COX-2 in the Kidney: Good, BAD or Both?

It is well established that non-steroidal anti-inflammatory drugs (NSAID) use can result in renal dysfunction but the mechanisms underlying the pathology are poorly understood. In this issue, a report by Kuper et al. provides new insights about the role of cyclooxygenase (COX)-2-in the adaptative process of renal medullary cells to hypertonic stress¹.

COX-2 is present throughout the different regions of the kidney and promotes the production of prostaglandins and prostanoids. The most common action ascribed to COX-2 is the conversion of arachidonic acid to an intermediate endoperoxide that is converted to PGE₂, PGI₂, TXA₂ or PGD₂ through the actions of specific synthases. The benefits or dangers of inhibiting COX-2 depend on the role of these prostaglandins within the kidney. Since Kuper and colleagues focus on the role of PGE₂ in the renal medullary response to hypertonicity, this commentary also will be limited to the signaling pathways and actions of PGE₂.

Cellular responses to PGE₂ are mediated by four subtypes of PGE₂-specific receptors, designated EP1-EP4². EP1 increases intracellular Ca²⁺. EP3 interacts with the G_i heterotrimeric G protein and attenuates cAMP production. Conversely, EP2 and EP4 are coupled to G_s which activates adenylyl cyclase, resulting in increased production of cAMP. PGE₂, therefore, can influence a variety of physiological functions in the kidney. It was found to regulate the amounts of Na-K-ATPase in the inner medulla and Na-ATPase and organic ion transporters in the proximal tubule. PGE₂ and EP4 are reportedly involved in podocyte injury and impairment of glomerular barrier function. PGE₂ with EP2 or EP4 has been directly implicated in the release of renin from the macula densa and a variety of vasodilatory effects throughout the kidney. Conversely, PGE₂ and EP1 have been shown to induce vasoconstriction. Apart from transport, filtration and vascular effects, PGE₂ also can modulate anti-fibrosis and epithelial cell injury responses. PGE₂ in the kidney, therefore, exerts effects that span from protective to deleterious depending largely on the receptor subtype that is activated but also on other inputs such as hormone signals, substrate availability or duration of exposure. Excellent reviews are available for an introduction into the literature on prostaglandins in the kidney^2,3.

Over the past decade, the renoprotective functions of prostaglandins in the kidney have become evident but our understanding of the responsible mechanism has been slower to emerge. Neuhofer and colleagues have contributed significantly to this area by examining the role of PGE₂ in the responses of renal medullary cells to hypertonic stress such as occurs during dehydration or antidiuresis. Even though the collecting duct is always hypertonic relative to serum, dehydration can increase the interstitial osmolality several fold to levels that, if prolonged, lead to cellular distress and death by apoptosis unless intercellular adaptions that enable survival occur (reviewed by Burg et al.⁴). In an earlier paper, Steinert et al. demonstrated that one such adaptive response to hyperosmolality in renal medullary cells is the induction of COX-2 through a feedback loop signaling pathway involving the EP2 receptor and a cAMP/PKA/CREB response⁵. Interestingly, they also found that hyperosmolar conditions cause the activation of PKA which, in turn, phosphorylates Ser-155 in the pro-apoptotic protein BAD (Figure 1). This observation was the first hint about how COX-2 protects inner medullary cells from apoptosis. Pro-apoptotic BAD binds to anti-apoptotic members of the Bcl-2 protein family like Bcl-xL. When active, these anti-apoptotic proteins prevent the release of cytochrome C from mitochondria, disabling the formation of an apoptosome complex which otherwise would activate the caspase cascade leading to cell death. Phosphorylation of BAD prohibits the association with the Bcl-2 family members thus allowing their anti-apoptotic function and promoting cell survival.

Figure 1.

Hypertonic stress signaling via PGE₂. The diagram depicts the signaling pathway proposed by Kuper et al. for PGE₂-mediated inactivation of the pro-apoptotic protein, BAD. Hypertonicity increases cyclooxygenase-2 (COX-2) activity which catalyzes the production of unstable prostaglandin H₂ (PGH₂) from arachidonic acid. PGH₂ is rapidly converted to prostaglandin E₂ (PGE₂) by PGE synthase (PGES). PGE₂ binds to its receptor, EP2 which is coupled to Gs_α. Gs_α activates adenylyl cyclase which produces cAMP, leading to activation of protein kinase A (PKA). PKA then phosphorylates and inactivates BAD.

In their latest work, Kuper et al. utilize mice and cells with genetically-manipulated levels of COX-2 to provide important pieces of this puzzle. In response hypertonic conditions due to water deprivation, Cox-2^−/− mice had lower PGE₂, less phosphorylated BAD and greater apoptosis of renal tubular epithelial cells compared to COX-2^+/+ mice. The group then utilized RNAi technology to knock down the level of BAD in MDCK cells and demonstrate that BAD phosphorylation was key for the PGE₂-mediated attenuation of caspase-3 activity in cells incubated in hypertonic media. As in the earlier studies, BAD phosphorylation and caspase-3 activity in hypertonically-stressed cells were sensitive to inhibition of PKA and EP2. From these elegant and well-designed studies, it is clear that induction of COX-2 in response to osmotic stress is a critical adaptation that activates cAMP-mediated anti-apoptotic mechanisms that allow renal tubular cells to survive in harsh extracellular environments. Details of the relevant signaling and apoptotic pathways are shown in Figure 1.

It should be noted that BAD serves as a nexus for a variety of signaling pathways including the those activated by growth factors (e.g., EGF-1), hormones (e.g., Ang II) and other physiological mediators. These pathways phosphorylate BAD on Ser-112, Ser136 and Ser-155 (reviewed by Danial⁸). Neuhofer and colleagues focused on serine 155 of BAD which is modified by PKA¹. In an earlier study (Steinert et al.⁵), this group reported that inhibition of PKA only partially prevented the PGE₂-induced phosphorylation of BAD. This suggests that other signaling mediators may be activated by hypertonicity and be involved in BAD-mediated apoptosis. For example, cAMP activates EPAC as well as PKA and EPAC activates the MEK/ERK/RSK signaling system. This is notable because RSK also phosphorylates BAD on serine 155 (Figure 2). PGE₂, acting through EP2/EP4 has been shown to activate Rap1 through an EPAC signaling pathway; Rap1 is upstream of MEK/ERK/RSK⁶. In unrelated studies, Wang et al. found that EPAC acts coordinately with PKA in renal inner medullary collecting duct cells to alter urea movement and potentially other transport processes that are important for renal cell osmoregulation⁷. While the study did not examine the potential involvement of PGE₂, it demonstrated that hypertonicity activates the EPAC pathway in the inner medulla. When considered altogether, these studies suggest that PGE₂ can modulate BAD activity and apoptosis in the renal tubular epithelial cells in the inner medulla by activating parallel signaling pathways involving EPAC and PKA.

Figure 2.

BAD signaling. This diagram shows possible alternative regulatory pathways that result in phosphorylation and inactivation of BAD resulting in enhanced cell survival. PGE₂ initiates an increase in cAMP that can activate PKA and EPAC, both of which can phosphorylate BAD on Ser-155 by different signaling pathways. RSK, which is downstream of EPAC can also be activated by PKC independently of cAMP. Other signaling pathways can influence BAD activity by phosphorylating serine residues other than Ser-155 (e.g., Ser 112, Ser-136). For example, AKT can be activated by phosphoinositide 3-kinase (PI3K), heat shock protein 90 (HSP90) and heat shock protein 27 (HSP 27). Akt phosphorylates BAD on Ser-136 which promotes inactivation by cytosolic sequestration. BAD binds to and inactivates anti-apoptotic proteins at the mitochondria resulting in release of cytochrome C and activation of the caspase cascade leading to cell death. Phosphorylation of BAD prevents this binding and thus leads to cell survival.

Although the hypertonicity-induced increase in COX-2 activity can antagonize the apoptotic action of BAD, the potential negative effects of COX-2 should not be ignored. In situations where increased PGE₂ would be detrimental, there are alternate means by which BAD can be phosphorylated that offer an alternate control of apoptosis for the kidney. Phosphorylation of serine155 can be stimulated by activation of RSK by protein kinase C (PKC, figure 2). PI3kinase and heat shock proteins 90 and 27 stimulate AKT to inactivate BAD by phosphorylating it on serine136.

In conclusion, the work of Kuper et al. has elucidated an important signaling pathway of PGE₂ in the inner medulla and provided strong evidence that BAD may be a central effector in the cell’s response to hypertonic stress. The study suggests that inhibition of COX-2 by NSAIDs could be detrimental to inner medullary epithelial cells by increasing the pro-apoptotic actions of BAD and ultimately cell death. Given the complexity of responses that can be evoked by COX-2-generated prostaglandins, and the potential for non-prostaglandin-mediated regulation of BAD function, caution should be exercised not to oversimplify the control of apoptosis in the renal medulla and the role of NSAIDS in this process.