The Alzheimer's disease risk factor CD2AP maintains blood–brain barrier integrity

Abstract

CD2-associated protein (CD2AP) is a leading genetic risk factor for Alzheimer's disease, but little is known about the function of CD2AP in the brain. We studied CD2AP^–/– mice to address this question. Because CD2AP^–/– mice normally die by 6 weeks from nephrotic syndrome, we used mice that also express a CD2AP transgene in the kidney, but not brain, to attenuate this phenotype. CD2AP-deficient mice had no behavioral abnormalities except for mild motor and anxiety deficits in a subset of CD2AP^–/– mice exhibiting severe nephrotic syndrome, associated with systemic illness. Pentylenetetrazol (PTZ)-induced seizures occurred with shorter latency in CD2AP^–/– mice, but characteristics of these seizures on electroencephalography were not altered. As CD2AP is expressed in brain-adjacent endothelial cells, we hypothesized that the shorter latency to seizures without detectably different seizure characteristics may be due to increased penetration of PTZ related to compromised blood–brain barrier integrity. Using sodium fluorescein extravasation, we found that CD2AP^–/– mice had reduced blood–brain barrier integrity. Neither seizure severity nor blood–brain barrier integrity was correlated with nephrotic syndrome, indicating that these effects are dissociable from the systemic illness associated with CD2AP deficiency. Confirming this dissociation, wild-type mice with induced nephrotic syndrome maintained an intact blood–brain barrier. Taken together, our results support a role of CD2AP in mediating blood–brain barrier integrity and suggest that cerebrovascular roles of CD2AP could contribute to its effects on Alzheimer's disease risk.

Introduction

CD2-associated protein (CD2AP) has been identified in genome-wide association studies as an important genetic risk factor for Alzheimer's disease (AD) in both clinical and clinicopathological cohorts (1,2) and is in the top 10 on the AlzGene meta-analysis (3). However, little is known about the role CD2AP plays in the brain or how it might affect AD risk.

CD2AP was first characterized as a scaffolding protein that facilitates clustering of CD2 in T-cells, containing three SH3 domains, a proline-rich domain, and an actin-binding region (4). Early work identified a critical role of CD2AP in maintaining glomerular integrity in the kidney, evidenced by death in CD2AP^–/– mice by 6–7 weeks of age from nephrotic syndrome (5). CD2AP contributes to intercellular junction integrity by conferring resistance to mechanical stress (6). Together, these studies suggest that CD2AP is a critical scaffolding protein that facilitates intercellular junctions. This role may extend to many tissue types, as CD2AP is expressed broadly in endothelial and epithelial cells (7).

There has been very little research on CD2AP's potential roles in the brain. Immunohistochemical studies show that it is localized primarily to cerebrovascular endothelial cells in the brain (7–9). mRNA data from the Allen Brain Atlas suggest that it may be expressed at low levels in neurons (10), but RNA-seq data from another database suggest 5-fold more prominent expression in endothelial cells than neurons (11). One study identified cindr, the Drosophila homolog of CD2AP, as a modulator of tau toxicity (12). Another found that knockdown of CD2AP in neuroblastoma cells decreased the ratio of secreted Aβ42/Aβ40, but found only subtle effects of CD2AP deficiency on Aβ levels in an AD mouse model, with no effect on plaque load (13). Therefore, the normal function of CD2AP in the mammalian brain remains unclear, which limits our understanding of its potential role in AD. To further evaluate the function of CD2AP in the brain, we capitalized on the availability of CD2AP-deficient mice expressing a CD2AP transgene in the kidney to delay lethality (14). Here, we characterized these CD2AP^–/– mice to elucidate the function of CD2AP in the brain.

Results

We studied CD2AP^–/– mice with a nephrin promoter-driven CD2AP transgene that expresses only in the kidney and muscle (14), which attenuates the nephrotic syndrome-driven systemic illness and early mortality seen in CD2AP^–/– mice (5). We first established that CD2AP was normally expressed in brain, and absent in brain tissue from CD2AP^–/– mice expressing the renal transgene (Fig. 1A). Next, we verified that expression of CD2AP in the kidney extended the survival of CD2AP^–/– mice. Compared with typical survival of 6–7 weeks in CD2AP^–/– mice without the transgene (5), CD2AP^–/– mice with the transgene lived up to 5 months (Fig. 1B; median survival 4.4 months). In the original report describing CD2AP^–/– mice with the renal transgene (14), mice lived to 12 months and this reduction in lifespan is likely due to downregulation of transgene expression in the 10 years since the original report. Consistent with systemic illness, body weight was reduced in CD2AP^–/– mice (Fig. 1C). To determine if nephrotic syndrome was developing, we measured proteinuria. CD2AP^–/– mice indeed exhibited varying levels of proteinuria (Fig. 1D). CD2AP^+/– mice never exhibited proteinuria and had no weight loss (Fig. 1C and D).

Figure 1.

Characteristics of CD2AP-deficient mice (all expressing the nephrin-CD2AP transgene in the kidney to attenuate lethality). (A) CD2AP expression is not detectable in CD2AP^–/– brain lysates with two different antibodies. (B) CD2AP^–/– mice show premature mortality (n = 68–161 mice per group, P < 0.0001 by Log-rank Mantel-Cox test for CD2AP^–/– versus CD2AP^+/+, P = 0.09 for CD2AP^+/– versus CD2AP^+/+). (C) CD2AP^–/– mice have decreased weight (n = 14–23 mice per group per sex, genotype effect P = 0.0002 by two-way ANOVA, ^+/+ versus ^–/–P = 0.0014 by Dunnett's post hoc). (D) Representative Coomassie-stained gel of 1 µl of urine per mouse demonstrating that CD2AP^–/– mice exhibit readily detectable proteinuria, which was never seen in control or CD2AP^+/– mice. Quantitation of the level of proteinuria revealed a significant effect of CD2AP deficiency (P < 0.0001 by χ² test for trend). For further analysis, mice were divided into high and low proteinuria groups based on separation at the median level of proteinuria.

To begin assessing the function of CD2AP in the brain, we determined if CD2AP^+/– or CD2AP^–/– mice exhibited behavioral abnormalities. We were concerned that renal dysfunction in CD2AP^–/– mice could produce behavioral effects simply due to systemic illness, rather than as a reflection of intrinsic neuronal dysfunction. Therefore, we measured proteinuria in each mouse and split the CD2AP^–/– mice into two groups based on the median level of proteinuria (Fig. 1D). We first assessed exploratory locomotor behavior. CD2AP^+/– mice and CD2AP^–/– mice with low proteinuria had normal ambulatory distance in the open field (Fig. 2A). However, CD2AP^–/– mice with high proteinuria had reduced ambulatory distance, and consistent with the idea that this was due to their systemic illness, there was a strong negative correlation between proteinuria and locomotor behavior (Fig. 2A). Likely reflecting the same underlying deficit in locomotor function, the same pattern was seen with total entrances in the elevated plus maze (Fig. 2B), velocity in the light/dark box (Fig. 2C) and latency to enter the dark side in the light/dark box (Fig. 2D). CD2AP^–/– mice with high proteinuria also exhibited a mild anxiety-like phenotype in the elevated plus maze (Fig. 2E) and open field (Fig. 2F), which was not seen in CD2AP^+/– mice or CD2AP^–/– mice with low proteinuria.

Figure 2.

CD2AP-deficient mice (all Tg nephrin-CD2AP) are behaviorally normal, except for a subset of CD2AP^–/– mice with high proteinuria levels. (A–D) CD2AP^–/– mice with high proteinuria levels have locomotor deficits. (A) CD2AP^–/– mice with high proteinuria levels covered less distance in the open field during a 10-min testing period (n = 24–25 mice per genotype, ANOVA P = 0.0273, CD2AP^+/+ versus CD2AP^–/– with high proteinuria levels P = 0.0273 by Dunnett's post hoc) and CD2AP^–/– mice as a whole (including the full spectrum of high and low proteinuria) showed a negative correlation between time in the open arms and proteinuria level (n = 24 mice, P = 0.0003, r² = 0.40 by Pearson correlation). (B) In the elevated plus maze, CD2AP^–/– mice with high proteinuria levels trended towards fewer total entrances (n = 24–25 mice per genotype, ANOVA P = 0.076), and CD2AP^–/– mice as a whole showed a negative correlation between total entrances and proteinuria level (n = 24 mice, P = 0.008, r² = 0.27 by Pearson correlation). (C) CD2AP^–/– mice with high proteinuria levels had decreased velocity in the light/dark box (n = 12 mice per genotype, ANOVA P = 0.0245, CD2AP^+/+ versus CD2AP^–/– with high proteinuria levels P = 0.0112 by Dunnett's post hoc). (D) There was also a significant genotype effect for CD2AP in the latency to dark in the light/dark box, driven by CD2AP^–/– with high proteinuria levels mice that sat lethargically for an prolonged period upon initial placement in the light (n = 12 mice per genotype, ANOVA P = 0.0097, CD2AP^+/+ versus CD2AP^–/– with high proteinuria levels P = 0.0125 by Dunnett's post hoc). (E and F) CD2AP^–/– mice with high proteinuria levels had a mild anxiety phenotype. (E) In the elevated plus maze, CD2AP^–/– mice with high proteinuria levels spent decreased time in the open arms in the elevated plus maze (n = 24–25 mice per genotype, ANOVA P = 0.00058, ^+/+ versus ^–/–P = 0.00014 by Dunnett's post hoc), and CD2AP^–/– mice as a whole trended strongly towards a negative correlation between time in the open arms and proteinuria level (n = 24 mice, P = 0.051, r² = 0.16 by Pearson correlation). (F) CD2AP^–/– mice with high proteinuria levels did not spend significantly less time in the center in the open field overall (n = 24–25 mice per genotype, ANOVA P = 0.243), but there was a significant negative correlation for CD2AP^–/– mice as a whole between time in the center and proteinuria level (n = 24 mice, P = 0.009, r² = 0.27 by Pearson correlation). (G) CD2AP^–/– with high proteinuria levels mice showed increased freezing before training in fear conditioning (n = 12 mice per genotype, genotype effect P = 0.0054 by two-way ANOVA, pre-training CD2AP^+/+ versus CD2AP^–/– with high proteinuria levels P = 0.002 by Dunnett's post hoc), but did not exhibit detectable differences in fear conditioning otherwise.

We also examined learning and memory using a fear conditioning paradigm. CD2AP^–/– mice with high proteinuria showed increased freezing upon initial placement in the box used for the fear conditioning task, before any shock was delivered (Fig. 2G), again consistent with reduced mobility or increased anxiety due to illness. Interestingly, CD2AP-deficient mice, even those with high proteinuria, had no detectable deficits in either contextual or cued fear conditioning (Fig. 2G).

Thus, CD2AP^+/– and CD2AP^–/– mice with low levels of proteinuria were not differentiable from CD2AP^+/+ mice in any of the behaviors we measured. Since the only apparent effects of its deficiency are indirect, due to systemic illness in a subset of mice, we conclude that CD2AP does not play a critical role in a variety of behavior paradigms, including learning and memory.

We also asked if CD2AP^+/– or CD2AP^–/– mice exhibited any abnormalities in excitability, which is altered in AD and AD models (reviewed in 15). We used the paradigm of pentylenetetrazol (PTZ)-induced seizures. After injection with PTZ, CD2AP^–/– mice exhibited shorter latency to seizures (Fig. 3A) and an increased rate of seizure-associated mortality (Fig. 3B). CD2AP^+/– mice had normal seizure susceptibility. Interestingly, the increased susceptibility to seizures, unlike the behavioral deficits in CD2AP^–/– mice, did not seem related to nephrotic syndrome and systemic illness, as there was no correlation between the severity of nephrotic syndrome and either seizure latency (Fig. 3C) or maximal severity stage (Fig. 3D). Thus, the increased seizure susceptibility appears to reflect a central nervous system (CNS) abnormality. To further investigate excitability, we recorded electroencephalograms before and after PTZ. Neither CD2AP^+/– nor CD2AP^–/– mice had any abnormalities on baseline EEG before injection with PTZ. EEG recordings during seizures were not qualitatively (Fig. 3E) or quantitatively (Fig. 3F and G) different between genotypes.

Figure 3.

CD2AP^–/– mice (with Tg nephrin-CD2AP) are more susceptible to pharmacologically induced seizures, which does not correlate with nephrotic syndrome. (A) CD2AP^–/– mice had decreased latency to PTZ-induced seizures (n = 34–41 mice per group, RMANOVA genotype P = 0.0451, ^+/+ versus ^–/–P = 0.0492 by Dunnett's post hoc, stage 7 P = 0.0208 and stage 8 P = 0.0080 by Dunnett's post hoc). (B) CD2AP deficiency increased overall death rate induced by PTZ (n = 34–41 mice per group, P = 0.0253 by χ² test for trend). (C) No significant correlation was detectable between latency to stage 4 (selected because all CD2AP^–/– mice reached stage 4, giving a common basis for comparison) and proteinuria (n = 17 mice, P = 0.9104, r² = 0.00 by Pearson correlation). (D) No significant correlation was detectable between PTZ max stage and proteinuria (n = 17 mice, P = 0.5711, r² = 0.02 by Pearson correlation). (E) Qualitatively (three representative examples are shown), seizure characteristics monitored by EEG did not detectably vary with CD2AP genotype (^+/+n = 19, ^+/–n = 8, ^–/–n = 16 mice). (F and G) Seizure characteristics monitored by EEG did not detectably vary with CD2AP genotype. (F) The number of spikes on EEG in the 20 min testing window was not detectably different between CD2AP genotypes (^+/+n = 19, ^+/–n = 8, ^–/–n = 16 mice, ANOVA P = 0.8690, error bars indicate SEM). (G) The time for EEG signal to recover after signal saturation from the first seizure, where longer time to recover indicates a more severe seizure, was not detectably different between CD2AP genotypes (^+/+n = 19, ^+/–n = 8, ^–/–n = 16 mice, ANOVA P = 0.3970, error bars indicate SEM).

The shorter latency to seizures without obvious electrographic differences in seizure characteristics, combined with the fact that CD2AP is highly expressed in cerebrovascular endothelium and may function in cell adhesion, prompted us to wonder if CD2AP^–/– mice might have a compromised blood–brain barrier, which would allow PTZ to more rapidly achieve an effective dose in brain tissue. To test this hypothesis, we injected mice retro-orbitally with fluorescein to assess blood–brain barrier integrity (16). Mice were then killed and perfused with saline to clear intravascular fluorescein. Therefore, any fluorescein detectable above low residual levels in brain lysates after perfusion is indicative of blood–brain barrier compromise. CD2AP^–/– mice exhibited increased fluorescein extravasation, indicating increased blood–brain barrier permeability (Fig. 4A). Consistent with the seizure data, CD2AP^+/– mice had normal blood–brain barrier integrity. Importantly, like the seizure phenotypes (and unlike the behavioral deficits that were associated with nephrotic syndrome), the compromised blood–brain barrier in CD2AP^–/– mice did not correlate with proteinuria (Fig. 4B), suggesting that the blood–brain deficit is not secondary to nephrotic syndrome or systemic illness, but rather arises as a direct result of cerebrovascular dysfunction. To more directly test if nephrotic syndrome alone was sufficient to disrupt the blood–brain barrier, we induced nephrotic syndrome chemically with doxorubicin, according to established protocols (17). As expected, and as with CD2AP deficiency, doxorubicin caused proteinuria (Fig. 5A) and weight loss (Fig. 5B). However, doxorubicin did not cause blood–brain barrier disruption (Fig. 5C), again dissociating nephrotic syndrome from the blood–brain barrier. We conclude that CD2AP plays an important role in maintaining the blood–brain barrier, and that this effect is dissociable from its role in maintaining glomerular integrity in the kidney.

Figure 4.

CD2AP^–/– mice (with Tg nephrin-CD2AP) have blood–brain barrier deficits that do not correlate with proteinuria. (A) CD2AP^–/– mice (both high and low proteinuria groups are included) had increased extravasation of sodium fluorescein after intravascular injection and subsequent perfusion (n = 22–25 mice per group, non-parametric Kruskal–Wallis ANOVA P = 0.0149, ^+/+ versus ^–/–P = 0.0205 by Dunn's post hoc, graph shows Tukey box plot), but there was no difference between the amount of extravasation in high versus low proteinuria groups (n = 12 mice/group, P = 0.288 by Mann–Whitney) (B) Sodium fluorescein extravasation did not significantly correlate with nephrotic syndrome in CD2AP^–/– mice (n = 24 mice, P = 0.319, r² = 0.05 by Pearson correlation).

Figure 5.

Doxorubicin-induced nephrotic syndrome does not disrupt the blood–brain barrier. (A) Proteinuria was readily detectable in doxorubicin-treated mice 7 days after injection, indicating induction of nephrotic syndrome. (B) Weight loss was also observed after doxorubicin treatment (n = 4 Veh. and 8 Dox. mice, P = 0.045 by t-test). (C) Sodium fluorescein extravasation in doxorubicin-treated mice did not differ from vehicle-injected mice (n = 4 Veh. and 8 Dox. mice, P = 0.804 by Mann–Whitney).

Discussion

To address how CD2AP may contribute to AD risk, we investigated the normal function of CD2AP in the brain by characterizing CD2AP-deficient mice. We found no behavioral abnormalities in CD2AP^+/– or CD2AP^–/– mice, except for reduced locomotor activity and increased anxiety in CD2AP^–/– mice with severe nephrotic syndrome, most likely reflecting their systemic illness. On the other hand, we found impaired blood–brain barrier integrity in CD2AP^–/– mice, which was associated with shorter latency to pharmacologically induced seizures, and these abnormalities were not correlated with nephrotic syndrome. Furthermore, induction of nephrotic syndrome with a nephrotoxic drug was not sufficient to disrupt the blood–brain barrier. Our data indicate that CD2AP plays a key role in the adult brain in maintaining the blood–brain barrier, a role that is likely independent from its role in maintaining glomerular integrity in the kidney.

These observations are consistent with several prior observations about CD2AP. CD2AP is highly expressed in cerebrovascular endothelial cells (7–9,11) which help form the blood–brain barrier (18). This proposed role of CD2AP is consistent with the finding that CD2AP maintains the integrity of adherens junctions (6), key components of the blood–brain barrier (19). CD2AP, through its actin-binding domain, regulates actin dynamics in several cell types (4,20), and the actin cytoskeleton plays an important role in adherens junctions.

A role for CD2AP in blood–brain barrier integrity could explain its genetic link to AD. Several other late-onset AD risk factors have been implicated in blood–brain barrier integrity or pathways, including APOE (21), PICALM (22), INPP5D and CLU (23). Blood–brain barrier breakdown has been reported in AD (24) and was recently demonstrated to occur in humans with aging (25). It is possible that loss-of-function variants in CD2AP may contribute to genetic risk of AD by facilitating age-related blood–brain barrier breakdown. This would link CD2AP and other genetic risk factors for AD to a growing body of evidence implicating vascular abnormalities as contributors to cognitive impairment and dementia, including AD (reviewed in 26).

We used CD2AP^–/– mice to probe the function of CD2AP, but should emphasize that it is not yet clear that the polymorphisms associated with AD reduce CD2AP expression. CD2AP expression is not associated with several measures of AD (27,28), which may be due to the relative scarcity of the allele associated with AD (29). Further studies on rare variants or utilizing epigenomic techniques are needed to determine the functional effects of genetic variants in CD2AP and test the hypothesis that loss of CD2AP function contributes to AD risk (30,31).

A vascular role for CD2AP is consistent with the fact that CNS expression of CD2AP is highest in the endothelium, but recent reports have suggested other possible roles for CD2AP in AD. Knockdown of CD2AP decreased the ratio of secreted Aβ42/Aβ40 in neuroblastoma cells, although only a subtle decrease in Aβ42/Aβ40 ratio was observed in 1-month-old mice APP/PS1 mice lacking CD2AP and no effect on plaque load was observed in 7-month-old APP/PS1 mice haploinsufficient for CD2AP (13). However, that study did not look specifically at Aβ in the vasculature, which may be important to assess given the findings here. CD2AP deficiency also exacerbated tau-mediated toxicity in Drosophila (12). As pathological tau can induce blood–brain barrier deficits, these pathways may not be mutually exclusive (32), and further research will be necessary to determine whether and how these observation might be linked to vascular changes.

In conclusion, the evidence provided here indicates that CD2AP helps maintain the blood–brain barrier and supports further investigation of vascular roles for CD2AP in AD models. In particular, cell-type specific knockout models would be of particular value in further elucidating both the endogenous role of CD2AP relevant to the CNS and its contribution to AD risk.

Materials and Methods

Mice

CD2AP^–/– mice with a nephrin promoter-driven CD2AP transgene that expresses in the kidney and muscle, but not brain (14), were obtained from the laboratory of Dr Andrey Shaw at Washington University. We bred CD2AP^+/– mice expressing this transgene together to obtain CD2AP^+/+ (wild-type), CD2AP^+/– and CD2AP^–/– siblings. Mice in this study were aged 2.5–4.5 months. All mice for CD2AP experiments expressed the nephrin-CD2AP transgene to reduce early mortality due to renal disease. Mice were on a congenic C57BL/6J background, with the exception of mice used for doxorubicin-induced nephrotic syndrome, which were on a BALB/cJ background, as C57BL/6J mice are known to be resistant to doxorubicin-induced nephrotic syndrome whereas BALB/cJ mice are susceptible (33). An approximately equal number of male and female mice were used in all experiments and no sex-dependent effects were observed. Mice were housed in a pathogen-free barrier facility on a 12-h light/dark cycle with ad libitum access to water and food (NIH-31 Open Formula Diet, #7917, Harlan). For postmortem analyses, mice were anesthetized by Fatal-plus (Vortech) and perfused with 0.9% saline. Brains were then removed, dissected into hemi-brains, and either drop-fixed in 4% PFA for 48 h before storage in phosphate-buffered saline (PBS) at 4°C, or flash-frozen on dry ice for storage at −80°C. All experimental protocols were approved by the Institutional Animal Care and Use Committee of the University of Alabama at Birmingham.

Antibodies and western blotting

Samples were prepared as previously described (34). Blots were probed with 1:200 rabbit anti-CD2AP clone H-290 (Santa Cruz) or 1:500 rabbit anti-CD2AP HPA003326 (Sigma), and 1:10 000 mouse anti-α-Tubulin clone B-5-1-2 (Sigma T5168). Fluorescent secondary antibodies (LICOR) were used along with an Odyssey infrared scanner (LICOR).

Proteinuria quantitation

Proteinuria was quantified by running 1 µl of urine per mouse on reducing NuPAGE gels and staining with SimplyBlue Safestain (Life Technologies). BSA standards (Thermo Scientific) were run in parallel and used for quantitation in ImageJ (NIH).

Behavioral testing

Behavior testing for elevated plus maze, open field and fear conditioning was performed as previously described (34,35), except that mice were allowed to explore for 5 min in the elevated plus maze in this study. Light/dark testing was conducted in a chamber 20 cm L × 40.5 cm W × 22 cm H, which was split into one-third dark (a black Plexiglas box) or two-thirds exposed to room lights. Mice were allowed to freely explore for 10 min and were scored by a video tracking system (CleverSys).

Electroencephalography

Electrode implantation surgery and EEG recordings were performed as previously described (34).

PTZ-induced seizures

PTZ (Sigma) was administered intraperitoneally in PBS at a dose of 40 mg/kg and scored on an 8 point severity scale as described previously (36).

Blood–brain barrier integrity assessment by fluorescein

Sodium fluorescein (Sigma F6377) was administered retro-orbitally at a dose of 50 mg/kg in PBS using an insulin syringe (Fisher 14-826-79). Fluorescein was allowed to circulate for 1 h, at which point mice were killed and perfused as earlier. Extravasation of sodium fluorescein was measured after homogenization of one hemi-brain in 800 µl 60% 6.1 N trichloroacetic acid (Sigma) in MilliQ water. Lysates were cleared by centrifugation at 18 000g for 20 min, and read by dispensing 50 µl of cleared lysate in triplicate into 384-well plates and comparing to a standard curve of sodium fluorescein dissolved in 60% trichloroacetic acid on a Synergy 2 plate reader (Biotek) by excitation at 485/20 nm and emission at 528/20 nm.

Doxorubicin-induced nephrotic syndrome

The DNA intercalator doxorubicin consistently induces nephrotic syndrome in mice within 1 week (17) (reviewed in 37). Doxorubicin was administered retro-orbitally at 10 mg/kg in 0.9% saline using an insulin syringe (Fisher 14-826-79).

Statistics

All of the datasets, except the sodium fluorescein extravasation data, were normally distributed. Therefore, unless otherwise noted, data are presented as mean ± standard error of the mean and the appropriate one- or two-way ANOVA with multiple corrections post hoc was performed. For survival analysis, the Log-rank Mantel-Cox test was used to separately compare CD2AP^+/+ to CD2AP^–/– or CD2AP^+/– mice. The non-normal distribution of the sodium fluorescein extravasation data required evaluation by Kruskal–Wallis ANOVA with Dunn's post hoc. For correlation of sodium fluorescein data, the values were Log₁₀ transformed prior to correlation to proteinuria data because of the large range of fluorescein data. All statistical tests were performed with Prism 6 (GraphPad).

Funding

This work was supported by National Institutes of Health grant R01NS075487 to E.D.R.

Acknowledgements

We thank Dr Andrey Shaw (Washington University) for the kind gift of CD2AP-deficient mice with nephrin promoter-driven CD2AP transgene. We thank James Black, Miriam Roberson, Hayden Pacl, Vincent Onyilo, and Andrew Arrant for excellent technical assistance, and other members of the Roberson Laboratory for critical reading of the manuscript.

Conflict of Interest statement. None declared.