Abstract
Antibiotic resistance is an increasing threat for public health, underscoring the need for new antibacterial agents. Antimicrobial peptides (AMPs) represent an alternative to classical antibiotics. TAT-RasGAP317-326 is a recently described AMP effective against a broad range of bacteria, but little is known about the conditions that may influence its activity. Using RNA-sequencing and screening of mutant libraries, we show that Escherichia coli and Pseudomonas aeruginosa respond to TAT-RasGAP317-326 by regulating metabolic and stress response pathways, possibly implicating two-component systems. Our results also indicate that bacterial surface properties, in particular integrity of the lipopolysaccharide layer, influence peptide binding and entry. Finally, we found differences between bacterial species with respect to their rate of resistance emergence against this peptide. Our findings provide the basis for future investigation on the mode of action of this peptide and its potential clinical use as an antibacterial agent.
Introduction
The spread of antibiotic resistance in many bacterial species is severely limiting the benefits of antibiotics and a growing number of infections are becoming harder to treat (O’Neill, 2016). Therefore, there is a need for new antimicrobials that could be used in the treatment of bacterial infections. Antimicrobial peptides (AMPs), several of which are already in clinical trials with promising results, represent a large source of antibacterial agents (Kumar et al., 2018). They are attractive alternatives to classical antibiotics due to their broad-spectrum activity that allows the targeting of a wide variety of bacterial species (Di Somma et al., 2020). In addition, AMPs have a relatively simple structure that can be bioengineered to increase, for example, their stability under physiological conditions or their resistance to degradation by gastrointestinal tract enzymes after oral administration (Kong et al., 2020).
AMPs were first described as naturally occurring peptides produced by many different organisms. Thousands have been identified (Wang et al., 2016, Kumar et al., 2018). In bacteria, AMP-producing strains have an advantage over other strains or species during competitive colonization of ecological niches (Hassan et al., 2012). In multicellular organisms, AMPs such as the human cathelicidin LL-37 and the bovine bactenecin, are part of the innate immune system involved in the destruction of various microorganisms (Gennaro et al., 1989, Xhindoli et al., 2016).
Despite their diversity, AMPs share a number of common features: they are short peptides rich in cationic and hydrophobic amino acids and display an overall positive charge. To exert their biological activity, positively charged AMPs first interact with the negatively charged bacterial surface through electrostatic interactions (Brogden, 2005). This initial interaction with the bacterial surface is followed, for a majority of AMPs described up to date, by permeabilization and disruption of the membrane bilayer resulting in bacterial death. For example, this mechanism of killing has been demonstrated for melittin isolated from bee venom (Hong et al., 2019), human cathelicidin LL-37 (Mendez-Samperio, 2010), and polymyxin B derived from the Gram-positive bacterium Bacillus polymyxa (Srinivas and Rivard, 2017, Kumar et al., 2018).
TAT-RasGAP317-326 is a recently identified antimicrobial peptide that kills both Gram-positive and Gram-negative bacteria and has antibiofilm activity in vitro (Heulot et al., 2017, Heinonen et al., 2021). This peptide is composed of a cell permeable moiety, the TAT HIV 48-57 sequence, and a 10 amino acid sequence derived from the Src homology 3 domain of p120 RasGAP. TAT-RasGAP317-326 was initially identified as an anticancer compound that sensitizes cancer cells to genotoxins (Michod et al., 2004, Michod et al., 2009) and to radiotherapy (Tsoutsou et al., 2017). This peptide also inhibits cell migration and invasion (Barras et al., 2014c) and possesses anti-metastatic activity in vivo (Barras et al., 2014b). It can also directly lyse a subset of cancer cells by targeting plasma membrane inner leaflet-enriched phospholipids (Serulla et al., 2020) in a manner that does not involve known programmed cell death pathways (Annibaldi et al., 2014, Heulot et al., 2016). We have previously shown that the tryptophan residue at position 317 of the TAT-RasGAP317-326 peptide is essential for its activity against both eukaryotic and bacterial cells (Barras et al., 2014a, Heulot et al., 2017). Furthermore, we have reported that, despite its potent in vitro antimicrobial activity, TAT-RasGAP317-326 showed limited protection in a mouse model of Escherichia coli (E. coli)-induced peritonitis (Heulot et al., 2017). Physiological factors may have contributed to the poor biodistribution and rapid clearance of TAT-RasGAP317-326 and, subsequently, to its low efficacy in this setting (Michod et al., 2009).
In this study, we assessed TAT-RasGAP317-326 activity under various experimental settings to better characterize the peptide activity as well as the bacterial response to peptide exposure. Our findings provide important initial insights into the activity of the TAT-RasGAP317-326 peptide that will pave the way for further investigation on its antimicrobial properties.
Results
Divalent cations reduce TAT-RasGAP317-326 surface binding and entry into bacteria
P. aeruginosa grown under low Mg2+ conditions is resistant to EDTA, gentamicin and polymyxin B via a mechanism that involves outer membrane modifications (Macfarlane et al., 1999, McPhee et al., 2003, Olaitan et al., 2014). To assess whether the antimicrobial activity of TAT-RasGAP317-326 is also affected by Mg2+ levels, we assessed how Mg2+ modulated its minimal inhibitory concentration (MIC) and concentration inhibiting growth by 50% (IC50) in three laboratory strains, E. coli MG1655, E. coli ATCC 25922, and P. aeruginosa PA14. The results of these experiments are summarized in Table 1 and shown in detail in Supplementary Figures 1–3. The MIC of TAT-RasGAP317-326 in standard Luria-Bertani (LB) medium for E. coli and P. aeruginosa was determined to be 8 μM and 32 μM, respectively. There was a small difference in peptide MIC levels between LB and BM2 medium supplemented with 2 mM MgSO4 (BM2 Mghigh) for both E. coli and P. aeruginosa. However, these two bacterial species displayed an 8-fold decrease in MIC of TAT-RasGAP317-326 in BM2 containing 20 μM MgSO4 (BM2 Mglow) relative to 2 mM MgSO4 (BM2 Mghigh). Low magnesium in BM2 medium resulted in a 4-fold increase of MIC of polymyxin B in P. aeruginosa but had no impact in E. coli. Our results agree with earlier data that low Mg2+ increases P. aeruginosa resistance to polymyxin B (Macfarlane et al., 1999, McPhee et al., 2003). Altogether these findings show that low Mg2+ in culture medium renders bacterial cells more susceptible to TAT-RasGAP317-326 but not to polymyxin B, suggesting that TAT-RasGAP317-326 and polymyxin B attack bacterial cells through different mechanisms.
Because BM2 is a defined bacteriological medium, we also investigated whether Mg2+ affects the sensitivity to TAT-RasGAP317-326 in complex medium such as LB. We compared peptide MIC in LB and LB supplemented with 2 mM MgSO4 (LB Mghigh) and found that high Mg2+ increased peptide MIC in both E. coli and P. aeruginosa (Table 1), consistent with the data obtained with these two bacterial strains in BM2 medium. Moreover, the ability of TAT-RasGAP317-326 to hamper E. coli growth rate was clearly inhibited by 2 mM MgSO4 (Fig. 1, panels A and B). Therefore, high Mg2+ levels decreased bacterial sensitivity to TAT-RasGAP317-326 and this result was independent of the medium used. We also determined that the MIC of polymyxin B for P. aeruginosa decreased in the presence of high Mg2+ in LB (Table 1), which is in accordance with our data for P. aeruginosa in BM2 medium.
Would cations other than Mg2+ also render bacteria less sensitive to TAT-RasGAP317-326? Addition of Fe2+ and Ca2+ in culture medium decreased the sensitivity of E. coli to peptide (Fig. 1 panels C and D). We also found that high Mg2+ concentrations decreased bacterial sensitivity to TAT-RasGAP317-326 both in the context of sulfate and chloride counterions (Fig. 1 panels B and E). In contrast, ammonium sulfate did not affect bacterial susceptibility to TAT-RasGAP317-326 (Fig. 1F). Collectively, these results indicate that the Fe2+, Ca2+ and Mg2+ divalent cations in culture medium both hamper the ability of TAT-RasGAP317-326 to kill bacteria.
Since divalent cations are known to influence outer membrane characteristics, such as lipopolysaccharide (LPS) integrity (Hancock, 1997) we questioned whether TAT-RasGAP317-326 binding and internalization were altered by these cations. Fe2+, and to a lower extent Ca2+ and Mg2+, decreased the levels of FITC-labelled TAT-RasGAP317-326 bound to the surface of E. coli as well as the amount of internalized peptide (Fig. 1G). However, we found that peptide binding and internalization were not affected by ammonium sulfate – a finding that is consistent with our data that ammonium sulfate does not impact peptide MIC (Fig. 1F). Altogether, this data suggests that divalent cations decrease bacterial sensitivity to TAT-RasGAP317-326 peptide via a mechanism that restricts peptide binding and entry in bacteria.
TAT-RasGAP317-326 is bactericidal against E. coli and P. aeruginosa
To determine the relationship between peptide exposure and the number of viable (culturable) bacteria, we performed colony formation unit (CFU) assays. For E. coli grown in LB, 10 μM of TAT-RasGAP317-326 induced an initial 2- to 5- fold decrease in the number of surviving bacteria but there was no further decrease upon longer incubation times (Fig. 2A). A more pronounced decrease in bacterial viability was observed at peptide concentrations ≥ 15 μM, indicating that the peptide is bactericidal at these concentrations (Fig. 2A). Confocal microscopy studies showed that E. coli accumulated TAT-RasGAP317-326 intracellularly when exposed to a concentration leading to bacterial killing (Fig. 2E). Furthermore, peptide exposure at this concentration led to changes in bacterial morphology as seen by electron microscopy (Fig. 2F). For P. aeruginosa grown in BM2 Mglow medium, 0.5-2 μM TAT-RasGAP317-326 had a small impact on bacterial growth relative to the no peptide control, while 5-10 μM strongly reduced bacterial numbers (Fig. 2B). In order to analyze the kinetics of peptide activity at early time points, we performed survival curves using 20 μM of TAT-RasGAP317-326 peptide for E. coli and 10 μM for P. aeruginosa. These concentrations correspond to 2.5 times the MIC of TAT-RasGAP317-326 (Table 1) and were shown to kill a majority of bacteria (Fig. 2A-B). We monitored bacterial killing for the first two hours of peptide exposure and compared bacterial killing by TAT-RasGAP317-326 and polymyxin B, the latter also added at 2.5 times its MIC (2.5 μg/ml for E. coli and 10 μg/ml for P. aeruginosa). Interestingly, TAT-RasGAP317-326 displayed slow time-kill kinetics in comparison to polymyxin B against E. coli (Fig. 2C), suggesting that these two peptides have different killing mechanisms in this bacterial species. In P. aeruginosa however, the killing kinetics were similar between TAT-RasGAP317-326 and polymyxin B (Fig. 2D).
TAT-RasGAP317-326 alters the transcriptional landscape of E. coli
RNA sequencing analysis was performed to evaluate the impact of TAT-RasGAP317-326 on E. coli transcriptome. For this, we used 10 μM of the peptide, a concentration that prevents E. coli proliferation but does not lead to a dramatic drop in bacterial numbers (Fig. 2A). Among the 4419 transcripts predicted from the E. coli MG1655 genome, 95.6% (n = 4223) were detected in at least one condition (Dataset 1). Figure 3A presents the fold change in gene expression between bacteria incubated with and without TAT-RasGAP317-326 as well as the average level of expression for each gene. We excluded from our analysis genes whose expression was below the threshold set at 16 reads per kilobase of transcripts per million reads (RPKM). Overall, TAT-RasGAP317-326 treatment affected the expression of 962 genes (fold change > 4): 11.0% of total detected genes were upregulated (red dots in Fig. 3A) while 11.8% were downregulated (blue dots in Fig. 3A). Detailed lists of upregulated and downregulated genes can be found in Supplementary Tables 1 and 2, respectively.
We assessed and validated twelve genes from the RNA-Seq data by qRT-PCR on RNA extracted under the same conditions as for the RNA-Seq analyses. Five of these (lpxL, fabF, marA, entB, and fepA, depicted in red in Fig. 3B) were reported by RNA-Seq as upregulated, five as downregulated (bssR, frdA, ompF, nuoE, and sdhC, depicted in blue in Fig. 3B) and two as unchanged according to the RNA-Seq analysis (tolC and ompR). One of the unchanged genes, ompR, was used as the housekeeping reference gene for normalization. We obtained good correlation between the fold changes obtained with RNA-Seq and with qRT-PCR, confirming the validity of the RNA-Seq data (Fig. 3B).
Using the gene expression profiles we obtained from RNA sequencing, we investigated which biological pathways were associated with the E. coli response upon exposure to TAT-RasGAP317-326. To accomplish this in a systematic manner, we performed Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses (Kanehisa and Goto, 2000, Kanehisa et al., 2019, Kanehisa, 2019, Ashburner et al., 2000, The Gene Ontology, 2019) on the subset of differentially expressed genes. The analysis of KEGG pathways revealed that several metabolic and information-processing pathways were enriched among differentially expressed genes (Fig. 3, panels C and D). For example, seven of the eight genes responsible for enterobactin synthesis in E. coli (included in “biosynthesis of siderophore group nonribosomal peptides” KEGG pathway) were upregulated upon peptide treatment. Other metabolic pathways such as carbon metabolism (citrate cycle, pyruvate metabolism) and oxidative phosphorylation were downregulated. Similarly, GO term analysis revealed that upregulated and downregulated genes in response to TAT-RasGAP317-326 were enriched in biological processes related to general bacterial metabolism and stress response (Supplementary Fig. 4). From our data, we could not distinguish peptide-specific gene expression changes from gene expression changes mediating general bacterial adaptation to stress. To address this question, we decided to perform a screening of a comprehensive E. coli deletion mutant library to determine which genes are directly involved in the bacterial response to TAT-RasGAP317-326.
Screening of the Keio E. coli deletion mutant library uncovers genes that affect bacterial responses to TAT-RasGAP317-326
The Keio collection of E. coli deletion mutants consists of single gene deletion clones for each non-essential gene in E. coli (Baba et al., 2006). To perform the screening of the collection, we exposed each Keio strain to 5 μM TAT-RasGAP317-326, a non-bactericidal concentration of peptide (Fig. 2A) and monitored bacterial growth by OD590 measurement at specific time points (detailed results of growth measurements for all individual mutants are available in Dataset 2). For each strain, we determined the relative growth of the deletion strain compared to the wild-type strain when incubated with TAT-RasGAP317-326 for 6 hours and 24 hours (Fig. 4, panels A and B). We identified 27 strains showing decreased sensitivity to the peptide, thus having a normalized growth at 6 hours higher than the average of 270 replicates of the parental strain + 2 times the standard deviation (Fig. 4A and Supplementary Table 3). Furthermore, we identified 356 hypersensitive strains (having deletions in 279 different genes) that showed a normalized growth at 24 hours lower than average of the parental strain – 3 times the standard deviation (Fig. 4B). While the wild-type strains grew more slowly in the presence than in the absence of TAT-RasGAP317-326, strains showing decreased sensitivity to the peptide grew similarly in both conditions and hypersensitive strains showed no detectable growth in the presence of the peptide (Fig. 4C). It has to be mentioned that Keio collection is composed of two independent deletion mutants for each gene. We could not observe decreased sensitivity, as defined by our criteria, in both strains having the same gene deleted (Supplementary Table 3). However, some decreased sensitivity, approaching the threshold, could be observed in the second strain for a few genes such as crr and rfaY, for example. The crr gene product is involved in glucose uptake and phosphorylation, and in carbon metabolism regulation (Deutscher et al., 2006). The rfaY gene product is part of the LPS biogenesis pathway (Yethon et al., 1998). Interestingly, inactivation of rfaY by transposon mutagenesis was shown to affect E. coli susceptibility to another AMP, LL-37 (Bociek et al., 2015).
On the other hand, 77 gene deletions caused hypersensitivity for both replicates present in the Keio collection (Supplementary Table 4). KEGG pathway and GO term analyses were thus performed on genes for which both deletion mutants showed hypersensitivity. The results of this analysis indicate that deletion of genes involved in bacterial metabolism and two component systems were associated with TAT-RasGAP317-326 bacterial sensitivity (Fig. 4D-E, Table 2).
Of interest, we found that a subset of less and more sensitive Keio strains were deletion mutants in LPS biogenesis genes. We confirmed differences in sensitivity for ΔrfaY and ΔlpxL deletion mutants by measuring MIC and IC50 of TAT-RasGAP317-326 on these mutants and could confirm that deletion of rfaY caused a decreased sensitivity and deletion of lpxL an increased sensitivity to the peptide (Fig. 5A). This raises the possibility that TAT-RasGAP317-326 directly interacts with bacterial LPS. In such a case, soluble LPS should compete with the peptide for binding to bacterial cells and reduce peptide efficacy, an effect that has been reported for polymyxin B (Domingues et al., 2012). Figure 5B shows indeed that soluble LPS greatly diminishes the efficacy of polymyxin B but has no impact on the sensitivity of E. coli towards TAT-RasGAP317-326. The potential role played by genes involved in LPS synthesis in TAT-RasGAP317-326 sensitivity remains therefore to be uncovered.
We further investigated whether LPS integrity was required for survival of E. coli in the presence of TAT-RasGAP317-326. For this purpose, we used EDTA to destabilize the LPS structure (Hancock, 1984) and measured how this impacted the MIC of TAT-RasGAP317-326 and polymyxin B on two E. coli strains lacking (the MG1655 strain) or not (the ATCC 25922 strain) O-antigen moieties (Eder et al., 2009). EDTA, at concentrations that do not affect bacterial proliferation (Supplementary Fig. 5), sensitized both strains to TAT-RasGAP317-326 (Figure 6B-C), suggesting that compromised LPS integrity favors the antimicrobial activity of the peptide. Polymyxin B sensitivity was less affected by EDTA, indicating again that polymyxin B and TAT-RasGAP317-326 use different mechanisms to inhibit bacterial growth or survival.
Transposon screening in P. aeruginosa
Since TAT-RasGAP317-326 is active against both E. coli and P. aeruginosa, we investigated whether some of the pathways that play a role in peptide resistance are shared between the two bacterial species. We thus exposed a P. aeruginosa transposon mutant library (Vitale et al., 2020) to 0.5 μM TAT-RasGAP317-326 for 12 generations and performed deep sequencing. This allowed us to compare level of transposons in different genes between a bacterial population treated with the peptide and another that was not. Prevalence of strains having a disruption of a gene required for growth in presence of the peptide would decrease compared to strains having integrated the transposon in an unrelated region (detailed results of this deep sequencing are presented as Dataset 3). We thus defined lower prevalence of transposon insertion as a read-out of hypersensitivity to TAT-RasGAP317-326. By this way, we identified 75 genes, for which prevalence of disruption via transposon insertion decreased in presence of the peptide (Supplementary Table 5). Interestingly, 26 of these (35%) are associated with hypersensitivity to other antimicrobial peptides (Vitale et al., 2020). Some of these genes code for LPS modifying enzymes such as ArnA, ArnB and ArnT, and for two-component regulators such as ParS and ParR that are involved in the regulation of LPS modifications (Fernandez et al., 2010). Among the genes, for which prevalence of transposon insertion was decreased in presence of TAT-RasGAP317-326 but not with other AMPs, we identified algJ, algK and algX, genes of the biosynthesis pathway of the extracellular polysaccharide alginate. We also observed that mutants in genes coding for the RND efflux transporter MdtABC and CusC, a component of the trans-periplasmic Cu2+ transporter CusCFBA, are potentially associated with hypersensitivity to TAT-RasGAP317-326. Other pathways that seem to be important for TAT-RasGAP317-326 resistance are related to carbon metabolism, redox reactions and translation regulation (Supplementary Table 5).
We next compared the lists of potential hypersensitive strains found in screenings in E. coli and in P. aeruginosa. We identified six gene orthologues, whose disruption is associated with hypersensitivity to the TAT-RasGAP317-326 peptide in both E. coli and P. aeruginosa (Table 3). Among them, four are coding for two-component system proteins: parR and parS (rtsA and rstB in E. coli), phoP, and pmrB (qseC in E. coli). These four mutants were associated with hypersensitivity to polymyxin B in P. aeruginosa (Vitale et al., 2020), indicating that these regulatory pathways may be required for a general adaptation to AMPs. This is of interest, since RstAB system is regulated by PhoQP system in E. coli (Ogasawara et al., 2007) and PhoQP system was shown to be involved in resistance to AMPs (Yadavalli et al., 2016). Two other genes conserved between P. aeruginosa and E. coli are associated specifically with hypersensitivity to TAT-RasGAP317-326. One is a transcriptional regulator and the other is involved in LPS biosynthesis, further highlighting a potential role for cell surface composition in the sensitivity of bacteria towards TAT-RasGAP317-326 (Table 3).
Effect of combining TAT-RasGAP317-326 with other AMPs
To determine whether TAT-RasGAP317-326 activity is affected by other AMPs, we performed growth tests of E. coli in presence of different combinations of TAT-RasGAP317-326, melittin, LL-37 and polymyxin B. Concentrations were chosen so that a clear difference in growth was observed when “half” concentrations were used (Supplementary Fig. 6A) as compared with “full” concentrations, that correspond to the double of “half” concentrations (Supplementary Fig. 6B). The effect of combining pairs of AMPs using “half” concentrations is shown in Supplementary Fig. 6C as percentage of growth compared to an untreated control. We did not observe an increase in the effect of TAT-RasGAP317-326 when combined with the three other AMPs. However, the combination of melittin and polymyxin B (2.7% of growth) and the combination of LL-37 and polymyxin B (0.6% of growth) showed increased activity. Notably, the combination of melittin and polymyxin B caused stronger growth inhibition (>95%) than obtained by either compound at the “full” concentration (~20% and ~35% growth inhibition for melittin and polymyxin B, respectively; Supplementary Fig. 6C). This observation is consistent with previous reports of the synergism between melittin and antibiotics such as doripenem and ceftazidime (Akbari et al., 2019). In contrast, we observed an apparent lower effect of TAT-RasGAP317-326 in presence of melittin (Supplementary Fig. 6C). Since this effect was very weak in these conditions, we combined the “half” concentration of melittin with the “full” concentration of TAT-RasGAP317-326 and could observe a clear inhibition of the antimicrobial activity of this peptide by melittin (Fig. 6A). To better understand the mechanism behind this observation, we assessed peptide binding and entry into bacteria using a fluorescently labelled version of TAT-RasGAP317-326 peptide. We found that, in the presence of melittin, binding of FITC-labelled TAT-RasGAP317-326 to E. coli bacteria was not decreased, but apparently slightly increased when compared to the control condition where bacteria were incubated with FITC-labelled TAT-RasGAP317-326 alone. However, we observed an apparent lower intracellular accumulation of the labelled version of TAT-RasGAP317-326 peptide in presence of melittin (Fig. 6B).
In vitro selection of resistant bacteria to TAT-RasGAP317-326 peptide
AMPs are less susceptible to bacterial resistance evolution than classical antibiotics (Lazar et al., 2018, Spohn et al., 2019, Lazzaro et al., 2020). To measure the propensity of bacteria to develop resistance against TAT-RasGAP317-326, we serially passaged several bacterial strains (E. coli, P. aeruginosa, S. aureus and S. capitis) in the presence of TAT-RasGAP317-326 peptide and recorded the number of passages required to detect the appearance of resistant mutants in each strain. First, we grew the parental bacterial strains overnight in presence of sub-inhibitory concentrations of the peptide. We then diluted this parent culture into two subcultures, one of which was exposed to an increased concentration of the TAT-RasGAP317-326 peptide while the other was kept in the same concentration of peptide as the parent culture. Once bacterial growth was detected in the culture exposed to an elevated concentration of the peptide, the process was repeated, thereby exposing the bacterial culture to sequentially increasing concentrations of peptide for a total of 20 passages. For each passage, we measured the corresponding MIC (Fig. 7A). Using this approach, we obtained strains with highly increased MICs (16-32 fold) for E. coli, S. capitis, but only a faint increase (2-4 fold) for P. aeruginosa (Fig. 7A, Table 4 and Supplementary Tables 6-8). It should be noted that the parental strain of S. aureus has a peptide MIC in the range 64-128 μM and this MIC rapidly increased to 256 μM (Supplementary Table 9). We did not expose bacteria to higher concentrations, as the peptide started to precipitate in these conditions. To test whether the strains recovered at passage 20 for E. coli, P. aeruginosa and S. capitis and passage 12 for S. aureus showed increased resistance to other AMPs as well, we determined the fold change of the MICs for polymyxin B, melittin and LL-37 relative to the corresponding parental strains that did not undergo selection (Table 4). Interestingly, peptide-resistant E. coli (gram-negative) did not show increased MICs to the other AMPs we tested as compared to the parental strain. In contrast, P. aeruginosa and the Gram-positive S. aureus and S. capitis selected for resistance to TAT-RasGAP317-326 showed increased MIC towards other AMPs (Table 4). Thus, our findings suggest that bacterial species differ in their tendency to develop cross-resistance to TAT-RasGAP317-326 peptide and other AMPs.
Finally, we sought to investigate whether the peptide-resistant bacteria we obtained in our selection process remain targets for alternative treatments such as combination therapy with other antimicrobial agents. In particular, we tested whether EDTA, an agent known to enhance the efficacy of antimicrobials via a mechanism that weakens the outer cell wall of bacteria, could potentiate the effect of TAT-RasGAP317-326 against peptide-resistant E. coli (Leive, 1965). Importantly, the presence of EDTA alone at the concentrations tested was not associated with any significant change in bacterial numbers (Supplementary Figure 5). However, EDTA in combination with TAT-RasGAP317-326 (Fig. 7B) potentiated the ability of the peptide against the peptide-resistant E. coli strain. Our findings suggest that peptide-resistance remains treatable in combination therapy with other antimicrobial agents.
Discussion
The activity of antimicrobial peptides can be affected by environmental factors, but we lack knowledge about how extracellular factors impact TAT-RasGAP317-326 activity. Here, we report that addition of divalent cations in LB medium resulted in decreased bacterial sensitivity to TAT-RasGAP317-326 peptide and reduced peptide binding and entry. The mechanism contributing to lower peptide binding might be due to competition between divalent ions in the culture medium and the cationic TAT-RasGAP317-326 peptide for binding to bacterial surface (Fig. 9). Alternatively, divalent cations, which are important for membrane stability, may influence binding and entry of TAT-RasGAP317-326 (Clifton et al., 2015).
RNA sequencing showed that genes involved in carbon metabolism were downregulated upon treatment with TAT-RasGAP317-326 (Fig. 3). Moreover, deletion or transposon mutants of genes involved in carbon metabolism and ATP production were more sensitive towards TAT-RasGAP317-326 (Fig. 4 and Table 3), indicating that energy production pathways may be important for resistance towards this peptide. Adaptation to environmental stimuli might also be of importance for survival to TAT-RasGAP317-326, since mutants lacking genes coding for some two-component systems show increased sensitivity towards TAT-RasGAP317-326 (Table 2). Several of these two-component systems are known to play role in resistance to antibiotics or AMPs, such as PhoPQ, whose importance in response to AMPs is well described (Bader et al., 2005, Yadavalli et al., 2016). This further highlights the importance of two-component systems for adaptability and survival of bacteria in harsh conditions.
Another pathway that may be involved in sensitivity to TAT-RasGAP317-326 is the LPS biosynthesis pathway. We found that some mutations affecting this pathway cause either moderate resistance or hypersensitivity to the peptide (Fig. 5A). We could further confirm the importance of LPS integrity for survival to TAT-RasGAP317-326 using EDTA that destabilizes LPS. This is consistent with the protective effect of divalent cations (Fig. 1), which can bind and stabilize LPS (Pelletier et al., 1994). Importance of bacterial surface composition in sensitivity towards TAT-RasGAP317-326 is further highlighted by the fact that P. aeruginosa transposon mutants affecting the alginate biosynthesis pathway are more sensitive to TAT-RasGAP317-326 than the control strain (Supplementary Table 5). Alginate is an anionic extracellular polysaccharide that is involved in virulence, antimicrobial resistance and biofilm formation in P. aeruginosa (Franklin et al., 2011).
Interestingly, screening of the Keio deletion collection did not allow to unearth mutants showing complete resistance towards TAT-RasGAP317-326. This indicates that resistance may not be obtained by the loss of function of one gene. Resistance towards TAT-RasGAP317-326 that we obtained by selection (Fig. 7A) may thus have acquired point mutations that modulate activity through activation of some pathways or modifications of essential components. This needs now further investigations in order to describe mechanisms of resistance towards TAT-RasGAP317-326 in particular and AMPs in general.
On the other hand, Keio collection screening highlighted pathways that are apparently required for E. coli to respond to TAT-RasGAP317-326. Whether these pathways are specifically required for response to TAT-RasGAP317-326 or play a role in a general response to AMPs needs further investigation. Interestingly, we observed, using a P. aeruginosa transposon mutants library, that 21% (16 out of 75) of the genes which mutation was association with hypersensitivity to TAT-RasGAP317-326 were associated with hypersensitivity towards other AMPs (Supplementary Table 5)(Vitale et al., 2020).
Combinatorial therapies are gaining interest in the treatment of multi-resistant bacteria (Leon-Buitimea et al., 2020). We thus investigated whether combination with other AMPs might influence the activity of TAT-RasGAP317-326. In general, activity of TAT-RasGAP317-326 was not influenced by other AMPs. However, melittin had an inhibitory effect on TAT-RasGAP317-326 activity, affecting its entry in bacteria (Fig. 6). This rather peculiar effect might be explained by the hypothesized mode of action of melittin (i.e. carpet model), in which melittin first interacts with the bacterial surface, before reaching a concentration threshold that leads to the disruption of the bacterial membrane (Lee et al., 2013). Sub-inhibitory concentrations of melittin might thus block binding of TAT-RasGAP317-326 to the bacterial membrane.
Finally, we investigated the potency of bacteria to develop resistance towards TAT-RasGAP317-326. Resistance could be obtained upon passages in sub-inhibitory concentrations of the peptide (Fig. 7A), but bacterial strains differed with respect to the rate of resistance emergence. Interestingly, peptide-resistant E. coli remains treatable by peptide in combination with EDTA, a chemical agent that compromises the integrity of the bacterial outer membrane. Future work should examine the mechanism of E. coli resistance to peptide and will help elucidate how EDTA, which targets the bacterial envelope, helps potentiate peptide activity in resistant backgrounds. Overall, our data highlight the potential benefit of combination therapies, which might not only prevent the development of such resistance, but also potentiate treatment of resistant strains, as shown here by EDTA in combination with TAT-RasGAP317-326.
The schemes presented in Figure 8 highlight the factors that may influence TAT-RasGAP317-326 activity and present hypotheses about underlying mechanisms. The positively charged TAT-RasGAP317-326 peptide interacts with the negative surface charges of the bacterial membrane, allowing its binding and entry in the bacterial cell (Fig. 8A). Presence of divalent cations in the culture medium compete with TAT-RasGAP317-326 peptide for binding to the negative charges on LPS, lowering the activity of the peptide. Similarly, modifications of LPS structure can also lower interaction between TAT-RasGAP317-326 and bacterial surface. We hypothesize this lower activity to be due to a decrease of the net charge of bacterial surface, causing a lower affinity of the peptide to bacteria (Fig. 8B). In contrast, destabilization of LPS by EDTA or by deletion of genes involved in biosynthesis of LPS precursors increases the bactericidal activity of TAT-RasGAP317-326. This is possibly due to a defect of the integrity of the bacterial envelope, decreasing bacterial defenses towards TAT-RasGAP317-326. (Fig. 8C).
In summary, the results presented in this article bring a better understanding of the factors that influence the antimicrobial activity of TAT-RasGAP317-326. We describe the importance of bacterial envelope integrity on the sensitivity towards TAT-RasGAP317-326. Factors such as divalent salts, EDTA and LPS structure influence the concentration of peptide needed to inhibit bacterial growth. Furthermore, we report the effect of TAT-RasGAP317-326 on the transcriptional landscape of E. coli and highlight the importance of a broad range of two-component systems in the adaptation of bacteria towards this AMP. We finally investigated the effect of other AMPs on the activity of TAT-RasGAP317-326 and could select TAT-RasGAP317-326-resistant bacteria. Our observation that sensitivity could be increased and resistance could be reversed by addition of EDTA is important in the perspective of a clinical use of this peptide to improve its efficiency and to prevent rapid emergence of resistance.
Limitations of the study
Results presented in this study originate from in vitro studies. They might thus only be partially representative of which interactions would happen in an in vivo model of infection. Indeed, several factors such as presence of endogenous AMPs, as well as proteins or other components with which TAT-RasGAP317-326 may interact are not present in our system. Moreover, interactions between TAT-RasGAP317-326 and other AMPs need to be investigated in further details using checkerboard assays, in order to determine putative synergisms. Similarly, mechanisms of action of the peptide and mechanisms of resistance towards the peptide that were selected need to be further investigated in the future, in order to describe how TAT-RasGAP317-326 interacts with bacteria at the molecular level.
Author contributions
MG, TH, NJ, AV, SH and SC performed experiments. MG, TH, LE, CW and NJ were involved in the planning of the project and discussed the results. MG, TH, NJ and TP analysed the results. MG, CW and NJ wrote the manuscript. All the authors proofread the manuscript.
Declaration of interest
The authors declare no competing interests.
Material and methods
Strains, growth conditions and chemicals
E. coli strains K-12 MG1655, ATCC 25922 and BW25113 were grown in LB or Basal Medium 2 (BM2; 62 mM potassium phosphate buffer [pH 7.0], 7 mM (NH4)2SO4, 10 μM FeSO4, 0.4% (wt/v) glucose and 0.5% tryptone) with high (2 mM) or low (20 μM) concentration of magnesium (MgSO4) (Fernandez et al., 2012). Pseudomonas aeruginosa strain PA14 was grown either in LB or BM2 medium. Staphylococcus capitis (Heulot et al., 2017) and S. aureus (ATCC 29213) strains were grown in tryptic soy broth (TSB) (Missiakas and Schneewind, 2013). All strains were stored at −80°C, in their respective medium, supplemented with ~25% glycerol. When required, antibiotics were added at final concentrations of 50 μg/mL (kanamycin), 20 μg/mL (gentamycin), or 100 μg/mL (carbenicillin). The retro-inverse TAT-RasGAP317-326 peptide (amino acid sequence DTRLNTVWMWGGRRRQRRKKRG) and the N-terminal FITC-labelled version of this peptide were synthesized by SBS Genetech (Beijing, China) and stored at −20°C. Chemicals were purchased from Sigma-Aldrich (St-Louis, MO, USA), unless otherwise specified.
MIC measurements
The minimum inhibitory concentration (MIC) of peptide was defined as the lowest concentration of peptide that resulted in no visible growth. Overnight cultures were diluted to OD600 = 0.1 and grown with shaking at 37°C for 1 hour. MICs were measured by diluting these cultures (1:20 for LB and TSB cultures and 1:8 for BM2 cultures) and then adding these dilutions to 2-fold sequential dilutions of the peptides in 96-well plates. Volume of media (with peptide) per well was 100 μl and 10 μl of diluted cultures were added to each well. Cell growth was monitored via OD590 measurement after overnight growth with shaking at 37°C. OD590 readings were measured by FLUOstar Omega microplate reader (BMG Labtech, Ortenberg, Germany). Peptide-free growth control wells and bacteria-free contamination control wells were included. First concentration at which no bacterial growth could be detected was defined as the MIC.
For MIC measurements in presence of E. coli LPS or EDTA, the indicated concentrations of these substances were dissolved in LB and distributed in 96-well plates prior to addition of the peptides. OD590 measurements of control wells without peptides were used to calculate the percentage of growth in presence of EDTA.
Growth curves
Overnight cultures were diluted to OD600 = 0.1 and grown with shaking at 37°C for 1 hour, before addition of peptide. Cell growth was monitored via OD600 measurement by Novaspec II Visible spectrophotometer (Pharmacia LKB Biotechnology, Cambridge, England) at 2, 4 and 6 hours.
Combinations of antimicrobial peptides were tested using the above methods and combining “half” concentrations of the different peptides (2 μM TAT-RasGAP317-326, 64 μg/ml melittin, 32 μg/ml LL-37 or 1 μg/ml polymyxin B) to produce supplementary Figure 6C.
CFU measurements
Overnight cultures were diluted to OD600 = 0.1 and grown with shaking at 37°C for 1 hour, before addition of the peptide. Each time point was taken by removing 10 μl and performing 10-fold serial dilutions. Dilutions of each condition were then plated in the absence of peptide and grown at 37°C overnight. CFU were measured by counting the number of colonies on the plates after overnight incubation.
Confocal microscopy
Overnight cultures of E. coli MG1655 were diluted to OD600 = 0.1, grown for 1 hour, incubated for 1 hour with 10 μM FITC-labelled TAT-RasGAP317-326, stained with 5 μg/ml FM4-64 and fixed with 4% paraformaldehyde solution. Incubation with DAPI was subsequently performed and pictures were acquired on a LSM710 confocal microscope (Zeiss, Oberkochen, Germany). Images were analyzed with ImageJ software (Schneider et al., 2012).
Electron microscopy
Bacteria were fixed with 2.5% glutaraldehyde solution (EMS, Hatfield, PA) in Phosphate Buffer (PB 0.1 M pH 7.4) for 1 hour at room temperature. Then, bacterial samples were incubated in a freshly prepared mix of 1% osmium tetroxide (EMS) and 1.5% potassium ferrocyanide in phosphate buffer for 1 hour at room temperature. The samples were then washed three times in distilled water and spun down in 2% low melting agarose, solidified on ice, cut into 1 mm3 cubes and dehydrated in acetone solution at graded concentrations (30% for 40 minutes; 50% for 40 minutes; 70% for 40 minutes and 100% for 3 times 1 hour). This was followed by infiltration in Epon at graded concentrations (Epon 1/3 acetone for 2 hours; Epon 3/1 acetone for 2 hours, Epon 1/1 for 4 hours and Epon 1/1 for 12 hours) and finally polymerization for 48 hours at 60°C in a laboratory oven. Ultrathin sections of 50 nm were cut on a Leica Ultramicrotome (Leica Mikrosysteme GmbH, Vienna, Austria) and placed on a copper slot grid 2×1mm (EMS) coated with a polystyrene film. The bacterial sections were stained in 4% uranyl acetate for 10 minutes, rinsed several times with water, then incubated in Reynolds lead citrate and finally rinsed several times with water before imaging.
Micrographs (10×10 tiles) with a pixel size of 1.209 nm over an area of 40×40 μm were taken with a transmission electron microscope Philips CM100 (Thermo Fisher Scientific, Waltham, MA) at an acceleration voltage of 80kV with a TVIPS TemCam-F416 digital camera (TVIPS GmbH, Gauting, Germany). Large montage alignments were performed using Blendmont command-line program from the IMOD software (Kremer et al., 1996) and treated with ImageJ software.
Flow cytometry
Overnight cultures of E. coli MG1655 were diluted 1:100 and grown to mid exponential phase (OD600 = 0.4-0.6) with shaking at 37°C. Each culture was then diluted to OD600 = 0.1, grown with shaking at 37°C for 1 hour and then treated with 10 μM FITC-labelled peptide for 1 hour. Following peptide treatment, bacterial cells were washed in PBS and diluted 1:5 before acquisition on a CytoFLEX benchtop flow cytometer (Beckman Coulter). For each sample, 10,000 events were collected and analyzed. Extracellular fluorescence was quenched with 0.2% Trypan Blue (TB). TB is an efficient quencher of extracellular fluorescence (Sahlin et al., 1983, Loike and Silverstein, 1983, Jevprasesphant et al., 2004, Wan et al., 1993) and allows quantification of fluorescent signal from intracellular peptide (not subject to quenching by TB). P values were calculated using ratio paired t-test.
RNA-Seq
Overnight cultures of E. coli MG1655 were diluted to OD600 = 0.1 and grown with shaking at 37°C for one hour to mid exponential phase (OD600 = 0.4-0.6). Cultures were then treated with TAT-RasGAP317-326 (10 μM) or left untreated (negative control), and grown with shaking at 37°C for an additional hour. For RNA extraction, protocol 1 in the RNAprotect Bacteria Reagent Handbook (Enzymatic lysis of bacteria) was followed using the RNeasy Plus Mini Kit (Qiagen) using TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0) containing 1 mg/ml lysozyme (AppliChem, Chicago, IL). In the last step, RNA was eluted in 30 μl RNase-free water. Next, any contaminating DNA was removed using the DNA-free™ DNA Removal Kit (Invitrogen, Carlsbad, CA). 10x DNase buffer was added to the 30 μl eluted RNA with 2 μl rDNase I. This mix was incubated for 30 minutes at 37°C followed by rDNase I inactivation with 7 μl DNase Inactivation Reagent for 2 minutes with shaking (700 rpm) at room temperature. Samples were then centrifuged for 90 seconds at 10,000 x g, supernatant was transferred to a new tube, and stored at −80°C. Integrity of the samples was verified using the Standard Sensitivity RNA Analysis kit (Advanced Analytical, Ankeny, IA) with the Fragment Analyser Automated CE System (Labgene Scientific, Châtel-Saint-Denis, Switzerland). Samples that met RNA-Seq requirements were further processed and sent for sequencing. Preparation of the libraries and Illumina HiSeq platform (1×50 bp) sequencing were performed by Fasteris (Plan-les-Ouates, Switzerland). Raw reads were trimmed with trimmomatic version 0.36 (Bolger et al., 2014) (parameters: ILLUMINACLIP: NexteraPE-PE.fa:3:25:6, LEADING: 28, TRAILING: 28 MINLEN: 30). Trimmed reads were mapped to the genome of E. coli K-12 MG1655 (accession: NC_000913.3) with bwa mem version 0.7.17 (https://arxiv.org/abs/1303.3997) using default parameters. Htseq version 0.11.2 (Anders et al., 2015) was used to count reads aligned to each gene (parameters: --stranded=no -t gene). Normalized expression values were calculated as Reads Per Kilobase of transcript per Million mapped reads (RPKM) with edgeR (Robinson et al., 2010).
Keio collection screening
Deletion mutants from the Keio collection (Baba et al., 2006, Yamamoto et al., 2009) were used, along with the corresponding wild-type, which was added as a control on each test plate. Overnight cultures were diluted 1:100 in LB medium. Bacteria were incubated at 37°C for 1 hour before adding TAT-RasGAP317-326 (5 μM final concentration). Plates were incubated statically at 37°C and OD590 was measured at 0 hour, 1.5 hour, 3 hours, 6 hours and 24 hours with FLUOstar Omega plate reader. Measurements were combined and analysed with R (version 3.6.1, (Team, 2019)). Data analysis and visualisation were performed with the dplyr (version 0.8.5) and ggplot2 (version 3.3.0) packages from the tidyverse (version 1.3.0) environment. Since starting OD590 (OD in equations) varied between strains and conditions, the OD590 starting values in each well was subtracted from corresponding measurements made at time t in the presence (P) or absence (noP) of TAT-RasGAP317-326. For each strain, NGmt(P), the normalized growth value for a mutant strain at time t in the presence of the peptide was calculated with the following formula:
Normalized growths of wild-type strain (mean WT), as presented on Fig. 4a and b were calculated by averaging normalized growths of all the wild-type controls performed (N=270). To normalize the growth of a mutant (m) to the growth of control (c) bacteria (wild-type) on the same plate, the NGmt(noP) factor was calculated with the following formula:
Gene ontology (GO) annotation (The Gene Ontology, 2019) was obtained from GO database (2020-09-01, “http://current.geneontology.org/annotations”) and assigned to the list of gene deletion inducing hypersensitivity with the GO.db package (version 3.10.0 (Carslon, 2019)). GO IDs were assigned to each gene and the corresponding GO names were obtained with the “Term” function. Additionally, the same set of genes was subjected to KEGG pathways analysis (Kanehisa and Goto, 2000) with the KEGGREST package (version 1.26.1). Briefly, the KEGG orthology (KO) and KEGG pathway annotation were obtained from the KEGG database (Kanehisa, 2019) for E. coli K-12 MG1655 (eco). The code is available on Github (https://github.com/njacquie/TAT-RasGAP_project).
Pseudomonas aeruginosa PA14 transposon library screening
The library of transposon (Tn) mutants in P. aeruginosa PA14 (Vitale et al., 2020) was grown in BM2 supplemented with 20 μM MgSO4 (Fernandez et al., 2012) and 0.2% L-rhamnose monohydrate (Sigma-Aldrich) in the absence or presence of 0.5 μM TAT-RasGAP317-326. Following growth for 12 generations, genomic DNA (gDNA) was extracted with the GenElute Bacterial Genomic DNA Kit (Sigma-Aldrich). The transposon sequencing (Tn-seq) circle method (Gallagher et al., 2011, Gallagher et al., 2013) was employed to sequence the transposon junctions. Briefly, the gDNA was sheared to an average size of 300 bp fragments with a focused-ultrasonicator. The DNA fragments were repaired and ligated to adapters with the NEBNext Ultra II DNA Library Prep Kit for Illumina (New England Biolabs). Following restriction of the Tn with BamHI (New England Biolabs), the fragments were circularized by ligation and exonuclease treatment was applied to remove undesired non-circularized DNAs (Gallagher et al., 2011). The Tn junctions were PCR amplified and amplicons were sequenced with the MiSeq Reagent Kit v2, 300-cycles (Illumina).
Following sequencing, the adapter sequences of the reads (.fastq) were trimmed with the command line “cutadapt -a adapter -q quality -o output.fastq.gz input.fastq.gz” (Martin, 2011). The software Tn-Seq Explorer (Solaimanpour et al., 2015) mapped the trimmed and paired reads onto the P. aeruginosa UCBPP-PA14 genome (Winsor et al., 2016), and determined the unique insertion density (UID, i.e. the number of unique Tn insertions divided per the length of the gene). The normalized UID between the treated and non-treated samples were compared and this ratio (log2-fold change, FC) was used to identify resistant determinants (log2-FC < − 1.0 and normalized UID > 0.0045).
Selection of resistant mutants
Bacteria were grown in the corresponding medium, diluted 1:100 and cultured overnight with 0.5x MIC of TAT-RasGAP317-326. The subculture was diluted 1:100 and incubated with 0.5x or 1x MIC overnight. Cells that successfully grew were diluted 1:100 in medium containing the same concentration or twice the concentration of peptide. Each dilution in fresh medium containing peptide is considered one passage. This process was repeated for up to 20 passages.
Acknowledgments
We would like to thank Sébastien Aeby and Yasmina Merzouk for technical support, Valentin Scherz for support in bioinformatics analyses and Prof. Gilbert Greub for sharing equipment and laboratories. This study was supported by an interdisciplinary grant of the Faculty of Biology and Medicine of the University of Lausanne.
Footnotes
↵# These authors share senior authorship
This new version of the manuscript contains new versions of figures which were improved with new experiments.