Retinoic acid-mediated mechanism in neural tube formation and further brain development

Retinoic acid is a non-peptide small lipophilic molecule derived from retinol—an active ingredient of vitamin A. Retinol gets converted to retinal and further into retinoic acid by the action of dehydrogenases which includes CYP1B1. Retinoic acid is taken to the nucleus by the cellular retinoic acid binding protein (CRABP) where it binds to retinoic acid receptors (RAR and RXR) on the promoter regions of specific genes (Balmer and Blomhoff, 2002; Rhinn and Dollé, 2012). RAR and RXR are important transcription factors (Rhinn and Dollé, 2012) and actively influence RARE consensus sequences in promoter regions of a plethora of genes involved in neural tube development in addition to serving other pertinent embryological functions (Balmer and Blomhoff, 2002; Rhinn and Dollé, 2012). These genes further activate a cascade of regulatory molecules involved in neural tube formation and brain development. Any dysregulation of this intricate molecular process at any step can lead to various degrees of neural tube defects and brain malformations (Grapin-Botton et al., 1998; Kam et al., 2012; Rhinn and Dollé, 2012). The RARE sequence is also spread all over human genome through commonly found ALU repeats which contain RARE sequence (Vansant and Reynolds, 1995). Retinoic acid enjoys the stature of being the initial-most molecular factor involved in the determination of the neural axis and, is the prime determinant of the anteroposterior axis owing to its interplay with Nodal gene in developing neural tube (Durston et al., 1989; Kam et al., 2012). Together with fibroblast growth factor (FGF), it is known to set a code for establishing the anterio-posterior axis (instituted by opposing gradients of retinoic acid and FGF along the neural tube) which is followed by other molecular factors (Martínez-Morales et al., 2011; Shimozono et al., 2013).

Nodal, which is thought to be the master regulator gene for deciding the axes of the developing neural tube has RARE consensus sequence in its intron-1. Retinoic acid is also involved in deciding dorsoventral axis by intricate crosstalk with Nodal (Kam et al., 2012). A plethora of literature suggests that retinoic acid is involved in FGF8 repression during body axis extension and regulation of the homeobox (HOX) genes involved in determining the position of the neural segments (Grapin-Botton et al., 1998). More than deciding the axes and positioning of the neural segments at the time of neural tube formation, retinoic acid (at a later stage) also has imperative role in the development of midline, posterio-dorsal brain structures, and forebrain components in addition to corticogenesis. Disruptions in the retinoic acid signaling pathway have been implicated in ensuing brain defects in developing fetus involving all the above mentioned structures and may, arguably, present as microcephaly (Wilson et al., 2007; Siegenthaler et al., 2009; Crandall et al., 2011; Gupta and Sen, 2015). Retinoic acid is also a teratogen and can cause the neural tube defect and other brain malformations if provided in abnormal doses to developing fetus (Yamamoto et al., 2003).

Plausible mechanism of ZIKV mediated retinoic acid signaling dysregulation resulting in fetal microcephaly

ZIKV is a positive sense RNA Flavivirus (Baronti et al., 2014). When an RNA virus infects a host cell, its genetic material is reverse-transcribed into cDNA by reverse transcriptase which, in turn, synthesizes complementary sense DNA strand, and finally, the so composed DNA gets integrated into the host DNA. We proceed with an important and coherent line of arguments that in acute viremia heavy doses of viral DNA (reverse-transcribed from RNA) will be integrating into the host DNA, and if the inserted DNA sequences match the regulatory regions of certain host genes then they may influence their expressions. By the transitive property of similarities, it can then be justified that, the RARE consensus sequence repeats (if present) in the genomic sequences of ZIKV strains would be inserted into the promoter regions of RARE dependent genes of the host DNA and, therefore, may influence their expression in a way that the developing fetus manifests with brain malformation like microcephaly (Figure 1).

Current understanding of the ZIKV-microcephaly association is improving as more and more relevant reports are coming up and many recent studies strongly agree to the notion that ZIKV-microcephaly may be a developmental brain malformation rather than a mere brain destruction by the virus (Hazin et al., 2016; Mlakar et al., 2016). So, in the light of comprehensive similarities between fetal brain defects in ZIKV infection and brain malformations caused by retinoic acid dysregulation in developing fetus, we considered it a plausible rationale to search for a possible retinoic acid-mediated mechanism involved in ZIKV-microcephaly association. We, for this important reason, searched for the RARE consensus sequence (5′–AGGTCA–3′) repeats in the genomic sequences of the ZIKV strains with a hypothesis that the virus might act through disruption of normal retinoic acid signaling mediated by incorporation of these sequences into the DNA of developing host brain cells. To validate this hypothesis robustly and to establish the cogency of our study, we also screened some other viruses (including members of the flaviviridae family and those viruses which are known to have maternal-fetal/perinatal transmission either causing congenital brain malformations or showing neurotropism) and subjected them to the same analysis.

Materials and methods

The genome sequence information of 17 strains of ZIKV (including “Brazil ZKIV-2015” which was recently implicated in causing microcephaly in Brazilian fetuses; Calvet et al., 2016; Mlakar et al., 2016), and many other viruses were retrieved from the NCBI/GenBank database (http://www.ncbi.nlm.nih.gov/genbank/) (Table 1) and screened for the presence of RARE consensus sequence (5′–AGGTCA–3′). Among the viruses other than ZIKV, selection was made under four groups viz: (1) Members of the flaviviridae family other than ZIKV (to check if RARE consensus sequence repeats were essentially present in the family) (2) Viruses showing maternal-fetal/perinatal transmission and causing congenital brain malformations (3) Viruses showing neurotropism but not known to cause fetal brain defects (4) Viruses chosen randomly from the RNA/DNA viruses not fitting in any of the above three groups (to account for the selection bias and to establish coherence and validity of the screening process). To identify the enrichment of RARE sequence (5′–AGGTCA–3′) repeats present in the viruses, a Perl script was written and executed.

In order to find and count the RARE sequences from the ZIKA virus genome, a PERL script was generated entitled “RARE_Seq_finder.pl.” The PERL script was based on the window shift algorithm. Genome sequence information of ZIKA virus isolates was downloaded from various genome databases. As input, the PERL Script requires only the list file carrying names of the genome sequence files. It searches the RARE sequences globally in the genome, and lists all genome file names along with number of RARE sequences present in the genome. A complete flow-chart with all the steps involved in the counting the RARE sequences is shown in Figure 2. The complete PERL script for this study is available as Supplementary Material.

Result

The ZIKV strains associated with microcephaly contained RARE consensus sequence repeats. The number of repeats corresponds to the virulence of each strain i.e., greater the number of RARE repeats higher the chances of microcephaly. Complete details of the number of RARE sequence repeats in each ZIKV strain are given in Table 2. For viruses other than ZIKV, the RARE consensus sequence repeats were found to be present only in those viruses known for maternal-fetal/perinatal transmission, involved in fetal brain defects and displaying neurotropism. The numbers of RARE sequence repeats was directly proportional to the virulence of screened viruses. The number of RARE consensus sequences repeats in the checked 17 strains of the ZIKV and other ZIKV viruses are provided in Tables 2, 3 respectively.

Fetal brain defects in ZIKV-microcephaly closely resemble to brain malformation caused by dysregulation of retinoic acid signaling

Fetal microcephaly has always been known as a developmental anomaly caused by multiple etiologies (Mochida and Walsh, 2001). Retinoic acid signaling dysregulation has also been implicated as a common mechanism in many environmental or developmental reasons of microcephaly (Lammer et al., 1985; Kot-Leibovich and Fainsod, 2009; Paganelli et al., 2010). The anatomical details of the brain of the fetuses suspected of ZIKV infection in recent reports (Aragao et al., 2016; Calvet et al., 2016; Hazin et al., 2016; Mlakar et al., 2016) closely resemble to those caused by retinoic acid signaling dysregulation and have many features in common i.e., cerebellar and brain stem hypoplasia (Yamamoto et al., 2003), ventriculomegaly (Micucci et al., 2014), hypoplasia of forebrain derivatives like frontal or parietal lobe of the brain (Halilagic et al., 2007), and hypogyration of cerebral cortex (Siegenthaler et al., 2009). The reason behind these striking resemblances could be that the retinoic acid signaling is crucially involved in development of the dorsal brain structures (Wilson et al., 2007; including cerebellum and brain stem), midline forebrain structures (Gupta and Sen, 2015; including choroid plexus which synthesizes CSF and a malformation of it may result in hydrocephalus and in turn ventriculomegaly), and the cortex (Siegenthaler et al., 2009; Crandall et al., 2011). So it is an obviously plausible concept in the light of our hypothesis as to why retinoic acid signaling dysregulation may result in malformations similar to that noted in ZIKV caused microcephaly (Calvet et al., 2016; Hazin et al., 2016; Mlakar et al., 2016).

The presentation of the brain defects in retinoic acid dysregulation varies depending on the stage of embryonic development. If the signaling disruption ensues at late gastrulation or early neurulation stage, the defect should be occurring in rostrocaudal sequencing of the brain structures in the developing neural tube i.e., there may be a larger hindbrain at the expense of anterior brain structures. But if the disruption is at late developmental stages that will affect dorsalization of the brain or hindbrain structures i.e., may cause cerebellar or brain stem hypoplasia or absence of some hind brain components (Wilson et al., 2007).

The presence of microcephaly or other brain malformations in the fetuses born of the ZIKV infected mothers (similar to the other viruses causing congenital brain malformations) may also depend upon the time-point of the fetal infection during neural tube formation, and also on the total viral load (Adams Waldorf and McAdams, 2013). In some recent reports of the cases with ZIKV-microcephaly, pregnant women presented with clinical features of ZIKV infection at 10th week of gestation or beyond and microcephalic features were detected on ultrasound beyond 20th week (Calvet et al., 2016; Mlakar et al., 2016). This matches with the retinoic acid signaling dysregulation at the late developmental stages when dorsal brain structures and hindbrain brain structures are still in development and cortical gyration still in progress. A disruption at this stage would essentially result in the features typically seen in ZIKV microcephaly (Calvet et al., 2016; Hazin et al., 2016; Mlakar et al., 2016) as has been noted above.

Due to striking resemblance to the embryological malformations, the recent reports on ZIKV-microcephaly (Calvet et al., 2016; Hazin et al., 2016; Mlakar et al., 2016) have stressed upon considering the fetal brain damage as a developmental problem rather than the brain destruction caused by viral invasion of neuronal cells. One important study has found substantial evidence of presence of ZIKV in the neurons through immuno-histological and microbiology investigations of the fetal brain tissue died of the severe microcephaly (Mlakar et al., 2016). Interstingly, other probable infectious causes of the microcephaly were successfully ruled out in this study.

Justification from maternal-fetal transmission

On an interesting note, our data also suggests that a virus which contains RARE consensus sequence repeats also shows some degree of neurotropism but its causal handcuff to microcephaly or any other type of fetal brain defect essentially depends upon its ability for maternal-fetal transmission. Chikungunya virus ((+)ssRNA, linear, 11811 bp: Table 3) seems to be a representative example. It contains four RARE consensus sequence repeats and shows neurotropism (Koyuncu et al., 2013), maternal-fetal transmission in humans has also been sufficiently demonstrated for this virus (Fritel et al., 2010; Gérardin et al., 2008), and is also known to cause microcephaly and/or other fetal brain defects (Gérardin et al., 2014). On the other hand Japanese encephalitis virus ((+)ssRNA, linear, 10970 bp: Table 3) which has no known evidence of maternal-fetal transmission in humans, although having shorter frame length and more number of RARE consensus sequence repeats (numbering to six) is known to show neurotropism (Kimura-Kuroda et al., 1993) but do not cause fetal brain defect in humans. The obligatory nature of ability of maternal-fetal transmission in a virus to cause fetal brain damage is empirically coherent and further suggests that presence of RARE consensus sequence repeats in a virus genome may be limited to confer it neurovirulence only, and may require additional contributing factors toward its maternal-fetal transmission ability.

Further notes in support of our hypothesis concerning ZIKV-mediated microcephaly

Another evidence in favor of the retinoic acid signaling dysregulation as a probable mechanism in ZIKV- microcephaly is that the mouse model study of a well-known fetal anomaly syndrome called “CHARGE” (in which dysregulation of retinoic acid signaling has been implicated) reported some anomalous features very similar to that found in fetal brains suspected of ZIKV microcephaly (Calvet et al., 2016; Hazin et al., 2016; Mlakar et al., 2016) such as cerebellar hypoplasia and ventriculomegaly (Micucci et al., 2014). In addition to that, hydrocephalus and malformation of corpus callosum noted in some cases of ZIKV microcephaly (Aragao et al., 2016) are also present in “CHARGE” syndrome. Furthermore, the widespread cortical anomalies in ZIKV microcephaly get explained by an interesting study involving retinoic acid (Siegenthaler et al., 2009). These investigators reported cranial meninges as a source of retinoic acid to the developing cortex (radial glial cells carrying migrating progenitor cells to the upper cortical layers attach their end feet to the covering meninx, from where they get retinoic acid as a diffusible morphogen and transfer to other cells involved in corticogenesis). Retinoic acid signaling has been extensively implicated in cortical genesis (Harrison-Uy et al., 2013), especially in the migration of progenitor cells to the upper cortical layers riding on the radial glial cells (Siegenthaler et al., 2009; Crandall et al., 2011). Many contemporary reports have confirmed that ZIKV has detrimental effects on human cortical progenitor cells in neural stem cell cultures (Crandall et al., 2011) and neurospheres or brain organoids developed from human induced pluripotent cells (Garcez et al., 2016). ZIKV has also been shown to induce cell death and attenuate growth mediated through vast transcriptomic alterations (Crandall et al., 2011) involved in apoptosis (Crandall et al., 2011; Garcez et al., 2016) and cell cycle regulations (Crandall et al., 2011). The disruption of the normal retinoic acid signaling in developing forebrain has been shown to induce death of neural precursor cells and attenuate their growth with somewhat similar mechanisms (Rajaii et al., 2008), raising a possibility that ZIKV might be disrupting retinoic acid signaling pathway. Our hypothesis, hence derives support from known literature with multiple justifications. It explains many mysteries and helps in understanding the embryological basis of ZIKV mediated microcephaly.

Additionally, the white matter hypodensity noted in ZIKV mediated microcephaly (Hazin et al., 2016) can also be explained by dysregulation of retinoic acid signaling due to its noted involvement in myelination in the developing brain. To augment this, adequate retinoid receptor expression on oligodendrocyte precursor cells is necessary which later myelinate CNS (Huang et al., 2011). Dysregulation of retinoic acid signaling has also been implicated in some demyelinating pathologies (Okuda and Prados, 2009; König et al., 2012). Also, there is sufficient evidence that the retinal or other ophthalmic lesions, a unique finding observed in some ZIKV infected fetuses (de Paula Freitas et al., 2016; Ventura et al., 2016) may precipitate through disruption retinoic acid signaling (Mic et al., 2004; Micucci et al., 2014).

Our claim that ZIKV-microcephaly may involve retinoic acid signaling dysregulation is further bolstered by the fact that many of the toxins that involve retinoic acid also have microcephaly as a presenting feature (Lammer et al., 1985; Mochida and Walsh, 2001; Kot-Leibovich and Fainsod, 2009; Paganelli et al., 2010) and may share other malformations too. This is especially relevant to malformations arising in fetus with exposure of pregnant women to the isotretinoin (an analog of retinoic acid known to cause retinoic acid embryopathy). These malformations match to almost all the features described under ZIKV-microcephaly spectrum including enlarged cistern magna and calcifications in brain parenchyma (Lammer et al., 1985; Irving et al., 1986).

With the above-described pieces of evidence, a plausible mosaic of a justified picture comes up supporting our proposition that ZIKV interferes with retinoic acid signaling in the developing brain. In view of (1) the bioinformatic evidence of the presence of RARE sequence repeats in genome sequences of ZIKV strains and in other viruses known to cause congenital brain malformation (Table 3) and (2) confirmed role of retinoic acid in neural tube formation and further brain development mediated through RARE sequences, it is scientifically coherent to assume that a RARE sequence dependent mechanism could be the underlying mechanism of microcephaly seen in ZIKV infected fetuses. We, however stress that there is a need of confirming the proposed hypothesis on animal models or neural cell cultures.

Why embryonic defects are limited to the central nervous system only, and no anomaly of the other organs or limbs are reported in ZIKV if it causes an embryopathy is quite surprising. It could probably be due to strong neurotropic nature of the virus as it has been reported to be completely absent in organs other than the brain. A recent study performed the autopsy on ZIKV infected fetuses with severe microcephaly confirmed the absence of this virus in organs other than brain (Mlakar et al., 2016). ZIKV infection has also been observed to be less prevalent in human precursor cells other than the neural precursors (Crandall et al., 2011). Again, why did USG studies in ZIKV infected fetuses show no abnormality of brain structures almost until 20th weeks of gestation is a mystery and needs to be explained. A probable explanation for the late involvement of the brain structures has been indicated in the recent study based on immune-histological features of affected brain tissue. It indicates that time is consumed in virus invasion of the neurones and accumulation of approximate lesions capable of arresting cortical growth (Mlakar et al., 2016).

We chose retinoic acid for its probable role in ZIKV-mediated microcephaly but there is obvious need of exploring other molecular factors which may be involved in neural tube axis determination and its further development. Also, retinoic acid signaling dysregulation is not able to explain multifocal calcifications widespread in the basal ganglia, thalamus and cortical regions as noted in histological and radiological studies of ZIKV-microcephaly cases (Aragao et al., 2016; Calvet et al., 2016; Hazin et al., 2016; Mlakar et al., 2016). It is, however, important to mention that a rare presentation of an embryological anomaly syndrome (“CHARGE” syndrome, in which retinoic acid signaling dysregulation; Micucci et al., 2014) has been noted and reported with bilateral basal ganglia calcifications (Jain et al., 2011). Calcifications in brain parenchyma have also been mentioned in retinoic acid embryopathy (Lammer et al., 1985). As the calcifications found in the ZIKV affected fetal brain are punctiform and appear as destroyed neuronal structure on histological examination (Mlakar et al., 2016), an alternative explanation for their formation could be that calcifications are due to the deposition of the calcium compounds in the accumulations of viral invaded dead neurons (Mlakar et al., 2016) as the process of degeneration.

To test our hypothesis, it was is essential to ascertain that the ZIKV strains which contain either equal or less number of RARE sequences than Brazilian strain (Table 2) in their genome can also cause microcephaly or not. As we discussed above based the ZIKV strain which was recently implicated in microcephaly in Brazil has close similarity with the Polynesian and other two strains (Mlakar et al., 2016; Schuler-Faccini et al., 2016; Table 2, iv, viii, xiv), and all three contained the highest numbers of RARE consensus sequences in our study. But to make a valid claim to relate number of RARE consensus sequences with virulence of the strains, further availability of knowledge on the ZIKV strains implicated in the microcephaly is indispensable.

Future prospects

The immediate application of our study are immense. If it is established that ZIKV caused microcephaly in infected fetus is a developmental condition arising out of retinoic acid signaling dysregulation it would not only help to understand the ZIKV–microcephaly pathogenesis but also may justify retinoic acid as a therapeutic target for preventing this condition.

Conflict of interest statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.