A MULTI-HIT MODEL OF LONG COVID PATHOPHYSIOLOGY: THE INTERACTION BETWEEN IMMUNE TRIGGERS AND NERVOUS SYSTEM SIGNALING

ABSTRACT

Early in the pandemic, clinicians recognized an overlap between Long COVID symptoms and dysautonomia, suggesting autonomic nervous system (ANS) dysfunction. Our clinical experience at Johns Hopkins with primary dysautonomia suggested heritability of sympathetic dysfunction, manifesting primarily as hyperhidrosis and as other dysautonomia symptoms. Whole exome sequencing revealed mutations in genes regulating electrical signaling in the nervous system, thus providing a genetic basis for the sympathetic overdrive observed. We hypothesize that dysautonomia in Long COVID requires two molecular hits: a genetic vulnerability to prime the ANS and a SARS-CoV-2 infection, as an immune trigger, to further disrupt ANS function resulting in increased sympathetic activity. Indeed, Long COVID patients show signs of chronic inflammation and autoimmunity. We have translated this two-hit concept to the clinic using ion channel inhibitors to target genetic susceptibility and immunomodulators to treat inflammation. This multi-hit hypothesis shows promise for managing Long COVID and merits further study.

INTRODUCTION

As of October 2023, over 698 million individuals had been infected globally with SARS-CoV-2 resulting in approximately 7 million deaths (1). The medical community, as early as Fall of 2020, issued an international online survey in 56 countries suspecting the presence in many patients of persistent heterogeneous sequelae involving multiple organs (2). But it was even earlier on May 20, 2020, when Elisa Perego, a health researcher from Lombardy, Italy, dubbed her experience of symptoms after SARS-CoV-2 as “#Long COVID” in a hashtag on Twitter (3). She penned “Long COVID” as a contraction of “long-term COVID illness” to describe the cyclical, progressive, and multiphasic nature of her symptoms that had persisted for over 40 days. This patient-originated term quickly spread through social media, gaining popularity as it legitimized other patients’ experiences and challenged medical assumptions that SARS-CoV-2 was a transient, short-term disease. Long COVID, also now known as post-acute sequelae of SARS-CoV-2 infection (PASC) and “Long-haul COVID,” is currently defined as ongoing, relapsing, or new symptoms or conditions present 30 days or more after COVID-19 infection (4-6). In this article, however, we will honor the sentiment of the British Medical Journal’s opinion of October 1, 2021, entitled, “Why we need to keep using the patient made term ‘Long COVID,’” and its claim that “Long COVID has clearly struck a chord.” Henceforth, we will refer to PASC, as we have in our title, as “Long COVID” (7).

THE PREVALENCE AND BURDEN OF LONG COVID

Long COVID has become a major public health issue affecting at least an estimated 65 million people globally, with cases increasing daily (8). This figure is most likely an underestimation due to the continued high rates of Long COVID among new SARS-CoV-2 cases around the world. In the United States, as of January 2023, the percentage of people who reported Long COVID symptoms after SARS-CoV-2 declined from 19% in June 2022 to 11% in January 2023 (9). Despite this decline, the prevalence of Long COVID in those who have had SARS-CoV-2 is still high. Moreover, its prevalence seems to depend on the severity of the underlying SARS-CoV-2 infection with 7.5–41% of nonhospitalized patients affected and over 50% of hospitalized patients debilitated by symptoms such as severe persistent anxiety and depression (10-12). In the United States, in 2022 the CDC estimated that the prevalence of Long COVID was about 6.9% of all adults with women (8.5%) more likely than men (5.2%) ever to have had Long COVID (13). In fact, most studies of Long COVID have shown that the prevalence among females is statistically significantly higher than among males (14,15). Interestingly, U.S. Census Bureau data suggest higher rates of Long COVID symptoms among Hispanic and Black adults compared to White adults. A recent survey found 31.1% of Hispanic, 28.7% of Black, and 27.6% of White adults reported Long COVID symptoms (16).

Population-based surveillance tools have been effective in monitoring the impact of the disease on daily living in the United States with 25.3% of over 18,000 individuals with Long COVID responding that day-to-day activities were impacted “a lot” (14). Importantly, 28.9% of individuals responded that their SARS-CoV-2 infection had occurred more than 12 months previously (14). This has negatively affected employment with two surveys showing that approximately 66% of individuals with Long COVID who previously were actively employed were now unemployed or were forced to reduce their working hours (2,17). Mental health and cognitive burdens have been a substantial contributor to this inability to function at work with over 60% of individuals in the United States having a neurological or psychiatric diagnosis six months after SARS-CoV-2 (18). One early estimate calculated the reduction of labor supply due to Long COVID could be more than $50 billion annually (19). A recent review of 12 publications suggested that Long COVID patients are increasingly utilizing health care systems, both primary and specialist care, much more than before they were infected (20). Using claims data, a recent estimate suggests that these more frequent visits for care mean that Long COVID adds an additional $224 in health care costs per month over a six-month period in any individual patient (21). Thus, Long-COVID has emerged as a major public health problem that will have considerable long-term consequences for population health, economic policy, and research initiatives.

Due to its relative novelty, major knowledge gaps persist that impede our ability to treat Long COVID effectively. Therefore, in this article, we will focus on research efforts into our understanding of the pathogenesis of Long COVID and of its underlying biological mechanisms.

LONG COVID AND ITS COMPLEX SYMPTOMATOLOGY

The Association Between Long COVID and Dysautonomia

Early in the pandemic, several neurologists realized that many of the symptoms of Long COVID were, in fact, symptoms of dysautonomia, a dysfunction of the autonomic nervous system (ANS) or components thereof such as the sympathetic nervous system. At the National Institutes of Health (NIH)’s National Institute of Neurological Disorders and Stroke Autonomic Medicine Section, Dr. David Goldstein quickly realized within two months of the pandemic’s onset in April 2020 that the symptoms of SARS-CoV-2 that he saw manifested in patients were multisystem, multidisciplinary complaints. Dr. Goldstein surmised that they originated from abnormal activation of the ANS leading to autonomic dysfunction in various organs and thereby causing myriad functional abnormalities, a phenomenon that he described as the “extended ANS” (22). One year later, he strengthened his theory with a patient case history in an article entitled, “The Possible Association Between COVID-19 and Postural Tachycardia Syndrome (POTS)” (23). Aided by new innovations in studies such as the COVID Symptom Study, where by May 2020 2.8 million users had documented their symptoms on a smartphone tracker app, more and more clinicians quickly began to recognize that clinical dysautonomia was a prominent feature of Long COVID (24-30).

At our Johns Hopkins Center for Sweat Disorders, we also were observing a sharp spike in the number of patients complaining about an increase in primary focal hyperhidrosis (PFH) after SARS-CoV-2 infection. Often, during close questioning, these patients would admit that they had indeed suffered from various degrees of underlying hyperhidrosis since childhood, but that after contracting SARS-CoV-2, their sweating had become intolerable and always interfered with daily activities [highest score on the Hyperhidrosis Disease Severity Scale (HDSS)] (31). It also became equally evident that sweating was not the only debilitating symptom that was occurring post infection with SARS-CoV-2. In addition to PFH, an important dermatologic manifestation of ANS dysfunction, our typical patient would complain of a constellation of debilitating symptoms including generalized anxiety disorder, cognitive impairment or “COVID-brain fog,” chronic fatigue, shortness of breath, heart palpitations, hyperhidrosis, irritable bowel, migraine-type headaches, insomnia, as well as lightheadedness. Over the last 13 years, we had observed this phenomenon quite regularly in our Center for Sweat Disorders in patients who presented with hyperhidrosis having similar dysautonomia symptoms, not after infection with SARS-CoV-2, but rather after contracting infectious diseases such as mononucleosis (32-34), Lyme disease (35-38), and malaria (39,40). As with Long COVID, however, patients reported that the most pervasive symptoms were often the neurological comorbidities of anxiety, depression, and cognitive impairment (41-44). It is important to note here that our PFH patients over the years have also frequently presented with postural orthostatic intolerance syndrome (POTS), which often occurs in adolescence when it can be triggered by an antecedent infection (45). Intriguingly, during the pandemic, multiple scientific reports started to appear suggesting that the majority of patients with Long COVID met the diagnostic criteria for POTS (26,46-49), a link that has yet to be fully explored.

Unanswered Questions

Despite our long experience with both POTS and post-infectious dysautonomia, what remained puzzling was that patients tended to have only marginal improvement of their symptoms, well short of a full cure, when given even potent immunomodulatory agents such as intravenous immunoglobulin to treat the autoimmunity or neuroinflammation comorbid with their disease (50-52). This prompted two yet unanswered questions. First, What is the association between Long COVID and dysautonomia symptoms? Second, Why does the inflammatory response that typically occurs after an infection with Epstein–Barr virus, Borrelia, and now SARS-CoV-2 not provoke dysautonomia symptoms such as anxiety, depression, and cognitive impairment in everyone who contracts these pathogens?

Q1: What Is the Association Between Long COVID and Dysautonomia?

A possible foundation for an answer to this first question comes from our years of clinical experience at the Johns Hopkins Center for Sweat Disorders. Three of the most important lessons we have learned from caring for hyperhidrosis patients are (1) hyperhidrosis is not just a mere cosmetic issue; (2) most patients do not complain about being stigmatized or embarrassed because they are “sweat bothered” [Instead, these patients consistently describe an impaired quality of life (53,54) in which the adverse impact of extreme sweating is thought to be greater than that of other chronic dermatologic disorders such as psoriasis (53,55,56).]; and (3) hyperhidrosis almost never occurs alone as an isolated symptom, but as mentioned previously, is habitually part of a constellation of systemic dysautonomia complaints. In fact, we found that 38% of all PFH patients who presented to our clinic were on prescribed psychotropic medications for anxiety versus 14% of controls (57). Since no morphological differences exist between the sweat glands of hyperhidrosis patients and healthy controls (58), these clinical observations of multiple accompanying dysautonomia symptoms fit a hypothesis held for over 50 years that PFH is indeed a disease of primary sympathetic dysfunction (59,60). In other words, a sympathetic overdrive systemically affects every organ of the body. However, the skin, as the largest organ in the body, can manifest hyperhidrosis as an early diagnostic sign of abnormal ANS function that may herald diseases such as posttraumatic stress disorder (PTSD). Moreover, our clinical observation was that hyperhidrosis occurred at a young age, probably at birth, with frequent patient histories from mothers describing sweaty footprints of a patient as a toddler. Given this early age of onset and frequently encountered family histories of PFH, physicians have long suspected an autosomal dominant genetic transmission, but with incomplete penetrance (61-63). Indeed, our clinical observations suggest that PFH may not only be inherited, but also that many of the above-mentioned dysautonomia comorbidity complaints of hyperhidrosis patients may run in families.

Propelled by our clinical observations, we embarked on a search for genetic causes of hyperhidrosis that affect the ANS. We started to construct family history pedigrees to document that not only was hyperhidrosis prevalent in families, but also that many of the family members shared dysautonomia symptoms such as generalized anxiety, an affliction that is applicable to virtually all our patients. Figure 1 illustrates two typical families where the probands presented to our clinic with severe hyperhidrosis, family histories of hyperhidrosis, and multiple family members showing a wide array of dysautonomia symptoms including generalized anxiety, orthostatic intolerance, chronic fatigue, and chronic itch symptoms. Many of their symptoms are validated using established clinical questionnaires such as the HDSS, COMPASS-31, the Zung Self-Rating Anxiety Scale, and the Social Phobia InveNtory (SPIN), thereby removing as much as possible the potential bias by the observer (31,64-67). The high scores across all four of these surveys in multiple family members suggest impaired autonomic regulation in multiple organ systems. Intrigued by these findings, we started performing whole-exome sequencing on families with hyperhidrosis who visited our clinic and whose relatives exhibited non-syndromic autonomic dysfunction. Preliminary data resulting from these efforts suggest the presence of heritable missense mutations in genes that are nearly all associated with generating or transmitting electrical signals throughout the body (68). These genes encode proteins that are directly involved in membrane excitability such as (voltage-gated) ion channels, modulate other proteins that can generate or propagate electrical signals such as ancillary subunits or transcription factors, and may trigger electrical or morphological remodeling of excitatory tissues. Notably, mutations in these genes may lead to altered ANS excitability as is often observed in these families and can be labeled clinically as a “high sympathetic tone.” The existence of a gain-of-function mutation in a protein expressed in the sympathetic ganglia, for example, would provide a plausible genetic explanation for the pathogenesis of hyperhidrosis and its accompanying dysautonomia.

Fig. 1.

Pedigree of two families with the proband presenting with severe hyperhidrosis. symptoms are indicated with scores mentioned in pedigree. Black arrow indicates sequenced proband. Crossed out symbol indicates deceased family member. Abbreviations: CS, compass-31 score; spin, Social Phobia Inventory; hdss, Hyperhidrosis Disease Severity Scale; Zung, Self-Rating Anxiety Scale.

Q2: Why Does a Post-Infectious Inflammatory Response Not Always Trigger Dysautonomia?

A possible answer to the second question of why Long COVID only presents in a small subset of patients who contract SARS-CoV-2 led to the formulation of a two-hit hypothesis for how Long COVID and other postinfectious dysautonomia disorders may develop. We currently hypothesize that Long COVID requires two molecular hits. First is a genetic predisposition in genes that regulate electrical signaling, such as an ion channel gain-of-function variant expressed in the brain and/or sympathetic ganglia that can act as a genetic primer to alter neuronal membrane excitability. Second is a SARS-CoV-2 infection that serves as an immunologic trigger transforming the nervous system into an oversensitive state characterized by lower thresholds to activation and a lessened ability to extinguish the debilitating dysautonomia symptoms associated with Long COVID (22,23,28,69-71). Indeed, influential interoception theories such as the James-Lange hypothesis (72) state that exaggerated sympathetic responsivity can augment one’s vigilance to physiological feedback, leading to dysautonomia symptoms (73,74). Thus, potentially mild symptoms associated with the underlying genetic predisposition can be exacerbated by immunologic triggers, which offers a plausible answer to our second question of why only a subset of people who contract SARS-CoV-2 develop Long COVID.

Growing evidence suggests that chronic inflammation, as a hallmark of autoimmunity, plays a significant role in Long COVID, and this fits well with our two-hit hypothesis theory. As established in POTS, autoantibodies have been detected against cytokines, complement proteins, cell surface proteins, as well as self-antigens in patients with Long-COVID with the implication being that immune dysregulation can be the secondary trigger of sympathetic overactivation (29,46,75-77). Long-COVID patients also have persistent elevation, compared to the onset of SARS-CoV-2, of proinflammatory cytokines like TNF-α, IFN-γ, IL-1β, IL-6, and IL-13 with significantly decreased levels of IP-10 (78,79). Upregulation of some of these cytokines has been observed in patients up to at least eight months after SARS-CoV-2 infection (79). Another study showed 83% of Long COVID patients with signs of latent autoimmunity and 62% with polyautoimmunity (80).

TREATMENT

In our clinic, we have used our concept of a two-hit hypothesis to guide treatment of Long COVID patients. Our rationale is that there must be both an inhibition of the genetic propensity for electrical hyperexcitability as well as an immunomodulation of the autoimmunity to minimize a further sensitization of the central and/or sympathetic nervous system. One small case report of 12 patients with a remarkable amelioration of neurocognitive impairments due to Long COVID used a two-drug combination of the α2A-adrenoceptor agonist, guanfacine, and the antioxidant, N-acetylcysteine (NAC) (81). Importantly, we recently showed that guanfacine may also partially inhibit multiple voltage-gated Na⁺ (Na_V) channels (82), and NAC is an established immunomodulator reducing inflammation (81). In our clinic, we had been using guanfacine along with other inhibitors of voltage-gated ion channels successfully for years in patients with PFH, POTS, and dysautonomia. Since most Na_V channel inhibitors lack subtype specificity and can sometimes act on other ion channel families or even receptors, clinically we favor repurposing multiple drugs to achieve ion channel inhibition (83). We also knew through interviews that many of our patients experienced significant symptom relief with marijuana smoking. Since synthetic cannabinoids are known to inhibit many ion channel subtypes (84-87), we combined synthetic cannabinoid therapy off-label in the form of dronabinol with repurposed drugs that inhibited ion channels.

Success in translating our two-hit hypothesis to the clinic is illustrated by a patient who presented to our Center for Sweat Disorders in the Fall of 2018 with complaints of lifelong hyperhidrosis. Strikingly, this patient also suffered from multisystem constitutional symptoms of dysautonomia including pre-syncope, light-headedness, chronic cardiac palpitations, migraine-type chronic headaches, chronic fatigue, mild brain fog, irritable bowel symptoms, and problems with poor sleep. Since this patient seemed typical of an individual with a genetic predisposition to an increased sympathetic tone, in 2018, we treated his dysautonomia symptoms with Na_V channel modulators: oxybutynin, guanfacine, and dronabinol. After being maintained successfully for four years on these medications, his dysautonomia symptoms, especially his brain fog and chronic fatigue, acutely worsened after a SARS-CoV-2 infection. He presented to our clinic seven months later with Long COVID; by simply increasing his guanfacine and adding NAC, we were able to control his Long COVID symptoms effectively. His testimony three weeks after this treatment further reinforces the potential clinical efficacy of this two-step approach of inhibiting ion channels along with the immune response to SARS-CoV-2 in these patients:

“As I’ve made you aware, I’ve dealt with the struggle of Long COVID this entire year going back to when I got sick in November 2022. Initial symptoms lingered into the New Year, but then I started having severe setbacks in the following months that significantly impacted my quality of life. At times, I would feel so physically and mentally lethargic for days at a time that leaving the house was a major challenge. This lasted through the winter and spring months, and into June. However, I began to notice some improvement after following your directive of taking the N-Acetyl-L-Cysteine and increased dosage of the guanfacine, in addition to the dosages of dronabinol & oxybutynin that I’ve taken since 2018. Since that regimen began in June, the setbacks quickly began to fade, and I’m happy to report that I’ve felt better these past 30 days than I have since before my birthday in November 2022, when I last got COVID. In fact, aside from some brief fatigue episodes which can also be attributed to something else, I haven’t felt sick at all in almost two months. It’s been a tremendous relief that has come at a time that admittedly had left me somewhat desperate in terms of treatment due to how severe the setbacks were. I wanted to share my success story in case other patients you care for share similar characteristics of mine that can be treated.”

CONCLUSION

We propose that post-acute sequelae of SARS-CoV-2 or Long COVID may require two molecular hits: a predisposing genetic vulnerability such as ion channelopathies or mutations in genes that support electrical signaling and an immune reactivity to SARS-CoV-2 as a secondary trigger. We also hypothesized here that pharmacological targeting of both the genetic and environmental hit may, in combination, show efficacy in patients. With some initial success, we have translated this concept to the clinic and are treating patients with drugs such as guanfacine to target the genetic aberration and the neuro-inflammation with NAC. The full clinical implementation will need individual genetic screening of each patient as well as the manufacturing of more specific drugs that can inhibit the underlying genetic defect. The full application of this multi-hit hypothesis may not only have applicability to Long COVID but also to other postinfectious dysautonomic diseases.

DISCUSSION

Schiffman, Providence: How do you reconcile your data and your hypothesis with the serotonin work that has been done which shows that serotonin depletion, especially in the gut, is associated with Long COVID symptoms and signs? Paxlovid use may lessen such symptoms and signs, some believe.

Brock, Baltimore: I think it’s all part of the same puzzle actually. Why don’t serotonin levels decrease in everyone? Why do only 10% of us go on to get Long COVID? For those who suffer from Long COVID and who are in this room, go home and take a really good family history of your first- and second-degree relatives. I think that you may be surprised at what you find.

Cushman, Colchester: Great, thank you, that’s a great talk. Have you considered looking into large cohorts with genomics data to see if you find similar patterns outside of these families? In other words, will the findings hold in the general population with post-acute sequelae of COVID?

Brock, Baltimore: Yes, we are trying to do that because some of these Long COVID data are now accumulating, as you can imagine. Previously, they weren’t readily available as we were developing this story. We have only been contemplating this two-hit hypothesis concept for the last couple of years. We do think that this is about membrane excitability from ion channelopathies in general, and not specifically, just sodium channelopathies. We think there are more than likely lots of other mutations involved here. I think the important take-home message is that there seemed to be genetic disruptions in the autonomic nervous system—this really hasn’t been looked at. Many genetic variants may be responsible for causing genetic predispositions to Long COVID, and we are talking with several groups now to investigate the prevalence of these variants.

Kraft, New York City: Great talk. You know we’re seeing a lot of manifestations of Long COVID that thankfully seemed to be improving in some domains. That always makes me wonder about epigenetics and about what the virus may be doing in that space. Have you looked at any of those mechanisms at all?

Brock, Baltimore: Yes, another great question. Any time you have an environmental insult, you have to consider chromatin remodeling, that is, you have to think about the fact that epigenetics may play a major role in this Long COVID phenotype. Yes, we’ve thought about it. I actually work in a cancer epigenetics lab as well, so epigenetic aberrations are a frequent topic of discussion. As we had to start somewhere, just understanding the genetics aspect of the project has taken much time and effort. We plan on adding the epigenetics piece hopefully as time permits. Interestingly enough, there aren’t really many specific genetic inhibitors of these ion channels. Similarly, most of the epigenetic drugs that we use in cancer have multiple effects. They’re not specific so there’s still the possibility that even though you think you’re doing one thing, there may be unknown off-target effects. At this point, we don’t know as much as we would like to know.

Kraft, New York City: Right, well, it keeps you employed.

Brock, Baltimore: Thank you very much.

Palmer, New York City: It has now been shown that the microbiome is important in various mouse models with anxiety. Is it possible that the human microbiome may be involved at one of the factors producing anxiety in this syndrome?

Brock, Baltimore: That’s a great question, and it dovetails with the first question about serotonin. If you look at the serotonin paper, it talks a lot about the microbiome. For example, it mentions tryptophan, which is a precursor for serotonin. Only 1% of dietary tryptophan actually synthesizes serotonin in the brain so dietary tryptophan doesn’t really translate into brain serotonin. There are lots of things about the microbiome that we don’t understand. Lower levels of plasma serotonin do reduce serotonin synthesis somewhat, but dietary intake of tryptophan has little effect on brain serotonin. This study does show that the microbiome is important. I think we’re bringing to the table this idea of a genetic predisposition as a contributing factor to the proposed mechanism. We hypothesize that a genetic vulnerability is necessary pre-infection and that COVID is really showing us what we have had evidence of before in other post-infectious dysautonomia syndromes—an exploitation of a predisposition to membrane excitability by an immune response.

Feldman, Philadelphia: Incredible talk. Really, really nice work. I’m not sure you explained everything to me that I need to know so I’m going to send you a bunch of sequences when I get home. We took a slightly different approach; we looked at patients who went into the ICU and had cardiopathy or heart failure as a part of their presentation of COVID and our expectation when we sequenced them was that we were going to see a whole bunch of cardiopathy genes. Instead, what we found were two big hits: one was in the gene that caused a hearing loss and the other was a hypertrophic cardiomyopathy gene. We’re scratching our heads, but I’d love to let you take a look at what we found to see what you think of it.

Brock, Baltimore: That would be fantastic. As you saw, we have an American Heart Association Grant Collaborative Science Award, and Gordon Tomaselli is the real expert here. What we essentially hypothesized in that grant was that these ion channelopathy mutations are not just in the heart muscle but they are also in the autonomic nervous system as well. This is why as surgeons we perform successful sympathectomies for patients with refractory ventricular tachyarrhythmias. In fact, 70% of our patients who receive a bilateral thoracic sympathectomy for ventricular arrhythmias are completely free of any ventricular arrhythmia or need for another cardioversion over a year later. What is the mechanism driving that outcome? At present, we don’t know, but we’re hypothesizing that it’s the autonomic innervation of the heart that is a contributing factor to the problem.

Feldman, Philadelphia: Ok, thanks.

Prabhu, St. Louis: Beautiful presentation from a patient to the lab. I want to ask you about the link between autoantibodies and this dysautonomia syndrome. Are there any data or did you find in your own research if there are agonistic antibodies that activate the sodium channel that you’re investigating? I thought about this in the context of a patient who was recently referred to me who had dysautonomia symptoms following the mRNA vaccine—this could suggest that there’s antibody generation with agonistic stimulation of these pathways.

Brock, Baltimore: That’s a great question. My partner in crime, Dr. Frank Bosmans, is actively working on the answer. The preliminary answer is yes, we believe we do have some evidence that certain cytokines, complement, and other immune factors may be involved in promoting membrane excitability. But we also believe there are a host of these channelopathies in our patients susceptible to immunoreactivity caused by cytokines or different autoantibodies. In summary, other hypotheses about Long COVID etiology being related to viral reservoirs or autoantibodies or even immunity from vaccines are not incompatible with the pathophysiology of a predisposition to membrane excitability caused by an ion channelopathy being exacerbated by the immune response to a virus.