Introduction Insulin-resistant type 2 diabetes mellitus (DM) leads to premature death and disability, primarily as a consequence of accelerated vascular disease. Shc homology 2-containing inositol 5 phosphatase-2 (SHIP2) is a lipid phosphatase that suppresses insulin signalling downstream of phosphoinositide-3-kinase (PI3K). Inhibition of SHIP2 has been proposed as a therapy for type II DM, but the potential impact on vascular function is currently undefined.

Methods Mice with endothelium-specific deletion of the SHIP2 catalytic domain (ECSHIP2/+) were generated by crossing mice with LoxP sites flanking exons 18–19 of the Inppl gene with Tie2-Cre mice. Pulmonary endothelial cells (PECs) were isolated from lungs using anti-CD146 microbeads. SHIP2 knockdown was studied in human umbilical vein endothelial cells (HUVECs) by transducing with SHIP2 (or control) shRNA. Immunoblotting was performed in basal conditions and after insulin stimulation, using appropriate primary antibodies. Superoxide generation was quantified using dihydroethidium (DHE) fluorescence, and endothelial nitric oxide synthase (eNOS) activity via a 13C-l-arginine to 13C-l-citrulline conversion assay (quantified as% of baseline). Aortic hydrogen peroxide abundance was assessed using an Amplex Red assay. Vasomotor function was studied ex vivo in aortic rings. Statistical analysis was performed with unpaired and paired t-tests assuming unequal variance, with significance defined by p-values<0.05.

Results PECs derived from ECSHIP2Ã&squ;â€&noentity;/+ mice had increased basal and insulin-stimulated activation of downstream signalling intermediates, including pPDK1, S473 pAKT and S1177 peNOS. Nox2 protein expression was increased (1.19 fold; p=0.04), in association with increased superoxide abundance (2.20 fold; p=0.004). This was normalised both with the Nox2 specific inhibitor Gp91ds-tat (22% reduction; p=0.05) and PI3K inhibitors, Wortmannin (35% reduction; p=0.03) and LY294002 (26% reduction; p=0.03). SHIP2 transduction in HUVECs achieved~70% knockdown. Findings were recapitulated, with more abundant S473 pAKT and S1177 peNOS. Moreover, there was enhanced Nox2 expression (1.39 fold; p=0.41) and superoxide generation (1.42 fold; p=0.02), which was suppressed in the context of Nox2 (34% reduction; p=0.03) and PI3K inhibition (39% reduction; p=0.01 [Wortmannin], 13% reduction; p=0.05 [LY294002]). eNOS activity was reduced in ECSHIP2/+ PECs treated with insulin (114% [±6] vs 136% [±3]; p=0.01). Aortic rings from ECSHIP2/+ mice had blunted insulin-mediated vasodilation, increased hydrogen peroxide abundance, and impaired vasoconstriction in response to the non-selective NOS inhibitor L-NMMA (indicating reduced NO bioavailability).

Conclusions Endothelial-specific SHIP2 inactivation causes PI3K- and Nox2-dependent oxidative stress and endothelial dysfunction. These data suggest that SHIP2 may not be an ideal therapeutic target for diabetes-associated vascular disease.