Molecular Psychiatry (2021) 26:6411–6426 https://doi.org/10.1038/s41380-021-01127-9 ARTICLE Kidins220 deficiency causes ventriculomegaly via SNX27-retromer- dependent AQP4 degradation Ana del Puerto1,2,16 Julia Pose-Utrilla1,2 Ana Simón-García1,2 Celia López-Menéndez1,2 Antonio J. Jiménez3,4 ● ● ● ● ● Eva Porlan5,6,7 Luis S. M. Pajuelo1,2 Guillermo Cano-García1 Beatriz Martí-Prado8 ● ● ● ● Álvaro Sebastián-Serrano1,2,17 Marina P. Sánchez-Carralero1,2 Fabrizia Cesca9 Giampietro Schiavo 10,11 ● ● ● ● Isidro Ferrer 2,12,13,14 Isabel Fariñas2,8 Miguel R. Campanero 5,15 Teresa Iglesias 1,2 ● ● ● Received: 14 July 2020 / Revised: 29 March 2021 / Accepted: 14 April 2021 / Published online: 17 May 2021 © The Author(s) 2021. This article is published with open access Abstract Several psychiatric, neurologic and neurodegenerative disorders present increased brain ventricles volume, being hydrocephalus the disease with the major manifestation of ventriculomegaly caused by the accumulation of high amounts of cerebrospinal fluid (CSF). The molecules and pathomechanisms underlying cerebral ventricular enlargement are widely 1234567890();,: 1234567890();,: unknown. Kinase D interacting substrate of 220 kDa (KIDINS220) gene has been recently associated with schizophrenia and with a novel syndrome characterized by spastic paraplegia, intellectual disability, nystagmus and obesity (SINO syndrome), diseases frequently occurring with ventriculomegaly. Here we show that Kidins220, a transmembrane protein effector of various key neuronal signalling pathways, is a critical regulator of CSF homeostasis. We observe that both KIDINS220 and the water channel aquaporin-4 (AQP4) are markedly downregulated at the ventricular ependymal lining of idiopathic normal pressure hydrocephalus (iNPH) patients. We also find that Kidins220 deficient mice develop ventriculomegaly accompanied by water dyshomeostasis and loss of AQP4 in the brain ventricular ependymal layer and astrocytes. Kidins220 is a known cargo of the SNX27-retromer, a complex that redirects endocytosed plasma membrane proteins (cargos) back to the cell surface, thus avoiding their targeting to lysosomes for degradation. Mechanistically, we show that AQP4 is a novel cargo of the SNX27-retromer and that Kidins220 deficiency promotes a striking and unexpected downregulation of the SNX27- retromer that results in AQP4 lysosomal degradation. Accordingly, SNX27 silencing decreases AQP4 levels in wild-type astrocytes whereas SNX27 overexpression restores AQP4 content in Kidins220 deficient astrocytes. Together our data suggest that the KIDINS220-SNX27-retromer-AQP4 pathway is involved in human ventriculomegaly and open novel therapeutic perspectives. Introduction developmental factors, viral infections, co-morbidities and aging [1–4]. Although hydrocephalus is the pathology that Hydrocephalus is a disease associated with cognitive reaches maximal brain ventricles enlargement, ven- impairment where cerebrospinal fluid (CSF) accumulates triculomegaly is also a co-morbidity factor in several psy- provoking brain ventricular enlargement or ven- chiatric and neurodegenerative disorders, including triculomegaly. There are multiple hydrocephalus forms, schizophrenia [5, 6], Parkinson´s disease [7] and Alzheimer including foetal, neonatal, paediatric and adult onset pre- ´s disease (AD) [8]. Dementia is one of the characteristic sentation, and diverse aetiologies such as genetic and symptoms of chronic hydrocephalus in adults, where the major form is idiopathic normal pressure hydrocephalus (iNPH), a disease recently proposed to be neurodegenera- Supplementary information The online version contains tive with dysfunctions shared with AD [9–11]. supplementary material available at https://doi.org/10.1038/s41380- There is an urgent need to identify and characterize the 021-01127-9. pathogenic mechanisms governing CSF dynamics and * Teresa Iglesias underlying brain ventricular enlargement in different neu- tiglesias@iib.uam.es ropathologies to uncover novel therapeutic strategies. Very Extended author information available on the last page of the article little is known about the molecules involved in the cellular 6412 A. del Puerto et al. processes that contribute to ventriculomegaly, although the SNX27-retromer components, which in turn controls AQP4 water selective channel aquaporin-4 (AQP4) appears as a turnover. critical player. AQP4 expression in the brain is prominent in the basolateral membranes of ependymocytes lining cere- bral ventricles and in the end-feet of astrocytes contacting Materials and methods brain blood vessels and the pia membrane [12, 13]. Aqp4−/− mice show sporadic brain ventricular enlargement, Human brain samples accelerated progression of induced hydrocephalus, and increased basal brain water accumulation [14–18], sug- Brain samples from iNPH patients (n = 6) and non- gesting that AQP4 is critical for controlling whole-brain normotensive hydrocephalus patients (n = 3), and control water homeostasis (reviewed in [11, 19–21]). While studies subjects (n = 8) (for data see Supplementary Table 1) were in iNPH brain revealed loss of astrocytic perivascular AQP4 obtained from the Institute of Neuropathology Brain Bank [22, 23], the underlying mechanisms of AQP4 down- (IDIBELL, Hospitalet de Llobregat, Spain), tissue Biobank regulation remain unknown. It is also unknown whether from Hospital Clinic de Barcelona and Institute d´Investi- AQP4 content at the ependymal ventricular lining is altered gacions Biomediques August Pi I Sunyer (HCB-IDIBAPS in iNPH. Biobank), and the Biobank Banco de Tejidos CIEN (PT17/ The recycling of endocytosed transmembrane proteins 0015/0014), integrated into the Spanish National Biobanks (cargos) from early endosomes back to the plasma mem- Network following Spanish legislation and local Ethics brane might be involved in ventriculomegaly development, Committee guidelines and processed following standard as suggested by the presence of severe postnatal hydro- operating procedures as described [42]. Briefly, one brain cephalus in mice lacking sorting nexin 27 (Snx27) [24]. hemisphere was fixed by immersion in 4% buffered for- SNX27 associated with vacuolar protein sorting 35 malin for 3 weeks. Samples of selected regions of the brain (VPS35), VPS29 and VPS26 forms the so-called SNX27- were embedded in paraffin and 15-μm sections were stained retromer, a complex specifically involved in the endosomal- for neuropathological studies using immunohistochemistry to-plasma membrane recycling of selected cargos, that are methods (see ‘Immunohistochemistry of mouse and human degraded at lysosomes upon SNX27 loss [25, 26]. Impor- brain samples’ section and details for antibodies and dilu- tantly, retromer impairment also contributes to aberrant tions used in Supplementary Table 2). All participants gave endocytic trafficking in various neurodegenerative their written consent, and the study was approved by the diseases [27]. local Ethics and Scientific Committees following the ethical Kinase D interacting substrate of 220 kDa (Kidins220 standards recommended by the Helsinki Declaration. [28]), also known as ankyrin-repeat rich membrane span- ning (ARMS [29]), is a SNX27-retromer cargo targeted to Experimental animals lysosomal degradation in the absence of this PDZ protein [25]. Kidins220 is an effector of several signalling path- Kidins220f/f mice and Kidins220−/− mice in C57BL/6 J ways, including those downstream neurotrophin and gluta- background had been previously generated in G. Schiavo mate N-methyl-D-aspartate (NMDA) receptors, and laboratory [43]. Male and female 2-month-old Kidins220+/f, modulates neuronal differentiation and synaptic activity [30, Kidins220f/f or WT littermates were employed in indepen- 31]. Neuronal survival depends on Kidins220 and the dent experiments for comparison purposes. Genotyping was expression of this molecule is altered in neurological and performed by PCR using specific pairs of primers [43]. All neurodegenerative disorders, including cerebral ischemia, animals were produced and housed at the animal care AD and Huntington´s disease [32–36]. In addition, rare facility at Instituto de Investigaciones Biomédicas ‘Alberto novel missense variants in KIDINS220 gene have been Sols’ (IIBM, CSIC-UAM, Madrid, Spain) and maintained associated with schizophrenia [37–39], a psychiatric dis- under 12/12 h light-dark cycle and with access to food and order with a strong linkage to ventriculomegaly [5, 6]. More water ad libitum in a temperature-controlled environment. recently, KIDINS220 nonsense and loss-of-function variants Overall mouse health was assessed by daily inspection for have been associated with SINO syndrome, a novel rare signs of discomfort, weight loss or changes in behaviour, autosomal disease characterized by spastic paraplegia, mobility and feeding or drinking habits. Procedures invol- intellectual disability, nystagmus and obesity, signs that ving animals had been approved by Institutional (IIBM and concur with different degrees of ventriculomegaly [40, 41]. CSIC) and local Ethical Committees, and were conformed Here we identify Kidins220 as a key determinant in the to the appropriate national legislations (RD 53/2013) and control of brain water homeostasis, ventricular enlargement the guidelines of the European Commission for the and hydrocephalus pathology by molecular mechanisms accommodation and care of laboratory animals (revised in that involve the unexpected expression regulation of Appendix A of the Council of Europe Convention ETS123). Kidins220 deficiency causes ventriculomegaly via SNX27-retromer-dependent AQP4 degradation 6413 Magnetic resonance imaging; Scanning electron micro- elevated in LVs and TV from Kidins220f/f mice relative to scopy; RNA isolation, reverse transcription and quantitative WT animals suggesting an imbalance in the circulation of real-time PCR; Plasmids; Cell culture, treatment and CSF (Fig. 1E). transfection of primary cortical astrocytes and The generation of Kidins220f/f animals involved HEK293T cells; Lentiviral production and transduction of genetic manipulations that integrated Kidins220 cDNA astrocytic cultures; Water permeability and cell-volume within exon 16 of the mouse gene (see details in reference measurements in cultured astrocytes; Commercial anti- [43] and scheme in Supplementary Fig. S1c). We hypo- bodies; Preparation of protein extracts, immunoprecipitation thesized that this strategy might potentially cause a defi- and immunoblot analysis; Immunofluorescence of cultured ciency in Kidins220 expression. To test this hypothesis, astrocytes; Immunohistochemistry of mouse and human we analysed Kidins220 levels in various brain regions. brain samples; Image acquisition and Quantitative and sta- Peroxidase-based immunostaining of brain slices detected tistical analysis were performed as described in the Exten- lower Kidins220 signal intensity in Kidins220f/f compared ded Data Supplementary Material and Methods section. to WT mice (Supplementary Fig. S1d). Immunoblot Pairs of primers for quantitative real-time PCR and sh- analysis showed that the cortex and cerebellum of oligonucleotides cloned in lentiviral vectors are detailed in Kidins220f/f mice exhibited decreased amounts of Supplementary Table 3. Kidins220 compared to WT animals (Fig. 1F). The cortex of Kidins220+/f, but not the cerebellum, also presented substantially reduced levels of this protein, indicating that Results some brain regions are potentially more susceptible to the interference induced by the transgenic modification. Kidins220 deficient mice present ventriculomegaly Importantly, we found an inverse correlation between Kidins220 levels and LVs and TV ventricular volume in Because Kidins220 complete ablation produces perinatal WT, Kidins220+/f, and Kidins220f/f mice (Fig. 1G), sug- lethality, we created Kidins220-floxed mice to generate gesting that Kidins220 deficiency and brain ventricle mouse models with a conditional deletion in different tis- enlargement are strongly linked. In addition to brain, sues and cells to study this molecule function in adulthood extracts obtained from several tissues showed decreased [43, 44]. We noticed that homozygous Kidins220-floxed levels of Kidins220 to various degrees between WT, mice (Kidins220f/f) were not present in the offspring of Kidins220+/f, and Kidins220f/f mice (Supplementary heterozygous (Kidins220+/f) crosses at Mendelian ratio, Fig. S1e). Kidins220f/f(s) mice were sacrificed immedi- with a greater lethality at weaning (Fig. 1A, B). Two- ately after MRI, and their brain tissue was not further month-old Kidins220f/f animals did not show differences in analysed due to the high degree of damage (see MRI of a body weight but presented an increment in brain weight highly severe phenotype in Supplementary Fig. 2). compared to their littermates (Supplementary Fig. S1a). It Together, these data indicate that Kidins220f/f mice con- was also visually apparent that a low percentage of stitute a hypomorphic model with marked deficiency of Kidins220f/f mice presented encephalomegaly (17.4%), Kidins220 in several tissues, supporting the notion that referred to as ‘severe’ (S) (Kidins220f/f(S)), suggesting pos- the decrease of Kidins220 brain levels is linked to ven- sible brain ventricular enlargement and hydrocephalus triculomegaly and hydrocephalus development. (Fig. 1C). MRI analysis confirmed the development of ventriculomegaly to various extents, from mild to severe, Loss of Kidins220 and AQP4 at the ventricular both in male and female Kidins220f/f mice (Fig. 1D and ependyma of Kidins220 hydrocephalus mice and Supplementary Fig. S1b). Volumetric quantification from human iNPH patients T2-weighted (T2-W) images rendered increments in the volume of lateral ventricles (LVs) and third ventricle (TV) Due to the importance of the ventricular ependyma in brain of male and female Kidins220f/f animals compared to wild water homeostasis and congenital hydrocephalus [3], we type (WT) or Kidins220+/f littermates (Fig. 1D and Sup- examined the presence of Kidins220 in this specialised plementary Fig. S1b). While no substantial changes were tissue by using a recently generated antibody [45]. Its detected in the fourth ventricle (FV), the aqueduct of Syl- specificity in immunofluorescence and immunoblot ana- vius (SA) volume was decreased in Kidins220f/f males and lyses was further validated by antibody neutralization with females (Fig. 1D and Supplementary Fig. S1b). Using the immunizing peptide and by using brain tissue from diffusion-weighted imaging MRI, we also calculated Kidins220−/− mice [43, 44] (Supplementary Fig. S3a–c). apparent diffusion coefficient (ADC) values as a measure- Confocal microscopy analysis showed higher ependymal ment of the magnitude of diffusion of water molecules in Kidins220 expression in the lateral ventricles of WT mice the ventricles and SA of male mice. ADC values were compared to Kidins220f/f mice (Fig. 2A). Accordingly, 6414 A. del Puerto et al. Kidins220 levels were significantly lowered in extracts hydrocephalus present developmental defects in ependy- prepared from the lateral periventricular area (PVA) of mocytes differentiation, ciliogenesis and maturation, or Kidins220f/f animals (Fig. 2B). denudation of the ependymal barrier [3, 48]. We, therefore, The ependymal barrier lines cerebral ventricles and is examined the morphological features of the ependymal cells formed by multiciliated ependymocytes that beat ventricular in LVs of 2-month-old Kidins220f/f mice by scanning CSF synchronously to drive its unidirectional flow across electron microscopy, finding no apparent macroscopic ventricles [46, 47]. Different types of congenital changes in their cilia compared to WT animals Kidins220 deficiency causes ventriculomegaly via SNX27-retromer-dependent AQP4 degradation 6415 Fig. 1 Ventriculomegaly in Kidins220 hypomorphic mice. A Total marks mature ciliated ependymal cells [50, 51] were number (n = 232) and percentage of WT, Kidins220f/+ and Kidins220f/ observed (Supplementary Fig. S4d, e). f mice alive at weaning vs their expected Mendelian ratio. B Percentage Next, we examined potential decreases of KIDINS220 of WT, Kidins220f/+ and Kidins220f/f mice death at weaning showing severe hydrocephalic phenotype (n = 232). C Representative images and AQP4 at the ventricular ependyma in iNPH patients of 2-month-old WT, Kidins220f/f, and severely hydrocephalic compared to control donors (Supplementary Table 1). We Kidins220f/f (Kidins220f/f(S)) mice. Right: Zoom detail of the head. found that ependymal cells were positive for D Representative in vivo T2-weighted (T2-W) MRI coronal images KIDINS220 signal in brain necropsies from control indi- showing lateral and third ventricles (LVs + TV; left panels), aqueduct of Sylvius (SA; central panels—dotted line) and fourth ventricle (FV; viduals and that this staining diminished significantly in right panels—dotted line) of 2-month-old WT, Kidins220f/+, samples from equivalent brain regions from iNPH patients Kidins220f/f and Kidins220f/f(S) male mice. Scale bar, 1 mm. Quantifi- (Fig. 2E and Supplementary Fig. S5a–S6). Samples from cation of ventricular and aqueduct volume (mm3) in 17 WT (circles), 6 adult non-normotensive hydrocephalus (Hyd) patients Kidins220f/+(squares) and 14 Kidins220f/f (triangles) mice. LVs and TV values from the same group of 17 WT (circles) animals have been similarly presented a fainter KIDINS220 ependymal barrier represented in a separate graph with those of three severe Kidins220f/f signal than aged-matched controls (Fig. 2E and Supple- (S) (diamonds) mice. E Apparent Diffusion Coefficient (ADC) repre- mentary Fig. S5a–S6). By contrast, S100β ependymal zone sentative coronal images showing LVs and TV (left panels), SA immunolabelling was unaltered in hydrocephalic versus (central panels—dotted line) and FV (right panels—dotted line) of 2- month-old WT, Kidins220f/f and Kidins220f/f(S) male animals. The control human samples (Supplementary Fig. S5b). Notably, pseudo-colour scale indicates maximum and minimum ADC values in AQP4 immunofluorescence analysis also showed a sharp μm2/s. Scale bar, 1 mm. Quantification of ADC values (μm2/s) at the signal drop in samples from iNPH patients (Fig. 2F and ventricles and aqueduct in WT (circles; LVs, TV and SA, n = 16; FV, Supplementary Fig. S5c). This is the first evidence of an n = 12), Kidins220f/f (triangles; LVs, TV and SA, n = 12; FV, n = 8) and Kidins220f/f(S) (n = 3, diamonds) mice. F Representative ependymal deficit of AQP4 and KIDINS220 in iNPH Kidins220 and α-tubulin (loading control) immunoblots of cortex (Cx) patients, supporting the involvement of the ependymal and cerebellum (Cb) lysates from WT, Kidins220f/+ and Kidins220f/f barrier and these two molecules in the development of adult mice; each lane represents one animal. Kidins220 levels are repre- chronic hydrocephalus. sented in arbitrary units after normalization with α-tubulin in Kidins220f/+ (Cx, n = 6; Cb, n = 4) and Kidins220f/f (Cx, n = 6; Cb, n = 5) relative to WT (Cx, n = 7, Cb, n = 4) mice. G Correlation Downregulation of AQP4 in astrocytes from between Kidins220 relative levels in the cortex (Cx, left panel) and Kidins220 hypomorphic mice cerebellum (Cb, right panel) represented in (F) and LVs and TV volumes quantified in T2-W MRI images depicted in (D), established by linear regression analysis (Cx, R2: 0.5954; Cb, R2: 0.6231). D, E, F To gain mechanistic insight into the regulation of S100β Data are mean ± s.e.m.; each data point denotes an individual mouse; and AQP4 expression by Kidins220, we analysed primary NS not significant; ***P < 0.001, by one-way ANOVA (D, F) and cultures of astrocytes because these cells contain high levels two-tailed unpaired Student’s t test (E). of these two molecules and, in contrast to ependymocytes, can be readily cultured, maintained and transfected/trans- (Supplementary Fig. S4a). In addition, organization and duced in vitro. RTq-PCR showed no significant decreases integrity of the ependymal barrier in Kidins220 deficient of Aqp4 or S100β mRNA in cultured astrocytes or in tissue mice was largely unaltered, as assessed by whole-mount from the PVA obtained from Kidins220f/f animals (Sup- preparation immunostaining and localization of β-catenin at plementary Fig. S7). However, we observed a substantial the intercellular contacts and γ-tubulin at the base of the decrease in Kidins220 and AQP4 in cultured astrocytes cilia (Supplementary Fig. S4b). obtained from Kidins220f/f mice (Fig. 3A, B), suggesting Although the ultrastructural appearance and planar that post-transcriptional mechanisms underlie AQP4 polarity of the ependymal barrier in Kidins220f/f animals downregulation. Of note, we could not detect S100β were normal, ependymal cells lacked expression of S100 protein. calcium-binding protein β (S100β), a protein expressed in Astrocytes transiently increase their volume upon ependymocytes and astrocytes [49] (Fig. 2C). We also hypotonic stress and this process is regulated by AQP4 [52]. examined levels of AQP4 in this area as this channel is also To determine whether AQP4 function is impaired in expressed in ependymal cells and plays a critical role in Kidins220f/f astrocytes, we treated astrocytes with a hypo- brain water homeostasis and ependyma function [19, 20]. tonic stimulus and measured their volume by using the Kidins220f/f mice exhibited a striking downregulation of calcein-quenching assay [52]. We found that Kidins220f/f AQP4 in ependymal cells of LVs compared to WT animals, astrocytes did not increase their volume as much as in immunofluorescence analysis of both brain sections and WT cells upon hypotonic stress, and showed a significant whole-mount preparations (Fig. 2D and Supplementary delay in bringing their volume back to baseline (Fig. 3C–E). Fig. S4c). In contrast, no substantial changes in GFAP, a These results strongly suggest that AQP4 function is greatly marker for neural stem cells and astrocytes, or FoxJ1 that hampered in Kidins220 deficient astrocytes. 6416 A. del Puerto et al. Fig. 2 Decreased AQP4 and Kidins220 levels at the ependymal (D; n = 6 WT and n = 7 Kidins220f/f mice) staining at the ependymal barrier of Kidins220 hypomorphic mice and human iNPH. A barrier of Kidins220f/f mice relative to WT mice. E, F Representative Representative confocal microscopy images of Kidins220 (green) and confocal microscopy images of KIDINS220 (E) and AQP4 DAPI (blue) staining at the ependymal barrier of brain lateral ven- (F) immunostaining (green) and DAPI nuclear labelling (blue) at the tricles (LV) of 2-month-old WT and Kidins220f/f mice, and (right ependymal barrier (dotted lines) of necropsies from iNPH patients and panel) quantification of Kidins220 immunostaining at the ependymal control subjects. Scale bars, 30 μm. Quantification of ependymal barrier of 4 Kidins220f/f mice relative to 3 WT mice. Scale bar, 50 μm. barrier immunostaining in human postmortem samples from hydro- B Kidins220 and α-tubulin (loading control) immunoblot analysis of cephalic patients relative to control individuals: KIDINS220 (E) the periventricular area (PVA) of WT and Kidins220f/f mice (n = 4 per [iNPH n = 6; Hyd n = 3; control n = 8] and AQP4 (F) [iNPH n = 3; genotype); each lane represents one mouse. Kidins220 levels, after control n = 3]. For quantifications, three sections per animal and normalization with those for α-tubulin, were represented relative to genotype and 3 ROIs per section (A, C, D), and 9 randomly selected WT mice (bottom panel). C, D Representative confocal microscopy ROIs per human section (E, F) were used; data are mean ± s.e.m.; each images showing S100β (C), AQP4 (D) immunostaining (green) and data point denotes an individual mouse/subject. (A–F) *0.01 < P < DAPI nuclear labelling (blue) of lateral ventricles (LV) ependymal 0.05, **0.001 < P < 0.01, ***P < 0.001; by two-tailed unpaired Stu- barrier from 2-month-old WT and Kidins220f/f mice. Scale bar, 50 μm. dent’s t test. Quantification of S100β (C; n = 7 animals per genotype) and AQP4 We then investigated the potential post-transcriptional increase in AQP4 levels both in WT and Kidins220f/f astro- mechanisms involved in AQP4 downregulation. As this water cytes, as determined by immunofluorescence analysis and channel can be degraded by the lysosome [53], we examined immunoblot quantification (Fig. 3F, G). Together, our data the effect of the vATPase inhibitor bafilomycin A1 on AQP4 strongly suggest that Kidins220 deficiency increases AQP4 levels. We found that bafilomycin A1 induced a significant lysosomal degradation. Kidins220 deficiency causes ventriculomegaly via SNX27-retromer-dependent AQP4 degradation 6417 Joint regulation of the SNX27-retromer and its could be associated with SNX27-retromer dysfunction. novel cargo AQP4 by Kidins220 Staining of VPS35, the key SNX27-retromer component [26], and SNX27 at the ependyma of LVs revealed a pro- Given that Kidins220 endosomal recycling to the cell sur- minent decrease of their levels in Kidins220f/f (Fig. 4A). face is controlled by the SNX27-retromer complex [25], and Accordingly, astrocytes from Kidins220f/f mice also showed that Snx27 genetic inactivation causes severe postnatal a severe downregulation of SNX27 and VPS35 (Fig. 4B). hydrocephalus in mice [24], we investigated whether We then checked whether Kidins220 could be directly hydrocephalus development in Kidins220 deficient animals involved in the control of SNX27 and VPS35 by 6418 A. del Puerto et al. Fig. 3 AQP4 downregulation in Kidins220-deficient astrocytes is [54] remained constant as well, both in Kidins220f/f and rescued by inhibition of lysosomal degradation. A Representative Kidins220 silenced astrocytes (Supplementary Fig. S10), confocal microscopy images of AQP4 (green), GFAP (red) and supporting the specificity of Kidins220 deficiency on DAPI (blue) staining of primary cortical astrocytes dissected from WT (n = 3) and Kidins220f/f (n = 3) mice. Scale bar, 50 μm. B (top retromer regulation. panels) Kidins220, AQP4 and GFAP (loading control) immunoblot The loss of SNX27 and VPS35 in Kidins220f/f deficient analyses of WT and Kidins220f/f cortical astrocytes; each lane astrocytes, contrary to that of AQP4, was not rescued by represents a culture obtained from one animal. (bottom panels) bafilomycin A1 treatment (Supplementary Fig. S11a). In Kidins220 and AQP4 levels in Kidins220f/f astrocytes expressed in arbitrary units after normalization with GFAP and relative to WT addition, Snx27 and Vps35 mRNA levels did not decrease in astrocytes (n = 4 of each genotype). C–E Primary astrocytes from PVA tissue or cultured astrocytes from Kidins220f/f mice, WT (n = 3) and Kidins220f/f (n = 3) mice were exposed to a hypo- while Vps35 transcripts only increased slightly in the glial tonic medium (ΔOsm = 150 and 75 mOsm) (n = 3 independent cells in vitro (Supplementary Fig. S11b, c), suggesting that experiments); C Calcein-quenching measurement of osmotically induced volume changes in these cells (n = 6 of each genotype) post-transcriptional regulation mechanisms distinct from during 300 s after a basal recording of 60 s. D Quantitative analysis lysosomal degradation mediate Kidins220-dependent of the initial volume increase osmotically induced in 7 WT and 14 SNX27 and VPS35 downregulation. Kidins220f/f astrocytes from calcein-quenching measurement data; Kidins220 is considered as a canonical cargo for the ordinate values represent mean volume changes at 65 s and 170 s (Ft) of hypotonic stress relative to WT baseline levels (Ft0). SNX27-retromer and its type I postsynaptic density-95/ E Representation of the final volume osmotically induced changes in discs large/zonula occludens-1 (PDZ)-binding motif or 8 WT and 12 Kidins220f/f astrocytes from calcein-quenching mea- ligand (PDZ-L) has been classified as a bona fide ligand surement data; ordinate values represent mean volume changes for SNX27 PDZ domain [25, 55]. Notably, AQP4 bears a incurred between the maximum volume at 170 s (Ft170) and the one registered at 300 s, at the end of the recording period (Ft300) to PDZ-L that is fundamental for its localization at the examine recovery capacity. See ‘Materials and Methods’ for quan- plasma membrane in astrocytes [56, 57]. AQP4 PDZ-L is tification details. Regulatory volume decreases (RVD) indicative of evolutionarily conserved (Fig. 4D) and highly similar to volume recovery are represented by negative values. F Repre- that of Kidins220 and other known SNX27 cargos sentative confocal microscopy images of AQP4 (green), GFAP (red) and DAPI (blue) staining of cultured astrocytes from WT and (Fig. 4E), suggesting that AQP4 could be a cargo for the Kidins220f/f mice incubated with vehicle (Control) or bafilomycin SNX27-retromer. In this scenario, it was feasible that A1 for 16 h. Scale bar; 50 μm. Quantification of AQP4 immunos- SNX27 downregulation in Kidins220f/f mice could lead to taining in GFAP-positive astrocytes under the indicated conditions; lysosomal processing of AQP4 in astrocytes, as it occurs each data point represents a single astrocyte (n = 60–80 astrocytes per condition; n = 3 independent experiments). G Representative with Kidins220 after SNX27 loss in tumour cells [25]. Kidins220, AQP4 and GFAP (loading control) immunoblot analysis However, alignment of AQP4 PDZ-L sequence with those of cultured cortical astrocytes incubated with vehicle (−) or bafi- from consensus SNX27-retromer cargos, including lomycin A1 (Baf, +) obtained from WT and Kidins220f/f mice. Kidins220, showed an apparently important mismatch at Quantification of Kidins220 and AQP4 levels after normalization with GFAP signal (n = 5 independent experiments) and relative to the acidic clamp for SNX27 interaction [55] (Fig. 4E). To WT cells is shown (bottom panels). Quantification data are shown as address whether AQP4 could be a cargo for the SNX27- mean ± s.e.m.; NS not significant; **0.001 < P < 0.01, ***P < 0.001; retromer, we first analysed the cellular sub-localization of by two-tailed unpaired Student’s t test in (B, D, E) and two-way AQP4 and some components of this complex in cultured ANOVA in (F, G). astrocytes from WT mice. Confocal microscopy imaging showed co-localization of AQP4 not only with SNX27 but transducing cultured astrocytes from WT mice with lenti- also with VPS35 (Fig. 4F) and Kidins220 (Supplementary viral particles encoding Kidins220 shRNA (shKidins220, Fig. S12). Moreover, AQP4 was present in VPS35- referred as shK) or a control shRNA (shC). Strikingly, immunocomplexes obtained from murine cerebral cortex Kidins220 knockdown strongly reduced SNX27 protein and cultured primary astrocytes (Fig. 4G). We could not levels without altering those of VPS35 (Fig. 4C). Consistent use SNX27 antibodies for co-immunoprecipitation with results obtained in Kidins220f/f astrocytes, experiments due to the difficulty of distinguishing the Kidins220 silencing similarly decreased AQP4 content SNX27 band from that of the immunoglobulin heavy (Fig. 4C). chain in the immunoblot analysis. In addition, we exam- Subcellular distribution of SNX27 was apparently ined AQP4 association to the SNX27-retromer by trans- preserved in Kidins220f/f astrocytes (Supplementary fecting HEK293T cells with plasmids encoding AQP4, Fig. S8). In addition, analysis of the early endosomal full-length SNX27 (SNX27-FL) or a mutant lacking its marker Rab5 revealed no substantial differences in PDZ-domain (SNX27-ΔPDZ), all fused to GFP. Analysis Kidins220f/f astrocytic endosomal network (Supplemen- of endogenous VPS35-immunoprecipitates revealed they tary Fig. S9). Together these data suggest that the early contained AQP4 (Fig. 4H). Co-expression of SNX27-FL endosomal compartment is not globally affected by is compatible with the association of AQP4 with VPS35- Kidins220 deficiency. The retromer-independent SNX17 immunocomplexes but co-expression of SNX27-ΔPDZ Kidins220 deficiency causes ventriculomegaly via SNX27-retromer-dependent AQP4 degradation 6419 disrupted this binding (Fig. 4H). Complementary experi- Kidins220-dependent AQP4 downregulation is ments showed that mCherry-SNX27 was present in GFP- mediated by SNX27 AQP4 immunoprecipitates (Supplementary Fig. S13). Together, these data indicate that the PDZ domain of To investigate whether SNX27 downregulation mediates SNX27 is necessary for the association of AQP4 with the decrease of AQP4 levels caused by Kidins220 VPS35 and that AQP4 is a novel cargo for the SNX27- deficiency, we generated lentiviral particles for retromer. SNX27 silencing in WT cultured astrocytes (Supplementary 6420 A. del Puerto et al. Fig. 4 Joint regulation of the SNX27-retromer and its novel cargo endosomes (Supplementary Fig. S15). Together our data AQP4 by Kidins220. A Representative confocal microscopy images support the notion that Kidins220 deficiency causes SNX27 of AQP4 (green), SNX27 (red), VPS35 (magenta) and DAPI (blue) and VPS35 loss that in turn mediates AQP4 lysosomal staining of the ependymal barrier from lateral ventricles (LV) of WT (n = 5) and Kidins220f/f (n = 7) mice; and (right panel) quantification degradation and water dyshomeostasis (see scheme in of SNX27 and VPS35 staining of these samples (three sections per Fig. 5D). animal). Scale bar, 50 μm. B Kidins220, SNX27, VPS35 and GFAP (loading control) immunoblot analyses of WT and Kidins220f/f cortical astrocytes; each lane represents a culture obtained from one animal. Graphs represent the quantification of SNX27 and VPS35 after nor- Discussion malization with GFAP. C Kidins220, SNX27, AQP4, VPS35 and α- tubulin (loading control) immunoblot analyses of cultured cortical Here we have uncovered Kidins220-SNX27-retromer as a astrocytes obtained from WT mice transduced with lentivirus encoding regulatory pathway for brain AQP4 expression and its shControl (shC) or shKidins220 (shK). Quantification of the indicated protein levels after normalization with α-tubulin: Kidins220 (n = 3), involvement in brain ventricular enlargement and hydro- SNX27 (n = 3), AQP4 (n = 4) and VPS35 (n = 3) independent cephalus. We have identified a surprising link between experiments. D Sequence alignment of the last 9 amino acids within SNX27-retromer down-regulation and AQP4 lysosomal AQP4 C-terminal end of different organisms and their access number. degradation induced by Kidins220 deficiency, and dis- E Sequence alignment of the last 8 amino acids of human AQP4 and the indicated known cargos of SNX27. PDZ-binding motif triplet covered that KIDINS220 and AQP4 expression is sharply residues (PDZ-L) are shown in grey boxes and critical amino acids at 0 decreased at the ependymal barrier in iNPH patients. We and −2 position are labelled in blue. Negatively charged amino acids also show that hypomorphic Kidins220f/f mice are a novel (aspartic and glutamic acid or phosphorylated serine, threonine resi- mouse model of hydrocephalus that display ven- dues) at −3 and −5 position enhancing SNX27 binding by forming an acidic clamp are shown in magenta. Note that in −3 position there is a triculomegaly and a marked AQP4 downregulation in brain leucine within AQP4 sequence or isoleucine within PTEN, and in the astrocytes and in the ependymal cells lining lateral ven- −5 position there is a glycine within GLUT1 and a cysteine within tricles, without apparent ependymocytes structural organi- OCLUDIN (labelled in red). F Representative confocal microscopy zation and ciliogenesis impairment. images showing AQP4 (green), SNX27 (red), VPS35 (magenta) and DAPI (blue) staining in cultured WT astrocytes (n = 3 independent The Kidins220 hypomorphic mouse model shows var- experiments). Scale bar, 50 μm. Zoom images from boxed regions are ious degrees of Kidins220 downregulation that positively also shown. Scale bar, 5 μm. G Representative immunoprecipitation correlate with ventriculomegaly penetrance. The decrease in (IP) analysis of VPS35 in lysates from cortical cultured astrocytes or the number of Kidins220f/f mice at weaning might be cortex (Cx) from WT mice and immunoblotting with VPS35 or AQP4 antibodies. Immunoprecipitation with a non-related immunoglobulin attributed to embryonic or perinatal hydrocephalus- (IgG) was used as control. Total lysates (TL) were run in parallel (n = associated death of mice with high downregulation of 3 independent experiments). Low and high exposure images of AQP4 Kidins220. Supporting this hypothesis, constitutive Kidins220−/− mice die at birth and their brains present a in cortical lysates are included. H VSP35 was immunoprecipitated in lysates from HEK293T cells transfected with GFP or GFP-AQP4 and GFP-SNX27 full-length (SNX27-FL) or GFP SNX27 lacking the PDZ pronounced ventricular enlargement [43]. The majority of domain (SNX27-ΔPDZ). Representative immunoblot analysis of total Kidins220f/f born mice reached adulthood without showing lysates (TL) or VPS35-immunoprecipitates (IP VPS35) using GFP, apparent external evidence of overt hydrocephalus, sug- AQP4 or VPS35 antibodies. Low and high exposures of GFP and gesting that ventricle enlargement progresses slowly in AQP4 immunoblots are shown in order to analyse properly bands signal (n = 3 independent experiments). Quantification data are shown these animals. While the brain of Kidins220−/− embryos as mean ± s.e.m.; NS not significant, *0.01 < P < 0.05, **0.001 < P < presents a pronounced enlargement of LVs, TV and FV 0.01; by two-tailed unpaired Student’s t test (A, B, C). [43], 2-month-old Kidins220 f/f animals only show enlar- gement of the LVs and TV. These results suggest an inci- pient nature of hydrocephalus in 2-month-old Kidins220f/f, Fig. S14a). Transduction of astrocytes with shSNX27-1 or which might develop progressively with aging. Ven- −2 lentivirus sharply decreased AQP4 content, as assessed triculomegaly in these mice resembles that observed in by immunofluorescence and quantitative immunoblot ana- Aqp4−/− animals, which show sporadic progressive lysis (Fig. 5A, B and Supplementary Fig. S14b, c). Note obstructive hydrocephalus with LVs and TV enlargement that SNX27 knockdown also diminished significantly (P < and aqueduct stenosis, without severe changes in FV 0.001) Kidins220 protein levels in astrocytes (Fig. 5B and volume and mild ependymal disorganization [14, 58]. In Supplementary Fig. S14c), as previously observed in HeLa contrast, ependymal-conditional Aqp4−/− mice reveal no cells [25]. In addition, AQP4 levels were restored after abnormalities in the ependymal lining and do not present lysosomal inhibition by bafilomycin treatment in SNX27- changes in brain ventricular volume or increased brain silenced astrocytes (Supplementary Fig. S14d). Conversely, water content, at least at the age studied [18]. A possibility forced SNX27 expression in Kidins220f/f astrocytes sub- is that ependymal-conditional Aqp4−/− mice could develop stantially increased AQP4 expression (Fig. 5C). Of note, ventricular enlargement during aging, a study that is indeed AQP4 colocalized with ectopic SNX27 in Rab5+ missing. Alternatively, these data could suggest that Kidins220 deficiency causes ventriculomegaly via SNX27-retromer-dependent AQP4 degradation 6421 ventriculomegaly in Kidins220f/f mice might rely not only Kidins220f/f animals because the latter do not completely on AQP4 downregulation but also on changes in additional lack SNX27, thus sparing its cargos from complete lyso- molecules. somal degradation. Constitutive Snx27−/− mice develop severe hydro- A very recent publication, that does not study AQP4, cephalus soon after birth, showing ependymal deciliation describes that young postnatal Vps35 conditional KO mice and differentiation defects with complete penetrance [24]. develop neonatal hydrocephalus, similar to that of Snx27−/− These mice could show a more severe phenotype than mice, with deficient ependymal cells differentiation and 6422 A. del Puerto et al. Fig. 5 AQP4 downregulation by Kidins220 deficiency is mediated Kidins220-silenced astrocytes, indicating a cell-autonomous by SNX27. A Representative confocal microscope images of GFP and direct role of Kidins220 on SNX27 regulation but not (green), SNX27 (red), AQP4 (magenta) and DAPI (blue) staining of on VPS35. cultured astrocytes transfected with plasmids encoding GFP and shC or shSNX27 (1 and 2). Scale bar 50 μm. (Lower panels) Quantification Future studies will determine whether the genetic inac- of SNX27 and AQP4 fluorescence intensity in GFP+ astrocytes in each tivation of Snx27 or Vps35 leads to AQP4 depletion and if condition is shown relative to shC-transfected cells (n = 20–30 restoration of SNX27-retromer activity in vivo can be of astrocytes per condition; n = 3 independent experiments); each data therapeutic interest to stabilise AQP4 levels and rescue point represents a single astrocyte. B Representative SNX27, AQP4, Kidins220 and GFAP (loading control) immunoblot analyses of cor- ventriculomegaly in Kidins220-deficient mice or in other tical astrocytes from WT mice transduced with shC- or shSNX27-1- hydrocephalus or neurological disease models that may encoding lentivirus. Quantification of AQP4, SNX27 and Kidins220 concur with AQP4 deficiency. We propose that AQP4 levels after normalization with GFAP is shown relative to shC values downregulation is critical for ventricular enlargement in (n = 4 independent experiments). C Representative confocal micro- scopy images of GFP (green), SNX27 (red), AQP4 (magenta) and Kidins220f/f animals, but we cannot rule out that other DAPI (blue) staining of cultured astrocytes from WT or Kidins220f/f proteins deregulated in these mice might also contribute to mice transfected with plasmids encoding GFP or SNX27-GFP. Scale this phenotype. This could be the case of transient receptor bar 50 μm. Quantification of AQP4 and SNX27 fluorescence intensity potential vanilloid 4 (TRPV4), a cation channel upregulated in GFP+ astrocytes is shown relative to shC-transfected cells (40–60 astrocytes per condition; n = 3 independent experiments); each data in Kidins220−/− astrocytes [60] that together with AQP4 is point represents a single astrocyte. Quantification data are mean ± s.e. essential for water cell-volume control in these cells [52]. m.; NS not significant; *0.01 < P < 0.05, **0.001 < P < 0.01, ***P < Kidins220 trafficking from the endosome to the plasma 0.001; by one-way ANOVA (A), two-tailed unpaired Student’s t test in membrane or to the lysosome is regulated in HeLa cells by (B) and by two-way ANOVA (C). D Scheme illustrating the role of Kidins220 in the control of AQP4 through SNX27 downregulation. In the SNX27-retromer [25], a complex responsible for the the presence of Kidins220, AQP4 levels are preserved by the inter- endosomal recycling of multiple cargos to the cellular sur- action with the SNX27-retromer that favours AQP4 endosomal recy- face [26]. Although regulation in the reverse direction (of cling to the plasma membrane. In the absence of Kidins220, SNX27 SNX27-retromer levels by Kidins220) was unexpected, and VPS35 levels decrease drastically and AQP4 is targeted for lysosomal degradation, consequently leading to brain water dysho- here we find that Kidins220 deficiency downregulates meostasis and contributing to the development of ventriculomegaly SNX27 in vivo and in vitro, and VPS35 in vivo, likely and hydrocephalus. through protein stabilization, as mRNA levels for both molecules are not substantially decreased. The possible role of Kidins220 on the regulation of SNX27-retromer stability ciliogenesis [59]. As in Kidins220f/f animals, S100β is is extremely attractive and deserves further investigation. downregulated in the ependymal barrier of Vps35-deficient Globally, our results point to Kidins220 as a critical mice. Locally activated microglia seems to be responsible regulator of SNX27-retromer dependent endosomal sorting for ependymal cells dyshomeostasis provoked by VPS35 of multiple cargos. Accordingly, a genome-wide analysis loss, as pathology diminishes following microglia depletion performed in Caenorhabditis elegans identified Kidins220 [59]. A similar environment, with activated microglia as a strong candidate regulator of endocytosis [61]. exerting cell non-autonomous effects might be found in Kidins220 and SNX27, therefore, regulate one another in an Kidins220f/f ependymal barrier where VPS35 levels are also intriguing reciprocal ‘feedback loop’ that may result in a severely decreased. Alternatively, as ependymal cells pre- fine-tuning of endocytic recycling versus lysosomal degra- cursors are S100β negative and mature ependymal cells are dation of various SNX27-retromer cargos. Whether S100β positive, the lack of S100β staining might indicate Kidins220 loss could affect AQP4 directly is an interesting changes in the different cell populations at the ependymal issue difficult to study since Kidins220 deficiency decreases barrier in Kidins220 deficient animals. Future studies to both SNX27 and AQP4. Our studies using SNX27 silencing analyse in detail the different cellular subtypes that con- and rendering AQP4 downregulation, suggest that AQP4 stitute the ventricular and subventricular zones (i.e. number downregulation in Kidins220-deficient cells and mice is of precursors and their proliferation) or the presence of mediated by SNX27 protein loss, but we cannot exclude activated microglia will be of high interest. Cell non- that an additional, direct regulation of Kidins220 on AQP4 autonomous effects might also modify the nature of astro- (SNX27-independent) may also exist. cytes in Kidins220f/f brain that may retain these altered Here we have identified AQP4 as a novel SNX27- features when cultured in vitro. This scenario would explain retromer cargo. AQP4 C-terminal amino acids constitute a why VPS35 levels are decreased in cultured Kidins220f/f canonical type I PDZ-L, but differ slightly with those astrocytes while they remain unaltered following defined as selective for SNX27 PDZ-binding bearing an Kidins220 silencing in cultured WT astrocytes. These acidic clamp between P-3 and P-5 [55]. Similar to other results are in striking contrast to those obtained for SNX27 cargos, such as the typical cargo GLUT1, or PTEN SNX27, which diminishes both in Kidins220f/f and [62] and OTULIN [63], AQP4 PDZ-L does not contain Kidins220 deficiency causes ventriculomegaly via SNX27-retromer-dependent AQP4 degradation 6423 acidic residues at both P-3 and P-5, suggesting that selec- shunt, saves the life of many patients but the failure rate of tivity and affinity of SNX27 for its cargos may rely not only this procedure is high and its outcome is not always posi- on their C-terminal peptide but also on their global context. tive. AQP4 downregulation had been shown at perivascular KIDINS220 heterozygous nonsense variants have been locations in iNPH patients [22, 23] and here we show that recently associated with SINO syndrome, a novel rare its expression and that of KIDINS220 (but not S100β) are autosomal disease characterized by spastic paraplegia, substantially decreased at the ventricular ependymal lining, intellectual disability, nystagmus and obesity in children suggesting that loss of these proteins at the ependyma might [40]. Similar to Kidins220f/f mice, the brain LVs and TV of also contribute to adult chronic hydrocephalus. Indeed, these children are enlarged but not the FV [40]. Moreover, AQP4 deficiency in this region might account for the homozygous KIDINS220 loss-of-function variants have reported enhancement in transependymal CSF flow in iNPH been detected in foetuses that also present cerebral ven- patients [64] that also present glymphatic (glia-lymphatic) triculomegaly [41]. We anticipate that truncated Kidins220 circulation system for brain waste clearance impairments forms lacking the PDZ-L, as that identified in SINO syn- [64, 65]. Since the glymphatic CSF circulation is controlled drome [40], will not interact with the SNX27-retromer by AQP4 and is altered in iNPH and AD [11], our results complex. Recycling to the cell surface of Kidins220 var- prompt to investigate the participation of KIDINS220 and iants devoid of PDZ-L would therefore be hampered and SNX27-retromer in the function of this system, and in the these variants would be mostly targeted for lysosomal dynamics and turnover of AQP4 and other cargos in iNPH degradation. Accordingly, this type of variant would not and different hydrocephalus types, AD, and other neuro- only greatly impair Kidins220 function at the cell surface degenerative and neuropsychiatric disorders that concur and other intracellular compartments, but also drop its with ventriculomegaly. Novel strategies designed to steady-state levels by increasing its degradation rate. potentiate AQP4 levels in ventricular ependymal cells and Moreover, as the interaction of Kidins220 PDZ-L with astrocytes, including the use of viral vectors, molecules or SNX27 could be key for maintaining SNX27-retromer compounds that enhance Kidins220 and/or SNX27- stability and levels, the lack of Kidins220 PDZ-L binding retromer expression or stabilization and therefore AQP4 could destabilize SNX27 and therefore decrease its half-life. recycling to the plasma membrane, could be promising The global recycling of SNX27 cargos (like AQP4) to the therapeutic approaches to prevent or revert ventricular cell surface would be consequently diminished, resulting in enlargement caused by AQP4 increased turnover. their increased lysosomal targeting and degradation. SINO syndrome, hydrocephalus and ventriculomegaly could thus Acknowledgements This work was supported by Spanish grants SAF2017-88885-R to TI and SAF2017-88881-R to MRC from the be all retromer-related disorders. Spanish Ministerio de Economía y Competitividad/Agencia Estatal de The novel missense variants identified in KIDINS220 Investigación/European Union Funds for Regional Development gene in schizophrenic patients [37–39] could be also asso- (MINECO/AEI/ FEDER, EU); by Comunidad de Madrid through the ciated with cerebral ventricular enlargement in these indi- European Social Fund (ESF)-financed programme NEUROMETAB- CM (B2017/BMD-3700) to TI and AORTASANA-CM (B2017/ viduals given it is a frequent feature in this psychiatric BMD-3676) to MRC; by ‘La Caixa’ Banking Foundation project disorder [5, 6], a possibility that should be examined. HR18-00068 to MRC; by Centro de Investigación Biomédica en Red Although it is currently unknown whether KIDINS220 de Enfermedades Neurodegenerativas (CIBERNED, Instituto de Salud variants in SINO and schizophrenic patients could lead to Carlos III, Spain) CB06/05/1122 to TI and collaborative grant CIBERNED PI2018/06 to TI and IFa; by grant PID2019-104763RB- similar defects in SNX27-retromer and AQP4 expression 100 and Ramón y Cajal programme contract (RYC-2014-15991), both and brain water dyshomeostasis as found in Kidins220 from MINECO/AEI/FEDER (EU) to EP; by Wellcome Trust Senior deficient mice, our data point at Kidins220f/f mice as an Investigator Award (107116/Z/15/Z), the European Union’s Horizon attractive model to study SINO syndrome, Kidins220- 2020 Research and Innovation programme under grant agreement 739572 and a UK Dementia Research Institute Foundation award, all linked schizophrenia and other neuropathologies accom- to GS; by grant PI19/00778 to AJJ from the Spanish Instituto de Salud panied by ventriculomegaly such as foetal/perinatal to Carlos III, co-funded by FEDER (EU); by Spanish grants from adulthood onset forms of hydrocephalus. FEDER/Ministerio de Ciencia e Innovación (MICI)–AEI (SAF2017- The complex etiopathology of ventriculomegaly and 86690-R), Network of Excellence iDIFFER (RED2018-102723-T), Instituto de Salud Carlos III (CIBERNED CB06/05/0086 and RETIC hydrocephalus and the limited knowledge of the molecular Tercel RD16/0011/0017) and Generalitat Valenciana (Prometeo 2017/ mechanisms involved hinder the diagnostic, prognostic and 030) to IFa. ADP was funded by a Juan de la Cierva Formación therapeutic strategies for patients. There is an urgent need of programme contract (FJCI-2014-19673; MINECO) and CIBERNED; appropriate tools for differential diagnosis and biomarkers JP-U was contracted by SAF2017-88885-R grant; AS-S was con- tracted by grants CIBERNED-PI2015-2/06 and PI2018/06, and for specific hydrocephalus types or different SAF2017-88885-R; AS-G was funded by a contract from B2017/ ventriculomegaly-associated diseases, especially in adults. BMD-3700 grant and LSMP by a PhD training contract (PRE2018- The most common procedure to treat the large ventricular 086989) associated to grant SAF2017-88885-R. The cost of this accumulation of CSF, the implantation of a ventricular publication has been paid in part by FEDER (European Union Funds 6424 A. del Puerto et al. for Regional Development) funds. We are particularly indebted to the 5. Johnstone EC, Crow TJ, Frith CD, Husband J, Kreel L. Cerebral patients and the Biobanks integrated in the Spanish National Biobanks ventricular size and cognitive impairment in chronic schizo- Network, Biobank Banco de Tejidos CIEN (PT17/0015/0014), the phrenia. Lancet. 1976;2:924–6. HCB-IDIBAPS Biobank and Institute of Neuropathology Brain Bank- 6. Kempton MJ, Stahl D, Williams SC, DeLisi LE. Progressive IDIBELL, for their collaboration in sample and data procurement. We lateral ventricular enlargement in schizophrenia: a meta-analysis are grateful to Professor M. Amiry-Moghaddam (University of Oslo, of longitudinal MRI studies. Schizophr Res. 2010;120:54–62. Norway) for his help with AQP4 reagents and methodological details, 7. Mak E, Su L, Williams GB, Firbank MJ, Lawson RA, Yarnall AJ, Professors P.J. Cullen (University of Bristol, UK) and F. Steinberg et al. Longitudinal whole-brain atrophy and ventricular enlarge- (Albert Ludwigs Universitaet Freiburg, Germany) for providing ment in nondemented Parkinson’s disease. Neurobiol Aging. SNX27 plasmids. We also acknowledge M. Rodríguez-Martínez and 2017;55:78–90. T. Navarro for performing MRI studies; Drs. P. Lopez-Larrubia 8. Nestor SM, Rupsingh R, Borrie M, Smith M, Accomazzi V, Wells (IIBM, CSIC-UAM) and A. Martín (Achucarro Basque Center for JL, et al. Ventricular enlargement as a possible measure of Alz- Neuroscience, Bilbao, Spain) for advice in MRI analysis; L. García- heimer’s disease progression validated using the Alzheimer’s Guerra, A. González-Martín and other laboratory members for help disease neuroimaging initiative database. Brain. and suggestions; L. Sánchez-Ruiloba for technical assistance in con- 2008;131:2443–54. focal images analysis, L. Jiménez for taking care of mice colonies; and 9. Silverberg GD, Mayo M, Saul T, Rubenstein E, McGuire D. to core facilities at Instituto de Investigaciones Biomédicas ‘Alberto Alzheimer’s disease, normal-pressure hydrocephalus, and senes- Sols’ de Madrid (IIBM, CSIC-UAM): Animal Care, Genomics, cent changes in CSF circulatory physiology: a hypothesis. Lancet Confocal microscopy. Neurol. 2003;2:506–11. 10. Martin-Laez R, Valle-San Roman N, Rodriguez-Rodriguez EM, Author contributions TI, MRC and ADP conceived and designed the Marco-de Lucas E, Berciano Blanco JA, Vazquez-Barquero A. experiments, interpreted the data and wrote the paper with input from Current concepts on the pathophysiology of idiopathic chronic all authors. ADP, JP-U, AS-G, CL-M, AJJ, EP, LSMP, GC-G, BM-P, adult hydrocephalus: are we facing another neurodegenerative AS-S and MPS-C performed experiments, analysed, interpreted and disease? Neurologia. 2018;33:449–58. discussed results. AJJ, FC, GS, IFe and IFa, contributed reagents, 11. Reeves BC, Karimy JK, Kundishora AJ, Mestre H, Cerci HM, animals, materials or analysis tools, and made intellectual contribu- Matouk C, et al. Glymphatic system impairment in alzheimer’s tions to experimental design and discussion. All authors approved disease and idiopathic normal pressure hydrocephalus. Trends submission of this work. Mol Med. 2020;26:285–95. 12. Nielsen S, Nagelhus EA, Amiry-Moghaddam M, Bourque C, Agre P, Ottersen OP. Specialized membrane domains for water Compliance with ethical standards transport in glial cells: high-resolution immunogold cytochemistry of aquaporin-4 in rat brain. J Neurosci. 1997;17:171–80. Conflict of interest The authors declare no competing interests. 13. Rash JE, Yasumura T, Hudson CS, Agre P, Nielsen S. Direct immunogold labeling of aquaporin-4 in square arrays of astrocyte Publisher’s note Springer Nature remains neutral with regard to and ependymocyte plasma membranes in rat brain and spinal cord. jurisdictional claims in published maps and institutional affiliations. Proc Natl Acad Sci U.S.A. 1998;95:11981–6. 14. Feng X, Papadopoulos MC, Liu J, Li L, Zhang D, Zhang H, et al. Sporadic obstructive hydrocephalus in Aqp4 null mice. J Neurosci Open Access This article is licensed under a Creative Commons Res. 2009;87:1150–5. Attribution 4.0 International License, which permits use, sharing, 15. Bloch O, Auguste KI, Manley GT, Verkman AS. Accelerated adaptation, distribution and reproduction in any medium or format, as progression of kaolin-induced hydrocephalus in aquaporin-4- long as you give appropriate credit to the original author(s) and the deficient mice. J Cereb Blood Flow Metab. 2006;26:1527–37. source, provide a link to the Creative Commons license, and indicate if 16. Iliff JJ, Wang M, Liao Y, Plogg BA, Peng W, Gundersen GA, changes were made. The images or other third party material in this et al. A paravascular pathway facilitates CSF flow through the article are included in the article’s Creative Commons license, unless brain parenchyma and the clearance of interstitial solutes, indicated otherwise in a credit line to the material. If material is not including amyloid beta. Sci Transl Med. 2012;4:147ra111. included in the article’s Creative Commons license and your intended 17. Haj-Yasein NN, Vindedal GF, Eilert-Olsen M, Gundersen GA, use is not permitted by statutory regulation or exceeds the permitted Skare O, Laake P, et al. Glial-conditional deletion of aquaporin-4 use, you will need to obtain permission directly from the copyright (Aqp4) reduces blood-brain water uptake and confers barrier holder. To view a copy of this license, visit http://creativecommons. function on perivascular astrocyte endfeet. Proc Natl Acad Sci U. org/licenses/by/4.0/. S.A. 2011;108:17815–20. 18. Vindedal GF, Thoren AE, Jensen V, Klungland A, Zhang Y, Holtzman MJ, et al. Removal of aquaporin-4 from glial and References ependymal membranes causes brain water accumulation. Mol Cell Neurosci. 2016;77:47–52. 1. Del Bigio MR. Neuropathology and structural changes in hydro- 19. Nagelhus EA, Ottersen OP. Physiological roles of aquaporin-4 in cephalus. Dev Disabil Res Rev. 2010;16:16–22. brain. Physiol Rev. 2013;93:1543–62. 2. Oi S. Hydrocephalus research update-controversies in definition 20. Papadopoulos MC, Verkman AS. Aquaporin water channels in the and classification of hydrocephalus. Neurol Med Chir (Tokyo). nervous system. Nat Rev Neurosci. 2013;14:265–77. 2010;50:859–69. 21. Nakada T, Kwee IL. Fluid dynamics inside the brain barrier: 3. Jimenez AJ, Dominguez-Pinos MD, Guerra MM, Fernandez- current concept of interstitial flow, glymphatic flow, and cere- Llebrez P, Perez-Figares JM. Structure and function of the epen- brospinal fluid circulation in the brain. Neuroscientist. dymal barrier and diseases associated with ependyma disruption. 2019;25:155–66. Tissue Barriers. 2014;2:e28426. 22. Eide PK, Hansson HA. Astrogliosis and impaired aquaporin-4 and 4. Leinonen V, Vanninen R, Rauramaa T. Cerebrospinal fluid cir- dystrophin systems in idiopathic normal pressure hydrocephalus. culation and hydrocephalus. Handb Clin Neurol. 2017;145:39–50. Neuropathol Appl Neurobiol. 2018;44:474–90. Kidins220 deficiency causes ventriculomegaly via SNX27-retromer-dependent AQP4 degradation 6425 23. Hasan-Olive MM, Enger R, Hansson HA, Nagelhus EA, Eide PK. nonsense variants cause spastic paraplegia, intellectual Loss of perivascular aquaporin-4 in idiopathic normal pressure disability, nystagmus, and obesity. Hum Mol Genet. hydrocephalus. Glia. 2019;67:91–100. 2016;25:2158–67. 24. Wang X, Zhou Y, Wang J, Tseng IC, Huang T, Zhao Y, et al. 41. Mero IL, Mork HH, Sheng Y, Blomhoff A, Opheim GL, Erichsen SNX27 deletion causes hydrocephalus by impairing ependymal A, et al. Homozygous KIDINS220 loss-of-function variants in cell differentiation and ciliogenesis. J Neurosci. fetuses with cerebral ventriculomegaly and limb contractures. 2016;36:12586–97. Hum Mol Genet. 2017;26:3792–6. 25. Steinberg F, Gallon M, Winfield M, Thomas EC, Bell AJ, Heesom 42. Santpere G, Nieto M, Puig B, Ferrer I. Abnormal Sp1 transcrip- KJ, et al. A global analysis of SNX27-retromer assembly and tion factor expression in Alzheimer disease and tauopathies. cargo specificity reveals a function in glucose and metal ion Neurosci Lett. 2006;397:30–4. transport. Nat Cell Biol. 2013;15:461–71. 43. Cesca F, Yabe A, Spencer-Dene B, Scholz-Starke J, Medrihan L, 26. Cullen PJ, Steinberg F. To degrade or not to degrade: mechanisms Maden CH, et al. Kidins220/ARMS mediates the integration of and significance of endocytic recycling. Nat Rev Mol Cell Biol. the neurotrophin and VEGF pathways in the vascular and nervous 2018;19:679–96. systems. Cell Death Differ. 2012;19:194–208. 27. Zhang H, Huang T, Hong Y, Yang W, Zhang X, Luo H, et al. The 44. Cesca F, Yabe A, Spencer-Dene B, Arrigoni A, Al-Qatari M, retromer complex and sorting nexins in neurodegenerative dis- Henderson D, et al. Kidins220/ARMS is an essential modulator of eases. Front Aging Neurosci. 2018;10:79. cardiovascular and nervous system development. Cell Death Dis. 28. Iglesias T, Cabrera-Poch N, Mitchell MP, Naven TJ, Rozengurt E, 2011;2:e226. Schiavo G. Identification and cloning of Kidins220, a novel 45. Lopez-Menendez C, Simon-Garcia A, Gamir-Morralla A, Pose- neuronal substrate of protein kinase D. J Biol Chem. Utrilla J, Lujan R, Mochizuki N, et al. Excitotoxic targeting of 2000;275:40048–56. Kidins220 to the Golgi apparatus precedes calpain cleavage of 29. Kong H, Boulter J, Weber JL, Lai C, Chao MV. An evolutionarily Rap1-activation complexes. Cell Death Dis. 2019;10:535. conserved transmembrane protein that is a novel downstream 46. Mirzadeh Z, Doetsch F, Sawamoto K, Wichterle H, target of neurotrophin and ephrin receptors. J Neurosci. Alvarez-Buylla A. The subventricular zone en-face: wholemount 2001;21:176–85. staining and ependymal flow. J Vis Exp. 2010; J Vis Exp. 2010; 30. Neubrand VE, Cesca F, Benfenati F, Schiavo G. Kidins220/ 39:e1938. ARMS as a functional mediator of multiple receptor signalling 47. Del Bigio MR. Ependymal cells: biology and pathology. Acta pathways. J Cell Sci. 2012;125:1845–54. Neuropathol. 2010;119:55–73. 31. Scholz-Starke J, Cesca F. Stepping out of the shade: control of 48. Roales-Bujan R, Paez P, Guerra M, Rodriguez S, Vio K, Ho- neuronal activity by the scaffold protein kidins220/ARMS. Front Plagaro A, et al. Astrocytes acquire morphological and Cell Neurosci. 2016;10:68. functional characteristics of ependymal cells following disruption 32. Lopez-Menendez C, Gamir-Morralla A, Jurado-Arjona J, Higuero of ependyma in hydrocephalus. Acta Neuropathol. AM, Campanero MR, Ferrer I, et al. Kidins220 accumulates with 2012;124:531–46. tau in human Alzheimer’s disease and related models: modulation 49. Raponi E, Agenes F, Delphin C, Assard N, Baudier J, Legraver- of its calpain-processing by GSK3beta/PP1 imbalance. Hum Mol end C, et al. S100B expression defines a state in which GFAP- Genet. 2013;22:466–82. expressing cells lose their neural stem cell potential and acquire a 33. Gamir-Morralla A, Lopez-Menendez C, Ayuso-Dolado S, Tejeda more mature developmental stage. Glia. 2007;55:165–77. GS, Montaner J, Rosell A, et al. Development of a neuroprotective 50. Jacquet BV, Salinas-Mondragon R, Liang H, Therit B, Buie JD, peptide that preserves survival pathways by preventing Kidins220/ Dykstra M, et al. FoxJ1-dependent gene expression is required for ARMS calpain processing induced by excitotoxicity. Cell Death differentiation of radial glia into ependymal cells and a subset of Dis. 2015;6:e1939. astrocytes in the postnatal brain. Development. 34. Gamir-Morralla A, Belbin O, Fortea J, Alcolea D, Ferrer I, Lleo 2009;136:4021–31. A, et al. Kidins220 correlates with Tau in alzheimer’s disease 51. Redmond SA, Figueres-Onate M, Obernier K, Nascimento MA, brain and cerebrospinal fluid. J Alzheimers Dis. 2017;55:1327–33. Parraguez JI, Lopez-Mascaraque L, et al. Development of epen- 35. Sebastian-Serrano A, Simon-Garcia A, Belmonte-Alfaro A, Pose- dymal and postnatal neural stem cells and their origin Utrilla J, Santos-Galindo M, Del Puerto A, et al. Differential from a common embryonic progenitor. Cell Rep. regulation of Kidins220 isoforms in Huntington’s disease. Brain 2019;27:429–41. e423 Pathol. 2020;30:120–36. 52. Benfenati V, Caprini M, Dovizio M, Mylonakou MN, Ferroni S, 36. Lopez-Benito S, Sanchez-Sanchez J, Brito V, Calvo L, Lisa S, Ottersen OP, et al. An aquaporin-4/transient receptor potential Torres-Valle M, et al. Regulation of BDNF release by ARMS/ vanilloid 4 (AQP4/TRPV4) complex is essential for cell-volume Kidins220 through modulation of synaptotagmin-IV Levels. J control in astrocytes. Proc Natl Acad Sci U.S.A. Neurosci. 2018;38:5415–28. 2011;108:2563–8. 37. Kranz TM, Goetz RR, Walsh-Messinger J, Goetz D, Antonius D, 53. Madrid R, Le Maout S, Barrault MB, Janvier K, Benichou S, Dolgalev I, et al. Rare variants in the neurotrophin signaling Merot J. Polarized trafficking and surface expression of the AQP4 pathway implicated in schizophrenia risk. Schizophr Res. water channel are coordinated by serial and regulated interactions 2015;168:421–8. with different clathrin-adaptor complexes. EMBO J. 38. Malaspina D, Kranz TM, Heguy A, Harroch S, Mazgaj R, 2001;20:7008–21. Rothman K, et al. Prefrontal neuronal integrity predicts symptoms 54. McNally KE, Faulkner R, Steinberg F, Gallon M, Ghai R, Pim D, and cognition in schizophrenia and is sensitive to genetic het- et al. Retriever is a multiprotein complex for retromer-independent erogeneity. Schizophr Res. 2016;172:94–100. endosomal cargo recycling. Nat Cell Biol. 2017;19:1214–25. 39. Kranz TM, Berns A, Shields J, Rothman K, Walsh-Messinger J, 55. Clairfeuille T, Mas C, Chan AS, Yang Z, Tello-Lafoz M, Chandra Goetz RR, et al. Phenotypically distinct subtypes of psychosis M, et al. A molecular code for endosomal recycling of phos- accompany novel or rare variants in four different signaling genes. phorylated cargos by the SNX27-retromer complex. Nat Struct EBioMedicine. 2016;6:206–14. Mol Biol. 2016;23:921–32. 40. Josifova DJ, Monroe GR, Tessadori F, de Graaff E, van der 56. Neely JD, Amiry-Moghaddam M, Ottersen OP, Froehner SC, Zwaag B, Mehta SG, et al. Heterozygous KIDINS220/ARMS Agre P, Adams ME. Syntrophin-dependent expression and 6426 A. del Puerto et al. localization of Aquaporin-4 water channel protein. Proc Natl Acad 61. Balklava Z, Pant S, Fares H, Grant BD. Genome-wide analysis Sci U.S.A. 2001;98:14108–13. identifies a general requirement for polarity proteins in endocytic 57. Nakahama K, Fujioka A, Nagano M, Satoh S, Furukawa K, Sasaki traffic. Nat Cell Biol. 2007;9:1066–73. H, et al. A role of the C-terminus of aquaporin 4 in its membrane 62. Shinde SR, Maddika S. PTEN regulates glucose transporter expression in cultured astrocytes. Genes Cells. 2002;7:731–41. recycling by impairing SNX27 retromer assembly. Cell Rep. 58. Li X, Kong H, Wu W, Xiao M, Sun X, Hu G. Aquaporin-4 2017;21:1655–66. maintains ependymal integrity in adult mice. Neuroscience. 63. Stangl A, Elliott PR, Pinto-Fernandez A, Bonham S, Harrison L, 2009;162:67–77. Schaub A, et al. Regulation of the endosomal SNX27-retromer by 59. Wu KY, Tang FL, Lee D, Zhao Y, Song H, Zhu XJ, et al. OTULIN. Nat Commun. 2019;10:4320. Ependymal Vps35 promotes ependymal cell differentiation and 64. Ringstad G, Vatnehol SAS, Eide PK. Glymphatic MRI in idio- survival, suppresses microglial activation, and prevents neonatal pathic normal pressure hydrocephalus. Brain. hydrocephalus. J Neurosci. 2020;40:3862–79. 2017;140:2691–705. 60. Jaudon F, Chiacchiarettla M, Albini M, Ferroni S, Benfenati F, 65. Eide PK, Ringstad G. Delayed clearance of cerebrospinal fluid Cesca F. Kidins220/ARMS controls astrocyte calcium signaling tracer from entorhinal cortex in idiopathic normal pressure and neuron-astrocyte communication. Cell Death Differ. hydrocephalus: a glymphatic magnetic resonance imaging study. J 2020;27:1505–19. Cereb Blood Flow Metab. 2019;39:1355–68. Affiliations Ana del Puerto1,2,16,16 Julia Pose-Utrilla1,2 Ana Simón-García1,2 Celia López-Menéndez1,2 Antonio J. Jiménez3,4 ● ● ● ● ● Eva Porlan5,6,7 Luis S. M. Pajuelo1,2 Guillermo Cano-García1 Beatriz Martí-Prado8 ● ● ● ● Álvaro Sebastián-Serrano1,2,17,17 Marina P. Sánchez-Carralero1,2 Fabrizia Cesca9 Giampietro Schiavo 10,11 ● ● ● ● Isidro Ferrer 2,12,13,14 Isabel Fariñas2,8 Miguel R. Campanero 5,15 Teresa Iglesias 1,2 ● ● ● 1 Instituto de Investigaciones Biomédicas “Alberto Sols”, Consejo 10 Department of Neuromuscular Disorders, UCL Institute of Superior de Investigaciones Científicas-Universidad Autónoma de Neurology, University College London, London, UK Madrid (CSIC-UAM), Madrid, Spain 11 UK Dementia Research Institute, University College London, 2 Centro de Investigación Biomédica en Red de Enfermedades London, UK Neurodegenerativas (CIBERNED), Instituto de Salud Carlos III, 12 Madrid, Spain Departamento de Patología y Terapéutica Experimental, Universidad de Barcelona, Barcelona, Spain 3 Departamento de Biología Celular, Genética y Fisiología, Facultad 13 de Ciencias, Universidad de Málaga, Bulevar Louis Pasteur, Instituto de Neurociencias, Universidad de Barcelona, Hospitalet Málaga, Spain de Llobregat, Barcelona, Spain 14 4 Instituto de Investigación Biomédica de Málaga (IBIMA), Instituto Servicio de Anatomía Patológica, Hospital Universitario de de Salud Carlos III, Málaga, Spain Bellvitge, Carrer de la Feixa Llarga, Barcelona, Spain 15 5 Centro de Biología Molecular “Severo Ochoa” (CSIC-UAM), Centro de Investigación Biomédica en Red en Enfermedades Madrid, Spain Cardiovasculares (CIBERCV), Instituto de Salud Carlos III, Madrid, Spain 6 Departamento de Biología Molecular, Facultad de Ciencias, UAM, 16 Madrid, Spain Present address: Departmento de Biotecnología, Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria (INIA), 7 Instituto de Investigación Hospital Universitario La Paz (IdiPAZ), Madrid, Spain Instituto de Salud Carlos III, Madrid, Spain 17 Present address: Departmento de Bioquímica y Biología 8 Departmento de Biología Celular, Biología Funcional y Molecular, Facultad de Veterinaria, Universidad Complutense de Antropología Física, Universidad de Valencia, Burjassot, Spain Madrid, Madrid, Spain 9 Department of Life Sciences, University of Trieste, Trieste, Italy