Patterson, John, authorGonzalez-Juarrero, Mercedes, advisorLyons, Mike, committee memberGhosh, Soham, committee member2024-05-272024-05-272024https://hdl.handle.net/10217/238449Targeted delivery of drugs to the lungs can improve TB chemotherapy and thus our goal is to develop TB-drug loaded nanoparticles tagged to pulmonary homing peptides. In a previous study, homing peptides to the lungs of TB diseased animals were identified using preclinical TB models (Balb/c and C3HeB/FeJ mice). The selection of homing peptides was carried out using a phage library containing peptides with known homing affinity in other diseases (e.g. cancer). Having identified and selected the homing peptide PL1 (PPRRGLIKLKTS) to granulomas present in the lungs of murine TB models, the PL1 peptide and a negative control (scrambled LinnTT peptide) were tagged to Fluorescein Amidites (FAM). To facilitate tracking in vivo of the nanoparticles to be loaded with TB drugs, silver nanoparticles (SNP) were conjugated to Cy3 fluorochrome, a fluorescent marker used in in vivo tracking studies, followed by functionalization with the PL1 homing peptide (PL1-SNP) or biotin as negative control (Ctrl-SNP). Tracking and homing of the PL1 peptide to granulomas was possible after in vivo administration via intraperitoneal (IP) or intravenous (IV) route of either the FAM tagged synthetic peptides or Cy3-SNPs to Mycobacterium tuberculosis (Mtb) infected C3HeB/FeJ mice. Visualization of the fluorescence-tagged carriers within the lungs was performed using microscopic slides affixed with lung sections from each mouse followed by whole slide imaging. The semi-quantitative analysis of the fluorescence whole slide images performed using the QuPath workflow confirmed that PL1-FAM, or PL1-SNP homed to the granulomas. Thereafter, a QuPath workflow was developed that uses machine learning approaches (MLP) for unbiased identification of tissue types. Other tools were used for characterization and quantification of FAM (synthetic peptides) and Cy3 (SNP) positive cells within granulomatous lesions of the C3HeB/FeJ TB mouse model. Moreover, it was important to quantify the penetration capacity of the FAM tagged peptide as well as the peptide coated SNP into granulomas. QuPath also includes a built in MLP pixel classifier for unbiased segmentation of the whole slide. In addition, a modified QuPath script was developed to segment the granulomas into concentric regions (outer, inner and center) followed by detection and quantification of positive cells for either fluorochrome within each region. Specific colocalization of PL1 with its known receptor (FN-EDB), either as a synthetic peptide or coupled to the SNP, was also studied using lung sections from mice treated with PL1-FAM or PL1-SNP and counter stained with Alexa 647 conjugated anti-FN-EDB monoclonal antibodies. The modified QuPath script was trained to quantify fluorescence from Alexa 647 in cells within granulomas and the Pearson coefficient and QuPath script was used to assess PL1 and FN-EDB colocalization within each region of the granuloma. The results demonstrated that when compared to their respective control samples, the IP route of administration provides equal or better homing of PL1 peptide to the granulomas than the IV route. Both the PL1-FAM and PL1-SNP home to the granulomas and specifically colocalize with its receptor FN-EDB. The FAM tagged peptide and SNP penetrate to the inner and center regions of the granuloma whereas the control SNP were unable to penetrate the barrier in the outer region of the granulomas. The QuPath workflow developed here can be used for tracking and quantification of other homing peptides and nanoparticles for development of new TB therapeutics.born digitalmasters thesesengCopyright and other restrictions may apply. User is responsible for compliance with all applicable laws. For information about copyright law, please see https://libguides.colostate.edu/copyright.nanoparticlesquantitative pathologymachine learningwhole slide imagingquantitative image analysisA QuPath workflow utilizing machine learning to analyze homing protein specificity and penetration into lung granulomas of Mycobacterium tuberculosis infected miceText