Introduction
Ependymal cells form a cuboidal monostratified epithelium that covers the wall of the cerebral ventricles and the central canal of the spinal cord. One of their distinctive morphological characteristics is that they are multiciliated, which allows them to perform an asymmetric shake that generates a constant flow of cerebrospinal fluid (CSF) [
1,
2]. Ependymal cells are linked by different types of cellular junctions, like cadherins and integrins [
3‐
5], that maintains the integrity of the ependymal epithelium. Also, the use of sialic acid-specific lectins established that the luminal surface of the ependyma is rich in sialic acid, a feature that might be relevant to various properties such as epithelial permeability and integrity, cell masking in innate immunity, or CSF dynamics [
6‐
8]. The integrity of the ependymal epithelium is essential for the stability of the ventricular system [
7,
8] and it has been shown that the absence of ependyma [
9,
10] or a lack of ciliary beating [
1,
11‐
15] can cause hydrocephalus.
Neuraminidase (NA) is a sialidase found in the membranes or cell walls of certain viruses and bacteria. Some of these microorganisms displaying NA in their coats induce damage in ependymal cells when invading the central nervous system (CNS), as occurs with
Streptococcus pneumoniae. This bacteria, which is the main cause of bacterial meningitis in humans [
16], also provokes ependymal damage [
17,
18]. In addition, NA-bearing viruses from orthomyxoviridae family (influenza viruses), and paramyxoviridae family (causing mumps and measles) have been associated to ependymal death and hydrocephalus [
19‐
28].
In rats, a single intracerebroventricular (ICV) injection of NA into the lateral ventricles induces an acute sterile neuroinflammation, which disappears about 2 weeks later [
6]. This neuroinflammation model is characterized by a strong activation of microglia and astrocytes in areas near the ventricles, and infiltration of neutrophils, monocytes, CD8 + T lymphocytes, and B lymphocytes from the bloodstream [
6]. In addition, a high dose of NA (Cat. nº 107,590, Boheringer, Mannheim GmbH, Biochemica, Germany) causes a massive destruction of the ependymal epithelium lining the ventricular walls, occlusion of the cerebral aqueduct and the development of hydrocephalus [
7]. With a lower dose of NA (Cat. nº 1 585 886, Roche Diagnostics GmbH, Germany), only limited patches of the ependyma are lost (53.5 ± 8.5%)[
29]. However, this loss is permanent since the ependyma does not regenerate; a glial scar appears in the nude surfaces [
30].
Recently, the activation of the complement system by NA has been demonstrated, as well as its participation in ependymal cell death; however, complement alone does not account for ependymal death, since some damage occurred even when the complement was not active [
29]. Due to the relevance of the ependyma in the stability of the ventricular system [
7,
8] and in preventing hydrocephalus [
9,
31‐
33], investigating the causes of the ependymal loss provoked by microbial NA is of outstanding interest. As activation of microglia in periventricular areas is an outstanding feature of NA-induced inflammation, we hypothesize that activated microglia might participate in ependymal cell death. Recently, the direct activation of microglia by NA acting through toll-like receptor 4 (TLR4) has been demonstrated [
34].
Explants obtained from ventricular walls represent a valuable and complex culture model where tissue cytoarchitecture is preserved. A layer of subependymal microglia lies underneath the ependymal epithelium [
35]. To assess the effect of NA-activated microglia on ependymal cells, the viability of ependymal cells in ventricular wall explants exposed to NA in vitro was quantified. To further investigate the impact of NA-activated microglia on ependymocytes, pure cultures of isolated ependymal cells were co-cultured with isolated microglial cells in the presence of NA. A possible mechanism of ependymal death involving pro-inflammatory cytokines was later inquired. Therefore, this work aimed to explore the contribution of NA-activated microglia to the impaired viability of ependymocytes, and the possible role of specific pro-inflammatory cytokines.
Material and methods
Animals
Male Wistar rats (9 weeks old; 350 g weight) were provided by Charles River Laboratories (Barcelona, Spain). These animals were maintained in the animal house at Universidad de Malaga, under a 12 h light/dark cycle, at 23 °C and 60% humidity, with food and water available ad libitum. To obtain primary cell cultures, rats were anaesthetized with 2,2,2-tribromoethanol (300 mg/kg i.p.) before decapitation. Animal care and handling was performed according to guidelines established by Spanish legislation (RD 53/2013) and the European Union regulation (2010/63/EU). All procedures were approved by the ethics committee of Universidad de Malaga (Comite Etico de Experimentacion de la Universidad de Malaga; reference 2012–0013). All efforts were made to minimize the number of animals used and their suffering.
Ventricular wall explants
Explants were obtained from the wall of the lateral ventricles of adult rats, following the procedure described by Grondona et al. [
36]. They were about 0.5 mm thick and included about 1.0–1.5 mm
2 of ventricular surface. These explants were used for: (i) obtaining pure cultures of ependymal cells, (ii) experiments of co-culture with microglial cells, (iii) cytokine receptor gene expression studies.
Pure cultures of ependymal cells
Pure ependymal cell cultures were obtained from ventricular wall explants as described by Grondona et al. [
36]. Briefly, the explants were initially washed with ice-cold HBSS without calcium and magnesium (Invitrogen, ref. 14170-088) for 30 min, and then incubated with TrypLE™ Express (Invitrogen, ref. 12605-010) following a specific sequence of temperatures: 4 °C [
5 min], 37 °C (20 min), and 4 °C (5 min). The explants were then washed for 10 min in ice-cold alpha-MEM (Invitrogen, ref. A10490-01) with the following additives: 0.2% Pluronic F-127 (Sigma, ref. P-2443), 0.3% glucose (Sigma, ref. D-7021), and 0.01 M HEPES (Invitrogen, ref. 15,630,056). Finally, they were incubated at 37 °C and 5% CO
2 during 24 h in a separation medium consisting of αMEM supplemented with 0.2% Pluronic F-127, 0.3% D-glucose, 0.01% DNase type I (Sigma, ref. DN-25) and 0.01 M HEPES. After 24 h, detached ependymal cells, which were free floating and moving in the medium, were harvested by centrifugation (300
g, 10 min) and suspended in the supplemented αMEM (described above). Isolated ependymal cells were used for viability studies in co-culture with microglial cells, as well as to determine the expression of specific cytokine receptors. For co-culture experiments, the ratio microglial cells/ ependymal cells was about 3.3. This ratio was estimated considering (i) the relative proportion of both cell types in vivo in the rat lateral ventricles (ratio ≈0.24), (ii) the intimate relationship of ependymal cells to subependymal microglia (which is lost in the co-culture system), and (iii) the relatively high dilution of microglial secretion within the culture media (a volume much larger than that of the ventricular cavities).
Isolation and culture of microglial cells
Microglial cells were isolated according to the Saura´s method [
37]. The mix cells cultures were obtained from of 3 to 5-day-old rats sacrificed by decapitation. The purity of these microglial cultures was checked by immunocytochemistry and was about 95%. The average yield with this method was about 20,000 cells/well in 12 multiwell plates (Sigma-Aldrich, TPP tissue culture plates, Z707775). Microglial cells survived in culture for at least 2 weeks (co-culture experiments were performed at shorter times) in the presence of conditioned media obtained from the previous mixed cultures. Rat microglial cultures were used for co-culture experiments with either explants or ependymocytes.
Viability assay for ependymal cells in explants
Lateral ventricular wall explants (septal and striatal sites) obtained from adult rats were used to assess ependymal cell viability under different experimental conditions. Explants were individually placed in 12-well plates in DMEM-F12 medium, supplemented with 10% FBS and 1% penicillin/streptomycin (1 ml per well). The explants were co-cultured with pure primary microglial cells (20,000 microglial cells per well). Four independent experiments with explants and microglial cultures obtained from different rats were performed.
Microglia activation was achieved by the addition to the culture medium of lipopolysaccharide (LPS; InvivoGen, LPS-EB Ultrapure E. coli 0111:B4; 1 μg/mL) [
38,
39] or NA (Neuraminidase, Roche,
C. perfringens 11 585 886 001; 50 mU/mL) [
40]. Other conditions consisted of: (i) explants treated with NA without microglia, and (ii) explants co-cultured with non-activated microglia. All these culture conditions were maintained for 24 h. Then, the viability assay was performed as follows. Explants were incubated for 10 min in a 0.4% solution of the vital stain trypan blue (Gibco; 15250061). After staining they were washed with HBSS for 2 min, immersed in Bouin´s fixative solution for 2 h (5% acetic acid, 9% formaldehyde, and 0.9% picric acid), and later embedded in paraffin wax. Five-micrometer paraffin sections were obtained from each explant, aiming to get a cutting plane perpendicular to the ependymal surface, so that ependymal cells could be clearly identifiable. Paraffin sections were mounted onto slides treated with poly-
l-lysine solution (Sigma-Aldrich; P8920). After deparaffinization, tissue sections were stained with hematoxylin to visualize the tissue and to stain live cells, while dead cells were distinguished by a blue staining (Fig.
2). Images were captured using an Olympus VS120 microscope through UPLSAPO 20 × objective. About 400 live (white) or dead (blue) ependymal cells were counted per explant; viability was expressed as the percentage of living cells.
Viability assay for cultured ependymal cells
Primary cultures of ependymal cells obtained from adult rats were placed in DMEM-F12 medium supplemented with 10% FBS and 1% penicillin/streptomycin. Cells were seeded at a density of 1500 cells/well in a 24-multiwell plate, using 0.5 ml of media per well. Some ependymal cells were placed in wells already containing a pure microglial culture (from rat) previously prepared. To trigger microglial activation, LPS or NA were added (same concentrations as described for explants); some wells were left without NA or LPS as controls. LPS or NA were also added to wells with ependymal cells but with no microglia. Incubation times were 1, 3, 6 and 24 h; six independent experiments with ependymal and microglial cultures from different rats were done. After incubation under the different experimental conditions, ependymal cells were immediately processed to evaluate their viability by treating them with a 0.4% trypan blue solution for 5 min. Then, they were centrifuged and the pellet fixed with 4% paraformaldehyde (PFA) (Sigma-Aldrich; ref. 1.00496) for 5 min, and washed with 0.9% sterile saline. Live (white) and dead (blue) cells were counted under the microscope by using a Neubauer chamber. Cells were counted following a standardized path along the Neubauer chamber, until reaching about 400 cells per sample, which represent approximately one third of the ependymal cells seeded in each well. All counting were performed blinded for experimental groups. Viability was expressed as percentage of living cells.
For experiments with functional blocking antibodies to cytokines anti-IL-1β (rat IL-1β/IL-F2 Antibody, R.D. Systems, AF-501-NA) and anti-TNFα (rat TNFα Antibody, R. D. Systems, AF-510-NA) were added to the culture medium at 5 μg/mL and 2.5 μg/mL respectively; ependymocytes viability was determined 24 h later. Eight independent experiments with ependymal and microglial cultures obtained from different rats were performed.
Immunocytochemistry
Ependymal and microglial cell cultures were fixed in 4% PFA for 10 min. The primary antibodies used were as follows: rabbit anti-IBA1 (1:1000, WAKO, 19-19741); goat anti-IL-1β (1:500, R&D Systems, AF501NA); and mouse anti-β-IV tubulin (1:1000, monoclonal Sigma T7941). The secondary antibodies used were as follows: biotinylated goat anti-mouse IgG (H + L) (1:1,000, Thermo Fischer Scientific 31800), donkey anti-rabbit Alexa 488 (1:1000, Thermo Fischer Scientific A-21206); and donkey anti-goat Alexa 594 (1:1000, Thermo Fischer Scientific A-11058). Both immunoperoxidase and immunofluorescence labelling was performed as described elsewhere [
34,
40]. Negative controls for the immunostaining consisted in equivalent cultures subjected to the same protocol but omitting the primary antibody.
Specificity of the primary antibodies
The specificity of anti-IBA1 (WAKO, 19-19741) (immunogen: synthetic peptide corresponding to C-terminus of Iba1) and mouse anti-β-IV tubulin (Sigma T7941, monoclonal antibody produced by the ONS.1A6 hybridoma; immunogen: synthetic peptide from the C-terminal sequence of β-tubulin isotype IV coupled to BSA) were validated by the manufacturer, as given in the specification sheets.
Gene expression by quantitative PCR
Total RNA from ependymal cells, choroid plexus (obtained from the lateral ventricles) and brain parenchyma tissue (the latter obtained from striatal wall) was isolated using TRIzol reagent (Invitrogen; ref. 15596026), following manufacturer´s instructions. The concentration of RNA was measured in a NanoDrop microvolume spectrophotometer (NanoDrop 1000, Thermo Fisher Scientific). The A260/280 ratio of the isolated RNA was usually about 1.8. cDNA synthesis from isolated RNA was performed using the SuperScript TM III First-Strand Synthesis (Invitrogen; 11752-050) according to the manufacturer´s protocol.
To quantify specific messenger ribonucleic acids (mRNAs) (Table
1) in the cDNA samples, the SYBR Green I based method for qPCR was employed. The hot start reaction mix FastStart Essential DNA Green Master (Roche; 06 402 712 001) was used for this purpose. qPCR reactions were prepared following manufacturer´s instructions. PCR reactions were carried out in a LightCycler® 96 Instrument (Roche), programmed with 45 cycles of melting at 95 °C for 10 s, annealing at 60 °C for 10 s, and elongation at 72 °C for 10 s. The information obtained (amplification curves, melting curves and crossing points, CP, o cycle threshold, Ct) for each transcript was processed using the software provided with the LightCycler® equipment. To estimate the PCR efficiency (E), serial dilutions of the cDNA samples were amplified, and E calculated according to the equation E = 10[−1/slope] [
41]. Relative quantification was based on the level of expression of a target gene relative to the level of expression of a reference gene (glyceraldehyde 3-phosphate dehydrogenase, GAPDH) [
42]. Thus, for each cDNA experimental sample, the expression of a particular target gene (Table
1) relative to GAPDH expression (gene mRNA rel. GAPDH) was calculated as follows:
$${\text{Gene mRNA rel}}.{\text{ GAPDH }} = \, {{\left( {E_{{{\text{target}}}} } \right)^{{\Delta {\text{CP target}}}} } \mathord{\left/ {\vphantom {{\left( {E_{{{\text{target}}}} } \right)^{{\Delta {\text{CP target}}}} } {\left( {E_{{{\text{GAPDH}}}} } \right)^{{\Delta {\text{CP GAPDH}}}} }}} \right. \kern-\nulldelimiterspace} {\left( {E_{{{\text{GAPDH}}}} } \right)^{{\Delta {\text{CP GAPDH}}}} }}$$
Table 1
Sequence of primers used in qPCR
GAPDH | CACTGCCACTCAGAAGACTG | GGCATGTCAGATCCACAAC |
FOXJ1 | GACTATGCCACCAACCCACA | CGGATGGAATTCTGCCAGGT |
IIIG9 | ACAACCCCAGCTATGTTCGG | GGCACGTCTCGATAGAAGGG |
IL-1βR1 | AGAAACTCAACATACTGCCTCA | CAGCCACATTCATCACCATC |
TNFαR1 | TGTTGCCTCTGGTTATCTT | ACCCTCCACCTCTTTGAC |
where E is the efficiency of the corresponding gene (target gene or GAPDH), ΔCP
target is the difference in crossing point values for the target gene obtained in the control sample and in the experimental sample, and ΔCP
GAPDH is the difference in crossing point values for GAPDH obtained in the control sample and in the experimental sample [
42].
Analytical methods
The statistical analysis of the data was carried out using SPSS Statistics software. The Kolmogorov–Smirnov normality tests, along with the Levene homoscedasticity test, were used to verify if data could be analysed by parametric methods. One-way or two-way analysis of variance (ANOVA) were used to compare mean values. Afterwards, the pairwise comparisons were done by the Tukey test. For the non-parametric datasets, the Kruskal–Wallis test was used, and in this case pairwise comparisons were done with Mann–Whitney U test. In all comparisons differences between means were considered significant when the P value obtained was < 0.05.
Discussion
It is well documented that a single intracerebroventricular (ICV) injection of NA provokes denudation and death of the ependymal epithelium, which is not restored over time [
7,
30]. In the most severe cases, that is when high NA doses are injected, the ependymal loss is complete, and an obstructive hydrocephalus develops [
7]. A lower dose of NA (the model used here) results in: (i) partial ependymal loss (53.5 ± 8.5%)[
29], (ii) strong microglial activation, and (iii) infiltration of leukocytes into the meninges, the cerebrospinal fluid and the brain parenchyma. Thus, this paradigm represents a model of acute aseptic neuroinflammation [
6].
A unique feature of this model is the damage to the ependymal epithelium. Among the various possible causes, the contribution of the complement system to ependymal damage and death has been demonstrated, as well as the direct impact of NA itself, which may also provoke mild ependymal damage without the aid of the complement [
29]. However, these causes do not fully account for all ependymal death. Thus, additional mechanisms, which are still not completely clarified, may underlie ependymal death.
Here, attention focused on microglial cells as players in ependymal damage, as they become activated upon exposure to NA. In fact, the direct activation of microglia by NA acting through the receptor TLR4 has been recently reported [
40], as well as the resulting induction of inflammatory cytokines such as IL-1β, IL-6 and TNFα [
40]. These cytokines could cause cellular damage, including the death of ependymal cells. The results obtained here provide evidence for this possibility. As (1) the ependymocytes isolates used here were 100% pure, (2) the expression of the receptors was detected by a quite sensitive technique such as qPCR, and (3) receptors expression was differentially detected in RNA extracts from other tissues (choroid plexus, brain parenchyma), it is very probable that ependymocytes constitutively express receptors for both IL-1β and TNFα cytokines. Furthermore, NA provoked an increase in the expression of IL-1β and TNFα receptors in ventricular wall explants, suggesting that this tissue becomes more susceptible to these inflammatory cytokines. However, demonstrating the protein expression of both receptors would be necessary to definitely confirm this finding. In this regard, the presence of IL-1βR1 in the ependyma in vivo has been previously described by other authors by both in situ hybridization and immunohistochemistry [
47,
48], confirming that both the mRNA and the protein for this receptor are expressed in ependymocytes. Regarding TNFα receptor, Nadeau and Rivest [
49] showed the mRNA expression in ependyma and choroid plexus by in situ hybridization, in accordance to our results. However, in a study using immunohistochemistry, TNFα receptor was not found in rat brain ependyma [
50]. Furthermore, personal communication of unpublished results confirms such observation, as ependymal cells and choroid plexus appeared negative in mouse brain immunohistochemistry for TNFα receptor. Therefore, our results along with other reports indicate that ependymal cells express the receptor for IL-1β. However, and even though TNFα receptor mRNA has been detected in ependymal cells by qPCR and in situ hybridization, these cells do not seem to express a functional receptor for TNFα.
The presence of cytokine receptors in ependymal cells allows us to speculate the possibility that cytokines may mediate ependymal death. In previous works functional blocking antibodies have been used to unravel the participation of cytokines in inflammation [
51,
52] and in brain cell death [
53]. This strategy was used here to investigate the possible involvement of IL-1β and TNFα in ependymocyte viability. Our results strongly suggest that IL-1β impairs ependymal cell viability, whereas TNFα does not. This is in agreement with the fact that ependymal cells express IL-1β receptor but not TNFα receptors, as discussed above. Increased cell survival in hippocampus when using neutralizing IL-1β antibodies has also been described in cases of transient global ischemia [
54]. Besides, the systemic infusion of IL-1β neutralizing antibodies reduced short-term brain injury after cerebral ischemia in the ovine foetus [
55]. However, with these results, we cannot rule out the participation of other molecules produced by microglia (or other cells) in the death of ependymal cells.
Although the main role of microglia in the brain is protective and homeostatic, activated microglia release inflammatory cytokines that can sometimes generate neurotoxic effects [
56,
57]. This fact has been described in neurons [
58,
59], in Purkinje cells [
60], in oligodendrocytes [
61,
62] and in retinal ganglion cells [
63]. Furthermore, these harmful effects of inflammatory cytokines have also been observed in experimental neuroinflammatory processes caused by inoculation of the influenza A virus, which has NA in its lipid envelope [
64,
65]. This evidence supports the results obtained in the present work, that suggest the involvement of IL-1β in ependymal cell death.
Ependymal cells form a continuous epithelial layer that covers the brain ventricles and the central canal of the spinal cord [
3,
8]. The integrity of the ependymal epithelium is essential for the stability of the brain ventricular system [
7,
9,
32,
66]. It is well established that the loss of the ependymal cells leads to stenosis and obliteration of the cerebral aqueduct, giving rise to hydrocephalus [
7,
33]. Moreover, even a dysfunction of the ciliary beating in ependymal cells generates hydrocephalus [
1,
11‐
15]. Numerous cases of hydrocephalus with ependymal loss have been described in infections with NA-bearing viruses such as influenza, mumps, and measles viruses. [
19,
20,
23‐
25,
27,
28]. In the model of NA-induced inflammation used here, several events concur that contribute to ependymal damage and loss: (i) activation of the complement system, with deposition of the membrane attack complex onto the apical ependymal membrane [
29]; (ii) microglial cell activation and IL-1β production [
34]. In vitro experiments showed that even NA by itself can cause some ependymal damage [
29]. Therefore, the ependymal cell death reported in infections by NA-bearing viruses could be ascribed to the activity of NA. Despite this evidence, we cannot rule out that the behaviour of ependymal cells in vitro could be different from that in vivo. However, in vivo data support the effect of activated microglia on the viability of ependymal cells. Thus, in mice injected ICV with LPS, where an important microglial activation occurs, the death of ependymal cells was observed [
67]. Also, in a similar model which uses ICV injected NA instead, there is a remarkable death of ependymal cells in vivo [
6,
7,
30]. Hence, these in vivo experiments are in accordance with the in vitro results shown here and support the involvement of activated microglia in ependymal cell death.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.