Simplified Autoinducing Peptide Mimetics with Single-Nanomolar Activity Against the Staphylococcus aureus AgrC Quorum Sensing Receptor

Abstract

Staphylococcus aureus is a leading cause of hospital-acquired infections worldwide, and cases of community-acquired infections are becoming more prevalent. The production of numerous virulence factors in S. aureus is under the control of the accessory gene regulator (agr) quorum sensing (QS) system. S. aureus upregulates agr at high cell density by secreting a peptide pheromone, or autoinducing peptide (AIP), which is detected by its cognate transmembrane receptor, AgrC. The extracellular AIP binding site of AgrC represents an attractive target for inhibition of the agr system, and thereby, QS-controlled virulence in S. aureus. Nonnative peptides and more recently, peptidomimetics, have been reported to inhibit the AgrC receptor and represent useful chemical tools to study the role of QS in S. aureus infections. We seek to expand beyond peptide-like scaffolds to generate AgrC modulators with enhanced stability, solubility, and synthetic accessibility relative to these compounds, while maintaining their high potencies. Toward this goal, we report herein a study of the structure-activity relationships responsible for the activity of a recently reported simplified AIP mimetic and AgrC antagonist, n7OFF, and the discovery of a new AIP mimetic, Bnc3, which has low- to sub-nanomolar inhibitory activity in all four S. aureus agr specificity groups. NMR structural studies of Bnc3 revealed hydrophobic and hydrophilic faces that are likely critical for AgrC antagonism, in agreement with prior studies of peptide-derived inhibitors. Bnc3 represents an important transition compound toward the development of small-molecule AgrC antagonists.

Graphical Abstract

Staphylococcus aureus is a common opportunistic pathogen that colonizes approximately 30% of the world’s population.¹ Most colonized individuals coexist with S. aureus as a commensal organism that inhabits the skin and nose. However, virulent S. aureus strains have been found in hospital-acquired infections for decades and are being isolated more frequently from community-acquired infections.² Combined with the growing incidence of multidrug-resistance in these strains, S. aureus presents a significant challenge for our healthcare system. New alternatives are desperately needed to prevent and treat S. aureus infections.^1–3

S. aureus produces numerous virulence factors that are responsible for many aspects of acute infections. A large proportion of these virulence factors, such as hemolysins and phenol-soluble modulins, are controlled by a cell-cell communication pathway known as quorum sensing (QS).^3–5 QS is a means by which bacteria assess their local population densities and initiate group-beneficial behaviors at high cell number. Bacteria use simple chemical signals, or autoinducers, for QS that are produced at a low, but constant basal level. Once the bacterial population is sufficiently large in a given environment, the local autoinducer concentration reaches a threshold level at which it can productively bind to its cognate receptor; this signal:receptor binding event effectively signals to the bacteria that their population has reached a quorum.⁶ The bacterial “group” will then initiate a diversity of behaviors, ranging from bioluminescence by marine bacteria, to antibiotic production by soil dwelling bacteria, to virulence factor production by pathogens, as is the case for S. aureus.^4–7

S. aureus uses the accessory gene regulator (agr) system for QS, which is considered an autocatalytic sensory transduction system.⁸ This system is found in many staphylococcal species, but is best understood in S. aureus and illustrated in Figure 1A.⁹ The agr operon encodes four proteins (AgrA−D), of which AgrC and AgrA are part of a classical two-component regulatory system. AgrD contains three domains: an amphipathic N-terminal domain that localizes the protein to the inner leaflet of the plasma membrane, a “pro-peptide” domain consisting of a linear precursor of the autoinducing peptide (AIP) QS signal, and a C-terminal recognition domain.¹⁰ AgrB is an integral membrane endopeptidase that recognizes the C-terminal domain of AgrD, cleaves this domain, and cyclizes the new C-terminal residue of AgrD to a conserved Cys residue sulfhydryl in the pro-peptide domain.¹¹ The modified AgrD is then transported outside the plasma membrane and the N-terminal domain is cleaved, liberating the mature AIP signal outside of the cell.¹² As the population grows, this newly formed QS signal accumulates in the local environment, and when a sufficient concentration of AIP is reached (and thus a quorum of S. aureus), the AIP binds to its target, AgrC, a transmembrane receptor histidine kinase. Once bound to AIP, AgrC transautophosphorylates and then phosphorylates AgrA, an intracellular transcription factor.^{13, 14} Binding of AgrA to the P2 promoter upregulates the agr operon and thereby amplifies the QS signal, in a typical autoinduction cycle that is a hallmark of QS systems. In turn, the binding of AgrA to the P3 and other promoters upregulates myriad virulence factors that are associated with S. aureus infections.³

Figure 1.

The agr QS system and associated AIP signals. A: Schematic of the agr system. (a) agr encodes production of AgrA−D. (b) AgrD contains the precursor for the AIP QS signal. (c) AgrB processes AgrD and liberates the mature AIP signal. (d) The AIP binds to AgrC. (e) AgrC transautophosphorylates and then phosphorylates AgrA. (f) AgrA drives transcription at the P2 and P3 promoters, which upregulates production of AgrA−D and activates virulence factor production. B: Structures of the native AIP signals used by the four groups of S. aureus, using single letter amino acid abbreviations.

The agr system of S. aureus has diverged evolutionarily into three common (I-III) and one rare (IV) specificity groups, each with a unique AIP signal (Figure 1B) and some sequence variability in the AgrA−D proteins, although the sequences of AgrA and the histidine kinase domain of AgrC are highly conserved.^15–17 Among the four AIPs, the five-amino acid macrocycle and thioester bridge from the C-terminus to a Cys is conserved, and each has at least two hydrophobic amino acids at the C-terminus. Otherwise, the identity of each amino acid and the length of the AIP tail varies between the four groups.

Because agr activity upregulates many important virulence factors in S. aureus, antagonism of this system has attracted interest as a non-bactericidal, or “anti-virulence,” strategy to block infection.^18–20 Indeed, in several recent reports by Williams, Horswill, and others, antagonism of agr has reduced the infectivity of virulent S. aureus strains,^21–26 underscoring the potential utility of this approach for infection control. Inhibition of agr via the AgrC receptor has been a focus of several research laboratories, including our own.^{23, 24, 27–32} A recognized method for antagonizing AgrC is to use AIP analogs, often with a single amino acid substitution or tail truncation, or combinations of these modifications, to competitively bind to AgrC and thereby block the native signal from binding.^31–35 Most of these reported AgrC antagonists are peptides, with sizes ranging from five to nine amino acids. While some are highly potent (sub-nanomolar IC₅₀ values), these peptides have several unfavorable attributes that impede their broad use as chemical probes. They are typically relatively large, only produced in small quantities, and hydrolytically unstable. The generation of non-peptide, or small molecule, AgrC inhibitors that circumvent these properties while maintaining high potencies would be a significant advance. Our laboratory is focused on identifying such scaffolds, and as a step toward this goal, we recently redesigned the structure of tail-truncated AIP-II (t-AIP-II), reported by the Muir lab,³⁶ to develop a simplified, four-amino acid peptide derivative (n7OFF; Figure 2) with comparable antagonistic activity against AgrC.³⁷ While n7OFF represents one of the first peptidomimetics of AIPs with appreciable biological activity,^{38, 39} it is 10−1000-fold less potent than lead peptide antagonists of AgrC. In the current study, we sought to delineate the chemical features of n7OFF that are critical to its activity in order to increase its potency. Additionally, we sought to identify structural motifs that can allow for the transition of this peptidomimetic into a non-peptide-derived AgrC antagonist, such as non-canonical amino acid motifs, which have been shown to have promise in related systems.^{24, 40}

Figure 2.

Structures of t-AIP-II (top) and n7OFF (bottom). The nomenclature for specific motifs in n7OFF is indicated.

Herein, we report a detailed structure-activity relationship (SAR) analysis of the AgrC antagonist n7OFF and the discovery of new peptidomimetics, Bnc3 and Bn-n7OCpa(3ClF), which have low- to sub-nanomolar potencies against AgrC in all four S. aureus agr groups. An amide-linked analog of Bnc3 was also identified as a potent AgrC antagonist, with both improved solubility and stability over the parent thioester. Notably, these potency profiles are now comparable to current lead, all peptide-derived antagonists of AgrC. NMR structural studies of Bnc3 revealed hydrophobic and hydrophilic faces that appear important for AgrC antagonism. This study advances the understanding of the chemical motifs that are critical for AgrC antagonism and will inform the design of small-molecule antagonists of AgrC.

RESULTS AND DISCUSSION

Compound design and synthesis.

To define the SARs for the n7OFF scaffold, we began by examining its (i) two endocyclic Phe residues, (ii) tail region, and (iii) thioester linkage (Figure 3). We made structural changes in two phases, and combined features that engendered enhanced potencies against AgrC as the study progressed, including the use of an aliphatic linker in place of the oxa-linker (i.e., n7O2; as was examined in our previous work).³⁷ To probe the role of sterics and electronics on the Phe aromatic rings, -Me, -OMe, -F, -Cl, and -NO₂ substituents were introduced. The aromatic rings also were replaced by five- and six-member aliphatic rings. To examine the tail region, the acetyl methyl group was replaced with a set of linear aliphatic chains (ethyl, butyl, and pentyl). Two acyclic, branched aliphatic tails (isopropyl and isobutyl) and two aryl tails (benzyl and phenyl) were also evaluated. To study the thioester linkage, we replaced it with an ester and the more hydrolytically stable amide isostere. We also were interested in replacing the thioester linkage with an olefin, and first tested this replacement strategy in the known peptide-derived AgrC antagonists, t-AIP-II and t-AIP-III-D2A (Figure 4).^{31, 36, 38} All compounds were prepared as linear precursors on a solid support followed by solution-phase macrocyclization reactions and purification to homogeneity by HPLC, using our previously reported methods (see Methods).^{37, 41}

Figure 3.

Structural modifications evaluated in n7OFF. Tail modifications indicated in blue at R₁, side chain modifications indicated in green at R₂ and R₃, thioester substitutions indicated in red, and the oxa-linker modification indicated in violet.

Figure 4.

Thioester-to-olefin linkage modifications examined within known AgrC antagonists, t-AIP-II and t-AIP-III-D2A.

Biological testing.

To determine compound activities against AgrC, we utilized reporter plasmids in S. aureus strains (groups-I−IV) that permit agr activity to be measured using fluorescence (via the production of yellow-fluorescent protein (YFP)).^{31, 37, 42} These strains are clinical isolates, contain the native agr system, and produce AIPs at wild-type levels. Therefore, in the absence of exogenous compound, these reporter systems produce YFP once a quorum is reached. Compounds were first screened for agr antagonism (and thus a reduction in YFP production) at 10 μM, and if deemed sufficiently active, screened again at 3.3 μM (see Figures S1−S7). These screens provided initial information about compound efficacy, and compounds with near-complete antagonism of AgrC at both 10 μM and 3.3 μM were further evaluated via dose-response analyses in these same reporters to determine their IC₅₀ values and comparisons of their relative potencies (see Tables S2−S3 for a listing of all values; key data is highlighted below).

Phase-one SAR analyses.

We first examined the effects of alterations to the Phe aromatic rings and the tail of n7OFF on potency (Table 1). Relative to the parent compound, the placement of an electronegative fluorine or chlorine at the 3-position of the C-terminal aromatic ring (Phe4) increased potency in groups-I−IV, suggesting that electronics play at least a partial role in receptor-ligand interactions. The 3-fluoro analog (n7OF(3fF)) was more potent in groups-I and -III, and the 3-chloro analog (n7OF(3ClF)) was more potent in groups-II and -IV. The same effect on potency was not universally observed with other substituents (-Me, -OMe or -NO₂; Tables S2−S3), suggesting that this effect is not strictly due to an increase in steric bulk at the 3-position of Phe4.

Table 1.

IC₅₀ values (in nM) for AgrC antagonism by selected phase-one n7OFF analogs in S. aureus groups-I−IV^a_. 95% confidence intervals (CI) in parentheses.

compound	group-I	group-II	group-III	group-IV
n7OFF	181 (154−213)	583 (468−725)	332 (280−395)	5938 (4221−8532)
n7OF(3fF)	30.6 (22.5–41.8)	445 (217–911)	222 (158–312)	~3500 (ND)b
n7OF(3ClF)	125 (90–173)	167 (118–237)	630 (364–1090)	2240 (1150–4460)
n7OChaF	439 (278–695)	627 (317–1238)	194 (148–255)	1250 (990–1580)
n7OCpaF	141 (107–185)	204 (169–246)	19.7 (14.8–26.3)	498 (414–598)
Bn-n7OFF	15.3 (12.4–19.0)	37.8 (29.8–47.9)	17.3 (14.1–21.2)	564 (435–731)
Et-n7OFF	129 (97.7–170)	197 (152–255)	531 (312–905)	--c
iPr-n7OFF	45.8 (34.0–61.9)	97.1 (65.6–144)	316 (192–522)	--c
Pr-n7OFF	57.9 (30.3–111)	103 (68–155)	203 (121–341)	--c
iBu-n7OFF	56.0 (38.6–81.2)	40.7 (35.3–46.9)	55.3 (37.7–81.0)	--c
Bu-n7OFF	34.7 (28.1–42.8)	72.3 (45.6–115)	78.1 (43.6–140)	--c
pentyl-n7OFF	15.8 (13.5–18.4)	17.0 (14.7–19.6)	15.7 (11.8–21.0)	--c
Ph-n7OFF	157 (123–201)	23.0 (15.6–34.1)	98.2 (75.0–129)	--c
n7OF(4fF)	375 (204–689)	354 (253–497)	2140 (1080–4220)	--c

^aData for parent n7OFF included for comparison. See Methods section for details of strains and assays. Comprehensive assay data for all compounds tested are provided in Tables S2−S3.

^bThe confidence interval was too wide to quantify accurately.

^cMinimal activity at concentrations up to 10 μM; IC₅₀ not calculated.

Replacement of the phenyl group of Phe3 with a cyclohexyl group (n7OChaF; Cha for cyclohexylalanine) caused only a slight loss of potency relative to n7OFF in groups-I and -II and an increase in potency in groups-III and -IV (Table 1). Interestingly, contracting the ring size to a five-member ring (n7OCpaF; Cpa for cyclopentylalanine) resulted in an AgrC antagonist that was more potent than n7OFF in all four groups. This result, in combination with our earlier work with Leu3 analogs of n7OFF,³⁷ indicates that a hydrophobic group of intermediate size with an sp³ γ-carbon (i.e., between an isobutyl and cyclohexylmethyl group) at the Phe3 position is preferred for pan-group AgrC antagonism in the n7OFF scaffold.

Addition of a benzyl group at the tail position (to give Bn-n7OFF) yielded another interesting SAR (Table 1). Relative to n7OFF, Bn-n7OFF exhibited a 10-fold improvement in potency for groups-I−IV. We note that Bn-n7OFF reduced agr activity by ~95% in groups-I−III, but only by ~70% in group-IV; other compounds discussed in this section reduced agr activity by ~95% in all four groups. Nevertheless, this result suggests that simple aryl tails, as opposed to peptide tails, can improve antagonistic activity against AgrC for AIP-type mimetics. Indeed, the addition of this third hydrophobic group, along with the two Phe groups on n7OFF, could generate a hydrophobic “triad” motif that previously has been deemed important to the activity profiles of other potent AgrC antagonists.^{35, 43}

Additional SAR analyses in S. aureus groups-I–III.

A number of n7OFF analogs were identified in the phase-one SAR studies that were potent AgrC antagonists in S. aureus groups-I−III, but were less potent in group-IV (Table 1). As group-I−III strains are far more common than group-IV, compounds that target the former three agr groups are still of importance, and we scrutinized SARs for these additional compounds further. The most noteworthy observation was that as the aliphatic tail was increased in length on n7OFF, including the incorporation of branched chains, more potent AgrC antagonists were generated. For example, pentyl-n7OFF has an IC₅₀ of ~16 nM in each of groups-I−III, which represents an 11- to 36-fold increase in potency relative to the parent n7OFF. Tails longer than five carbons were not evaluated as the solubility in water became problematic. The 4-fluoro derivative n7OF(4fF) was found to display potency against groups-I−III also (Table 1), albeit more modestly than the tail modified compounds. We found that introduction of a nitro group at any position of the endocyclic Phe rings yielded no potent antagonists (i.e., no IC₅₀ < 3 μM), which is perhaps understandable if the aromatic rings bind in a hydrophobic binding site on AgrC. In addition, substitution of a methoxy group on the endocyclic Phe rings was largely not tolerated by AgrC, except for the 4-OMe Phe4 derivative that was a modest antagonist of AgrC in group-II (IC₅₀ = 256 nM). Additional SAR trends were more subtle and are detailed in the SI (Tables S2−S3).

Phase-two SAR analyses.

The most interesting SARs identified in the phase-one analyses were combined to generate a set of second-round n7OFF analogs, and the effects on potency were largely found to be additive. Key compounds are highlighted in Table 2. First, the Cha and 3fF residues were combined to give n7OCha(3fF), which was surprising in that it was less potent than its parents n7OChaF and n7OF(3fF) in group-II, but otherwise was a more potent antagonist in groups-I, -III, and -IV than n7OChaF and more potent than n7OF(3fF) in groups-III−IV. As hydrophobicity appears to be important for AIP-AgrC interactions,^{31, 38} a more hydrophobic analog of n7OCha(3fF) was prepared with a saturated carbon chain in place of the n7O linkage (Figure 3), to give n7Cha(3fF). Except in group-III, this analog was less potent in each agr group, so this linker modification was not considered further. Cpa was next incorporated alongside 3fF to give n7OCpa(3fF), which yielded the most potent compound described so far. Relative to n7OFF, n7OCpa(3fF) was four-fold more potent in group-I, two-fold more potent in group-II, 60-fold more potent in group-III, and 30-fold more potent in group-IV. n7OCpa(3fF) also displays high efficacy, reducing AgrC activity by ~95% in all four groups, marking it as a promising new scaffold for further study.

Table 2.

IC₅₀ values (in nM) for AgrC antagonism by selected phase-two n7OFF analogs in S. aureus groups-I−IV^a_. 95% confidence intervals (CI) in parentheses.

compound	group-I	group-II	group-III	group-IV
n7OFF	181 (154–213)	583 (468–725)	332 (280–395)	5938 (4221–8532)
n7OCha(3fF)	88.3 (70.9–110)	1240 (830–1840)	28.9 (22.5–37.2)	432 (371–503)
n7Cha(3fF)	334 (264–421)	1850 (950–3640)	12.1 (9.8–15.0)	960 (780–1182)
n7OCpa(3fF)	48.0 (41.1–56.2)	207 (110–388)	5.33 (3.84–7.39)	190 (149–242)
pentyl-n7OCpa(3fF)	2.56 (2.17–3.02)	14.1 (11.0–18.0)	0.148 (0.119–0.186)	30.4 (15.7–58.9)
Bn-n7OCpa(3fF) [Bnc3]	2.01 (1.57–2.57)	17.9 (12.9–24.9)	0.0918 (0.0788–0.107)	4.75 (2.98–7.59)
Bn-n7OCpa(3ClF)	4.88 (3.16–7.52)	6.47 (4.88–8.59)	0.387 (0.325–0.461)	3.87 (2.37–6.30)
Bn-n7OCpa(3fF)-amide	81.3 (62.5–106)	362 (137–952)	0.634 (0.509–0.789)	142 (79–255)
Bn-n7OCpa(3fF)-ester	401 (312–516)	870 (621–1220)	2.08 (1.69–2.56)	463 (298–720)

^aData for parent n7OFF included for comparison. See Methods section for details of strains and assays. Comprehensive assay data for all compounds tested are provided in Tables S2−S3.

We next examined the introduction of various tails into n7OCpa(3fF), and found the pentyl-n7OCpa(3fF) and Bn-n7OCpa(3fF) (hereafter named Bnc3) derivatives to display markedly improved potencies (18−53 fold improvements; Table 2). Notably, Bnc3 was an exceptionally potent AgrC antagonist, with IC₅₀ values in the low- to sub-nanomolar range across the four agr groups. This activity profile was comparable to previously reported lead peptide-based AgrC inhibitors, such as AIP-I D5A and AIP-III D4A.^{31, 44} Critically and in support of the goals of the current study, Bnc3 was approximately 90-, 32-, 3320-, and 1250-fold more potent than n7OFF and 50-, 5-, 100-, and 30-fold more potent than t-AIP-II in groups-I−IV (Table S2),³⁷ respectively, underscoring the value of our SAR-driven analog design. Another promising combination compound, Bn-n7OCpa(3ClF), had a chlorine in place of the fluorine of Bnc3, and was a single-digit nanomolar antagonist in each of groups-I, -II, and -IV, and a sub-nanomolar antagonist in group-III.

Examining the thioester in Bnc3 was our next target. To investigate the utility of an olefin replacement for the thioester in the Bnc3 AIP scaffold, we first examined an olefin in the known antagonists, t-AIP-II and t-AIP-III-D2A (Figure 4). We found that the olefin was an unsuitable replacement for the thioester, as these analogs were only potent antagonists in group-III (Tables S2−S3). We thus did not pursue olefin replacements in n7OFF analogs in this study. To determine if an amide linkage could be tolerated, diaminopropionic acid was incorporated in place of the thioester-forming Cys to make Bn-n7OCpa(3fF)-amide (Table 2). This compound was relatively less potent than Bnc3 in each agr group, but still a strong AgrC antagonist with potency similar to or better than t-AIP-II.^{33, 34, 37, 38} Further, as evidenced by its retention time from analytical HPLC (see Table S1) and previous amide versus thioester stability studies,⁴⁵ Bn-n7OCpa(3fF)-amide possesses improved solubility and hydrolytic stability in aqueous media that make it a valuable new probe compound moving forward. Lastly, to further explore isosteres of the thioester, an ester linkage was incorporated to yield Bn-n7OCpa(3fF)-ester. This compound was also significantly less potent than Bnc3 in all four groups. However, in group-IV, the ester analog reduced agr activity by ~90% at the highest concentrations tested, compared to the ~70% reduction obtained by Bnc3 and other Bn-n7OFF analogs.

Summarizing the results of these two phases of n7OFF analog analysis, we found that the incorporation of a Bn tail, Cpa at position 3, 3fF or 3ClF at position 4, and an amide linkage into n7OFF generated new AgrC antagonists that should advance the design of small-molecule ligands. Bnc3, Bn-n7OCpa(3ClF), and Bn-n7OCpa(3fF)-amide represent the lead compounds from this SAR study. We note that, while highly potent, these compounds display lower efficacy in group-IV (~70% inhibition) relative to other analogs in this study. However, as Bnc3, Bn-n7OCpa(3ClF), and Bn-n7OCpa(3fF)-amide still reduce agr activity by ~95% in the other more common agr groups-I−III, this activity profile is not detrimental to their utility as probes or lead scaffolds moving forward. In addition, several analogs were identified that selectively inhibit agr in one group with minimal activity in the other groups (see Tables S2−S3); such selectivity profiles could prove useful to explore the role of these different groups in S. aureus infections.⁴⁶

NMR solution structure of Bnc3.

In our past studies of non-native peptides capable of modulating AgrC, we have relied on NMR methods to characterize their solution-phase structures and develop mechanistic rationales for the modes by which these peptides could interact with AgrC receptors.^{37, 43, 47} As many macrocyclic peptides are found to be conformationally rigidified in aqueous media, and we could connect trends in cell-based reporter activity to specific structural features in our peptides as revealed by NMR, we reasoned that these NMR structures could be suggestive of their biologically active structures. We therefore applied these NMR techniques to Bnc3 in 1:1 CD₃CN:H₂O and characterized this mimetic as in our previous studies to gain insights into the structural features that may engender its high potency.

Using an all-atom alignment, the average RMSD for the NMR solution structure ensemble of Bnc3 was 0.87 Å, and using a heavy-atom alignment, the average RMSD for the ensemble was 0.68 Å (see Figure S8 for complete ensemble). The RMSD value is smaller than that previously observed for n7OFF,³⁷ suggesting that Bnc3 is more rigid than n7OFF. We suspect that the Bn tail constrains the amide position and corresponding sidechain of Cys1, reducing the conformational freedom of Bnc3 relative to n7OFF.³⁷ A representative 3D NMR structure that most closely matches the ensemble “average” structure is shown in Figure 5B, alongside its 2D chemical structure in Figure 5A.

Figure 5.

2D structure and 3D NMR structure of Bnc3. A: Chemical structure of Bnc3, with tail and amino acid residues identified. B: Representative NMR structure of Bnc3, with the Connolly surface shown. Sulfur is gold, oxygen is red, nitrogen is blue, fluorine is cyan, amide hydrogens are white, and carbon is grey.

In the NMR structure of Bnc3, we clearly see its amphipathic nature, with one face comprised of lipophilic side chains and the opposing face comprised of numerous heteroatoms. As the ligand binding site of AgrC has been proposed to be a hydrophobic cleft in the extracellular transmembrane segments,^{33, 34, 48} it is logical to envisage such an amphipathic compound to engage with AgrC via its lipophilic face. This ampipathicity for Bnc3 is very similar to that observed in NMR structures of its parent compound, n7OFF, and other peptide-derived AIP mimetics.^{37, 43, 47} Indeed, the relative orientation of the side chains was found to coincide with patterns previously observed in our NMR structures of other AIP mimetics;^{37, 43, 47} for example, the two endocyclic hydrophobic residues of Bnc3 overlay with the two endocyclic hydrophobic residues of t-AIP-II and n7OFF (Figure 6A and 6B), and the orientation of the Cys residues of Bnc3 and t-AIP-II are also similar (Figure 6A). Additionally, a comparison of the hydrophilic Ser residues of t-AIP-II and the hydrophilic n7O linker of Bnc3 shows that they occupy similar spaces (Figure 6A). These structural trends support the hypothesis that one face of Bnc3 interacts with the proposed hydrophobic cleft of the receptor site, while the other face participates in favorable solvent interactions, thus stabilizing the ligand-AgrC complex. It is plausible that if an AgrC ligand does not have such favorable solvent interactions, it would likely pay an entropic penalty for perturbing the hydrogen bonding network of the surrounding water and interact less strongly with the receptor. This may partially explain the observation that n7Cha(3fF), which lacks the same hydrophilic face as n7OCha(3fF) due to the absence of the n7O linker, is less potent than n7OCha(3fF) (Table 2). We note that the structure adopted in solution of course may not be the same as the structure adopted when the ligand binds to AgrC, but for these rigidified systems, it is very likely that the conformations would be similar.

Figure 6.

NMR structure overlays. A: Overlay of representative NMR structures for t-AIP-II and Bnc3, using the nine most similar atoms for alignment (0.86 Å RMS difference). B: Overlay of representative NMR structures for n7OFF and Bnc3, using the 34 most similar atoms for alignment (1.49 Å RMS difference). Sulfur is gold, oxygen is red, nitrogen is blue, fluorine is cyan, amide hydrogens are white, and carbon is magenta (t-AIP-II), grey (Bnc3, or cyan (n7OFF).

CONCLUSION

This study reports our progress toward the goal of developing simplified, small molecule AgrC antagonists through the analysis of a reported peptidomimetic scaffold (n7OFF) with moderate potency against AgrC. The design, synthesis, and biological testing of a series of close n7OFF analogs revealed SARs important to its activity: namely, amino acids with highly lipophilic side chains at the 3 and 4 positions (with sp³ and sp² carbons at the respective γ-carbons), a bulky hydrophobic tail group, and an oxa-linker that creates an amphipathic nature are necessary elements for strong inhibitory potency against AgrC in all four groups of S. aureus (Figure 7). Critically, this work revealed new antagonists, such as Bnc3, which have low- to sub-nanomolar potencies against all four agr groups of S. aureus. We also discovered that an amide analog, Bn-n7OCpa(3fF)-amide, was only slightly less potent and should have utility as a probe compound with increased stability and solubility.

Figure 7.

The evolution of n7OFF to Bnc3 and key SAR features discovered in this study. The new AIP mimetic Bnc3 has a phenyl group added to the tail (in blue), 3-fluorophenylalanine adjacent to the thioester (in green), and cyclopentylalanine (in red) in place of Phe3 of n7OFF.

A solution NMR structure of Bnc3 revealed its amphipathic nature, with the relative orientation of the side chains overlapping with patterns previously seen in peptide AIP mimetics. This structural data illuminates the directions that functionality should protrude from the Bnc3 scaffold to effectively antagonize AgrC, and provides a 3D framework from which to begin to design non-peptide, small molecule modulators of AgrC. At a minimum, we reason that a thioester or suitable isostere and a hydrophilic ring structure need to be maintained in the structure, with lipophilic functionality present on the opposite side. The design and synthesis of compounds with such features is currently in progress and will be reported in due course.

METHODS

Reagents and general methods.

Chemical reagents and solvents were purchased from commercial sources (Sigma-Aldrich and Chem-Impex International) and used without further purification, with the exception of dichloromethane, which was distilled and stored over activated 3 Å molecular sieves. Dawson Dbz AM and Wang solid-phase resins were purchased from Novabiochem/EMD Millipore. 18 MΩ water was prepared using a Sartorius Arium Pro water purifier. Brain-heart infusion (BHI) medium powder was purchased from Teknova.

Peptide synthesis.

Thioester-, ester-, and amide-containing peptidomimetics were prepared on solid-phase resin, purified using reverse-phase semi-preparative HPLC, and characterized using high-resolution mass spectrometry (HRMS) and analytical HPLC as reported previously.³⁷ The olefin-containing peptidomimetics were prepared by solution-phase coupling of a linear peptide precursor prepared on solid support with a C-terminal amino olefin building block, followed by cyclization via ring-closing metathesis using Grubbs’ Generation-II catalyst and established procedures.^{41, 49} See SI for additional experimental details and characterization data.

Fluorescent reporter assay.

Compounds were evaluated for activity in AgrC using S. aureus YFP agr reporter strains generously provided by A. Horswill (strains AH1677, AH430, AH1747, and AH1872 for groups-I−IV, respectively, containing the pDB59 plasmid with YFP_10B gene under control of the P3 promoter of agr).⁴² Overnight cultures were diluted 1 to 50 into fresh BHI medium and added to a 96-well microtiter plate containing the compound (1% DMSO), as a single-concentration or serial dilution series, with DMSO and growth medium controls. The plates were incubated for 24 h with shaking at 200 rpm and 37 °C. OD₆₀₀ and YFP fluorescence (excitation 500 nm, emission 540 nm) were read with a BioTek Synergy 2 plate reader. Compounds were screened at a single concentration and, if deemed interesting, over a 1 to 3 serial dilution with at least 3 biological replicates. Each biological replicate was the average of 3 or 4 technical replicates and prepared from a fresh overnight culture. Raw fluorescence measurements were processed by subtracting the background (BHI medium), correcting for OD₆₀₀, and normalizing to the DMSO control to obtain percent agr activity for each compound concentration. A four-parameter sigmoidal dose response curve was fitted using GraphPad Prism 6, with the top of the curve set to 100% and the bottom of the curve constrained to remain above 0%, and IC₅₀ values with the 95% confidence intervals and Hill slopes (h) were calculated (see Tables S2 and S3, respectively).

Solution NMR structural studies.

Bnc3 was dissolved to a concentration of 2 mM in 1:1 CD₃CN:H₂O at pH 5 and characterized by NMR to assign chemical shifts and obtain interproton distances using ROESY and NOESY spectra, with scripts written for NIH Xplor (v 2.35) to determine a solution NMR structure as previously described (see SI for additional details).^{37, 43, 47} Structural images and statistics were generated using PyMOL v4.3.0, with the representative structure shown in Figure 5B having the most similarity to the average structure and lowest relative energy selected from the ensemble.⁵⁰