Mechanisms of viral inflammation and disease in humans

Abstract

Disease and accompanying inflammation are uncommon outcomes of viral infection in humans. Clinical inflammation occurs if steady-state cell-intrinsic and leukocytic immunity to viruses fails. In the confusing battle between a myriad of viruses and cells, studies of human genetics can separate the root cause of inflammation and disease from its consequences. Single-gene inborn errors of cell-intrinsic or leukocytic immunity underlying susceptibility to diverse infections in the skin, brain, or lungs can help to clarify the human determinants of viral disease. The genetic elucidation of immunological deficits in a single patient with a specific vulnerability profile can reveal mechanisms of inflammation and disease that may be triggered by other causes, inherited or otherwise, in other patients. This human genetic dissection of viral infections is giving rise to a new understanding of biology and new medicine.

Clinically apparent tissue inflammation during viral infection attests to an excessive influx of leukocytes, and activation of tissue-resident leukocytes. Systemic inflammation reflects an excessive activation of circulating leukocytes, particularly in response to viremia. These two types of clinical inflammation can co-exist. They result from a failure of cell-intrinsic immunity to control the intracellular virus in leukocytes, or, more commonly, in other cells. They also result from a failure of steady-state, subclinical leukocytic immunity, which is normally mediated by tissue-resident or circulating leukocytes, to suppress extracellular viruses and other virus-infected cells. Clinical inflammation may alleviate or aggravate disease, may have no effect on disease or may both alleviate and aggravate disease at different time points, in ill-defined ways, depending on the patient.

Most viruses rarely cause severe disease, or severe inflammation, in human populations (1). Viruses causing life-threatening disease in >1% of infected individuals in the absence of treatment are rare at any given time in history. There are currently no more than a dozen such virulent viruses, including rabies, smallpox, the Ebola, Marburg, and Nipah viruses, hantavirus, some influenza viruses, human immunodeficiency virus (HIV), severe acute respiratory syndrome coronavirus (SARS-CoV), Middle-East respiratory syndrome (MERS)-CoV, and SARS-CoV-2, the most recent addition to the list. By contrast, a myriad of viral pathogens cause life-threatening illness in <1% of humans, with most posing a threat to <0.1% of infected individuals.

Improvements in hygiene, and vaccination against smallpox, yellow fever, influenza, polio, mumps, rubella, hepatitis A and B, and human papillomavirus (HPV)-driven cervical cancer, have contributed to this decline in life-threatening viral illnesses, as, it is hoped, will the current program of vaccination against SARS-CoV-2. Evolution has probably done the rest, as in the case of non-viral infectious diseases (1). Consistent with this hypothesis, homozygosity for fucosyltransferase 2 (FUT2) loss-of-function variants, which underlie the non-secretor phenotype, is currently common in all populations because it confers protection against intestinal viruses (2). Likewise, homozygous C-C motif chemokine receptor 5 (CCR5) deletion confers resistance to infection with HIV (3). If HIV had been spreading uncontrollably for millennia, the surviving population would probably have gradually become enriched in CCR5-deficient, naturally resistant individuals. Overall, only a minority of individuals infected with most viruses develop life-threatening disease. Nevertheless, there are millions of deaths from viral infection annually, due to the high diversity of viruses and the large number of infected patients.

What are the human determinants of virus-mediated inflammation, disease, and death? Viruses, like other pathogens, are necessary but not sufficient to trigger disease. This is the ‘infection enigma’ (4). Several acquired immunodeficiencies, caused by immunosuppressive drugs or HIV, for example, favor the development of viral diseases (5). These deficiencies typically target adaptive immunity, particularly that mediated by T cells. Moreover, they generally underlie ‘opportunistic’ viral diseases (seen exclusively or mostly in patients with identifiable immunodeficiencies), and have been implicated in only a small subset of ‘idiopathic’ viral diseases (seen in patients without known immunodeficiency). Indeed, most tissue-specific, severe viral diseases affecting otherwise healthy individuals are not hallmarks of acquired immunodeficiency. Thus, despite the major insight provided by these immunodeficiencies, the determinants and mechanisms of most life-threatening viral diseases in humans remain unexplained.

Single-gene inborn errors of immunity (IEI) have proved more useful for shedding light on virus-induced disease because they have been shown to derail a greater variety of molecular and cellular pathways of host defense. In the 1950s, IEI were thought to be collectively rare and to underlie multiple infections, including viral diseases, in individual patients (1). Beginning in 1996, many new IEI were found to underlie specific infections, including a growing diversity of viral illnesses in otherwise healthy individuals (1) (Table 1). These IEI, and others underlying different types of infection, are more individually diverse and collectively common than initially appreciated, in part because they remain clinically silent until their phenotype is revealed by a specific infection. These studies were inspired by pioneering investigations in plants (6) and mice (7, 8), which revealed the essential role of cell-intrinsic immunity to viruses (notably in cells other than leukocytes), extending host defense from the ‘immune system’ (innate and adaptive leukocytes) to the whole organism (9).

Table 1: Human inborn errors of immunity to viruses.

This table presents inborn errors of immunity conferring a selective predisposition or resistance to a given virus resulting in a specific clinical phenotype, with the corresponding mutated genes and mode of inheritance. Only inborn errors that confer selective predisposition or resistance to viral infections are included. The much broader field of inborn errors underlying multiple viral infections, including inherited DOCK8 deficiency, GATA2 haploinsufficiency, and CXCR4 gain of function, has been reviewed elsewhere (68).

Clinical infections	Virus responsible	Mutated human gene	Inheritancea
Skin and mucosae
“Tree-man” syndrome	HPV-2	CD28	AR
Recalcitrant warts	HPV-4	CD28	AR
Epidermodysplasia verruciformis (EV)b	Human beta-papillomavirus (HPV)	EVER1, EVER2, CIB1	AR
Recurrent respiratory papillomatosis	HPVc	NLRP1	ARd
Kaposi sarcoma	Human herpes virus 8	OX40	AR
Brain
Forebrain encephalitis	Herpes simplex virus-1 (HSV-1)	UNC93B, IFNAR1 TLR3, TICAM1 TRAF3, TBK1, IRF3, SNORA31	AR AR, AD AD
	Varicella zoster virus (VZV)	POL3RA, POL3RC	ADe
Brainstem encephalitis	HSV-1, influenza virus, norovirus	DBR1	AR
Mollaret’s meningitis	HSV-2	ATGA4, LC3B2	AD
Lung
Critical influenza	Seasonal influenza virus Avian influenza virus	IRF7, IRF9 TLR3 MX1	AR AD AD
Critical COVID-19f	SARS-CoV-2	IFNAR1, IRF7 TLR3, UNC93B1, TRIF, TBK1, IRF3, IFNAR2 TLR7	AR, AD AD XR
Recurrent/severe infections	Rhinovirus, other respiratory viruses	IFIH1	AR, AD
Others
X-linked lympho-proliferative disease; B-cell lymphoma	Epstein-Barr virus	SH2D1A, XIAP, MAGT1 CD27, CD70, ITK, CTPS1, TNFRSF9	XR AR
Fulminant hepatitis	Hepatitis A virus	IL18BP	AR
Adverse reaction	Live measles and yellow fever vaccines	IFNAR1, IFNAR2, STAT1, STAT2, IRF9	AR
Disseminated disease	Cytomegalovirus	NOS2	AR
Resistance to infection	Human immunodeficiency virus-1 (R5-tropic)	CCR5	AR
Resistance to infection	Norovirus	FUT2	AR

^aAR: Autosomal recessive; AD: Autosomal dominant; XR: X-linked recessive.

^bTypical EV underlies a specific vulnerability to skin-tropic β-HPVs. Atypical EV, not covered here, refers to patients who are also vulnerable to various other infections, typically due to inborn errors of T cells.

^cRecurrent respiratory papillomatosis is usually due to HPV-6 or HPV-11. No HPVs were identified in the two siblings homozygous for the gain-of-function NLRP1 variant.

^dThe NLRP1 allele is gain-of-function, whereas all other alleles presented in this table are loss-of-function.

^eInheritance is monogenic or digenic. One patient with both POL3RA and POL3RC mutations had VZV pneumonia.

^fAutoantibodies neutralizing type I IFNs mimic inborn errors of type I IFNs and can underlie critical COVID-19 and adverse reactions to live attenuated yellow fever vaccine.

This human genetic approach is recent. The classic view of viral inflammation and disease in humans was based on observational studies in infected humans and experimental studies in animal models. Combined virological, immunological, pathological, and clinical studies of patients during natural infections have provided a wealth of knowledge about the spread of viruses in the human body and the ensuing tissue and systemic inflammation, with its innate [myeloid and natural killer (NK) lymphocytes] and adaptive (T and B lymphocytes) leukocytic components. In parallel, experimental inoculations of animals, particularly mice, have made it possible to analyze the molecular and cellular mechanisms involved in more detail. The possibility of manipulating the genetic background of both the virus and the animal in reproducible conditions has enabled scientists to make great strides forward. However, the inoculation of single homozygous knockout strains of mice carrying various other deleterious mutations with viruses that only rarely display a natural tropism for and co-evolution with mice is unlikely to reproduce the characteristics of natural infection in very large numbers of individually different humans (10–13). Recent progress in human genetics has transformed this immense heterogeneity from a formidable obstacle into a unique asset for understanding viral infections (14). In this Review, we discuss how the study of IEI has revealed general mechanisms of inflammation and disease in response to viral infections of the skin, brain, and lungs.

Skin and mucosal lesions

Human papillomaviruses (HPVs).

Cutaneous warts, condylomas, and respiratory papillomas develop on stratified squamous epithelia and can progress to cancers (Figure 1). These cellular proliferations are triggered by HPVs, double-stranded DNA (dsDNA) viruses that exclusively infect keratinocytes (15). The completion of their viral life cycle involves both early (E) and late (L) viral genes and requires all layers of the epithelium. There are more than 200 genotypes of HPVs, grouped into five genera. Some, but not all, HPVs are oncogenic. Skin-tropic β-HPVs cause no visible lesions in most people. By contrast, ‘common warts’, typically triggered by α-HPVs (HPV-2) or γ-HPVs (HPV-4) occur in >20% of the general population, at some point during the individual’s lifetime, usually disappearing within a few weeks or months. Histological analyses have revealed the presence of inflammatory leukocytes, including T cells, in the dermis underlying warts, whereas both keratinocytes and steady-state T cells probably control these viruses in lesion-free skin. Indeed, patients with inherited or acquired deficits of T cells or myeloid antigen-presenting cells (APCs) often have persistent or recurrent, frequently disseminated, ‘recalcitrant’ warts, together with many other infections (16).

Figure 1. Inborn errors of immunity underlying severe cutaneous HPV infections.

Inborn errors of immunity underlying severe α/γ-HPV infection (red), β-HPV infection (blue), or both (in red/blue) are shown. Mutations impairing T-cell activation confer a predisposition to infections with α/γ-HPVs, which can underlie common warts or giant horns, and with β-HPVs, which can underlie flat warts or tinea versicolor-like lesions, and can progress to non-melanoma skin cancer. Inborn errors of immunity that impair the T-cell CD28 costimulation pathway (CD28, CARMIL2, CARD11, MAGT1) result in a predisposition to α/γ-HPV infections. EVER-CIB1 mutations impair keratinocyte-intrinsic immunity and confer a specific predisposition to infection with β-HPVs. GATA2 is not shown here because it is expressed by hematopoietic progenitors, and its deficiency can confer predisposition to both α/γ− and β-HPV infections due to progressive leukopenia. Inborn errors of immunity partially impairing TCR rearrangement (RAG1, RAG2, LIG4, NHEJ1, ATM, DCLRE1C, SMARCAL1) or T-cell differentiation (IL7, IL2RG, JAK3) are not shown, but can also underlie both α/γ- and β- HPV infections. APCs: antigen-presenting cells.

Inherited dedicator of cytokinesis protein 8 (DOCK8) deficiency confers a predisposition to almost all viral infections of the skin, because DOCK8-deficient T cells are insufficiently flexible to migrate in this collagen-rich environment (16). Instead, these T cells die by mechanical fragmentation, or ‘cytothripsis’, in high-density skin tissues (17). Another IEI of interest is warts, hypogammaglobulinemia, infections and myelokathexis (WHIM), due to heterozygous gain-of-function C-X-C motif chemokine receptor 4 (CXCR4) mutations. Warts are common in patients with these multiple myeloid and lymphoid abnormalities, and their pathogenesis probably involves a myeloid APC deficiency (18). Likewise, haploinsufficiency for GATA-binding protein 2 (GATA2) probably also underlies warts through myeloid APC defects (19). By contrast, IEI not affecting myeloid APC or T cells, such as selective disorders of granulocytes and/or B cells, have been associated with many infections, but not common warts (20). In general, IEI that disrupt T-cell adaptive immunity in the skin can underlie inflammatory, recalcitrant common warts, indicating that steady-state cutaneous APCs and T cells are required for the subclinical control of the causal HPVs.

Giant horns and ‘tree-man’ syndrome.

In rare cases, recalcitrant common warts may strike otherwise healthy subjects. Even more rarely, they may progress to ‘giant horns’, underlying the ‘tree-man’ phenotype (21). Only five unrelated cases have been reported to date, four of which were caused by HPV-2. In an Iranian adult, the causal genetic disorder was found to be an autosomal recessive deficiency of the T-cell costimulatory molecule CD28 (21) (Figure 1). This is surprising, because a keratinocyte-intrinsic deficiency was predicted based on both the severity and specificity of the lesions. Moreover, given the key role of CD28 in T-cell activation, its deficiency was predicted to underlie susceptibility to many infections, viral and otherwise. The giant horns of the CD28-deficient patient are driven by episomal (not integrated into the human genome) and wild-type HPV-2 (viruses without variants that increase virulence). They form a multifocal benign tumor without human somatic mutations, but with devastating consequences, due to the expression of viral oncogenes in the basal layer of the epidermis. Two CD28-deficient relatives of the patient suffered from disseminated warts. Skin lesions remain present in the youngest of these patients, an adolescent, but most of the lesions of the adult patient regressed spontaneously. The current lesions in these two patients are driven by HPV-4.

Based on the findings for these patients, we can conclude that the CD28 T-cell pathway is required for protective cutaneous immunity to at least two HPVs, from two different genera. Nevertheless, immunity to HPV-4 is impaired and delayed, but not entirely abolished, in the absence of CD28; it is probable, but not definitively proven, that CD28-deficient T cells eventually drove spontaneous wart regression in these patients. Remarkably, the three CD28-deficient patients identified to date do not seem to have suffered from any other severe infections, mucocutaneous or otherwise, due to viruses or other pathogens. Intriguingly, all three CD28-deficient patients are seropositive for other cutaneous and mucosal HPVs, but no clinical manifestations of these infections were observed at the time of evaluation. The study of other patients with disseminated and/or persistent common warts, including patients with giant horns, may reveal inborn errors of adaptive immunity, possibly related to CD28 or T cells, or perhaps inborn errors of keratinocyte-intrinsic immunity to HPVs underlying common warts. These rare patients provide unique insight into the essential mechanisms by which humans control these ubiquitous HPVs. CD28-deficient steady-state and newly recruited T cells cannot correctly control the infection of skin keratinocytes with HPV2 and HPV4, resulting in recalcitrant common warts or even giant horns. CD28-independent inflammation is insufficient for the healing of giant horns and only sufficient, after substantial delay, for that of common warts.

Epidermodysplasia verruciformis.

Epidermodysplasia verruciformis (EV) is an autosomal recessive and specific predisposition to skin lesions caused by β-HPVs (22). These viruses are not very virulent and persist silently in the general population, without causing visible lesions. In patients with EV, β-HPVs cause flat warts and tinea versicolor-like lesions, which often progress to non-melanoma skin cancer when they occur in areas exposed to the sun. Some patients with ‘atypical EV’ suffer from other infections and their underlying deficiency is typically an acquired or inherited T-cell deficit, indicating that T cells are also required for protective immunity to β-HPVs (16). Patients with bona fide ‘typical EV’, are not vulnerable to other HPVs (apart from an α-HPV, or HPV-3), other viruses, or infectious agents. Most patients with classic EV carry biallelic loss-of-function mutations of transmembrane channel-like protein 6 (TMC6; also called EV protein 1, EVER1), TMC8 (also called EVER2), or calcium and integrin-binding protein 1 (CIB1) (23, 24) (Figure 1). EVER1, EVER2, and CIB1 form a multimeric protein complex, with the presence of each component required to ensure the stability of the whole complex. Within the skin, this complex is most abundant in keratinocytes.

This complex interacts with the E5 and E8 virulence factors of α- and γ-HPVs, respectively, but these virulence factors are absent from β-HPVs. This complex may, therefore, act as a restriction factor for these viruses (i.e. a human factor, the absence of which specifically releases the checks on the corresponding viral virulence factor). In the general population, E5⁺ (or E8⁺) HPVs can trigger warts in the presence of this complex, whereas E5⁻ and E8⁻ HPVs remain silent. In the absence of this complex, warts are triggered not only by E5⁺ (or E8⁺) HPVs, but also by E5⁻ and E8⁻ HPVs, because the viral deficiency of virulence factors E5 and E8 is compensated by lack of the human EVER-CIB1 restriction factor. This model has yet to be tested experimentally in vitro. The EVER-CIB1 complex is not induced by type I IFNs, consistent with the lack of β-HPV lesions in patients with inborn errors of type I IFN immunity. Human cell-intrinsic immunity to viruses can, therefore, operate via restriction factors that are not dependent on type I IFNs. The occurrence of EV lesions after successful hematopoietic stem cell transplantation and T-cell reconstitution in patients with severe combined immunodeficiency due to inherited Jun N-terminal kinase 3 (JAK3) or interleukin-2 receptor subunit-γ (IL2Rγ) deficiency suggests that an IL2Rγ- and JAK3-dependent cytokine signaling may control the expression or function of the EVER-CIB1 complex in keratinocytes (16). The silent persistence of defective β-HPVs in keratinocyte stem cells requires both cell-autonomous checking and steady-state T-cell patrols. If insufficient, inflammatory flat warts and even cancers can occur.

Herpes simplex virus encephalitis

Epidemiology.

Herpes simplex virus (HSV)-1 is a dsDNA virus that infects most individuals in a silent or benign manner, with occasional flare-ups of herpes labialis (cold sores) due to reactivation from a latent reservoir in the trigeminal ganglia. However, in rare cases, HSV-1 can also reach the forebrain, via the olfactory bulb, or the brainstem, via the trigeminal nerve, to cause devastating HSV-1 encephalitis (HSE), with massive inflammation that is a consequence of neuronal cell death due to viral replication and spread within the central nervous system (CNS). HSE cases are highest in early childhood, before the peak age for primary infection with HSV-1, implying that the risk of this inflammatory disease decreases with age (25). HSE is typically sporadic and is the most common sporadic viral encephalitis, at least in the West. HSE occurs in ~1/10,000 children, and in patients of diverse ancestries. The natural outcome of HSE is very poor, with ~80% mortality. Since the advent of acyclovir antiviral therapy, most patients have survived, albeit often with dreadful neurological sequelae.

Remarkably, HSE is typically isolated: there are no other HSV-1 lesions before, during, or after HSE, and there is no detectable viremia during disease. Moreover, these patients are generally not prone to other infectious diseases, viral or otherwise. Finally, none of the many inborn errors of leukocytes, including innate and/or adaptive subsets, and myeloid and/or lymphoid subsets, underlie HSE, whereas patients with disorders of T cells or myeloid APCs can suffer from disseminated HSV-1 disease that affects the skin, oral and esophageal mucosae, and liver, with viremia. Children on immunosuppressants or living with HIV are not prone to HSE either. Overall, HSE is a devastating idiopathic condition that neatly illustrates the ‘infection enigma’ (1): Why does HSV-1 attack the brain of some otherwise healthy children who are normally resistant to other infectious agents but also display resistance to HSV-1 in other tissues, when the same virus is harmless in most infected individuals?

Inborn errors of type I IFN immunity.

The causes of HSE were first elucidated in 2003 with the genetic study of a single patient (26). Paradoxically, this patient also had mycobacterial disease, which led to the discovery of an inborn error of type II IFN — autosomal recessive signal transducer and activator of transcription 1 (STAT1) deficiency — which abolishes cellular responses to type II IFN by preventing the formation of STAT1 homodimers (26). This finding serendipitously suggested that abolition of the transcriptional response to type I IFNs — the only IFNs known to activate STAT1-STAT2-IRF9 (interferon regulatory factor 9) heterotrimers (known as interferon stimulatory gene factor 3, ISGF-3) at the time, type III IFNs being characterized later — was causal for HSE. With hindsight, this interpretation was premature but correct, because another patient was reported in 2021 to have HSE caused by inherited IFN-α/β receptor 1 (IFNAR1) deficiency (27). A breakthrough in deciphering the pathogenesis of the typically isolated form of HSE was achieved by combining linkage and candidate gene approaches, which led to the identification of autosomal recessive UNC93B deficiency in two unrelated patients (28). This deficiency impairs cellular responses to endosomal, type I IFN-inducing Toll-like receptors (TLRs): TLR3, TLR7, TLR8, and TLR9 (29).

Cells from patients with autosomal recessive interleukin-1 receptor-associated kinase 4 (IRAK4) or myeloid differentiation primary response protein (MyD88) deficiency respond normally to the stimulation of TLR3, but not to the stimulation of TLR7–9, and these patients do not develop HSE upon HSV-1 infection. This suggested that TLR3 might be involved in protecting against HSE, and tests of this hypothesis rapidly led to the identification of autosomal dominant and recessive TLR3 deficiencies in children with HSE (30, 31). In turn, germline mutations of other genes controlling the TLR3 pathway were discovered: TLR adaptor molecule 1 (TICAM1), TANK binding kinase 1 (TBK1), TNF receptor associated factor 3 (TRAF3), and IRF3 (30, 32). TLR3 acts as an endosomal sensor of dsRNAs, which can be produced as intermediates or by-products in the course of viral infection; it induces type I IFNs. The penetrance of HSE is typically incomplete, more so for the dominant (TLR3, TICAM1, TBK1, TRAF3, and IRF3) than for the recessive (TLR3, TICAM1, and UNC93B) disorders (Figure 2). This accounts for the typically sporadic, as opposed to familial, nature of HSE. However, penetrance remains to be quantified precisely, and its determinants remain to be deciphered. These genetic studies thus suggest that HSE of childhood is due to impaired TLR3-dependent induction of type I IFNs, resulting in HSV-1 replication and subsequent inflammation in the brain.

Figure 2. Inborn errors of immunity of antiviral pathways underlying HSV-1 encephalitis.

Illustrations of the TLR3-type I IFN circuit and snoRNA31-dependent immunity in forebrain HSV-1 infection, and DBR1-mediated RNA lariat metabolism in brainstem viral infection. HSV-1 enters the forebrain and brainstem via the olfactory and trigeminal (TG) neurons, respectively. TLR3 controls basal levels of IFN-β-mediated anti-HSV-1 immunity in cortical neurons. TG neurons are susceptible to HSV-1 regardless of TLR3 genotype, whereas an exogenous TLR3 agonist or type I IFN renders healthy control TG neurons resistant to HSV-1 infection (35). DBR1 controls brainstem-specific immunity to viruses (HSV-1, influenza virus, norovirus), presumably in brainstem neurons. Genes for which mutations have been found to underlie forebrain HSV-1 encephalitis (red) or brainstem encephalitis (blue) are shown.

Inborn errors of cerebral neuron-intrinsic immunity.

Most of the leukocyte subsets of patients with TLR3-pathway deficiencies respond normally to dsRNA, probably through other dsRNA sensors, consistent with the lack of HSV-1 dissemination in HSE patients (33). By contrast, TLR3-deficient induced pluripotent stem cell (iPSC)-derived central nervous system (CNS)-resident cells, including cortical neurons and oligodendrocytes, are unresponsive to dsRNA and susceptible to HSV-1 infection (34). This phenotype was rescued by adding exogenous type I, but not type III IFN, further suggesting that insufficient amounts of type I IFN in the CNS are key to HSE pathogenesis. Moreover, iPSC-derived trigeminal neurons with and without TLR3 expression are equally susceptible to HSV-1 (35). These findings implicate CNS- and cell-intrinsic immunity to HSV-1 in the pathogenesis of HSE. Fibroblasts and iPSC-derived cortical neurons lacking TLR3 have low baseline concentrations of tonic type I IFNs (36), suggesting that their susceptibility to HSV-1 may be due to these concentrations being too low, rather than a failure to recognize virus-derived dsRNAs efficiently.

Other children with forebrain HSE carry heterozygous mutations of the small nucleolar RNA 31 (snoRNA31) gene (37). Deficiencies of snoRNA31 do not seem to interfere with the TLR3-dependent production of, or response to, type I IFN. However, iPSC-derived cortical neurons carrying snoRNA31 mutations are unable to control HSV-1 normally. This suggests that snoRNA31 is a type I IFN-independent restriction factor for HSV-1 in the forebrain. About 5 to 10% of children with forebrain HSE studied to date have been found to carry germline mutations impairing cell-intrinsic immunity in cortical neurons. HSV-1 can also cause brainstem infections, which were surprisingly found to be due to inherited debranching RNA lariats 1 (DBR1) deficiency in a multiplex kindred (38). DBR1 is the only known enzyme responsible for degradation of the RNA lariats formed during mRNA splicing, and its levels are highest in the brainstem (Figure 2). The defect in the patients is not complete, with 3 to 10% residual debranching activity. The mechanism by which lariat accumulation underlies HSE of the brainstem is unclear. However, this mechanism is not specific to HSV-1, because other DBR1-deficient patients have presented brainstem infections with influenza virus or norovirus (38). Moreover, inherited deficiency of TLR3 or melanoma differentiation-associated protein 5 (MDA5) was recently reported in children with enterovirus brainstem infection (39). Different molecular pathways govern cell-intrinsic immunity to HSV-1 in different territories of the brain, and their genetic disruption underlies viral growth, cell death, and secondary inflammation.

Severe viral pneumonia: influenza and COVID-19

Viral pneumonia.

An understanding of the basis of life-threatening influenza pneumonia has long remained enigmatic. Seasonal influenza viruses, which evolve by antigenic drift, with the gradual accumulation of mutations in viral glycoprotein genes, are common RNA viruses that infect the respiratory tract, causing benign illness in most infected individuals (40). Influenza pneumonia is rare, and may lead to inflammatory, acute respiratory distress syndrome (ARDS), which is even rarer. Pandemic influenza viruses, which emerge by antigenic shift due to drastic changes in viral glycoproteins, are more virulent (40). The 1918 influenza pandemic killed ~50 million people, but its victims constitute only a small proportion (~10%) of the total number of individuals infected, providing another striking example of the ‘infection enigma’. It is also intriguing that patients with the most severe forms of inherited leukocyte deficiency are not prone to severe influenza pneumonia. Even children lacking T cells and/or B cells are not prone to influenza-related respiratory failure. They respond poorly to influenza vaccination, but control influenza viruses normally. Patients with autosomal dominant GATA2 deficiency and severe influenza have been described (41). This condition may result from the lack of type I IFN-producing plasmacytoid dendritic cells (pDCs) in these patients (42).

The rarity of severe influenza in patients with most IEI strongly suggests that pulmonary defense against influenza virus in humans is dependent on other, unknown mechanisms. Not surprisingly, similar epidemiological findings were recently documented for COVID-19 pneumonia. Most individuals (90%) infected with SARS-CoV-2, another RNA virus that infects the respiratory system, remain asymptomatic or develop a benign, self-healing, ambulatory illness of the respiratory tract. However, a minority of infected individuals develop hypoxemic pneumonia (~10%), which can progress to ARDS (~2%), with exuberant pulmonary and systemic inflammation. Again, patients with inborn errors of leukocytes do not seem to be particularly prone to severe COVID-19 pneumonia. The parallel with severe influenza is striking. However, although seasonal influenza preferentially strikes pediatric and elderly populations, typical of most infectious diseases, the risk of life-threatening COVID-19 pneumonia increases with age from childhood (43).

Inborn errors of type I IFN immunity.

The elucidation of the pathogenesis of influenza pneumonia began in 2015, with the discovery of a single child with inherited IRF7 deficiency (44). Influenza virus replication rates were higher in her iPSC-derived pulmonary epithelial cells (PECs) than in control cells. Moreover, her pDCs did not produce type I and III IFN, even when stimulated with influenza virus, confirming that the high constitutive expression of IRF7 by this leukocyte subset underlies its production of most of the antiviral IFN in the blood (45). Surprisingly, this patient has not suffered from any other severe viral illness since her hospitalization in 2011 at the age of two years, with only annual vaccinations against influenza. Other children with severe influenza have been found to carry mutations of IRF9 (46), which encodes a component of the ISGF-3 transcription factor induced by type I and III IFNs, and mutations of TLR3 (47). Critical influenza and HSE are therefore allelic; they can share the same genetic etiology, possibly because TLR3 governs basal type I and III IFN production by both cortical neurons and PECs (36). In both conditions, insufficient type I IFN-dependent cell-intrinsic immunity results in viral spread, cell death, and secondary inflammation. By inference, it is probable that these two conditions are allelic at other, related loci, such as IRF7, IFNAR1 or IRF9. By contrast, deleterious mutations of ISGs may be more specific, as suggested by MX1 mutations underlying critical pneumonia caused by avian influenza viruses but not seasonal influenza viruses (48).

Two other IRF7-deficient patients recently suffered from critical COVID-19 pneumonia at the ages of 49 and 50 years of age (49). Another two patients with critical COVID-19 pneumonia at the ages of 26 and 38 years had inherited IFNAR1 deficiency, and other patients with critical COVID-19 pneumonia were found to be heterozygous for mutations of TLR3- and type I IFN-related genes (49). These patients had no prior episodes of severe viral disease, including influenza. Fibroblasts from patients with mutations of TLR3, IRF7 or IFNAR1, were found to support massive SARS-CoV-2 replication, and pDCs from an IRF7-deficient patient displayed no response to either influenza virus or SARS-CoV-2. It is unknown whether patients with inherited deficiencies of a cytosolic dsRNA sensor, MDA5, who are susceptible to rhinovirus (50, 51), are also susceptible to influenza virus or SARS-CoV-2. TLR3, IRF7, and type I IFNs may, therefore, be essential for the control of SARS-CoV-2 and influenza virus in the lungs, and HSV-1 in the brain in some patients, albeit with incomplete penetrance. By contrast, an unbiased analysis of patients with critical COVID-19 pneumonia found that ~1% of male patients under the age of 60 years had X-linked recessive TLR7 deficiency (52). TLR3 is expressed in PECs but not pDCs, whereas TLR7 is expressed by pDCs but not PECs. TLR7-deficient pDCs induce only small amounts of type I IFN in response to SARS-CoV-2 (52). Conversely, pDCs from UNC93B-deficient patients (unresponsive to TLR7 stimulation) responded normally to the seasonal influenza virus tested (44). Thus, hyperinflammatory ARDS caused by seasonal influenza virus or SARS-CoV-2 can be due to mutations that impair type I IFN cell-intrinsic immunity in PECs (Figure 3). Critical COVID-19 seems to be preferentially caused by the selective disruption of TLR7-dependent blood pDC type I IFN production, whereas severe avian influenza seems to be preferentially caused by the selective disruption of the ISG MX1 in PECs (48).

Figure 3. Type I IFN immunity in viral pneumonia.

Type I IFN responses in influenza A virus (IAV) and SARS-CoV-2 infections of the lung. Type I IFNs are produced by plasmacytoid dendritic cells (which do not express TLR3) via an IRF7-dependent pathway, and by respiratory epithelial cells (which express TLR3) via a TLR3- and IRF7-dependent pathway. Inborn errors of immunity affecting these pathways can cause life-threatening pneumonia, including influenza pneumonia (IRF9, red), COVID-19 pneumonia (UNC93B1, TRIF, TBK1, IRF3, IFNAR1, and IFNAR2, blue), or both (TLR3 and IRF7, red/blue), whereas autoantibodies against type I IFN constitute a phenocopy of inborn errors of immunity of type I IFN that can cause life-threatening COVID-19 pneumonia. ISGs: interferon-stimulated genes.

Autoimmunity to type I IFNs.

Critical COVID-19 pneumonia has turned out to be commonly caused by autoantibodies against type I IFNs. These antibodies were first described in the 1980s and were not widely thought to underlie viral diseases (53, 54). However, one patient with neutralizing autoantibodies against type I IFNs reported in 1984 had severe disease caused by varicella zoster virus (55). Further support for the pathogenicity of these antibodies is provided by the demonstration that autoantibodies against other cytokines underlie infectious phenocopies of the corresponding IEI (56). In this context, at least 10% and 15% of patients with COVID-19 pneumonia were found to have neutralizing autoantibodies against high (54) and lower, more physiological concentrations of type I IFNs (57), respectively. These autoantibodies are very rare in infected individuals without pneumonia. They are found in ~20% of cases of life-threatening COVID-19 pneumonia in patients over the age of 80 years and in ~20% of deaths across all ages. These autoantibodies recognize the 13 different IFN-α subtypes and/or IFN-ω, more rarely IFN-β, and even more rarely IFN-ε, and -κ.

Autoantibodies are present before SARS-CoV-2 infection and are causal for life-threatening pneumonia. They were found in ~0.5% of a sample of the general population aged 20 to 60 years, but their frequency reached 4% in people over the age of 70 years and 7% in those aged between 80 and 85 years (57). They were also shown to underlie severe or critical COVID-19 in patients with mutations of autoimmune regulator (AIRE) (58), all of whom harbor these autoantibodies because of defective central T-cell tolerance, and to underlie adverse reactions to the live attenuated yellow fever vaccine in other patients (59). Paradoxically, life-threatening COVID-19 pneumonia in many patients may, therefore, be considered an autoimmune condition, with adaptive B-cell immunity disrupting cell-intrinsic (in PECs) and innate (via pDCs) type I IFN immunity (Figure 3). Overall, inborn errors of and autoantibodies against type I IFN account for ~20% of cases of life-threatening COVID-19 pneumonia, inborn errors being more common in patients under the age of 60 years and autoantibodies being more common in patients over the age of 70 years. The autoantibodies were shown to decrease type I IFN activity in the nasal mucosae early in SARS-CoV-2 infection (60), as also expected for the corresponding inborn errors. These findings have suggested a unifying mechanism of life-threatening COVID-19, with the first stage characterized by insufficient type I IFN immunity in the respiratory tract, leading to viral dissemination, and an inflammatory second stage (61, 62). This suggests that the early administration of IFN-β in patients infected with SARS-CoV-2 may be of clinical benefit, whereas late administration has been shown to be ineffective (63).

Concluding remarks

The discovery of a causal germline genetic lesion points to a primary cause that precedes viral infection and responsibility for the various infectious and inflammatory phenotypes that together define an individual’s viral disease. The molecular, cellular, and immunological mechanisms of disease can be unraveled from this genetic starting point. Of course, it is unlikely that severe viral disease in every patient can be attributed to a monogenic lesion, even with incomplete penetrance. Nevertheless, IEI point to a physiological mechanism that can be disrupted by other causes, inherited or otherwise, and may, therefore, be of general importance. A single patient can be sufficient to decipher the underlying mechanism of inflammation and disease (64). Moreover, a rare inborn error can lead to the discovery of a common autoimmune phenocopy. Studies of a patient with IRF7 deficiency underlying severe influenza (44) led to the detection of inborn errors of and autoantibodies against type I IFN in ~20% of patients with severe COVID-19 (49, 52, 54, 57). Other viral infections may benefit from similar genetic approaches. Such studies provide new biological insights, moving the frontiers of immunity from the immune system to the whole organism (9, 65–67), while clarifying devastating medical problems.