Introduction
Hydrocephalus represents excess accumulation of cerebrospinal fluid (CSF) in the cerebral ventricles and subarachnoid space. In humans, the prevalence of congenital hydrocephalus is high and markedly geographically variable. In sub-Saharan Africa, there are approximately 750 new cases per 100,000 live births, whereas in Europe, there are only 110 cases of infantile hydrocephalus per 100,000 live births[
1]. Congenital hydrocephalus is the most common neurological disorder requiring surgery in children [
2]. Despite the high prevalence of congenital hydrocephalus, knowledge about the pathophysiological mechanisms leading to this disorder remains extremely limited. A large body of evidence suggests that congenital hydrocephalus is an inheritable disorder and characterization at the molecular level should greatly increase the understanding of the disease and lead to new therapies. However, relatively few genes in humans, including
L1CAM,
MPDZ, and
CCDC88C have been linked with hydrocephalus to date [
3‐
5]. Less than 5% of mutations in these genes account for primary congenital hydrocephalus cases [
6]. As animal models of inherited disease provide opportunities to identify potential genetic causes of disease in humans, a variety of genes causing hydrocephalus in rodents have been identified and characterized over the past decades in transgenic animal models, including the Hydrocephalic-Texas (H-Tx) rat, the Wistar-Lewis rats, the hydrocephalus-1 mouse, the hydrocephalus with hop gait mouse, and the TGF-beta1 mouse [
7‐
13].
Among these, the H-Tx rat strain, first described in 1981, with congenital hydrocephalus resulting from a spontaneous mutation is widely used as a model of the pathogenesis of the disease [
14]. Hydrocephalus in H-Tx rats develops in early gestation via a complex mode of inheritance and with a prevalence of 30–50% [
15]. Ventricular dilatation starts to occur from days 17 to 18 of gestation and is associated with aqueductal stenosis [
16,
17]. Pups with severe hydrocephalus are detected at birth by a domed head and die at 4–6 weeks of age. Breeding data from crosses between H-Tx rats and Sprague-Dawley (SD) rats have indicated that hydrocephalus in H-Tx rats is a single autosomal recessive gene disease [
18], but a series of studies by Jones et al. indicated that the phenotypic expression of congenital hydrocephalus in them is controlled by a combination of genetic and epigenetic factors [
15,
19‐
23].
The genetic abnormality causing congenital hydrocephalus in H-Tx rats remains only partially understood. In this study, we performed copy number analysis to investigate candidate genes potentially responsible for hydrocephalus in H-Tx rats. Under this approach, we identified a copy number loss in the
Ptpn20 gene in hydrocephalic H-Tx rats, generated
Ptpn20-knockout (
Ptpn20−/−) mice and then tested them for genotypic and phenotypic differences from the wild-type (WT) mice. Our study found significant differences in the expression of phosphorylated Na-K-Cl cotransporter 1 (pNKCC1) in the choroid plexus (CP) of
Ptpn20−/− mice compared to the WT. NKCC1 is encoded by Slc12a2 and belongs to the SLC12 family of cation-chloride cotransporters, which show a wide distribution in different tissues and cell types. In most secretory epithelial cells NKCC1 localizes at the basolateral membrane and mediates chloride transport as an important mechanism of cell volume regulation [
24,
25]. In contrast, in the central nervous system (CNS), cotransporters are atypically located on the apical membrane of the CP [
26,
27]. Because of this, the function of NKCC1 in the CP epithelium is unclear, and whether NKCC1 functions primarily as a secretory, permissive, or absorptive (clearance) fluid system there has been highly controversial [
28‐
30]. Based on our findings, we suggest a hypothesis about the function of NKCC1 and propose a pathophysiological mechanism that may be related to the development of hydrocephalus and together with a possible therapeutic target.
Discussion
This study identified copy number loss of
Ptpn20 in the brain of H-Tx (+) rats.
Ptpn20 was enriched in the CP of non-hydrocephalic rats, but significantly decreased in the CP of H-Tx (+) rats. We then generated
Ptpn20-deficient mice to investigate whether the deletion of
Ptpn20 is a risk for hydrocephalus.
Ptpn20−/− mice did not exhibit the characteristics of congenital hydrocephalus observed in H-Tx rats after birth. However, further analysis in adult
Ptpn20−/− mice revealed ventriculomegaly in more than half of these mice without significant blockage of CSF circulatory pathways, thus indicating that
Ptpn20−/− mice had developed communicating hydrocephalus. Since hydrocephalus is not associated with stenosis of the cerebral aqueduct, we further investigated the CP, which is mainly affected by
Ptpn20 deletion. Previous studies have shown that structural or functional abnormalities of CP may affect CSF production and thus lead to the development of hydrocephalus [
31‐
33]. To evaluate possible CP abnormalities, we analyzed the ultrastructure of the CPs of the lateral and third ventricles in
Ptpn20−/− mice using TEM. A normal appearance of the CP epithelium and intact junctions between cells suggested that passive diffusion of molecules from the CP vasculature into the CSF is unlikely. However, in analyzing the channels and transporters involved in ion and water transport for CP epithelial cells, we found that expression of pNKCC1 protein was mainly at the apical surface of CP epithelial cells and was significantly increased in about half of
Ptpn20−/− mice as compared with WT mice. This result may provide useful insights into the pathogenesis of hydrocephalus, and the individual differences in the degree of pNKCC1 expression seem to explain why only half of the knockout mice developed hydrocephalus.
NKCC1 is located on the basolateral membrane and is involved in active transportation of Na
+, K
+, and Cl
−, which play important roles in fluid absorption and secretion in most epithelial tissues. However, NKCC1 in the brain is highly expressed in the apical membrane of CP epithelial cells [
34]. Several studies have shown that inhibition of NKCC1 by bumetanide leads to decreased CSF production in the CP [
29,
35‐
37]. Ex vivo and in vivo studies have shown that NKCC1 contributes approximately half of the CSF production that is independent of the osmotic gradient [
29]. In addition, previous studies have shown that increased NKCC1 activity is associated with direct phosphorylation of NKCC1 [
38], suggesting that enhanced or continued NKCC1 phosphorylation may result in greater CSF production by the CP. This hypothesis is supported also by a rat model of posthemorrhagic hydrocephalus in which inflammation-induced phosphorylation of NKCC1 caused hypersecretion of CSF [
39].
Ptpn20 is a member of the PTPs family, which, together with protein tyrosine kinases, is responsible for regulating the phosphorylation state of intracellular protein [
40]. The function of
Ptpn20 is not yet completely understood. Limited studies have shown that
Ptpn20 is expressed in a variety of cell lines, showing a dynamic subcellular distribution in response to various extracellular stimuli targeting sites of actin polymerization [
41]. Previous studies have shown that inhibition of actin polymerization with cytochrome D increases NKCC1 activity in colonic cells [
42,
43]. However, our results did not show obvious differences in F-actin staining (Fig.
3C), suggesting that
Ptpn20 correlates with actin, but has no significant effects on actin polymerization or depolymerization.
On the other hand, NKCC1 activity is known to be regulated by phosphorylation of its specific Serine/Threonine residues in other cell types, and Ste20/SPS1-related proline/alanine-rich kinase, as an upstream regulator of NKCC1, can phosphorylate the N terminus of NKCC1 at Thr203, Thr207 and Thr212 [
44,
45].
It is not clear why NKCC1 can be activated by tyrosine kinase in the absence of tyrosine sites. However, based on this finding, we can hypothesize that NKCC1 may also be dephosphorylated by the protein tyrosine phosphatase encoded by
Ptpn20. In addition, a portion of intracellular protein tyrosine phosphatases in the nervous system also function as dual specificity phosphatases that can dephosphorylate both phosphotyrosine and phosphoserine or phosphothreonine [
46]. That is, if
Ptpn20 also possesses the properties of dual specificity phosphatase, serine or threonine residues that are phosphorylated in NKCC1 and its upstream regulators can also be directly dephosphorylated by
Ptpn20. Therefore, deletion of
Ptpn20 may thus cause NKCC1 to fail to dephosphorylate and persistently remain in the phosphorylated state, and our results showing increased expression of pNKCC1 in
Ptpn20−/− mice from 8 to 72 weeks support this concept. This change would result in a continuous excess of CSF entering the ventricular system. In humans, CSF hypersecretion secondary to CP hyperplasia or CP papilloma disturbs the homeostasis of CSF production and absorption in the brain, leading to hydrocephalus [
47,
48], and higher secretion rates correlate with more severe hydrocephalus [
49]. We therefore consider that CSF hyperproduction may be a direct cause of hydrocephalus in this study. To further confirm this hypothesis, we also examined pNKCC1 expression in H-Tx rats, and we found that pNKCC1 expression was significantly increased in the CP of H-Tx rats compared to SD rats (Additional file
3: Fig. S3B). This result suggests that increased expression of pNKCC1 due to
Ptpn20 deficiency is a potential factor contributing to hydrocephalus in both rats and mice. Notably, the interpretation of ventriculomegaly is complicated by a range of potential causative factors. In addition to CP hypersecretion, which leads to lateral ventricular dilatation, other factors can contribute to hydrocephalus, such as cerebral compliance and arachnoid drainage [
50‐
53].
However,
Ptpn20−/− mice did not exhibit severe hydrocephalus after birth like H-Tx rats, and we hypothesize that this may be because
Ptpn20−/− mice did not develop aqueductal stenosis during development or that the overproduction of CSF caused by NKCC1 in the genetic background of C57BL/6J was insufficient to cause severe hydrocephalus. Aqueductal stenosis is associated with abnormal development and dysfunction of the subcommissural organ (SCO) in H-Tx rats. The SCO is a secretory gland of epithelial cells dorsal to the aqueduct of the brain, and in H-Tx rats with congenital hydrocephalus, immunoreactivity of the SCO and glycoproteins of midbrain epithelial cells is decreased starting from embryonic day 16, before ventricular enlargement. Abnormal secretion of glycoprotein is known to interfere with the formation of Reissner’s fiber, obstructing the cerebral aqueduct and leading to ventricular enlargement on embryonic day 17 [
54‐
57]. This finding is consistent with other animal models regarding the link between SCO function and hydrocephalus [
58]. Although the gene for hydrocephalus in H-Tx rats has not been identified, quantitative trait analysis suggests that the loci associated with the hydrocephalus phenotype are present on chromosomes 9, 10, 11, and 17 [
23]. In contrast, the
Ptpn20 gene, located on chromosome 16, may not be involved in the developmental process of SCO.
Furthermore, in a study of L1-deficient mice, significantly enlarged ventricles were observed only in mutant mice backcrossed to the C57BL/6J genetic background, whereas the ventricular system appeared normal when the same mutant mouse strain was backcrossed to the 129 genetic background [
59]. In addition, Cai et al. [
18] found a lower than expected incidence of hydrocephalus and milder disease than in H-Tx rats when they cross-mated hydrocephalic H-Tx rats, and suggested that this result might be due to the effects of gene modifications in different genetic backgrounds. Similarly, the role of the
Ptpn20 gene may be enhanced in the genetic background of H-Tx rats and diminished in C57BL/6J mice. In addition, another study on L1 mutants showed that ventricular enlargement preceded the onset of aqueductal stenosis, and that massive enlargement of the ventricles caused deformation of the brain, which in turn compressed the aqueduct, leading to severe hydrocephalus [
60]. Based on this hypothesis, we consider that although phosphorylation of NKCC1 may produce excess CSF, its production rate might be relatively mild and constant, and so may be insufficient to create an unbalanced pressure differential within the ventricular system that would result in brain deformation. As a result, only moderate hydrocephalus developed.
The interpretation of ventriculomegaly is complicated by a range of potential causative factors. While excessive CP secretion undoubtedly leads to lateral ventricular dilatation, other factors can also contribute to ventriculomegaly, such as cerebral compliance and arachnoid drainage, both acute and chronic. NKCC1 is also expressed in the arachnoid and is likely involved in fluid formation in arachnoid cysts [
61,
62]. In the current study we did not evaluate NKCC1 expression on the arachnoid. However, if the hypothesis about NKCC1 is reasonable, it is also worthwhile to investigate in depth whether
Ptpn20 deletion causes functional alterations in arachnoid NKCC1. This is because it is possible that arachnoid NKCC1-mediated changes in water and osmolarity could affect the absorptive function of arachnoid granules or cause an increase in subarachnoid water production. Whether this is also responsible for the chronic state of hydrocephalus in H-Tx rats also requires further investigation.
As previously mentioned, the direction of NKCC1 transport has been a controversial topic. As proposed by Steffensen et al., CP apical NKCC1 works in a net efflux mode, co-transporting ions and water and directly contributes to the continued production of CSF [
29]. Gregoriades et al. objected that under basal conditions CP apical NKCC1 works in a net instream mode, transporting ions and associated water into the cell, maintaining intracellular Cl− concentrations and volume of cytosolic water required for CSF secretion [
28]. However, Steffensen’s group refuted Gregoriades et al. by suggesting that the results they derived for the direction of NKCC1 endocytic transport may stem from the treatment of the cells prior to the experiment and the contents of the solutions tested during the experiment [
63]. As a further basis for the inward NKCC1 transport pattern, Xu et al. found that CP NKCC1 mediates the clearance of CSF during early postnatal development in mice, and in addition, it was found that overexpression of CP NKCC1 in a model of postnatal obstructive hydrocephalus resulted in a reduction in ventricular enlargement [
30]. It has to be said that this finding provides additional strong evidence for the inward transport of CSF at a particular stage in mice. However, for the current study, this inward transport mechanism is inconsistent with our latest findings, and in the absence of other factors contributing to ventricular enlargement, outward transport of CSF by CP NKCC1 is the most plausible explanation at this stage.
About 100 years ago the CP was thought to be the main site of CSF production, and since then choroid plexus cauterization (CPC) has been developed to reduce CSF production for the treatment of hydrocephalus. In the early twentieth century, CPC alone was used with some success in patients with communicating hydrocephalus, but it was abandoned later due to the high mortality rate and the poor outcome observed [
64‐
66]. With the advance in neurosurgical techniques, CPC has received renewed attention, but is mostly used as an adjunct to endoscopic third ventriculostomy (ETV) in the treatment of hydrocephalus [
67‐
71]. Although it seems that CPC alone has become an obsolete procedure, a question remains if there is still a value of CPC in the treatment of any specific form of hydrocephalus. ETV + CPC is reported to be effective in pediatric hydrocephalus. Many of the reported patients were considered successful because intracranial pressure was controlled during the acute phase and shunt placement was not necessary, however detailed evaluation of psychomotor development was not performed. Therefore currently, the indicated cases are considered to be children with extremely high rates of shunt dysfunction and patients with hydranencephaly, in whom future psychomotor development is not expected [
72]. The rationale for the success of ETV + CPC is the reduction of intraventricular pulsatility due to the CP, in addition to the reduction of CP CSF production [
73]. The confirmation of this consideration requires accumulation of basic research on intraventricular pulsatility [
74]. However, we must note that other potential functions of the CP besides CSF secretion should not be neglected. In particular, NKCC1, the focus of attention in this study, has been shown to play a key role in regulating cell volume and maintaining ionic homeostasis in the CNS [
75,
76]. Furthermore, the CSF produced from the CP is also thought to play additional role. The CP releases various hormones, cytokines, and other proteins as CSF, which act on the thalamus, hypothalamus, periventricular organs (pineal gland), and some neuronal nuclei (suprachiasmatic nucleus).In that way it plays an important role in maintaining homeostasis, including autonomic daily rhythm, emotion, and stress response [
77,
78]. Furthermore, the role of amyloid-β elimination in the CSF has attracted attention in relation to the pathogenesis of Alzheimer’s disease, and it has been reported that amyloid-β metabolizing enzyme system proteins are released from the CP in conjunction with the circadian rhythm and that amyloid-β is absorbed directly into the blood from the CP [
79].
This study had several limitations. First, since accurately determining whether H-Tx rats at P18 have hydrocephalus based solely on appearance is difficult, hydrocephalic H-Tx rats may have been present among the current non-hydrocephalus samples, masking other genetic risk factors. Future studies will be needed, with larger sample sizes and improved sensitivity of the results, to identify other possible causative genes. Eventually, paired or multiple knockout models can be established to further clarify the molecular mechanisms underlying hydrocephalus. Second, since reliable methods to accurately measure CSF production remain lacking, it is not possible to test the hypothesis of CSF overproduction in Ptpn20−/− mice. Third, we focused primarily on changes in membrane transport proteins located on the apical side of the CP epithelium, however, Na+-driven chloride bicarbonate exchanger located on the basolateral side of the epithelium also plays a role in CSF production. Whether Ptpn20 deletion leads to any functional changes on the basolateral side warrants further investigation. In addition, the lack of imaging tools in our study makes the degree of ventricular dilatation impossible to determine directly on live animals. Isolated or dehydrated brain tissue may result in varying degrees of morphological and structural alteration of the ventricles, which may in term have affected the accuracy of our results.