HSV-1 hijacks peroxisome biogenesis in human neurons and brain organoids

What the work claims

The central claim is that HSV-1 co-opts peroxisomal biogenesis and lipid metabolism to promote replication in human neurons, rather than simply triggering peroxisomes as part of an antiviral defense.¹ The authors report that infection of SH-SY5Y neuroblastoma cells upregulates the peroxisome biogenesis regulator PGC-1α and the peroxisomal import factors PEX13, PEX14, and PEX19, increases peroxisome number, and alters peroxisome morphology. Pharmacologically increasing peroxisome abundance with 4-phenylbutyrate (4-PBA) boosts infection, while blocking PEX3-PEX19-dependent biogenesis with NCC 55-0396 suppresses it. Lipidomics shows selective increases in plasmalogens and sphingolipids. The same remodeling appears in hiPSC-derived neurons and in human cortical organoids, is tied to productive replication and reactivation, and is absent during latency. The broader implication is that peroxisome-derived ether lipids are a metabolic dependency HSV-1 exploits in neurons.

How it works

Peroxisomes are small metabolic organelles that, among other tasks, synthesize plasmalogens, a class of ether phospholipids abundant in neuronal membranes and myelin. The authors start in SH-SY5Y cells because they are easy to manipulate, then move to hiPSC-derived neurons and 42-day cortical organoids as more physiologically relevant models. In SH-SY5Y cells infected with HSV-1 strain McIntyre at an MOI of 1, quantitative PCR shows PGC-1α mRNA rising roughly 100-fold by 24 hours post-infection, PEX13 up approximately 2-fold at 4 hours and 5-fold at 48 hours, PEX14 up about 4-fold at 48 hours, and PEX19 up about 2-fold at 16 hours and 10-fold at 48 hours. Immunofluorescence for the peroxisomal marker PEX14 reveals a 14% increase in peroxisome number per cell at 24 hours, along with enlarged and more elongated organelles.

To test whether this remodeling is pro-viral or anti-viral, the authors treat cells with 4-PBA, a known peroxisome proliferator, or with NCC 55-0396, which blocks the PEX3-PEX19 interaction required for peroxisome membrane protein import. 4-PBA raises peroxisome abundance from about 41.5 to 60.4 arbitrary fluorescence units per cell, increases the percentage of infected cells by roughly 50%, and approximately doubles viral titers. NCC 55-0396 cuts the fraction of HSV-1-positive cells from about 80% to less than 5% and reduces viral yield. The same directionality is seen in AGPS-knockout SH-SY5Y cells: deleting the alkylglycerone-phosphate synthase required for plasmalogen synthesis reduces infection by about 50%, and 4-PBA no longer enhances infection when plasmalogen synthesis is disabled.

Lipidomics at 16 hours post-infection shows a 40% increase in PE- and PC-plasmalogens and a 2.5-fold increase in ceramide, with concurrent rises in sphingomyelin and total PC. In hiPSC-derived neurons infected with the neurotropic HSV-1 17syn+ strain, PEX13 transcript doubles by 6 hours and PEX14 fluorescence rises by about 55% at 12 hours. In cortical organoids, the remodeling is observed during productive infection and upon reactivation from latency, but not during latency itself, suggesting the metabolic dependency is linked to active replication rather than viral persistence.

Where a skeptic should push

The single most load-bearing assumption is that the peroxisome-plasmalogen pathway is a specific metabolic dependency of HSV-1 in neurons rather than a generic stress or lipid-remodeling response that accompanies many viral infections. The comparison with Zika virus helps: 4-PBA reduced Zika infection and titers by 95%, the opposite of its effect on HSV-1. That suggests the pro-viral role is not universal. Still, one control virus is not enough to establish specificity across the herpesvirus family or across enveloped viruses more broadly.

Second, SH-SY5Y is a neuroblastoma line with aberrant metabolism. The authors correctly move to hiPSC neurons and organoids, but most of the quantitative dose-response, knockout, and lipidomic work is in SH-SY5Y cells. The organoid data are presented as confirmatory; the exact number of organoids, donors, and replicates is not stated in the accessible main text, and the latency-reactivation protocol comes from the same laboratory, which is a strength for consistency but also a potential bias.

Third, NCC 55-0396 and 4-PBA have pleiotropic effects. 4-PBA is a histone deacetylase inhibitor and chemical chaperone, not a clean peroxisome-selective agent. NCC 55-0396 was reported to block PEX3-PEX19 interaction, but off-target effects on other cellular membranes or signaling pathways are possible. The AGPS knockout is the cleanest genetic evidence, yet even there the infection reduction is about 50%, not the near-complete block seen with NCC 55-0396, which hints that the small molecule may be acting through additional mechanisms.

Fourth, the latency model uses 5-bromovinyl-deoxyuridine and interferon-alpha to suppress lytic replication, followed by sodium butyrate or LY294002 to reactivate. Those reagents themselves alter metabolism and signaling. Showing that peroxisomal remodeling re-emerges upon reactivation is informative, but it does not prove that the same pathway operates during natural latency in human trigeminal ganglia, where viral genomes persist for decades with a very different cellular environment.

Peroxisomal dependency as a brain-organoid antiviral target

For organoid models of human organs and the drug-discovery work built on them, the opportunity is to repurpose brain organoids as a screening platform for host-directed antivirals. Most antiviral development targets viral enzymes, which is effective but vulnerable to resistance and often narrow in spectrum. A host metabolic dependency such as peroxisome biogenesis offers a higher genetic barrier to resistance and could be relevant across HSV-1 strains. Brain organoids that recapitulate the peroxisome-plasmalogen response could be used to test whether candidate antivirals block the metabolic remodeling or the replication that depends on it, particularly for drugs aimed at the central nervous system where traditional cell lines fail to model neuronal lipid metabolism.

The non-obvious implication is that organoid fidelity may now need to be judged by metabolic state, not just morphology and electrophysiology. A brain organoid that looks structurally neuronal but has immature peroxisomal function or atypical plasmalogen composition could give misleading antiviral readouts. Conversely, this paper provides a functional benchmark: an organoid that responds to HSV-1 with PEX13/PEX14 upregulation and plasmalogen production is, in at least one metabolic dimension, behaving like the infected human brain tissue the model is meant to represent. That could become a validation marker for organoid protocols intended for infectious-disease or neurodegeneration research.

The genuine threat is therapeutic window. Peroxisomes and plasmalogens are essential for normal neuronal function, myelination, and redox homeostasis. A drug that broadly suppresses peroxisome biogenesis could be neurotoxic, especially during development or in a brain already stressed by aging or amyloid pathology. The organoid model can help measure this trade-off by comparing antiviral potency against neuronal viability, lipid homeostasis, and synaptic markers in the same 3D tissue. If the window is narrow, the mechanistic insight may remain academically interesting without becoming a drug target.

Another threat is overextension to Alzheimer’s disease. The paper notes the HSV-1-Alzheimer’s association and suggests recurrent reactivation could drive peroxisomal and ether-lipid reprogramming that contributes to neurodegeneration. That is a plausible hypothesis, but the organoid latency model lasts days, not decades, and does not capture the amyloid-tau-inflammatory cascade of human aging. Brain organoids could be used to test whether HSV-1 reactivation alters secreted Aβ or phosphorylated tau, but those experiments are not reported here. The drug-discovery implication is antiviral; the neurodegeneration implication remains speculative.

The bottom line

Established: HSV-1 infection of SH-SY5Y cells, hiPSC-derived neurons, and human cortical organoids upregulates PGC-1α and PEX genes, increases peroxisome number, raises plasmalogen and sphingolipid levels, and depends on peroxisome biogenesis for efficient replication. Pharmacological enhancement of peroxisomes promotes infection, while PEX3-PEX19 inhibition or AGPS knockout suppresses it. Hypothesis: that peroxisome-derived ether lipids are a druggable host dependency for HSV-1 in the nervous system. What would confirm the case is a structure-activity relationship showing that antiviral potency correlates with target engagement of PEX3-PEX19 or AGPS, plus efficacy in an animal model of HSV-1 encephalitis or reactivation without unacceptable neuronal toxicity. What would break the case is finding that the small-molecule effects are off-target, that primary human neurons do not require plasmalogens for HSV-1 replication, or that blocking the pathway is too toxic to be useful. The mechanism is solid; the therapeutic translation needs careful pharmacology.