Inhibition of SHIP2 in CD2AP-deficient podocytes ameliorates reactive oxygen species generation but aggravates apoptosis

Lack of CD2-associated protein (CD2AP) in mice increases podocyte apoptosis and leads to glomerulosclerosis and renal failure. We showed previously that SHIP2, a negative regulator of the PI3K/AKT signalling pathway, interacts with CD2AP. Here, we found that the expression level and activity of SHIP2 and production of reactive oxygen species (ROS) are increased in cultured CD2AP knockout (CD2AP−/−) mouse podocytes. Oxidative stress was also increased in CD2AP−/− mouse glomeruli in vivo. We found that puromycin aminonucleoside (PA), known to increase ROS production and apoptosis, increases SHIP2 activity and reduces CD2AP expression in cultured human podocytes. PDK1 and CDK2, central regulators of AKT, were downregulated in CD2AP−/− or PA-treated podocytes. Downregulation of PDK1 and CDK2, ROS generation and apoptosis were prevented by CD2AP overexpression in both models. Notably, inhibition of SHIP2 activity with a small molecule inhibitor AS1949490 ameliorated ROS production in CD2AP−/− podocytes, but, surprisingly, further reduced PDK1 expression and aggravated apoptosis. AKT- and ERK-mediated signalling was diminished and remained reduced after AS1949490 treatment in the absence of CD2AP. The data suggest that inhibition of the catalytic activity of SHIP2 is beneficial in reducing oxidative stress, but leads to deleterious increase in apoptosis in podocytes with reduced expression of CD2AP.

Thus far the functional interrelationship of CD2AP and SHIP2 in podocytes has remained unclear. Here, we found that CD2AP-deficient podocytes, characterized by increased expression of TGF-β1 and apoptosis 9 , show elevated level of ROS and increased expression and activity of SHIP2. We then hypothesized that inhibition of SHIP2 activity reduces oxidative stress and apoptosis in the absence of CD2AP. Notably, we observed that inhibition of SHIP2 activity with a specific small molecule inhibitor AS1949490 reduces ROS, but contrary to our expectations, increases podocyte apoptosis in the absence of CD2AP. This suggests that inhibition of the catalytic activity of SHIP2 provides a means to reduce oxidative stress in podocytes, but aggravates rather than ameliorates podocyte injury by increasing apoptosis when CD2AP is depleted.

Results
Absence of CD2AP in mouse podocytes induces ROS production and apoptosis. To analyse whether CD2AP and SHIP2 play a role in oxidative stress-induced apoptosis in podocytes, we first analysed whether lack of CD2AP enhances ROS production. 2′, 7′-Dichlorofluorescein diacetate (DCFH-DA) fluorescent probe assay revealed that lack of CD2AP increases ROS production by 25% (Fig. 1A), and increases apoptosis by 2.8-fold compared to wild type cells (Fig. 1B) as shown by fluorescence-activated cell sorting (FACS) with annexin V and 7-AAD, labelling apoptotic and necrotic cells, respectively. Reintroducing CD2AP back to the knockout cells, even at low level, reduced apoptosis to the same level as in wild type cells (Fig. 1B,C). In line with an increase in apoptosis, we observed that threonine 308 (T308) phosphorylation of AKT was 30% lower at basal state in the absence of CD2AP compared to wild type podocytes (Fig. 1D,E. Supplemental Fig. S1A). In contrast, we did not observe any difference in the phosphorylation of serine 473 (S473) of AKT between WT and CD2AP−/− podocytes (Fig. 1F). We also observed that phosphorylation of extracellular-signal related kinase (ERK), a member of the mitogen-activated protein kinase (MAPK) family that regulates cell survival and apoptosis (reviewed in refs 14 and 15), was significantly lower in podocytes lacking CD2AP (Fig. 1G).
Inhibition of SHIP2 activity reduces ROS production in CD2AP−/− podocytes. We next defined whether CD2AP and SHIP2 are functionally interrelated, and found that the expression of SHIP2 was increased by 30% in CD2AP−/− podocytes ( Fig. 2A,B). In line with this, SHIP2 activity was 26% higher in CD2AP−/− podocytes compared to wild type cells (Fig. 2C). Treatment of CD2AP−/− podocytes with AS1949490, a SHIP2-specific small molecule inhibitor 16 , reduced SHIP2 activity by 33% compared to wild type podocytes, and by 47% compared to CD2AP−/− podocytes without treatment (Fig. 2C). The differences detected in the activities were not due to different amounts of immunoprecipitated SHIP2 used for the malachite green phosphate assay (Figs 2D and S1B). AS1949490 also reduced ROS production in CD2AP−/− podocytes from 122% back to the level observed in wild type cells (Fig. 2E). Contrary to our expectations, inhibition of SHIP2 did not ameliorate apoptosis caused by lack of CD2AP but rather increased it by 40% (Fig. 2F).
To define the reason for this, we investigated the expression of 3-phosphoinositide-dependent protein kinase 1 (PDK1) and cyclin-dependent kinase 2 (CDK2), both shown to induce apoptosis in podocytes when knocked down 17,18 . Interestingly, the expression of PDK1 was downregulated by 35% and CDK2 by 65% in CD2AP−/− podocytes when compared to wild type cells (Fig. 2G,H). Notably, AS1949490 treatment led to further downregulation of PDK1 expression in CD2AP−/− podocytes but not in wild type podocytes (Fig. 2I). Collectively, PDK1 and CDK2 are downregulated in the absence of CD2AP and may contribute to increased apoptosis. Inhibition of SHIP2 further reduces the total amount of PDK1 in the absence of CD2AP, which may aggravate apoptosis.
To further investigate the signalling pathways associated with an increase in apoptosis, we determined the activity of AKT and ERK after AS1949490 treatment. T308 phosphorylation of AKT, shown to be mediated by PDK1 19 , was increased in both WT and CD2AP−/− podocytes after the AS1949490 treatment (Fig. 3A,B). In contrast, AS1949490 increased S473 phosphorylation of AKT only in WT podocytes but not in CD2AP−/− podocytes (Fig. 3C). As inhibition of SHIP2 increases the amount of PI(3,4,5)P 3 which activates PDK1 19-21 , we investigated whether inhibition of SHIP2 increases the activity of PDK1. AS1949490 treatment increased the phosphorylation of serine 241 residue of PDK1, mandatory for the activation of PDK1 22 , in both WT and CD2AP−/− podocytes (Fig. 3D,E). Phosphorylation of ERK was increased in both cell types after AS1949490 treatment but significantly less in CD2AP−/− podocytes compared to WT podocytes (Fig. 3F).
Lack of CD2AP in mice in vivo enhances ROS production in glomeruli. As lack of CD2AP increased oxidative stress in cultured podocytes (Fig. 1A), we stained kidneys of three weeks old wild type and CD2AP−/− mice for 8-hydroxyguanosine (8-OHdG), a marker of oxidative damage (Fig. 4A,B). Quantification of the signal in glomeruli revealed 3.5-fold increase in ROS generation in the absence of CD2AP (Fig. 4C).
PA-treatment downregulates CD2AP, upregulates SHIP2 and increases ROS production and apoptosis in cultured human podocytes. Since puromycin aminonucleoside (PA), a highly potent podocyte toxin, is known to cause podocyte injury by induction of DNA damage mediated via ROS 23 and to increase apoptosis in a time-and dose-dependent manner 24 , we treated differentiated human podocytes in culture with PA   and analysed the effect on CD2AP and SHIP2. PA-treatment of podocytes for 48 hours decreased the expression level of CD2AP by 80% (Fig. 5A,B), and increased SHIP2 expression by 34% (Fig. 5C) and its phosphatase activity by 45% (Fig. 5D). PA increased ROS generation by 60% as shown by DCFH-DA fluorescent probe assay (Fig. 5E). FACS analysis with annexin V and 7-AAD double labelling indicated that the level of podocyte apoptosis was increased by 5-fold after PA-treatment ( Fig. 5F) confirming previous findings 24 .
SHIP2 inhibition in PA-treated cultured human podocytes does not reduce ROS production or apoptosis. We further tested whether inhibition of SHIP2 reduces ROS production and apoptosis induced by PA. We first confirmed that AS1949490 treatment prevents the increase in the activity of SHIP2 caused by PA (Fig. 6A). As expected, the activity of SHIP2 remained at the level of the control when podocytes were treated simultaneously with PA and AS1949490 for 48 hours (Fig. 6A). However, this did not prevent an increase in ROS production ( Fig. 6B) or rescue podocytes from apoptosis (Fig. 6C).
CD2AP overexpression reduces ROS production and apoptosis induced by PA in cultured human podocytes. To define whether CD2AP is able to rescue podocytes from the effects of PA, including downregulation of PDK1 and CDK2 17 , we overexpressed CD2AP simultaneously with PA-treatment of cultured human podocytes. PA-treatment reduced the expression of PDK1 by 80% and CDK2 by 40% compared to the empty vector (Fig. 7A,B), and overexpression of CD2AP significantly prevented PA-induced downregulation of both PDK1 and CDK2 (Fig. 7A,B). Overexpression of CD2AP alone had no significant effect on the expression level of any of the analysed proteins (Fig. 7A,B).
We then defined whether reintroduction of CD2AP rescues podocytes from PA-induced ROS production and apoptosis. PA-treatment increased ROS production by 64% compared to empty vector-transfected podocytes (Fig. 7C), and indeed, overexpression of CD2AP simultaneously with PA-treatment significantly reduced the increase in ROS generation, leading to only 19% increase (Fig. 7C). PA-treatment increased apoptosis by 2.9-fold compared to non-treated podocytes transfected with empty vector (Fig. 7D). The apoptosis rate of PA-treated podocytes overexpressing CD2AP was increased only by 1.5-fold, which did not significantly differ from the rate of apoptosis observed in empty vector-transfected cells (Fig. 7D). Overexpression of CD2AP alone caused no significant difference in ROS generation or apoptosis compared to empty vector-transfected podocytes (Fig. 7C,D). Collectively, the data show that reintroduction of CD2AP back to podocytes treated with PA, showing downregulation of CD2AP, is able to revert an increase in ROS production and apoptosis caused by PA.

PA-treatment reduces CD2AP and increases SHIP2 expression in PA-induced nephrotic rats.
To define whether PA-treatment leads to downregulation of CD2AP and upregulation of SHIP2 in vivo, we immunostained kidneys of rats with PA-induced nephrosis. The rats had significantly increased proteinuria 9 days after PA administration 25 , and as previously shown 25 , greatly diminished expression of nephrin 10 days after induction of nephrosis (Fig. 8B,E,H,K). Double immunofluorescence staining for nephrin and CD2AP or SHIP2 revealed that the expression of CD2AP appears diminished (Fig. 8A-F) and SHIP2 increased (Fig. 8G-L) in podocytes in rats with PA-induced nephrosis, confirming the data obtained in cultured cells.

Discussion
CD2AP-deficient podocytes are more prone to undergo apoptosis than wild type podocytes 9 . Here, we show that lack of CD2AP in podocytes induces oxidative stress, which may be one of the contributing factors leading to an increase in apoptosis. We also observed that the activity of SHIP2, an interaction partner of CD2AP 11 , is increased in the absence of CD2AP. Notably, we further found that inhibition of SHIP2 activity ameliorates oxidative stress caused by lack of CD2AP. However, SHIP2 inhibition alone was not able to rescue CD2AP-deficient podocytes from apoptosis.
We showed previously that CD2AP associates with SHIP2 11 , and here we found that SHIP2 expression and activity are upregulated in CD2AP−/− podocytes. Together these data suggest that CD2AP has an inhibitory role on SHIP2. Our previous data revealed that overexpression of SHIP2 in podocytes reduces AKT activation and enhances apoptosis 11 , suggesting that increased expression of SHIP2 could contribute to increased apoptosis in CD2AP−/− podocytes. Elevated SHIP2 increases apoptosis also in neuronal cells in SHIP2 overexpressing mice 26 and hepatocytes 13 . In line with the elevated activity of SHIP2 in CD2AP−/− podocytes, we observed that the activity of the AKT and ERK cell survival pathways at basal state were reduced (as visualized by reduced AKT and ERK phosphorylation) in the absence of CD2AP reflecting the increased rate of apoptosis. This is in line with earlier literature showing that CD2AP interacts with p85 subunit of PI3K and, together with podocyte proteins nephrin and podocin, stimulates the PI3K/AKT pathway 27 . The data are also consistent with a previous report revealing reduced base-line activation of ERK in the absence of CD2AP 28 . Reduced ERK activation may be due to an increase in SHIP2 expression, as a previous study showed decreased insulin-stimulated ERK activation in SHIP2 overexpressing cells 12 .
Interestingly, we found that lack of CD2AP reduces the expression of PDK1, a central activator of AKT 19 , providing a further link between CD2AP and cell survival signalling (Fig. 9). Pancreatic β cells depleted of PDK1 are more prone to undergo apoptosis 29 and in line with this, knockdown of PDK1 in podocytes increases apoptosis by inhibiting antiapoptotic and stimulating proapoptotic pathways 18 . Here, we observed that lack of CD2AP also reduces the expression of CDK2, a cell cycle regulatory protein previously shown to activate AKT during  30 . Notably, knockdown of CDK2 downregulates PDK1, decreases activation of AKT and increases apoptosis 17 . Furthermore, knockdown of PDK1 downregulates CDK2 and thus proposes a regulatory loop between the proteins 17 . The data suggest that PDK1 and CDK2 play central roles in CD2AP-mediated control of the AKT cell survival pathway, but the exact mechanisms leading to their downregulation in the absence of CD2AP require further studies.
Excess generation of ROS has been shown to lead to apoptosis and progression of kidney injury 31,32 . Our study shows that an increase in ROS may be one of the mediators of podocyte injury in the absence of CD2AP, as we observed increased ROS in CD2AP−/− kidneys at the age of three weeks, which coincides with an increase in podocyte apoptosis and the development of albuminuria and glomerular lesions reminiscent of FSGS 9 . A recent study proposed that podocytes bind plasminogen and activate it, which then evokes production of superoxide anion (a type of ROS) that further contributes to podocyte injury 33 . Interestingly, the authors found increased concentrations of plasminogen in the urine of CD2AP knockout mice which could lead to increased ROS in CD2AP-deficient kidneys 33 , in line with our findings. Resembling CD2AP-deficient podocytes, PA-treatment leads to downregulation of CD2AP and PDK1 (this study) and increases ROS generation 23 .
The key finding of this study is that inhibition of SHIP2 activity with a small-molecule inhibitor AS1949490 in the absence of CD2AP reduces ROS generation (Fig. 9). AS1949490 is highly specific towards SHIP2 and does not inhibit other phosphatases that regulate the PI3K-pathway such as SHIP1, a close homologue of SHIP2, or phosphatase and tensin homolog (PTEN), a lipid 3′-phosphatase 16 . Previous studies in hepatocytes indicated that catalytically inactive form of SHIP2 protects against palmitate or high glucose -induced excessive production of ROS and apoptosis 13,34 . In the absence of CD2AP, however, inhibition of SHIP2 activity alone was not able to rescue podocytes from apoptosis, but rather increased it. This could be due to the fact that SHIP2 inhibition increased only T308 but not S473 phosphorylation of AKT in CD2AP-deficient cells (Fig. 9), which is not enough to fully activate AKT 35 . Phosphorylation of T308 may sound contradictory as phosphorylation of this site occurs by PDK1 19 and PDK1 is downregulated in the absence of CD2AP. However, as the catalytic activity of SHIP2 is to hydrolyse PI(3,4,5)P 3 to PI(3,4)P 2 , SHIP2 inhibition is expected to lead to an increase in PI(3,4,5)P 3 in cells similarly as after SHIP2 knockdown 36 . Increase in PI(3,4,5)P 3 , in turn, may lead to activation of PDK1 (as we observed here), in line with studies showing that PDK1 is activated by both PI(3,4,5)P 3 and PI(3,4)P 2 19-21 , but that it binds PI(3,4,5)P 3 more strongly than PI(3,4)P 2 21 . PI(3,4,5)P 3 also binds AKT, apparently altering its conformation, and thereby enhances its binding to PDK1 contributing to an increase in T308 phosphorylation 37 . Furthermore, even though AS1949490 treatment slightly increased the phosphorylation of ERK in the absence of CD2AP, its activity remained lower compared to WT podocytes treated with AS1949490. In a similar fashion, a previous study showed an increase in AKT and ERK phosphorylation by AS1949490 in lymphatic endothelial cells 38 . Interestingly, the authors found that knockdown of SHIP2 in these cells, contrary to expectations, increased apoptosis 38 , resembling our finding here that SHIP2 inhibition in the absence of CD2AP increases apoptosis. As ERK activation may also trigger apoptosis 14 and ERK activation has been associated with podocyte injury 39 , it is also possible that SHIP2 inhibition in the absence of CD2AP triggers proapoptotic pathways leading to reduced cell survival (Fig. 9).
The inability of SHIP2 inhibition to rescue CD2AP-deficient podocytes from apoptosis does not rule out the potential of SHIP2 inhibitors to reduce ROS and apoptosis in proteinuric diseases presenting normal expression levels of CD2AP. Increased levels of ROS have been observed, for example, in podocytes in diabetic kidney disease 2 . Our previous finding that SHIP2 is upregulated in glomeruli in diabetic rodent models 11 together with this study suggests that elevated SHIP2 could contribute to an increase in ROS in podocytes in diabetes. In future studies it will be of interest to define whether inhibition of SHIP2 in diabetic kidney disease reduces the increased levels of ROS, and thereby potentially prevents or slows down the progression of the disease.
Reduced expression of CD2AP apparently associates with multiple pathways leading to podocyte injury, shown here with both CD2AP-deficient podocytes and the PA model of podocyte injury. The latter was our model of choice to be used side by side with CD2AP−/− podocytes, as the PA-induced nephrosis in rats is a commonly used model for FSGS 40 , and the glomerular changes observed in CD2AP knockout mice resemble those observed in human FSGS 5 . In addition, mutations in CD2AP, detected in a subgroup of patients with FSGS, lead to reduced expression of CD2AP [6][7][8] . The molecular changes, including decrease of CD2AP, PDK1 and CDK2 expression, and increase of SHIP2 activity, oxidative stress and apoptosis, were shared between the two models. In the harsh PA injury model, apparently affecting a multitude of proteins in addition to CD2AP, inhibition of SHIP2 did not rescue podocytes from excess ROS generation and apoptosis, whereas overexpression of CD2AP did. In CD2AP−/− podocytes, SHIP2 inhibition did reduce ROS but aggravated apoptosis. Resembling this, we observed previously increased activity of β-catenin in podocytes in the absence of CD2AP, but found that inhibition of the Wnt/β-catenin pathway did not ameliorate but further aggravated kidney injury 41 . These data reinforce the central role of CD2AP as a regulator of multiple pathways in podocytes and reveal that inhibition of the harmful pathways may have unexpected adverse effects.
Collectively, SHIP2 inhibitors may be effective in preventing excessive generation of ROS. However, this may not be a suitable approach to treat kidney diseases associated with reduced expression or lack of CD2AP. Further studies are necessary to define the applicability of SHIP2 inhibition as a part of therapy in kidney diseases presenting with oxidative stress, such as diabetic kidney disease.  Induction and detection of apoptosis. Differentiated human podocytes were exposed to PA (Sigma-Aldrich, 50 µg/ml) for 48 hours. After 24 h, podocytes were treated with 10 µM AS1949490 or DMSO (diluent) as a control. CD2AP−/− and wild type podocytes were treated with 10 µM AS1949490 or DMSO for 24 h prior to detection of apoptosis. Apoptosis was detected by flow cytometry using annexin V-FITC and 7-AAD double staining with FACSAria (BD Biosciences, Franklin Lakes, NJ) or CyAn ADP (Beckman Coulter, Brea, CA). Cells positive for annexinV-FITC and negative for 7-AAD were deemed apoptotic. A total 1 × 10 4 cells were detected by FACS in each sample.
Immunoblotting and antibodies. Immunoblotting was performed as described previously 11 . Primary antibodies used were mouse monoclonal anti-8-hydroxyguanosine (8-OHdG) (15A3), rabbit polyclonal anti-CD2AP H-290, mouse monoclonal anti-CDK2 and goat anti-SHIP2 I-20 IgGs (Santa Cruz Biotechnology, Dallas, Texas, USA), rabbit polyclonal, anti-phospho-PDK1 (Ser241), anti-PDK1, anti-phospho-p44/p42 MAPK (T202/Y204), anti-p44/p42 MAPK (referred to as ERK), and anti-phospho-AKT (Thr308) IgGs, and mouse monoclonal anti-phospho-AKT (Ser473) IgG (Cell Signaling Technology, Danvers, MA), mouse monoclonal anti-Pan AKT IgG (R&D Systems, Minneapolis, MN) and mouse monoclonal anti-tubulin IgG (Sigma-Aldrich). Rabbit polyclonal anti-CD2AP IgG raised against mouse CD2AP was described previously 43 . Alexa Fluor 680 (Invitrogen, Carlsbad, CA, USA) and IRDye 800 (LI-COR, Lincoln, NE) donkey anti-rabbit, anti-goat or anti-mouse IgGs were used as secondary antibodies. Detection and quantification was performed with an Odyssey Infrared Imager (LI-COR). Full-length lanes of blots probed with antibodies against SHIP2, CD2AP, panAKT, p-AKT (Thr308), PDK1 and p-PDK1 (Ser241) are shown in Supplemental Fig. S2. CD2AP associates with SHIP2. CD2AP inhibits the production of ROS, and SHIP2 negatively regulates the pathways leading to AKT and ERK phosphorylation. The balance between the prosurvival and proapoptotic signals is maintained. (B) CD2AP−/− podocytes in basal state. The expression and activity of SHIP2 and generation of ROS are increased in the absence of CD2AP. Increased SHIP2 activity leads to low phosphorylation level of AKT and ERK. The expression of PDK1 is downregulated contributing to degreased phosphorylation of T308 of AKT. Decrease in prosurvival signalling manifests as increased apoptosis of the cells. (C) Inhibition of SHIP2 activity with AS1949490 in CD2AP−/− podocytes. Inhibition of the activity of SHIP2 leads to reduced generation of ROS. AS1949490 attenuates the potential of SHIP2 to negatively regulate AKT and ERK activity, yet the phosphorylation of S473 of AKT does not increase. Despite of low expression level of PDK1, the phosphorylation of T308 of AKT increases. As phosphorylation of both sites of AKT is required for its full activity, remains the balance between prosurvival and proapoptotic signalling disrupted and podocytes undergo apoptosis. It is also possible that increased ERK activation may contribute to an increase in apoptosis (see Discussion for details).