Assistant Professor Georgia Institute of Technology, United States
Introduction:: Respiratory infections are caused by a wide variety of pathogens, and pathogen identification is a crucial step for informed, targeted treatment. However, the current lead time to pathogen identification is days due to time-intensive clinical assays that require the microbe be grown up from patient samples [1]. To eliminate this time-intensive step, we are developing a diagnostic test that leverages existing pathogen growth inside the infected lungs. Together, the pathogen and host cells produce a repertoire of proteases in the infected tissue microenvironment that can be used for pathogen classification [2]. To harness pulmonary protease activity for a rapid disease readout, we have developed nanosensors that can be administered into the lungs and that will release volatile reporter payloads upon degradation by specific proteases (Fig. 1) [3]. Nanosensors are comprised of volatile reporter molecules tethered to a nanocarrier via protease-cleavable peptide linkers. Upon peptide cleavage, reporters are liberated, partition into the alveolar space in vapor form, and are subsequently exhaled for quantification in breath via mass spectrometry. For pathogen specificity, we are developing multiplexed sensing capabilities in which nanosensors designed to detect different proteases are barcoded with volatile reporters of distinct mass. Altogether, this approach provides a mechanism through which pathogen-specific breath biomarker signatures can be engineered.
Materials and Methods:: To demonstrate feasibility of pathogen identification, three mouse models of lung infection were established via intratracheal instillation of one of three bacteria responsible for hospital-acquired pneumonia – Staphylococcus aureus, Pseudomonas aeruginosa, or Klebsiella pneumoniae. To identify peptide sequences cleaved by infection-associated proteases, a fluorogenic peptide substrate library was screened using bronchoalveolar lavage fluid (BALF) collected from infected mice. Using principal component analysis and hierarchical cluster analysis, the peptide library was downselected to 10 peptides with classification power for the pathogens of interest and orthogonal protease sensitivity. Volatile reporters with distinct mass were conjugated to the C-terminus of each peptide. Volatile release kinetics and protease specificity were characterized using in vitro cleavage assays in which volatile-barcoded peptides were incubated with recombinant proteases in glass vials. Reaction headspaces were sampled using gas-tight syringes and analyzed using a proton transfer reaction-mass spectrometer (PTR-MS). Multiplexed protease sensing was assessed similarly.
Results, Conclusions, and Discussions:: Results & Discussion: Ex vivo cleavage assays combining BALF and the fluorogenic substrate library revealed a set of 10 peptide substrates that could be used for pathogen identification (Fig. 2). The 10-plex includes peptide substrates previously shown to be cleaved by either host or pathogen proteases, suggesting possible contribution from both sources for pathogen classification. In vitro cleavage assays confirmed that volatile-barcoded peptides generate protease concentration-dependent volatile reporter signal in the reaction headspace and are cleaved predominantly by their intended protease with minor cleavage by other proteases. Volatile-barcoded peptides were also successfully used in vitro for multiplexed protease sensing. Next steps include nanosensor synthesis by surface conjugation of volatile-barcoded peptides to 8-arm polyethylene glycol (PEG) nanocarriers, intrapulmonary administration of multiplexed nanosensors into mouse models of infection for breath studies, and development of a random forest classifier for pathogen identification based on measured reporter quantities exhaled in breath.
Conclusions: In summary, we have developed a mechanism for multiplexed sensing of lung infection-associate proteases, and we will assess the ability of nanosensors to generate pathogen-specific breath biomarkers in mouse models of infection. If successful, our approach could expedite pathogen identification for lung infections to shift from the current paradigm of empirical antibiotic treatment towards more informed, targeted treatment. This would reduce infection-related morbidities and mortalities and minimize the misuse of antibiotics.
Acknowledgements (Optional): : This work is supported by a K99/R00 Pathway to Independence Award (NIH EB028311).
References (Optional): : 1. Bouzid, D. et al. Rapid diagnostic tests for infectious diseases in the emergency department. Clin. Microbiol. Infect.27, 182–191 (2021).
2. Anahtar, M. et al. Host protease activity classifies pneumonia etiology. Proc. Natl. Acad. Sci. U. S. A.119, 1–12 (2022).
3. Chan, L. W. et al. Engineering synthetic breath biomarkers for respiratory disease. Nat. Nanotechnol.15, 792–800 (2020).