Discussion
Branched-chain fatty acids, specifically a 15:0 and a 17:0, are the predominant fatty acids in Gram-positive bacterial membranes. They are essential for growth and virulence in pathogenic species including S. aureus (Annous et al., 1997; Beck 2005; Pendleton & Yeo et al., 2022; Singh et al., 2008; Whaley et al., 2023). Indeed,a 15:0 fatty acid is essential for full activity of the majorS. aureus virulence two-component regulatory system SaeRS, andlpdA mutant cells are attenuated for virulence (Pendleton & Yeo et al., 2022). Given their critical roles, it is difficult to imagine that there is no redundancy in the pathways that lead to their synthesis. On the other hand, disrupting the BKDH complex results in BCFA auxotrophy. In the present study we took advantage of this phenotype and looked for mutants that might reveal an alternative route to BCFA synthesis.  Extragenic suppressor mutants isolated in TSB medium map to the regulatory region of a putative acyl-CoA synthetase gene (now referred to as mbcS ). These mutations result in overexpression ofmbcS and resolve the BCFA auxotrophy (Table 1, Figs 2-4) . Using genetic and biochemical approaches, we demonstrate herein that MbcS is an acyl-CoA synthetase with specificity for short, branched carboxylic acid substrates, and MbcS activity is required for salvaging exogenous carboxylic acids and aldehydes for BCFA synthesis when the BKDH complex is inactivated (Table 2, Figs 4-7) .
Our data are in strong agreement with a recent report by Whaley et al. , demonstrating MbcS is an AMP-forming acyl-CoA synthetase that selectivity catalyzes the activation of isobutyrate and 2-methylbutyrate. In that manuscript, they clearly show extracellular carboxylic acids are converted to their CoA derivatives and flow into the FASII elongation cycle, and argue that MbcS serves in a salvaging capacity (Figs 4-5) (Whaley et al., 2023). Our suppressor screen also points to MbcS, and kinetic analysis of the enzyme indicates a high (i.e., low micromolar) affinity fori C4 and a C5 substrates. This is consistent with the finding that TSB medium contains a trace amount of these compounds. Interestingly, unlike in TSB, the lpdAsingle mutant does not exhibit auxotrophy for BCFAs when grown in lysogeny broth, prompting Whaley et al. to hunt for mbcS.This is likely due to the relative levels of carboxylic acids in these two media. Indeed, this accounts for the distinct membrane fatty acid profiles seen for WT cells (Sen et al., 2016; Whaley et al., 2023). Moreover, BCFA auxotrophy is resolved in our suppressor mutants by incorporating i 14:0 BCFA into membranes derived fromi C4. This makes physiological sense - while FabH prefers a C5-CoA to initiate fatty acid synthesis, incorporating iso even fatty acids into the bilayer reflects not only the affinity of MbcS for substrate, but also the availability of the acyl-CoA precursor pool (Whaley et al., 2023). We note that in contrast to the Whaley et al. study where they report a ~5-fold increase in affinity fora C5 over i C4, we did not measure a significant difference in MbcS affinity fori C4 or a C5. The discrepancy may be explained by the assay used to determine kinetic constants. Our coupled assay may simply be not sensitive enough to resolve the Km values below 5 µM.
Our data validate and expand the idea that S. aureus salvages BCFA precursors by showing that branched-chain aldehydes are catabolized(Figs 6-7) . At the same time, it is intriguing that strains lacking a functional BKDH complex with heightened MbcS enzyme activity are able to grow, albeit poorly, in unsupplemented, defined medium. We posit this growth is the result of a cryptic alternative de novosynthesis pathway. Previous studies investigating BCFA synthesis in related Gram-positive bacteria inform our working model for an alternative route to BCFAs synthesis (Fig 8) . For instance, the Gram-positive bacterium Lactococcus lactis , which is largely used for the fermentation of dairy products, harbors an α-keto acid decarboxylase KdcA that converts the branched-chain α-keto acids into their respective branched-chain aldehydes, and these are recognized to be important for flavor formation (Smit et al., 2005). In USA300 strains, SAUSA300_0190 is annotated to encode an indole-pyruvate decarboxylase (IpdC). IpdC is 64% similar to L. lactis KdcA that has highest activity for the α-keto acid of valine (Smit et al., 2005), consistent with our GC-FAME data showing incorporation of valine derived, i1 4:0 BCFA in the lpdA mbcS1 and lpdA mbcS2  strains (Fig 3) . We propose that IpdC is a branched-chain α-keto acid decarboxylase. Three additional facts support this hypothesis: i) IpdCs are known to have broad specificity (Parsons et al., 2015); ii) ipdC expression is repressed by S. aureus CodY, which controls BCFAs synthesis by regulating the genes that direct the synthesis of the short BCFAs precursors (Waters et al., 2016); and iii) S. aureus synthesizes tryptophan but, unlike other organisms, does not catabolize the amino acid (Proctor & Kloos,1973; Spaepen et al., 2007; ). AlsSD (acetolactate synthase/decarboxylase) and CidC (pyruvate oxidase) also have significant similarity to KdcA. However, how these enzymes would fit into the model is unclear.
As described above, other staphylococcal species are commonly used in the food industry due to their ability to form aldehydes, which are important for flavor formation (Beck, 2005; Beck et al., 2002). S. aureus itself was described to synthesize nine different aldehydes among them the branched chain 2-methylbutyraldehyde, 3-methylbutanal, and 3-methylpropanal (Bos et al., 2013; Filipiak et al., 2012). Mass spectrometry data from S. xylosus showed these cells can convert 3-methylbutanal to its respective carboxylic acid, the 3-methylbutanoic acid (Beck et al., 2002). In addition, human commensals likeCutibacterium acnes and S. epidermidis also produce these compounds on skin (Bos et al., 2013; Duffy & Morrin, 2019; Tait et al., 2014; Verhulst 2011) and salvage them in laboratory culture in an MbcS-dependent manner (Fig 7) , suggesting these compounds are salvaged by S. aureus using this pathway during infection. Taken together, data from our lab and other labs indicate that S. aureus oxidizes the branched-chain aldehyde to the carboxylic acid. TheS. aureus genome encodes several potential aldehyde dehydrogenases (Alds); some of which have been characterized and are known to have broad specificity (Imber et al., 2018). Thus, we posit that one or more of these Alds converts the aldehyde derivatives of the branched-chain α-keto acids to their cognate carboxylic acids. MbcS then would serve to activate the carboxylic acid to its acyl-CoA to feed fatty acid synthesis. The α-keto acid decarboxylase and the aldehyde dehydrogenase(s) involved in this proposed alternative pathway are still to be found but this is currently an active focus of our laboratory.
Why has this pathway remained hidden until now? Overexpression ofmbcS (either by mutation of the native promoter or by artificial overexpression using an inducible promoter) restores growth in the BKDH-deficient strain (Figs 2-4) . This suggests that MbcS activity in laboratory cultures is normally too low to support growth. Interestingly, MbcS is a member of the AMP-forming family of acyl-CoA synthetases. These enzymes (including Rp IbuA used in this study) are known to be regulated by reversible lysine modification of a conserved lysine residue by acetylation (Crosby & Escalante, 2014; Crosby et al., 2010; Gardner et al., 2006). Acetylation down-regulates enzyme activity (Gardner et al., 2006; Starai et al., 2002). It is conceivable that MbcS is acetylated under our laboratory conditions, and overproducing MbcS increases enzyme activity because the fraction of MbcS that escapes the acetylation machinery is increased. Indeed, the GCN5-related N- acetyltransferase AcuA was recently shown to acetylate acetyl-CoA synthetase in S. aureus (Burckhardt et al., 2019). The role of AcuA in bacterial physiology is unknown. Whether MbcS is acetylated and is a substrate for AcuA is not known. We are actively pursuing these questions in our laboratory.