Peptide Sharing Between Viruses and DLX Proteins: A Potential Cross-Reactivity Pathway to Neuropsychiatric Disorders

The present study seeks to determine potential associations between viral infections and neuropsychiatric diseases. To address this issue, we investigated the peptide commonalities between viruses that have been related to psychiatric and neurological disorders—such as rubella, human immunodeficiency virus, and herpesviruses—and human distal-less homeobox (DLX) proteins expressed in developing brain—namely, DLX1, DLX2, DLX5, and DLX6. Peptide matching analyses revealed a high degree of pentapeptide sharing. From an immunological perspective, this overlap is relevant because pentapeptides are endowed with immunogenicity and antigenicity—that is, they are immune determinants. Moreover, infection-induced immune cross-reactions might have functional, spatial, and temporal implications related to the functions and expression patterns of DLX1 and DLX5 in the fetal and adult human brain. In sum, our data support the hypothesis that viral infections may be linked to neuropsychiatric diseases through autoimmune cross-reactions caused by molecular mimicry between viral proteins and brain-specific DLX self-antigens.


INTRODUCTION
Infections, neuropsychiatric diseases, and language disorders are often concomitant pathological events that can have early etiological roots in fetal life and then become apparent in any stage across the life span of the individual, from the postnatal period to the late age (Yolken and Torrey, 1995;Coplan et al., 1998;Arias et al., 2012;Brown, 2012Brown, , 2015Khandaker et al., 2013). However, it is unclear how these pathological events are mechanistically interlinked and temporally related, most likely because the wide spectrum of infectious agents and the varying extent of the numerous neuropsychiatric symptoms make it difficult to dissect the molecular correlations between infections and brain damage (Ludlow et al., 2016).
During the past few decades, scientific-clinical research examined the assumption that infections may relate to neuropsychiatric disturbances through autoimmune mechanisms (Knuesel et al., 2014;Severance et al., 2014;Blomström et al., 2015;Estes and McAllister, 2016;de Haan et al., 2017). More recently, it was suggested that anti-pathogen immune responses cross-reacting with the human NMDA receptor 2A subunit-alterations of which are involved in language dysfunctions (Carvill et al., 2013;Turner et al., 2015)-might represent a pathologic background for infections and many neurodegenerative disorders, ranging from schizophrenia to frontotemporal dementia (Lucchese, 2016).
The current study is extended to four members of the DLX transcription factor (TF) family-namely DLX1, DLX2, DLX5, and DLX6-that have been thoroughly investigated in numerous studies on neurodevelopment. Indeed, the four TFs are expressed during early fetal neurodevelopment (Merlo et al., 2000;Panganiban and Rubenstein, 2002) and are associated with the specification of γ-aminobutyric acid (GABA)ergic interneurons in the vertebrate forebrain subventricular zone (SVZ) as well as with granule neurons in the subgranular zone (SGZ) (Simeone et al., 1994;Anderson et al., 1997a,b). The issue appears to be important especially when considering that cognitive and emotional tasks occur in the neurogenic areas (Aimone et al., 2011;Ming and Song, 2011;Miller and Hen, 2015), and that altered adult neurogenesis and hyppocampal lesions have been repeatedly related to neuropsychiatric conditions (Parent and Murphy, 2008;Gonçalves et al., 2016;Inta et al., 2016;Kang et al., 2016;Yun et al., 2016;Kohman and Rhodes, 2017) and language disturbances (Sass et al., 1992;MacKay et al., 1998;Covington and Duff, 2016;Piai et al., 2016).
In this context, we hypothesized that an infection-induced maternal immune response may cross-react with DLX proteins, thus possibly causing a first subclinical immune-mediated damage of the developing nervous system. Later, successive encounters in adulthood with pathogens able to induce crossreactions with DLX proteins might further damage regions of the adult brain (the subventricular zone and dentate gyrus), where DLX proteins may be expressed (Lim and Alvarez-Buylla, 2016) thus triggering the onset of neuropsychiatric clinical manifestations.
Focusing on infections as a trigger of DLX alterations and seeking for a possible mechanism, we proceeded along three steps. Firstly, we investigated whether or not infectious pathogens have the molecular basis to react with human DLX proteinsthat is, we searched for shared peptides that might lead to crossreactions. Secondly, we analyzed the immunological potential of the viral vs. DLX peptide overlap. Thirdly, we collected data on the expression level of DLX proteins in the fetal and adult human brain.

RESULTS AND DISCUSSION
A sample library formed by 25 virus proteomes was analyzed for pentapeptide sharing with DLX proteins. Pentapeptides were used as probes, for five aa residues represent a minimal immune-biological determinant in humoral and cellular immune recognition (Kanduc, 2012, 2013 andfurther refs. therein). NREP, a protein involved in neuronal regeneration (Fujitani et al., 2004), was used as a neural control protein.

Pentapeptide Sharing Between DLX Proteins and Potential Viral Pathogens
The quantitative and qualitative pentapeptide overlap between the four human TF DLX proteins and NREP vs. the set of 25 virus proteomes is shown in Table 1.
At a first glance, the following points become apparent when considering Table 1: • the neural proteins, including the neuronal regenerationrelated protein NREP, have pentapeptides in common with all viral proteomes, excluding Borna disease virus and Influenza B virus; • DLX2 is the main target of the peptide sharing by being 49% the level of DLX2 pentapeptide similarity to the 25 proteomes, i.e., 159 out of 324 DLX2 pentapeptides are shared with the viral proteomic ensemble; • the peptide sharing mostly occurs with herpesviruses in general, and with HCMV in particular. Instead, the peptide sharing with HeV, HBV-C, RV-C, and RV ADRV-N is restricted to a few pentapeptides thus indicating a scarce contribution of such infectious agents in crossreactivitytriggered DLX alterations and consequent neurological manifestations; • the viral pentapeptide distribution is not stochastic. For example, Vaccinia virus pentapeptides represent 27% of NREP peptide sharing (i.e., 3 out of 13 pentapeptides) and 4.4% of the DLX2 peptide sharing (i.e., 7 out of 159 pentapeptides); • the extent of the peptide sharing is independent of the virus protein length. For example, Rubella virus (3,179 aa) and the eleven times longer HHV3 (35,782 aa) share exactly the same number of pentapeptides-namely, three-with DLX1 protein; • quantitatively, the extent of the peptide sharing is unexpectedly high compared to the mathematical expected value of the pentapeptide sharing between the five neural proteins and the 25 viral proteomes. The expected value can be calculated as follows: given two protein entities (for example, DLX1, and the set of 25 viral proteomes) and assuming that all aa occur with the same frequency, the expected probability for the two entities to share a pentapeptide is expressed by the formula mn/N 2 where: m is the number of pentapeptides present in the DLX1 protein (i.e., 251); n is the number of pentapeptides present in the set formed by the 25 viral proteomes (i.e., 418,854), and N is the number of all possible pentapeptides. Since each residue can be any of 20 aa, then N is 20 5 (i.e., 3,200,000).
Developing the equation, the expected pentapeptide sharing between DLX1 and the 25 viral proteomes is equal to 1,02668314453125e-5 or, practically, zero.
The Immunologic Potential of the Viral vs.

DLX Proteins Peptide Overlap
The pentapeptide matching between viral and neural proteins illustrated in Table 1 has an experimentally documented immunologic potential. As a matter of fact, exploration of the Immune Epitope DataBase (IEDB; www.immuneepitope.org; Vita et al., 2015) shows that almost all shared pentapeptides are also part of hundreds of epitopes that have been experimentally validated as immunopositive in humans. Using the epitope aa length as a filter, Table 2 is restricted to n-mer sequences with n < 12.

Comparative DLX Transcript Expression in Fetal and Adult Human Brain
A comparative analysis of DLX expression in fetal and adult neurogenic areas of the human brain was conducted using the online database and resources of the Allen Institute for Brain Science (Lein et al., 2007;Miller et al., 2014). Figure 1 reports laser microdissection microarray analyses showing that the transcript expression of the four TFs ranges from medium to high in the fetal transient structures of forebrain (ventricular zone and ganglionic eminence) (Figure 1A), and reaches the lowest but still detectable value in the adult neurogenic dentate gyrus ( Figure 1B, subareas CA1, CA2, CA3, and CA4). Notably, only DLX5 and DLX6 appear to be expressed in basal ganglia of adult brain, that is, substantia innominata, caudate nucleus, nucleus accumbens, and putamen ( Figure 1B). The control neuronal regeneration-related protein NREP is widely expressed through almost all fetal and adult brain areas, except the fetal ventricular zone and ganglionic eminence ( Figure 1A) as well as the dentate gyrus area in the adult brain ( Figure 1B).

DLX Protein Expression in Fetal and Adult Human Brain
On the whole, Figure 1 theoretically supports the possibility that the cross-reactivity scenario outlined in Tables 1, 2 may occur in neurogenic areas in the fetal life of an individual and then possibly recur in adulthood. However, from an immunological point of view, a condition that is necessary for the cross-reactivity hypothesis to be biologically plausible depends on sufficient DLX antigen expression in the brain. In other words, data of Figure 1 need to be substantiated in a protein context. Actually, few data are available on DLX protein expression in humans. Rakic and Zecevic (2003) studied DLX expression in the late human embryonic period (Carnegie stages 19-20) and showed that DLX2 protein was widely distributed through the ganglionic eminence and dorsal telencephalon. Moreover, immunocytochemistry based on a pan-DLX antibody that recognizes DLX 1, 2, 5, and 6 revealed that, in the developing brain, 11 gestational week, DLX protein expression is present in all cortical layers, including layer I and the subpial granular    layer (SGL). Almost all small GABAergic cells of the SGL were labeled with the pan-DLX antibody. Successively, using the same pan-DLX antibody, Jakovcevski et al. (2011) showed labeling of the neocortical VZ/SVZ and of the cortical plate in human fetal forebrains during the first half of gestation. Such experimental results obtained in human fetal developing brain are flanked by data collected from the Human Protein Atlas (https://www.proteinatlas.org/on DLX1 and DLX5 protein expression in the adult human brain. The data are shown in Figure 2. It can be seen that human DLX1 and DLX5 have a protein expression from low to medium level, that is, sufficient to sustain immune cross-reactions. No protein expression data were available for DLX2 and DLX6 proteins. The control NREP had the highest levels of protein expression (from medium to high in the cerebral cortex and the cerebellum).
In essence, we found a vast and unexpected peptide sharing between DLX proteins and numerous infectious agents that constellate human life, from prenatal time to adulthood. The peptide platform defined in Table 1 has also a high immunologic potential, as documented in Table 2, so that, on the whole, data from Tables 1, 2 show the existence of a wide immunologic peptide platform common to viral and human DLX proteins. Moreover, data on protein expression from literature (Rakic and Zecevic, 2003;Jakovcevski et al., 2011) and Figure 2 (www. proteinatlas.org; Uhlén et al., 2015;Thul et al., 2017), although few in numbers and fragmentary, support the possibility that mild, subclinical anti-DLX autoimmune damage in the fetal brain structures evoked by maternal viral infections (and consequent maternal immune activation) may be followed in the adult brain by additional damage after a second encounter with the same pathogen or novel infection with a different agent sharing the same epitopic sequences.
In this context, different immunological pathogenic mechanisms might be theoretically account for the neuronal damage according to the type of immune response, i.e., humoral vs. cell-mediated, and the timing of infection-induce maternal immune activation in relationship to the expression patterns of DLX proteins in the fetus (see Figure 1). The main hypothesis that we considered here relates to maternal infection and consequent immune activation that may also be antecedent to pregnancy and are followed by antibody-mediated neuronal damage in the fetus due to cross-reactions with DLX proteins. In such a scenario, the passage in the fetus of maternal memory B cells (Vernochet et al., 2005(Vernochet et al., , 2007 against epitopes shared between the pathogen(s) and DLX proteins might induce an immune response targeting the developing nervous systems, where DLX proteins are expressed early (see Figure 1). Cellular damage from (auto)antibodies targeting intracellular antigens, like the DLX family of TFs may be, not only plays a pathogenic role in a variety of autoimmune diseases (Racanelli et al., 2011) but even represents a promising therapeutic strategy for cancer treatment (Weisbart et al., 2012;Wang et al., 2015;Chan et al., 2016). The notion that autoantibodies can penetrate living cells is not new. Alarcon-Segovia et al. (1978) showed that antibodies can penetrate living cells. In more recent years, more evidence has accumulated showing autoantibody penetration into different types cell, including neurons, and proposing  FIGURE 1 | Comparative analyses of DLX1, DLX2, DLX5, and DLX6 transcript expression in fetal (15 post-conception weeks, pcw; A) and adult (24 years; B) human brain. The symbol localizes neurogenic areas of developing and adult brain. The figure assembles images and data from the Allen Institute. Further details on donors, DNA probes, complete transcriptome profiles, and methodology can be found at http://www.brainspan.org/ and http://human.brain-map.org/ (Lein et al., 2007;Miller et al., 2014).
mechanisms that may explain a pathogenic role of internalized immunoglobulins in autoimmune diseases (Koren et al., 1995;Koscec et al., 1997;Portales-Pérez et al., 1998;Ruíz-Argüelles et al., 1998Ronda et al., 2002;Proulx et al., 2009). Moreover, a nuclear-penetrating lupus anti-DNA autoantibody, 3E10, has been found to inhibit DNA repair and selectively kill certain cancer cells that are highly vulnerable to DNA damage (Weisbart et al., 2012), and, of special importance, nuclear-penetrating anti-dsDNA autoantibodies could possibly function as a pathogenic autoimmune factor for lupus nephritis (Im et al., 2015). Bearing even more relevance to the present discussion, antibodies targeting intracellular antigens, like for instance the glutamic acid decarboxylase, appear to be also involved specifically in neuropsychiatric disorders, like CNS lupus (Karimifar et al., 2013), limbic encephalitis (Matà et al., 2008), schizophrenia (Najjar et al., 2012), and autism (Rout et al., 2012). Indeed, the glutamic acid decarboxylase isoforms (Gad1 and Gad2), which regulate GABA synthesis from the excitatory neurotransmitter glutamate and whose expression is activated by DLX1 and/or DLX2 (Stühmer et al., 2002a,b;Le et al., 2017), FIGURE 2 | DLX1, DLX5, and NREP protein expression in adult human brain. Estimate of the protein expression are: not detected (n), low (l), medium (m), or high (h). Data for DLX2 and DLX6 proteins were not available or pending. The figure assembles images and data from www.proteinatlas.org (Uhlén et al., 2015;Thul et al., 2017).
share numerous pentapeptides with the 25 viral proteomes (see Supplementary Table 1). Hence, a scenario emerges where immune responses following infections might cause a cascade of multiple crossreactions at multiple levels (i.e., DLX, GAD) of the intracellular mechanisms regulating the function of GABAergic neurons and altering the excitation and inhibition ratio, which is necessary for normal neural circuit function and whose imbalance contributes to neurodevelopmental diseases (Kang, 2017;Maffei et al., 2017;Ye and Kaszuba, 2017;Catavero et al., 2018;Garret et al., 2018). On the other hand, a cell-mediated mechanism could also theoretically be implied in the cross-reactive immune-mediated subclinical damage of the fetal nervous systems, since memory T-cell trafficking between mother and fetus is also a wellknown phenomenon (Jeanty et al., 2014). Nevertheless, the hypothesis of a cell-mediated response would need to take into account the late MHC expression in the fetal CNS (Elmer and McAllister, 2012;Zhang et al., 2013Zhang et al., , 2015McAllister, 2014) that might not sit well with the very early pattern of expression of the DLX-proteins in the fetus seen in Figure 1. However a later cell-mediated damage, and even the possible occurrence of both humoral and cell-mediated responses at different stages of the fetal neural development, can still be hypothesized.
Based on data from Figure 2 and, consequently, confining our discussion to DLX1 and DLX5 proteins, we observe that infection-induced immune cross-reactions might have functional, spatial, and temporal implications: • Functionally, infection-induced immune cross-reactions would affect two TFs that, according to numerous studies, are implicated in crucial functions and fundamental processes during neurodevelopment and adult neurogenesis, and are potentially relevant to language competence and other higher cognitive functions (see Box 1); • Spatially, infection-induced immune cross-reactions would damage brain structures where adult neurogenesis occurs and that are involved in the neural circuitry of language and memory, and in cognitive and emotional functions (Ming and Song, 2011). Altered human neurogenesis is linked to neuropsychiatric conditions and to impaired cognition (Aimone et al., 2014). Also, alterations of the SVZ and hippocampus have been specifically related to some of the pathogenic and symptomatic aspects of schizophrenia (Reif et al., 2006(Reif et al., , 2007Duan et al., 2007;Toro and Deakin, 2007;Christian et al., 2010;Aimone et al., 2014;Allen et al., 2016;Kang et al., 2016;Yun et al., 2016;Iannitelli et al., 2017;Terrillion et al., 2017).
• Temporally, infection-induced immune cross-reactions suggest a two-hit model that, depending on the DLX protein expression profiles, comprehends targets allocated in two time-windows in the life of an individual with a subclinical damage in fetal life and clinical onset in adulthood.

CONCLUSION
In synthesis, the present study confirms previous reports (Kanduc et al., 2008;Lucchese et al., 2014;Lucchese, 2017) and supports the hypothesis that an autoimmune connection exists at the molecular level among infections, autoimmune reactions and neuropsychiatric disorders. Such a connection implies a vast viral vs. neural proteins peptide overlap and operates through cross-reactivity mechanisms. To test this hypothesis in vivo, the possibility of obtaining animal models BOX 1 | DLX1 and DLX5 functions and relevance to neuropsychiatric disturbances.
• regulates the fate switch between cortical and striatal interneurons: cells that ordinarily would become cortical interneurons transform toward a subtype of GABAergic striatal interneurons, thus reducing glutamatergic input to the hippocampus (McKinsey et al., 2013).
• its loss leads to subtype-specific loss of inhibitory interneurons with a reduction in inhibitory currents and generalized seizures in mice (Cobos et al., 2005;Jones et al., 2011).
• contributes to promote cortical interneuron migration from the basal forebrain by direct repression of the semaphorin receptor neuropilin-2 (Le et al., 2017).
• regulates development of peripheral and central components of the olfactory system (Long et al., 2003).
• participates to the regulation of the expression of the glutamic acid decarboxylases (Stühmer et al., 2002a).
of neuropsychiatric disorders by immunizing pregnant animals with DLX proteins could be examined. Moreover, analyses of sera from human patients with neuropsychiatric diseases, for example schizophrenia, are warranted to measure immunoreactivity against the peptides shared between viruses and DLX proteins. Possibly, the results of these joint basic and clinical in vivo approaches might also help design new therapeutic approaches in neuropsychiatry.

AUTHOR CONTRIBUTIONS
All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

ACKNOWLEDGMENTS
We acknowledge support by the German Research Foundation and the OpenAccess Publication Funds of the Freie Universität Berlin. We thank the two reviewers for the constructive criticism that contributed to increase the clarity and quality of this article.