Evaluation of peptide-based probes towards in vivo diagnostic imaging of bacterial biofilm-associated infections

Abstract

The clinical management of bacterial biofilm infections represents an enormous challenge in today’s healthcare setting. The NIH estimates that 65% of bacterial infections are biofilm related and therapeutic outcomes are positively correlated with early intervention. Currently, there is no reliable imaging technique to detect biofilm infections in vivo and current clinical protocols for accurate and direct biofilm identification are non-existent. In orthopedic implant-associated biofilm infections, for example, current detection methods are based on non-specific X-ray or radiolabeled white blood cell imaging, coupled with peri-prosthetic tissue or fluid samples taken invasively and must be cultured. This approach is time consuming and often fails to detect biofilm bacteria due to sampling errors and lack of sensitivity. The ability to quantify bacterial biofilms by real-time, non-invasive imaging is an urgent, unmet clinical need that would revolutionize the management and treatment of these devastating types of infections. In the present study, we assembled a collection of fluorescently labeled peptide candidates to specifically explore their biofilm targeting properties. We evaluated these fluorescently labeled peptides using various in vitro assays for their ability to specifically and non-destructively target biofilms produced by the model bacterial pathogen Pseudomonas aeruginosa. The lead candidate that emerged, 4Iphf-HN17, demonstrated rapid biofilm labeling kinetics, a lack of bactericidal activity, and biofilm targeting specificity in human cell infection models. In vivo, fluorescently labeled 4Iphf-HN17 showed enhanced accumulation in biofilm-infected wounds, thus warranting further study.

The classic picture of infection is that of acute infection where bacteria exist as independent cells that proliferate in the host. A new picture has now emerged where invading bacteria organize into multicellular assemblies known as biofilms, causing untreatable, chronic infections ¹. The National Institutes of Health estimates that 65% of bacterial infections are biofilm related ^2-4, with sites such as wounds, catheters, implanted devices, and the lungs of cystic fibrosis patients notoriously affected ⁵. Biofilms are composed of extracellular polymeric substances (EPS), mainly polysaccharides, extracellular DNA, proteins, nucleic acids, and lipids, immediately surrounding the microorganisms embedded inside ⁶. These structures afford bacterial cells an enhanced ability to survive antimicrobial agents, evade clearance by the host immune system, and disseminate to other sites of the body ¹.

The treatment of biofilms is an important medical challenge because therapeutic outcomes are positively correlated with early intervention, yet detecting their presence in the body is difficult. Currently, there is no reliable imaging technique to detect biofilms in vivo. In orthopedic implant-associated biofilm infections, for example, current detection methods are based on non-specific X-ray or radiolabeled white blood cell imaging, coupled with peri-prosthetic tissue or fluid samples taken invasively and must be cultured. This approach is time consuming and often fails to detect biofilm bacteria due to sampling errors ^7-8. The ability to quantify bacterial biofilm burden by real-time, non-invasive imaging is an urgent, unmet clinical need that would revolutionize the management and treatment of these devastating types of infections.

A potential diagnostic strategy for the specific detection of bacterial biofilms is the utilization of targeted probes that can be localized in the body by medical imaging (Figure 1). Current infection-specific probes that are in the development pipeline such as the PET tracer ¹⁸F-fluorodeoxysorbitol ⁹ are based on targeting acute infections by mechanisms that do not translate to chronic, biofilm infections. Antibiotic based imaging probes are also being considered ^10-11, but these probes may be restricted to one class of bacteria and their ability to penetrate into biofilms and engage their cellular target is unproven ¹². The challenge of delivering a large quantity of imaging probe to biofilm-associated infections, which are sessile aggregates of bacteria, requires new breakthroughs.

Figure 1.

Specific detection of bacterial biofilms by targeted probes that can be localized in the body by imaging. Following probe administration, non-invasive in vivo imaging allows the infection site to be identified and localized. Cartoon on the right shows the labeled probe recognizing and accumulating in biofilms surface-attached to eukaryotic cells.

An ideal probe for imaging bacterial biofilms would possess several key properties. First, the probe would ideally display non-destructive biofilm targeting in order to avoid the induction of bacterial dispersal or hinder probe accumulation due to biomass loss. Second, the probe would have the ability to penetrate into biofilms, potentially allowing for increased accumulation and amplification of imaging signal. Third, given estimates that the EPS makes up approximately 80% of the biofilm ¹³ and that the embedded cells are metabolically quiescent ¹⁴, the probe would ideally recognize and bind the biofilm matrix and potentially also the embedded cells, with no requirement for microbial metabolic activity.

In this work we focused on peptides as targeting vectors. Inspired by the discovery of naturally occurring antimicrobial peptides (AMPs) found in humans and other organisms, there is now a long history of efforts in developing peptides into potent antimicrobial agents for therapeutic and imaging purposes ^15-18. AMPs are a heterogeneous class of compounds that generally conform to three different classes, α-helical, β-sheet, and flexible peptides, with α-helical peptides being the dominant type studied ¹⁹. In aqueous solutions α-helical peptides are typically unstructured but assume their amphipathic α-helical conformations when associated with a cell membrane ^20-21. The positively charged 12 amino acid fragment of the natural α-helical AMP ubiquicidin (UBI_29-41) has been investigated as an infection imaging agent ²². Other AMPs studied as infection imaging agents include a depsidomycin-derived compound ²³ and human neutrophil-derived peptides ^24-25. While these probes have been studied in the context of various clinically relevant bacterial pathogens, if and how they target biofilm-associated infections is an important and understudied question. Given the prevalence of biofilm-associated infections clinically, it is important to understand how infection imaging probes interact with biofilm.

A far greater emphasis on targeting biofilm can be found in peptide therapeutic drug development. In this context, several groups have focused on AMPs, both naturally occurring and synthetic, as therapeutic agents aimed at preventing or eliminating established biofilms ^26-27. For example, some peptides interfere with the early events of biofilm formation by preventing adhesion of bacterial cells ²⁸ or interfere with bacterial communication signals that stimulate biofilm formation ²⁹. Others interfere with the synthesis and/or accumulation of biofilm EPS ^30-31. Given their success in preventing biofilm formation in vitro and reducing established biofilm infections in in vivo infection models ³², we hypothesized that one or more peptides of this type might effectively target bacterial biofilms and do so at lower, sub-biocidal concentrations required for in vivo imaging.

In the present study, we assembled a collection of fluorescently labeled peptide candidates to specifically explore their biofilm targeting properties. The peptides studied (Table 1) include UBI_29-41 as well as an amino acid scrambled version of UBI_29-41 (UBI_sc). Compared to naturally occurring AMPs, synthetic AMPs can provide superior properties such as reduced host cell toxicity and susceptibility to inactivation in high salt conditions ³³. Given the positive correlation between peptide length and antibacterial activity ^34-35, we chose to study the 8 residue peptides composed of Trp and Arg or Trp and Lys, referred to as WR8 and WK8, respectively, as proxy peptides for this class. We reasoned that the 8 residue peptides might achieve a balance where they attract to bacterial membranes without exerting enough membrane destabilization to kill cells. Given that short, lipidated peptides can achieve a unique localization in biofilms ^36-37, we studied a lipophile-enhanced peptide inspired by a previously discovered peptide identified from a phage display assay ³⁸ referred to as 4Iphf-HN17. A 21 amino acid sequence known as LG₂₁ previously identified to target a specific exopolysaccharide of Pseudomonas aeruginosa biofilm EPS ³⁹ was studied, as well as the natural AMP LL37 to serve as a reference peptide. We evaluated this focused collection of peptides using various in vitro assays for their ability to specifically and non-destructively target P. aeruginosa biofilms, used as a model biofilm forming pathogen in this study. The lead candidate that emerged from these studies was further evaluated in vivo in a mouse wound model.

Table 1 –

Fluorescently labeled peptides studied with their primary sequences and observed molecular masses in Daltons.

Amino Acid Sequence	Name	Molecular wt (observed)	Reference
aFITC-WWRRWWR	WR8-FITC	1,890.4	[28]
aFITC-WWKKWWKK	WK8-FITC	1,778.4	[28]
aFITC-WPRRPWRR	WR8-Pro-FITC	1,712.4	N/A
aFITC-LLPIVGNLLKSLLGWKRKRFG	LG21-FITC	2,911.8	[34]
(4Iph)(f)LPNSNHIKQGL-FITCb	4Iphf-HN17-FITC	2,104.1	[33]
(4Iph)(f)LPNSNHIKQGL-Cy5b	4Iphf-HN17-Cy5	2,338.9	[33]
aFITC-TGRAKRRMQYNRR	UBI29-41-FITC	2,195.2	[56, 60]
aFITC-KRNQRMARYRRGT	UBIsc-FITC	2,195.2	N/A
aFITC-LLGDFFRKSKEKIGKEFKRIVQRIKDFLRNLVPRTES	LL37-FITC	4,995.9	[64]
Cy5c-G-4Abs-F*QWAVGHΔL	Bombesin-Cy5	1,913.9	[42]

^adenotes FITC conjugation to the N-terminus via amino hexanoic acid linker

^*Denotes d-configuration, Δ denotes (3s, 4s) 4-amino-3-hydroxy-6-methylheptanoic acid, f=Fmoc, 4Iph=4-iodophenyl

^bdenotes dye conjugates were to the lysine

^cdenotes dye conjugation was to the glycine

EXPERIMENTAL DESIGN AND METHODS

Synthesis and labeling of peptides

All peptides were synthesized via solid phase 9-fluorenylmethoxycarbonyl (Fmoc) chemistry and procured from Genscript Co. (Piscataway, NJ), with the exception of 4Iphf-HN17 and Bombesin which were synthesized in house (Table 1). The fluorescent dye fluorescein-5,6-isothiocyanate (FITC) was conjugated to all the peptides at the N-terminus via Ahex chemistry with the exception of 4Iphf-HN17 which was dye-conjugated via the lysine. Fractions with >90% purity were pooled via preparative high-performance liquid chromatography (HPLC, Shimadzu, LC-8A) and analyzed using mass spectral analysis. Sulfo-Cyanine5 NHS ester (Lumiprobe Corporation) was substituted for FITC for lead peptide evaluation in vivo to enhance detectability. Lyophilized peptides were sequentially dissolved in DMSO and water to a final concentration of 1 mM, aliquoted, and stored at −20°C until use.

Bacterial growth inhibition studies

Antimicrobial activities of the peptide candidates were tested in vitro via growth inhibition assays. Cultures of P. aeruginosa (strain PAO1) were prepared as follows. An overnight culture of P. aeruginosa was diluted 1:100 in fresh Luria–Bertani (LB) without NaCl (LBNS) and incubated at 37°C and 5% CO₂ with shaking at 200 rpm. Once the culture reached mid-log phase (OD₆₀₀=0.5), the bacteria were harvested by centrifugation at 18,000xg for 5 min and the pellets resuspended in fresh Mueller-Hinton broth (MHB) to an OD₆₀₀ of 0.1 (approximately 10⁷ CFU/mL). 180 μL of the culture was pipetted into a 96 clear bottom well plate. Bacteria were treated, in triplicate, with serially diluted peptides at final concentrations of 10, 5, or 1 μM. Gentamycin (10 μg/mL) was included as a positive killing control and PBS-treated wells were included as a mock treatment group. Immediately after peptide and antibiotic addition, the OD₆₀₀ was measured using a Spectramax i3 plate reader (Molecular Devices LLC) every 30 min for 16 h, with a 5 sec shaking period prior to each read. The plate was maintained at 37°C and 5% CO₂. Peptide-mediated growth inhibition was calculated by normalizing mean, background-corrected OD₆₀₀ values to OD₆₀₀ values of PBS-treated wells and presented as a percentage.

Peptide targeting flow-grown P. aeruginosa biofilms

Biofilm labeling by the peptide candidates was assessed by confocal laser scanning microscopy (CLSM). Biofilms of P. aeruginosa were generated in flow cells as previously described ⁴⁰. Briefly, overnight cultures of P. aeruginosa PAO1 were normalized to OD₆₀₀ 0.05 and 150 μL transferred to uncoated μ-Slide VI^0.4 flow cell chambers (catalog no. 80601; IBIDI GmbH). Bacteria were statically incubated at room temperature for 1 h to allow adherence prior to initiation of flow of 5% (vol/vol) LBNS in water. Flow was then initiated and maintained through the channels for 24 h at a flow rate of 0.15 mL/min. Channels were washed twice with PBS and treated with 2 μM peptide solutions in PBS, statically for 1 h at room temperature in the dark. Channels were washed twice with PBS and biofilms were stained with 10 μg/mL FM4-64 membrane dye (Molecular Probes, Invitrogen) for 20 min at room temperature. FM4-64 is a red-fluorescent membrane localizing dye that rapidly enables bacterial visualization in biofilms. Biofilms were washed twice in PBS and fixed with 4% paraformaldehyde for 15 min. Fixative was removed and biofilms were maintained in PBS. Biofilms were imaged on an Olympus FV3000 confocal system using a 60x oil objective. The 488 nm and 543 nm lasers were used for FITC and FM4-64 excitation, respectively. Approximately 10-15 representative images were acquired for each peptide in two independent experiments. Biofilm structures were consistently imaged at mid-height using a single optical section.

Probe binding to P. aeruginosa cells

We used flow cytometry to directly assess binding of the fluorescently labeled peptide candidates to planktonic P. aeruginosa cells. An overnight culture of P. aeruginosa (PAO1) carrying a constitutive TdTomato-producing plasmid (pMQ400) ⁴¹ was diluted 1:100 in fresh Luria–Bertani (LB) without NaCl (LBNS) and incubated for 3 h at 37°C with shaking (200 rpm). At mid-log phase, bacteria were harvested by centrifugation at 18,000xg for 5 min. Pellets were resuspended in PBS and diluted to 10⁸ CFU/mL. 225 μL of the culture was transferred to microcentrifuge tubes and treated with 10, 5, or 1 μM of FITC labeled peptides for 30 min at room temperature. Bacteria were pelleted and washed with PBS. After a final centrifugation step, bacteria were fixed with 4% paraformaldehyde for 20 min, transferred to flow tubes, and analyzed on a Fortessa flow cytometer (BD Biosciences) with 10,000 events collected for each sample. FITC and TdTomato signals were analyzed using the 488 and 560 nm lasers on the Fortessa instrument, respectively. Two gates were defined for analysis: an initial gate to reject debris, based on forward and side scatter, and a second gate to reject events based on low TdTomato fluorescence intensity. FITC mean fluorescence intensity (MFI) of cells captured in the gates was measured and normalized to cell autofluorescence in the FITC channel. Two biological replicates were performed and combined results expressed as mean ± standard deviation.

Probe specificity to P. aeruginosa in an epithelial cell infection model

Human A549 type II alveolar epithelial cells (ATCC) were seeded into a 24 well plate (IBIDI GmbH) at a density of 20,000 cells per well. The cells were cultured in RPMI 1640 supplemented with 10% FBS and 100 U pen/strep for 3 days until the cells reached approximately 70% confluency. At this time cells were stained with 1 μM cell tracker blue (Life Technologies) diluted in warm RPMI 1640 without pen/strep for 20 min at 37°C. A culture of P. aeruginosa PAO1 carrying a constitutive mCherry-producing plasmid pMRP9 was grown overnight in LBNS supplemented with 50 μg/mL gentamicin. The next day the epithelial cells were washed with warm PBS and infected with a normalized culture of PAO1-mCherry at a multiplicity of infection of 50:1 for 2 h at 37°C statically. Epithelial cells were washed twice with warm PBS and stained with 2 μM of FITC-labeled peptides for 1 h at 37°C. Cells were washed 2 times with warm PBS and fixed with 4% paraformaldehyde for 30 min. Cells were maintained in 300 μL of PBS during image acquisition. CLSM was performed on an Olympus FV3000 confocal system using a 60x oil objective (N.A. 1.4) with an optical magnification of 1.5X. A minimum of 8 random fields of view were imaged for each probe for two independent experiments. Images were analyzed to estimate colocalization of the probes with P. aeruginosa and epithelial cells using MIPAR v2.2.5 (MIPAR, Worthington, OH). Our approach to quantify colocalization included a dilation of the bacterial aggregates to account for the non-fluorescent EPS component ^42-43 that appeared to be targeted by several probes (Supplementary Methods). Data presented as mean ± standard deviation of two biological replicates. The FITC dye used to label the peptides for the screening phase was substituted for Cy5 on the lead probe to facilitate more efficient detection in the follow on human neutrophil and in vivo studies given the longer wavelength emission of this dye. The retention of key properties of the Cy5 conjugated lead probe was confirmed ahead of in vivo studies. For comparison purposes, an additional irrelevant peptide of similar molecular weight and matched conjugated fluorophore was included as a control probe.

Probe specificity towards P. aeruginosa-infected human primary neutrophils

We used imaging flow cytometry to examine 4Iphf-HN17-Cy5 in the context of a human primary neutrophil infection model. Healthy control donors were recruited, and each donor gave written informed consent in accordance with a protocol approved by The Ohio State University Institutional Review Board #2009H0154. Neutrophils were purified from peripheral venous blood as described previously ⁴⁴. Briefly, blood was collected into heparinized tubes, overlaid with Ficoll-Hypaque (Amersham, Pittsburgh, PA), and centrifuged at 400 x g for 40 min at room temp. Following centrifugation, the neutrophil-rich pellets were fractionated by dextran sedimentation and resuspended in hypotonic solution to remove residual erythrocytes. Isotonicity was restored with 1.8% sodium chloride and tubes centrifuged at 233 x g for 3 min. Pellets containing neutrophils were resuspended in HBSS without calcium or magnesium, counted in the presence of trypan blue in a hemocytometer chamber, and adjusted to a concentration of 4×10⁶ cells/mL. The cells were kept on ice until ready for experimentation.

A mid-log phase culture of P. aeruginosa transformed to constitutively express green fluorescent protein (PAO1-GFP⁺) was diluted to 3x10⁸ cells/mL. Cells were pelleted, resuspended with 20% pooled normal human serum (Complement Technologies) diluted in PBS to opsonize the bacteria, and incubated for 20 min at 37°C with gentle rocking. Following opsonization, bacteria were washed twice in PBS and diluted to an OD₆₀₀ of 0.3. 4x10⁶ neutrophils were seeded into 15 mL conical tubes and infected with approximately 2x10⁷ serum-opsonized PAO1-GFP⁺ cells corresponding to a MOI of 5:1. Uninfected neutrophils were treated with equal volumes of PBS. Cells were incubated at 37°C for 30 min on a nutator, after which neutrophils were pelleted by centrifugation at 800×g for 5 min at 4°C and resuspended in HBSS. Neutrophils were incubated with a 2 μM solution of 4Iphf-HN17-Cy5 (or PBS as control) for 20 min at 37°C on a nutator. To mark non-viable neutrophils, zombie live/dead dye (Biolegend, San Diego, CA) was added to the tubes during the last 10 min of peptide staining at a 1:250 final dilution of the stock solution. Neutrophils were pelleted and washed twice in PBS. Neutrophils were fixed in 4% paraformaldehyde for 10 min, pelleted by centrifugation, and resuspended in 50 μL of PBS for imaging flow analysis.

Imaging flow cytometry was performed on an ImageStreamX Mark II (Amnis Corporation). Zombie live/dead (Channel 8), GFP (Channel 2), and Cy5 (Channel 11) fluorescence was recorded using the 405, 488, and 647 nm lasers, respectively. All laser intensities were kept constant for all samples analyzed. 60x brightfield images were collected in Channel 1 (camera 1) and Channel 9 (camera 2). A total of 10,000 events were collected for each sample. Two replicate experiments were conducted using cells from different donors. Data were analyzed using IDEAS v6.0 software (Amnis Corporation). Details of the cell gating scheme are reported in Supplementary Methods. In each experiment, single stained controls were analyzed using the same instrument settings to obtain a compensation matrix, applied to all files prior to analysis.

From the gated cells we specifically determined: 1) whether 4Iphf-HN17-Cy5 shows an increased association for the infected neutrophil population compared to the uninfected population, 2) whether the probe associates with morphologically irregular neutrophils considered a surrogate of cell activation regardless of infection status, and 3) whether 4Iphf-HN17-Cy5 colocalizes with the phagocytosed bacteria. To address the first two questions, neutrophil staining by 4Iphf-HN17-Cy5 was calculated as the mean fluorescence intensity (MFI) and assessed on 4 distinct populations: i) morphologically normal and uninfected neutrophils, ii) morphologically normal and infected neutrophils, iii) morphologically irregular and uninfected neutrophils, and iv) morphologically irregular and infected neutrophils. Note that infected neutrophils were defined as having 1-4 GFP⁺ "spots" using the built-in spot analysis algorithm and irregular vs. normal morphological populations were readily distinguished and empirically defined by differences in cell aspect ratio. Visual confirmation was performed on all sub-populations for quality control. Neutrophils possessing more than 4 spots were rare but were excluded from the analysis to avoid possible artifacts. Note that we did not attempt to distinguish bound from intracellular bacteria in this study. To address the third question, the similarity feature was used to measure the degree of colocalization between the punctate 4Iphf-HN17-Cy5 signals and GFP⁺ spots using all infected neutrophils regardless of morphology.

In vivo and ex vivo imaging of probe distribution in a mouse infected wound model

To determine if 4Iphf-HN17-Cy5 targets bacterial biofilms in vivo, we used a wound infection mouse model pre-approved by The Ohio State University Institutional Animal Care and Use Committee (IACUC) under the protocol #2017A00000033 ⁴⁵. Briefly, anesthetized female BALB/c athymic nude mice were administered buprenorphine analgesia subcutaneously and subjected to a 6 mm biopsy, creating two identical full-thickness dorsal wounds that were covered with an occlusive dressing (Opsite Flexfix). After a 48 h recovery period, the wounds were infected with a 60 μL inoculum of approximately 10⁷ cells consisting of a 50:50 mixture of lux-tagged and TdTomato-expressing P. aeruginosa. The lux reporter plasmid construct was introduced into PA01 by the mini-Tn7 vector to drive constitutive expression. The use of these two strains permitted dual modality bacterial detection in the wounds by bioluminescence and fluorescence imaging, respectively. Three groups of mice were studied. Mice in group 1 (n=4) were infected in both wounds and injected with 60 nmoles of 4Iphf-HN17-Cy5. Group 2 mice (n=3) were mock infected in both wounds with PBS and injected with 60 nmoles of 4Iphf-HN17-Cy5. Group 3 mice (n=2) were infected in both wounds and injected with sterile PBS. All injections occurred 24 h after the wounds were infected or PBS-treated and were administered through the tail vein. Live whole-animal bioluminescence (BL) and fluorescence (FL) imaging was performed on an IVIS Lumina II optical imaging system (PerkinElmer, Waltham, MA) to assess bacterial burden and image probe distribution. Mice were anesthetized using isoflurane gas for all the imaging acquisitions. BL images were acquired with the following parameters: integration time, 2 sec; binning, medium, field of view, D; and f/stop, 1. FL images were acquired using the following parameters: excitation filter, 640 nm; emission filter, open; exposure time, auto-exposure; binning, medium; field of view, D; f/stop, 8. Animals were imaged at 3, 6, and 18 h after probe injection. These times were chosen based on serum stability analyses and blood circulation studies reported previously ³⁸. Region of interest (ROI) analysis was performed on the images using Living Image 4.4 software (Caliper Life Sciences, Hopkinton, MA) to quantify lux and Cy5 wound signals relative to surrounding non-wounded areas. ROI-derived wound radiant efficiency was normalized to the radiant efficiency in a surrounding non-wound ROI and reported as a unitless wound to non-wound signal ratio. This analysis was performed for all mice across the three groups.

Following the last imaging time point, mice were euthanized by CO₂ inhalation. Wounds were harvested and stained with DAPI (1 μg/mL) for 10 min. Tissue was washed twice in PBS, fixed in 4% paraformaldehyde for 24 h, and embedded in OCT media for cryo-sectioning. CLSM of the sectioned tissue was performed on the Olympus Fluoview 3000 confocal microscope using a 60x oil objective (N.A. 1.4) . Images were processed using the Olympus Fluoview software.

Statistical analysis

All statistical analyses were performed using Prism version6.0 software (GraphPad Software, Inc.). One-way ANOVA followed by Dunnett’s or Tukey’s post hoc test was performed for comparisons between multiple groups. A two-tailed unpaired Student’s t-test was performed for comparisons between mouse groups in the in vivo imaging study. Linear regression was used to compare potentially correlated continuous variables. For all statistical tests, a p value less than 0.05 was considered statistically significant. Grouped data are expressed as mean ± SEM unless otherwise noted.

Results

Synthesis and labeling of peptides

The peptide analogs prepared in this study are shown in Table 1. The predicted molecular weights of the peptides were confirmed by mass spectrometry (MALDI). Chromatographic evaluation by reverse-phase HPLC confirmed >90% purity of the peptide conjugates.

In vitro antibacterial activity of FITC-labeled peptides

To measure potential bactericidal activity of the probes, 16 h growth curves were generated to track P. aeruginosa growth in the presence of the FITC-labeled peptides applied at 3 different concentrations. A significant decrease in normalized optical density (OD₆₀₀) was observed for cultures exposed to gentamicin (Fig 2). There was no growth inhibition observed for any peptides over the concentration range tested. 4Iphf-HN17-FITC showed a significant increase in normalized optical density at the highest concentration tested (Fig 2). Complete 16 h growth curves are shown in Fig. S1.

Figure 2.

Normalized OD₆₀₀ measurements of P. aeruginosa cultures following 16 h exposure to FITC-labeled peptides applied at 8 μM (A), 4 μM (B), and 2 μM (C). Data are presented as the fold change in OD₆₀₀ compared to PBS-treated P. aeruginosa cultures. The FITC-labeled peptide candidates do not affect growth of P. aeruginosa, with the exception of 4Iphf-HN17-FITC applied at applied at the highest concentration (8 μM). The positive control gentamicin significantly inhibited growth as expected. Statistical significance was determined by one-way ANOVA followed by Dunnett’s multiple-comparison test. Data derive from two independent experiments and are presented as mean ± SEM. **p < 0.0001 and *p < 0.01.

FITC-peptide labeling of P. aeruginosa biofilms

We used confocal microscopy to screen the probes for biofilm targeting. We generated and imaged flow cell-grown P. aeruginosa biofilms following treatment with the FITC-labeled peptides. As shown in Fig. 3, there was wide variability in biofilm labeling among the probes. WR8-pro-FITC, UBI_29-41-FITC, and UBI_sc-FITC failed to label biofilms (Fig. 3) above the weak, non-specific signal. In contrast, biofilm labeling was observed for WR8-FITC, WK8-FITC, 4Iphf-HN17-FITC, LG₂₁-FITC, and LL37-FITC, though with distinctive labeling patterns. Specifically, LG₂₁-FITC, and LL37-FITC localized to the periphery of biofilms (arrow heads) while WR8-FITC, WK8-FITC, and 4Iphf-HN17-FITC penetrated into biofilms (arrows). To understand the specific property mediating bacterial targeting, we synthesized an altered version of WR8-FITC with second and third Trp amino acids in the sequence each substituted with Pro. We observed a complete lack of binding of WR8-Pro-FITC indicating that disrupting the α-helical arrangement is sufficient to abrogate binding, even while maintaining the same cationicity.

Figure 3.

Representative CLSM images showing FITC-peptide labeling of flow cell-generated P. aeruginosa biofilms. The biofilms were grown for 24 h and treated with FITC-peptides at 2 μM. The green and red channels show FITC-peptide and FM4-64 staining, respectively. The images show a single slice through the biofilms at mid-height of the structures. Scale bar = 20 μm. FITC-peptides that localized to the periphery of biofilms are indicated by arrow heads while FITC-peptides that penetrated biofilms are indicated by arrows.

As a means to screen the probes for targeting of single cells, which are associated clinically with bacteremia and acute infections, the binding of the FITC-probes to P. aeruginosa planktonic cells was measured by flow cytometry. The peptides that labeled biofilms in the previous assay also labeled planktonic P. aeruginosa cells in a dose-dependent manner (Fig. S2). Interestingly, the probes that localized to the periphery of biofilms, LG₂₁-FITC and LL37-FITC, showed the greatest targeting of planktonic cells. The probes that penetrated biofilm, WR8-FITC, WK8-FITC, 4Iphf-HN17-FITC, also labeled P. aeruginosa planktonic cells, but to a lesser degree than LG₂₁-FITC and LL37-FITC (Fig S2). Similar to the biofilm staining result, the peptides WR8-pro-FITC, UBI_29-41-FITC, and UBI_sc-FITC showed minimal targeting of P. aeruginosa cells.

Probe specificity towards P. aeruginosa in an epithelial cell infection model

Next, we determined whether the peptide candidates specifically target bacterial biofilms rather than human cells. This was addressed in an in vitro cell-based infection model. In this experiment, epithelial cells were infected with mCherry-expressing P. aeruginosa for 2 h, washed, and treated with FITC-labeled peptides. Following a final wash, the cells were fixed and imaged to assess peptide distribution. The majority of P. aeruginosa aggregates appeared attached to the epithelial cells, with few aggregates observed that were not epithelial cell associated (Fig 4). It is of note that several groups report these epithelial cell-attached bacterial aggregates as having biofilm properties with respect to growth rates and resistance to antibiotics ⁴⁶. We therefore use the term aggregates interchangeably with biofilms in this context. Image analysis estimated the colocalization of FITC-labeled probes with epithelial cells and P. aeruginosa aggregates. Images of the following FITC-labeled peptides were observed as having 0-1% of pixel intensities above the background threshold set to exclude autofluorescence and therefore did not meet inclusion criteria for further analysis: LL37-FITC, WR8-pro-FITC, UBI_29-41-FITC, and UBI_sc-FITC. The probes 4Iphf-HN17-FITC, LG₂₁-FITC, and WK8-FITC colocalized with P. aeruginosa aggregates, yielding percent area overlap values of 70.3% ± 8.7%, 61.2% ± 7.4%, and 56.6% ± 6.9%, respectively (Figure 5A). With respect to epithelial cell targeting, 4Iphf-HN17-FITC and LG₂₁-FITC had the lowest colocalization percentages at 13.1% ± 7.3%, and 9.7% ± 2.4%, respectively (Figure 5B). WK8-FITC colocalized with epithelial cells to a slightly higher degree (14.8% ± 4.9%). WR8-FITC did not significantly colocalize with bacterial aggregates (8.9% ± 4.2% area overlap) and showed the highest epithelial cell colocalization among the probes (63.5% ± 3.6%). Notably, 4Iphf-HN17-FITC and LG₂₁-FITC did not target epithelial cells in the absence of infection while WK8-FITC and WR8-FITC showed moderate targeting of these cells (Fig. S8).

Figure 4.

Representative CLSM images used for evaluating FITC-peptide distribution in a co-culture model of epithelial cells with mCherry-expressing P. aeruginosa. 4Iphf-HN17-FITC and LG₂₁-FITC demonstrated selectivity for bacteria while other probes showed minimal bacterial targeting or showed targeting to both bacteria and epithelial cells. Each panel shows a representative image of at least two independent experiments for each peptide, with >8 images acquired per experiment. Epithelial A549 cells were prelabeled with cell tracker blue (CTB) for visualization. Scale bar = 20 μm.

Figure 5.

CLSM image colocalization analysis of select FITC-labeled peptides. Probe colocalization is reported as the % colocalization with epithelial cells (A) or bacteria (B) normalized by total detected pixel area of the probe. Data shown are combined from two separate experiments and error bars represent SEM. FITC-probes that showed no detectable signal above a background threshold (LL37-FITC, WR8-pro-FITC, UBI_29-41-FITC, and UBI_sc-FITC) were excluded from analysis. *p < 0.0003 and **p = 0.001.

The cumulative properties of P. aeruginosa biofilm targeting, lack of growth inhibition, and bacterial targeting specificity in an epithelial cell infection model led to the identification of 4Iphf-HN17-FITC and LG₂₁-FITC as candidates for further study. We chose to further advance 4Iphf-HN17 given its smaller size, clear ability to penetrate the interior of biofilms, and favorable serum stability previously reported ³⁸. To enhance detectability in tissue required for in vivo evaluation, we synthesized and tested 4Iphf-HN17-Cy5. Similar to 4Iphf-HN17-FITC, 4Iphf-HN17-Cy5 exerted no bactericidal effects against P. aeruginosa (Fig. S3) and retained similar biofilm targeting properties (Fig. S4), including biofilm specificity in the epithelial cell infection model (Fig. S5). A Cy5-labeled irrelevant control peptide known as Bombesin, previously studied as a cancer tracer ⁴⁷, showed no targeting in this model, ensuring that Cy5 was not mediating the bacteria-specific distribution pattern observed for 4Iphf-HN17-Cy5.

4Iphf-HN17-Cy5 specificity in a phagocyte cell infection model

Advancing on our findings in the epithelial infection model, we next sought to study 4Iphf-HN17 in the context of neutrophil infection. Neutrophils are generally the predominant cell type associated with sites of infection and even sterile inflammatory conditions. We used imaging flow cytometry to examine 4Iphf-HN17 interactions with human primary neutrophils infected or not with GFP-expressing P. aeruginosa. Neutrophils were incubated with 4Iphf-HN17-Cy5 following infection and analyzed on an imaging flow cytometry system. Representative images of neutrophils infected by PAO1-GFP⁺ in the absence of probe is shown in Figure 6A, verifying no detectable Cy5 fluorescence. Images of uninfected neutrophils incubated with the probe are shown in Figure 6B. Discrete punctate spots of Cy5 fluorescence are apparent in the majority of these uninfected neutrophils. Figure 6C shows infected neutrophils that were incubated with 4Iphf-HN17-Cy5. The punctate Cy5 signals within neutrophils occasionally colocalized with GFP signals but generally appeared non-overlapping. Quantification of the colocalization between Cy5 and GFP for every cell was <0.2, indicating poor spatial colocalization between the signals. Neutrophil viability was >85% and did not fluctuate between samples regardless of exposure to P. aeruginosa or probe. The percentage of neutrophils that were infected by 1 or more bacterium ranged from 41.9% to 48.6%, while the percentage of neutrophils that were not infected ranged from 51.4% to 58.1%. Specifically, the percentage of neutrophils that were uninfected, harbored 1-4 PAO1-GFP⁺ cells, or harbored 5 or more PAO1-GFP⁺ cells was 57.2%, 33.0%, and 9.8%, respectively.

Figure 6.

4Iphf-HN17-Cy5 preferentially labels P. aeruginosa-infected human neutrophils in vitro. Panels show representative images of (A) neutrophils infected with PAO1-GFP⁺ in the absence of probe, (B) neutrophils incubated with probe in the absence of infection, and (C) infected neutrophils incubated with 4Iphf-HN17-Cy5. Images are from one representative donor. Scale bar = 7 μm. (D) Normalized Cy5 mean fluorescence intensity (MFI) on four different neutrophil populations based on infection status and neutrophil morphology. Data shown are from two different donors and error bars represent the standard deviation of the means. *p = 0.008.

Probe uptake was quantified on 4 distinct neutrophil populations (Figure 6D). The cell population with the greatest Cy5 MFI, which is reported as a fold increase above cellular autofluorescence in the same channel, was infected neutrophils with irregular morphology (20.7 ± 5.9, *p = 0.008 compared to uninfected neutrophils with normal morphology). In this context we considered irregular morphology to be an indicator of viable but activated cells ⁴⁸. The population with the next highest Cy5 MFI was observed on infected neutrophils with normal morphology (18.0 ± 3.7). Among neutrophils that were uninfected, Cy5 MFI was approximately equal in the populations with irregular (12.9 ± 1.1) vs. normal (11.1 ± 1.8) morphology. The percentage of neutrophils that were Cy5⁺ was greater than 87% for each of the four populations, with the highest percentage associated with infected neutrophils with irregular morphology (99.8%) and lowest associated with uninfected cells with normal morphology (87.8%).

In vivo and ex vivo imaging of 4Iphf-HN17-Cy5 in a mouse infected wound model

We next sought to evaluate 4Iphf-HN17-Cy5 in vivo in a mouse wound infection model ⁴⁹. 48 h after the mice were subjected to the wounding procedure, the wounds were infected with approximately 10⁷ cells consisting of a 50:50 mixture of lux-tagged and TdTomato-expressing P. aeruginosa. 24 h following wound infection, the mice were administered 60 nmoles of 4Iphf-HN17-Cy5 via the tail vein. Optical imaging of the wounds was performed out to 18 h following probe administration. Wound Cy5 signals generally tracked with lux signals 18 h after injection, as shown in Fig. 7A for a representative wound-infected animal from group 1. There was a positive correlation between individual wound lux signals and wound to non-wound Cy5 fluorescence ratios for all mice in this group (R²=0.80, Fig. S6). In contrast, probe-injected mice with uninfected wounds (group 2) showed wound Cy5 signals that were near background levels (Fig 7B). Images of a representative mouse from group 3 show wound lux signals with no Cy5 signal as expected (Fig 7C). Of the nine mice that were wounded, two mice lacked detectable wound lux signals and were excluded from the study ahead of receiving probe. One mouse showed prominent lux signal only in one of the wounds but was included in the analysis. ROI analysis revealed a mean Cy5 fluorescent signal ratio in infected wounds, normalized to surrounding non-wounded tissue, of 3.2 compared to a ratio of 1.7 measured in PBS-treated wounds at 18 h post injection (Fig. 7D; p<0.001). The mean Cy5 fluorescent signal ratio measured in mice in group 3, which did not receive probe, was 0.85 ± 0.12. Cy5 signal enhancement in infected wounds, measured in the same manner, also reached statistical significance compared to PBS-treated wounds at 6 h post probe injection (p<0.0035, Fig. S7), though the difference was not as significant compared to the 18 h timepoint.

Figure 7.

In vivo optical imaging evaluation of 4Iphf-HN17-Cy5 in a mouse model of wound infection. Panels show bioluminescence and epi fluorescence images of representative mice from each group detecting lux-tagged P. aeruginosa and Cy5 distribution, respectively. (A) Mouse with P. aeruginosa-infected wounds injected with 4Iphf-HN17-Cy5, (B) mouse with PBS-treated wounds injected with 4Iphf-HN17-Cy5, and (C) mouse with lux-tagged P. aeruginosa-infected wounds injected with PBS. All images were acquired 18 h after probe injection. The min/max scale was set equally for all images. (D) ROI-derived Cy5 wound fluorescence normalized to surrounding non-wound tissue 18 h after injection for the three study groups. *p < 0.001.

Ex vivo CLSM imaging of harvested tissue enabled high-resolution visualization of 4Iphf-HN17-Cy5 in infected and uninfected wounds. As shown in the representative image in Fig. 8A, no TdTomato fluorescence and minimal Cy5 fluorescence was detected in sectioned tissue from an uninfected wound. In contrast, a representative image from an infected wound showed areas of TdTomato fluorescence appearing as single, rod-shaped cells and small multi-cellular aggregates (Fig 8B). Cy5 signals appeared as intense, multi-focal pockets that were generally localized near areas of TdTomato signals and occasionally resembled the shape of eukaryotic cells (Fig 8B, arrowheads).

Figure 8.

In situ CLSM images of 4Iphf-HN17-Cy5 distribution in wound tissue. Representative confocal micrographs of tissue sections of 4Iphf-HN17-Cy5 injected mice with (A) PBS-treated wounds and (B) P. aeruginosa-infected wounds. The images revealed a punctate localization pattern of 4Iphf-HN17-Cy5 in areas immediately adjacent to TdTomato-positive P. aeruginosa cells and multi-cellular aggregates. Arrowheads indicate cell-shaped localization patterns of Cy5 fluorescence near pockets of P. aeruginosa aggregates. Scale bar = 10 μm.

Discussion

The NIH estimates that at least two-thirds of bacterial infections are biofilm related ^{2, 4}. Unfortunately, there is no reliable medical imaging technique to detect biofilms in vivo and protocols for accurate and direct biofilm identification are non-existent, hindering early interventions that are crucial to therapeutic success. Current imaging approaches in infection are anatomical-based and thus non-specific or are based on targeting acute infections using mechanisms that do not translate to chronic, biofilm infections ^50-52. The use of modified quantum dots that robustly label biofilm EPS have been reported ^53-54; however, these probes have not yet demonstrated the appropriate specificity or suitability for in vivo applications. Other groups have explored the use of antibodies ⁵⁵, antibiotics ¹¹, siderophores ^56-57, bacteriophages ⁵⁸, and sugars ^{9, 22}. In the present work, we screened fluorescently labeled peptides to identify candidates suitable for biofilm imaging. The concept of our approach to target the biofilm matrix that makes up ~80% of biofilm volume ¹³ is innovative and may circumvent challenges encountered by other probes due to sparse concentration and/or metabolic inactivity of the target microbes. Of the two lead peptides that were identified in our screens, LG₂₁ and 4Iphf-HN17, both demonstrated selective targeting of biofilms with no bactericidal activity at relevant concentrations. Having demonstrated the unique ability to penetrate into biofilms for potentially enhanced signal amplification, we chose 4Iphf-HN17 for further development and in vivo testing.

CLSM imaging revealed that biofilm staining of 4Iphf-HN17-FITC was prominent and the pattern distinct from the other probes. Specifically, 4Iphf-HN17-FITC/Cy5 penetrated the interior of biofilms, appearing to associate with the biofilm matrix rather than the embedded cells. In the epithelial cell infection model, 4Iphf-HN17-FITC/Cy5 highly colocalized with aggregates of P. aeruginosa, while individual P. aeruginosa cells did not appear labeled. Using imaging flow cytometery, we observed punctate 4Iphf-HN17-Cy5 fluorescent signals in P. aeruginosa infected primary neutrophils, suggesting that the probe may target intracellular bacteria. In contrast to the other peptides studied, this peptide carries an overall slight negative charge (Isoelectric point = 8.26), making its targeting mechanism likely unique and not dependent on electrostatic interaction typically described for AMPs. In vivo, HN17-Cy5 localized to biofilm-infected wounds 3.2 times greater than surrounding normal tissue and showed significantly higher uptake in biofilm-infected wounds compared to uninfected wounds. An optical signal enhancement of 3.2-fold is considered excellent for discriminating diseased tissue from healthy tissue in cancer surgery applications ⁵⁹. Ex vivo CLSM revealed that compared to uninfected wounds, intense, multi-focal Cy5 signals were observed throughout the infected tissue samples. These Cy5 signals were almost always immediately adjacent to bacterial cells and aggregates, a pattern consistent with earlier observations of 4Iphf-HN17-FITC in the epithelial infection model. Based on prior observations showing functional bacterial targeting enhancements for short AMPs by the attachment of lipophilic molecules ^{37, 60}, our operating hypothesis is amphiphilicity and polarization of the hybrid peptide is responsible for biofilm targeting and accumulation. Future studies on 4Iphf-HN17 will focus on understanding targeting mechanism and pathogen range.

Our initial peptide screen included LL37, the first human amphipathic α-helical peptide isolated ⁶¹. Interestingly, LL37-FITC localized to the outer perimeter of flow-grown biofilms, yet showed no bacterial targeting in the epithelial cell infection model, perhaps due to differing matrix compositions in these environments ⁶². UBI_29-41 is an antimicrobial peptide fragment that has been investigated as an infection imaging probe with demonstrated affinity for the cell wall of bacterial ⁶³ and fungal pathogens ⁶⁴. In this work we observed no detectable binding of UBI_29-41-FITC to P. aeruginosa cells or biofilms, perhaps due to the use of FITC as the optical label. Previous studies have reported large discrepancies in UBI_29-41 binding depending on the choice of the conjugated dye ^{63, 65}. Chin, et al. reported on a peptide known as LG₂₁ that specifically interacts with Psl, a polysaccharide produced by P. aeruginosa ³⁹. The peripheral biofilm localization of LG₂₁-FITC we observed is consistent with Psl expression patterns previously shown for P. aeruginosa biofilms ⁶⁶. The bacterial specificity demonstrated by LG₂₁-FITC in the epithelial infection model makes this peptide a viable candidate for further study. Interestingly, while both WR8-FITC and WK8-FITC labeled flow cell-grown biofilms and planktonic bacteria, bacterial labeling in the epithelial infection model was divergent. WK8-FITC preferentially targeted bacterial aggregates while WR8-FITC preferentially targeted the epithelial cells. The mechanism underlying WK8-FITC bacterial specificity is currently unknown, though our data on WR8-Pro-FITC suggest that secondary α-helical rearrangement plays a critical role.

Conclusion

The clinical management of bacterial biofilm infections represents an enormous challenge in today’s healthcare setting. Currently, there is no reliable imaging technique to detect biofilms in vivo, hampering the opportunity for early treatment intervention of implant-related infections, for example, when more conservative treatment is possible. We identified two fluorescently labeled peptides that possess many desirable biofilm-targeting properties including specific and non-destructive targeting of the biofilm matrix. Due to its unique ability to penetrate biofilms, we selected the small, lipophile-enhanced peptide probe (4Iphf-HN17) for further study in vivo where it showed homing potential to sites of biofilm-infected wounds. 4Iphf-HN17 may, therefore, be a promising candidate as a biofilm-specific in vivo imaging agent that would change the diagnostic paradigm for at least 65% of clinical infections.