeI/XbaI and cloned in the ts plasmid pPR27 [23], generating pFR18 plasmid. This plasmid was used to transform M. smegmatis cells harbouring the plasmid pFRA42B carrying the TetR/Pip OFF system (WT-pFRA42B) [24]. One of the Sm/Am-resistant transformants was grown at 308C and plated at 428C to promote plasmid recombination. The recombination event that left fas under the control of ATc was confirmed by PCR and the resultant strain was named fas cKD.
Fatty acid synthesis was allowed to proceed for 30 min at 378C. Fatty acids were separated following saponification (0.5 ml 20% KOH-50% methanol, 30 min, 1008C), acidification (0.3 ml 6 M HCl) and extraction ( petroleum ether, four times with 1 ml each time). The extracted fatty acids were transferred to scintillation vials, dried at 508C and dissolved in 0.2 ml of diethyl ether. Radioactivity was measured by scintillation counting after the addition of 1 ml of scintillation liquid (Perkin Elmer).

RNA techniques
RNA was extracted from M. smegmatis fas cKD and WT-pFRA42B grown in the presence and absence of 200 ng ml 21 ATc using Direct Zol RNA MiniPrep (Zymo Research).
qRT-PCR was performed using second strand cDNA as template, generated with SuperScript III Reverse Transcriptase (Invitrogen), random primers and a green fluorochrome as the indicator dye (qPCR master mix, Biodynamics). The expression of fas, acpS, fasR, inhA, fabH, hadB, fabD, acpM, kasA and kasB was quantified after normalization of RNA levels to the expression of sigA gene as previously described [27]. qPCR cycling conditions were as follows: 958C for 2 min followed by 40 cycles of 958C for 15 s, 588C for 15 s and 688C for 20 s. qPCR data are presented as fold difference of expression in fas cKD grown with ATc over that in the same strain grown without ATc 200 ng ml 21 , using the Pfaffl method [28]. sigA was the reference gene used for normalization of the expression level of each target gene. The sequences of all the primers used are listed in electronic supplementary material, . Fatty acid methyl esters (FAMEs) and mycolic acid methyl esters (MAMEs) were obtained after treatment of the radiolabelled cell pellets containing the same number of cells, with aqueous tetrabutyl ammonium hydroxide followed by esterification with methyl iodide and extraction with dichloromethane, as previously described [29]. The resulting solution of FAMEs and MAMEs was assayed for radioactivity in a Beckman liquid scintillation counter and then subjected to thin layer chromatography (TLC), using silica gel plates (TLC silica gel 60 F 254 , Merck) and hexane:ethyl acetate (9 : 1, v/v) as the developing solvent. Autoradiograms were produced by overnight exposure of the TLC plates to Carestream Kodak BioMax MR films. The developed films were then digitalized and the spots of 14 C-labelled FAMEs and MAMEs were quantified using Gel-Pro ANALYZER software.

Triacylglycerol analysis
The biosynthesis of triacylglycerols (TAGs) was analysed by incorporation of [

Fatty acid analysis
To analyse the fatty acid content, the FAMEs were extracted as previously reported [29] and quantified by GC-MS in a Shimadzu GCMS-QP2010 Plus chromatography-mass spectrometer on a SPB-1 capillary column (28 m Â 0.25 mm inside diameter). Helium at 1 ml min 21 was used as the carrier gas and the column was programmed at 68C min 21 from 408C to 3108C.

Acyl-CoA analysis
Acyl-CoAs were extracted following the protocol described by Cabruja et al. [31]. Bacterial cultures were harvested by centrifugation at 5800g for 15 min at room temperature. 0.625 nmol of 13 C 2:0 -CoA and 13 C 16:0 -CoA were added to each sample as internal standards. Separation of the different chain-length acyl CoAs was made by ion pairing-reverse phase high performance liquid chromatography (IP-RP-HPLC) with an rsob.royalsocietypublishing.org Open Biol. 7: 160277 Agilent 1200 SL instrument (Agilent Corporation, Santa Clara, CA, USA) having a Hypersil GOLD C 18 column (dimensions: 2.1 mm Â 150 mm Â 3 mm, Thermo Fisher Scientific, Waltham, MA, USA) at 308C. High-resolution mass spectrometry was performed in a Bruker micrOTOF-QII, a Q-TOF instrument with an electrospray ionization source (ESI). ESI parameters were optimized for acyl CoA's detection as follows: nebulizer pressure 1.0 bar, desolvation gas (N 2 ) flow 4.0 l min 21 , dry heater 2008C, capillary voltage 2800 V, end plate offset 2500 V. The target mass scan was set from 50 to 3000 m z 21 . The mass spectrometer parameters were set to detect the entire acyl CoA's ions in the single deprotonated form (M2H)2 operating in the negative mode.

Total lipid analysis
Total lipid extractions for lipidomic analysis were prepared as follows: cell pellets were washed with 50 mM ammonium acetate, pH 7.8 and transferred to 10 ml glass tubes containing 5 ml CHCl 3 /CH 3 OH (2 : 1, v/v). The samples were incubated overnight at 48C with gentle agitation. After centrifugation, bacterial pellets were subjected to an additional extraction using CHCl 3 : CH 3 OH (1 : 2, v/v) for 2 h. Organic extracts were pooled and dried under nitrogen at 48C. Lipids were resuspended in 3 ml of CHCl 3 , washed with 3 ml of H 2 O and the organic phase was transferred to preweighted glass tubes, dried under nitrogen at 48C and reweighted on a microbalance. Extracts were dissolved in CHCl 3 :CH 3 OH (1 : 1, v/v) at 1 mg ml 21 and centrifuged at 3000g for 5 min. Lipids were analysed in an Agilent 1200 series HPLC system with a Reprospher (Dr Maish) 100 C8 column (1.8 mm Â 50 mm Â 2 mm). The flow rate was 0.3 ml min 21 in binary gradient mode with the following elution programme: the column was equilibrated with 100% mobile phase A [CH 3 OH:H 2 O (99 : 1, v/v), containing 0.05 mM AcNH 4 ]. 2 ml of each sample, containing 13 mg of total lipids in CHCl 3 , were injected, and the same elution conditions continued for 1 min, followed by a 12 min gradient to 100% mobile phase B (isopropanol:hexane:H 2 O (79 : 20 : 1, v/v), containing 0.05 mM AcNH 4 ) and holding 1 min at that condition. The column was equilibrated for 2 min with 100% mobile phase A before injection of the next sample. An Agilent 6500 series Q-TOF instrument with a Dual AJS ESI was used for mass analysis. Ionization gas temperature was maintained at 2008C with a 14 l min 21 drying gas flow, a 35 psig nebulizer pressure and 3500 volts. Spectra were collected in positive and negative mode from m/z 115 to 3000 at 4 spectra s 21 . Continuous infusion calibrants included m/z 121.051 and 922.010 in positive-ion mode and m/z 119.035 and 955.972 in negative-ion mode. CID-MS was carried out with an energy of 50 volts. For the comparative analysis, the column is conditioned by four successive mock injections with solvent cycling before randomized QCs and mycobacterial samples are analysed.

Statistical analysis
Data are reported as arithmetic means of the results obtained from three independent experiments + standard deviations. Statistical significance was calculated using t-test or ANOVA analysis as described in the figure legends. Statistical significance was accepted at p , 0.05.

Construction and growth behaviour of a fas conditional mutant in Mycobacterium smegmatis
In order to study the effect of impairing the de novo FA biosynthesis on M. smegmatis lipid homeostasis, we constructed an M. smegmatis conditional knockdown mutant of the fas-acpS operon. For this, the fas promoter was replaced by the repressible promoter Pptr in a strain carrying the TetR/Pip OFF repressible system (WT-pFRA42B), using an adaptation of the method developed by Boldrin et al. [24] (figure 1a). The correct recombination event leading to the replacement of Pfas by the regulated promoter Pptr was rsob.royalsocietypublishing.org Open Biol. 7: 160277 confirmed by genomic PCR (figure 1b). In the resulting conditional mutant strain, named fas cKD, expression of fas is downregulated by the addition of anhydrotetracycline (ATc) to the culture media, which leads to a partial depletion of the multi-domain FAS I. When this strain was plated on 7H10 solid medium containing 200 ng ml 21 ATc, no growth was detected, suggesting that the repression of fas-acpS expression from Pptr was stringent enough to prevent FAS levels compatible with cell viability (electronic supplementary material, figure S2). Growth of the parental strain WT-pFRA42B was unaffected by the presence of ATc at this concentration (electronic supplementary material, figure S2).
To study the effect of FAS depletion at different growth stages, the conditional mutant strain fas cKD was grown in 7H9 medium in the absence and in the presence of ATc 100 or 200 ng ml 21 . As depicted in electronic supplementary material, figure S2c, growth of the fas cKD strain was inhibited when ATc was added at the same time of the inoculum. Therefore, in order to find a condition where we could explore the impact of fas depletion in lipid metabolism, ATc was added at different time points after inoculation of the culture. Electronic supplementary material, figure S2, shows that in the absence of ATc, the mutant strain exhibited a typical exponential growth curve, comparable with the isogenic strain WT-pFRA42B; however, when ATc was added at the beginning of the exponential phase, at OD 600 0.2 (T0), cells started to clamp and grow at a lower rate to finally stop dividing at OD 600 of approximately 0.8 -1 (figure 2a). Exogenous FA supplementation ( palmitic acid, oleic acid or a mix of them) was not able to restore growth of the fas cKD mutant after the addition of ATc (data not shown). When cell viability, determined as the number of colony forming units (CFU), was analysed before and after addition of ATc 200 ng ml 21 , the fas cKD strain showed a reduction of 31% in CFU numbers 12 h after the addition of ATc (figure 2a), thus confirming the essentiality of fas for the survival of M. smegmatis in axenic cultures.
In order to correlate cell growth arrest to the expression level of fas, the relative amount of fas mRNA was measured by quantitative RT-PCR after 12 h of exposure to ATc, before cells stopped growing and viable counts decreased (T3). Relative amounts of fas mRNA in fas cKD were compared in cell cultures grown in the presence or absence of ATc, representing the physiological level of fas mRNA (figure 2b). In the presence of ATc, the levels of fas mRNA in the fas cKD strain decreased dramatically (approx. 90%) compared with the levels found in the absence of ATc. In addition, it is worth noting that the level of fas mRNA in the parental strain WT-pFRA42B was not affected by the presence of ATc in the medium (data not shown). rsob.royalsocietypublishing.org Open Biol. 7: 160277 FAS II activity. As observed in figure 3, in the presence of ATc the activity of the FAS I system was 25% lower at T3 and barely detected after 24 h of incubation (T4) when compared with the same enzymatic activity determined in the protein extracts the fas cKD cultures grown without ATc. Surprisingly, when the same cell-free extracts were used to analyse FAS II activity, we found that this enzyme activity was unaffected at T3 in the cultures grown with ATc and showed only a slight reduction (17%) at T4, probably due to a pleiotropic effect on cell metabolism triggered by the strong depletion of FA biosynthesis (electronic supplementary material, figure S3).
In order to determine the levels of metabolic activity in the fas cKD strain, we performed [ 3 H] leucine labelling experiments at different time points after the addition of ATc 200 ng ml 21 and compared with the culture grown without ATc. As shown in electronic supplementary material, figure  S3, the metabolic activity of the mutant cells was comparable with that of the control strain for about 12 h after the addition of ATc. However, at the time the cells stop dividing (T4), we observed a 35% reduction of [ 3 H] leucine incorporation, suggesting a decrease in cell viability.
To further analyse the consequences of a reduced FAS I activity in lipid metabolism, we determined the acyl-CoA composition in the fas cKD mutant strain in the presence or absence of ATc. Acyl-CoAs extracted from a fixed number of cells were analysed by LC-MS as previously described [31]. As shown in figure 4a, an increase in the content of acetyl-CoA and malonyl-CoA was observed in the cultures grown in the presence of ATc. This result is consistent with an accumulation of these metabolites as a consequence of a reduced FAS I activity under these growth conditions. Furthermore, the lower levels of FAS I activity strongly impacted on the synthesis of the complete repertoire of medium-chain-length acyl-CoAs ( figure 4b). Surprisingly, the levels of the longchain acyl-CoAs synthesized by the mutant showed no differences for C 20 -CoA and C 22 -CoA, and a relatively small difference (25%) for C 24 -CoA (figure 4c). Moreover, the levels of a-carboxy-C 24 -CoA, a key intermediate for MA biosynthesis, showed increased levels in the ATc treated cultures (figure 4c), confirming that the levels of C 24 -CoA were less affected than those corresponding to the medium-chain acyl-CoAs.
The FA composition of the fas cKD mutant strain grown in the presence or absence of ATc was analysed by GC-MS. As shown in figure 5 and electronic supplementary material, figure S4, the relative composition of FA changed upon the addition of ATc: the levels of medium-chain-length FA were reduced approximately 10% with the concomitant increase (approx. 12%) of the long-chain C 24 . These results are in agreement with the differences observed in the acyl-CoA content of the mutant strain.

Loss of FAS I activity differentially affected the de novo FA and MA biosynthesis
To study the impact of FAS I depletion in the de novo synthesis of FA and MA, [ 14 C] acetate labelling experiments were carried out with the fas cKD mutant strain. Cultures were grown with and without ATc 200 ng ml 21 and their lipid content analysed by radio-TLC. As shown in figure 6, de novo synthesis of FA was progressively and significantly reduced after 12 and 16 h exposure to ATc (T3 and T4, respectively). Remarkably, MA content did not decrease at these time points, although there were differences between the fas cKD mutant strain grown with and without ATc with regard to the relative abundance of the three classes of MAMEs (figure 6a). The relative percentages for a-, a 0 -and epoxy-mycolates from the fas cKD strain grown without ATc were 60, 30 and 10%, respectively (figure 6a). Similar profiles were obtained for the WT-pFRA42B isogenic strain grown with and without ATc. The percentages for the mutant strain fas cKD grown with ATc 200 ng ml 21 were 80, 14, 7% and 85, 12, 3% in T3 and T4, respectively, showing a significant increase in the abundance of a-mycolates (25% in T4) with a compensatory reduction of a 0 -mycolates and epoxy-mycolates ( figure 6). In addition to this alteration, in the presence of ATc the TLC mobility of each of the MA species showed a slight but reproducible increment, compared with the mutant strain grown without ATc, suggesting the prevalence of MA of longer carbon chain lengths. When the same samples were run under different solvent conditions, in order to improve the separation of the different species of mycolates, the fas cKD strain grown with ATc clearly showed a different distribution of MA species and all of them had a bigger Rf compared with the ones extracted from the cultures grown without ATc or from the isogenic strain WT-pFRA42B (figure 6b).

Lipidomics of the fas cKD mutant
Based on the previous results showing important qualitative and quantitative modifications on the acyl-CoA pools and on the FA and MA contents of the fas cKD mutant strain, compared with the isogenic WT-pFRA42B strain, we decided to study in more detail the impact of low FAS I activity levels on the global lipid profile. For this, we performed comparative lipidomic analyses between the conditional fas mutant and its isogenic strain. Briefly, we extracted cell-associated lipids from the fas cKD mutant strain grown in the absence and in the presence of ATc (T3) and analysed the total lipid content by high-performance liquid chromatography-mass rsob.royalsocietypublishing.org Open Biol. 7: 160277 spectrometry (HPLC-MS). As a control, we extracted and analysed lipids from the isogenic strain WT-pFRA42B grown in the same conditions. Figure 7 shows the results obtained for the lipids analyses carried out in and negative ion mode. The most significant differences in the content of MA between the wild-type and mutant strains grown in the presence or absence of ATc were those found for the relative composition of a, a 0 and epoxy mycolates, as had been previously observed in the radio-TLC, and in the size of the molecules, on average 1-2 carbons longer in the mutant with reduced levels of FAS I activity (figure 7; electronic supplementary material, figure S5a,b). Furthermore, tandem  Another relevant change observed in the qualitative composition of the PL was the significant increase in the double bonds of the acyl residues, most significant being those of the CL, phosphatidylglycerol (PG) and PI (electronic supplementary material, figure S8). Analysis carried out in the positive mode was more difficult in terms of the relative content of each of the molecular species identified, such as diacylglycerol (DAG), monomeromycolyl DAG (MMDAG), GroMM, TMM, phosphatidylinositol mannoside (PIM) and TAG, due to the high content of the latter compound in the samples.

Central role of triacylglycerides in the sustained synthesis of mycolic acids in the fas cKD mutant
One of the most intriguing results of the studies carried out with the fas cKD mutant strain was the continuous synthesis of MA at T3 and T4 in the presence of ATc, when the cells are deprived of most of the FAS I enzyme and when the de novo FA biosynthesis appears strongly compromised ( figure 6). Considering that FAS I provides the acyl-CoAs that are further elongated by FAS II with malonyl-ACP to synthesize the meromycolic acids, it was expected that MA biosynthesis would follow the same detrimental impact as the de novo FA biosynthesis in the ATc-treated cultures. Therefore, in order to find out the source of acyl-CoAs under these conditions we analysed the impact of the reduction of FAS I activity in the de novo TAG biosynthesis. For this, cultures of the fas cKD mutant strain were grown with and without ATc 200 ng ml 21 , labelled with [ 14 C] acetate and their TAG content analysed by TLC. As shown in figure 8a, de novo synthesis of TAG was progressively reduced after 12 and 16 h exposure to ATc (T3 and T4, respectively). The TLC also showed an accumulation of free FA that increased throughout time. This result could suggest that besides a reduced TAG biosynthesis, the cells started to actively degrade the accumulated ones in order to compensate the lack of FAS I activity and provide FA to be used in other metabolic pathways (e.g. in MA biosynthesis). This hypothesis was tested treating the cultures with tetrahydrolipstatin, an inhibitor of lipases marketed as Orlistat, which was previously shown to inhibit TAG hydrolysis in mycobacteria [32]. As shown in figure 8b, total TAG content in the fas cKD strain grown in the presence of ATc and Orlistat is higher than in the strain grown in the presence of ATc but in the absence of Orlistat, and this difference is bigger when higher concentrations of Orlistat are used, suggesting that the inhibitor is clearly having an inhibitory effect in the lipase/s involved in TAG hydrolysis. Moreover, [ 14 C] acetate labelling experiments with the fas cKD mutant strain treated with Orlistat revealed a clear reduction in MA biosynthesis when the cultures were grown with ATc (figure 8c), suggesting that MA biosynthesis relies in TAG hydrolysis when the acyl-CoAs produced by the FAS I system become limited.

Cross-talk between the FAS I and FAS II systems
Mycobacterium has an extremely complex lipid metabolism. In order to preserve lipid homeostasis it is essential for these bacteria to maintain a tight regulation between the FAS I and FAS II components. Therefore, to unveil the cross-talk between these two systems and to understand the response of the FAS II genes to the knock down of fas expression, we performed a quantitative RT-PCR experiment with RNAs extracted from the fas cKD strain grown in the presence or absence of ATc. As shown in figure 9, cultures grown in the presence of ATc showed a markedly reduced expression of the fas-acpS operon, 90% lower expression for the fas gene and 85% for acpS than that found in the absence of ATc. The higher level of expression observed for the transcriptional activator of the fas-acpS operon, fasR, is probably a response to try to keep fas and acpS expression at their normal levels. The fact that MA biosynthesis remained active after the treatment of the mutant with ATc suggested that the MA biosynthesis genes should be actively transcribed in this conditions. Indeed, as shown in figure 9, the complete fasII operon genes ( fabD-acpM-kasA-kasB) showed an increased level of transcription in the mutant strain treated with ATc. Moreover, the expression level of fabH, a key component that links FAS I and FAS II activities, was also higher upon treatment of the fas cKD strain with ATc ( figure 9).

Discussion and conclusion
Mycobacterial survival depends on membrane lipid homeostasis and on the ability to adjust lipid composition to adapt the bacterial cell to different environments. This is especially relevant for M. tuberculosis, a pathogen with a very complex lifestyle capable of modulating its metabolism in response to the environmental changes within the host, where several components of the unusual lipid-rich cell wall of this organism play a key role in the outcome of the infection [33]. Therefore, understanding the interaction between FAS I and the metabolic pathways that rely on FAS I products to synthesize the complex cell wall lipids, as well as the membrane PL and the storage compounds TAG, is a key step to better understand how lipid homeostasis is regulated in mycobacteria.
In order to gain information regarding this critical regulatory process of lipid metabolism in mycobacteria, we initiated this study by constructing a conditional mutant in the fas-acpS operon of the model organism M. smegmatis. The genetic and physiological characterization of this mutant showed that FAS I is a housekeeping component of the mycobacteria lipid metabolism and that the absence of this function cannot be compensated in vitro by the addition of FA to the culture medium. This result is consistent with those obtained for Gram-negative and most Gram-positive bacteria, where the inhibition of de novo FA biosynthesis by specific FAS II inhibitors cannot be overcome by supplementing FA to the media or by the host lipids in infection models [34,35].
The conditional fas cKD mutant showed a neat growth response to the addition of ATc. The mutant strain significantly slowed down its growth rate 12 h after the addition of ATc and stopped dividing soon after. This growth defect is well correlated with the lower expression of fas and the concomitant and specific reduction of FAS I activity in the first hours after ATc treatment. As could be expected, the lower levels of FAS I provoked an accumulation of its substrates, acetyl-CoA and malonyl-CoA, and a strong reduction on the absolute levels of the C 12 -C 18 acyl-CoAs. Interestingly, the impact of the reduced FAS I activity on the levels of the long-chain acyl-CoAs (C 19 to C 24 ) was rather weak. It is well established that FAS I exhibits a bimodal product behaviour and that the two dominant products are C 16-18 -CoA and C 24 -CoA in M. smegmatis. However, it has also been demonstrated that the overall rate of FA biosynthesis as well as the product distribution of this enzyme becomes affected by endogenous polysaccharides or by interactions of the mycobacterial FAS I with other FA processing enzymes like the FAS II system [25,[36][37][38]. Thus, the unexpected accumulation of long-chain acyl-CoAs, when FAS I was downregulated, could be the result of a biochemical change in the bimodal behaviour of the FAS I system when the cells become stressed by the deficiency of de novo FA biosynthesis. Alternatively, the accumulation of long-chain acyl-CoAs could also represent a differential degradation of the different chain-length acyl-CoAs through the b-oxidation cycle. The relative composition of total FA in the fas cKD mutant also shows a bias towards the long-chain FA, being the most abundant C 24 FA under conditions of fas downregulation, suggesting that TAG and PL may have been enriched in long acyl chains.
A very intriguing result was the continuous synthesis of MA at T3 and T4 of the growth curve, where de novo FA biosynthesis appears severely compromised. In fact one would expect that the inhibition of de novo FA biosynthesis would also result in the inhibition of MA synthesis, because the FAS II system and the long-chain acyl-CoA carboxylase 4 utilize the products of FAS I as substrates of their corresponding reactions to generate the meromycolic acid and the carboxy-C 24 -CoA, respectively, before the final step of condensation mediated by Pks 13 [39]. Therefore, to solve this puzzle and to seek for the possible source(s) of these precursors we  rsob.royalsocietypublishing.org Open Biol. 7: 160277 inhibited the degradation of TAG with Orlistat, a strong lipase inhibitor. The results showed that the inhibition of TAG degradation impaired, at least partially, de novo synthesis of MA under conditions of FAS I reduction, suggesting that FA released from TAG can compensate, to some extent, for the absence (or reduced levels) of de novo FA synthesized by FAS I. It has long been known that under starvation conditions mycobacteria depend on TAG mobilization for energy production [40]. However, to our knowledge, this is the first time that TAG are experimentally demonstrated to be a reservoir of FA as precursors for the biosynthesis of more complex molecules in bacteria.
A detailed analysis by LC-MS of the MA species synthesized at late time points under conditions of low levels of FAS I revealed that a MA not only became more abundant with respect to the other species, a 0 and epoxy, but also longer on average. This result suggests that FAS II probably loses its natural regulation either because of the absence of substrate competition by FAS I or due to other more complex regulatory processes. In fact, it was noticeable that almost all the FAS II encoding genes, with the exception of inhA, were upregulated as a response to fas repression. This condition is similar to that where KasA is overexpressed and where the outcome was the synthesis of longer molecules of meromycolic acids [41,42]. The low levels of FAS I activity provoked a clear unbalance on the relative composition of the acyl-CoAs pool, which not only are the precursors for the biosynthesis of other complex lipids but also the signal molecules sensed by transcriptional regulators of lipid metabolisms in several species [43][44][45][46], including FasR of M. smegmatis, the positive regulator of the fas-acpS operon [16]. Moreover, we recently found that these effector molecules are also sensed by MabR, the transcriptional regulator of the fasII operon of mycobacteria (to be published elsewhere). Therefore, by manipulating the expression of fas to induce a conditional depletion of FAS I we showed that  rsob.royalsocietypublishing.org Open Biol. 7: 160277 not only the de novo synthesis of FA is affected, but that this metabolic stress produces a strong and pleiotropic effect on the regulation of other lipid related genes (lipid homeostasis) that inevitably should lead to cell death.
In previous work, we also showed that a modification of the expression levels of the fas-acpS or the fasII transcriptional regulators, fasR and mabR, respectively, also provoked a loss of the natural cross-regulation between the FAS I and FAS II systems [16,17]. A detailed lipidomic analysis of the fas cKD mutant strain showed some remarkable features besides the ones related to the MA that were discussed above. LC-MS analysis of the PL synthesized under conditions of fas repression revealed a strong reduction on the relative composition of PI, which dropped almost four-fold compared with the fas cKD strain growing without ATc or the isogenic strain. PI and its metabolically derived products such as PIM and linear and mature branched lipomannan are prominent PL/lipoglycans of Mycobacterium sp. and have been shown to be essential for cell viability [47]. The preferential impact in the biosynthesis of this PL is intriguing and might be related to the regulation of the PI synthase. When analysing the length of the acyl moieties esterified to the different PL of the mutant, it was noticeable that only PE and CL contained C 24 FA. These FA were not present in the PL of either the mutant strain grown in the absence of ATc or the isogenic strain WT -pFRA42B. Another interesting feature that came out of this analysis was that the FA present in the PL of the ATc treated mutant contained higher levels of unsaturation. This phenomenon might occur to mitigate the rigidity of the plasma membrane imparted by the presence of long-chain FA in the PL [48].
A number of studies indicate that M. tuberculosis utilizes a restricted set of host-derived nutrients to persist within its host cell. Most notably, FA and cholesterol are important  21 , and in the presence or absence of Orlistat 75 and 100 mM. Culture aliquots containing the same number of cells were centrifuged at T3 for total lipid extraction followed by analysis on silica gel TLC plates developed in hexane/diethylether/acetic acid (75 : 25 : 1, v/v/v). Total lipids extracted from WT-pFRA42B were loaded as control. Chemical staining with Cu-phosphoric stain was used for detection. (c) Reduced TAG hydrolysis upon treatment with tetrahydrolipstatin (Orlistat) limits the de novo mycolic acid biosynthesis. Aliquots from fas cKD cultures grown with or without ATc 200 ng ml 21 , and in the presence or absence of Orlistat 75 and 100 mM, were labelled with [1-14 C]-acetate at T3 for 1 h at 428C. The 14 C-labelled methyl esters of mycolic acids (MAMEs) and fatty acids (FAMEs) were extracted and analysed on silica gel TLC plates developed in hexane/ethyl acetate (9 : 1 v/v). Aliquots extracted from the same number of cells for each sample were loaded.  Figure 9. The reduced expression of fas in the fas cKD mutant strain leads to a global upregulation of genes required for mycolic acid biosynthesis. The relative expression profile of fas, acpS, fasR, fabD, acpM, kasA, kasB, inhA, hadB and fabH was determined by qRT-PCR at T3 and T4. The cycle threshold (CT) value of each gene of interest was normalized by using the housekeeping gene sigA. Values represent the mean difference between the fas cKD strain grown with and without ATc 200 ng ml 21 . The quantification was performed in triplicate + standard deviation. rsob.royalsocietypublishing.org Open Biol. 7: 160277 nutrients during infection in vivo or in infected macrophages in cultures [49]. Developing a conditional fas mutant in M. tuberculosis will help to define the essentiality of de novo FA biosynthesis in the survival of this pathogen during infection, and in that way it should help to unambiguously establish if FAS I is a good target for drug development. Furthermore, a conditional mutant of FAS I in M. tuberculosis will also help us to study how the complex regulatory network that links the different lipid-related pathways in this deadly pathogen responds to or tries to compensate for a deficiency of de novo synthesized FA.
Authors' contribution. M.C., S.M., Y.T.T. and J.L. carried out the experiments and participated in data analysis; H.G. and G.G. conceived, designed and coordinated the study, participated in data analysis and wrote the manuscript. All authors gave final approval for publication.
Competing interests. We declare we have no competing interests.