We analysed the expression of various HERVs families in different subtypes of FTD and AD patients using post-mortem brain tissue. HERVR-env was increased in FTD-MAPT, FTD-GRN, sFTD-tau, sFTD-TDP43, sAD, and AD-PSEN1 compared to controls, presenting all of them a small fold change effect, except for FTD-MAPT. FTD-MAPT showed also elevated levels of HERVK transcripts (-pol, -gag, -env1, -env3), HERVR-env and HERVW-env. Due to the consistency found in the vast majority of HERVs in FTD-MAPT subgroup, we decided to focus in this FTD subtype and go deeper in the understanding of HERVs expression in this subgroup. The ELISA results were in the same line and the HERVK-env protein was elevated in FTD-MAPT post-mortem prefrontal cortex samples, further substantiating our observations.
Several mechanisms could explain the observed increased expression of HERVs in the brain tissue of FTD-MAPT patients. Global control of HERVs can be achieved through either heterochromatin silencing or/and piwi-interacting RNA (piRNA) degradation. These mechanisms have been identified as dysregulated in tauopathies20. One plausible hypothesis might be that the hyperphosphorylated tau, produced by MAPT carriers, and also found in sAD and sFTD-tau, induces a decondensation of heterochromatin via global relaxation (Fig. 3). This relaxed state could facilitate HERVs transcription, while tau may concurrently reduce the efficacy of the piRNA mechanism, thereby enhancing HERVs viability21. HERVs can be controlled by CpG methylation3. Thus, if HERVs are upregulated because of a combination of chromatin relaxation and increased CpG methylation caused by pathological tau, a deeper understanding of pathological tau subtypes might be crucial to comprehend why we do not detect an upregulation of the HERVs used in this study in sAD and sFTD-tau.
HERVK activation can be inhibited using anti-retroviral reverse transcriptase inhibitors in cultured cells12,22. This approach has been recently implemented in clinical trials. Anti-retroviral drugs targeting neurodegenerative disorders are under investigation in a Phase 2 clinical trial for amyotrophic lateral sclerosis (NCT02868580)23 and recent initiation of Phase 1 (NCT04500847) and Phase 2 trials for AD (NCT04552795). However, our results suggest that the impact of HERV inhibition might vary depending on the underlying proteinopathy that might rely on tau depending on the isoform. Genetic forms of FTD are excellent candidates for targeted therapies because they can be detected in the presymptomatic phase of the disease, allowing for a prompt intervention when structural brain damage and other biological changes are in their early stages. Mutations in MAPT position 301 are the most common causes of FTD-MAPT24. P301L and P301S mutations in exon 10 are the cause of approximately 25% of disrupted tau proteins, leading to an overproduction of 4R tau repeat isoforms, which bind more tightly to microtubules than 3R. Although related, abnormal tau proteins are not identical among tauopathies, and they do not behave the same way, which is interesting when considering clinical trials that target this protein. Different possibilities could be that pathological tau proteins from genetic and sporadic patients behave differently. Thus, tau can present with distinct aggregation pattern, it could be a change in 3R to 4R tau repeat isoforms, or mutations that promote tau assembly into filaments25–27. Interestingly, in a study by Ramirez et al., they found that only tau-transgenic mice expressed higher levels of ERVs in contrast with APP-carrying mice, where there was not an ERV upregulation28. In post-mortem brain tissue from tauopathies caused by mutations in MAPT gene, several HERVs are increased. Furthermore, some HERVs were also elevated in sFTD-tau, supporting a common pathological mechanism between altered tau. In fact, when tau is mutated, seeding, phosphorylation and folding status vary29. In AD, the six isoforms of tau co-aggregate, while on FTD-MAPT it is not clear if 3R tau can propagate. Tau hyperphosphorylation in AD leads to tau misfolding ending up in insoluble aggregates, but in FTD-MAPT, hyperphosphorylation of tau is not needed prior to folding, thus in FTD-MAPT patients, mutant tau misfolding precedes hyperphosphorylation29. These differences in tau properties might partially explain the diversity observed upon HERVs activation. Ideally, by understanding the underlying molecular mechanism behind different tauopathies, we could develop treatments that target specific tau subtypes.
The data assessment from microarray methylome profiling led to the identification of HERV-related genes (NUPR1, PGBD5, CBX1, CBX3 and ARC) exhibiting hypo/hypermethylation specifically in FTD-MAPT patients when compared to controls, which were not found in the other FTD subgroups. We further analysed genome-wide DNA methylation data and we detected hypermethylated CpGs in common in all FTD subgroups in ERVMER34-1, PIWIL3, and CHD4 genes. NUPR1 gene expression is upregulated in brain tumours and overexpressing HERVK-env cells30,31. NUPR1 acts as an essential component during stressed cells and has been of interest as a target gene in different cancers and in HERVK-env transfected human colorectal cancer cells31.Interestingly, this gene was found as hypomethylated in FTD-MAPT patients, alluding for a possible increase in its protein expression. Another gene of interest was PGBD5, the most evolutionarily conserved transposable element in vertebrates, found as hypermethylated in FTD-MAPT patients, therefore a reduction of its expression could be hypothesised, mainly expressed in the brain32. However, no relationship has been established between HERVs and PGBD5 yet. On a different note, CBX1 and CBX3, heterochromatin protein-1 family members (HP1s), are found to be hypermethylated. A reduced expression of their encoded proteins might be expected, and subsequently a dysregulation at an epigenetic level, which then would lead to a de-repression of HERVs and subsequently increases in FTD-MAPT patients compared to controls. Both proteins are involved in gene silencing via epigenetic control, and their loss can lead to impaired neuromuscular and cerebral cortex development33. Notably, the CBX1 isoform is required to repress immune genes and HERVs34. Moreover, if less CBX1 is present, immune signals might increase, promoting hyperstimulation of the immune system, which is one of the main characteristics of neurodegenerative diseases. In fact, Galectine-3, a protein that promotes microglial activation in the central nervous system, is increased in both cerebrospinal fluid (CSF) and post-mortem brain tissue from FTD-MAPT patients (Borrego-Ecija, accepted). Finally, we identified the neuronal gene ARC, with a retroviral origin, whose protein localises in activated synaptic sites in an NMDA receptor-dependent manner and has a critical role in learning and memory molecular processes35. It has been found that HERVW-env protein can alter NMDA trafficking, disrupting physiological homeostasis and thus, leading to neuronal deficits36. Therefore, we can hypothesize that a downregulation of ARC might be a consequence rather than a cause of HERV env release.
In contrast to brain tissue, we observed decreased expression of several HERVs in whole blood RNA in FTD-MAPT group. Different hypothesis could explain this finding. It might be that the immune system in blood contributes to HERVs silencing, such as autoantibodies against HERVs that cannot cross the brain blood barrier. Another possible explanation could be the methylation state, which is cell-type and age dependent37. Moreover, post-mortem brain tissue is typically obtained in the late stages of the disease, whereas whole blood samples are obtained at earlier disease stages. Therefore, we could speculate that HERVs activation might be a specific event found in prefrontal brain tissue in FTD-MAPT. Arru et al. examined HERVs levels in different neurodegenerative diseases and only amyotrophic lateral sclerosis presented increased HERVs levels in blood and CSF38. As mentioned previously, HERVs might express differentially depending on the disease, disease stage, cell type and age, which might be due to DNA methylation-dependent mechanisms. More studies should be performed to fully understand HERV expression in blood among different FTD and AD subtypes.
In our study, we did not find a significant relevant increase in sAD when compared to control subjects. In contrast, Sun et al., identified HERVK dysregulation in sAD and progressive supranuclear palsy because of pathogenic tau, which in turn caused neuronal death20. They found HERVK family as the most over-represented using RNA-seq on cerebellum and frontal cortex from AD. One possible explanation about these differences might be due to the HERV sequence design, as they analysed distinct HERVK subtypes such as HERK22I and HERVKC4, whereas in our study we used HERVK113 and HERVK118. Better computational tools to target HERVs might help to understand the variations.