OUP user menu

Susceptibility genes are enriched in those of the herpes simplex virus 1/host interactome in psychiatric and neurological disorders

Chris J. Carter
DOI: http://dx.doi.org/10.1111/2049-632X.12077 240-261 First published online: 1 December 2013


Herpes simplex virus 1 (HSV-1) can promote beta-amyloid deposition and tau phosphorylation, demyelination or cognitive deficits relevant to Alzheimer's disease or multiple sclerosis and to many neuropsychiatric disorders with which it has been implicated. A seroprevalence much higher than disease incidence has called into question any primary causal role. However, as also the case with risk-promoting polymorphisms (also present in control populations), any causal effects are likely to be conditional. During its life cycle, the virus binds to many proteins and modifies the expression of multiple genes creating a host/pathogen interactome involving 1347 host genes. This data set is heavily enriched in the susceptibility genes for multiple sclerosis ( P = 1.3E−99) > Alzheimer's disease > schizophrenia > Parkinsonism > depression > bipolar disorder > childhood obesity > chronic fatigue > autism > and anorexia ( P = 0.047) but not attention deficit hyperactivity disorder, a relationship maintained for genome-wide association study data sets in multiple sclerosis and Alzheimer's disease. Overlapping susceptibility gene/interactome data sets disrupt signalling networks relevant to each disease, suggesting that disease susceptibility genes may filter the attentions of the pathogen towards particular pathways and pathologies. In this way, the same pathogen could contribute to multiple diseases in a gene-dependent manner and condition the risk-promoting effects of the genes whose function it disrupts.

  • Alzheimer's disease
  • multiple sclerosis
  • schizophrenia
  • mood disorders


The herpes simplex virus 1 (HSV-1) has been implicated in Alzheimer's disease and is one of several pathogens (the human cytomegalovirus, Chlamydia pneumoniae, Borrelia burgdorferi and related spirochaetes and treponemes) able to promote beta-amyloid deposition and tau phosphorylation, its key pathological features (Little et al., 2004; Miklossy, 2011a; Wozniak et al., 2011; Alvarez et al., 2012; Lurain et al., 2013). It has also been implicated in multiple sclerosis relapse (Ferrante et al., 2000) and Parkinson's disease (Marttila et al., 1981) and with cognitive deficits in bipolar disorder and schizophrenia (Dickerson et al., 2004; Yolken et al., 2011). In multiple sclerosis, HSV-1 seropositivity has been associated with increased risk in those without the DRB1*15 allele or decreased risk in DRB1*15-positive patients, illustrating conditioning by gene/environment interactions (Waubant et al., 2011). In animal models, HSV-1 can promote demyelination (Townsend & Collins, 1986), and human oligodendrocyte cell lines are susceptible to HSV-1 infection (Bello-Morales et al., 2005). Such effects are relevant to multiple sclerosis and to the demyelination problems observed in many psychiatric disorders (Fatemi et al., 2009; Bartzokis, 2012). The virus can also induce rotational behaviour and modifications in cerebral dopamine function (Paivarinta et al., 1993). A viral protein (US3) also inhibits mitochondrial electron transport (Derakhshan et al., 2006), a property shared by the pesticides and toxins that selectively kill dopaminergic neurones (Duty & Jenner, 2011). Prenatal infection in animal models can induce neurodevelopmental changes relevant to autism-related disorders and to schizophrenia (Boksa, 2010). Such effects can also be induced by nonspecific viral mimics, cytokines or fever (Hinoue et al., 2001; Zuckerman & Weiner, 2005; Nawa & Yamada, 2012; Pacheco-Lopez et al., 2013). The viral mimic polyriboinosinic–polyribocytidylic acid also provokes beta-amyloid deposition and tau phosphorylation and memory loss in mice (Krstic et al., 2012), while gamma-interferon can induce nigrostriatal degeneration (Chakrabarty et al., 2011). Throughout life's course, many infections could potentially affect neurodevelopment or modify systems relevant to adult psychiatric and neurological disorders, either directly or via collateral immune and inflammatory effects.

Pathogens are implicated in many diseases, although their high seroprevalence relative to the incidence of disease has perhaps militated against a primary causal involvement. For example, HSV-1 has a seroprevalence of 68% in the American population aged more than 12 (Schillinger et al., 2004), far above the incidence of the diseases mentioned above. The problem of seroprevalence also applies to Helicobacter pylori, which indubitably causes stomach ulcers (Marshall et al., 1985) and likely gastric cancers (Rathbone & Rathbone, 2011), although not all of the many it infects [ c. 50% of the world population (Brown, 2000)], succumb to these conditions. Host and bacterial genetics, related to diverse pathogenic strains, play a key role in determining the outcome of infection (Israel et al., 2001; Snaith & El Omar, 2008).

Gene/environment interactions are also likely to influence the outcome of infection, and it has recently been noted, in a study of the Epstein–Barr virus and human papillomavirus, that the human targets of viral proteins reside in networks that overlap with those also occupied by susceptibility genes of a number of cancers (Gulbahce et al., 2012) where a role for these agents is suspected. A recent genome-wide association study (GWAS) has also highlighted an enrichment of the Epstein–Barr virus/host interactome in relation to multiple sclerosis susceptibility genes (Mechelli et al., 2013). Several susceptibility genes in Alzheimer's disease, bipolar disorder, schizophrenia and multiple sclerosis are also related to the life cycles of pathogens implicated in these conditions (Carter, 2007a, 2008, 2009, 2011a, 2012a). The toxoplasmosis parasite, Toxoplasma gondii, has been implicated in schizophrenia and related disorders (Torrey & Yolken, 2003; Henriquez et al., 2009), and a study of an extensive T. gondii host/pathogen interactome has shown that this data set is heavily enriched in susceptibility genes reported for schizophrenia and a number of related psychiatric conditions, but also in susceptibility genes related to multiple sclerosis, Alzheimer's and Parkinson's disease, where the pathogen's role is less clear (Carter, 2012b).

As shown below, the host genes implicated in the herpes simplex virus 1 (HSV-1) life cycle are also highly enriched in the products of susceptibility genes for the numerous conditions with which this pathogen has been associated. The pathways deranged by the virus are relevant to each condition, and subsets of the extensive HSV-1 host/pathogen interactome are specific for distinct diseases and to pathways relevant to their underlying pathologies. The close association between viral interactomes and disease susceptibility genes suggests multiple gene/environment interactions that might allow pathogens to cause certain aspects of disease in genetically susceptible individuals, where susceptibility genes might be able to orient the attentions of the pathogen towards particular pathways, pathologies and endophenotypes.


The genes implicated in Alzheimer's disease, attention deficit hyperactivity disorder (ADHD), autism, bipolar disorder, chronic fatigue syndrome, depression, schizophrenia, multiple sclerosis, Parkinson's disease, anorexia and childhood obesity have been culled from literature surveys and from data at the autism database at Mindspec [AutDB; Basu et al., 2009), the Bipolar database at the University of Chicago (Piletz et al., 2011), Alzgene, MSGene, PDgene and SZgene (Bertram et al., 2007; Allen et al., 2008; Lill et al., 2010, 2012)]. Genome-wide association data can be accessed at the National Human Genome Research Institute http://www.genome.gov/gwastudies/ (Hindorff et al., 2010) and from a noise-reduction genome-wide analysis of autism yielding a large number of genes involved in the outgrowth and guidance of axons and dendrites (Hussman et al., 2011). Kegg pathway analysis of these gene lists for each disease is available at PolygenicPathways (http://www.polygenicpathways.co.uk). The selection includes all genes for which at least one association study concluded that positive association was detected and is based on the interpretation of the original authors. All genes thus identified are hereinafter described as ‘susceptibility genes’, although the gradation of importance is appreciated. This include-all policy allows the presentation of a pool of genes whose implication may be supported/contested by other data. The reasons for this are related to an assumption of genetic heterogeneity in different populations and to possible gene/gene and gene/environment interactions, which may have influenced both negative and positive association studies. The collection assumes that both negative and positive association studies in different populations may be correct (assuming an adequate methodological statistical study) and that there may be an underlying reason for this heterogeneity. As shown by the example of the peppered moth, whose dark and light genes can each convey both risk and protection, in relation to its avian predators, depending upon the similar colours of the trees on which it alights (Kettlewell, 1955), environmental influences can radically modify genetic effects. In this context, herpes simplex infection is but one such variable (of many more) in relation to the diseases studied. In a postmortem study, the odds ratio for APOE4 was shown to be 1.67 in Alzheimer's disease patients without cerebral HSV-1 DNA and 16.8 in patients where viral DNA was detectable (Itzhaki et al., 1997), illustrating such influences in a disease context.

Thus, although replication problems exist for many of these genes, this may in part be due to the conditioning effects of gene/gene and gene/environment interactions [(Carter, 2007b, 2011d) as also suggested by the results of this survey]. The functional significance of many of these risk variants has also been questioned due to their location in noncoding regions, but as the ENCODE project has demonstrated, such genomic ‘dark matter’ is also endowed with functional properties capable of affecting the transcription and biology of relevant genes (Dunham et al., 2012).

Collection of the host/pathogen interactome

Host/pathogen interactions for HSV-1 were collected by literature survey, with no favouring of any particular viral strain. These included published protein/protein or yeast/two-hybrid interactions, host gene deletion experiments related to infectivity, viral replication or nocivity and the effects of infection on protein or mRNA expression in single or bulk microarray or proteomics experiments. The collection included data from experiments in laboratory animals and human cell lines. A similar strategy underpins the interactome data for the HIV-1/host interactome collated by NCBI, which contains more than 2500 interactions (Ptak et al., 2008). The interactome, with links to the original references, is posted at http://www.polygenicpathways.co.uk/HERPEshost.html. Other relevant web pages are listed at http://www.polygenicpathways.co.uk/hsv1.htm. Pathway analysis of this interactome was performed using KEGG mapper (Goto et al., 1997) http://www.genome.jp/kegg/tool/map_pathway2.html and the resultant pathways posted at http://www.polygenicpathways.co.uk/herpeSKEGG.htm.

Enrichment analysis

Given the number of genes influenced by HSV-1, or defined as susceptibility genes, one would expect a certain degree of overlap between the host/pathogen interactome and the susceptibility gene lists. Any significant over-representation of such overlap was estimated as follows. At the time of writing, the human genome contains 26 846 genes and the HSV-1/host interactome 1347. Other data sets would be expected to contain 1347/26 846 interactome-related genes (5.01%). Similarly, for N disease susceptibility genes, N/26 846 should appear in the host/pathogen interactome, providing the expected numbers in each data set (interactome in susceptibility genes and susceptibility genes in interactome). The P-values stem from the chi-squared test, derived from the observed/expected ratios.

Pathway enrichment analysis

Statistical analysis for KEGG pathway enrichment (over-representation analysis) was performed using the Consensus Path database (CPDB) http://cpdb.molgen.mpg.de/CPDB where the methodology is described. A P-value is calculated according to the hypergeometric test based on the number of physical entities present in both the predefined set (total number of genes in each pathway) and the list of interactome genes (Kamburov et al., 2011). Overlapping data sets were identified using Venn diagrams from the online Venny tool http://bioinfogp.cnb.csic.es/tools/venny/index (Oliveros, 2007). Throughout the text, HUGO gene nomenclature committee gene symbols are used (White et al., 1997).


Enrichment of interactome genes within susceptibility gene data sets

Interactome genes were significantly enriched in the susceptibly gene data sets for all diseases except ADHD representing from c. 9.5% (anorexia) to 23% (multiple sclerosis) of the genes analysed, with enrichment values from 1.19-fold (autism) to 4.12-fold (multiple sclerosis), in order of overlap were multiple sclerosis, Alzheimer's disease, schizophrenia, Parkinsonism, depression, bipolar disorder, childhood obesity, chronic fatigue, autism and anorexia (Table 1).

View this table:
Table 1

A statistical analysis of the overlap between human genes in the HSV-1 interactome and the susceptibility genes in various diseases

HSV-1 interactome in diseaseDisease genes in interactome
ConditionN disease genes% in interactome data setObserved gene/interactome overlapExpected NFold enrichment (A)Expected NFold enrichment (B)Mean enrichment (A + B)/2P-value
Multiple sclerosis40823.39526.43.6020.54.634.121.3E−99
Childhood obesity7316.4124.72.543.73.32.914.07E−08
Chronic fatigue9511.6116.11.794.
  • ADHD, attention deficit hyperactivity disorder.

  • The number of susceptibility genes analysed ( N genes) is shown for each disease, together with the observed and expected values for each condition, the fold and mean enrichments (observed/expected), and the P-value derived from the chi-squared test. The HSV-1 interactome used contains 1347 host genes.

KEGG pathway analysis of the HSV-1/host interactome

Infection pathways including herpes simplex infection ( P = 2.2E−72) but also other pathogen life cycle pathways [e.g. HTLV-I infection P = 9.3E−28, influenza A ( P = 11.8E−42), measles ( P = 2.5E−34), Epstein–Barr virus infection ( P = 4.4E−32), hepatitis C ( P = 1.3E−27), toxoplasmosis P = 2.2E−22)] were defined by the interactome, as well as many immune-related and pathogen defence pathways, stemming perhaps from the common defence mechanisms related to infection (immune activation, inflammation/oxidative stress etc.) and to similar access routes (Supporting Information, Tables S1 and S2; see http://www.polygenicpathways.co.uk/herpeskegg.htm). These included, for example, cytokine and chemokine signalling pathways ( P = 3E−22) and T- and B-cell receptor signalling ( P = 2.5E−21 and 2.3E−13, respectively) and natural killer cell–mediated cytotoxicity ( P = 2.6E−15).

The involvement of microbial defence pathways in the interactome [Toll receptor ( P = 2.54E−21), NOD ( P = 3.13E−16), RIG1 ( P = 9.07E−16) and cytosolic DNA sensing ( P = 3.68E−15)] is likely to generally impact upon pathogen defence. Immune activation by one pathogen might well protect against other invaders, but other interactions also exist. For example, herpes simplex infection activates Epstein–Barr viral replication (Wu et al., 2012), while a viral protein (US11) activates retroviral glycoprotein expression (HTLV-1 and HIV-1; Diaz et al., 1996). Such effects and the shared pathways highlight the potential cross-talk between elements of the microbiome.

Cancer-related pathways are highly represented (pathways in cancer P = 2.9E−27; Table S1), as are growth factor, cell cycle, p53 and other relevant networks (Table S2): HSV-1 and other viruses have been implicated in cancers (Saddawi-Konefka & Crawford, 2010), but certain HSV-1 strains also possess oncolytic qualities and are being engineered for this purpose (Nawa et al., 2008; Todo, 2008).

Several autoimmune disease pathways are affected by this interactome [notably rheumatoid arthritis ( P = 1.7E−10), autoimmune thyroid disease ( P = 4.4E−07) and type I diabetes mellitus ( P = 1.2E−08)]. Other pathways involved within the interactome included Alzheimer's disease ( P = 1.3E−11), amyotrophic lateral sclerosis ( P = 3.1E−16), Parkinson's disease ( P = 1.18E−27) and cardiac myopathy pathways [Dilated cardiomyopathy ( P = 2.6E−34)].

The interactome implicates a large number of signalling networks and processes (Table S2), many of which are relevant to degenerative processes (e.g. apoptosis P = 8.1E−25) or to neuropsychiatric disorders [e.g. dopaminergic synapse ( P = 2.0E−08), long-term potentiation ( P = 6.8E−13), long-term depression ( P = 0.0002), cholinergic synapse ( P = 6.75E−06)] or to neurodevelopment [e.g. neurotrophin signalling ( P = 4.77E−18), Wnt signalling ( P = 2.88E−08), ErbB signalling ( P = 3.79E−14), axon guidance ( P = 0.001) and cell adhesion molecules ( P = 3.14E−09)]. These are further discussed below.

Disease selectivity of interactome compartments

Within and across diseases, the types of susceptibility genes influenced were either common to several diseases or relatively specific for a particular disease. This was assessed by comparing shared and specific overlapping interactome/disease genes across all diseases and by statistical analysis of the enrichment of KEGG pathways in the disease-specific arms of the interactome (Tables 24). The genes within the KEGG pathways for each of these disease/interactome overlaps are posted at http://www.polygenicpathways.co.uk/hsv1.htm.

View this table:
Table 2

Neurological diseases: HSV-1 interactome genes specifically overlapping with individual neurological disease gene data sets

DiseaseGenes and classification
Multiple sclerosisApoptosis: CASP8 CASP9 FASLG
Growth: RPS6KB1
Myelin: OLIG3
Adhesion: SELPLG
Proteasome: PSMB8 PSMB9
Inflammation/ox stress: NFKBIA PTGER4
CPDB physiology analysis (specific to MS):
Apoptosis P = 5.56E−10, Osteoclast differentiation P = 3.52E−07, Neurotrophin signalling P = 5.68E−05, Prion diseases P = 0.00085, MAPK signalling P = 0.00284, Cell adhesion molecules P = 0.00556
APP processing and signalling APP: BACE1 MAPK8IP1 NCSTN PSEN1 PSENEN
Lipoprotein: APOA1 LPA OLR1
Apoptosis: CST3 TP73
Growth: VEGFA
Proteoglycan: ACAN HSPG2
Retinol: RXRA TTR
Immune: C11orf30 CD33 CD36 CFH LCK MAL2 MICA TLR2 ZBP1
Cell cycle: CCNT1 CDK5
CPDB analysis (specific to ALZ): Alzheimer's disease P = 2.53E−07: PPAR signalling pathway P = 0.000186, Notch signalling P = 0.0008: Hematopoietic cell lineage P = 0.0133
Parkinson's diseaseChaperone: HSPA1A TOR1A
Growth and development: BMP4 CDKN2C NR4A2(Nurr1) PCGF3
Vitamin B: PDXK
Pathogen defence: NOD2
Diverse: EIF4G1 SSBP3
CPDB analysis (specific to PD): no significant enrichment
  • APP, amyloid precursor protein; PPAR, peroxisome proliferator–activated receptor.

  • Genes are broadly classified by function. Grey shaded genes are primary receptors [or putative receptors (proteoglycans, sialic acid-binding lectins, etc., or genes involved in heparan sulphate metabolism)] for the virus. Beneath each primary disease set is the CPDB enrichment analysis restricted to physiological KEGG pathways etched out by the interactome genes that are specific for each disease (pathway name and P-value). Other pathways (infection, immune and other diseases) can be found in Tables S1–S6.

View this table:
Table 3

Psychiatric diseases: HSV-1 interactome genes specifically overlapping with individual psychiatric disease gene data sets

DiseaseGenes and classification
Autism Dopamine: PITX1
Heparan sulphate: HS3ST3A1 HS3ST5
Krebs cycle/propanoate: SUCLA2 SUCLG2
Circadian: PER1
Growth/development: EYA2 RCOR1 RPS6KA2
Apoptosis: DFFB PDCD6 RIPK2
Inflammation/ox stress: NOS2A PTGER2
Immune: CCL20 CCL25 CCL26 CD38 CXCR3 IL17RA
Specific to autism: CPDB enrichment:
Processes: Focal adhesion P = 0.000563, Regulation of actin cytoskeleton P = 0.000734, ECM–receptor interaction P = 0.00305, Osteoclast differentiation P = 0.00954, Cell adhesion molecules P = 0.0106, Adherens junction P = 0.026, Apoptosis P = 0.0345, Progesterone-mediated oocyte maturation P = 0.0352
Physiology: mTOR signalling P = 2.84E−05, Neurotrophin signalling P = 0.000903, Insulin signalling P = 0.00123, Long-term potentiation P = 0.0228, VEGF signalling P = 0.0267, ErbB signalling P = 0.036, Glycosaminoglycan biosynthesis – heparan sulphate P = 0.00356
Metabolism: Citrate cycle (TCA cycle) P = 0.00473, Propanoate metabolism P = 0.00537
Bipolar disorderCircadian: CSNK1E
Inflammation/ox stress: GPX3
Immune: SIGLEC6
No KEGG CPDB enrichment for Bipolar disorder
Adhesion: VCAN
Circadian: CRY1
Apoptosis: APAF1
Immune: IDO1
Lysosome: M6PR
Diverse: DPP4 PRKCB
No CPDB enrichment for depression
Schizophrenia Ubiquitin: SKP2 UFD1L
DNA related Chromatin/repair: SMARCA2 SMARCC1 XRCC1 ZNF184
Neural: CHGB SLC1A2
Inflammation/ox stress: GSS
Specific to schizophrenia:
Processes: Adherens junction P = 5.38E−06, Axon guidance P = 0.000126, Osteoclast differentiation P = 0.000132, Apoptosis P = 0.000172, Focal adhesion P = 0.00022, Protein processing in endoplasmic reticulum P = 0.00345, Circadian rhythm P = 0.00803, Endocytosis P = 0.00807, Regulation of actin cytoskeleton P = 0.00983
Signalling: MAPK signalling P = 5.92E−09, VEGF signalling P = 5.83E−06, GnRH signalling P = 3.50E−05, Neurotrophin signalling P = 0.000126, ErbB signalling P = 0.00201, Calcium signalling P = 0.00444
Neural: Dopaminergic synapse 1.21E−06, Cocaine addiction P = 7.75E−06, Amphetamine addiction P = 5.93E−05, Glutamatergic synapse P = 0.000111, Long-term potentiation P = 0.00849, Long-term depression P = 0.00919
  • Genes are broadly classified by function. Grey shaded genes are primary receptors [or putative receptors (proteoglycans, sialic acid-binding lectins, etc.)] for the virus, or genes involved in heparan sulphate metabolism. Beneath each primary disease set is the CPDB enrichment analysis restricted to physiological KEGG pathways etched out by the interactome genes that are specific for each disease (pathway name and P-value). Other pathways (infection, immune and other diseases) can be found in Tables S1–S6.

View this table:
Table 4

HSV-1 interactome genes specifically overlapping with ADHD, childhood obesity, anorexia and chronic fatigue disease gene data sets

DiseaseGenes and classification
ADHD (no enrichment in relation to the HSV-1 interactome) Neural: ARRB2 CAMK2A
Myelin: MOBP
Pancreas: PDX1
Childhood obesity
No CPDB KEGG enrichment
Calcium: ITPR3 TRPV1
Growth: IGFBP3
Inflammation/ox stress: PTER
Immune: IL15
Chronic fatigue
No CPDB KEGG enrichment
Neural: SCN9A
Immune: IL17F
Other: ARHGAP20 MSH3
  • ADHD, attention deficit hyperactivity disorder.

  • Genes are broadly classified by function. Grey shaded genes are primary receptors [or putative receptors (proteoglycans, sialic acid-binding lectins, etc.)] for the virus.

Shared genes and processes in the interactome/susceptibility gene data sets

Many diseases share significant aspects of pathology and endophenotypes. For example, apoptosis or growth factor malfunction is typical feature of neurodegenerative disorders (Connor & Dragunow, 1998; Fuchs & Steller, 2011). Demyelination is a feature of multiple sclerosis, but also of psychiatric and neurodevelopmental disorders (Bartzokis, 2012). Neurotransmitter imbalance dictates behavioural symptomatology in psychiatric conditions where circadian disruption is also relevant (Lamont et al., 2010). Immune activation and inflammation (Muller et al., 2009; Pizza et al., 2011), and autoimmunity and oxidative stress (Jou et al., 2009; Melo et al., 2011) are also common to these disorders.

Degenerative diseases: Alzheimer's disease, Parkinson's disease and multiple sclerosis

For these diseases (Table S3), few of the shared genes were outside a key principal group relating to apoptosis or growth, inflammation/oxidative stress and immunity, supporting the notion that neurodegeneration might be associated with the collateral damage inflicted by the activation of general defence mechanisms common to pathogen invasion (Deleidi & Isacson, 2012). Certain exceptions included groups of genes related to metalloproteases (multiple sclerosis, Alzheimer's), adhesion (multiple sclerosis/Alzheimer's) and lipoprotein-related genes and GSK3B (all degenerative disorders).

Psychiatric diseases: ADHD, autism, depression, bipolar disorder and schizophrenia

In the psychiatric group, inflammation/oxidative stress, immune- and growth factor-related, but not apoptosis, genes were the principal groups of shared genes (Table S4). Neural [glutamate (NMDA), GABA, dopamine, acetylcholine and synaptic] and circadian groups were also commonly represented across most diseases.

Common interactome genes in the childhood obesity/anorexia/chronic fatigue group only concerned APOE and TNF (all) and the oestrogen receptor, ESR1 (obesity/anorexia).

Certain genes were common to many diseases (all subsets), including APOE (all except autism), GSK3B (all except autism, ADHD and the obesity/anorexia/fatigue subset) and the oestrogen receptor ESR1 (all except bipolar disorder and Parkinsonism). APOE is concerned with the transport of cholesterol, a major myelin constituent, also indispensable for cerebral function and plasticity (Bjorkhem & Meaney, 2004; Holtzman et al., 2012). APOE also transports fats and the vitamin A precursor retinyl palmitate (Weintraub et al., 1987).Vitamin A derivatives also play an important role in myelination, plasticity and neurite outgrowth (Huang et al., 2011; Puttagunta et al., 2011, 2011). Oestrogens also modulate neuroplasticity and promote axon and myelin survival (McEwen, 2002; Spence & Voskuhl, 2012). These genes affect several aspects of cerebral function that are pertinent to most psychiatric and neurological disorders, perhaps explaining this ubiquity.

HSV-1 interactome overlaps specific to certain diseases

Several HSV-1 interactome genes were limited to a particular disease (see Tables S5 and S6 for full data sets). The pathways and processes within these specific data sets are relevant to each disease, as illustrated by the pathway enrichment analysis of the genes within these compartments (Tables 24, and Tables S5 and S6) and by the Venn diagrams relating to the pathway enrichment analysis across groups of four diseases (Fig. 1). The genes within these disease-specific data sets are distinct, but the pathways they occupy define both specific and shared processes and pathologies across diseases.

Figure 1

Venn diagrams showing the common and specific KEGG pathways across permutations of four diseases etched out by the genes of the HSV-1 interactome that overlapped with each disease (same pathways, different genes: from Tables S5 and S6). The pathway names within each are illustrated below the figures. Aut, autism; SZ, schizophrenia; DEP, depression; MS, multiple sclerosis; PD, Parkinson's disease. Only significantly enriched pathways (from the CPDB analysis) are shown in this analysis. There is overlap in other pathways, for example in relation to infections, but not all are significantly enriched in relation to individual interactome/susceptibility gene overlaps. Psychiatric and Multiple Sclerosis ( left): Autism alone: mTOR signalling, insulin signalling, Fc epsilon RI signalling, progesterone-mediated oocyte maturation, arrhythmogenic right ventricular cardiomyopathy, hypertrophic cardiomyopathy, dilated cardiomyopathy, ECM–receptor interaction, hematopoietic cell lineage, glycosaminoglycan biosynthesis – heparan sulphate, TCA cycle, propanoate metabolism, renal cell carcinoma. Schizophrenia alone HTLV-I infection, amoebiasis, Salmonella infection, dopaminergic synapse, glutamatergic synapse, long-term depression, calcium signalling, cocaine/amphetamine addiction, axon guidance, GnRH signalling, protein processing in endoplasmic reticulum, circadian rhythm, endocytosis Bladder cancer. Depression alone African trypanosomiasis. Multiple sclerosis alone Jak-STAT signalling pathway, adipocytokine signalling, natural killer cell–mediated cytotoxicity, Staphylococcus aureus infection, viral myocarditis, cytosolic DNA sensing, primary immunodeficiency, colorectal cancer prion diseases, allograft rejection. Autism, schizophrenia and MS Herpes simplex infection, Leishmaniasis, chemokine signalling, NOD-like receptor signalling, B-cell receptor signalling, T-cell receptor signalling pathway, Toll-like receptor signalling, neurotrophin signalling, osteoclast differentiation, endometrial cancer, acute myeloid leukaemia, apoptosis, prostate cancer. Schizophrenia and MS Chagas disease (American trypanosomiasis), tuberculosis, toxoplasmosis, hepatitis C, epithelial cell signalling in Helicobacter pylori infection, Influenza A MAPK signalling, pathways in cancer, pancreatic cancer. Autism and schizophrenia Adherens junction Focal adhesion Regulation of actin cytoskeleton, Long-term potentiation, ErbB signalling, VEGF signalling, nonsmall cell lung cancer, glioma, melanoma. Autism and MS Intestinal immune network for IgA production, cytokine–cytokine receptor interaction, cell adhesion molecules, chronic myeloid leukaemia. Neurology and schizophrenia ( Right) Alzheimer's disease alone Alzheimer's disease pathway, PPAR signalling, Notch signalling, hematopoietic cell lineage, malaria ALZ and MS Prion diseases. Multiple sclerosis alone Cytokine–cytokine receptor interaction, Jak-STAT signalling, natural killer cell–mediated cytotoxicity, cytosolic DNA sensing, colorectal cancer, chronic myeloid leukaemia, primary immunodeficiency, allograft rejection, intestinal immune network for IgA production, S. aureus infection, cell adhesion molecules, adipocytokine pathway, viral myocarditis. No specific Parkinson's pathways: Schizophrenia alone HTLV-I infection, amoebiasis, Salmonella infection, adherens junction, ErbB signalling pathway, VEGF signalling, calcium signalling, GnRH signalling, circadian rhythm, dopaminergic synapse, amphetamine addiction, cocaine addiction, glutamatergic synapse, long-term potentiation, long-term depression, axon guidance, focal adhesion, regulation of actin cytoskeleton, protein processing in endoplasmic reticulum, endocytosis on Bladder cancer, nonsmall cell lung cancer, glioma. Schizophrenia and MS Influenza A, herpes simplex infection, hepatitis C, Chagas disease (American trypanosomiasis), Shigellosis, Tuberculosis, Legionellosis, Leishmaniasis, Pertussis, epithelial cell signalling in H. pylori infection, NOD-like receptor signalling, RIG-I-like receptor signalling, Toll-like receptor signalling, T-cell receptor signalling, B-cell receptor signalling, chemokine signalling, neurotrophin signalling, apoptosis, osteoclast differentiation, small cell lung cancer, prostate cancer, acute myeloid leukaemia, pancreatic cancer, endometrial cancer, amyotrophic lateral sclerosis. MS, PD and SZ: Toxoplasmosis, Measles, Pathways in cancer, MAPK signalling. PD and SZ Melanoma.

Neurodegenerative diseases

In Alzheimer's disease, the Alzheimer's disease pathway itself, peroxisome proliferator–activated receptor (PPAR) and Notch signalling and hematopoietic cell lineage pathways [regulated by Notch and PPAR's (Greene et al., 2000; Caolo et al., 2012)] and malaria pathways are specifically affected by the HSV-1 interactome (Table 2 and Table S5). Only one overlapping pathway was identified (the prion pathway shared with multiple sclerosis). PPARs regulate many processes including glucose/insulin homeostasis, fatty acid oxidation, cholesterol metabolism, immune responses, oxidative stress and amyloid precursor protein (APP) processing that are all relevant to Alzheimer's disease (Chen et al., 2012). Notch similarly plays an important role in multiple developmental pathways and in plasticity, survival and neurogenesis in the adult brain (Ables et al., 2011). Notch signalling also regulates T-cell development and immune function and also the regulatory T cells (Treg's) that control autoimmunity (Haque et al., 2012). The APP processing network also plays a key role in immune defence, as beta-amyloid is an antimicrobial and antiviral peptide (Lukiw et al., 2010; Soscia et al., 2010). The gamma-secretase components that generate beta-amyloid are localised in dendritic cells of the immune network, which scout for invading pathogens and present antigens to immunocompetent cells. A number of HSV-1 and other pathogen receptors are substrates for this complex (Carter, 2011a). The normal form of the prion protein is also expressed in immunocompetent cells, suggesting a role in immune function (Nitta et al., 2009). This immune-related spectrum appears to be a common theme of this group.

In multiple sclerosis, a large number viral, bacterial and parasitic pathways were related to the specific overlap, as were dedicated (Toll, RIG-1, NOD and viral DNA sensing) and more general immune defence pathways. With respect to bacterial infections, one of the HSV-1 viral receptors, the herpes viral entry mediator (TNFRSF14) is expressed on monocytes and neutrophils, and its ligand, LIGHT (TNFSF14), a gene within the multiple sclerosis specific compartment of the interactome, promotes bactericidal activity against Listeria monocytogenes and Staphylococcus aureus, an effect blocked by HSV-1 glycoprotein D (Heo et al., 2006). Many viruses or bacteria have been linked to multiple sclerosis or demyelinating diseases and are capable of inducing demyelination in animal models, effects that may in part relate to the autoimmune problems related to pathogen/human protein mimicry (Westall, 2006; Harkiolaki et al., 2009; Ochoa-Reparaz et al., 2011; Owens et al., 2011; Carter, 2012a).

Numerous cancer pathways were represented in the viral interactome compartment specific to multiple sclerosis, although the role of the herpes virus in controlling such associations is unclear. Adhesion molecules, which control axonal/glial relations and survival (Coleman, 2011), and neurotrophin pathways within this pathway set and are also of relevance.

Interactome genes specific to Parkinsonism included NR4A2 (Nurr1) and BMP4, which both regulate dopaminergic phenotypes (Zeng et al., 2004). PDXK is involved in the synthesis of vitamin B6, a cofactor in the enzymes of dopamine synthesis (de Lau et al., 2006). Chaperones (HSPA1A and torsin TOR1A) play an important role in the protein-folding problems in Parkinson's disease (Ebrahimi-Fakhari et al., 2012). The NOD2 gene again suggests an influence of pathogens although in this relatively small gene set, there was no significant pathway enrichment.

Psychiatric diseases

In Autism, as with degenerative diseases, a number of pathogen- and immune-related pathways were incorporated within the disease-specific arm of the viral interactome, as were a number of cancer-related pathways (Table 3 and Table S6 for full data set). Adhesion, actin and extracellular matrix pathways were also implicated. These play a key role in synaptic growth and organisation during neurodevelopment and in autism (Melom & Littleton, 2011). Apoptosis also figured in this group and is used, in a beneficial rather than a degenerative way, to sculpt pathways during neurodevelopment (De Zio et al., 2005). Growth factor and long-term potentiation pathways in this group can all be related to synaptic plasticity and/or myelination (Lim et al., 2003; Talmage, 2008; Gipson & Johnston, 2012). Cardiomyopathy pathways in this data set are relevant to links between autism and cardiac myopathy disorders (Burusnukul et al., 2008; Connolly et al., 2010). The transcription factor PITX1 regulates dopaminergic development (Freed et al., 2008), and a number of specific genes are involved in methylation, an epigenetic process relevant to autism and related disorders (Houston et al., 2012). Viruses and bacteria can provoke host gene methylation and infection may be a driving force in this process (Paschos & Allday, 2010).

Schizophrenia-specific interactome genes are implicated in a number of infection, defence and immune pathways. As mentioned above, cancer pathways were well represented as were apoptosis, axon guidance, adhesion and actin pathways, and growth factor and LTP pathways, highlighting similarities between autism and schizophrenia. In addition, more specific to schizophrenia were calcium signalling, dopamine and addiction pathways, glutamate and long-term depression pathways, and circadian rhythm in line with the key role of these processes in the disorder (Seeman, 2009; Lamont et al., 2010).

In depression, adhesion, apoptosis, immune-related and circadian genes were implicated in the disease-specific data set, and circadian and immune genes in bipolar disorder, although there was no KEGG pathway enrichment in either data set.

Fewer genes and not specifically enriched KEGG pathways were specific to ADHD, chronic fatigue, anorexia or childhood obesity, and these are not further discussed (Table 4).

Although the genes within the disease-specific interactome data sets are distinct, the pathways they occupy are specific or common to various diseases as illustrated by the VENN diagrams in Fig. 1. These are discussed above, and the diagrams illustrate this in graphic form.

Herpes simplex receptors specific to disease data sets

It is noteworthy that several primary and potential HSV-1 receptors appear in the disease-specific interactome lists (Tables 24; see http://www.polygenicpathways.co.uk/herpeshost.html for references). Experimentally validated HSV-1 entry receptors include the herpes virus entry mediator (TNFRSF14), poliovirus relayed receptors (PVRL1 and PVRL2), growth factor receptors (FGFR1 and IGF2R), a related mannose-6-phosphate receptor (M6PR), several integrins (ITGA5, ITGB3) and the mac-1 dimer (ITGB2/ITGAM), myelin-associated glycoprotein (MAG), myosin MYH9, the syndecans (SDC1 and SDC2) and the paired immunoglobulin-like receptor, PILRA , inter alia (see database for references). HSV-1 also binds to heparan and chondroitin sulphates, which are themselves tied to several heparan or chondroitin sulphate proteoglycans (HSPGs and CSPGs), which might also be considered as potential receptors. HSPGs include agrin (AGRN) biglycan (BGN), glypicans (GPC1-5) neuropilin (NRP1) and perlecan (HSPG2). The CSPGs include aggrecan (ACAN), appican, a glial form of APP, brevican (BCAN), CSPG3–5, neurocan (NCAN), neuropilin (NRP1), syndecans (SDC1-5) and versican (VCAN): the HSV-1 virion also contains alpha 2,3 and alpha 2,6-linked sialic acids, which play a role in viral entry (Teuton & Brandt, 2007). SIGLEC1, SIGLEC6, CD22, CD33 and MAG bind to alpha 2,3- and/or alpha 2,6-linked sialic acids (Brinkman-Van der Linden & Varki, 2000). The disease-specific entities are PVRL1 and the integrin ITGB3 in autism (HS3ST3A1 and HS3ST5 involved in heparan sulphate synthesis), CSPG5, FGFR1, GPC1, MAG and NRP1 (schizophrenia), nucleolin and SIGLEC6 (bipolar disorder), M6PR and versican (depression), ITGAM and a heparan sulphate enzyme (HS3ST3B1; multiple sclerosis); APP, aggrecan (ACAN), CD33, HSPG2 and PVRL2 (Alzheimer's disease), and ITPR3 (childhood obesity). Other receptors shared across diseases (not shown) include MYH9 (multiple sclerosis, Alzheimer's and schizophrenia), neurocan (bipolar disorder and schizophrenia), GPC5 (autism, schizophrenia, bipolar disorder, multiple sclerosis) and insulysin (IDE; Alzheimer's, multiple sclerosis and Parkinson's disease). Complement receptor CR1, a receptor for many pathogens (Powers et al., 1995), appears in Alzheimer's, multiple sclerosis, Parkinson's, schizophrenia and depression data sets.

Pathogen receptors are the primary points of entry and govern all downstream events. They are expressed at specific times and locations during foetal and childhood development and adulthood. This receptor/disease specificity suggests a means by which the same pathogen could differentially affect common pathways, but in specific tissues at particular times, in a manner that would depend upon host genetics and development, and the route, site and timing of infection. Viral strain and route of infection are clearly important. For example, intranasally administered HSV-1 (strain 17 syn+) in BALB/c mice provokes entorhinal cortex and hippocampal neuronal loss, decreased brain volume, cortical atrophy, astrogliosis and memory deficits (Armien et al., 2010), while the intraperitoneal administration of the viral KOS strain in C57BL/6 mice produces fewer neurotoxic effects and slight memory deficit (Guzman-Sanchez et al., 2012).

Analysis of genome-wide association studies

GWAS data (reported genes column) were downloaded from the Genome.gov Website (April, 2013), and the genes overlapping with those of the HSV-1 interactome are shown in Table 5 (no data were available for anorexia, childhood obesity or chronic fatigue). The HSV-1 interactome was also significantly enriched in the multiple sclerosis and Alzheimer's disease data sets, but not in others. Interestingly, as with the larger gene sets, different known (PVRL2/Alzheimer's) or putative HSV-1 entry receptors (heparan or chondroitin sulphate proteoglycan laced proteins (GPC5/multiple sclerosis, NCAN/bipolar disorder; NRP1/schizophrenia; VCAN/depression) or sialic acid-binding receptors (CD33/Alzheimer's) appeared as specific entities in certain data sets.

View this table:
Table 5

Genome-wide association study genes related to the HSV-1/host interactome: as with the larger data set, a figure of 5.01% (interactome genes/human genome genes) might be expected within these data sets

Interactome genesFold enrichmentP-value
Multiple sclerosis 20/160 (12.5%)CD40 CXCR4 EN1 GPC5 HLA-B HLA-DRB1 IL12B IL7 IL7R KPNB1 MAPK1 MYC NFKB1 OLIG3 PTGER4 RPS6KB1 TNFRSF1A TNFSF14 YWHAG ZIC12.49< 0.0001
Alzheimer's disease (6/60) (10%)APOE CD33 CR1 LUZP2 MMP3 PVRL21.990.02
Schizophrenia 7/80 (8.75%)CXCL12 HLA-DRB1 IL3RA NFKB1 NRP1 PARD3 ZNF1841.750.066
Bipolar disorder 12/156 (7.6%)CTNNA2 CTSH HNRNPC KIF5B NCAN PC RAB1B RBM14 SF3B2 SFMBT1 SNAP91 TLR91.52NS
Depression (5/68) 7.35%ABCF3 EIF3F ITGB1 PARD3 VCAN1.47NS
Parkinson's (4/64) 6.25%ATF6 BMP4 MAPT SNCA1.25NS
ADHD 3/90 (3.3%)MOBP PDX1 YWHAZ0.66NS
Autism (0/8)NS
  • ADHD, attention deficit hyperactivity disorder.

  • The observed percentages are provided (first column) together with the fold-enrichment and P-values (from the chi-squared test; last two columns). Grey shaded genes are known or potential primary receptors for HSV-1.

Pathway enrichment analysis of all GWAS genes for each disease (with the exception of ADHD, autism, anorexia and chronic fatigue or childhood obesity, due to a paucity of genes) was also performed using the consensus path database (Table 6). A number of pathogen-related pathways, including the HSV-1 pathway, were significantly enriched in the multiple sclerosis and schizophrenia GWAS data sets. The appearance of the Epstein–Barr virus pathway as the most significant in the multiple sclerosis data set concords with a recent publication highlighting a significant relationship between the Epstein–Barr viral/host interactome and genes from the latest GWAS of the International Multiple Sclerosis Genetics Consortium and the Wellcome Trust Case Control Consortium (Mechelli et al., 2013). The toxoplasmosis pathway also figures in the multiple sclerosis and schizophrenia GWAS data sets, concordant with previous data on a larger set of genes (Carter, 2012b). Pathogen pathways, but not the HSV-1 pathway, also figured in the Parkinson's disease data set. The HSV-1 pathway was also absent from the Alzheimer's disease GWAS data set, although many of the key GWAS genes that do not figure in the KEGG pathway can related to HSV-1 or other pathogens implicated in Alzheimer's disease (Carter, 2011a). Immune- and defence-related pathways were also represented in multiple sclerosis, schizophrenia, depression, Alzheimer's and Parkinson's disease, but not in bipolar disorder.

View this table:
Table 6

Pathway enrichment analysis of genome-wide association study genes (not partitioned in relation to the HSV-1 interactome); KEGG pathways preceded by P-values are shown

Multiple sclerosisInfections: 1.91E−08 Epstein–Barr virus infection, 7.44E−08 Toxoplasmosis, 2.45E−06 Viral myocarditis, 3.09E−06 HTLV-I infection, 3.55E−06 Tuberculosis, 3.25E−05 Herpes simplex infection, 0.83E−05 Leishmaniasis, 0.00015 Influenza A, 3 0.001 Measles, 0.004 Pertussis, 0.006 Hepatitis C, 0.008 Malaria, 0.009 Viral carcinogenesis, 0.012 Staphylococcus aureus infection, 0.013 Chagas disease (American trypanosomiasis), 0.022 Epithelial cell signalling in Helicobacter pylori infection, 0.035 African trypanosomiasis, 0.041 Hepatitis B
Immune: 2.66E−08 Allograft rejection, 1.51E−07 Intestinal immune network for IgA production, 1.94E−05 cytokine–cytokine receptor interaction, 0.00012 Hematopoietic cell lineage, 0.00036 T-cell receptor signalling, 0.00053 B-cell receptor signalling, 0.0014 Fc gamma R-mediated phagocytosis, 0.0024 Toll-like receptor signalling, 0.00463 Antigen processing and presentation, 0.00615 Chemokine signalling, 0.02267 Adipocytokine signalling, 0.024 Fc epsilon RI signalling, 0.024 RIG-I-like receptor signalling, 0.033 Natural killer cell–mediated cytotoxicity, 0.045 Phagosome
Processes and signalling: 4.3E−07 Cell adhesion molecules, 1.2E−06 NF-kappa B signalling, 8.6E−06 PI3K-Akt signalling, 9.1E−06 Jak-STAT signalling; 0.00016 Osteoclast differentiation, 0.0004 Endocytosis, 0.00476 Riboflavin metabolism, 0.007 ErbB signalling, 0.01 Steroid biosynthesis 0.014 HIF-1 signalling, 0.034 TGF-beta signalling, 0.026 Adherens junction, 0.042 Apoptosis
Cancers: 9.9E−06 Acute myeloid leukaemia, 2.6E−05 Transcriptional misregulation in cancer, 0.00082 Proteoglycans in cancer, 0.00203 Thyroid cancer, 0.00382 Chronic myeloid leukaemia, 0.01063 Endometrial cancer, 0.017 Colorectal cancer, 0.02 Pancreatic cancer, 0.043 Prostate cancer, 0.043 Bladder cancer, 0.02 Pathways in cancer
Autoimmune diseases: 1.8E−06 Type I diabetes mellitus, 5.7E−06 Autoimmune thyroid disease, 2.74E−05 Graft-versus-host disease, 0.00013 Asthma, 0.00109 Rheumatoid arthritis, 0.006 Systemic lupus erythematosus, 0.039 Primary immunodeficiency
SchizophreniaInfection: 0.003 Leishmaniasis, 0.015 Toxoplasmosis, 0.018 HTLV-I infection, 0.019 Staphylococcus aureus infection, 0.029 Influenza A, 0.030 Viral myocarditis, 0.032 Tuberculosis, 0.034 Herpes simplex infection, 0.042 Epstein–Barr virus infection, 0.046 Viral carcinogenesis
Immune: 0.00083 Intestinal immune network for IgA production, 0.005 Hematopoietic cell lineage, 0.035 Antigen processing and presentation, 0.036 Chemokine signalling pathway
Processes and signalling: 0.045 Apoptosis 0.048 NF-kappa B signalling pathway
Autoimmune diseases: 0.002 Systemic lupus erythematosus, 0.005 Rheumatoid arthritis, 0.006 Asthma, 0.009 Allograft rejection, 0.011 Graft-versus-host disease, 0.012 Type I diabetes mellitus, 0.017 Autoimmune thyroid disease
Alzheimer's disease0.01 Cell adhesion molecules
Immune: 0.037 Hematopoietic cell lineage
Major depressive disorderProcesses and signalling: 2.70E−05 Serotonergic synapse, 0.0008 Synaptic vesicle cycle, 0.006 Glutamatergic synapse, 0.009 Notch signalling, 0.036 Melanogenesis, 0.046 Neuroactive ligand–receptor interaction
Immune: 0.018 Chemokine signalling
Disorders: 0.009 Cocaine addiction 0.016 Alcoholism
Bipolar disorderProcesses and signalling: 0.0009 Glycosphingolipid biosynthesis – lacto and neolacto series, 0.005 Glycosphingolipid biosynthesis – globo series, 0.016 Tight junction, 0.03 Glycine, serine and threonine metabolism, 0.045 Calcium signalling
Diseases: 0.002 Arrhythmogenic right ventricular cardiomyopathy, 0.003 Hypertrophic cardiomyopathy, 0.028 Dilated cardiomyopathy, 0.048 Type II diabetes mellitus
Parkinson's diseaseInfection: 0.017 Staphylococcus aureus infection, 0.03 Viral myocarditis, 0.03 Leishmaniasis, 0.04 Epstein–Barr virus infection
Immune: 0.009 Lysosome, 0.01 Intestinal immune network for IgA production, 0.04 Hematopoietic cell lineage, 0.03 Antigen processing and presentation
Autoimmune diseases: 0.005 Asthma, 0.008 Allograft rejection, 0.009 Graft-versus-host disease, 0.01 Type I diabetes mellitus, 0.01 Autoimmune thyroid disease, 0.04 Rheumatoid arthritis
Other diseases: 0.02 Basal cell carcinoma, 0.02 Alzheimer's disease
Processes and signalling: 0.01 Hedgehog signalling, 0.01 Cell adhesion molecules, 0.019 Hippo signalling, 0.02 Synaptic vesicle cycle
ADHD/Autism/childhood obesity/anorexia/chronic fatigue, No KEGG pathway enrichment

With respect to Alzheimer's disease, a recent microarray analysis of gene expression in postmortem brains has identified an immune network dominated by genes involved in pathogen phagocytosis, with TYROBP (aka DAP12) as a key regulator. TYROBP (TYRO protein tyrosine kinase binding protein) is a ligand for triggering receptor expressed on myeloid cells 2 (TREM2), a receptor for bacterial lipopolysaccharide. The TYROBP/TREM2 complex triggers activation of the immune responses in macrophages, microglia and dendritic cells (Paloneva et al., 2002; Zhang et al., 2013). TYROBP signalling, presumably via TREM2, increases the viral brainstem titre of systemically infected HSV-1 and increases mortality in infected female mice (Geurs et al., 2012). Thus, genetic and brain gene expression data both highlight an important role for pathogens and the immune system in Alzheimer's disease, as does a substantial enrichment of HSV-1-related proteins in Alzheimer's disease plaques and neurofibrillary tangles (Carter, 2010a).

Natural selection and the role of pathogens in the evolution of the human genome

It is clear, if only from the KEGG pathways illustrated in the PolygenicPathways site that multiple genes in many diseases etch out clearly defined signalling networks that are relevant both to the disease and to the pathogenic risk factors. From the above analysis, it appears that many pathways are forged at the host/pathogen interface.

Mutation and natural selection drive the appearance and maintenance of gene variants in the population. Infections as well as mutations in immune-related genes have played a major role in human evolution (Leal & Zanotto, 2000; Van Blerkom, 2003; Wang et al., 2012). Several have commented on the possibility that the maintenance of disease susceptibility genes in the human population infers that they must convey some benefit (Balaresque et al., 2007; Finch & Morgan, 2007).

Selection pressure particularly applies to lethal infections, as illustrated by the CCR5-delta mutation conferring resistance to the bubonic plague, smallpox and HIV-1 (Galvani & Slatkin, 2003). Descendants of the survivors of such pestilence, endowed with CCR5-delta, might, however, be more susceptible to myocardial infarction or cervical cancer (Singh et al., 2008; Karaali et al., 2010) but resistant to multiple sclerosis or rheumatoid arthritis (Prahalad, 2006; Otaegui et al., 2007), illustrating how survival from infection pandemics could shape the destiny of future generations. In relation to Alzheimer's disease, APOE4 protects against hepatitis C infection (Hishiki et al., 2010), but favours cerebral infection by herpes simplex (Burgos et al., 2003). Hepatitis C kills many, while herpes simplex does not, potentially creating a relatively greater selective pressure for the maintenance of APOE4 within the population. Such selection pressure is likely to vary according to the global distribution of the most prevalent forms of lethal pathogens. Such effects may explain the numerous pathways related to dangerous pathogens in the pathway analyses of the susceptibility genes ( per se; e.g. amoebiasis, influenza, trypanosomiasis, tuberculosis, malaria), which may have shaped this genomic landscape (Table 7).

View this table:
Table 7

The number of susceptibility genes in various diseases that are implicated in KEGG pathogen entry pathways

PathogenMultiple sclerosisSchizophreniaAlzheimer'sParkinson'sDepressionBipolarAutismChronic fatigueChildhood obesityADHDAnorexia
Herpes simplex34211299864332
Epstein–Barr virus2122101249124020
Hepatitis C18101244551100
Chagas disease2616178141043302
African trypanosomiasis1499511502302
Staph aureus139681122020
Helicobacter pylori58311510020
Bacterial invasion of epithelial cells345011100010
Escherichia coli24310350010
Vibrio cholerae14004530001
Dedicated antiviral or antibacterial pathways
Toll pathway2371063441101
NOD pathway126954521101
RIG-1 pathway115421131101
Cytosolic DNA sensing84532200000
General immune pathways
Cytokine–cytokine receptor interaction52311471211134431
Hematopoietic cell lineage3010825873301
Complement and coagulation cascades2711841010140011
Chemokine signalling19166551373301
Fc epsilon RI signalling pathway1812683545511
Antigen processing and presentation17131185483301
T-cell receptor signalling pathway158573263210
Intestinal immune network for IgA production1173235102010
Fc gamma R-mediated phagocytosis11103039110000
B-cell receptor signalling95332640000
Natural killer cell–mediated cytotoxicity811723982201
Leucocyte transendothelial migration651132240010
  • ADHD, attention deficit hyperactivity disorder.

  • The table is ranked from left to right in terms of the mean number of genes for all pathways and from top to bottom by the multiple sclerosis data set. The figures in bold in the grey cells denote the top three pathogens in each class (viruses, parasites and bacteria). The number of susceptibility genes implicated in dedicated antiviral and antibacterial pathways (Toll, NOD, RIG-1 and cytosolic DNA sensing) or general immune-related pathways is also shown for each disease. The numbers refer to the number of genes in each pathway and have not been subjected to pathway enrichment analysis.

Pathogen-derived autoimmunity

Recent studies have shown that the human proteome contains multiple sequences (pentapeptides or longer contiguous or gapped consensus) identical to those within numerous viral and microbial proteins, with the human matches concentrated within networks relevant to diseases in which the pathogen is implicated (Kanduc et al., 2008; Bavaro et al., 2011; Trost et al., 2011). This phenomenon concerns all human proteins. For example, there are 18 000 pentapeptide overlaps between the poliovirus and the human proteome (Capone et al., 2012), while a single immunogenic pentapeptide (VGGVV) within beta-amyloid is shared by HSV-1 and 68 other viruses (Carter, 2010b). Host/pathogen interactomes likely result from this homology, which enables pathogen proteins to mimic their human counterparts and compete with their usual binding partners. This homology, and more particularly slightly differing rather than identical peptides (which are more likely to be recognised as nonself; Kanduc, 2010, 2011), may also contribute to autoimmunity problems that are evident in many diseases. In Alzheimer's disease, multiple sclerosis, schizophrenia and AIDS, antigenic regions of several autoantigens particular to each disease are homologous to proteins expressed by the relevant suspect pathogens (Carter, 2010b, c, 2011b, c, 2012a).

Autoimmunity may play a role in more disorders than currently appreciated. For example, using an array of 9486 human proteins (approximately one-third of the human proteome), even control blood samples averaged more than 1000 autoantibodies, although with extreme intersample variation. It was estimated that we may accumulate more than 3000 autoantibodies, irrespective of any particular disease. A district autoantibody profile was, however, observed in both Alzheimer's and Parkinson's disease, which proved to have diagnostic and predictive value (Nagele et al., 2011; Han et al., 2012). Autoimmune signatures have also been reported in multiple sclerosis (Cameron et al., 2009) and cancer (Wu et al., 2010).

The immune system is trained in early life to recognise the body's own proteins as self (Male et al., 2010). These homologies suggest that the multiple autoantibodies observed, even in the absence of disease, may stem not from some inherent malfunction of the immune system, but from antibodies raised to the numerous pathogens encountered during our lifetime. A study of 600 antiviral monoclonal antibodies generated to 11 viruses estimated that c. 3.5% cross-reacted with host tissue, often targeted at multiple organs (Srinivasappa et al., 1986). This is likely to relate to host/pathogen protein homology, and even after pathogen elimination, continued encounter of these human homologues would sustain an autoimmune response. In this way, pathogens might be able to influence disease processes, long after their successful elimination. The ability of pathogen antibodies to react with specific human proteins could be tested in bulk using the proteome arrays described above.

Antibodies can enter cells via endocytosis (Zhou & Marks, 2012) or transported by the pathogens to which they bind (Baravalle et al., 2004) and can also traverse the blood–brain barrier (Pardridge, 2008). Antibodies can have devastating pathological consequences. In transgenic mice engineered to express nerve growth factor antibodies only in lymphocytes, the blood–brain barrier is soon disrupted, with cerebral antibody entry provoking cortical degeneration, cholinergic neuronal loss, tau hyperphosphorylation and beta-amyloid deposition (the pathology of Alzheimer's disease; Capsoni et al., 2000). If autoantibodies play a key role in the pathogenesis of many diseases, their removal may be of benefit. However, given the large number of autoantibodies, many of which may be required for pathogen defence, this may be no trivial task. However, the number of autoantibodies specific to a particular disease appears more limited, allowing scope for analysis of their pathological or redemptive properties.

Other pathogens

Many other pathogens have been implicated in these diseases. As can be seen from a summary of the KEGG pathways etched out by susceptibility genes (not partitioned in relation to the HSV-1 interactome and not subjected to statistical analysis), these too can be related to the genetic networks, as can a number of dedicated antiviral and antimicrobial defence pathways, and many related to the immune system. These are particularly concentrated in multiple sclerosis, Alzheimer's and schizophrenia and to a large extent in other neurological, psychiatric conditions, but to a lesser extent in chronic fatigue, childhood obesity, ADHD or anorexia (Table 7). The interactomes of such agents (and of others not treated by KEGG) are likely to be more extensive than defined by the KEGG pathways, and further work is necessary to define their relative importance.

It would also be useful to compare the host/pathogen interactomes of other viruses and pathogens with susceptibility gene data sets for these and other diseases where a contributory role is suspected or irrelevant. A centralised database for all host/pathogen interactomes might perhaps also be considered.

Comparisons with the T. gondii host/pathogen interactome

A similar study in relation to the T. gondii/host interactome has recently been published with the same set of diseases and susceptibility genes (Carter, 2012b). In general, the relative contribution of the viral interactome to each of these diseases was greater than that of the protozoan parasite, in terms of enrichment and significance (based on observed/expected values) although in all cases, with the exception of ADHD, the number of overlapping genes was greater for the parasite (a mathematical consequence reflecting its larger interactome; Table 8). Such comparisons of gene lists do not, however, take into account the functional significance of each gene or effect. For example, the ability of herpes simplex to promote beta-amyloid deposition and tau phosphorylation is highly relevant to Alzheimer's disease, while the ability of T. gondii to modulate dopamine function (Prandovszky et al., 2011) most relevant to schizophrenia. It is also likely that certain genes within the interactome/disease overlaps may reflect deleterious effects of the pathogen, while others might well be harmless or even beneficial, a factor not appreciable in terms of gene listing.

View this table:
Table 8

A comparison of the enrichment of HSV-1 or Toxoplasma gondii/host interactome enrichment in the diseases studied. The viral and parasite interactomes contain 1347 and 2792 genes, respectively. Certain but not all disease-related genes were common to both interactomes, and this number is recorded. See subsequent tables for a more detailed analysis. The overall sum and percentage of the combined interactomes relative to each disease gene data set are also shown

N susceptibility genesSpecific overlap with HSV-1 interactome (from 1347)Specific overlap with T. gondii interactome (from 2792)Common to bothCumulative (sum of both interactomes) (%)Fold enrichment HSV-1Fold enrichment T. gondii
Multiple sclerosis407296966164 (40.2)4.12 P = 1.3E−992.83 P = 1.22E−71
Alzheimer's433377049156 (36)3.48 P = 2.19E−672.33 P = 2.26E−41
Schizophrenia7554910653208 (27.5)2.4 P = 1.05E−381.8 P = 3.06E−27
Parkinson's26023312175 (28.8)2.96 P = 1.05E−261.69 P = 3.82E−08
Depression22115292266 (29.8)2.88 P = 4.63E−212.01 P = 2.41E−13
Bipolar443256232119 (26.8)2.27 P = 1.26E−191.81 P = 5.36E−17
Childhood obesity73717731 (42.4)2.91 P = 4.02E−082.69 P = 2.32E−12
Chronic fatigue9567518 (18.9)2.05 P = 0.00051.88 NS
Anorexia7549316 (21.3)1.67 P = 0.0471.38 NS
Autism11174811528191 (17)1.19 P = 0.0111.08 P = 0.013
ADHD232940251 (25.4)0.82 NS1.51 P = 8.01E−05
  • ADHD, attention deficit hyperactivity disorder.

For the diseases where the interactome overlaps were significant for both the virus and the parasite, a certain number of overlapping interactome/susceptibility genes were either specific to each or common to both pathogens. However, there was a cumulative increase in the overall number of susceptibility genes when the overall effects of the two pathogens were summated, in some cases exceeding 40% of the genes sampled (Table 8). Given the number of other pathogens implicated in these disorders, each with distinct life cycles, one might expect further percentage increases when other relevant host/pathogen analyses are taken into account.

Pathways etched out by genes shared by the HSV-1 and T. gondii interactomes

CPDB pathway analysis of the herpes simplex or T. gondii interactome (irrespective of any disease gene overlaps) showed that many of the common genes were primarily concerned with the immune system and with pathogen defence as might be expected (Table 9). They also included several physiological pathways relevant to the diseases in this study, for example, Erbb and growth factor signalling and long-term potentiation in relation to myelination and plasticity, dopamine and cholinergic pathways relevant to neurotransmitter abnormalities, and diverse pathways relevant to neurodevelopment (axon guidance, adhesion, growth factors, apoptosis; Table 9). This overlap suggests that co-infection might be an important consideration. Just as there are epistatic interactions between genes, diverse pathogens are likely to influence each other's risk-promoting potential.

View this table:
Table 9

CPDB pathway enrichment relating to the subset of genes shared by the HSV1 and Toxoplasma gondii host/pathogen interactomes. KEGG pathways and P-values are shown

Immune and pathogen relatedPhysiologyDisease
8.0E−40 Herpes simplex infection
5.5E−37 Tuberculosis
1.6E−36 Leishmaniasis
2.1E−34 Toxoplasmosis
2.2E−34 Hepatitis B
6.4E−33 Influenza A
1.6E−30 Chagas disease
3.6E−30 Toll-like receptor signalling
2.5E−29 Measles
1.2E−28 Cytokine–cytokine receptor interaction
1.2E−25 Pertussis
2.9E−23 Legionellosis
2.2E−21 NOD-like receptor signalling
4.1E−21 Chemokine signalling
1.1E−20 Malaria
1.4E−20 HTLV-I infection
3.1E−19 T-cell receptor signalling
1.2E−17 Salmonella infection
4.6E−17 Natural killer cell–mediated cytotoxicity
1.2E−15 B-cell receptor signalling
3.2E−15 Hematopoietic cell lineage
7.7E−15 Epstein–Barr virus infection
9.0E−14 Hepatitis C
4.0E−13 Phagosome
4.6E−13 Amoebiasis
8.2E−13 African trypanosomiasis
4.8E−12 Viral myocarditis
6.6E−11 RIG-I-like receptor signalling
7.6E−10 Cytosolic DNA sensing
1.0E−09 Leucocyte transendothelial migration
1.8E−08 Antigen processing and presentation
2.0E−08 Intestinal immune network for IgA production
3.9E−08 Fc epsilon RI signalling
5.4E−08 Shigellosis
2.5E−07 Adipocytokine signalling
1.1E−06 Staphylococcus aureus infection
1.5E−06 Epithelial cell signalling in H. pylori infection
0.00089 Fc gamma R-mediated phagocytosis
0.012 Bacterial invasion of epithelial cells
0.023 Complement and coagulation cascades
0.031 Pathogenic Escherichia coli infection
4.8E−31 Apoptosis
1.2E−30 Osteoclast differentiation
9.2E−23 NF-kappa B signalling
2.7E−15 HIF-1 signalling
1.1E−13 Jak-STAT signalling
2.7E−13 PI3K-Akt signalling
4.1E−13 MAPK signalling
2.4E−11 VEGF signalling
1.4E−10 Cell adhesion molecules
1.8E−08 ErbB signalling
5.2E−08 Neurotrophin signalling
1.9E−07 Focal adhesion
2.1E−07 p53 signalling
3.7E−07 mTOR signalling
6.8E−07 Oestrogen signalling
1.8E−05 Regulation of actin cytoskeleton
8.4E−05 Axon guidance
8.8E−05 Long-term potentiation
9.6E−05 ECM–receptor interaction
0.00015 Insulin signalling
0.00026 Cholinergic synapse
0.00029 Dopaminergic synapse
0.00054 Adherens junction
0.00058 Base excision repair
0.00073 Wnt signalling
0.00075 GnRH signalling
0.0029 Protein processing in endoplasmic reticulum
0.0031 Long-term depression
0.004 Regulation of autophagy
0.004 Glutamatergic synapse
0.006 Calcium signalling
0.0064 Endocytosis
0.0080 Aldosterone-regulated sodium reabsorption
0.010 Oocyte meiosis
0.012 Circadian entrainment
0.013 Proteasome
0.015 TGF-beta signalling
0.019 Cell cycle
0.021 Progesterone-mediated oocyte maturation
0.026 PPAR signalling
0.030 Serotonergic synapse
0.034 Ubiquitin mediated proteolysis
0.045 Melanogenesis
7.0E−20 Pathways in cancer
5.4E−19 Proteoglycans in cancer
3.6E−16 Rheumatoid arthritis
1.9E−15 Small cell lung cancer
1.9E−15 Amyotrophic lateral sclerosis
4.3E−15 Prostate cancer
1.1E−13 Alzheimer's disease
6.1E−12 Colorectal cancer
1.7E−11 Acute myeloid leukaemia
2.1E−11 Viral carcinogenesis
3.9E−11 Prion diseases
5.9E−11 Allograft rejection
1.1E−10 Chronic myeloid leukaemia
5.1E−10 Type I diabetes mellitus
1.9E−09 Pancreatic cancer
4.5E−08 Graft-versus-host disease
6.9E−08 Endometrial cancer
9.5E−07 Non-small cell lung cancer
1.3E−06 Transcriptional misregulation in cancer
2.5E−06 Bladder cancer
5.2E−06 Autoimmune thyroid disease
1.3E−05 Amphetamine addiction
4.0E−05 Glioma
4.6E−05 Renal cell carcinoma
6.93E−05 Hypertrophic cardiomyopathy
0.00013 Primary immunodeficiency
0.00013 Type II diabetes mellitus
0.00033 Asthma
0.00044 Melanoma
0.00063 Dilated cardiomyopathy
0.0013 Systemic lupus erythematosus
0.0021 Thyroid cancer
0.0025 Arrhythmogenic right ventricular cardiomyopathy
0.005 Cocaine addiction
0.034 Huntington's disease
  • PPAR, peroxisome proliferator–activated receptor.

Pathways etched out by genes specific to the HSV-1 or T. gondii interactomes

With the exception of five common physiological pathways (PI3K-Akt signalling, salivary secretion, NF-kappa B signalling pathway, cell adhesion molecule and the Krebs cycle (same pathways/different genes), those etched out by the genes specifically implicated in the HSV-1 or T. gondii host/pathogen interactome were divided primarily into signalling networks for the virus and metabolic pathways for the parasite (Table 10). This illustrates a major difference between viral and cellular pathogens, the latter exerting a significant effect on host metabolism by both scavenging nutrients and donating chemical compounds to various metabolic chains. Regardless of the differences between shared and common genes and their derived pathways, it should be noted that each individual interactome is composed of a summation of both specific and common gene data sets.

View this table:
Table 10

CPDB pathway enrichment analysis (physiological pathways and metabolism) in relation to the host/pathogen interactome genes specific to herpes simplex or Toxoplasma gondii: pathways common to both pathogens (although by definition with different genes) are highlighted in bold

Herpes simplexToxoplasma gondii
2.83E−11 Cell cycle
3.53E−10 GnRH signalling
1.76E−09 MAPK signalling
1.05E−08 Long-term potentiation
2.90E−08 Neurotrophin signalling
1.23E−07 Oocyte meiosis
5.11E−07 ErbB signalling
5.86E−07 Mismatch repair
1.08E−06 Oestrogen signalling
2.04E−06 Endocytosis
5.01E−06 Homologous recombination
7.45E−06 Dopaminergic synapse
1.09E−05 Gap junction
1.15E−05 Adherens junction
3.63E−05 PI3K-Akt signalling
6.07E−05 DNA replication
7.90E−05 Nonhomologous end-joining
0.00010 RNA transport
0.00017 Insulin signalling
0.00029 Wnt signalling
0.00045 Vasopressin-regulated water reabsorption
0.00045 Basal transcription factors
0.00054 Nucleotide excision repair
0.00057 TGF-beta signalling
0.00062 NF-kappa B signalling
0.00080 Osteoclast differentiation
0.00100 VEGF signalling
0.00132 Long-term depression
0.00166 Salivary secretion
0.00239 Circadian rhythm
0.00312 Progesterone-mediated oocyte maturation
0.00337 Glycosaminoglycan biosynthesis – heparan sulphate/heparin
0.00346 Insulin secretion
0.00494 Melanogenesis
0.00512 Cholinergic synapse
0.00654 Ubiquitin mediated proteolysis
0.00759 HIF-1 signalling
0.01073 Citrate cycle (TCA cycle)
0.01090 p53 signalling
0.0120 Focal adhesion
0.0163 Dorso-ventral axis formation
0.0245 Apoptosis
0.0320 Notch signalling
0.0384 Cell adhesion molecules
2.82E−05 Purine metabolism
0.0006 Pyrimidine metabolism
0.0008 Glycolysis/Gluconeogenesis
0.0008 Glyoxylate and dicarboxylate metabolism
0.0017 Jak-STAT signalling
0.0023 Propanoate metabolism
0.0028 Amino sugar and nucleotide sugar metabolism
0.0032 Fatty acid metabolism
0.0040 beta-Alanine metabolism
0.0040 Galactose metabolism
0.0044 Tryptophan metabolism
0.0055 Lysine biosynthesis
0.0056 Cell adhesion molecules (CAMs)
0.006 Fructose and mannose metabolism
0.006 PI3K-Akt signalling
0.0083 ECM–receptor interaction
0.009 Carbohydrate digestion and absorption
0.009 Glutathione metabolism
0.011 Arginine and proline metabolism
0.011 Mineral absorption
0.012 Butirosin and neomycin biosynthesis
0.017 Citrate cycle (TCA cycle)
0.02 RNA polymerase
0.02 Endocrine/other factor-regulated calcium reabsorption
0.025 Valine, leucine and isoleucine degradation
0.029 NF-kappa B signalling
0.029 Pentose phosphate
0.035 Nicotinate and nicotinamide metabolism
0.041 Proximal tubule bicarbonate reclamation
0.046 Starch and sucrose metabolism
0.046 Glycerolipid metabolism
0.049 Salivary secretion

Population genetics and a proposed gene/environment interaction model

The mechanisms described above provide a general example of multiple gene/environment interactions in relation to a single pathogen interactome with more than 1000 interactions (Fig. 2). Even for a simple population genetics model, with two genes and three environmental factors, varying permutations can dramatically influence the eventual outcome. For example, the light- and dark-coloured genes of the peppered moth, or the light and dark colours of the clean or polluted trees on which they alight, can all be either risk promoting or protective vs. predation by birds, depending on the varying permutations of genes and coloured trees (Kettlewell, 1955). Neither gene nor tree colour is relevant if there are no hungry birds, while predation of the moths by bat species at night renders these same genes and tree colours totally irrelevant to a different cause of death. From an epidemiological standpoint, the real causes of death can be rendered invisible, as the birds or bats are always present, in all gene/tree colour conditions, whether the moths are alive or dead. Given that several different pathogens can induce beta-amyloid deposition, demyelination or neurodevelopmental problems (see Introduction), the same model might well apply to complex diseases where polygenic may also equate to ‘polymicrobial’, with different subsets of susceptibility genes dictating the relevance of particular pathogens. The same principle also applies to other environmental factors. For example, hypercholesterolaemia, hyperhomocysteinaemia, streptozotocin-induced diabetes, cerebral hypoperfusion (ischaemia, hypoglycaemia, hypoxia), vitamin A, oestrogen or nerve growth factor deficiency are all able to induce beta-amyloid deposition in laboratory models and subsets of genes implicated in Alzheimer's disease relate to each of these parameters (Carter, 2011d).

Figure 2

A model of the host pathogen interactome illustrating how multiple gene/environment interactions might direct the attentions of the pathogen towards distinct pathways processes and diseases. For any pathogen, immune and pathogen defence pathways as well as inflammatory processes will be activated to counter the infection. Although the pathogen can interact with hundreds of host genes and proteins, those chosen will depend upon the strain of pathogen, the timing and localisation of infection and on whether prior immune barriers exist. In turn, which human elements are available for interaction will depend upon their expression (time and location, as with the receptor gates) and upon their functional quirks dictated by gene variants. This selection process, involving a genetic sieve and individual interaction probabilities, enables similar interactome selectivity, allowing the pathogen to specifically affect different series of pathways in different circumstances (illustrated by the number of human proteins ending their route in a particular pathway bin). The differential modification of particular pathways will in turn affect particular processes and endophenotypes, whose final assembly constitutes the eventual disease. This triage, involving both human and pathogen genes and proteins, as well as environmental factors explains how the same pathogen could cause or otherwise influence a variety of diseases, depending upon genetic factors and a series of co-incidences (see text for further details).

The high prevalence of common pathogens, such as herpes simplex, has militated against their implication as causes of relatively uncommon diseases: However, such gene/environment filtering effects could well explain how such pathogens could cause certain features of disease in genetically susceptible individuals, and in certain environmental conditions. Certain environmental factors, equating to tree colour in the peppered moth model, might also be relevant to HSV-1 infection. For example, HSV-1 entry is cholesterol dependent and blocked by statins (Bender et al., 2003); viral replication can be reduced by retinoic acid derivatives (vitamin A; Isaacs et al., 1997) and also by curcumin derivatives (Zandi et al., 2010) or the ingredient of red grapes and other fruits, resveratrol (Docherty et al., 1999). It is also recognised that drugs of abuse (cannabis, opiates, cocaine and nicotine) have immunomodulatory effects and that chronic drug abuse is associated with increased levels of infections (Friedman et al., 2003). Vitamins A and D,omega-3 polyunsaturated fatty acids are known to exert general anti-inflammatory effects and also play an important role in regulatory T-cell function (Friedman et al., 2003). Many of these environmental factors also have a significant influence on disease incidence.

The large number of gene/environment interactions for a single pathogen is also likely to condition the risk-promoting effects of polymorphic human genes, a factor that is relevant to replication problems in genetic studies. Polymorphic genes will have an effect not only on their human function but also on the ability of pathogens to affect this function.

If one splits a complex disease into its component parts (e.g. myelination, cell death, inflammation, neurotransmitter abnormalities) and given the number of processes involved in the HSV-1/host interactome, this single pathogen could act either as a provoking or as a protective agent, for particular aspects of pathology, depending upon the pathways influenced the most.

Such effects are based on a simple concept that each interaction has an effect on the processes and pathways regulated by the human protein concerned, which is deprived of function if bound to a viral protein (or antibody), perhaps explaining why, despite the fact that most susceptibility gene variants are single nucleotide polymorphisms, certain endophenotypes can reliably be produced by gene knockout (e.g. Desbonnet et al., 2009), a situation that does not exist in the human condition, at the gene expression level.

This suggests a model that may have general application to many other pathogens implicated in disease. If one imagines the viral components as a number of spheres, each with particular affinity for certain human targets, and their human partners as a further series of spheres perched on a genetic ledge whose characteristics and apertures are regulated by gene variants, the trajectory of each, dropped through this genetic sieve, and falling through the apertures, will be influenced by the strain of pathogen, host/pathogen affinities, the dropping point (route of infection and viral receptors), the timing of infection, when and where different human genes are expressed, and by the polymorphic genes.

Each human gene controls a particular element of one or many pathways, differentially positioned in reception bins beneath the sieve. Depending upon permutations of these factors, the number of spheres in each bin will vary, resulting in a spectrum of pathway disturbance. Assembly of this pathway mosaic leads to particular endophenotypes or subpathologies, which together constitute a particular disease. The same pathogen could thus produce diverse effects ranging from cause to prevention depending on a permutation of circumstance.

The genes and risk factors, as well as the immune system thus work together to determine the final outcome, while neither per se, are likely to provoke a particular disease. Clearer understanding of these effects, involving multiple pathogens and genes, could eventually lead to disease prevention or cure in multiple conditions.


This analysis shows that the HSV-1 host/pathogen interactome is highly concentrated within susceptibility genes for many psychiatric and neurological disorders, and to a lesser, but appreciable, extent in chronic fatigue, childhood obesity and anorexia, but not ADHD. Many pathways etched out by susceptibility genes lie at the host/pathogen interface, suggesting that evolutionary pressure and natural selection against infections may have contributed to the maintenance of risk-promoting polymorphisms in many diseases. The herpes virus or its inflammatory consequences is able, per se, to induce many key pathological features of psychiatric and neurological disorders, suggesting that susceptibility genes relate to the causes of disease rather than (and as well as) to the disease itself. Viral detection, antiviral drugs and vaccination might be expected to affect the incidence of many diseases, although host/pathogen protein mimicry suggests that the effects of such treatments may be mitigated by the problems of established autoimmunity that is potentially of pathogen origin. The herpes virus is but one of multiple species of an extensive microbiome, whose influence on disease is increasingly apparent (Cho & Blaser, 2012). A fusion of genetic, microbiological and autoimmune data is essential to understand the complexity of polygenic diseases where genes, environment, the microbiome and the immune system together, and not singly, appear to influence the genesis and severity of disease.

In certain instances, where a clear pathological effect of the HSV-1 virus has been established, for example, its ability to provoke beta-amyloid deposition and tau phosphorylation, relevant to Alzheimer's disease, there is clearly a case for clinical trials with antiviral agents, as argued by others (Miklossy, 2011b; Wozniak et al., 2011; Ball et al., 2013; Kobayashi et al., 2013; Wozniak & Itzhaki, 2013). Given the evident ability of this and other relatively ‘benign’ viruses and other pathogens to influence disease-relevant pathways, it is also possible that vaccine development could have a marked influence on the incidence of a variety of disorders.

Supporting Information

Table S1. Results of the KEGG pathway analysis of the HSV-1 host/pathogen interactome: immune and defence pathways, diseases and other infections.

Table S2. Results of the KEGG pathway analysis of the HSV-1 host/pathogen interactome: signalling networks, Tissue and cellular process, and neuronal related pathways.

Table S3. HSV-1 interactome genes that were shared across various neurodegenerative disease gene datasets.

Table S4. Psychiatric disorders (ADHD, autism, depression, bipolar disorder and schizophrenia): HSV-1 interactome genes that were shared across various neurodegenerative disease gene datasets.

Table S5. Neurological diseases: HSV-1 interactome genes specifically overlapping with individual neurological disease gene datasets.

Table S6. Psychiatric diseases: HSV-1 interactome genes specifically overlapping with individual psychiatric disease gene datasets.


I am indebted to the KEGG staff for their inspiration and for permission to postpathways on the PolygenicPathways Website and to the numerous authors who have provided reprints. The author runs the PolygenicPathways Website, curating genes, risk factors and interactomes related to this and other publications. Data are referenced with links to PubMed. The database is funded by the author and by the advertisements on the site.


View Abstract