CD200-CD200R imbalance correlates with microglia and pro-inflammatory activation in rat spinal cords exposed to amniotic fluid in retinoic acid-induced spina bifida

Spina bifida aperta is a congenital malformation characterized by the failure of neural tube closure resulting in an unprotected fetal spinal cord. The spinal cord then undergoes progressive damage, likely due to chemical and mechanical factors related to exposure to the intrauterine environment. Astrogliosis in exposed spinal cords has been described in animal models of spina bifida during embryonic life but its relationship with neuroinflammatory processes are completely unknown. Using a retinoic acid-induced rat model of spina bifida we demonstrated that, when exposed to amniotic fluid, fetal spinal cords showed progressive astrogliosis with neuronal loss at mid-gestation (E15) compared to unexposed spinal cords. The number of microglial cells with a reactive phenotype and activation marker expression increased during gestation and exhibited progressive disruption in the inhibitory immune ligand-receptor system. Specifically we demonstrate down-regulation of CD200 expression and up-regulation of CD200R. Exposed spinal cords demonstrated neuroinflammation with increased tissue water content and cytokine production by the end of gestation (E20), which correlated with active Caspase3 expression in the exposed layers. Our findings provide new evidence that microglia activation, including the disruption of the endogenous inhibitory system (CD200-CD200R), may participate in the pathogenesis of spina bifida through late gestation.

Involvement of the immune ligand-receptor CD200-CD200R complex has been demonstrated in the regulation of cytokine production 7 . The CD200 transmembrane glycoprotein, mostly expressed in neurons, interacts with its receptor, CD200R which is expressed in the CNS almost exclusively in microglia as well as other CNS macrophages, to inhibit microglial priming and holds microglia in a quiescent state 8 . During normal development, this cell-cell communication reinforces an anti-inflammatory response, and is responsible for microglial development and renewal in the developing CNS 8 .
However, disruption of the CD200-CD200R signaling can cause microglial over reactivity [9][10][11][12] . In spina bifida the loss of neural tissue 2,13 can disrupt the CD200-CD200R ratio, thus leading to over activation of the microglia, induction of the neuroinflammatory process, and neurodegeneration 14,15 . Knowing the exhisting neural tissue damage in utero in spina bifida and given the importance of this immune ligand-receptor system in regulating microglial expression during brain development and microglial priming; we hypothezised that spina bifida suffer a progressive astrroglial and microglial activation during gestation and the microglia priming may be related to the alteration of the CD200-CD200R system. This study examines progression during gestation of astrogliosis and neuroinflammation analyzing the effect of CD200-CD200R on microglia reactivity during gestation in the rat fetal model of retinoic acid-induced spina bifida aperta.

Material and Methods
All experimental protocols were approved by the Institutional Animal Care and Use Committee at the Children's Hospital of Cincinnati, and followed guidelines set forth in the National Institutes of Health Guide for Care and Use of Laboratory Animals (IACUC 2013-0293).
Retinoic acid (RA)-induced spina bifida animal model. 36 timed-pregnant Sprague-Dawley rats weighing 200-250 g (Charles River Laboratories, Inc, Wilmington, MA) were housed individually at 22°C on a standard dark:light schedule (12:12 hours,i.e., light 7:00-19:00) with ad libitum access to water and standard chow. Mating date was defined as E-1 and plug date as 0. Trans retinoic acid (RA) (Sigma-Aldrich Chemical, St. Louis, MO) solubilized at room temperature in olive oil was used within 1 h of preparation and protected from light. On E10 at 10 am, 20 timed-pregnant rats were gavaged with RA 100 mg/kg in olive oil to induce spina bifida in the litter and 16 timed-pregnant rats were gavaged with olive oil as a sham.
Tissue Processing. Litters from timed-pregnant rats were harvested at three timepoint in gestation (E15, E17, E20). Tissue samples were collected and included spinal cords exposed and unexposed to amniotic fluid in spina bifida fetuses and spinal cords from sham fetuses. For gene and protein expression, spinal cords were removed, immediately snap frozen, and stored at −80 °C until use. Specimens for histological analysis were isolated and fixed in 4% paraformaldehyde for 24 hours, processed, and embedded in paraffin. For flow cytometry and water content experiments, fresh cords were extracted and processed according to specific protocols described below. (a) Exposed cord at gestational days E15, E17, and E20 (4×). (b) At E20 (sagittal view), exposed cord shows altered cellular density and blood vessel extravasation (arrows) (4×).
SCiEnTiFiC REPORts | (2018) 8:10638 | DOI:10.1038/s41598-018-28829-5 RNA Extraction. Tissues were homogenized (IkaT10 basic Ultra-Turrax homogenizer, USA) in RLT buffer and RNA extracted using an RNeasy Plus Mini Kit (Qiagen Science, Hilden, Germany) following manufacturer's protocol. Samples were quantified by spectrophotometry (Epoch Biotek, Biotek Instruments, Winooski, Vermont). A 1-µg RNA/sample was reverse transcribed into cDNA using the RT 2 First Strand Kit (Qiagen Sciences, city, Maryland, USA). A 1-µg cDNA sample was used to set up RT-qPCR using TaqMan R gene expression assay (Applied Biosystems, Foster City, CA, USA) (Supplementary Table 1) and 7500 Fast Real-Time PCR System. Target genes were normalized using HPRT1 as endogenous control; their relative quantification of transcript expression was performed using the 2 −Δ ΔCt method (C t represents the threshold cycle). Samples were run in triplicate.
Protein extraction. Total protein extracted from spinal cord samples was 20-30 mg. Tissue samples were homogenized by sonication (Fisher Scientific, Pittsburgh, PA) in N-PER (Thermo Fisher Scientific, Rockford, IL) + Proteinase K + PhosphoSTOP (Roche Diagnostics GmbH, Mannheim, Germany) in 1.5-mL tubes, and centrifuged at 10,000 rpm for 5 minutes. Protein concentration was assessed using the BCA-mini method (Thermo Fisher Scientific, Rockford, IL).

Results
In our rat model, retinoic acid at 100 mg/kg on gestational day E10 induced spina bifida in 71% of the fetuses.
No fetuses developed spina bifida in the 16 sham animals. Litter size was significantly smaller in the RA-induced model compared to sham, 8.6 ± 4.2 and 13.9 ± 2.1, respectively (*p < 0.05) (Supplementary Fig. 1). Fetuses at E15, E17, and E20 with open RA-induced spina bifida showed classic defects that included open posterior arc, dysraphic spinal cord (i.e., lacking dura), and exposure to the ammonitic fluid environment. Variations were observed in the shape of the open spinal cord and the extent of neural tissue loss. Spina bifida exposed cords showed progressive tissue loss throughout gestation (Fig. 1).
Effects of amniotic fluid exposure. After RA exposure and non-closure of the posterior arc, the neural tissues remained exposed to the amniotic fluid. Compared with sham controls at E15 and E20, gene expression of the neural microtubule-associated protein 2 (MAP2) and RNA binding protein Fox-1 Homolog 3 (Rbfox3) were downregulated in the exposed spinal cord but significantly increased in the unexposed spinal cord at E20 (Fig. 2a). NeuN expression was significantly (*p < 0.05) lower in exposed spinal cords than either unexposed or sham cords (Fig. 2b) at E 20. βIII-Tub and NeuN staining identified unprotected neurons exposed to the amniotic environment, which caused neural loss and resulted in abnormal cytoarchitecture of the exposed spinal cord (Fig. 2c). This architecture was also affected in regions where reduced cellular density was clearly distinguished with extravasation of the blood vessels (arrows) (Fig. 1b).

Astrogliosis in the injured spinal cord.
At each of the timepoints, GFAP expression was dramatically higher in the exposed spinal cords than either the unexposed region of cord from the same spina bifida fetus or sham cords, especially at E20 (p < 0.05) (Fig. 3a); this was confirmed by Western blot analysis (Fig. 3b). Distribution of GFAP-positive astrocytes appeared as numerous aggregates throughout the lesion near the exposed neural tissue in spina bifida fetuses, but followed normal distribution in the spinal cords for unexposed and sham at E20 (Fig. 3c). GFAP expression showed a progressive pattern through gestation from E15 to E20. At E15, high expression of GFAP was detected in some areas in the exposed cords but not distributed equally in the exposed layers compared with incipient low expression in the non-exposed and sham cords (Fig. 4a). Higher expression of GFAP in exposed layers also correlated with higher levels of proliferation identified by immunofluorescence for Ki67 (Fig. 4b).
Neuroinflammation and reactive microgliosis in exposed neural spinal cord. To confirm the microglia activation in spina bifida fetuses, we assessed morphological changes and quantified microglial-specific markers Iba1 and CD11b (both constitutive markers). CD11b and CD68 gene expression was progressively significantly up-regulated after E17 of gestation in spinal cords exposed to the amniotic fluid compared with unexposed spinal bifida or sham spinal cords (p < 0.05) (Fig. 5a). In Western Blot analyses, Iba1 protein expression was higher in the spinal cords exposed to amniotic fluid than those unexposed at E20 (p < 0.05) (Fig. 5b). Furthermore, of the microglia present, an increase in the population with activated morphology was detected by immnohistochemistry at E17 and E20 but not at E15 (Fig. 5c). More Iba1 cells with activated morphology were seen at E20 showing the progressive increase in microglial cells (Fig. 6a). Mixed populations of microglial cells were also present in the spina bifida exposed spinal cord, including Iba1 + and MHCII + cells, and Iba1 + cells MHCII-cells; this finding thus differentiated between activated and unactivated cells. In contrast, only Iba1 + MHCII-cells were observed in unexposed spinal cord as resting phenotype with extended projections (Fig. 6b).

CD200-CD200R alteration correlates with cytokine production in spina bifida. The relationship
between the imbalance of CD200-CD200R and the pro-inflammatory cytokines in spina bifida was demonstrated by the expression of the ligand CD200, mostly by neurons, and the expression of receptor CD200R, mostly by microglia. Compared with unexposed and sham spinal cords, exposed neural tissue showed reduced CD200 expression, increased CD200R expression, and disruption of the normal CD200R-CD200 ratio throughout gestation with more significance at E20 (p < 0.05) (Fig. 7a). In the exposed spina bifida spinal cords, this activation Figure 6. Progressive microglia activation in spinal cord exposed tissue in spina bifida. (a) Immunoreactive astrocytes GFAP (red), microglia Iba1 (green), and DAPI (blue) in exposed and un-exposed spina bifida spinal cords; top panel (magnification 20×). Activated rounded microglial morphology in exposed spinal cord in the layers exposed to the amniotic fluid compared with resting morphology in the unexposed spinal cord in spina bifida pups; lower panel (magnification 40×). (b) Iba1 (red), MHCII (green) and DAPI (blue) immunostaining in exposed and unexposed spina bifida cords. Iba1+/MHCII− (arrows) and Iba1+/MHCII+ (arrowheads) microglial cells in layers of the spinal cord exposed to the amniotic fluid (magnification 100×) also showed activated phenotype compared with Iba1+/MHCII−(arrows) microglial cells in resting phenotype with extended projections in unexposed spinal cords (magnification 100×). correlated with pro-inflammatory cytokine production as demonstrated by the gene expression of IL1β at E20 and by the increase of IL1β, IL6 and IFNγ protein expression at E20 (p < 0.05) (Fig. 7b,c).

Flow cytometry evaluation of activated microglia/macropahges cell populations. Cells were
initially sorted on the basis of viability and the presence of CD45 staining. Microglia/Macrophages cell population were selected for CD45 and then positively selected for CD11b/c. Within populations of CD45/CD11b/c cells on cords in spina bifida and sham spinal cords, exposed spinal cords in spina bifida fetuses had increased CD200R + cells (14.54%) compared with unexposed spina bifida and sham, respectively (9.07% and 2.27%). Within the CD200R + positive subpopulation, MHCII + was also increased in exposed cords (14.86%) compared with unexposed cords in spina bifida (0.17%) and sham (0.00%) (p < 0.05) (Fig. 8). Fetuses they comprised Water Content. In assessing water content and percentage by dry and wet methods, respectively, significant differences were found in the cord tissues of the same spina bifida pup with exposure to the amniotic environment ( Supplementary Fig. 2a). Water content was significantly higher in exposed than unexposed spinal cords in the same spina bifida fetuses. A fairly small change in percentage water content actually reflected large changes in tissue water content ( Supplementary Fig. 2b) (*p < 0.05, paired Student's T-test).

Spina bifida exposed spinal cord induced apoptosis.
To test if spinal cord exposure to amniotic fluid induced apoptosis by activation of the neuroinflammatory process, tissue sections were stained for cleaved caspase 3. In spina bifida fetuses, spinal cords exposed to the amniotic fluid exhibited a greater density of cleaved-caspase-3 + cells in the exposed layers at E20 of gestation. No positive cells were found in the unexposed spina bifida spinal cords and sham controls (Fig. 9).

Discussion
Our study describes the disruption of normal prenatal CNS evolution with presence of neuroinflammatory processes and microglia activation in an RA-induced rat model of spina bifida. One of the possible mechanism involved in priming microglia is the immune suppressor ligand-receptor CD200-CD200R. The imbalance on CD200-CD200R is related to the activation of microglia and pro-inflammatory cytokines. Our findings provide evidence for the first-time into the presence of the microglia with activated phenotype in the pathophysiology of spina bifida. Our characterization of neuroinflammation in a retinoic-acid induced animal model of neural tube defects builds upon other in vivo studies that have profiled lesions in spina bifida 18 . By performing analysis at the gestation time points (E15, E17, and E20) we were able to identify progressive effects of exposure of the spinal cord to Comparison of CD200R, CD200, and CD200/CD200R ratio relative expression in exposed neural tissue, unexposed, and sham animals at different gestational age (E15, E17 and E20). Means ± SE (n = 6 fetuses/group) of relative expression (2 − ΔΔCt) for RA-treated and time-matched controls; (*p < 0.05). (b) IL1β gene relative expression confirmed the upregulation and correlation between microglial activation, neuroinflammation, and cytokine production in exposed spina bifida spinal cord at E20 of gestation. (c) Cytokine (IL1β, IL6, INF γ) protein expression by Luminex exhibit higher cytokine expression in the exposed spinal cord of spina bifida compared with unexposed or sham rats (n = 5 fetuses/group) at E20. Values are means ± SD (*p < 0.05).
SCiEnTiFiC REPORts | (2018) 8:10638 | DOI:10.1038/s41598-018-28829-5 amniotic fluid. Firstly, pathophysiological changes along the spinal cord by analysis of neural cell types (i.e., astrocytes, neurons, microglia) could be compared between exposed and unexposed spinal cords in the same spina bifida fetuses. The progression of the astrogliosis and neural loss in the exposed spinal cords started very early in gestation at E15 compared with the normal development and differentiation at E18 and after birth. Second, understanding the braking effect of the CD200-CD200R interaction resulted; specifically, microglia maintained in the quiescent state led to abnormal activation in the injured tissue just before birth, especially at E20, inducing a pathological neurinflammatory response.
In spina bifida the fetal spinal cord is exposed to the enzymatic action of amniotic fluid 3 . Chemical and physical erosion result in loss of neural tissue, both aggravating the primary lesion and changing the cytoarchitecture of the spinal cord 13 .
In histological samples, regions of reduced cellular density were also observed; importantly, these regions, which contained edema and extravasation of the blood vessels, were clearly identifiable in the sagittal view along the spinal cord. The increased water content in exposed Vs. unexposed neural tissues in the same animal may reflect a process that occurs with spinal cord injuries in adults 19 . For example, when tissues in the blood-brain-barrier are broken, extravasation of the vessels, activation of the inflammatory and apoptotic processes increase the water content in injured neural tissue.
In this RA-induced rat model, the most common neuropathological feature described was the presence of reactive astroglia in spinal cords of fetuses with spina bifida exposed to the amniotic fluid. Astrogliosis has been described before in this model 8 and other spina bifida animal models 20,21 , but in our model we detected earlier activation at E15 in the exposed spinal cords from spina bifida fetuses.
Astrocytes, which are the neuromodulatory and neurotrophic elements in the CNS, respond to any stress or tissue damage by proliferation, hypertrophy, and migration. Given the strong association of glial fibrillary acidic protein (GFAP) with astrocytic cell proliferation and differentiation 22,23 , assessment of its expression may elucidate the pathophysiology of the neural tissue in open neural tube defects. The GFAP is up-regulated in reactive astrocytes 24,25 Astrogliosis has been previously described in animal models of spina bifida 18,20 at E16.5-E18. Normal astrogenesis in rat fetuses starts at E18 and continues after birth until post-natal day7 26 . However, mechanisms that control gliosis in spina bifida remain unclear. This reactive astrogliosis, which is not only a marker of neuropathology, also plays a role in the response to the injury by regulating neuroinflammation and reparation 17,25,[27][28][29][30] . Our data suggest the formation of a gliosis in injured tissues in spina bifida rat fetuses. Specifically, the reactive astrocytes proliferate in the exposed layers trying to protect the neural tissue from the hostile environment from mid-gestation (E15). Microglial cells migrate, mix with the astrocytes in the exposed layers in the spina bifida fetuses, and thus create the inflammatory response to the injury.
Microglia, the resident macrophage population of the CNS, maintain homeostasis for normal function during fetal development; their involvement includes removal of apoptotic cells and debris 31 , synaptic pruning 32 , modulation of the synaptic transmission 33 , and angiogenesis 34 . Responding quickly after injury or stress, they offer protective effects. However, if unrestricted, these effects can be deleterious leading to neural degeneration in the injured tissues. In prenatal cases, disruption of normal CNS maturation by neuroinflammation and microglia activation can cause serious cognitive and bahavioral defects [35][36][37][38] . These mechanisms, which may play an important role in spinal bifida, have not been investigated for this pathology. Microglial cells are the sentinel cells derived from the primitive yolk sac; these cells migrate and invade the CNS during early gestation E8-E8.5 36 and are the macrophages resident in the CNS when lack of closure is induced in retinoic acid models (E10). Here, using other microglial activation markers (e.g., CD11b, MHCII, CD68), we demonstrated that microglial cells are activated in the exposed spinal cord in spina bifida fetuses. More importantly, this microgliosis and neuroinflammatory response is progressive during gestation. As shown by a previously described flow cytometry strategy 17 , we corroborated the increase of activated microglia/macrophages in exposed neural tissues in the spina bifida fetuses at day 20 of gestation.
Disruption of the inhibitory ligand-receptor system, CD200-CD200R. Our findings provide new evidence that microglia activation may participate in the etiology and pathogenesis of spina bifida, especially towards the end of gestation. One mechanism implicated in microglial activation in the lesion was the disruption of the inhibitory ligand-receptor system (CD200-CD200R). Many studies suggest that CD200-CD200R signaling down-regulates the activity of myeloid cells and that microglia derived from myeloid progenitor cells migrated at an early stage from the yolk sac 39 . In analyzing the expression of CD200-CD200R, we demonstrated that disruption of this inhibitory immune ligand-receptor system in the lesion led to microglial activation at E20. Downregulation of CD200, probably due to loss of neural tissue by erosion and cell death caused by amniotic fluid exposure. Activation of astrocytes and microglia in the damaged tissue upregulated the expression of CD200R 27,40 . These dysregulations disrupted the balance of the immune ligand:receptor ratio (CD200-CD200R) 41,42 . The disruption then exacerbated microglial activation which in turn increased MHCII expression and pro-inflammatory cytokine (IL1β, IL6, and IFNγ) [10][11][12] production that was detected in the exposed spinal cords at E20.

Study limitations.
It is unlikely that CD200-CD200R system is solely responsible for priming microglia, especially at early gestational stages when the neural tube defect occurs, as CD200 and CD200R are not expressed at that time. Therefore, additional experiments are needed to elucidate the role and mechanism of macrophage/ microglia in the injured spinal cord at earlier gestational ages.

Conclusion
Neural tube defects, including spina bifida aperta, are the most common permanent disabling birth defects. Despite fetal surgery to repair or cover the spinal cord at mid gestation, significant anomalies already established persist throughout pregnancy and beyond birth. The physiopathology of the complex inflammatory activation remains poorly understood. Characterization of CNS alterations in the open lesion of RA-induced spina bifida in rats in this study provides evidence for the first-time of a relationship between reactive astrocytes, neural tissue loss, and neuroinflammation in Spine Bifida. Furthermore, we have demonstrated that activation of the neuroinflammatory mechanisms and primed microgliosis involves the disruption of the endogenous inhibitory system (CD200-CD200R). Chronic microglial activation is an important component of neurodegenerative diseases, and this chronic neuroinflammatory component likely contributes to injury in spina bifida aperta after birth.