Enhanced avidity from a multivalent fluorescent antimicrobial peptide enables pathogen detection in a human lung model

Rapid in situ detection of pathogens coupled with high resolution imaging in the distal human lung has the potential to provide new insights and diagnostic utility in patients in whom pneumonia is suspected. We have previously described an antimicrobial peptide (AMP) Ubiquicidin (fragment UBI29–41) labelled with an environmentally sensitive fluorophore that optically detected bacteria in vitro but not ex vivo. Here, we describe further chemical development of this compound and demonstrate that altering the secondary structure of the AMP to generate a tri-branched dendrimeric scaffold provides enhanced signal in vitro and ex vivo and consequently allows the rapid detection of pathogens in situ in an explanted human lung. This compound (NBD-UBIdend) demonstrates bacterial labelling specificity for a broad panel of pathogenic bacteria and Aspergillus fumigatus. NBD-UBIdend demonstrated high signal-to-noise fluorescence amplification upon target engagement, did not label host mammalian cells and was non-toxic and chemically robust within the inflamed biological environment. Intrapulmonary delivery of NBD-UBIdend, coupled with optical endomicroscopy demonstrated real-time, in situ detection of bacteria in explanted whole human Cystic Fibrosis lungs.

Here, we describe further development of this compound, engineering a multivalent (tri-branched) form, while utilising a modified portion of the bacterial binding fragment (UBI [29][30][31][32][33][34][35][36][37][38][39][40][41] ) of the 59 amino acid sequence in which the metabolically labile methionine residue in replaced with a norleucine analogue. The arms of the trivalent scaffold were capped with the environmentally sensitive fluorophore, 7-nitrobenz-2-oxa-1,3-diazole (NBD) 16 that self-quenches any residual fluorescence within the aqueous environment and then generates excellent signal-to-noise ratios when dispersed in the distal human lung, with fluorescent amplification only upon entry into the hydrophobic environment of the bacterial membrane. When compared to its' monovalent equivalent, NBD-UBI dend selectively labelled a diverse panel of pathogenic bacteria and Aspergillus fumigatus, and remained chemically stable in the complex environment of the inflamed and injured lung. It detected bacteria in an ex vivo large animal model of bacterial burden and in infected explanted cystic fibrosis (CF) human lungs.

Results
NBD-UBI dend specifically labels bacteria and fungi with high signal-to-noise ratio over inflammatory cells. The multivalent peptide (NBD-UBI dend ) comprising three modified monomeric UBI peptides, coupled to the environmentally responsive fluorophore NBD (NBD-UBI) (Fig. 1A) was synthesised by solid-phase peptide synthesis (Fig. 1B).
NBD-UBI dend demonstrated fluorescent amplification in vitro with increasing environmental hydrophobicity (Fig. 1C). Bacterial labelling was demonstrated in a concentration dependent manner (Fig. 1D) and of a panel of clinically relevant lung pathogenic bacteria, including both Gram-positive and Gram-negative species (Fig. 1E) with varying labelling intensity (Fig. 1F).
Furthermore, to ensure the compound retains specify and does not label the inflammatory cell infiltrate observed with pneumonia, NBD-UBI dend was co-cultured with freshly isolated primary human neutrophils, monocytes, lymphocytes and human alveolar macrophages ( Fig. 2A-D) with no labelling seen. Confocal imaging was undertaken in the continued presence of the NBD-UBI dend demonstrating the fluorescent amplification in bacteria. To further demonstrate selectivity in human lung tissue, bacterial labelling was demonstrated   www.nature.com/scientificreports www.nature.com/scientificreports/ by confocal microscopy of human alveolar tissue 17 . Three-dimensional optical reconstructions confirmed bacterial-specific labelling within this complex environment (Fig. 2E).
NBD-UBI dend also labelled the pulmonary pathogenic fungus Aspergillus fumigatus in human lung tissue (Fig. 2F). The pattern of labelling of Aspergillus fumigatus, imaged with OEM ( Supplementary Fig. S1) was clearly discrete from bacteria, supporting the utility of NBD-UBI dend and OEM to discriminate bacteria from Aspergillus fumigatus.
NBD-UBI dend demonstrates high avidity, stability when compared to its linear fragment and is non-toxic. NBD-UBI dend demonstrated high avidity for bacteria as evidenced by retention of bacterial labelling despite washing steps in vitro. Compared to 15 μM of the analogous linear moiety (NBD-UBI), 5 μM of NBD-UBI dend retained bacterial labelling after washing (Fig. 3A). Hydrophobic phospholipoprotein components of pulmonary surfactant have the potential to generate off-target fluorescence. In order to investigate this, synthetic surfactant was co-cultured with bacteria and imaged. NBD-UBI dend labelled bacteria in the presence of synthetic surfactant with good signal-to-noise ratios (Fig. 3B,C). Chemical stability was assessed by incubating NBD-UBI dend and NBD-UBI with bronchoalveolar lavage fluid (BALF) from ARDS patients in ICU and performing mass spectroscopic analysis. NBD-UBI was rapidly degraded in ARDS BALF, while NBD-UBI dend remained intact (Fig. 3D). Furthermore, the multivalent peptide provided enhanced fluorescent labelling when compared to equimolar levels of the analogous monomeric construct NBD-UBI (Fig. 3E,F).
NBD-UBI dend demonstrated no overt biological toxicity, evidenced by absence of erythrocyte hemolysis, and no preclinical in vivo toxicity: murine intratracheal instillation of high concentrations (over 700 times greater than the final intended human pulmonary dosing), resulted in no pulmonary inflammatory cell recruitment, pulmonary toxicity or systemic toxicity at early and late time-points (Supplementary Fig. S2A-D).

NBD-UBI dend rapidly detects bacteria with topical microdosed endobronchial delivery and OEM
in a large animal ex vivo lung model. Using a previously reported model 13 we evaluated NBD-UBI dend against a diverse panel of clinically relevant pathogenic bacteria (Fig. 4). NBD-UBI dend was able to label bacterial segments, over control segments and the linear NBD-UBI compound within a clinically relevant limit of detection (LoD) of 1 × 10 5 CFU/mL bacteria retrieved on BALF (

NBD-UBI dend detects bacteria in situ in whole explanted human lungs.
To further support clinical utility, we performed bronchoscopy and delivery of NBD-UBI dend in freshly-explanted whole lungs from patients with cystic fibrosis (CF) undergoing transplantation (Fig. 5A). The CF lungs were extremely damaged with large amounts of mucopurulent secretions. Discrete distal bronchopulmonary segments were instilled with NBD-UBI dend or equimolar monomeric NBD-UBI (negative control). The characteristic bacterial signal of punctate twinkling speckles and also distinct and intense colonies were detected in the alveoli in the NBD-UBI dend instilled segments, but not the monomeric instilled segments (Fig. 5B,C and Video 5). BAL confirmed the presence of pathogenic bacteria in all segments of the CF lung (Fig. 5D). This confirmed the ability of NBD-UBI dend to rapidly detect bacteria in situ in the human lung using a microdosing regime and OEM.

Discussion
Immediate-point-of-care bedside methodologies to enable clinicians to determine the presence or absence of bacteria or pathogenic fungi in the distal lung are required. Clinicians are faced with significant uncertainty in relation to triggers to commence antibiotic treatment 7 , the choice of agents 18 and especially when to de-escalate therapy 19 . These issues are barriers to effective antibiotic stewardship because of the association between delayed and inadequate antibiotic therapy and adverse clinical outcomes 20 .
Optical imaging of bacteria is an emerging and attractive modality [21][22][23] and utilising OEM with localised microdosed delivery of SmartProbes in distal bronchopulmonary segments potentially offers an immediate method for bacterial detection. This approach shows particular promise for the future as it can be multiplexed using fiber based systems capable of multispectral imaging 24,25 , used in combination with other SmartProbes for Gram-negative bacteria 13 or inflammatory cells 26,27 and compounds can be delivered directly into the imaging field of view 28 . Furthermore, with ongoing refinements to the image analysis algorithms through multiple methods [29][30][31] , automated readouts of the signals generated may help decision making and advance our understanding of the pathophysiology during suspected pneumonia.
UBI [29][30][31][32][33][34][35][36][37][38][39][40][41] , in its monomeric form, has been applied in human infection imaging in the form of radionuclide imaging with technetium 32 . We demonstrated that this monomeric scaffold was rapidly degraded in the inflamed bronchoalveolar compartment, and have demonstrated that a monomeric ligand remains ineffective for bacterial labelling in assays using fragments of human lung tissue 14 . In contrast, the trivalent dendrimeric scaffold afforded enhanced proteolytic stability, partitioning in surfactant and high avidity in the whole human lung.
Although this compound demonstrates promise for clinical utility, it now requires clinical translation to determine utility alongside other SmartProbes in patients with suspected pneumonia.
greater fluorescence intensity at eqimolar dye concentrations. (F) Quantification of flow cytometry data for the monomeric form (white bars, normalised) and dendrimeric forms (grey bars) demonstrating between a 2-10 fold increase in fluorescence at eqimolar dye equivalents. Bars represent means (+/− SEM) from three independent experiments, analysis is by students t-test, **p < 0.01, ***p < 0.001. www.nature.com/scientificreports www.nature.com/scientificreports/ In summary, we have synthesized a multivalent fluorescent UBI analogue that demonstrates excellent optical properties for the detection by optical imaging platforms already in clinical use and with microdosed delivery selectively labels a range of pathogens in vitro and in an ex vivo ovine model. Importantly, we have also shown

Materials and Methods
Ethics statement. All experiments using human samples were performed following approval of the appro-  Fungal culture. Aspergillus fumigatus (clinical isolate) was stored in potato glucose agar (Fluka Analytical, Gillingham, UK) at 4 °C. A wet harvest of conidia was performed into 1 mL of potato dextrose broth (Fluka Analytical, Gillingham, UK) and conidia were counted on a haemocytometer. 1 × 10 6 conidia were added to 9 mL potato dextrose broth and grown overnight in a shaking incubator at 37 °C. Germinating hyphae were washed for confocal assays.
Neutrophil and mononuclear cell isolation. Neutrophils and mononuclear cells were isolated from the peripheral blood of healthy human volunteers using dextran sedimentation and discontinuous Percoll gradients, as previously described 33 .

MALDI-TOF and FTMS.
NBD-UBI or NBD-UBI dend were added to saline or pooled BALF from three patients with ARDS incubated for 30 min. ARDS BALF was retrieved from patients in ICU, as previously described 34  Haemolysis assay. Performed as previously decribed 13 . Briefly, erythrocytes (resuspended to 20 vol % in PBS) were cultured with varying concentrations of NBD-UBI dend, PBS (negative control) or 0.2% Triton X-100 (positive control) and incubated at 37 °C for 1 h. Wells were diluted with 150 µl of PBS, centrifuged and absorbance of supernatant read at 350 nm.
Single dose intra-tracheal rodent study. Adult male CD-1 mice aged 8-12 weeks were given direct intra-tracheal administration of a single dose of NBD-UBI dend (100 µg) in 50 µl PBS (equivalent to 300 µM) or PBS vehicle control. Animals were monitored and then sacrificed at 48 hours or 14 days (n = 3 per group per time point). Airway resistance was assessed via whole body plethysmography. Following necroscopy, bronchoalveolar lavage fluid (BALF) was harvested via 3 × 0.8 ml PBS flushes of the lung and cytospin slides were prepared and stained with Diff-Quick (Thermo Fisher Scientific). BALF cell types were quantified using a light microscope. Lung, liver and kidney were removed, fixed and paraffin-embedded. Slides were prepared and stained with hematoxylin and eosin.
Confocal imaging and analysis. Imaging and analysis was conducted as previously reported 13,14 . 'Green' fluorescence (for NBD) was excited with a dedicated 488 nm line, Syto nuclear acid dyes were excited with a dedicated 543 nm line and Cellvue Claret labelled surfactant were excited with a dedicated 633 nm line. For affinity assays, following initial imaging, the fluid was aspirated from wells and two gentle washes with PBS were performed. The chamber was re-imaged on benchtop confocal. Flow cytometry. NBD-UBI or NBD-UBI dend were incubated with bacteria for 5 minutes at 37 °C prior to analysis using BD FACSCalibur (Becton Dickenson, San Jose, CA, USA) flow cytometer capturing 50,000 of counterstained gated events. Analysis performed with FlowJo version 7.6.5 (TreeStar Inc., Ashland, OR).
Ovine lung model and image analysis. The ovine model and analysis algorithm have been reported previously 13 . Briefly, surplus stock animals, which were destined for cull, had their lungs removed, which were ventilated and kept at 37 °C, humidity of 65% and ventilated using a Pressure Controlled Ventilator (Vivo PV www.nature.com/scientificreports www.nature.com/scientificreports/ 403, Breas Medical, Sweden). Segments were bronchoscopically identified and bacteria (or PBS control) instilled. After >1 hour segments were re-identified with subsequent SmartProbe instillation and OEM imaging at 488 nm excitation (Cellvizio Lab, Mauna Kea Technologies) using a 1.5 mm diameter S-1500 fiber (Mauna Kea Technologies), followed by BALF retrieval. Frame-by-frame analysis was undertaken 13 after removal of redundant frames; frames were considered positive if there were >80 dots per frame. Data is presented as the proportion of frames over a video sequence containing >80 dots per frame.
Ex vivo cystic fibrosis human lung. Whole explant CF lungs were received fresh from theatre (<1 hour) and main the bronchi were identified. Bronchoscopy and suction of the airways was performed prior to instilling NBD-UBI/NBD-UBI dend into anatomically distinct distal bronchopulmonary segments. The bronchoscope was proximally wedged and the S-1500 fiber passed into distal segments. Following imaging in the distal lung (alveolar regions) as described above, a BAL was undertaken, sample was centrifuged at 200 g for 5 minutes to remove erythrocytes and cellular material and was plated for bacterial quantification.
Statistical analysis. All experiments were performed at least three times unless otherwise stated and results expressed as mean ± SEM. Data was analysed by Student's t-test or ANOVA, significance was determined as p < 0.05 (GraphPad Prism version 5.01 for Windows, GraphPad Software, San Diego California USA).

Data Availability
The data that support the findings of this study are presented in the published article and any additional data is available from the corresponding author upon reasonable request.