Herpes and Alzheimer’s Disease: Subversion in the Central Nervous System and How It Might Be Halted

REACTIVATION OF HSV1 IN BRAIN

One of the most important aspects of the concept is that of reactivation of HSV1 in brain. Several studies were described in detail in the recent review [1], including a finding that provides major direct evidence of reactivation in brain: an examination of CSF samples sent to a reference laboratory for HSV testing [4], revealed, unexpectedly, that 26 of the 3200 samples were positive for the viral DNA. HSV DNA is present in CSF of herpes simplex encephalitis (HSE) patients but the disease could not explain the relatively very high proportion of samples that were viral DNA-positive (8 per thousand) as the prevalence of HSE in the population is much lower, only ∼2 per million. A further reason why HSE could not account for the results is that after HSE, HSV1 DNA remains in the CSF only briefly, for about a week (in contrast to the long life of intrathecal antibodies to HSV). The data thus not only confirm HSV1 presence in brain but also suggest that HSV1 reactivation in brain is not so infrequent. The finding is consistent with an earlier study using in situ hybridization, which suggested that immunosuppression causes latent HSV to reactivate and that subsequent replication leads to its amplification [5]: HSV DNA was detectable in postmortem brain specimens of subjects who had been immunosuppressed and were seropositive for HSV, but not in those who were seronegative or who had not been immunosuppressed.

Another relevant early study [6] described cases of “mild” encephalitis: patients’ symptoms were less severe than usual and recovery was almost complete, with only minor sequelae, even though the patients were not treated with antivirals. In the days before antiviral treatment became routine, most cases of HSE were usually fatal, and those who survived often suffered severe neurological problems, including memory loss. Klapper et al. referred also to recurrent HSE, i.e., reactivation of latent HSV1 present in brain, and they suggested that other recurrences might not always have been recognized. They presciently speculated: “Is it possible that one or more reactivation events [of HSV1] resulting in mild disease could play an etiological role in such conditions [chronic psychiatric illness]”? Cases of recurrent HSE are still quite often reported, but of course they would not be seen by neuropathologists, who examine only fatal cases, i.e., the most severe, and who might well conclude therefore that “mild” encephalitis does not exist.

In mice, HSV1 latency in brain is established a few weeks after inoculation of the virus. Many early investigations indicated that its reactivation in brain was far rarer than in the trigeminal ganglia (TG), as determined by assays of dissociated and minced tissue such as the ex vivo (explant) assay of reactivation frequency (in which latently infected cells are cultured with susceptible uninfected ones). However, two recent very interesting studies indicate that HSV1 in mouse brain can in fact be reactivated relatively easily. Yao et al. [7] examined the brain stem and TG of HSV1-infected animals during latency and unexpectedly found in brain a greater number of copies of the viral genome, and also more frequent reactivation, than in the TG. They attributed this difference from previous results to a more rapid loss of viability of brain stem cells than of TG cells after dissociation, and especially after mincing, so that ex vivo measurements would not accurately assess viral reactivation in brain. The data of Ramakrishna et al. [8] were equally striking: they investigated HSV1-infected immunodeficient mice (lacking B and T cells) which were treated with intravenous immunoglobulin (IVIG) to promote long-term survival (via IVIG’s immunomodulatory and antiviral activities). After high dose HSV1 inoculation, mice in which viral latency had been established in brain showed spontaneous reactivation of the virus; this was suppressed by T cells but not B cells. Hyperthermic stress caused HSV1 reactivation in brain of most of the animals, with subsequent occurrence of HSE.

Productive HSV1 infection causes damage via inflammatory processes as well as by direct viral action. Some of these processes can occur also during latency [9]: in HSV1-infected mice, several inflammatory markers such as toll-like receptor-4, interferon α/β, and p-IRF3, characteristic of viral replication, were all detectable in brain at a time well after virus inoculation, and therefore after the establishment of viral latency in brain, thus indicating that reactivation had occurred. The authors concluded that HSV-1 presence in the CNS could cause chronic neuroinflammation through recurrent reactivation, leading to activation of toll-like receptors and thence to cumulative neuronal dysfunction. All these data support the proposal that HSV1 reactivates–not just in the periphery, but also in the brain.

CELL BIOLOGICAL STUDIES

In another type of approach–cell biological studies aiming to find if the characteristic abnormal molecules found in AD brains can be produced by HSV1 infection–HSV1-infected cell cultures revealed accumulation of amyloid-beta (Aβ) [10–12], and of AD-like tau (P-tau) occurs [13–16]. Implicating HSV1 further in AD was the discovery that in AD brains, most of the HSV1 DNA is very specifically localized in amyloid plaques [17]: in brain of elderly controls, a much lower proportion of the viral DNA is present in plaques, presumably reflecting a lower extent of synthesis of Aβ, or else a more efficient removal of the peptide. The HSV1-induced increases in Aβ and P-tau were accounted for by increases in the relevant enzymes via, in the case of BACE, HSV1-induced PKR activation followed by phosphorylation of eukaryotic translation initiation factor 2-alpha (eIF-2α) [18]; eIF2α shuts off general protein synthesis, but reverses the inhibitory effect of the BACE1 5’ untranslated region (5’UTR) in the BACE promoter on BACE expression. The PKR polymorphisms in AD patients discovered by Bullido et al. [19] could affect this process, thereby leading to the observed high level of activated PKR in AD brains.

Santana et al. [20] investigated the effects of mild oxidative stress combined with HSV1 infection of cells (which itself causes oxidation). They found that oxidative stress significantly augmented the HSV1-induced accumulation of Aβ and its secretion, as well as the inhibition of autophagy, although it did not increase the degradation of long-lived proteins. These oxidative effects were not attributable to enhanced virus replication as, surprisingly, oxidation reduced viral DNA replication and reduced even more the formation that leads to the neurodegeneration seen in AD.

Civitelli et al. [21] found that HSV-1 infection of cultured mouse cortical neurons and SH-SY5Y neuroblastoma cells causes the production of several APP fragments, including the APP intracellular domain (AICD). AICD binds the promoter region of both neprilysin (NEP), the major Aβ-degrading enzyme, and GSK3β, the enzyme causing hyperphosphorylation of tau. NEP level and enzyme activity were initially stimulated by infection but later were down-regulated. GSK3β level and activity remained almost constant, although at late stages of infection the enzyme was inactivated through being phosphorylated at Ser9. However, a second study by the same group [22] showed that HSV1 caused activation of phosphorylated GSK3. The activation of pGSK3 was Ca2+-dependent and was essential for the HSV-1-dependent phosphorylation of APP at Thr668, leading then to its subsequent degradation and to the intraneuronal accumulation of Aβ. A very significant finding was that HSV-1 infection reduced the expression of the presynaptic proteins synapsin-1 and synaptophysin, and depressed synaptic transmission. By using 4G8 antibody which binds to Aβ, and also by infecting APP-knockout mice, the authors showed that these inhibitory effects on synaptic function were dependent on GSK-3 activation and intraneuronal accumulation of Aβ.

The increase in Aβ that HSV1 causes raised the possibility that at least initially, the peptide at low levels might function as part of the innate immune system, acting protectively as a “bioflocculant”, i.e., binding neurotoxic agents, as previously suggested by Robinson and Bishop [23], or as an anti-microbial peptide [10]; however, in the latter study, although Aβ appeared to have antiviral activity, it was attributable to its toxic effect on the cells. Furthermore, virucidal assays, which assess the capacity of the test molecule to inactivate virus particles, showed no effect on viral infectivity. However, in view of recent positive findings (see below), the antiviral activity is probably determined by the method of its preparation and its state of aggregation. In any case, though, Aβ eventually becomes toxic when over-produced and when oligomerization occurs.

DOES Aβ HAVE ANTIMICROBIAL PROPERTIES?

There is now evidence that Aβ, which structurally resembles antimicrobial peptides (AMPs) and, like them, can cause activation of immune cells, does indeed have antiviral activity. A number of studies have implicated certain bacteria–spirochetes [24] and Chlamydia pneumoniae (C. pneumoniae) [25], as well as HSV1, in the development of AD. Both types of bacteria elicit the formation of Aβ and P-tau, and components of both colocalize with AD pathology. The antibacterial activity of Aβ was detected first by Soscia et al. [26] and is discussed later in this section together with a recent study from the same group.

The first paper on the antiviral properties of Aβ [27] investigated its effect on influenza virus A during infection of several human and canine epithelial cell cultures, used as model systems. The authors found that the activity of Aβ₄₂ was much greater than that of Aβ₄₀, and that the maximum antiviral effect of Aβ₄₂ was achieved when it was pre-incubated with the virus, thereby indicating that it acts on the virus rather than on the cell. Also, Aβ caused aggregation of the virus, reduced viral protein synthesis, and modulated its interaction with phagocytes.

As to the effect of Aβ on HSV1, Bourgade et al. [28] infected cell cultures with the virus and found that both Aβ₄₀ and Aβ₄₂ inhibited HSV1 DNA replication when added to the cultures. This occurred either when the peptides were added before the virus or when added together with it, but not when added after virus addition. Also, in a cell-free system, Aβ interacted directly with HSV1 (as Aβ₄₂ did with influenza virus), indicating that in the cell cultures, it prevented HSV1 entry into cells. Both this and the influenza virus study showed also that Aβ acts selectively against enveloped viruses as opposed to non-enveloped viruses, and Bourgade et al suggested that this might reflect Aβ insertion into the viral envelope. In a second study, Bourgade et al. [29] used co-cultures of neuroglioma (H4) and glioblastoma (U118-MG) cells as an in vitro model, and found that the H4 cells secreted Aβ₄₂ in response to HSV-1 challenge, and that U118-MG cells could rapidly internalize Aβ₄₂. Extraneous Aβ₄₂ induced strong production of cytokines in the cell lines, and a combination of Aβ₄₂ and HSV-1 induced the production of the pro-inflammatory cytokines TNFα and IL-1β, and IFNα in the cell lines. Aβ₄₂-conditioned medium from HSV-1-infected H4 cells, when added to cultures of H4 cells, conferred Aβ-dependent protection against HSV-1 replication when the cells were challenged with HSV-1. The authors proposed that in human brain, Aβ₄₂ acts as an AMP against neurotropic enveloped viruses such as HSV1; also, in agreement with the present author’s suggestions, they considered that eventual overproduction of Aβ peptide might contribute to amyloid plaque formation.

Intriguingly, α-synuclein (Asyn), another AMP-like peptide, has very recently been shown by Beatman et al. [30] to have antiviral activity against certain enveloped RNA viruses. Infection of primary neurons with West Nile virus (WNV) or with Venezuelan equine encephalitis virus caused an increase in Asyn expression, and infection of Asyn knock-out mice resulted in a huge increase in number of infectious viruses, and much greater subsequent mortality, compared with wild-type and heterozygous litter mates. The authors suggested that WNV-induced Asyn inhibits viral replication, growth, and injury in the CNS and that the peptide has a novel and important functional role in the development of Parkinson’s disease.

Both the influenza and the HSV1 studies tested Aβ efficacy as an antiviral by assaying virus level, using quantitative PCR on viral DNA extracted from the cell cultures. However, PCR has the disadvantage of measuring DNA not only from “live” but also from inactivated virus, thereby over-estimating the virus level. Also, in the HSV1 studies, the Aβ concentration used was high (20μg/ml) probably very much greater than the levels in brain cells. It would therefore be well worth extending the studies using a much lower Aβ concentration and assaying virus levels by standard virological methods, such as the plaque assay (the method used by Beatman et al. [30]).

Further strong evidence for the protective role of Aβ, although unexpectedly in its oligomeric form, has been obtained in an interesting, very detailed study by Kumar et al. [31]. This followed work by the same group examining the effect of synthetic Aβ on the growth of eight pathogens, the yeast Candida albicans (C.albicans) and seven common types of bacteria, in culture, which indicated that Aβ has a protective role in innate immunity [26]. In the more recent study, the microbes investigated were the bacterium, Salmonella typhimurium (S. typhimurium), and the yeast C. albicans. The targets were transfected human neuroglioma cells (H4) over-expressing Aβ, transgenic (Tg) nematodes, Caenorhabditis elegans (C. elegans), expressing Aβ in body wall muscle, and Tg mice overexpressing Aβ. The authors showed that Aβ protected the cultures of transfected cells and also the Tg nematodes, greatly increasing their survival when infected by C. albicans. Similarly, the Tg mice survived infection with S. Typhimurium for a far longer time period than did wild-type and APP knock-out mice. To examine the protective mechanism, the authors compared Aβ with an antimicrobial peptide (AMP), LL-37, which is known to protect against microbes by oligomerizing and binding to their surface, thereby preventing their attachment to the target cells, and then forming fibrils round them so that they are immobilized. On infecting the transfected H4 cells with C. albicans, the authors found that the transfected cells bound fewer yeasts than did non-transfected H4 cells, and that the Aβ bound to the yeast cell walls, but only if it was in oligomeric form; then, like LL-37, the Aβ wrapped up the yeast. Similarly, on infection of the nematodes, the yeasts became entrapped and the clumps thus formed were stainable with thioflavin S, as are amyloid plaques in human brain. Further, in Tg mice–animals that normally develop amyloid plaques only at a later age–plaques were seen in young mice at just 2 days after infection with S. Typhymurium. The authors commented that the same features, oligomerization, fibrillization, and carbohydrate binding, are associated also with the pathophysiological effect of Aβ, and they suggest that dysregulation of the normal protective activity of Aβ leads to AD pathology.

Unfortunately, the authors did not investigate infection with HSV1 in either study, despite its being the pathogen most frequently implicated in AD, and implicated via diverse approaches. In fact their immobilized pathogen model strikingly resembles the pictures of HSV1 DNA embedded within amyloid plaques in AD brains, a finding published in 2009 [17], and the bioflocculent model proposed by Robinson and Bishop in 2002 [23]. Further, the bacteria and the yeast investigated in the studies described above have never been associated in any way with AD, yet the two types of bacteria that are strongly implicated in the disease (spirochetes and C. pneumoniae) both of which, significantly, are intracellular, were not investigated.

HSE AND Aβ

Interestingly, a link between HSE and Aβ was discovered by Bearer et al. [32], who investigated autopsy brain tissue from three HSE patients: a 9-day-old, a 8-year-old, and a 76-year-old (the latter showing no evidence of AD). They detected Aβ but no P-tau in brain of each subject. Aβ was not detected in cases of non-herpetic viral encephalitis. They concluded that HSV can induce the formation of Aβ deposits, and recommended future follow-up of HSE patients who survive to find if the plaques and HSV1 infection persist, i.e., if more Aβ is deposited.

ARE THERE OTHER HERPESVIRUSES IN THE ELDERLY BRAIN?

There have been very few studies on the possible involvement of other herpes viruses in AD. Most of these viruses, if detected at all in brain, were found in a relatively low proportion of AD patients and elderly controls, compared to HSV1, apart from human herpesvirus type 6 (HHV6) which, in the author’s laboratory was detected in brain of 70% and 40% of AD patients and age-matched controls respectively [28]. It was suggested that as there was considerable overlap of HHV6 and HSV1 in brain, HHV6 might act together with HSV1 in the development of AD. Previously, the same laboratory found no varicella zoster virus (VZV) in brain [34], but detected HSV2 in 13% and 20%, respectively, of patients and controls, cytomegalovirus (CMV) in 36% and 34%, respectively [33], and in 93% of vascular dementia patients [35].

Carbone et al. [36] sought the presence of the DNA of CMV, Epstein Barr virus (EBV), and HHV6 in peripheral blood leucocytes (PBL) and in brain. No CMV was detected in any samples, but EBV was detected in 45% of PBL from AD patients, 31% from controls, and in 6% of AD brains. HHV6 was detected in 23% PBL from AD patients, 4% from controls, and in 17% of AD brains. In subjects followed for 5 years, the percentage positive for EBV and HHV6 increased in those who developed AD, as did serum IgG levels for CMV and HHV6. They considered that the non-detection of CMV DNA, in contrast to their anti-CMV antibody detection and to the data of Lin et al. [33], possibly reflected the inability of their technique to detect low levels of CMV DNA. They concluded that EBV, HHV6, and perhaps CMV might all be implicated in the progression to AD.

In a later study [37], the authors examined AD patients and elderly controls over a five-year period for cognitive performance and for clinical diagnosis of AD, investigating genetic factors regulating antiviral response, such as IFN-λ3. They found that the genes responsible were associated with increased risk of cognitive decline and AD, again implicating EBV and HHV6, and they proposed that impaired immunity against persistent viruses, such as herpesviruses, in genetically predisposed elderly people might cause recurrent virus reactivation from latency, hence activating brain microglia, and increasing Aβ production and accumulation. An earlier publication, by Carter [38] had discussed putative antiviral host responses, specifically to HSV1, which could affect its infectivity or replication; these included nitric oxide, cysteine protease inhibitor cystatin C and certain cytokines, namely, IL1A, IL2, IL1RN, IL6, IL18, and TNF and as Carter commented, their effects would be influenced by any polymorphisms.

Recently, the effects of HSV2, another herpes virus highly homologous to HSV1, have been studied in cultured human neuroblastoma cells [39]. HSV2 was found, like HSV1, to cause increased accumulation of abnormally phosphorylated tau and Aβ, altered APP processing, and impaired autophagy. The authors suggested that HSV2 (and other herpesviruses) might play a role in AD as it remains latent in sensory neurons but is capable of reactivating, and it can infect the brain and cause neurological symptoms, just as HSV1 infection does. However, they acknowledged that HSV2 usually causes HSE only in neonates, not in adults, and that serological data show that the virus infects a much lower proportion of the population, and resides in far fewer elderly human brains than does HSV1.

All these results, together with the discovery that in AD brains, almost all the HSV1 DNA resides within amyloid plaques [17], suggests that in many AD patients, HSV1 in brain is responsible for the abnormal processing of amyloid precursor protein (APP), for the formation of Aβ, its toxic aggregates, and of plaques, for abnormal phosphorylation of tau, and for synaptic dysfunction: the major features of AD. Whether the other herpesviruses contribute remains to be confirmed; possibly in the case of CMV, its action might be through immune dysregulation, as proposed by Stowe et al. [40].

EPIDEMIOLOGICAL STUDIES

There have been further epidemiological investigations on anti-HSV1 IgG and IgM antibodies in serum from AD patients. The rationale for the serum antibody work is that while the presence of IgG shows that the person has been infected with HSV1, the presence of IgM indicates that recent reactivation of HSV1 has occurred. However, serum antibody levels reflect the response to the virus in the periphery; whether or not they reflect response to the virus in brain is unknown because at present, no imaging method can detect either latent HSV1 in brain, or reactivated virus if present at very low levels. It does seem likely though that events that cause reactivation in the periphery, such as stress and immunosuppression, would cause reactivation also in the brain, but perhaps less severely.

Many antibody studies have shown an association between systemic infections and cognitive decline, with HSV1 as the main suspect [41–46] – but see comment on [46] by Itzhaki and Klapper [47]. Letenneur et al. [44], Feart et al. [45], and Lovheim et al. [46] mainly implicated IgM, thereby suggesting that HSV1 reactivations were the events leading to the development of AD, However, a second study by Lovheim et al. [49] found, surprisingly, an association of IgG, but not IgM, with AD, thus implicating HSV1 presence rather than activity in AD. This result contradicted the authors’ previous data and those of others, so to explain the difference they suggested that it might result from the different approaches used–the previous ones being cohort studies, and their present one a case-control study. Alternatively, it could be because of HSV1 affecting an early stage in AD development, or perhaps it reflected their paucity of IgM-positive subjects.

Two investigations have been made on the possible association, in young subjects, of infection by a specific virus, or of infectious burden (I.B. - seropositivity to several microbes), with cognition or AD. One study investigated 612 soldiers in the Israeli military (59% male and 41% female, aged 19–21) for HSV-1 infection and possible association with cognitive functioning and language abilities [50]. After controlling for education, immigration status, and sex (although not for socio-economic status), and removing those with mild to moderate mental illness, the 62% who were seropositive for HSV-1 infection were found to have lower IQ and lower language skills. The second study on young to middle aged subjects [51] examined serum IgG antibodies to toxocariasis, toxoplasmosis, hepatitis A, hepatitis B, and hepatitis C, CMV, HSV1, and HSV2, in over 5,000 subjects aged 20–59 years. Cognition was assessed by three tests: the Third National Health and Nutrition Examination Survey computer-based simple reaction time (SRT), symbol-digit substitution (SDS), and serial-digit learning (SDL) tasks. The infectious burden index was found to be associated with two of the three cognitive function measures, SDS and SDL, on controlling for age, sex, race-ethnicity, educational attainment, and the poverty-to-income ratio (an estimate of socioeconomic status). HSV1, CMV, and hepatitis A were the main contributors to the association, that of hepatitis C was very low, and those of HSV2, toxoplasmosis, toxocariasis, and hepatitis B were intermediate.

Possible microbial associations with cognition or with AD in older subjects have recently been investigated in three studies. D’Aiuto et al. [52] used functional MRI to evaluate brain activation during a working memory task, and found an association between “nonencephalitic HSV-1 infection”, assessed by serum IgG, and functional brain changes linked with working memory impairment. Barnes et al. [53], in a longitudinal study, implicated CMV in an increased risk of AD, and stated that HSV1 was not related to AD incidence. However, Itzhaki and Klapper [54] pointed out that Barnes et al used a far less sensitive assay for HSV1 than for CMV, detecting only a single viral glycoprotein for HSV1 whereas for CMV, all of its proteins were detectable. From data obtained in another longitudinal study, Nimgaonkar et al. [55] maintained that CMV, HSV2, or Toxoplasma gondii exposure, but not HSV1 exposure, were associated with cognitive decline in older persons; however, in their discussion they alluded to the lack of sensitivity of their assays for HSV1, in any case adding that not finding an association between exposure to HSV-1 and cognitive decline did not preclude a role for HSV1. Such differences in sensitivity of assays used for detecting antibodies to various viruses should obviously be taken into account when comparing different viruses or estimating infectious burden.

FURTHER ANTIVIRAL STUDIES

Further investigations have been pursued on antiviral treatment of cells in culture during HSV1 infection, following the studies on acyclovir (ACV), penciclovir (PCV), foscarnet [56], and BAY 57–1293 [57], all of which inhibit viral DNA replication. Each of these agents greatly reduced HSV1-induced formation of P-tau and Aβ (and of HSV1, as expected), P-tau dropping to almost zero, but Aβ decreasing to 20–30% of the value without the antiviral. This showed that HSV1 DNA replication is needed for the abnormal phosphorylation of tau, but not for Aβ formation, so the decrease in the latter caused by the antivirals was attributed instead to the antiviral causing a reduction in viral spread, because of reduced viral replication. Another agent, IVIG, also reduced P-tau and Aβ, probably through preventing HSV1 entry into cells, and treatment with a combination of IVIG and ACV was found to be particularly effective [58]. The authors then tried a type of anti-HSV1 antiviral known to prevent HSV1 entry, namely, fucans-sulphated polysaccharides, which are derived from various types of brown algae. The most efficient of these in reducing P-tau and Aβ was an extract from Undaria pinnatifida, and this, when used in combination with ACV (even at a very low ACV dose, only one tenth of that in the ACV-PCV-foscarnet study), lead to a marked synergistic effect [59]. Fucans are much more readily obtainable than is IVIG, so that treating AD patients with the fucan from Undaria together with valacyclovir (the biodrug of ACV, which is far better absorbed in the body than is ACV) would be particularly suitable, as well as relatively inexpensive.

CONCLUSIONS

It is sometimes asserted that HSV1 presence in AD brain–the basis of the viral concept–and the effects of the virus, are a consequence either of the disease itself or of APOE-ɛ4 carriage conferring particular susceptibility to HSV1 infection of the brain. However, the former suggestion is rebutted by the fact that the virus is present in brain of a high proportion of elderly controls as well as AD patients, and the latter by the fact that many elderly controls harbor HSV1 in brain but only a few carry an APOE-ɛ4 allele [2]. Thus, the data strongly indicate that HSV1 is a cause, not an effect, of the disease (nor an effect of having the “wrong” APOE allele). Also, as mentioned above, the APOE-ɛ4-HSV1 association in cold sores (as well as APOE’s influence on microbial diseases [60]) support the concept, as do the data described above and in previous reviews [1, 61], in particular, work on HSV1-APOE interactions [62–64]. And the diversity of the types of study lends further credence to the concept. Whether or not HSV1 acts in combination with another microbe is unknown but should be investigated. And whether HSV1 augments the effect of a non-microbial factor is unknown also, but cannot usefully be discussed, as no other factor has been proposed that is known to be more damaging specifically in those who will develop AD than in those fortunate enough to evade it. Whatever the answers to these questions, a clinical trial treating patients with an antiviral to slow or halt disease progression is now surely warranted.

DISCLOSURE STATEMENT

The author’s disclosure is available online (http://j-alz.com/manuscript-disclosures/16-0607r1).