Mitochondrial dysfunction a prelude to Neuro-pathogenesis of SARS-CoV-2

Abstract

The SARS-CoV-2 virus is notorious for its neuroinvasive capability causing multiple neurological conditions. The neuropathology of the SARS-CoV-2 virus is increasingly attributed to mitochondrial dysfunction of brain microglia cells. However, the changes in biochemical content of mitochondria, which drive the progression of Neuro-COVID, remain poorly understood. Here we introduce a Raman microspectrometry approach, enabling the molecular profiling of single cellular organelles, to characterize the mitochondrial molecular makeup in the infected microglia cells. We found that microglia treated with either spike protein or heat inactivated SARS-CoV-2 triggers a dramatic reduction in mtDNA content and an increase in phospholipid saturation levels. At the same time, no significant changes were detected in Golgi apparatus and in lipid droplets, - the organelles which accommodate biogenesis and storage of lipids. We hypothesize that transformation in mitochondria were caused by increased synthesis of reactive oxygen species in these organelles. Our findings call for development of mitochondria targeted therapeutic approaches to limit neuropathology associated with SARS-CoV-2.

Graphical Abstract

Biochemical analysis of single mitochondria by Raman microspectrometry

Introduction

A significant number of COVID-19 patients develop neurological symptoms, attributed to viral encephalitis, resulting in neuro-inflammation, neuronal damage and neurocognitive impairment. The microglia, which are the resident macrophages in the central nervous system, are the major players in the brain’s immune response to SARS-CoV-2 infection. Furthermore, it has been shown that functional mitochondria are integral to initiation and maintenance of immune responses by microglia, while neurological damage in COVID patients is attributed to mitochondrial dysfunction. Mitochondria are the primary site of ATP production and also regulate basic metabolic functions and participate in homeostasis, cellular proliferation, apoptosis, as well as in the synthesis of amino acids, lipids, and nucleotides. In microglia these organelles also mediate the anti-viral immune response by releasing pro-inflammatory cytokines, which limit viral survival, viral replication and trigger inflammation (1–4). Strikingly, SARS-CoV-2 can evade the innate immune response of host cells via the modulation of mitochondrial functions. The spike protein of the SARS-CoV-2 binds to the angiotensin-converting enzyme-2 (ACE-2) receptor on the human host cell (3) to enter the host, and the transmembrane serine protease 2 (TMPRESS 2) facilitates this attachment by priming the spike protein (5). Notably, ACE-2 receptor regulates the mitochondrial function (6). Reduced expression of ACE-2 correlates with decreased ATP synthesis and activation of NADPH oxidase 4, which contributes to the production of reactive oxygen species (ROS) (4). Consistent with that, an invasion of SARS-CoV-2 via the ACE-2 receptor compromises the mitochondrial regulation. Excessive ROS production exacerbates neuro-inflammation, initiating apoptosis in infected cells, which results in neurocognitive impairments. It is known that SARS-CoV-2 infection results in a massive inflammatory response in the brain by triggering the release of cytokines such as interleukin (IL)-10, TNF-α, and INF-γ, which, in turn, further increase mitochondrial ROS production through the upregulation of mitochondrial genes and the modulation of the electron transport chain (ETC) (7). The mitochondrial ROS stimulate additional proinflammatory cytokine production (8) in the face of viral persistence, leading to a “cytokine storm syndrome”, which underlies viral encephalopathy (7). We recently observed an increased oxygen consumption rate (OCR) in microglial cells treated with SARS- CoV-2 spike protein (9). Our data suggested that SARS-CoV-2 induced a robust inflammatory response, significantly increasing oxidative stress and OCR, all of which contributed to neuroinflammation and associated neuropathology of an encephalitic coronavirus infection.

In order to evade host cell immunity and facilitate virus replication, SARS-CoV-2 viral open-reading frame (ORF) −9b, localizes in mitochondria and can directly modulate mitochondrial function, thereby contributing to COVID-19 disease progression (4).

Thus, we hypothesize that modulating mitochondrial activity may prevent mitochondrial dysfunction following SARS-CoV-2 infection and that mitochondria-targeted pharmacological interventions regulating the mitochondrial functions may enhance an immune response in SARS-CoV-2 associated neuro-pathogenesis.

Towards verification of this hypothesis, we analyzed the molecular composition in mitochondria of the infected cells. It is worth noting that a characterization of the mitochondria metabolic variations by standard biochemical approaches is extremely challenging. Traditional molecular profiling approaches rely on cellular fractioning and extraction of the analyte proteins or lipid molecules from the studied organelles, which is a cumbersome procedure prone to contamination. Additionally, the molecular extraction approach inherently produces data averaging, thus masking heterogeneity between organelles obtained from different cells.

Remarkably, the capabilities of biochemical analysis have been recently expanded with optical biosensing tools. Raman spectrometry, one of the most valuable biosensing technologies, relies on analysis of molecular vibrations spectra, and permits for identification of diverse molecular groups in biological samples. Because of inherently non-invasive properties and independence from labels, Raman-based techniques open new dimensions in systemic studies of cells and tissues (10, 11). The recently developed Biomolecular Component Analysis (BCA) of the Raman spectra enables selective detection and concentration measurement of the major categories of biomolecules that include lipids, proteins, nucleic acids and saccharides (12, 13) in the studied samples. High 3D resolution available in modern confocal Raman spectrometry setup has been validated for characterization of microscopic subcellular structures, such as single organelle, including the identification of abnormal biomolecular signatures associated with disease progression (14–20).

In this study we have employed Raman spectrometry together with BCA algorithm to characterize the changes in the molecular composition of mitochondria in response to treatment with the SARS-CoV-2 virus, which has been heat inactivated, or the SARS-CoV-2 spike protein. In addition, we studied key organelles involved in lipid metabolism: Golgi Apparatus (GA) and lipid droplets (LD). The role lipids play in viral infection include viral endocytosis and exocytosis, viral entry into the host cell via membrane fusion and viral replication, therefore, we were interested in potential changes of the lipid signatures in these organelles.

Our data indicate that infection with SARS-CoV-2 causes the mitochondrial dysfunction in microglia cells, that triggers metabolic alterations resulting in a substantial increase in glycolysis (9). These findings suggest that mitochondrial dysfunction and an energy deficit in microglia is compensated by a metabolic switch to glycolysis and a consequence of this metabolic change is an enhanced inflammatory response that contributes to neuropathology associated with COVID-19. At the same time, the molecular content of GA and LD was not significantly changed, apparently due to the lack of specific interactions between these organelles and the components of SARS-CoV-2 virus.

Overall, our findings support a view that the viral infection of a host cells results in higher metabolic alterations to cope with the increased anabolic demand of the cell for viral replication. Furthermore, SARS-CoV-2-induced manipulation of the host cell metabolic machineries alters transcriptional regulation of key metabolic pathways.

Methods

Cell culturing and sample preparation.

Human Microglia cells (HMC3) were obtained from ATCC (Cat # ATCC^® CRL-3304^™) and were grown in luminescence-free 35-mm glass-bottom dishes (Fisher Scientific Co, Hanover Park, IL). Culture media used was Eagle’s Minimum Essential Medium (EMEM) (Cat #ATCC^® 30–2003^™) supplemented with 5% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 μg/ml streptomycin and cells were grown to 70% confluence at 37°C in a humidified atmosphere containing 5% CO₂. The mitochondria, and Golgi Apparatus (GA) were labeled with MitoTracker Green FM, and NBD C6 ceramide-BSA, (ThermoFisher Scientific) respectively, as per manufacturer provided protocols. After labeling, the cells were thoroughly washed in sterile PBS.

The cells were treated with the following viral constructs: 0.5 μg/ml of recombinant Spike protein from SARS-Related Coronavirus 2 Wuhan-Hu-1 (BEI resources Inc Cat # NR-52308; Lot: 70034410) or 5μl/ml of the heat inactivated SARS-Coronavirus 2 (HI-SARS), Isolate USA-WA1/2020, (BEI resources Inc, Cat # NR-52286, Lot: 70033548; Pre-Inactivation Titer by TCID50 Assay in Vero E6 Cells= 1.6 × 10⁵ TCID50 per mL), as specified.

To target Raman spectra acquisition to specific organelles the mitochondria, ER, and GA were labeled using MitoTracker Green FM, ER-Tracker Green, and NBD C6 ceramide-BSA (ThermoFisher Scientific), respectively, as per manufacturer provided protocols. Then, the cells were thoroughly washed in sterile PBS, and Raman spectra were acquired in the labeled organelles.

The Raman microscope.

The spectra measurements were performed on a DXR2 Raman microscopy setup (Thermo Fisher Scientific, WI), equipped with ~60 mW @ 633 nm laser source unit (ROUSB-633-PLR-70-1; Ondax), 50 μm pinhole shaping the laser beam to a 0.7×0.7×1.5μm FWHM (full width half maximum) and a Plan N 100x Olympus objective lens (NA= 1.25). In addition, the Raman microscope was equipped with a fluorescence illumination system (5-UR7005; Olympus), green fluorescence cube (488/561EX) and a fluorescence lamp (X-Cite 120 PC; Photonic Solutions).

Raman Spectra Acquisition.

Prior to the measurements live cells were transferred into an optically transparent DMEM medium (ThermoFisher Scientific) and mounted on the microscope stage. The spectra were acquired from the labeled organelles in live cells, as recently described (15). Fluorescence-labeled organelles were visualized using the 488/561EX fluorescence cube. To generate the spectra, the Raman excitation laser was overlapped with single labeled organelles. To warrant high quality signal/noise ratio, the spectra accumulation parameter was set to 6 × 20 s; Importantly, no measurable phototoxicity was observed at this irradiation dose. During the experiments, the cells were maintained under physiological conditions at 37°C. We visually verified the XYZ position of the cell before and after each measurement to ensure the spatial precision of Raman spectra acquisition.

Biomolecular Component Analysis of Raman Spectra.

The calibration of Raman bands intensity on the concentration of biomolecules in the sample was performed as previously described (15, 21). Quantitative analysis of cellular spectra was performed by using BCAbox software (ACIS LLC, Buffalo, NY). The description, interface of BCAbox software, and schematics for spectra processing algorithm are shown in Supplementary Material (Figures S1, S2). The representative examples of raw and pre-processed mitochondria spectra are shown at Figure S3.

Results and Discussion

In our experiments we incubated the cultured microglia cells with SARS-CoV-2 spike protein or heat inactivated SARS-CoV-2 to imitate viral neuro-invasion. Untreated microglia were used as an experimental control. Mitochondria and the GA, were stained with specific fluorescence probes, thus enabling acquisition of Raman spectra in these organelles, while the lipid droplets were identified by transmitted light imaging.

The obtained Raman spectra were processed with BCA algorithm, to quantify concentrations of major groups of biomolecules (Figure S6–S7). The measurements were performed as recently described (18). It is worth noting, that although the Raman spectra are collected within a submicron volume of an excitation laser focused on specific organelles, the adjacent cytoplasm may also overlap with a laser probe and contribute to the spectra. Nevertheless, despite this potential contribution, there were statistically significant differences between the molecular profiles obtained in various organelles, which supports the sensitivity of Raman microspectrometry to subcellular biochemical environment. The measured values obtained in single mitochondria of control and treated cells are shown in Tables S1–S3.

We found that the treatment with SARS-CoV-2 spike protein or heat inactivated SARS-CoV-2 virus induced significant alterations in the concentrations of diverse types of biomolecules in the mitochondria. First, the concentration of mitochondrial DNA was reduced almost twice in the infected cells, from ~2.2 mg/ml in control cells to ~1.2 mg/ml in the cells treated with either viral agent (Figure 1), which manifests the decrease in mitochondrial DNA copy number. At the same time, the concentration of mtRNA was increased from ~2.25 mg/ml in control, to 2.8 mg/ml in HI-SARS treated cells and to ~4.0 mg/ml in cells treated with the spike protein, the latter difference was statistically significant. This increase of RNA is consistent with previous reports on mitochondrial genome upregulation in cells infected by the SARS-CoV-2 virus (7). We also found a significant reduction in mitochondria saccharides from ~1.5 mg/ml in control to ~0.9 mg/ml in the HI-SARS treated cells and to ~0.7 mg/ml in the cells treated with the spike protein. The mitochondrial saccharide fraction includes glucose and pyruvate and its reduction suggests a decrease of the respiratory function of mitochondria.

Figure 1:

Comparative analysis of the mitochondria molecular content in nontreated (control) and treated with either the heat-inactivated SARS protein or the SARS-CoV-2 spike protein (as indicated). The upper row shows absolute concentrations of Proteins, DNA, RNA, Saccharides (Gly) and Lipids in live cell mitochondria. The bottom chart shows a decrease in the number of C=C bonds in mitochondrial phospholipids (phospholipids unsaturation parameter) in both groups of the treated cells. The statistically significant differences are indicated by horizontal brackets and p value.

Furthermore, we detected a significant perturbation in the saturation of phospholipids populating the mitochondria lipidome. The average number of unsaturated C=C bonds per phospholipid was significantly reduced from ~4.3 in control, to ~3.8 in the cells treated with the HI-SARS viral construct and to ~3.7 in the cells treated with the spike protein. At the same time, we did not record any significant change in the total concentration of lipids in mitochondria (Figure 1). We thus concluded that the shift the lipidome saturation occurs due to biochemical processes inside mitochondria and likely is not caused by trafficking of the saturated phospholipids to this organelle.

In parallel, we investigated an impact of the heat inactivated SARS-CoV-2 on the major organelles involved in the metabolism of lipids such as Golgi apparatus. However, it appears that SARS -CoV-2 virus does not directly influence the lipid biogenesis. We found that all resolvable lipidome characteristics in control and treated cells for these organelles were remarkably uniform. Similarly, the composition of lipid droplets in the treated cells remained largely unchanged. We found, however, that HI-SARS induces an increase in the number of double carbon bonds in the pool of unsaturated phospholipids stored in LD (Figure S5–S7).

In the interpretation of our data, we point to the fact that predominantly mitochondrial lipids are synthesized in the endoplasmic reticulum and then transported to the mitochondria, through GA. While these organelles show no differences in their molecular composition between control and treated cells, the mitochondria demonstrate substantial differences not only in phospholipids saturation but also in abundance of RNA, saccharides and mtDNA content (Figure 1). We propose that these changes originate in virus-induced ROS production, in part via oxidative damage to lipids, oxidation of respiratory chain proteins, affecting metabolism and protein import, which then induces DNA damage as reflected in a sharp decrease of mtDNA levels. Furthermore, the mechanistic link between lipid metabolism and inflammation is well established, wherein lipids can directly activate inflammatory pathways (22). Thus, significant changes in the composition and distribution of lipids within the brain are believed to contribute to neurocognitive decline (23). Furthermore, SARS-CoV-2 induced oxidative stress impacts phospholipid membranes causing additional perturbations of biological processes (24). We propose that an increased oxidative stress impacts the fluidity of phospholipid membranes, which can affect the interactions and activity of metabolic enzymes, resulting in membrane remodeling. The membrane fatty acid composition is thought to be altered in response to oxidative stress, by reducing the number of C=C bonds, which results in higher saturation of organellar lipidome (24, 25). The physiological relevance of membrane remodeling remains unclear, but it may be an adaptive response to cellular stress. These data support our hypothesis that mitochondrial dysfunction, oxidative stress, and inflammation could lead to an increase in COVID associated neurological dysfunction. In addition, our data support the premise that SARS CoV-2 induces release of pathogen associated molecular patterns (PAMPS) and danger associated molecular patterns (DAMPS), ATP, oxidized lipids, heat shock proteins all of which are associated with apoptosis and autophagy (26, 27).

Overall, our study clarifies the role of mitochondrial dysfunction in SARS-CoV-2 induced neuropathology. Our data suggest that mitochondrial dysfunction is among the earliest and the most prominent features of neurodegeneration. In addition, the absence of any significant changes in the lipidome of GA and LD indicate a targeted impact of SARS infection on mitochondria. Therefore, examining mitochondrial function or mitochondrial damage markers in the microglia cells in response to interactions with SARS-CoV-2 spike protein may help identify pathways of viral pathogenesis, and unravel mechanisms of cellular vulnerability and aid discovery of mitochondrial biomarkers relevant to SARS-CoV-2 neuro-inflammation and progression to neuro-pathogenesis. Furthermore, therapeutics strategies that will modulate mitochondrial processes, may be efficacious in treating patients with Neuro-COVID. Our study calls for the development of mitochondria-targeted pharmaceutical drugs, capable to neutralize virus induced ROS production in these cellular organelles.