The human body is not sterile. Beyond the trillions of bacteria that make up the microbiome, scientists now recognize a vast community of viruses, collectively called the virome, that inhabits tissues, mucosal surfaces and the bloodstream. Most of these viruses, particularly the herpesviruses, establish lifelong latent infections after the initial exposure and are normally held in check by the immune system. What has remained unclear is whether these resident viruses behave differently in people experiencing acute disease, and in particular whether their reactivation plays any role in stroke, a condition long known to involve inflammation but rarely studied through the lens of virology.

To interrogate this question, the team, led by Michael D. Clarke and senior author Glen C. Jickling, performed RNA sequencing on blood samples from 37 patients who had suffered an ischemic stroke, the type of stroke caused by a blocked blood vessel in the brain, and 32 age- and sex-matched controls with no stroke history. RNA sequencing reads the full complement of genetic transcripts in a sample, both human and viral. The researchers aligned the millions of sequencing reads against the human reference genome first to strip out host sequences, and then against a comprehensive curated database of human virus genomes to detect and quantify any viral transcripts present. Applying strict quality filters, requiring that a viral species be found in more than 3 percent of samples and with raw read counts above a threshold of 2, the team identified transcripts from six distinct viral families.

The list of detected viruses reads like a catalog of humanity’s most common passengers: several members of the human herpesvirus family, adenoviruses, papillomaviruses, human pegivirus, respiratory syncytial virus, and human endogenous retrovirus K, an ancient retroviral fossil embedded in our own genome that is occasionally transcribed. But the most striking finding concerned the herpesviruses. When the researchers pooled the read counts from all human herpesviruses together, they found that stroke patients carried 2.13-fold more herpesvirus transcripts than controls. In other words, these viruses were not merely present; they were more transcriptionally active, actively manufacturing viral RNA in the hours and days surrounding the stroke.

Just as notable was the pattern of coinfection. Many people carry a single herpesvirus, such as herpes simplex virus, cytomegalovirus, or Epstein-Barr virus, but harboring several simultaneously is less common. The study found that stroke patients had, on average, 1.23 times more distinct herpesvirus species detectable in their blood than controls. This simultaneous reactivation of multiple herpesviruses suggests a coordinated loss of immune surveillance rather than a chance flare-up of one virus, hinting that the immune system of stroke patients may be broadly distracted or dysregulated at the moment of the cerebrovascular event.

The analysis then drilled down to the level of individual viral genes, and two names emerged from the noise. The first was UL95, a gene from cytomegalovirus, a herpesvirus that infects the majority of adults worldwide and typically causes no symptoms in healthy hosts. The second was EBNA2, a gene from Epstein-Barr virus best known for its role in driving B-cell proliferation and its association with multiple sclerosis and several cancers. Transcripts from both of these genes were significantly increased in the blood of stroke patients compared with controls. EBNA2 is particularly interesting to immunologists because it acts as a master regulator of latency, rewiring the host cell’s transcriptional machinery and orchestrating immune evasion. Its elevated expression in stroke blood raises the possibility that EBV-infected cells are not passive bystanders during a stroke but are actively modulating the host response.

The most provocative part of the study came when the researchers examined what was happening to human gene expression in the same samples. They found that a set of host genes already implicated in stroke biology, including APOE, the famous Alzheimer’s and cardiovascular risk gene; C3, a central component of the complement cascade that bridges innate immunity and inflammation; PDGF, a growth factor involved in vascular repair; and CXCL2, a chemokine that recruits immune cells to sites of injury, were differentially expressed specifically in stroke samples that contained high counts of UL95 or EBNA2 transcripts. This correlation between active viral gene expression and altered host immune signaling provides the first tentative mechanistic thread connecting the blood virome to the inflammatory response after stroke. It suggests, though does not prove, that reactivating herpesviruses may help shape how the immune system responds to a cerebrovascular event, and perhaps even influence how that event unfolds.

The idea that infections can precipitate vascular events is not new. Epidemiological studies have repeatedly observed that acute infections, including influenza and bacterial pneumonia, are followed in the weeks afterward by transient spikes in heart attack and stroke risk. Vaccination against influenza has been associated with reduced stroke incidence in some studies, and chronic infections have been proposed as contributors to atherosclerosis itself. What has been missing is a direct, unbiased look at which viruses are actually active in stroke patients’ blood, and that is precisely what metagenomic RNA sequencing now makes possible. Rather than testing for a handful of preselected pathogens with antibody assays or targeted PCR, this approach captures everything with viral RNA in the sample, revealing the virome’s composition without preconceptions.

The technical achievement should not be understated. Viral transcripts are vanishingly rare in a blood transcriptome dominated by human hemoglobin genes, immune cell transcripts and ribosomal RNA. Detecting them requires deep sequencing, careful bioinformatic alignment against large viral reference databases, and conservative filtering to eliminate false positives from sequencing artifacts or contaminants. The authors’ use of prevalence and count thresholds reflects this caution, ensuring that only reproducibly detected viral sequences were analyzed. The linearization of viral genomes for read mapping, described in the study’s supplementary methods, further addresses the technical difficulty of aligning short sequencing reads against circular or complex viral genomes.

The researchers are careful about interpretation, and so should readers be. This is an observational study: it cannot determine whether herpesvirus reactivation contributes to causing strokes, whether the stress of an acute stroke somehow wakes dormant viruses from their latency, or whether both phenomena are driven by a shared upstream factor such as systemic inflammation or an immune-triggering event. All three scenarios remain biologically plausible. Herpesviruses are exquisitely sensitive to physiological stress hormones like cortisol, which can suppress cellular immunity and permit reactivation, so a stroke itself, or the conditions leading up to it, might plausibly trigger the viral activity observed. Conversely, a viral flare could promote a prothrombotic, proinflammatory state that raises the risk of a clot forming in a cerebral artery, a mechanism that has been proposed for other infections.

Future work will need longitudinal designs, following people over time to see whether rising herpesvirus transcription precedes stroke onset, and larger cohorts to confirm these findings across diverse populations. The Canadian and American research teams behind the study, funded by the Canadian Institutes of Health Research, the NIH, the Heart and Stroke Foundation of Canada and the University Hospital Foundation, suggest that such work could eventually open new therapeutic avenues. If viral reactivation turns out to modulate stroke risk or outcome, antiviral drugs or vaccination strategies might one day join the arsenal of stroke prevention tools. For now, the study stands as a vivid reminder that stroke is not purely a disease of blood vessels and brain tissue, but an event that unfolds across the entire ecological landscape of the human body, including the ancient viruses we have carried within us for a lifetime.

The findings also add to a growing literature implicating Epstein-Barr virus in neurological disease, following recent large-scale evidence linking EBV infection to multiple sclerosis. Whether the virus plays any comparable role in cerebrovascular disease will be a question for the next generation of studies. What this work establishes firmly is that the human blood virome is a real and measurable entity, that it shifts detectably in acute illness, and that ignoring it may mean missing a layer of biology that matters for stroke.

Traumatic brain injury begins with a mechanical insult, but much of the lasting damage can arise later. The initial impact can rupture blood vessels, disturb the blood–brain barrier, damage axons and neurons, and trigger excitotoxicity, oxidative stress, and cellular injury. The resulting chemical environment activates microglia, the central nervous system’s resident immune cells, while the damaged barrier permits circulating immune cells to enter the brain. This inflammatory response is not inherently harmful: activated microglia can remove dead cells and release signals that support tissue repair. The problem is that activation may persist or become excessively pro-inflammatory, leading to the release of cytokines, chemokines, and reactive nitrogen and oxygen species that injure nearby neurons and amplify immune recruitment. The researchers describe this process using markers associated with pro-inflammatory and reparative states, including iNOS and CD80 on one side and CD206 and Arg1 on the other. They also emphasize that the familiar “M1” and “M2” labels are an oversimplification. Modern single-cell studies show that microglia and macrophages occupy a continuum of overlapping, changing states. The therapeutic goal, therefore, is not to force every cell into a rigid category, but to shift the injured tissue toward a less damaging and more repair-supportive inflammatory environment.

MAPK9 is a member of the mitogen-activated protein kinase family, a network of serine/threonine enzymes that converts cellular stress into changes in gene expression and behavior. MAPK9, or JNK2, can be activated by injury-associated signals and influence transcription factors, cytokine production, apoptosis, and other processes relevant to brain trauma. To test whether it was a useful intervention point, the team first screened several candidate genes in mouse bone-marrow-derived macrophages stimulated with lipopolysaccharide and interferon-gamma, laboratory signals that induce a strongly inflammatory state. Among the targets examined, Mapk9 suppression produced one of the clearest changes in cell phenotype. Treated macrophages displayed more CD206-positive cells and fewer CD80-positive cells than inflammatory control cultures. Gene-expression tests showed reductions in Ccl2, Ccl3, Ccl4, Ccl5, Cxcl1, and Il1b, molecules that help recruit or activate additional immune cells. The researchers then repeated the test in primary mouse microglia. In those cultures, CRISPR treatment reduced Mapk9 messenger RNA and lowered Nos2, Cd80, Ccl2, and Nlrp3 expression, while increasing Mrc1, which encodes CD206, and Arg1. Western blotting confirmed that the corresponding iNOS protein was also reduced. These experiments suggested that MAPK9 is not merely a marker of inflammation but a potential molecular lever for changing the response of injured myeloid cells.

The delivery system was designed to solve one of the central problems in brain gene therapy: getting a large, fragile molecular payload into the right cells without exposing the entire body. The nanoparticles contained phospholipids, cholesterol, an ionizable lipid used to package nucleic acids, polyethylene glycol-linked lipid, and a fluorescent lipid that allowed the researchers to track them. Inside the particles were Cas12a protein and a guide RNA designed to recognize the mouse Mapk9 gene. Cas12a is an RNA-guided nuclease related to the more widely known Cas9. Once inside a cell, the guide directs the nuclease to a matching DNA sequence, where it can cut the genome and disrupt the targeted gene. The nanoparticles were then decorated with an antibody against Iba-1, a protein expressed by microglia and macrophages. This surface modification was intended to increase the likelihood that the particles would interact with these myeloid cells. Physical characterization found particles about 160 nanometers in diameter, with antibody attachment changing their surface charge but not substantially altering their size. The treatment was given through the nose, 30 minutes after injury, at a dose containing 20 milligrams per kilogram of the total lipid and CRISPR formulation. Intranasal administration can provide access to the brain through olfactory and trigeminal pathways and associated tissue spaces, potentially reducing the systemic exposure associated with injection into the bloodstream.

The mouse experiments used adult male C57BL/6J mice subjected to a controlled cortical impact over the left motor and somatosensory cortex. The injury was produced with a three-millimeter impact tip moving at 3.25 meters per second to a depth of 1.5 millimeters. The investigators compared intranasal delivery with retro-orbital injection and examined the distribution of fluorescent nanoparticles in the injured hemisphere. After nasal dosing, roughly 90 percent of nanoparticle-positive cells were associated with Iba-1-positive cells, compared with about 55 percent after retro-orbital administration. Within the injured cortex, approximately 35 percent of fluorescent particle-positive cells were associated with Iba-1-positive myeloid cells, whereas about 13 percent were associated with NeuN-positive neurons. The result indicates preferential targeting, not exclusive targeting: some particles still reached cells that did not express Iba-1. RNA imaging showed that Mapk9 transcripts increased in the peri-lesional cortex after injury and that the signal was associated with Iba-1-positive cells. In treated mice, Mapk9 messenger RNA was visibly reduced, including within the Iba-1-positive compartment. A separate analysis of publicly available single-cell RNA-sequencing data supported the biological rationale. At 24 hours after injury, the fraction of Mapk9-positive microglia rose from 4.39 percent in uninjured samples to 8.01 percent after cortical impact, while Mapk9-positive bone-marrow-derived monocytes and macrophages increased from 4.90 to 6.93 percent. The proportion of Mapk9-positive neurons changed little, strengthening the case for concentrating treatment on myeloid cells.

The molecular changes were accompanied by alterations in the appearance and abundance of immune cells around the lesion. In untreated injured mice, Iba-1-positive cells became more numerous and hypertrophic, with enlarged cell bodies and shorter, less-branched processes. This morphology is commonly associated with activation. Compared with vehicle-treated injured animals, mice receiving the CRISPR nanoparticles had fewer Iba-1-positive cells, fewer hypertrophic cells, and smaller cell bodies. The composition of the myeloid response also shifted. The proportion of iNOS-positive Iba-1 cells fell, while the proportion expressing CD206 rose. RNA in situ hybridization detected lower levels of tumor necrosis factor and interleukin-1 beta messenger RNA in the injured cortex after treatment. These cytokines are potent inflammatory signals capable of affecting neurons, blood vessels, glial cells, and infiltrating leukocytes. The therapy was also associated with fewer TUNEL-positive cells, a measure of DNA fragmentation commonly used to estimate cell death. On the rotarod, injured mice normally fell sooner than they had before trauma, but treated animals remained on the rotating apparatus longer than vehicle-treated injured mice at one day after injury. However, the intervention did not significantly reduce the gross lesion volume at this early time point. That distinction matters: a treatment can influence inflammatory signaling and short-term function before it produces a measurable reduction in the amount of tissue lost.

The researchers also looked for signs that the nanoparticles caused harm outside the brain. Traumatic brain injury itself raised serum amyloid A, an acute-phase inflammatory protein, and treatment reduced this increase. Serum aspartate aminotransferase was elevated after injury but did not differ significantly between treated and untreated injured animals, while liver AST activity remained unchanged across groups. Microscopic examination of the liver, kidney, spleen, heart, and lungs found no obvious treatment-associated abnormalities. The mice showed no reported differences in body weight or overt behavior during the short observation period. These results provide an initial safety signal, but they do not establish that the approach is safe for long-term use. CRISPR nucleases can create unintended edits, and the study primarily measured suppression of Mapk9 messenger RNA rather than directly sequencing the genome to quantify editing efficiency or detect off-target changes. The antibody against Iba-1 also cannot distinguish resident microglia from infiltrating macrophages. That distinction could be important because the two populations arise from different sources and may have different effects on recovery. The work used only male mice, relatively small experimental cohorts, and a single acute time point, leaving open questions about sex differences, dose, treatment timing, repeated administration, and effects on cognition.

The study’s most striking feature is the combination of a programmable gene-editing payload, cell-enriching nanoparticle design, and a route of administration that does not require surgery or direct injection into the brain. Yet the results remain a proof of concept rather than evidence of a ready-to-use treatment for people with head injuries. The researchers will need to determine how long MAPK9 suppression lasts, whether it improves neurological recovery over weeks or months, and whether early motor benefits translate into preserved cognition and reduced neurodegeneration. Future experiments should directly measure genomic edits in purified nanoparticle-positive cells, map uptake across neurons, astrocytes, neutrophils, and peripheral organs, and test the system in both sexes and larger animal cohorts. Human-relevant models, including induced-pluripotent-stem-cell-derived microglia, could reveal whether the same pathway operates in human immune cells. The distinction between preferential and exclusive targeting will also be crucial for clinical development: Iba-1 is shared by multiple myeloid populations, and broad suppression of inflammatory signaling could theoretically interfere with useful immune functions. Even with those caveats, the findings offer a vivid example of how nanomedicine and CRISPR technology are converging. Instead of attempting to silence the entire post-traumatic immune response, the approach seeks to deliver a molecular instruction to the cells most responsible for sustaining inflammation, potentially turning the nose into an unexpected gateway for treating the injured brain.

Immune checkpoint inhibitors, which have transformed the treatment of various malignancies, function by liberating cytotoxic T cells from inhibitory signals, essentially releasing the brakes on the immune system to attack cancer cells. Despite their revolutionary potential, a significant portion of patients experience primary or acquired resistance, limiting the broad applicability of these therapies. Identifying the molecular culprits behind such resistance remains a critical challenge in oncology research, and the involvement of microRNAs—small regulatory RNA molecules known for their capacity to fine-tune gene expression—has increasingly attracted attention.

MicroRNA-25 (miR-25) has emerged from this study as a master regulator of immune suppression within the tumor microenvironment. Through meticulous experimental analysis, the researchers demonstrated that elevated levels of miR-25 in tumor cells correlate strongly with diminished immune checkpoint therapy response. The microRNA executes its immunomodulatory role by targeting Syndecan-3, a heparan sulfate proteoglycan expressed on immune cells that plays multifaceted roles in immune cell signaling and activation.

Syndecan-3 was previously recognized for its contributions to cell-cell and cell-matrix interactions, but its immunological role has been relatively underexplored. This study reveals that Syndecan-3 facilitates critical cross-talk between innate immune components such as natural killer (NK) cells and macrophages, alongside humoral immunity mediated by B cells. By downregulating Syndecan-3 expression, miR-25 effectively blunts both arms of the immune response essential for robust antitumor activity.

Functionally, the suppression of Syndecan-3 impairs the recruitment and activation of innate immune effectors, leading to reduced production of cytokines and chemokines that would otherwise potentiate the immune assault on cancer cells. Concurrently, B cell function is perturbed, compromising antibody-mediated targeting mechanisms that contribute to tumor eradication. This dual inhibitory effect cultivates an immunosuppressive niche that shields tumor cells from destruction despite immune checkpoint blockade.

The mechanistic insights were corroborated using a combination of in vitro assays and in vivo murine models genetically engineered to modulate miR-25 expression. Knockdown of miR-25 restored Syndecan-3 levels and reinvigorated immune function, overcoming resistance and enhancing tumor regression when combined with checkpoint inhibitors. Conversely, overexpression of miR-25 conferred resistance even in otherwise responsive tumors, underscoring its potent modulatory role.

Molecular pathway analysis further underscored the downstream consequences of Syndecan-3 repression, highlighting alterations in signaling cascades such as NF-κB and STAT pathways, which are crucial for orchestrating innate immunity and antibody production. This comprehensive signaling disruption effectively debilitates multiple layers of antitumor immunity, contributing to the resilience of resistant neoplasms.

Importantly, the clinical implications of these findings extend beyond mere mechanistic curiosity. Measuring miR-25 expression levels or Syndecan-3 status could serve as predictive biomarkers for patient stratification, identifying individuals at heightened risk of failing checkpoint blockade monotherapy. This stratification could guide personalized treatment regimens incorporating miR-25 inhibitors or Syndecan-3-targeted therapies.

Potential therapeutic strategies derived from this work envision employing antisense oligonucleotides, small molecule inhibitors, or CRISPR-based approaches to specifically reduce miR-25 levels in tumor cells or augment Syndecan-3 activity on immune cells. Such interventions might reinstate effective immune surveillance and responsiveness, turning resistant tumors into immunologically susceptible ones.

Further research is warranted to delineate the full spectrum of immune cell subsets influenced by the miR-25/Syndecan-3 axis and to explore combinatorial regimes that optimize therapeutic benefit. Exploring synergy with existing immune modulators and conventional treatments such as chemotherapy and radiation could enhance overall outcomes.

This study exemplifies the power of integrating molecular biology, immunology, and translational research to unravel complex resistance mechanisms. As immunotherapy continues to evolve, overcoming intrinsic and acquired resistance remains paramount for extending its benefits to a wider patient population, particularly those battling aggressive and refractory cancers.

With microRNAs increasingly recognized as critical players in cancer biology, the identification of miR-25 as a resistance driver represents a significant advance. By targeting fundamental regulatory nodes, it offers a promising strategy to circumvent immunosuppression and reinvigorate the immune army against tumors.

As cancer immunotherapy pivots towards more precision-based approaches, understanding molecular determinants like miR-25-induced Syndecan-3 repression will be central to tailoring interventions and improving durability of response. This research paves the way for innovative and potentially practice-changing modalities that enhance the immune system’s power to conquer cancer.

The implications resonate beyond oncology, as elaborating microRNA-mediated immune modulation could inform treatments for infectious diseases, autoimmunity, and transplantation medicine. Harnessing these molecular insights might recalibrate immune responses in diverse pathological contexts.

In summary, Zhu and colleagues have unlocked a novel molecular mechanism underlying immune checkpoint therapy resistance, with miR-25 acting as a suppressor of both innate and humoral immunity through Syndecan-3 downregulation. This work charts a compelling course for new therapies aimed at dismantling tumor immune evasion and boosting the effectiveness of life-saving immunotherapies.

Subject of Research:
The role of microRNA-25 in mediating immune checkpoint blockade resistance through suppression of innate and humoral immunity by targeting Syndecan-3.

Article Title:
microRNA-25 drives immune checkpoint therapy resistance by repressing innate and humoral immunity via Syndecan-3.

Article References:
Zhu, Z., Han, W., Deng, Y. et al. microRNA-25 drives immune checkpoint therapy resistance by repressing innate and humoral immunity via Syndecan-3. Nat Commun (2026). https://doi.org/10.1038/s41467-026-73339-y

Image Credits: AI Generated