EPCO-10. SYSTEMS BIOLOGY-BASED THERAPEUTIC PREDICTIONS WITH GBMSYGNAL AND CLINICAL CORRELATES IN THE REAL-WORLD LONGITUDINAL OUTCOMES REGISTRY XCELSIOR

Glioblastoma is an aggressive disease with multiple disrupted signaling networks in the same tumor, demanding the development of personalized combination regimens. Using SYstems Genetics Network AnaLysis we analyzed TCGA multiomics datasets for 422 glioblastoma patients to generate a predictive disease network model (gbmSYGNAL). This model seeks to uncover how subsets of mutations causally modulate regulators (transcription factors, miRNAs, etc.) that mechanistically regulate gene modules (“regulons”) associated with disease progression. Using the real-world registry XCELSIOR (NCT03793088), we identified 55 anti-cancer therapies utilized in glioblastoma treatment (N = 354 patients) and associated them with positive or negative regulon activity within individual patients across the TCGA cohort. A median of 12 regulons were targeted by each drug. Interestingly, nilotinib targeted only 5 regulons but they were found to be active in 38% of patients and regulons overactive in >35% of patients were targeted by repurposed anti-cancer drugs including celecoxib, hydroxychloroquine, and tocilizumab. Standard-of-care treatments lomustine and bevacizumab were predicted to be active in only 28% and 23% of patients, respectively. Using gbmSYGNAL, we then prioritized drugs for 12 additional patients from the XCELSIOR registry. Median overactive regulons per patient was 110 (range 55-254) represented by a median 28 molecular targets (range 17-37). Surprisingly, regulons targeted by metformin were predicted to be overactive in all patients. Certain regulons displayed a binary pattern (high “on” activity or completely “off” per patient) including those targeted by pemetrexed, pembrolizumab, and valproic acid while others showed a gradient of activity across patients including selinexor, panobinostat, palbociclib, and alpelisib. Systems biology analysis of commercially available NGS data combined with the XCELSIOR direct patient engagement platform yielded therapeutically actionable insights for glioblastoma patients in real time. Correlation with clinical outcomes is ongoing for ~50 additional patients and will be presented at the meeting.

Mycobacterium abscessus biofilms produce an extracellular matrix and have a distinct mycolic acid profile

A non-tuberculous mycobacterium, Mycobacterium abscessus is an emerging opportunistic pathogen associated with difficult to treat pulmonary infections, particularly in patients suffering from cystic fibrosis. It is capable of forming biofilms in vitro that result in an increase of already high levels of antibiotic resistance in this bacterium. Evidence that M. abscessus forms biofilm-like microcolonies in patient lungs and on medical devices further implicated the need to investigate this biofilm in detail. Therefore, in this study we characterized the M. abscessus pellicular biofilm, formed on a liquid-air interface, by studying its molecular composition, and its transcriptional profile in comparison to planktonic cells. Using scanning electron micrographs and fluorescence microscopy, we showed that M. abscessus biofilms produce an extracellular matrix composed of lipids, proteins, carbohydrates and extracellular DNA. Transcriptomic analysis of biofilms revealed an upregulation of pathways involved in the glyoxylate shunt, redox metabolism and mycolic acid biosynthesis. Genes involved in elongation and desaturation of mycolic acids were highly upregulated in biofilms and, mirroring those findings, biochemical analysis of mycolates revealed molecular changes and an increase in mycolic acid chain length. Together these results give us an insight into the complex structure of M. abscessus biofilms, the understanding of which may be adapted for clinical use in treatment of biofilm infections, including strategies for dispersing the extracellular matrix, allowing antibiotics to gain access to bacteria within the biofilm.

EPCO-23. A SINGLE-CELL BASED PRECISION MEDICINE APPROACH USING GLIOBLASTOMA PATIENT-SPECIFIC MODELS

Glioblastoma is a heterogeneous tumor made up of cell states that evolve over time. We modeled tumor evolutionary trajectories during standard-of-care treatment using multimodal single-cell analysis of a primary tumor sample, corresponding mouse xenografts subjected to standard of care therapy, and recurrent tumor at autopsy. We mined the multimodal data with single cell SYstems Genetics Network AnaLysis (scSYGNAL) to identify a network of 52 regulators that mediate treatment-induced shifts in xenograft tumor-cell states that were also reflected in recurrence. By integrating scSYGNAL-derived regulatory network information with transcription factor accessibility deviations derived from single-cell ATAC-seq data, we developed consensus networks that regulate subpopulations of primary and recurrent tumor cells. Finally, by matching targeted therapies to active regulatory networks underlying tumor evolutionary trajectories, we provide a framework for applying single-cell-based precision medicine approaches in a concurrent, neo-adjuvant, or recurrent setting. Our proof-of-concept work herein provides the basis for the development of a modeling and analytical system that enables single-cell characterization of an individual patient’s tumor and inferred therapeutic vulnerabilities. Although further validation is required, in the form of in vivo studies of these putative druggable targets, our preliminary analysis and results suggest that systems biology techniques can be used to infer and predict therapeutic vulnerabilities that are either selected or induced during standard-of-care treatment. Ultimately, the information gathered from such systematic modeling and analysis of individual tumors may inform clinical treatment in a more targeted manner and enable a rational, tailored precision medicine that accounts for intratumoral cell heterogeneity.

Identifying innate immune regulators of Mycobacterium tuberculosis pathogenesis in macrophages

Tuberculosis (TB) is a leading infectious disease that contributes to approximately 1.4 million deaths annually worldwide. In the present situation of the COVID-19 pandemic and disrupted healthcare services, mathematical modelling predicts that there will be an additional 6.3 million new cases and 1.4 million more TB deaths by the end of 2025. Innate immune cells like macrophages are the frontline responders in host defense against Mycobacterium tuberculosis (MTB) and may also play a major role in disease establishment. A major roadblock in developing effective strategies against TB is the inability of current anti-TB drugs to target both replicating and non-replicating bacteria. We have found that lack of specific components of the host innate sensing machinery has a major impact on intracellular bacterial growth and the production of IL-1β and IFN-β during MTB infection of macrophages; the latter cytokines play major roles in host-protective and host-detrimental outcomes of infection respectively. RNA seq reveals that these innate immune modules differentially control the expression of genes implicated in host protein synthesis and MTB dissemination and escape from the granuloma. Ongoing work is utilizing Path-Seq for quantitative profiling of the MTB transcriptome within infected macrophages and investigating macrophage-dependent mechanisms of MTB restriction and persistence in vivo. Our study will potentially identify immune regulators that make the macrophage environment either permissive or restrictive for intracellular replication of MTB with important implications for developing combinatorial targeting approaches for host-directed treatment of TB.

Biofilms of the non-tuberculous Mycobacterium chelonae form an extracellular matrix and display distinct expression patterns

Mycobacterium chelonae is an environmental, non-tuberculous mycobacterial species, capable of causing infections in humans. Biofilm formation is a key strategy used by M. chelonae in colonising niches in the environment and in the host. We studied a water-air interface (pellicle) biofilm of M. chelonae using a wide array of approaches to outline the molecular structure and composition of the biofilm. Scanning electron micrographs showed that M. chelonae biofilms produced an extracellular matrix. Using a combination of biochemical analysis, Raman spectroscopy, and fluorescence microscopy, we showed the matrix to consist of proteins, carbohydrates, lipids and eDNA. Glucose was the predominant sugar present in the biofilm matrix, and its relative abundance decreased in late (established) biofilms. RNA-seq analysis of the biofilms showed upregulation of genes involved in redox metabolism. Additionally, genes involved in mycolic acid, other lipid and glyoxylate metabolism were also upregulated in the early biofilms.

Distinct host immune programs in lung macrophages elicit cell-type specific Mycobacterium tuberculosis transcriptional responses

During Mycobacterium tuberculosis (Mtb) infection, bacteria are inhaled into the lung where they first infect resident, alveolar macrophages (AM). The majority of the infection remains in AM until D14, when other cells, including recruited monocyte-derived macrophages (MDM) begin to become infected. By the peak of infection (D28) MDM represent the major infected cell type. However, even at this time point, a small population of infected AM remain. This led us to ask how these cell types control Mtb. Using microscopy, we observed that AM harbor more bacteria than MDM on a per cell basis at multiple time points, including after the initiation of the T cell response. Using RNA-seq, we identified multiple differentially expressed pathways between the two cell types. While infected MDM upregulate proinflammatory signaling pathways associated with Mtb control, infected AM are enriched for proliferation and fatty acid metabolism pathways. We performed validation studies using dyes to track cell division and mitochondrial metabolism and observed that AM were more proliferative and had sustained engagement of mitochondrial metabolism relative to MDM. In parallel, we analyzed the bacterial transcriptome and found that Mtb in MDM have a signature associated with late hypoxia. These bacterial responses also suggested that Mtb may differ in its drug tolerance in a cell type specific manner, which we are currently investigating. Together, this work is in agreement with other studies demonstrating differences in the response to Mtb by tissue resident and recruited macrophages and suggests that these differences may lead to distinct transcriptional changes in the bacteria, which could have important implications for therapeutic interventions.

Antibody-recruiting protein-catalyzed capture agents to combat antibiotic-resistant bacteria

Antibiotic resistant infections are projected to cause over 10 million deaths by 2050, yet the development of new antibiotics has slowed. This points to an urgent need for methodologies for the rapid development of antibiotics against emerging drug resistant pathogens. We report on a generalizable combined computational and synthetic approach, called antibody-recruiting protein-catalyzed capture agents (AR-PCCs), to address this challenge. We applied the combinatorial protein catalyzed capture agent (PCC) technology to identify macrocyclic peptide ligands against highly conserved surface protein epitopes of carbapenem-resistant Klebsiella pneumoniae, an opportunistic Gram-negative pathogen with drug resistant strains. Multi-omic data combined with bioinformatic analyses identified epitopes of the highly expressed MrkA surface protein of K. pneumoniae for targeting in PCC screens. The top-performing ligand exhibited high-affinity (EC₅₀ ∼50 nM) to full-length MrkA, and selectively bound to MrkA-expressing K. pneumoniae, but not to other pathogenic bacterial species. AR-PCCs that bear a hapten moiety promoted antibody recruitment to K. pneumoniae, leading to enhanced phagocytosis and phagocytic killing by macrophages. The rapid development of this highly targeted antibiotic implies that the integrated computational and synthetic toolkit described here can be used for the accelerated production of antibiotics against drug resistant bacteria.

Alveolar and monocyte-derived macrophages differentially engage antibacterial programs during adaptive immunity to Mycobacterium tuberculosis

Cognate interactions between antigen-specific CD4 T cells and Mycobacterium tuberculosis (Mtb)-infected cells are critical for mucosal immunity to tuberculosis (TB). However, Mtb resides intracellularly within a variety of myeloid cell populations, and the relative impact of CD4 T cells on distinct infected cell types remains unknown. Here we report that intratracheally-transferred, Mtb-specific T cells are unable to reduce the bacterial burden in the first week after aerosol infection, when Mtb resides almost exclusively within alveolar macrophages (AMs). However, transferred T cells do mediate protection by day 14, at which point Mtb has disseminated to monocyte-derived macrophages (MDMs). Using an Mtb “replication clock” strain, which loses plasmid at a fixed rate with bacterial division, we observe that Mtb continues to replicate in AMs even after the onset of adaptive immunity, whereas replication in MDMs is curtailed. We also find that MDMs are less reliant on oxidative metabolism and express more NOS2 than AMs, which may explain their superior antibacterial activity. Finally, RNA-Seq analysis of these two macrophage populations reveals enriched expression of proinflammatory pathways (i.e. NFkB signaling) in Mtb-infected MDMs compared to AMs. Together, our results indicate that AMs may serve as a privileged niche for Mtb, despite the presence of Mtb-specific T cells. As many current TB vaccine candidates seek to induce lung-resident T cells that can recognize and control Mtb early after aerosol infection, the relative resistance of Mtb-infected AMs to T cell-mediated immunity could represent a previously unappreciated barrier to this strategy.

Constraint-based modelling captures the metabolic versatility of Desulfovibrio vulgaris

A refined Desulfovibrio vulgaris Hildenborough flux balance analysis (FBA) model (iJF744) was developed, incorporating 1016 reactions that include 744 genes and 951 metabolites. A draft model was first developed through automatic model reconstruction using the ModelSeed Server and then curated based on existing literature. The curated model was further refined by incorporating three recently proposed redox reactions involving the Hdr-Flx and Qmo complexes and a lactate dehydrogenase (LdhAB, DVU 3027-3028) indicated by mutation and transcript analyses to serve electron transfer reactions central to syntrophic and respiratory growth. Eight different variations of this model were evaluated by comparing model predictions to experimental data determined for four different growth conditions - three for sulfate respiration (with lactate, pyruvate or H₂ /CO₂ -acetate) and one for fermentation in syntrophic coculture. The final general model supports (i) a role for Hdr-Flx in the oxidation of DsrC and ferredoxin, and reduction of NAD⁺ in a flavin-based electron confurcating reaction sequence, (ii) a function of the Qmo complex in receiving electrons from the menaquinone pool and potentially from ferredoxin to reduce APS and (iii) a reduction of the soluble DsrC by LdhAB and a function of DsrC in electron transfer reactions other than sulfite reduction.

Identifying key regulators of CD4 T cell function in the Mycobacterium tuberculosis-infected lung using systems analysis

Mycobacterium tuberculosis (Mtb) infection is the leading infectious killer worldwide and an effective vaccine remains elusive. Although CD4 T cells are known to be important for protection, the specific requirements for a protective CD4 response are unclear. Our lab has shown that differences in antigen exposure drive significant phenotypic variation within Mtb-specific CD4 T cells. CD4 T cells specific for the Mtb protein Ag85B, which is only expressed early during infection, remain capable of robust cytokine production. Conversely, T cells specific for ESAT-6, which is expressed for the duration of infection, exhibit blunted cytokine production. To identify regulators that govern this T cell exhaustion, we employed a systems biology approach. We performed RNA-seq on unstimulated and stimulated Ag85B and ESAT-6-specific CD4 T cells isolated from the lungs of Mtb-infected mice and used cMonkey, a biclustering algorithm, to create gene clusters based on expression patterns and identify transcription factors predicted to regulate the clusters. This analysis revealed eight transcription factors that were themselves differentially expressed between Ag85B and ESAT-6-specific CD4 T cells and also predicted to regulate clusters of differentially expressed genes. We are currently focusing our analysis on two of these transcription factors, VDR and Lyl-1. Mtb infection of WT:VDR−/− mixed bone marrow chimeras have validated the systems-based predictions by showing that VDR regulates distinct processes in Ag85B-specific compared to ESAT-6-specific CD4 T cells. Understanding the disparate regulatory pathways operant in functional vs. exhausted T cells may inform new avenues for host-directed therapy of tuberculosis.

Allosteric modification of oxygen delivery by hemoglobin

Hemoglobin affinity for oxygen is altered by pH, temperature, and high altitude, making oxygen more readily available to the tissues. RSR13 (Allos Therapeutics, Denver, CO), an analog of the drugs clofibrate and bezofibrate, causes a dose-dependent, rightward shift of the oxygen dissociation curve in animals and humans. We tested the safety, pharmacodynamic, and pharmacokinetics of RSR13, an allosteric modifier of hemoglobin, in patients having general surgery in a prospective, randomized, double-blinded, placebo-controlled, dose-escalation clinical trial. After the induction of general anesthesia with endotracheal intubation, 26 patients who consented were randomly assigned to receive an infusion of RSR13 or placebo (2:1) in an ascending dose scheme. Doses studied were 10, 20, 30, 40, 50, 60, 75, and 100 mg/kg infused for 30--60 minutes. Samples were taken for determination of RSR13 concentration in plasma, red blood cells, and urine, as well as for determination of the p50 in blood by using three-point tonometry at frequent intervals after the infusion of the study drug. The RSR13 administration resulted in a dose-dependent rightward shift of the oxygen dissociation curve, with the target p50 shift of 10 mm Hg achieved at the 75- and 100-mg/kg doses. No differences were seen between RSR13 and placebo groups in laboratory or hemodynamic findings, with the exception of a transient, limited increase in serum creatinine in 3 patients who received RSR13. These increases peaked at 48 h (2.2, 3.5, and 4.5 mg/dL respectively), were not associated with oliguria, did not require treatment, and did not prolong hospitalization in any patient. The reasons for the unexplained increases in serum creatinine were not evident, but potentially included surgery itself (nephrectomy), patient condition, or the concomitant administration of renally cleared medications or drugs that affect renal blood flow.

IMPLICATIONS

We studied the safety and tolerance of an investigational drug, RSR13 (Allos Therapeutics, Denver, CO), in general surgery patients. This drug, which increases the amount of oxygen available to the body, was well tolerated by the 17 patients who received it. There were clinically relevant increases in serum creatinine in 3 patients, indicating a decrease in renal function, but these increases were short-lived and resolved without treatment.

Improving the thromboresistivity of chemical sensors via nitric oxide release: fabrication and in vivo evaluation of NO-releasing oxygen-sensing catheters

The development and in vivo analytical performance of a nitric oxide (NO)-releasing amperometric oxygen sensor with greatly enhanced thromboresistivity are reported. Gas permeable coatings formulated with cross-linked silicone rubber (SR) containing NO-generating compounds (diazeniumdiolates) are shown to release NO for extended periods of time (> 20 h) while reducing platelet adhesion and activation. Oxygen-sensing catheters prepared by dip-coating the NO-releasing films over the outer SR tubes of the implantable devices display similar analytical response properties in vitro (sensitivity, selectivity, response times) when compared to analogous sensors prepared without the NO release coatings. Superior analytical accuracy (relative to blood PO2 values measured in vitro) and greatly reduced thrombus formation on the outer surface of the sensors are observed in vivo (in canine model) with the NO release PO2 sensors compared to control sensors (without NO release) implanted simultaneously within the same animals. Based on these preliminary studies, the use of NO release polymers to fabricate catheter-style chemical sensors may be a potential solution to lingering biocompatibility and concomitant performance problems encountered when attempting to employ such devices for continuous intravascular measurements of blood gases and electrolytes.

Determination of low-molecular-weight heparins and their binding to protamine and a protamine analog using polyion-sensitive membrane electrodes

A polycation-sensitive membrane electrode based on the ion-exchanger dinonylnaphthalene sulfonate has previously been developed and used as an end-point detector for the determination of unfractionated heparin in whole blood samples via simple potentiometric titration with protamine. Herein, we report the application of the same methodology for the quantitation of a commercial low-molecular-weight heparin (LMWH) preparation (Fragmin) in whole blood samples at concentrations up to 2 U/ml. Further, an analogous polyanion (heparin)-sensitive electrode is used to estimate the binding constants between protamine and various LMWH preparations. The equilibrium constants (Keq) and the number of binding sites per mole of heparin (n) are determined by recasting the data in the form of a Scatchard plot. Results show that the average molecular weight and molecular weight distribution of the LMWH preparation are important parameters affecting their binding with protamine. Comparable binding constants are obtained for the same LMWH preparations titrated with a synthetic protamine analog, [+18RGD] [acetyl-EA(R2A2R2A)4R2GRGDSPA-NH2].

Improved protamine-sensitive membrane electrode for monitoring heparin concentrations in whole blood via protamine titration

An improved protamine-sensitive electrode based on a polymeric membrane doped with the charged ion exchanger dinonylnaphthalenesulfonate (DNNS) is used for monitoring heparin concentrations in whole blood. The electrode exhibits significant nonequilibrium potentiometric response to polycationic protamine over the concentration range of 0.5-20 mg/L in undiluted whole-blood samples. The sensor can serve as a simple end point detector for the determination of heparin via potentiometric titrations with protamine. Whole-blood heparin concentrations determined by the electrode method (n > or = 157) correlate well with other protamine titration-based methods, including the commercial Hepcon HMS assay (r = 0.934) and a previously reported potentiometric heparin sensor-based method (r = 0.973). Reasonable correlation was also found with a commercial chromogenic anti-Xa heparin assay (r = 0.891) with corresponding plasma samples and appropriate correction for whole-blood hematocrit levels. Whereas a significant positive bias (0.62 kU/L; P < 0.001) is observed between the anti-Xa assay and the protamine sensor methods, insignificant bias is observed between the protamine sensor and the Hepcon HMS tests (0.08 kU/L; P = 0.02). The possibility of fully automating these titrations offers a potentially simple, inexpensive, and accurate method for monitoring heparin concentrations in whole blood.