Deciphering Hypoxic Adaptations in Cutibacterium acnes using Multi-omics Profiling and Systems Modelling

Abstract

The opportunistic skin anaerobe Cutibacterium acnes (C. acnes) contributes to skin homeostasis with its antioxidant and immunomodulatory activities. C. acnes (formerly known as Propionibacterium acnes) is an aerotolerant, gram-positive anaerobe that grows inside the pilosebaceous units present on the skin. In acne, the orifices of these units get clogged by secretions from the sebaceous glands to form a perfect lipid-rich, hypoxic niche for the proliferation of lipophilic, anaerobic C. acnes. Elucidating the molecular mechanisms by which C. acnes switches between growth and dormancy will help in understanding its adaptation to different oxygen environments and ultimately finding strategies to kill the bacteria in the pathogenic niche. This is addressed by studying the in-vitro growth phenotype of C. acnes cultured in hypoxic (anaerobic) and normoxic (aerobic) conditions, profiling its transcriptome data, and further modeling the transcriptome-integrated networks and genome-scale metabolic models to identify oxygen-responsive expression patterns and adaptation mechanisms in C. acnes. This work aims to explore the genome-wide adjustments and survival strategies employed by C. acnes to remain viable in anaerobic conditions. To achieve this, first and foremost, sequence and structure-based genome annotation was carried out, and the latest findings were presented in a C. acnes-specific database, CAcnesDB. Chapter-2 describes the constructed annotation pipeline, the summary of the new findings, and the functionalities embedded into the constructed database that stores all the obtained results. The reference genome sequence of Cutibacterium acnes (C. acnes) KPA171202 was selected and subjected to gene calling and an initial draft of automated annotations was carried out. Based on the comparison with the existing ‘GenBank’ annotation, the genes were categorized into ‘Known /previously annotated’, ‘Putative/ poorly annotated’, and ‘Hypothetical/ un-known’ sets. A systematic, multi-level annotation pipeline was devised, which included both sequence- and structure-based querying and analysis of individual proteins. Thereupon, all the genes were assigned domains/signatures, functional descriptions, and KEGG pathways (if any). Next, using ColabFold (AlphaFold and MMSeq2), the entire C. acnes proteome was modeled. The quality of the modeled proteome was assessed following AlphaFold’s reported pLDDT score (per-residue local distance difference test). Thereafter, a high-confidence pocketome was extracted from the qualitative modeled proteome using a combination of three pocket-detection algorithms. The function of a protein can be correlated with the knowledge of binding metabolites to their corresponding pockets. Hence, a ligandome was constructed that corresponds to the C. acnes predicted metabolome (CApM). Overlaying the knowledge from the predicted ligands (the final outcome of the structure-based annotation pipeline) with the assigned functional descriptions (derived from the sequence-based annotation pipeline) significantly enriched the cues associated with a given annotation. To assess the assigned functional annotation, an evidence-based annotation-scoring schema was formulated. The summarized annotation scores for the entire C. acnes genome, underscore the extensive coverage of various aspects of annotations reported for individual proteins. By comparing with the existing annotation of C. acnes from global resources, namely, KEGG, UniProt, NCBI, and TBDB, the CAcnesDB has yielded up-to-date scoring-based genome annotation and modeled proteome, pocketome, and predicted metabolome. Thus, CAcnesDB holds the enriched annotations, modeled proteome, High-Confidence Pocketome, and the predicted C. acnes metabolome (CApM). C. acnes is an aerotolerant-to-anaerobe, and hence, in Chapter-3, this feature was utilized to study the anaerobic adaptations at phenotypic, genetic, and systems levels. The bacteria were cultured in anaerobic (hypoxia) and aerobic (normoxia) in-vitro lab conditions, and the growth curves were studied. Relatively stable and uniform growth in hypoxia indicates that C. acnes inherently adapts to the anaerobic environment. Further, to analyze the genetic level adjustments, RNA was isolated from the mid-log phase, stabilized, and high-throughput sequencing (RNAseq) was carried out. The obtained raw data was normalized for downstream analysis. The principal component analysis captured the majority of the variance (78.98%) that indicated transcriptionally distinctive features between hypoxia and normoxia samples. Genome-wide transcriptomic profiles of C. acnes in hypoxia and normoxia resulted in 407 differentially expressed genes (DEGs), with 200 upregulated (log2FoldChange ≥ 1.5, adjusted P.Value ≤ 0.05)and 207 downregulated genes (log2FoldChange ≤ −1.5, FDR ≤ 0.05). To obtain a systems perspective of the functional implications of these gene expression alterations, a transcriptome-integrated functional interactome was constructed. By using in-house developed network mining and path tracing algorithms, top-ranked perturbed paths were identified. Classifying the top-ranked perturbed response network based on the functional classes (from Chapter-2) resulted in a larger portion of activated genes belonging to ‘Intermediary Metabolism and Respiration’, followed by ‘Cell wall and Cell Processes’, and ‘Information Pathways’. On the contrary, the majority of the repressed genes belonged to ‘Information Pathways’, followed by ‘Stable RNAs’. Enrichment analysis for Gene Ontology (Biological Processes) and KEGG pathways was carried out using the annotations in CAcnesDB. Transport, localization, and vitamin metabolic processes were upregulated. Ribosomal activities, detoxification, and oxidative phosphorylation were downregulated. Analysis of the top-perturbed response network resulted in 342 genes, out of which 102 were upregulated DEGs, and 70 were downregulated DEGs. Several alterations were evident, such as activated porphyrin metabolism, increased anaerobic metabolism linked to the activated ornithine fermentation, and arginine catabolism in C. acnes. An increased activity in hypoxia in the pentose phosphate pathway was observed with a decrease in the expression in the lower parts of glycolysis and oxidative phosphorylation. Concomitantly, an increase was seen in the ABC transporter activity, indicating its ability to acquire essential nutrients and regulate metabolic pathways essential for survival and growth in an oxygen-depleted environment. To characterize this further, the constructed C. acnes predicted metabolome (CApM), identified from the small-molecule binding pockets of modeled 2280 proteins (Chapter-2), was used to map the metabolites. The ligandome of activated ABC transporters and metabolic transporters in hypoxia shows enhanced lipid activity, vitamin translocation, and possible metabolic exchange with the host skin. Additionally, increased transportation of squalene, testosterone, actinonin, and retinol were also identified. Put together, these investigations suggest the metabolic reprogramming and the molecular mechanisms that are altered in hypoxia in C. acnes. The identification of systems-level metabolic perturbations in Chapter-3 indicated a reprogrammed metabolism in C. acnes. To understand the purpose of this adapted response towards the anaerobic environment, an in-silico based modeling approach was opted to investigate the genome-wide metabolic adjustments. Chapter-4 explains the utilization of an extended Genome-Scale Metabolic model (GSM), the simulation strategy, and the profiled fluxes to study the cell’s response as a whole. Hence, to further study the metabolic alterations, a previously published model was extended by the addition of (1) one reaction catalyzed by Phosphoenolpyruvate kinase (PPCK), (2) adding a gene (zwf) to two reactions- Glucose-6-phosphate dehydrogenase and rxn01975. The extended model’s (iCA845, 845 metabolic genes, 1 spontaneous gene) stoichiometry consistency was checked using MEMOTE, and the transcriptome was integrated to make context-specific models: ‘hypoxia-iCA845’ and ‘normoxia-iCA845’. After FBA simulation (with biomass maximization) of the context-specific models, the fluxes (metabolite flow) in hypoxia were analyzed. The magnitude [∆F lux = Flux(Hypoxia-iCA845)-Flux(normoxia-iCA845)] and direction of the flux were analyzed to study the metabolism of C. acnes in hypoxia. An overall increase in flux through the Pentose Phosphate Pathway, lower parts of the glycolysis, along with alternate carbon metabolism, was observed. The flux enters the reverse TCA cycle and supports anaerobic respiration (activation of fumarate reductase) in hypoxia. An increase in flux through the Wood-Werkman cycle/succinate pathway indicated an increase in propionate production. Propionyl-coa, which is an intermediate in the Wood-Werkman cycle, is a precursor for odd-chain fatty acid synthesis. Many of the reactions specific to anaerobic metabolism were identified to be engaged. Overall, the study showed the coordinated set of metabolic adjustments carried out by the C. acnes. In summary, this thesis achieves multi-omics data acquisition and presents it in an accessible biological database- CAcnesDB. The ‘dry’-lab generated multi-omics data, such as structural-proteome, pocketome, ligandome, and ‘wet’-lab generated transcriptome data, are computationally integrated at multiple stages. It also demonstrates the applicability of integrated structural bioinformatics approaches to study the genome-wide perturbations in a given system. Broadly, this work provides genome-wide structural, transcriptional, and pathway-level insights into the perturbed activity in hypoxia. The presented data are explored with the common goal of elucidating the genome-wide adjustments carried out by C. acnes in anaerobic conditions.