Simultaneous Control of Infection and Inflammation with Keratin-Derived Antibacterial Peptides (KAMPs) Targeting TLRs and Co-Receptors

Abstract

Controlling infection-driven inflammation is a major clinical dilemma due to limited therapeutic options and possible adverse effects on microbial clearance. Compounding this difficulty is the continued emergence of drug-resistant bacteria, where experimental strategies aiming to augment inflammatory responses for enhanced microbial killing are not applicable treatment options for infections of vulnerable organs. As with corneal infections, severe or prolonged inflammation jeopardizes corneal transparency leading to devastating vision loss. We hypothesized that keratin 6a-derived antimicrobial peptides (KAMPs) may be a two-pronged remedy capable of tackling bacterial infection and inflammation at once. We used murine peritoneal neutrophils and macrophages, together with an in vivo model of sterile corneal inflammation, to find that non-toxic and pro-healing KAMPs with natural 10- and 18-amino-acid sequences suppressed lipoteichoic acid (LTA)- and lipopolysaccharide (LPS)-induced NF-κB and IRF3 activation, proinflammatory cytokine production, and phagocyte recruitment independently of their bactericidal function. Mechanistically, KAMPs not only competed with bacterial ligands for cell surface TLR and co-receptors (MD-2, CD14, and TLR2), but also reduced cell surface availability of TLR2 and TLR4 through promotion of receptor endocytosis. Topical KAMPs treatment effectively alleviated experimental bacterial keratitis, as evidenced by substantial reductions of corneal opacification, inflammatory cell infiltration, and bacterial burden. These findings reveal the TLR-targeting activities of KAMPs and demonstrate their therapeutic potential as a multifunctional drug for managing infectious inflammatory disease.

One Sentence Summary:

Bifunctional keratin peptides deliver a one-two punch to alleviate inflammation and infection and avoid functional damage in corneal tissue.

INTRODUCTION

Although infection-driven immune cell infiltration is a natural inflammatory response to fight invading microbes, it can easily cause serious damage to vulnerable tissues leading to permanent organ dysfunction. To reduce the risk of long-term disability and mortality, the infection itself and associated acute inflammation should both be controlled as early as possible (1–3). Unfortunately, this treatment goal is hampered not only by the ever-growing problem of antibiotic resistance, but also by limited anti-inflammatory drug options. For instance, even though corticosteroids are potent immunosuppressive agents that alleviate inflammation, their usage in conjunction with active antimicrobial therapy can result in potential harmful effects including impairment of microbial clearance and severe complications. Indeed, the clinical use of adjunctive corticosteroids in infections is far from optimal as it can be beneficial or harmful depending on a number of conditions, such as comorbidities, dosing, timing of treatment initiation, and type of causative pathogen (4–7). Undoubtedly, additional therapeutic options are needed for simultaneous control of infection and inflammation.

Bacterial keratitis is among the infectious inflammatory diseases that affect ocular tissues, which are functionally vulnerable to the deleterious effects of inflammation. In particular, scarring of the cornea impacts its optical clarity and proper refractive shape that are required for good vision, leading to devastating visual impairment or blindness. The first-line treatment is topical antibiotics, such as fortified vancomycin and fluoroquinolone monotherapy, which are the most common empirical choices (7, 8). However, both are limited by their respective drawbacks of high ocular toxicity and low susceptibility of methicillin-resistant Staphylococcus aureus (8, 9). Notably, rising antibiotic resistance among ocular pathogens is associated with worse clinical outcomes (10, 11). Nevertheless, even if infectious organisms are susceptible to treatment, inflammation can continue to progress and cause severe corneal opacification and ulcers that irreversibly impair vision (12). To mitigate inflammation-associated damage, current guidelines recommend initiation of adjunctive topical corticosteroids for culture-positive non-Nocardia bacterial keratitis 24 to 48 hours after initial antibiotic administration when infection is under control (13), although their potential negative effect on re-epithelialization of the cornea still exists (14–16). Notably, early addition of corticosteroids to the treatment course is critical to improved visual outcomes (3), as in the case of bacterial meningitis, hearing loss, and neurologic damage can be prevented only when adjunctive steroids are initiated before (by 10-20 minutes) or concomitantly with the first dose of antibiotics (2, 17). Therefore, treatments possessing dual antibacterial and anti-inflammatory functions to enable effective bacterial clearance and inflammation control at the earliest possible could minimize the debilitation and mortality of infectious inflammation.

Human keratin 6a (K6a) is an intermediate filament protein highly expressed in various epithelial cells. We previously reported a direct antibacterial response of epithelial cells involving reorganization of the endogenous K6a filament network to upregulate the amount of cytosolic K6a, which is processed by the ubiquitin-proteasome system to generate a series of short bactericidal peptides (18). The glycine-rich short peptides from the carboxyl terminal region of K6a (residues 515 to 559), termed keratin-derived antimicrobial peptides (KAMPs), are structurally flexible (single coils), salt-tolerant, and bactericidal via cell envelope disruptions and, potentially, intracellular targeting mechanisms (19, 20). As such, KAMPs are the only human example among the structurally unique non-αβ class of host defense peptides (20, 21). In our previous study that characterized the bactericidal activities of two specific KAMPs, KAMP18C (RAIGGGLSSVGGGSSTIK) and the core sequence KAMP10 (GGLSSVGGGS), we found that the former caused major damage to the cell envelope of Staphylococcus aureus, whereas the latter perforated Pseudomonas aeruginosa membrane (20). Given antimicrobial peptides often employ multiple bactericidal mechanisms including rapid cell lysis, the likelihood of bacterial resistance development or transmission is admittedly lower compared to the case of conventional antibiotics (22–24). Moreover, a number of studies have showed that natural antimicrobial peptides can be modified by amino acid substitution to reduce cytotoxicity, enhance the activity or spectrum of antimicrobial activity, or incorporate anti-biofilm or immunomodulatory functions (25, 26).

Studies examining potential of host defense peptides as anti-infective alternatives commonly use animal models of skin and lung infections to determine bacterial clearance efficacy. On the other hand, the immunomodulatory potential of engineered antimicrobial peptides is often examined alone using animal models of endotoxin-induced sterile sepsis, which focus on mortality rate as the treatment outcome. As such, there is a paucity of preclinical research exploring the use of antimicrobial peptides to reduce tissue damage and functional loss through concomitant control of infection and inflammation. Given our previous study on the characterization of KAMP10 and KAMP18C structures and bactericidal activities (19, 20), we set out to examine the immunomodulatory capacity of the two peptides in this work using mouse peritoneal neutrophils and macrophages as well as a murine model of sterile corneal inflammation. Our findings highlight the multiple intervention mechanisms of KAMPs in TLR activation and support that concomitant control of infection and aggressive inflammation by KAMPs could be an effective strategy to reduce tissue damage and severe debilitation.

RESULTS

Safety assessment of KAMPs in vitro and in vivo

Before assessment of KAMP eye drops in vivo, we tested whether KAMP10 and KAMP18C would pose undesirable effects on cell viability and corneal integrity. We incubated human telomerase-immortalized corneal epithelial (hTCEpi) cells (27) for 24 h with KAMPs or scrambled control SC10 at bactericidal concentration (200 μg/ml corresponding to 0.02% (w/v)) determined by in vitro antimicrobial assay (19), then assessed cell viability using MTT (3-[4,5-dimethylthiazol-2-yl]-2,5 diphenyltetrazolium bromide) assays. Benzalkonium chloride (BAK, 0.001%) was used as a positive control as it is commonly used as a preservative (0.004-0.025%) in topical ophthalmic formulations, although its cytotoxicity and deleterious effects have raised concerns (28). In contrast to BAK, hTCEpi cell viability after treatment with KAMP10, KAMP18C, or SC10 was comparable to the baseline of untreated cells, indicating that KAMPs did not induce cell death (Fig. S1A). We also found that both KAMPs did not induce apoptosis in hTCEpi cells during the 24-h incubation period, as assessed by TUNEL assay of DNA fragmentation (Fig. S1B). Next, we examined potential hemolytic effects by measuring the amount of hemoglobin released from lysed red blood cells after KAMP10 or KAMP18C treatment. We found no difference from the baseline of untreated samples (Fig. S1C).

Because KAMPs appeared to be safe in vitro, we examined their effects, if any, on corneal epithelium integrity in vivo. Anesthetized mice were topically applied once with saline containing BAK, SC10, KAMP10, or KAMP18C at 0.1% (the concentration used for the subsequent treatment efficacy studies), and then checked for barrier function using fluorescein staining. We detected a high degree of fluorescein penetration in the BAK-treated corneas, whereas KAMP10-or KAMP18C-treated eyes remained stain-free similarly to the SC10- or saline-treated controls (Fig. S1D). Given impaired wound healing is a major concern over the use of immunosuppressants (topical steroids) in the treatment of bacterial keratitis, we also examined the effects of KAMPs on corneal re-epithelialization in a mouse wound healing model. Epithelia on central corneas were mechanically removed, then topically treated with saline containing 0.1% SC10, KAMP10, or KAMP18C. At 6 h and 24 h after wounding, we found that barrier restoration was substantially accelerated by both KAMPs in contrast to saline and SC10 (Fig. S1E). Taken together, the results support the safety and tolerability profile of KAMP eye drops, which could be used to treat infection and inflammation of barrier-compromised sites without the undesirable side effect on wound healing.

KAMPs suppress TLR-mediated cytokine secretion from murine neutrophils and macrophages

Next, we tested whether KAMPs have immunomodulatory activity that could potentially benefit the management of bacterial keratitis. Thioglycolate-elicited mouse peritoneal neutrophils and macrophages were stimulated with toll-like receptor (TLR) ligands in vitro, alone or in combination with KAMPs, followed by quantitative measurements of secreted cytokines (IL-6, TNFα, and G-CSF) and chemokines (CXCL1, CXCL2, CXCL10) by ELISA. Purified bacterial ligands most relevant to bacterial keratitis, Gram-negative P. aeruginosa lipopolysaccharide (LPS) and Gram-positive S. aureus lipoteichoic acid (LTA) were added to culture media to activate cell surface/endosomal TLR4 and TLR2, respectively. At 8 h post-ligand stimulation, we found that pretreatment or post-treatment with KAMP10 or KAMP18C, but not SC10, inhibited proinflammatory cytokine and chemokine production by neutrophils (Fig. 1A) and macrophages (Fig. 1B). The inhibitory effects were dose-dependent (Fig. S2). At 24 h post-ligand stimulation, we continued to observe effective suppression by KAMPs treatment (Fig. S3). To further examine the attenuation effect of KAMPs on LPS- and LTA-activated TLR signaling, we performed flow cytometric analyses of NFκB p65 and IRF3 phosphorylation in neutrophils at 15 min, 1 h, and 8 h post-stimulation. In accordance with the suppression of proinflammatory cytokine production, the percentage of phospho-p65⁺ or phospho-IRF3⁺ neutrophils were reduced to baseline by KAMP10 administered either before or after stimulation (Fig. 1C and fig. S4). To demonstrate the selectivity of KAMPs for TLRs, we utilized synthetic double-stranded RNA analog poly(I:C) that activates cell surface/endosomal TLR3 when added to culture media but cytoplasmic dsRNA sensors RIG-1/MDA-5 when transfected into cells. KAMP10 inhibited p65 and IRF3 phosphorylation in neutrophils stimulated with media poly(I:C) (Fig. 1D). Notably, the inhibitory effect was not evident when poly(I:C) and KAMP10 were co-transfected into the cells (Fig. 1E).

Fig. 1. KAMP10 and KAMP18C attenuate LPS- and LTA-induced inflammatory responses in murine neutrophils and macrophages.

Peritoneal neutrophils and macrophages were stimulated with purified LPS or LTA (500 ng/ml) alone or in combination with 200 μg/ml of KAMP10 (K10), KAMP18C (K18C), or scrambled KAMP10 (SC10). The peptides were applied either 30 min before (Pre) or 30 min after (Post) the stimulant. Cells treated with medium only served as baseline controls. (A-B) Cytokines in culture supernatants of neutrophils (A) and macrophages (B) after 8 h stimulation were measured by ELISA. See also fig. S2 for dose-dependent suppression of additional three cytokines, and fig. S3 for 24-h post-challenge measurement. Each independent stimulation experiment was performed with one cell type, one stimulant, and one timepoint. A total of four independent 8-h experiments are shown here. (C-E) Time-dependent representative flow cytometric histogram plots showing phospho-p65 and phospho-IRF3 of neutrophils after stimulation. LPS (C) or poly(I:C) (D) was added to media. See also fig. S4 for LTA stimulation. Poly(I:C) was co-transfected (E) with K10 or SC10 and intracellular staining was performed 1 h after stimulation. Mean ± SD (N = 3 replicates). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, not-significant (ns): P > 0.05 (one-way ANOVA with Dunnett’s post hoc test against sham-treated LPS- or LTA-challenged group).

Topical KAMPs suppress bacterial ligand-induced corneal inflammation in mice

To examine if cytokine-suppressive activity of KAMPs could be observed in vivo, we employed an established mouse model of sterile corneal inflammation induced by purified bacterial ligands (LPS or LTA) or heat-killed bacteria (P. aeruginosa or S. aureus). At 24 h post-induction, we quantified cytokines and innate immune cells in whole corneas by ELISA and flow cytometry, respectively, to assess early inflammatory responses. LPS alone robustly induced cytokine (IL-6, TNFα) and chemokine (CXCL1, CXCL10) production in mouse corneas; unlike the scrambled control SC10, a topical dose of KAMP10 or KAMP18C applied 30 min either before or after stimulation significantly reduced their amounts (P < 0.0001) (Fig. 2A). Quantification of total leukocytes (CD45⁺), which were primarily composed of neutrophils (CD45⁺Ly6B.2⁺F4/80⁻) and macrophages (CD45⁺F4/80⁺), showed that both KAMPs markedly reduced innate cell infiltration to the inflamed corneas (P < 0.05) (Fig. 2B), consistent with suppressed cytokine and chemokine production. Similar observations were made when LTA (Fig. 2, C and D), heat-killed P. aeruginosa (Fig. S5, A and B), or heat-killed S. aureus (Fig. S5, C and D) was used to induce inflammation. Collectively, our in vitro approach employing TLR ligand stimulation of primary myeloid cells as well as in vivo approach using a mouse model of sterile corneal inflammation indicate that KAMP10 and KAMP18C are bifunctional peptides – they have potent anti-inflammatory activity that is independent of their previously characterized antimicrobial function (19, 20).

Fig. 2. KAMP10 and KAMP18C suppress LPS- and LTA-induced corneal inflammation in a murine model.

(A-B) Purified LPS or (C-D) LTA (20 μg) or PBS only was topically applied to scarified mouse corneas. Inoculated eyes were sham-treated or given one topical application of KAMP10, KAMP18C, or scrambled KAMP10 (SC10) 30 minutes either before (100 μg/ml) or after inoculation (200 μg/ml). Concentrations were chosen based on in vitro data. At 24 hours post-inoculation, mice were euthanized, and whole corneas were dissected for cytokine and immune cell analysis. ELISA quantification of cytokines per cornea (A, C). Representative FACS bivariate dot plots (B, D) showing Ly6B.2 and F4/80 gated cells among total CD45⁺ cells in the inflamed corneas received peptide pre- or post-treatment. Numbers represent percentage of cells in each gate relative to total CD45⁺ cells. Total leukocytes (CD45⁺), neutrophils (CD45⁺Ly6B.2⁺F4/80⁻) and macrophages (CD45⁺F4/80⁺) per cornea were quantified. Mean ± SD (n = 3-7 mice (corneas)/group). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, not-significant (ns): P > 0.05. Statistical significance was determined by one-way ANOVA (A, B, D) or Welch and Brown-Forsythe ANOVA (C) with Dunnett’s post hoc test against sham-treated LPS- or LTA-challenged group. See also fig. S5 for stimulation with heat-inactivated P. aeruginosa or S. aureus.

KAMPs compete with LPS and LTA for MD2, TLR2, and CD14 binding

We next questioned how KAMPs dampen TLR-mediated cytokine production. In our previous report, KAMP10 and KAMP18C exhibited only modest binding affinity for LPS and LTA, respectively (20). Thus we hypothesized that KAMPs exert anti-inflammatory effects via other routes apart from ligand sequestration, which is the typical mechanism shared among various antimicrobial peptides (29, 30). To test the hypothesis that direct interactions of KAMPs with MD2/TLR4 and CD14/TLR2 prevent activation of these receptors by LPS and LTA, respectively, we conducted in vitro blocking and displacement assays using latex beads conjugated with either human MD2/TLR4 complex or TLR2 on the surface. Flow cytometric detection of receptor and bacterial ligand fluorescence after 30-min preincubation of MD2/TLR4 complex-conjugated beads with KAMP10 or KAMP18C, but not SC10, effectively inhibited subsequent LPS binding to the PBS-washed beads, as indicated by >90% reduction of LPS⁺ receptor-conjugated bead population compared to SC10 and media controls (Fig. 3A). Notably, this inhibitory effect was abolished when soluble MD2 was included in the media during preincubation (Fig. 3B), suggesting that KAMP10 and KAMP18C block the LPS binding site on MD2 to prevent LPS-MD2/TLR4 complex formation.

Fig. 3. KAMP10 and KAMP18C directly interacts with MD2, CD14 and TLR2 to disrupt LPS-MD2/TLR4 and CD14-LTA-TLR2 binding.

(A-B) MD2 blocking assay. Representative flow cytometric histogram plots and percentage of LPS⁺ beads among TLR4/MD2-conjugated latex beads. Receptor-conjugated latex beads were incubated with peptides (60 μM) in the absence (A) or presence (B) of soluble MD2 (sMD2; 1 μg/ml), washed, then incubated with FITC-LPS (1 μg/ml). (C-D) TLR2 and CD14 blocking assays. Representative flow cytometric histogram plots and percentage of LTA⁺ beads among TLR2-conjugated latex beads. (C) TLR2-conjugated beads were incubated with peptides (60 μM), washed, then incubated with FITC-LTA (1 μg/ml) and CD14 (0.1 μM). (D) CD14 (0.1 μM) was preincubated with equal molar of peptides (0.1 μM) before mixing with TLR2-conjugated beads and FITC-LTA (1 μg/ml). (E-F) LPS and LTA displacement assays. Representative flow cytometric histogram plots and percentage of bacterial ligand-bound receptor-conjugated beads. TLR4-MD2 complex-conjugated beads (E) and TLR2-conjugated beads (F) were preincubated with FITC-LPS and FITC-LTA (1 μg/ml) respectively, washed and mixed with peptides (60 μM). Mean ± SEM (n = 3-5 independent experiments). **P < 0.01, ***P < 0.001, ****P < 0.0001, not-significant (ns): P > 0.05 (one-way ANOVA with Dunnett’s post hoc test against media control or K10).

Likewise, we found that preincubation of TLR2-conjugated beads with KAMP10 or KAMP18C effectively prevented subsequent binding of LTA-CD14 to the PBS-washed beads (Fig. 3C), suggesting that these peptides directly block the LTA-binding site on TLR2. As CD14 is known to bind LTA and facilitate LTA transfer to TLR2, we performed another experiment in which CD14 was preincubated with equal molar (0.1 μM) of KAMPs before mixing with LTA and TLR2-conjugated beads. CD14 preincubation with KAMP10 and KAMP18C, as opposed to SC10, disrupted LTA binding to TLR2 (Fig. 3D). It is noteworthy that the peptide concentration here (0.1 μM) was equivalent to 1/600 of the one used in the aforementioned TLR2 blocking experiment, and that we had confirmed preincubation of TLR2-conjugated beads with KAMPs at this low concentration did not have appreciable blocking effects on subsequent LTA-CD14 binding to the PBS-washed beads, suggesting that KAMP-CD14 interaction directly inhibits CD14-mediated transfer of LTA to TLR2. It is also possible that KAMP-CD14 interaction enhances TLR2 blocking by KAMPs.

Next, we examined whether KAMPs can displace receptor-bound bacterial ligands. We first preincubated MD2/TLR4 complex- or TLR2-conjugated beads with LPS or LTA, respectively. Subsequent addition of KAMP10 and KAMP18C to the PBS-washed beads caused substantial reduction in LPS⁺ and LTA⁺ receptor-conjugated bead population, with a displacement of >90% within 30 min obtained at peptide concentration of 60 μM (Fig. 3, E and F), as opposed to SC10 and media controls. Taken together, the findings of competition and displacement assays strongly support the notion that direct interactions of KAMPs with TLRs suppress host inflammatory responses to bacterial ligands.

KAMPs induce endocytosis of bacterial ligand-free TLR2 and TLR4

In addition to cell surface activation of TLR proinflammatory signaling, bacterial ligands complexed with MD2/TLR4 (LPS) or TLR2 (LTA), together with CD14, are internalized to induce endosomal signaling that activates interferon (IFN) regulatory factors (IRFs) for type I IFN gene expression (31, 32). Thus, TLR endocytosis is also considered an important negative regulatory mechanism for controlling the magnitude of inflammatory responses (33–36). To examine whether KAMPs modulate TLR endocytosis, thioglycolate-elicited mouse peritoneal neutrophils and macrophages were treated with KAMPs alone, followed by flow cytometric analyses of extracellular and intracellular TLRs. As in the case of LPS, incubation with KAMP10 or KAMP18C but not SC10 alone for 1 h reduced cell surface staining but increased intracellular staining of MD2/TLR4 for both neutrophils (Fig. 4A) and macrophages (Fig. 4B). Similarly, we observed that KAMP10 and KAMP18C alone triggered internalization of TLR2 into neutrophils (Fig. 4C) and macrophages (Fig. 4D) as did LTA. These findings, together with those demonstrating the ability of KAMPs to engage TLRs, suggest that the anti-inflammatory mechanisms of KAMPs are multifaceted, which include direct blocking of MD2, CD14, and TLR2 against LPS and LTA binding, as well as direct induction of receptor internalization before they are engaged by bacterial ligands.

Fig. 4. KAMP10 and KAMP18C alone induce endocytosis of bacterial ligand-free TLR4 and TLR2.

Thioglycolate-elicited mouse peritoneal (A, C) Ly6B.2⁺ neutrophils and (B, D) F4/80⁺ macrophages were stimulated with LPS or LTA (500 ng/ml), or KAMP10, KAMP18C or SC10 (200 μg/ml) alone for 1 hour. Sham (media) treatment served as baseline. Cell surface and intracellular TLR4 (A, B) and TLR2 (C, D) were detected by flow cytometry. Representative flow cytometric histogram plots and percentage of neutrophils or macrophages that were stained positive for the receptors are shown. Mean ± SEM (n = 3-6 independent experiments). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, not-significant (ns): P > 0.05 (one-way ANOVA with Dunnett’s post hoc test against media treatment).

To obtain structural clues to KAMPs’ direct interaction with TLRs, we modeled their binding structures computationally using HPEPDOCK (37), one of the best-performing peptide-protein docking programs with regard to docking accuracy for blind global docking (38). The top five favorable docked poses of KAMP10 were found in the hydrophobic barrel-like cavity of soluble and TLR4-associated MD2 (Fig. S6, A and B), and KAMP18C was predicted to interact with the hydrophobic pockets at TLR2 central domain (Fig. S6C) and CD14 N-terminal domain (Fig. S6D). It is well characterized that LPS (the acyl chains of its lipid A core) binds to the hydrophobic barrel-like cavity of MD2 in both soluble form (sMD2) and TLR4-associated form (MD2/TLR4) (39–42), and that acylated LTA binds to the hydrophobic pockets at both TLR2 central domain and CD14 N-terminal domain (43–46). The predicted docking sites suggest that KAMPs could potentially interact with MD2, CD14, and TLR2 at their LPS and LTA binding sites, which mirror their observed activities in the binding and displacement assays. Further investigations using molecular dynamics simulation and cryo-electron microscopy will be valuable in characterizing the structural basis and dynamics of their interactions at the atomic level.

Prophylactic KAMPs restrict bacterial infection and inflammation in a murine keratitis model

It has been shown that antibiotic-killed P. aeruginosa and S. aureus are still capable of triggering a high magnitude of neutrophil recruitment in the murine keratitis model (47, 48). Therefore, suppressing local inflammatory responses in conjunction with anti-infective treatment may conceivably minimize tissue damage and improve clinical outcomes. Given that bifunctional KAMPs have potent bactericidal and anti-inflammatory activities, we first evaluated their prophylactic efficacy against bacterial keratitis in a murine model. Saline (sham treatment), 0.1% (5 μg) of LL-37, or KAMPs were topically applied once to scarified mouse corneas 30 min before inoculation with live Gram-negative bacteria (amikacin-resistant P. aeruginosa keratitis isolate) or Gram-positive bacteria (standard laboratory strain of S. aureus). LL-37 is a well-characterized 37-amino-acid human cathelicidin antimicrobial peptide expressed in many cell types and tissues including leukocytes and corneal epithelia. As a reference, human LL-37 was included here based on its structural similarity to KAMPs (single helix versus single coil), high expression by corneal epithelial cells, characterized bactericidal activities in high-salt media, and its known LPS- and LTA-neutralizing activities (49, 50). In the P. aeruginosa keratitis model, we observed that KAMP10- or KAMP18C-pretreated mice at 24 h post-infection had minimal and superficial corneal opacification represented by a marked reduction of disease severity scores compared to the sham-, SC10-, and LL-37-pretreated groups, which all had prominent opacities and epithelial defects (Fig. 5, A and B). Along with disease severity, both KAMPs reduced bacterial load in the infected corneas by >95% (Fig. 5C). Flow cytometric analysis of corneal immune cells showed that the numbers of infiltrated leukocytes (CD45⁺), neutrophils (CD45⁺Ly6B.2⁺F4/80⁻) and macrophages (CD45⁺F4/80⁺) were significantly reduced in KAMP10- or KAMP18C-pretreated mouse corneas (P < 0.01) (Fig. 5, D and E). Similarly, in the S. aureus keratitis model, disease severity (Fig. 5, F and G), bacterial burden (Fig. 5H), and immune cell infiltration (Fig. 5, I and J) at 24 h post-infection were alleviated by pretreatment with KAMP10 or KAMP18C. The data supported that prophylactic application of bifunctional KAMPs on wounded (susceptible) corneas may lower the incidence of severe corneal scarring, attributed to concomitant bacteria killing and immunosuppressive effects.

Fig. 5. KAMP10 or KAMP18C prophylaxis reduces severity of bacterial keratitis in a murine model.

Scarified mouse corneas (one per mouse) were topically applied with saline, LL-37, KAMPs or SC10 (5 μg peptide), and 30 minutes later inoculated with (A-E) amikacin-resistant P. aeruginosa (10⁵ CFU) or (F-J) S. aureus (10⁶ CFU). Disease presentation (A, F) at 24 h post-inoculation, and severity scores (B, G) corresponding to the extents of opacification (area and density) and corneal surface irregularity. Numbers of viable bacteria (C, H), total leukocytes (CD45⁺), neutrophils (CD45⁺Ly6B.2⁺F4/80⁻) and macrophages (CD45⁺F4/80⁺) (D, I) per infected corneas. Representative FACS bivariate dot plots (E, J) showing Ly6B.2 and F4/80 gated cells among total CD45⁺ cells in the infected corneas received various prophylactic treatment. Numbers represent percentage of cells in each gate relative to total CD45⁺ cells. n = 8-10 mice (corneas)/group from two independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, not-significant (ns): P > 0.05. Statistical significance against vehicle-treated groups was determined by one-way ANOVA (B, G) or Welch and Brown-Forsythe ANOVA (C, H) with Dunnett’s post hoc tests, or Kruskal-Wallis with Dunn’s post hoc tests (D, I).

Therapeutic KAMPs abrogate established bacterial infection and inflammation in a murine keratitis model

We next assessed the potential of bifunctional KAMPs as a therapeutic agent for the treatment of ongoing bacterial keratitis. When infection and inflammation of the mouse corneas had been established one day after bacterial inoculation, we began topical treatment with 0.1% (5 μg) peptides three times daily for 1 or 3 days. Because poor absorption is a major concern for ocular drugs (51), liposomes (52, 53) in lieu of subconjunctival injection were used here as a vehicle to deliver peptide into the corneal stroma, where bacteria and immune cell infiltrates were accumulating.

Mouse corneas infected by amikacin-resistant P. aeruginosa showing similar disease severity (mean clinical score 5.3-5.5, P > 0.05) were topically treated with liposome vehicle alone or mixed with LL-37, SC10, KAMP10 or KAMP18C. One day after treatment (1 DAT), we began to observe a significant reduction of disease severity in both groups treated with KAMP10 or KAMP18C (P < 0.001), with the effect more prominent in those that received KAMP10 (Fig. 6, A and B). On the contrary, severity progression or lack of improvement over the 3-day treatment course was observed in eyes received vehicle alone or mixed with SC10 or LL-37. Consistent with disease presentations, flow cytometric analysis of acute inflammatory cells in the infected mouse corneas showed that total leukocytes, neutrophils, and macrophages were substantially reduced by KAMPs as early as one day post-treatment (Fig. 6, C–F). Notably, the suppressed immune responses did not result in uncontrolled bacterial growth. Indeed, both KAMPs demonstrated significant contribution to bacterial eradication (P < 0.0001), as shown by ~2-log and ~3-log reduction of P. aeruginosa in corneas on day 1 and day 3 post-treatment, respectively, when compared with the control groups (Fig. 6G). The results showed that topical KAMPs achieved simultaneous control of the Gram-negative bacterium and inflammation-associated tissue damage.

Fig. 6. KAMP10 or KAMP18C treatment effectively controls bacterial burden and alleviates inflammation in a murine P. aeruginosa keratitis model.

Scarified mouse corneas (one per mouse) were inoculated with amikacin-resistant P. aeruginosa (10⁵ CFU). One day after infection, topical treatment (3 times daily) commenced with liposome alone or liposome-associated LL-37, SC10, or KAMPs (5 μg peptide). Mice were euthanized one day or three days after treatment (DAT). (A-B) Disease severity scores on 1 DAT (A) and 3 DAT (B) (solid circles) with reference to the day treatment commenced (open circles). (C-F) Flow cytometric analysis of innate immune cells from whole infected corneas. Quantification of total leukocytes (C), Ly6B.2⁺F4/80⁻ neutrophils (D), and F4/80⁺ macrophages (E) per infected cornea. Representative FACS dot plots (F) showing CD45⁺ leukocytes (top) were gated in and analyzed for expression of Ly6B.2 and F4/80 surface markers (bottom). SSC-A: Side Scatter - Area. Numbers represent percentage of cells in each gate relative to total CD45⁺ cells. (G) Numbers of viable bacteria per infected corneas. n = 9-10 mice (corneas)/group from two independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, not-significant (ns): P > 0.05. Statistical significance was determined by two-way ANOVA (A-B) or Welch and Brown-Forsythe ANOVA (E-1DAT, G) with Dunnett’s multiple comparison test, or Kruskal-Wallis with Dunn’s multiple comparison test (C, D, E-3DAT).

To examine whether KAMPs are also effective against Gram-positive bacterial keratitis, we infected mouse corneas with S. aureus. One day after bacterial inoculation we commenced topical treatment with vehicle alone or mixed with LL-37, SC10, KAMP10, or KAMP18C (Fig. 7). Consistent with the findings in P. aeruginosa keratitis model, both KAMPs, but not SC10 or LL-37, provided significant clinical improvement beginning one day post-treatment (P < 0.0001), with KAMP18C more effective against S. aureus infection (Fig. 7, A and B). Furthermore, both KAMPs effectively suppressed infiltration of total leukocytes, neutrophils, and macrophages (Fig. 7, C–F) while substantially reducing the infectious load (Fig. 7G) in the S. aureus-infected corneas. The Gram-positive bacterial keratitis model again demonstrated concomitant control of infection and inflammation by dual-functional KAMPs is efficacious.

Fig. 7. KAMP10 or KAMP18C treatment effectively controls bacterial burden and alleviates inflammation in a murine S. aureus keratitis model.

Scarified mouse corneas (one per mouse) were inoculated with S. aureus (10⁶ CFU). One day after infection, topical treatment (3 times daily) commenced with liposome alone or liposome-associated LL-37, SC10, or KAMPs (5 μg peptide). Mice were euthanized one day or three days after treatment (DAT). (A-B) Disease severity scores on 1 DAT (A) and 3 DAT (B) (solid circles) with reference to the day treatment commenced (open circles). (C-F) Flow cytometric analysis of innate immune cells from whole infected corneas. Quantification of total leukocytes (C), Ly6B.2⁺F4/80⁻ neutrophils (D), and F4/80⁺ macrophages (E) per infected cornea. Representative FACS plots (F) showing CD45⁺ leukocytes (top) were gated in and analyzed for expression of Ly6B.2 and F4/80 surface markers (bottom). Numbers represent percentage of cells in each gate relative to total CD45⁺ cells. (G) Quantification of viable bacteria per infected corneas. n = 5-10 mice (corneas)/group from two independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, not-significant (ns): P > 0.05. Statistical significance was determined by two-way ANOVA (A-B), one-way ANOVA (G-1DAT) or Welch and Brown-Forsythe ANOVA (C- to E-1DAT, G-3DAT) with Dunnett’s multiple comparison test, or Kruskal-Wallis with Dunn’s multiple comparison test (C- to E-3DAT).

DISCUSSION

Inflammatory reactions to bacterial infections are aggressive and must be controlled as early as possible to avoid collateral tissue damage. This is important when treating infection-driven inflammation of delicate organ systems such as the brain and the eye. To avoid functional loss of the affected organs and achieve best possible clinical outcomes, effective antibiotics and swift alleviation of inflammation are both essential. Although corticosteroids are potent first line immunosuppressants, their potential side effects, such as elevated risks of exacerbating infection, warrant careful clinical consideration especially when patients with co-morbidities are involved. Unfortunately, the prudent and timely use of corticosteroids is further complicated by the increasing number of antibiotic-resistant bacteria that are difficult to eradicate. Clearly, optimal treatment of infectious inflammation faces two distinct but interconnected clinical challenges, control of infection and inflammatory cell infiltration. Using murine sterile and bacterial keratitis models, we have revealed the anti-inflammatory activities of native bactericidal peptides derived from human keratin 6a (KAMPs) and demonstrated their potential as multi-function therapeutic agents that may address the unmet needs in the management of infectious inflammation through eradication of drug-resistant bacteria, early control of harmful inflammation, and restoration of epithelial integrity.

The first challenge for optimal treatment of infectious inflammation is antibiotic resistance. Each year over two million cases of drug-resistant bacterial infections costing at least 23,000 lives are reported in the United States alone (54). Its negative impact on national and global health has been staggering rapidly and will continue to intensify unless new treatments are made available (55). Furthermore, since drug-resistant bacteria are often more virulent and more inflammatory to innate immune cells, especially after exposure to antibiotics which they have low susceptibility to (56), these infections are far more difficult to control and inevitably delay initiation or increase the risk of adverse effects of adjunctive steroid therapy, leading to substantial burden of morbidity and long-term disability in survivors (57). Antimicrobial peptides (AMPs) are one of the few promising alternatives to conventional antibiotics due to their rapid, multifaceted and broad-spectrum bactericidal activities that confer lower propensity for resistance development (58, 59). A number of natural and designed AMPs have been identified or developed over the past three decades and several of them are currently under evaluation in clinical trials for localized infections (60–62). Yet, the main challenges for translating most of the AMP-based drug candidates to the clinical phase of development include their inherent susceptibility to inhibition by in vivo environmental factors, such as physiological salts and proteases, as well as potential cellular toxicity and high cost of production (26, 63). Although technological advances in peptide engineering, nanocarrier formulation and chemical synthesis continue to demonstrate great promise for improving the safety, efficacy, stability, and manufacturing process of AMPs (64), our study shows that KAMP10 and KAMP18C - short peptides with native and non-chemically modified sequences from human cytokeratin 6a protein - have no detectable toxic effect on corneal epithelial cells, immune cells, or blood cells, and they also remain bactericidal on the ocular surface and in the infected cornea where bacterial and host proteases are present (65, 66). It is worth noting that the antibacterial effectiveness of KAMPs was not undermined by the antibiotic (amikacin) resistance of the P. aeruginosa clinical isolate used in this study. Similar to most peptide-based drugs (67), the major factor that determines therapeutic efficacy of KAMPs is bioavailability at the site of action. With the aid of regular (non-custom-made) phosphatidylcholine/stearylamine/cholesterol liposomes as the topical delivery vehicle, we have been able to forgo invasive subconjunctival injections and yet enhance tissue penetration of KAMPs to the mouse corneal stroma where bacteria and immune cells accumulate during active infection. The proof-of-concept evidence presented here supports further investigations into the range of drug-resistant bacterial species susceptible to KAMPs (alone or in cooperation with current antibiotics), as well as the optimization of physiochemical features of liposomes that enhance efficacy and bioavailability (tissue penetration in the case of topical use) of KAMPs (64, 68). The prospect of utilizing the antimicrobial characteristic of KAMPs to treat infection is also improved by increasing evidence showing that combined applications of membranolytic AMPs with conventional antibiotics that target intracellular machineries often produce additive or even synergistic bactericidal effects, which in turn afford reduction of drug use and potentiation of activity against drug-resistant bacteria (69).

The second challenge for optimal treatment of infectious inflammation is timely control of immune cell influx. While actively participating in innate defense, several AMPs are also known for their roles in immunomodulation because abnormal expression of these peptides is associated with various inflammatory skin and gastrointestinal diseases (70, 71). Typically, these AMPs enhance the recruitment of leukocytes to the site of microbial invasion through their direct chemotactic activity or by acting on local structural and immune cells to selectively modulate cytokine and chemokine milieu (72). This feature is particularly valuable for clearing drug-resistant bacterial pathogens, as demonstrated by engineered AMPs that reduced pathogen load and mortality in murine peritoneal infection (73). However, the potential risks of collateral tissue damage and resulting functional loss that are associated with amplified inflammatory responses and cellular recruitment remain a major drawback, rendering these AMPs inapplicable to treating infectious inflammation of delicate organs. Indeed, functional outcomes of these diseases are tremendously important, yet often overlooked or insufficiently measured in most studies that aimed to develop newer and better anti-infectives or anti-inflammatories, such as AMPs and their engineered derivatives. These studies have singularly focused on either bacterial killing or amelioration of cytokine production in immune cells for the control of deadly sepsis. In the case of acute infectious inflammation, bifunctional KAMPs can prove particularly useful – not only do KAMPs kill bacteria as demonstrated in the corneal infection model, but the most notable outcome of their concomitant anti-inflammatory activities is the substantial reduction and even prevention of immune cell-mediated corneal ulceration and opacification, a leading cause of blindness globally (74). Conceivably, extended studies using animal models of various infectious and non-infectious inflammatory disorders would be necessary to investigate potential indications of KAMPs beyond keratitis for prevention of permanent disabilities.

In this study, we also investigated the anti-inflammatory mechanisms of KAMPs which control bacteria-induced cytokine secretions from neutrophils and macrophages, the cells that make up the first wave of immune infiltrates. KAMPs were found to intervene both the cell surface and the endosomal signaling of TLR2 and TLR4. Instead of direct binding to and neutralization of exogenous bacterial ligands, which is the most common mechanism shared by anti-inflammatory AMPs, KAMPs act as receptor antagonists to simultaneously block multiple cell surface receptors for LPS and LTA. Specifically, KAMPs competitively bind MD2, the obligatory co-receptor that directly interacts with LPS in the activation of TLR4, and TLR2, the most prominent LTA receptor in a heterodimeric complex with TLR1 or TLR6. KAMPs also bind CD14, the co-receptor that facilitates the transfer of these two bacterial ligands to their respective cell surface TLR complexes. In addition, KAMPs induce internalization of bacterial ligand-free TLR2 and TLR4, thereby reducing cell surface availability of these receptors before bacterial activation. These findings help explain the suppressive effects of KAMPs on LPS- and LTA-induced NFκB activation. Given that the physical interactions between CD14 and LPS/LTA are required for their induction of TLR2 and TLR4 endocytosis (33, 75), KAMPs’ competitive binding to CD14 can also explain the inhibition of LPS- and LTA-induced IRF3 activation.

There are limitations to this study. Although our in vitro data strongly support the notion that KAMPs bind to TLR2 and TLR4, future experiments employing biophysical and computational methods, such as cryo-EM and molecular dynamic simulation, are still needed to further determine the precise TLR binding regions for KAMPs and to predict stability of complexes under different environmental conditions. The latter is expected to provide useful insights for screening ligands based on predicted affinity and stability and for modeling atom-atom interactions over time within ligand-receptor complexes. In future work, we will conduct extensive molecular dynamic simulation studies to guide activity optimization and formulation development of KAMP peptides.

In conclusion, the multifaceted mechanisms of KAMPs enable effective prevention or reduction of leukocyte infiltration when these antibacterial peptides are given either prophylactically or therapeutically. Considering that most of the TLR-targeting agents currently in clinical trials are immune stimulatory agonists indicated as vaccine adjuvants or cancer therapies, and that the few immunosuppressive biologics among them are neutralizing antibodies or classic small molecules (76), short bactericidal peptides from human keratin 6a as TLR2/4 antagonists would represent unique drug candidates with potential to treat both infectious and non-infectious inflammation.

MATERIALS AND METHODS

Study design

The goal of this study was to investigate the independent anti-inflammatory capacity of keratin-derived antimicrobial peptides (KAMPs) and their potential application as a therapeutic agent for simultaneous control of infection and inflammation. Thioglycolate-elicited peritoneal neutrophils and macrophages, as well as the corneas of live mice, were challenged with LPS and LTA to demonstrate the anti-inflammatory function of KAMPs. In vitro assays using toll-like receptor (TLR)-conjugated latex beads and mouse immune cells were used to show the TLR-targeting activities of KAMPs against bacterial ligand binding. To test the efficacy of bifunctional KAMPs in prevention and treatment of bacterial keratitis, we employed a well-established in vivo mouse infection model with P. aeruginosa and S. aureus, two commonly found causative agents representing both Gram types. Each experiment was performed as independently as possible: as such, experiments were repeated on different days with different cell stocks, reagents (including different batches of KAMPs and bacterial ligands), and mouse litters. Data pooled from multiple independent experiments was analyzed. For all in vivo experiments, mice were randomly assigned to experimental groups and one cornea per mouse was treated as one biological sample. As guided by power analysis (G*power) for one-way ANOVA using estimated effect size from means, significance < 0.05 and power of 80%, minimum sample sizes to generate reproducible data were determined. Disease grading, data collection, and analysis were conducted by a team of experimenters in a blinded fashion.

Biosafety and ethics

All work was performed with materials, procedures and safety measures approved by the Cleveland Clinic Institutional Biosafety Committee (IBC protocol 1419) and the Institutional Animal Care and Use Committee (IACUC protocol 2020-2324). In vivo studies were conducted in compliance with the Public Health Service (PHS) Policy on Humane Care and Use of Laboratory Animals, and the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research. As such, one eye of each animal was manipulated. The Cleveland Clinic’s animal care and use program, including the housing and ABSL2 facilities, is fully accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care, International (AAALACi).

Synthetic peptides

KAMP10 (GGLSSVGGGS), KAMP18C (RAIGGGLSSVGGGSSTIK), and scrambled KAMP10 (SC10) control (GLSGVGSGSG) in HCL salt were custom synthesized by GenScript USA. LL-37 was purchased from Biotechne Tocris Bioscience. Net Peptide content (>70%), peptide purity (>95%) and peptide sequences were confirmed by amino acid analysis, HPLC and mass spectrometry respectively. Stock solutions (4.8 mg net peptide/ml) were prepared in sterile distilled water and stored at −20°C. Aliquots were limited to one thaw prior to use.

In vitro stimulation of neutrophils and macrophages

To obtain thioglycolate-elicited peritoneal cells, C57BL/6J mice (Jackson Laboratory) were intraperitoneally injected with 1 ml of sterile 3% Brewer thioglycolate medium (Sigma-Aldrich). Elicited peritoneal neutrophils and macrophages were collected 1 day and 3 days after injection, respectively. Mice were euthanized immediately prior to the collection of neutrophils and macrophages. Peritoneal cells were collected by injecting 10 ml of sterile PBS into the peritoneal cavity, followed by gently massaging the peritoneum, then collecting with a 18g needle/syringe. Cells in PBS were pelleted and resuspended in red blood cell lysis buffer for 10 min. Primary neutrophils and macrophages were resuspended in Dulbecco’s Modified Eagle Medium and RPMI respectively (both containing 2% FBS) and placed in 96-well plate (1x10⁵ cells in 100 μl/well). Macrophages were additionally cultured at 37°C for 6 h to select adherent cells for subsequent treatments. Both neutrophils and macrophages were incubated with LPS or LTA (500 ng/ml; Sigma-Aldrich), or high-molecular-weight poly(I:C) (1 μg/ml; Invivogen), or KAMPs (50, 100 or 200 μg/ml), alone or in combination (15 min, 1 h, 8 h, or 24 h of co-incubation) at 37°C/5% CO₂. KAMPs were added 30 min either before or after bacterial ligand stimulation. In some experiments, low-molecular weight poly(I:C) (100 ng; Invivogen) and KAMPs (20 μg) were co-transfected to cells by Lipofectamine 3000 (Invitrogen). Cytokines in cell culture supernatants were measured by ELISA (DuoSet; R&D). Cells were analyzed by flow cytometry for NFκB and IRF3 phosphorylation.

Murine models of corneal wound healing, sterile inflammation, and bacterial infection

Both female and male C57BL/6J mice (Jackson Laboratory) at the age of 10-12 weeks were used. Induction of anesthesia was achieved by one intraperitoneal injection of ketamine (50 mg/kg BW) and dexmedetomidine (0.375 mg/kg BW) cocktail, which was reversed by atipamezole (3.75 mg/kg BW).

In the corneal wound healing model, the epithelium (2 mm diameter) at the central cornea was gently removed using a Beaver6400 mini blade (Beaver Visitec). Wounds were treated topically with saline or KAMPs (2.5 μg peptide) three times within 24 h. Fluorescein sodium 0.38% was used to stain compromised barrier immediately, 6 h, and 24 h after injury. The stained area was measured by ImageJ.

In the inflammation models, three parallel abrasions were made on the cornea using a 26g needle to compromise the epithelial barrier before inoculation with inflammatory or infectious agents. Scarified corneas were topically applied once with 2 μl saline (scratch control), purified P. aeruginosa LPS/S. aureus LTA (20 μg; Sigma-Aldrich), or heat-killed P. aeruginosa/S. aureus to induce sterile inflammation. These mice were topically applied once (5 μl with saline (sham treatment) or KAMPs 30 min either before (100 μg/ml) or after (200 μg/ml) challenge. In the infection model, live P. aeruginosa (10⁵ CFU in 5 μl saline) or S. aureus (10⁶ CFU) was inoculated to scarified corneas. As infection prophylaxis, 5 μl of KAMPs or LL-37 (0.1% in saline) was topically applied once 30 min before bacterial inoculation. As treatment, a liposome kit with lyophilized lipid powder containing L-α-phosphatidylcholine (63 μmoles), stearylamine (18 μmoles) and cholesterol (9 μmoles) (Sigma Aldrich) was reconstituted in 300 μl of saline, following by gentle mixing with 700 μl of saline (vehicle control) or KAMPs (1.5 mg/ml in saline). The resulting mixture (5 μg peptide in 5 μl was topically applied to infected corneas three times daily for 1 or 3 day(s) beginning at 24 h post-infection. Infected corneas were examined daily under a stereomicroscope equipped with digital camera to monitor disease progression. A five-point grading system (0-4 points) was used in blinded fashion to assess each of the four disease characteristics, including area of opacity, density of central opacity, density of peripheral opacity, and quality of the epithelial surface (77). The sum of four scores represented a single disease score (0-16 points).

At experimental endpoint, whole corneas were dissected with a trephine and a surgical micro-scissor, then homogenized in sterile PBS (150 μl each) using Precellys CK14 lysing kit and tissue homogenizer (Bertin Instruments). After centrifugation to remove debris, bacterial loads (CFUs) were quantified by serial dilution and plating on tryptic soy agar, and cytokines were measured by ELISA (DuoSet, R&D Systems) in the half area 96-well microplate format (Corning Costar). For flow cytometric analysis of immune cells, a single mouse cornea was incubated in type I collagenase (Sigma Aldrich) at 82 U/cornea for 2 hours at 37°C, then passed through a 30 μm filter once to prepare single cell suspension.

Statistical analysis

GraphPad Prism (version 9) was used to perform statistical tests. Normal distribution of data was determined by Shapiro-Wilk test or D’Agostino & Pearson test. Equality of variances was determined by Brown-Forsythe test. Outliers were detected by Grubb’s test. Three or more groups of data with normal distribution and equal variances were compared by one-way ANOVA with Dunnett’s post hoc test (Fig. 1; fig. 2, A, B and D; fig. 3; fig. 4; fig. 5, B and G; fig. 7, G-1DAT; fig. S1A; fig. S2; fig. S3; fig. S5), or two-way ANOVA with Dunnett’s (against vehicle) or Sidak (against T0) post hoc test (Fig. 6, A and B; fig. 7, A and B; fig. S1E). Normally distributed data with unequal variances were compared by Welch and Brown-Forsythe ANOVA with Dunnett’s post hoc test (Fig. 2C; fig. 5, C and H; fig. 6, E-1DAT and G; fig. 7, C-, D-, E-1DAT, and G-3DAT). To compare multiple groups of non-normal data, Kruskal-Wallis ANOVA with Dunn’s post hoc test was used (Fig. 5, D and I; fig. 6, C, D, and E-3 DAT; fig. 7, C-, D- and E-3DAT). A significance threshold of P < 0.05 was considered statistically significant.

Data and materials availability:

All data associated with this study are in the paper or supplementary materials. Sources of commercially available reagents are indicated in Materials and Methods section. PA3346 and samples of KAMPs are available upon request after completion and approval of a material transfer agreement.