Autoantibodies cause some Long-Covid symptoms

Experience with other viral infections suggests that increased autoimmunity following infection with COVID-19 might contribute to the symptoms of Post-Acute Sequelae of SARS-CoV-2 infection (PASC), more commonly known as Long COVID (LCOVID).  LCOVID is characterized by the presence of autoantibodies of diverse specificities and clinical improvement is correlated with their decline.  Immunoglobulin G (IgG) antibodies from fibromyalgia patients has been shown to cause pain in mice.  Therefore, this group tested the effects on mice of IgG antibodies from LCOVID patients. 

Most LCOVID patients’ plasma showed reduced IFN-gamma (contrast this report) and many showed elevated Glial Fibrillary Acidic Protein (GFAP, which is neuroprotective); other measured proteins were unchanged (Fig 1).  Patient sera were separated into 3 groups based on neuronal damage and astroglial activation markers (Long Covid-1, LC-1), and type I IFNs (LC-2 higher IFNa2a, beta, LC-2 lower).  Principal component analysis of sera proteome signatures confirmed groupings (Figs 3 & 4).  

 IgG from patients with elevated neuronal proteins (LC-1) or leukocyte activation markers (LC-3) caused sensory hypersensitivity (Fig 5, shown; LC1 red, LC3 yellow ).  They claim that IgG from patients with higher muscle proteins also reduced locomotor movements in mice, though the extent appears small (Fig 6).  This careful study is an important guide to such analyses.  

Hung-Jen Chen, Brent Appelman, Hanneke Willemen, Amelie Bos, Judith Prado, Chiara. E. Geyer, Patrícia Silva Santos Ribeiro, Sabine Versteeg, Mads Larsen, Eline Schüchner, Marije K. Bomers, Ayesha H.A. Lavell, Amsterdam UMC COVID-19 biobank, Braeden Charlton, Rob Wüst, W. Joost Wiersinga, Michèle van Vugt1, Gestur Vidarsson, Niels Eijkelkamp, Jeroen den Dunnen . Transfer of IgG from Long COVID patients induces symptomology in mice. BioRxiv, May 31. 2024.

Fumarate, mitochondria, and inflammation

Fumarate hydratase (FH), also known as fumarase, catalyses hydration of fumarate to malate and is found in mitochondria, where it performs in the Krebs cycle, and the cytosol, where it performs in the urea cycle and in catabolizing amino acids. It is encoded by homologous genes in bacteria, yeast, and eukaryotes. Fumarase deficiency in humans is involved in a range of symptoms from neurological abnormalities in newborns to tumors in adults.

These authors examined what happens immediately after loss of FH. They generated a mouse line with floxed Fh1 gene (homolog of the human FH) and crossed it with a line with ubiquitous, inducible, Cre recombinase (Rosa26). Induction of Cre caused the genetic loss of FH expression within 5 days and metabolic changes within 10 days (Fig 1). They found mitochondria swell and lose their genome (mtDNA) into the cytosol, where it triggers innate immunity (STING) and stimulates inflammation (RIG-I) including expression of interferon stimulated genes (ISGs). 

Treating cultured cells with a cell-permeable derivative of fumarate, monomethylfumarate (MMF), caused similar releases of mtDNA into the cytosol and expression of ISGs (Fig 3 Shown), strongly suggesting that the accumulation of fumarate causes the changes observed after loss of FH. They also show that mtDNA is released through ‘mitochondria-derived vesicles’ (MDVs), normally used to transport ‘content without affecting the integrity of the membranes’ (Fig 4), so the process occurs under some control. Finally, they show activation of DNA-sensor and innate immunity pathways in FH-deficient renal cancers, suggesting relevance for humans.

Fig 3a, Fumarate causes mitochondrial changes.  Overlay panels. Cells were treated for 8 days with solvent alone (vehicle) or with fumarate (200 or 400 uM MMF doses) then stained for DNA (green) and mitochondria (outer membrane protein TOM20, purple). Scale bars 10 um. Boxed area in left column expanded in right column.

One wonders what happens after the loss of other ancient and obscurely important cellular infrastructure.

 

Zecchini V, Paupe V, Herranz-Montoya I, Janssen J, Wortel IMN, Morris JL, Ferguson A, Chowdury SR, Segarra-Mondejar M, Costa ASH, Pereira GC, Tronci L, Young T, Nikitopoulou E, Yang M, Bihary D, Caicci F, Nagashima S, Speed A, Bokea K, Baig Z, Samarajiwa S, Tran M, Mitchell T, Johnson M, Prudent J, Frezza C. Fumarate induces vesicular release of mtDNA to drive innate immunity. Nature. 2023 Mar;615(7952):499-506. doi: 10.1038/s41586-023-05770-w. Epub 2023 Mar 8. PMID: 36890229; PMCID: PMC10017517.

Suppressing suppressors to improve lupus?

Lupus patients have more inflammation and more interferon type I (alphas, beta).  SLE monocytes have less NLRP12, which is part of certain inflammasome.and contributes to the activation of pro-inflammatory caspases.  TLR7 is important in lupus.  The authors note that NLRP12 was ‘recently identified’ (citing a 2012 paper) as a negative regulator of TLR and NFkB activation. 

They found (Fig 1) NLRP12 expression is lower in lupus patients than in healthy controls and it is inversely correlated with type I interferon (IFN-a2) expression; (Fig 2) RUNX1 binding sequences in the NRLP12 promoter reduced expression of a reporter gene and CRISPR knockout of RUNX1 increased NLRP12 expression induced by IFN or virus; (Fig 3) there is more RUNX1 and less NLRP12 in monocytes from lupus patients (panels I, shown) and more RUNX1 binding to NLRP12 promoters in PBMC of lupus patients (panel J). (Fig 4) IFN-induced suppression of NLRP12 is mediated by histone acetylation; (Fig 5) RUNX1 reduces NLRP12, which increases IFN, which is (also) observed in SLE; (Fig 6) NLRP12-KO mice are more pro-inflammatory and (Fig 7 & 8) develop worse disease in mouse models of lupus (pristane injection or Fas-lpr).

Figure 3, panels I&J. Panel I immunoblot of proteins in lysates of CD14+ monocytes, quantified on right. Panel J ChIP (chromosome immunoprecipitation) of PBMC.

They do a very good job connecting pointillist dots.

Tsao YP, Tseng FY, Chao CW, Chen MH, Yeh YC, Abdulkareem BO, Chen SY, Chuang WT, Chang PC, Chen IC, Wang PH, Wu CS, Tsai CY, Chen ST. NLRP12 is an innate immune checkpoint for repressing IFN signatures and attenuating lupus nephritis progression. J Clin Invest. 2023 Feb 1;133(3):e157272. doi: 10.1172/JCI157272. PMID: 36719379.  

Can’t just “get over it”: Long COVID

Many people have experienced lingering problems after recovery from COVID infection, a condition known as long COVID or post-COVID.  A Norwegian study found that symptoms persisted for 6 months in most patients (189 of 312), including most young adults.  (‘Long COVID’ preferred because ‘post-COVID’ is ambiguous.) 

What is long COVID? There are many symptoms, led by fatigue, loss of smell/taste, breathlessness, and cognitive impairment, with no obvious common cause or relationship. Increased blood clotting (prothrombotic) has been suspected, suggested by the involvement of the receptor for the COVID-19 spike protein, ACE2, but strong evidence and mechanisms have been lacking.  

These authors tested the blood of 21 patients with post-COVID syndrome (PCS), averaging nearly 2 years after onset of infection. They modeled real blood flow through vessels by collecting blood samples (treated with the anti-coagulant citrate) and sending it through narrow tubes coated with particular proteins. They looked at the binding of platelets, which are abundant cellular products that initiate blood clots upon being triggered by binding collagen. They found striking increases in platelets binding to collagen (shown, Figure 1a top panel). Antibodies against von Willebrand factor (VWF) produced equivalent binding though apparently with different patterns (middle panels). 

Figure 1A: Binding of platelets (yellow) from healthy control blood (left) or PCS blood (right) to collagen (top), anti-VWF (middle), or VWF (bottom).

 

Although these intriguing findings await confirmation (by others) and follow-up, of course, they are of the utmost importance given the enormous impact of COVID.  

Constantinescu-Bercu A, Kessler A, de Groot R, Dragunaite B, Heightman M, Hillman T, Price LC, Brennan E, Sivera R, Vanhoorelbeke K, Singh D, Scully M. Analysis of thrombogenicity under flow reveals new insights into the prothrombotic state of patients with post-COVID syndrome. J Thromb Haemost. 2023 Jan;21(1):94-100. doi: 10.1016/j.jtha.2022.10.013. Epub 2022 Dec 22. PMID: 36695401; PMCID: PMC9773628.  

Identifying Mitochondrial Functions by ‘Multiomic’ Profiling

Mitochondria make ATP through oxidative phosphorylation, thus providing energy for nearly all cellular functions. Many human disorders are attributed to mitochondrial dysfunction. Their functions seem narrow and their genomes, known for decades, were reduced to encoding a mere 13 proteins after transferring most to the nuclear genome. However, these authors note that ‘hundreds of mitochondrial proteins lack clear functions’. They previously (Stefely 2016) applied mass spectroscopy (MS) ‘multiomics’ to assign functions to mitochondrial uncharacterized (x) proteins (MXPs) in yeast. Here, they generated using CRISPR over 200 knockout (KO) cell lines, targeting 50 nuclear genes encoded MXPs plus 66 with known functions, and assessed in each line over 8,000 proteins, over 3,000 lipids and over 200 metabolites by MS (epic undertaking!). They found high reproducibility and dynamic range, with ‘many molecules showing regulation over 3-4 orders of magnitude’.

Fig 2e. Relative protein abundance in SLC30A9 KO cells compared to WT (“wild type”, i.e., normal) cells versus statistical significance with noted mitochondrial ribosome (black), OxPhos (blue), and mtDNA-encoded (red) proteins.

Some assessments confirmed expectations or were mild surprises, e.g., the importance of ALDH18A1 or NADK1 in proline synthesis (Fig 2a). Others revealed ‘new biology’, such as a key role for the putative zinc transporter SLC30A9 in mitochondrial ribosome and OxPhos proteins (Fig 2e, shown). They also found that one ’upstream (open) reading frame’ (PYURF) is a chaperone essential for complex I and coQ synthesis, linked a transporter (SLC30A9) to ribosomes, and found a second gene (RAB5IF) contributing to developmental disorders. They offer their “8.3 million distinct biomolecule measurements” online to help others ascribe additional functions, a promising resource.

Rensvold JW, Shishkova E, Sverchkov Y, Miller IJ, Cetinkaya A, Pyle A, Manicki M, Brademan DR, Alanay Y, Raiman J, Jochem A, Hutchins PD, Peters SR, Linke V, Overmyer KA, Salome AZ, Hebert AS, Vincent CE, Kwiecien NW, Rush MJP, Westphall MS, Craven M, Akarsu NA, Taylor RW, Coon JJ, Pagliarini DJ. Defining mitochondrial protein functions through deep multiomic profiling. Nature. 2022 Jun;606(7913):382-388. doi: 10.1038/s41586-022-04765-3. Epub 2022 May 25. PMID: 35614220; PMCID: PMC9310563.
NB PubPeer comment raises concerns regarding the methods, interpretations, and conclusions. 

Seasonal flu vaccination: short term protection, long term risk?

Influenza (flu) sickens millions and kills many thousands of Americans each year. Vaccination is intended to reduce the number cases and severity of illness. The flu virus changes its coat proteins each year, a ‘shape-shifting’ behavior that challenges the timely production of vaccines that are effective against flu variants. The US CDC evaluates vaccine effectiveness (VE) in thousands of outpatient respiratory illness patients, usually finding substantial vaccine protection with little risk Consequently, its Advisory Committee on Immunization Practices (ACIP) as well as the World Health Organization (WHO) recommend annual immunization of everyone over 6 months old unless contraindicated.
 
Here, the authors surveyed vaccination over 10 years among several thousand Japanese school children and several hundred school staff adults. Of the many forms of flu vaccine, Japan uses a quadrivalent (4 strains; trivalent prior to 2014) “split” vaccine based on influenza hemagglutinin (HA). “Split” vaccine means the virus was ‘disrupted’ by detergent (the equivalent of ‘heat killed’; viruses aren’t alive). The flu viruses used to prepare the vaccine are grown in eggs; alternative quadrivalent preparations are available for those with allergies to eggs, including Flublok and Flucelvax.

They found that morbidity was reduced in vaccinated elementary school students but elevated in middle school students (Fig 1). Most people who had been vaccinated ‘from infancy’ were also vaccinated in the 2019-2020 season (Fig. 2). Moreover, they “found that morbidity was significantly higher among elementary (P < 0.001) and middle (P < 0.05) school students who had been vaccinated since infancy than among those who had not been vaccinated since infancy” (Figure 3, shown).

Fig. 3 Relationship between morbidity and vaccination from infancy.

These data are self-reported via questionnaires, and therefore extra subjective. The authors propose no mechanism for how annual vaccination could cause increased morbidity. It seems probable that the association is not direct, not causative, but indirect through other behavioral or health status factors. This is an intriguing finding that should be analyzed and study that should be repeated.

Kajiume T, Mukai S, Toyota N, Kanazawa I, Kato A, Akimoto E, Shirakawa T. Effectiveness of seasonal influenza vaccine in elementary and middle schools: a 10-year follow-up investigation. BMC Infect Dis. 2022 Dec 6;22(1):909. doi: 10.1186/s12879-022-07898-y. PMID: 36474168; PMCID: PMC9724312.

Protein folding by AI: wrinkles

Tech giants Alpha and Meta (Google and Facebook) applied their Artificial Intelligence (AI) to fold proteins computationally, predicting 3-dimensional shapes from the 1-dimensional sequence data. Meta’s paper is still paywalled (preprint) but AlphaFold’s Nature papers from last year are available (Jumper, Tunyasuvunakool). 

The AlphaFold authors noted that the ~100,000 protein structures determined by conventional experimental means are a small portion of the “billions” extant in nature. Previous approaches “focus on either the physical interactions or the evolutionary history”, which they say relies on the availability of close homologues or works for (only) a few, small proteins and is otherwise “computationally intractable” (too hard). They evaluated in the 87 protein domains comprising the 14th Critical Assessment of (protein) Structure Prediction (CASP14) dataset, structures not yet deposited in the public Protein Data Bank (PDB). This permits a ‘blind’ (apriori) comparison of AI methods, by comparing their predictions with the newly-solved structures. 

Fig 1. a. Scores. b. Backbone. c. Side chains

 By this measure, AlphaFold is much better than its competitors (Fig 1a, shown, predicted vs experimental). It gains accuracy on backbone and side chains (1b, c) “by incorporating novel neural network architectures and training procedures based on the evolutionary, physical and geometric constraints of protein structures”. Using “multiple sequence alignments (MSAs) and pairwise” comparisons, it “predicts the 3D coordinates of all heavy atoms for a given protein using the primary amino acid sequence and aligned sequences of homologues as inputs”. So give it a bunch of similar sequences and structures and voila! it gives the ‘new’ one. Thy describe the process and you can download to code to inspect, modify, run yourself (open source).  

While impressive, this is a very constrained set of structures, nothing justifying the claims made in the popular press of solving all proteins. To be comprehensive, it seems that AI will have to consider biology, implement means of including the amino-terminal-first synthesis, nucleation, domain folding, insertion into a membrane, and above all interaction with chaperone proteins. 

Jumper J, Evans R, Pritzel A, Green T, Figurnov M, Ronneberger O, Tunyasuvunakool K, Bates R, Žídek A, Potapenko A, Bridgland A, Meyer C, Kohl SAA, Ballard AJ, Cowie A, Romera-Paredes B, Nikolov S, Jain R, Adler J, Back T, Petersen S, Reiman D, Clancy E, Zielinski M, Steinegger M, Pacholska M, Berghammer T, Bodenstein S, Silver D, Vinyals O, Senior AW, Kavukcuoglu K, Kohli P, Hassabis D. Highly accurate protein structure prediction with AlphaFold. Nature. 2021 Aug;596(7873):583-589. doi: 10.1038/s41586-021-03819-2. Epub 2021 Jul 15. PMID: 34265844; PMCID: PMC8371605.

Leptin boosts protective vaccine responses

Follicular helper T cells (Tfh) support B cell development and production of antibodies, essential for a protective vaccination response. Metabolism has been linked to T cell development and the metabolic hormone leptin varies up to 10-fold among healthy people. Here, the investigators asked whether leptin levels might influence T cell development and contribute to variability in vaccine responses. 

Within a cohort of 76 healthy adults, they found non-responders to influenza vaccination had on average 2.5-fold lower serum leptin levels, with non-responders 10-fold more frequent in the low leptin group (fig 1ab). Tfh counts correlate with leptin levels (fig 2). Similar observations were made among older flu vaccine recipients (age >64 yr) and young Hepatitis B vaccine (HBV) recipients. Adding leptin to T cells cultured in vitro increased Tfh markers and production of IL-21 (fig 2e). 

In mice, they found leptin in areas of B cell development and leptin receptors on Tfh cells. Leptin receptor deficiency reduced antibody responses (fig 3b, c) and (consequently) allowed viral growth (panel a) in mice infected with H1N1 influenza. Tfh in leptin-receptor-deficient mice produced less IL-21 (fig 5b) and supplemental IL-21 restored most antibody production (fig 5a). IL-21 production is abrogated in T cells lacking STAT3 (fig 5g), strongly supporting a mechanism involving STAT3 and IL-21. 

Fig 7. Leptin protects from fasting-induced susceptibility to influenza. 

 

They could transiently reduce serum leptin levels by ‘fasting’ (starving) mice on alternate days 5 to 15 days after infection with influenza (Fig 7a, shown above). This timing chosen to avoid interfering with T cell priming (d 0-5) and focus on peak Tfh development (starting d5). Supplemental leptin protected against influenza (panel b), underscoring the significance of this pathway.

Deng J, Chen Q, Chen Z, Liang K, Gao X, Wang X, Makota FV, Ong HS, Wan Y, Luo K, Gong D, Yu X, Camuglia S, Zeng Q, Zhou T, Xue F, He J, Wei Y, Xiao F, Ma J, Hill DL, Pierson W, Nguyen THO, Zhou H, Wang Y, Shen W, Sun L, Li Z, Xia Q, Qian K, Ye L, Rockman S, Linterman MA, Kedzierska K, Shen N, Lu L, Yu D. The metabolic hormone leptin promotes the function of TFH cells and supports vaccine responses. Nat Commun. 2021 May 24;12(1):3073. doi: 10.1038/s41467-021-23220-x. PMID: 34031386; PMCID: PMC8144586.

Physical activity increases gut bacteria diversity

Previous work established associations in humans between physical activity and reduced obesity, reduced mortality, and improved cardiovascular health. Physical activity has been also associated with the microbiome in animals. Here, the relationship between physical activity and microbiome in humans was investigated.

The authors studied a cohort of 720 adults, citizens of Wisconsin, average age 55 years, 83% White, 10% Black, 42% male. Gut microbial, (bacterial) composition was assessed using sequencing the V3-V4 region of 16S rRNA extracted from stool samples. Note this is only a subset of the ‘microbiome’, not include non-bacterial components such as fungi, viruses, etc. 

They monitored physical activity using accelerometers worn on the hip (activity) or wrist (sleep). Participants also self-reported whether in a typical week they walked or biked at least 10 minutes continuously to get around. Those who responded ‘yes’ were classified as participating in ‘active transportation’. Note this is a threshold of less than half a mile a week, walking only about 100 m per day. 

Table 2. Linear mixed effects models (adjusted for characteristics, Table 1). CI, confidence interval; SD, standard deviation; MVPA, moderate to vigorous physical activity; ** p<0.05. *** p < 0.01.

They identified 865 unique bacterial taxa, largely encompassed by about 20 abundant phyla (Fig 1). They observed no change in bacterial diversity in participants who engaged in moderate-to-vigorous activity (line 2, Table 2, shown) or active transportation (line 3). However, when they analyzed those participants who engaged in higher levels of active transportation, at least 1 standard deviation (SD) above the average, they observed significant increases in bacterial diversity (line 4).

They also found the abundance of an unknown family from order Clostridiales was associated with increased weekly MVPA minutes. They conclude that their results “point to a potential pathway by which the gut micro- biota may be linked to physical activity and other well established health benefits”.

Holzhausen EA, Malecki KC, Sethi AK, Gangnon R, Cadmus-Bertram L, Deblois CL, Suen G, Safdar N, Peppard PE. Assessing the relationship between physical activity and the gut microbiome in a large, population-based sample of Wisconsin adults. PLoS One. 2022 Oct 26;17(10):e0276684. doi: 10.1371/journal.pone.0276684. PMID: 36288361; PMCID: PMC9605031.

When viruses cohabit: Flu + RSV = Hybrid Frankenvirus

Experts worry that this winter might be made miserable by unwelcome visitors: something new, coronavirus variants, something flu (influenza A virus, IAV) and something blew, respiratory syncytial virus (RSV). What happens when somebody hosts IAV and RSV at the same time?

These investigators infected cultured human lung cells, A549 cells, and confirmed previous reports that coinfection reduces RSV but not IAV replication (Fig 1). Despite producing lower titer, they observed that infection with IAV appeared to increase the rate of coinfection by RSV.

Fig 3b. Filament with features of IAV and RSV.

IAV and RSV are enveloped viruses that bud from the cell membrane with characteristic glycoproteins hemagglutinin (HA) and fusion (F), respectively. Having detected HA in areas of RSV budding from coinfected cells, the authors hypothesized that some virions would contain components of both viruses. Indeed, they observed many filaments, typical of RSV, with proteins from both viruses, albeit segregated (Fig 2a-e). A remarkable scanning electron micrograph appears to show hybrid viral particles (HVP) budding from the filaments (2f, red arrows). They analyzed the hybrid buds using cryo-ET and were able to ‘segment’ features of both viruses (shown, Fig 3b): mostly IAV virions budding from mostly RSV filaments. 

Amazing biology, but what does it mean clinically? The authors found that the hybrid virions contained IAV capable of infecting cells that had been depleted of their sialic acids, which bind HA, by treatment with neuraminidase (NA), Fig. 4-5). This could be an important mechanism widening the range of infected cells.

Haney J, Vijayakrishnan S, Streetley J, Dee K, Goldfarb DM, Clarke M, Mullin M, Carter SD, Bhella D, Murcia PR. Coinfection by influenza A virus and respiratory syncytial virus produces hybrid virus particles.  Nat Microbiol. 2022 Oct 24. doi: 10.1038/s41564-022-01242-5. Epub ahead of print. PMID: 36280786.