Permissive Fatty Acid Incorporation in Host Environments Promotes Staphylococcal Adaptation to FASII Antibiotics

Development of fatty acid synthesis pathway (FASII) inhibitors against the major human pathogen Staphylococcus aureus hinges on the accepted but unproven postulate that an endogenously synthesized branched chain fatty acid is required to complete membrane phospholipids. Evidence for anti-FASII efficacy in animal models supported this view. However, restricted test conditions used previously to show FASII antibiotic efficacy led us to investigate these questions in a broader, host-relevant context. We report that S. aureus rapidly adapts to FASII antibiotics without FASII mutations when exposed to host environments. Treatment with a lead FASII antibiotic upon signs of infection, rather than just after inoculation as commonly practiced, failed to eliminate S. aureus from infected organs in a septicemia model. In vitro, addition of serum facilitated rapid S. aureus FASII bypass by environmental fatty acid (eFA) replacement in phospholipids. Serum lowers membrane stress, leading to increased retention of the two substrates required for exogenous fatty acid (eFA) utilization. In these conditions, eFA occupy both phospholipid positions 1 and 2, regardless of anti-FASII selection. This study revises conclusions on S. aureus fatty acid requirements by disproving the postulate of fatty acid stringency, and reveals an Achilles’ heel for using FASII antibiotics to treat infection in monotherapy.


Introduction
Fatty acid synthesis (FASII) pathway enzymes are priority targets for ongoing drug development 21 against methicillin resistant Staphylococcus aureus (MRSA) (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11). However, anti-FASII efficacy 22 remains a critical point of debate (12)(13)(14). Remarkably, S. aureus FASII sensitivity versus tolerance 23 hinges on a single issue; whether environmental fatty acids (eFA) can occupy the presumably 24 stringent 2-position of membrane phospholipids (the 1-position is permissive) (7,14). Anti-FASII 25 resistance due to mutations affecting the target enzyme or to horizontal transfer of an antibiotic-26 resistant gene homologue may occur and are often antibiotic-specific (15,16). In contrast, resistant 27 mutants that allow compensatory fatty acid utilization at both phospholipid positions were isolated 28 and also found in clinical isolates; these mutants map in FASII initiation genes distinct from the gene 29 encoding the antibiotic target protein (17)(18)(19). Despite emergence of mutations, continued FASII-30 targeted drug development is rationalized by the accepted postulate that the general wild type S. 31 aureus population must synthesize branched chain fatty acid anteiso 15:0 (ai15) to complete 32 membrane phospholipids (1, 3, 5, 8-11, 14, 20). We investigated this postulate, and report here that 33 an alternative antibiotic adaptation mechanism is functional in host environments, which enables S. tests are decisional checkpoints for antibiotic development. FASII antibiotic challenge tests to date 40 administer treatments within minutes to a few hours post-infection (summarized in (19)), i.e., prior 41 to bacterial dissemination to host organs, and before clinical symptoms would call for antibiotic 42 treatment (21,22). This consideration guided the design of the infection and treatment protocol 43 used here (Fig. 1A). S. aureus methicillin resistant strain USA300 was administered by the intravenous 44 route. Antibiotic treatments were initiated 16 h (T16) post-infection, at which time animals exhibited 45 signs of sickness (lethargy and ruffled fur). Group 1 received no treatment. At T16, Group 2 received 46 AFN-1252 a pipeline FASII antibiotic AFN-1252 targeting FabI, an enoyl-acyl-carrier-protein-47 reductase, following recommended dosing (23). Group 3 received vancomycin, which was used to 48 validate that treatment starting at T16 was feasible. At T40 (i.e., 24 h post antibiotic treatment), 49 bacterial counts were significantly lower in organs of both antibiotic-treated animals compared to 50 the untreated group, attesting to AFN-1252 activity (Fig. 1B). In contrast, at T88, whereas 51 vancomycin-treated animals were essentially free of bacteria, all organs from AFN-1252-treated mice 52 still contained S. aureus CFUs. Bacterial counts were increased 10-fold in kidneys (to 5x10 6 ; p= ≤0.01), 53 cultures, to 8 h and 10 h for USA300 and Newman respectively, and was followed by near-normal 66 growth ( Fig. 2A, Fig. S1A). If S. aureus outgrowth were due to triclosan titration by serum, FASII 67 would remain active, so that bacterial fatty acid composition would be endogenous. However, the 68 contrary occurred: bacterial fatty acid profiles during outgrowth in SerFA-Tric medium were totally 69 exogenous (Fig 2B, Fig. S1B). Addition of albumin, a major serum constituent, rather than full serum 70 also resulted in FASII bypass (available upon request). Similarly, when USA300 was grown with liver 71 or kidney extracts (without added fatty acids) and triclosan, outgrowth kinetics were similar to those 72 of SerFA-Tric cultures, and cells bypassed the FASII block by incorporating fatty acids from organ 73 lipids (Fig. S2). 74 Importantly, pre-incubation in SerFA prior to FASII antibiotic treatments shortened the time 75 prior to S. aureus outgrowth. USA300 was challenged with the AFN-1252, which led to a longer (10 h) 76 latency phase prior to outgrowth than did triclosan. However, pre-incubation in serum shortened 77 latency compared to that in non-selective medium to around 6½ h for both drugs (Fig. S3), indicating 78 that bacterial pre-exposure to the lipid-rich host environment contributes to limiting FASII antibiotic 79 efficacy. 80 Staphylococcus epidermidis, haemolyticus, and lugdugensis are emerging pathogens that, like 81 S. aureus, synthesize branched chain fatty acids (Fig. S4). Representative strains were grown in SerFA 82 and treated with AFN-1252 as above. All cultures grew after overnight incubation and displayed 83 exogenous fatty acid profiles, indicating that these staphylococcal species also bypass FASII 84 inhibitors. 85 These results show that in serum, S. aureus and other staphylococcal species escape anti-86 FASII inhibition and maintain robust growth by replacing endogenously synthesized fatty acids with 87 eFA. They indicate that serum does not prevent, but actually enhances eFA incorporation by S. 88 aureus. 89 90 Adaptation to FASII inhibitors is not due to FASII mutations. Mutations in FASII initiation genes may 91 lead to antibiotic resistance (14,(17)(18)(19). We monitored mutations in FASII antibiotic-adapted USA300 92 or Newman strains by DNAseq, using FASII inhibitors triclosan or AFN-1252 (Table S1). DNAseq of 93 USA300 grown in BHI and SerFA, and Newman grown in BHI were used as references. Adaptation 94 was confirmed by exogenous fatty acid profiles of antibiotic-grown samples in 12-14 h cultures. Eight 95 of nine adapted strains carried wild type FASII genes (the exception was mutated in SAUSA300_1476 96 encoding FASII initiation gene accB). One isolate displayed no detectable genome mutations. The 97 other clones carried SNPs corresponding to commonly found variants and are thus likely unrelated to 98 FASII antibiotic adaptation (described in Table S1). The absence of FASII mutations in FASII-antibiotic-99 adapted clones in serum distinguishes this adaptation mechanism from resistance due to FASII 100 mutations (15,18). S. aureus evasion of FASII antibiotics in serum identifies a novel strategy of 101 condition-dependent adaptation. 102 103

Serum lowers fatty-acid-induced bacterial membrane permeability and improves fitness. 104
Numerous fatty acids perturb bacterial membrane integrity and are a source of stress, while serum 105 albumin neutralizes these effects (24, 25, 27, 28). Accordingly, serum abolished the eFA-provoked 106 growth lag in non-selective medium ( Fig. 2A). Serum effects on S. aureus cell vitality and 107 permeability, and cell state, were examined. Free fatty acids had strong permeabilizing effects on 108 cells from FA and FA-Tric, as compared to BHI cultures, as evaluated by fluorescence microscopy; 109 these effects were offset by serum in SerFA and SerFA-Tric cultures (Fig. 2C, Fig. S5A, S5B). Plating 110 efficiency was ~10 3 -fold higher after 6 h growth in SerFA-Tric compared to FA-Tric (Fig. 2D). The 111 accumulation of tetrads comprising mixed-stained cells in triclosan-treated cultures suggested a cell 112 division block, which could explain the observed latency prior to outgrowth (Fig. S5C). Importantly, 113 serum facilitates fatty acid incorporation in the latency period, as seen in 6 h FA-Tric and SerFA-Tric 114 cultures (Fig. 2E). The bacterial stress state in FA-Tric supplemented or not with serum was also 115 assessed by a proteomics approach (Fig. S6). Differences in stress-related protein abundance 116 between anti-FASII-treated versus control cultures were overall more pronounced when serum was 117 absent. Serum therefore improves S. aureus fitness in fatty acid-containing environments and 118 contributes to FASII antibiotic adaptation via eFA incorporation. 119 120 Retention of the FASII bypass precursors acyl carrier protein (ACP) and eFA is increased in serum. 121 adaptation, ACP and eFA. ACP is required for both de novo fatty acid synthesis via FASII, and for eFA 123 incorporation in the phospholipid 2-position ( Fig. 3A) (18,29). eFA induce membrane leakage that 124 depletes S. aureus ACP pools, and could limit eFA incorporation during FASII inhibition (14,28). S. 125 aureus ACP pools were compared by immunoblotting using anti-ACP antibodies in total extracts from 126 cultures grown without and with serum and triclosan (Fig. 3B). ACP levels were lower in extracts from 127 cells grown in eFA compared to BHI medium, as reported (28). ACP was barely detected in FA-Tric-128 grown cells after 2 h, and remained low at 4 h. In contrast, serum addition reversed this effect, 129 leading to greater intracellular ACP availability.  (Fig 2B, Fig. S1B). However, eFA incorporation in 143 both S. aureus phospholipid positions could be a last resort choice when the preferred substrate ai15 144 is unavailable. Alternatively, the use of eFA versus ai15 may simply depend on intracellular substrate 145 availability, which increases in serum. eFA incorporation was assessed in non-selective conditions to 146 discriminate between these alternative hypotheses. To remove ambiguity in distinguishing 147 endogenous from exogenous fatty acids in phospholipid identifications, a cocktail of unsaturated eFA 148 17:1trans (tr) and 18:1cis prepared in delipidated serum (dSer2FA medium) was used to supplement 149 S. aureus USA300 growth. Although structurally distinct from S. aureus endogenous fatty acids, 17:1tr 150 and 18:1cis did not interfere with growth in the serum-containing medium (OD 600 = 6.4 for both BHI 151 and dSer2FA cultures at 6 h). In this non-selective growth condition, 17:1tr and 18:1cis comprised 152 about 60% of the total fatty acid content (Fig. 4A, Fig. S7A) at the expense of straight chain saturated 153 fatty acids, which decreased from 50% in BHI, to about 10% in dSer2FA cultures. Importantly, eFA 154 occupied both positions in phosphatidylglycerol (PGly) phospholipids after 6 h without antibiotic (Fig.  155 4B, Table S2). These results prove that wild type S. aureus incorporates dissimilar fatty acids in both 156 phospholipid positions without the need for FASII antibiotic selection, and rule out previously 157 assumed fatty acid selectivity of phospholipid-synthesizing enzymes. 158 Fatty acid and PGly profiles were determined 4, 6, 8, 10, and 15 h after USA300 treatment 159 with AFN-1252 or triclosan, in parallel with OD 600 (Fig. S7B, Fig. S8). For both FASII antibiotics, 160 bacterial transition from mixed to exclusively exogenous fatty acids was concomitant with outgrowth 161 from latency starting at 8 h post-treatment, and was completed at 10 h (Figs. S8A, S8C). For both 162 FASII antibiotic treatments, predominant phospholipid species at 10 h were totally exogenous 163 (18:1cis|18:1cis, 17:1tr|17:1tr, and 18:1cis|17:1tr; Fig. S8B and S8D, Table S3). The sharp increase in 164 exclusively eFA-containing phospholipids coincides with the exit from latency, as expected from 165 coordination between membrane phospholipid synthesis and cell growth (33, 34). These results 166 show that when S. aureus membrane integrity is maintained, as by host constituents, eFA 167 incorporation is not stringent, and reflects competition between fatty acids synthesized by S. aureus     24), this report), and improve membrane integrity. Higher ACP and/or eFA-PO 4 pools drive PlsX directionality to eFA-ACP production. eFA-ACP, and endogenous acyl-ACP (if not blocked by anti-FASII), compete for phospholipid synthesis at the 2-position via PlsC. eFA occupy both phospholipid positions even without FASII antibiotics (Fig. 4B). eFA (green) and ACP (blue) abundance is represented by font size; phospholipids, "π" form; thick arrows, favored reactions; thin arrows, reduced reactions. Dashed circle, permeable membrane; solid circle, intact membrane. PlsY mediates eFA incorporation in the phospholipid 1-position; PlsX and PlsC, catalyze fatty acid insertion in the phospholipid 2-position. Only eFA processing is presented. Tested USA300 derivatives from the Nebraska library contained insertions in the following genes: spa (SAUSA300_0113 encoding immunoglobulin G binding protein A), SAUSA300_0226, SAUSA300_0242, SAUSA300_0407, SAUSA300_1177, and SAUSA300_1684 (see Table S1).
Growth media. BHI medium was used for S. aureus growth except in media containing liver extracts.
The two fatty acid mixture comprising C17:1trans and C18:1cis was used at a 1:1 ratio (0.25 mM each); the more rigid 17:1trans species was added to limit membrane fluidity due to C18:1cis.
Newborn calf serum (Sigma-Aldrich, France), or delipidated calf serum (Eurobio, France) was added to growth medium (10% final concentration) as indicated. Triclosan (referred to as "Tric"; Irgasan; Sigma-Aldrich) was added at 250 ng/ml and 500 ng/ml in medium without and with serum respectively; this corresponds to 15-30 times the minimum inhibitory concentrations (MIC) as determined on BHI medium (2). AFN-1252 (referred to as "AFN"; MedChem Express) was added at 500 ng/ml in all media, which corresponds to about 100-fold the reported S. aureus MIC (~4-8 ng/ml; (3,4)). FASII inhibitors were prepared in DMSO (1 mg/ml) for all in vitro experiments.   Bacteria are generally exposed to the lipid-rich host environment prior to antibiotic treatment. Here, USA300 was precultured in BHI and in SerFA medium, and then challenged with either triclosan (green triangles) or AFN-1252 (orange squares). Growth of antibiotic-challenged cultures without preadaptation (empty markers) or with preadaptation (preadapt; filled markers) was followed by optical density (OD 600 ) for 13 h. All FASII-antibiotic-treated cultures displayed exogenous fatty acid profiles at this time (available on reqques). Growth curves show the average and range of three biological replicates from one of two independent comparable experiments.       Table S3A (dSer2FA-AFN) and Table S3B (dSer2FA-Tric). BHI and dSerFA cultures (Fig. 4) were analyzed together with this experiment.