Differential proteomic profiling towards elucidation of TB-IRIS pathogenesis

Doctoral Thesis

2022

Permanent link to this Item
Authors
Journal Title
Link to Journal
Journal ISSN
Volume Title
Publisher
Publisher
License
Series
Abstract
Background Up to 59% of tuberculosis (TB)/human immunodeficiency virus (HIV) co-treated patients develop paradoxical TB-associated immune reconstitution inflammatory syndrome (TB-IRIS) after addition of antiretroviral (ARV) therapy to anti-tuberculous therapy (ATT). The course can be prolonged and the average mortality rate is 2% (75% for TB-IRIS involving the central nervous system (CNS)). Immune elements – including neutrophils - involved in the anti-Mycobacterium tuberculosis (Mtb) response are implicated in pathogenesis, which remains incompletely understood. Diagnosis is one of exclusion, no reliable laboratory markers exist, corticosteroid-mediated prophylaxis and therapy are only partially effective, and no treatment targets tissue damage. Disentangling cause and effect in complex disorders such as TB-IRIS requires techniques capable of interrogating complex biological systems. Neutrophils are the major circulating leukocyte population, the earliest innate system responders, and exhibit various unusual immunometabolic functional specialisations. Proteins represent the most functionally-proximal and commonly pharmacologically-targeted cellular biomolecules. Label-free high-performance liquid chromatography-coupled tandem mass spectrometry (HPLCMS/MS) is well-suited to differentially profiling the ex vivo neutrophil proteome in an unbiased manner, in order to investigate TB-IRIS predisposition and pathogenesis. Methods Applying first principles to existing human literature, the most parsimonious holistic hypothetical model regarding paradoxical TB-IRIS predisposition and pathogenesis was inferred. A clinical cohort of control (CTRL) and matched TB-IRIS case (IRIS) study participants was assembled. Demographic, clinical, and biochemical characteristics were analysed for statistically-significant differences and to identify potential risk/protective factors (relative risk (RR) with 95% confidence interval (CI)). A small group (n = 9) of TB-HIV- healthy volunteers (HVs) was also assembled. Phlebotomy occurred at two timepoints: just prior to ARV initiation (week 0) and at the typical time of IRIS manifestation (week 2). Neutrophils were isolated and lysed, proteins underwent on-filter protein trypsinisation, peptide salts and detergents were removed, and neutrophil-optimised HPLC-MS/MS was conducted. Spectra were submitted to MaxQuant for parent protein identification and quantitation. Comparisons of (a) CTRL0 and IRIS0 to HV (and resultant differences) identified class-differential impacts of partial ATT-treated coinfection (IRIS predisposition) and of (b) CTRL2 to CTRL0 and IRIS2 to IRIS0 (and resultant differences) identified class-differential impacts of ARV therapy (IRIS pathogenesis). Class-discriminating proteomic differences were visualised using principal components analysis (PCA), protein differential expression analysis was performed (including for detectable/undetectable and significantly differentiallyexpressed (SDE) proteins), and results informed differential functional profiling via gene ontology overrepresentation analysis (GO-ORA) and pathway activation state prediction. To address shortcomings of current knowledgebases and automated tools, a novel deep manual analysis approach focused on key inference-friendly proteins, convergent findings, and neutrophil-specific functional modules. Integrated findings extended the literature-derived TB-IRIS model, generating testable novel hypotheses, one of which was partially validated using live-cell fluorescence microscopy. Proteomic data were additionally analysed to detect Mtb proteins, preliminarily analyse variable post-translational modifications (PTMs) of interest, and identify candidate prognostic and diagnostic biomarkers. Finally, mechanistic hypotheses facilitated identification of novel potential prophylactic and therapeutic targets. Results (1) Literature suggests advanced TB/HIV-coinfection (including a higher Mtb/antigen load) as the major TB-IRIS risk factor. Attendant significant immunometabolic state perturbations include myeloid overactivation, metabolic stress (possibly including adaptogen depletion), a lack of regulatory receptors, impaired pro-inflammatory signal transduction, and impaired antigen clearance. These likely predispose to lytic cell death - including release of host- and pathogen-derived inflammatory/cytotoxic molecules and proteolytic enzymes - and less restrained/more abnormal inflammation as well as tissue damage, on restoration of HIV-suppressed inflammatory signalling pathways by ARV therapy. (2) Regarding the clinical cohort, a sample size of > 42 participants per characteristic-matched comparison class provides > 95% power to detect a two-fold change with 99% confidence. Cases exhibited known TB-IRIS risk factors, and largely expected white cell count (WCC), body mass index (BMI), and C-reactive protein (CRP) level changes in response to ARV therapy (e.g. WCC increase and BMI decrease). None of the few participants on alternate (efavirenz (EFV)- or tenofovir (TDF)-lacking) regimens developed TB-IRIS. Pre-ARV prednisone (or incidental antihistamine/anti-fungal) use was associated with a non-significantly decreased TB-IRIS risk. Interestingly, smoking is associated with a significant decrease in TB-IRIS risk by 60%. (3) Regarding sample processing and analysis, the average sample collection to neutrophil isolation interval was within recommended limits. Isolation yield exceeded 15 x 106 and 25 x 106 per sample in the CTRL and IRIS groups, respectively. Isolated neutrophil purity exceeded 80% in both groups; the few low-purity samples were excluded from subsequent proteomic analysis. Lysates from 5 x 106 neutrophils routinely yielded over 100μg total protein, tryptic digestion was efficient (on average < 96% missed cleavages), and equivalent peptide injection volumes yielded comparable total ion chromatogram (TIC) profiles and intensities. An average 23% spectral identification rate resulted in a total of 2532 protein group identifications, the deepest neutrophil proteome coverage achieved to date without pre-fractionation, representing ~12% of human protein coding genes, and ~25% of the detectable human proteome. Samples were analysed in two (randomised) batches, producing independent datasets A (N = 37) and B (N =74). Withindataset technical replicates exhibit excellent agreement in protein identities and quantities; betweendataset protein identities and functional inferences also exhibit excellent agreement. We identify a number of proteins apparently not known to be expressed by human neutrophils, as well as one predicted human protein never before observed empirically. Overall, parent pathways of level-altered proteins suggest perturbation of nine major neutrophil function modules at both time-points: (a) signal transduction, (b) pattern recognition receptor (PRR) and cytokine signalling, (c) the eicosanoid cascade, (d) neutrophil antimicrobial functions, (e) carbon-energy metabolism, (f) protein homeostasis, (g) integrated nitrogen-sulfur-B-vitamin and redox/xenobiotic/glyoxal metabolism, (h) gene expression, and (i) cytoskeletal dynamics. (4) Regarding impacts of partially ATT-treated co-infection (week 0), neutrophil proteomic profiles successfully distinguish between HV and IRIS0 or CTRL0. Many differences from HV are shared between IRIS0 and CTRL0 (i.e. driven by partially ATT-treated co-infection), but some are class-unique (i.e. driven by factors predisposing to or protecting from TB-IRIS). Findings are supported by a head-to-head comparison of the CTRL0 and IRIS0 proteomes, including changes suggesting: more prevalent type I IFN, TGFβ, and Th2-type cytokine signalling; poorer capacity for restraint of alternate complement activation; mitochondrial and oxidative stress (including proneness to necrosis); impaired function (e.g. microbicidality, TLR/IL-1R-MyD88-NFκB signalling, and caspase 1- mediated IL-1β and IL-18 maturation) of activated neutrophils; and enhanced lipid and upstream (but inhibited downstream) isoprenoid synthesis (including decreased steroidogenesis). Candidate biomarkers distinguish CTRL- and IRIS-class partially ATT-treated neutrophils from HV neutrophils and from each other. (5) Regarding impacts of ARV therapy (week 2), both IRIS and CTRL neutrophil proteomes exhibit significant changes in response to ARV therapy. Many changes are shared between IRIS and CTRL (i.e. driven by ARV therapy and declining viral load (VL)), but some are class-unique (i.e. driven by factors preventing/contributing to TB-IRIS pathogenesis). Findings are supported by a head-to-head comparison of the CTRL2 and IRIS2 proteomes, including changes suggesting: a slower decline in type I IFN signalling; increased inflammatory cytokine (e.g. IL-6, TNFα, and IFNγ) signalling and protease (e.g. MMP-8) activity; decreased sensitivity to immunoregulatory glucocorticoids and vitamin A; and increased mitochondrial, endoplasmic reticulum (ER), and oxidative stress. Candidate biomarkers distinguish CTRL- and IRIS-group ARV-exposed neutrophils from baseline and from each other. (6) Livecell fluorescence microscopy of HV neutrophils suggests that in vivo-equivalent levels of EFV rapidly alter mitochondrial, lysosomal, and aggresomal architecture in a manner consistent with organelle and protein folding stress, and suggesting cell death commitment. (7) Integrated neutrophil immunometabolic changes suggested by proteomic findings support and extend the biologically compelling literature-derived model. Model highlights include more advanced baseline TB/HIV (including higher type I IFN, TGFβ, and possibly Th2-type cytokine levels), with consequent impaired myeloid-mediated Mtb antigen clearance and depletion of cellular adaptogens. The resultant abnormal immunometabolic state produces myeloid cells less able to counteract metabolic stress and primed for less-restrained inflammation. Introduction of mitotoxic ARV drugs and rapid lifting of HIVmediated immune embargoes escalates myeloid metabolic (including oxidative) stress and overactivation (including via NLRC4, CASP4/5, TLR/IL-1R-MyD88-NFκB, and MAPK-AP1 signalling), producing - instead of Mtb clearance - inflammatory cell death with release of immune-activating and tissue-damaging host- and Mtb-derived molecules. Reactivation of Mtb lymphocyte memory responses likely only produces clinically-apparent inflammation (TB-IRIS) when multiple simultaneous but incompatible immune programmes (e.g. overzealous myeloid activity, Th1, Th2, Th17, and Treg) coexist. Based on this model, existing compounds with the potential for rational, safe, effective TB-IRIS prophylaxis/therapy are identified (e.g. glutathione, vitamins B-complex and A/D/E, rapamycin, and metformin) which may assist in restoring system homeostasis.
Description

Reference:

Collections