A new target for squamous cell skin cancer?

Prostaglandins (PGs) derived from arachidonic acid of cell membrane are synthesized by PG G/H synthase (cyclooxygenase; COX-1/-2) and signal as autocrine/paracrine lipids (1, 2). Similar to other tissues, COX-1 is constitutively expressed in keratinocytes of normal epidermis, whereas COX-2 expression is more variable and regulated (3, 4). COX-2 is induced for example by cytokines and growth factors; accumulated prostaglandins from COX-2 regulate pain, inflammation, and cancers (2, 3). In the case of cancer, increased levels of PGs disrupt differentiation and thus contribute to the sensitization of cells to carcinogens and ensuing hyperplasia (5, 6). Therefore, the inhibition of COX-2 activity with aspirin like compounds has been suggested as a potent chemo-preventive therapy to suppress tumor development, not just in skin but particularly for gastrointestinal cancers (2, 7).

While in most cases COX-2 appears to support oncogenesis, opposite evidence that constitutive expression of COX-2 prevents the development of skin tumor has been reported suggesting that distinct downstream PGs mediated by COX-2 in vivo might have opposing effects and thus explain these variable results (8). This is quite classic for this pathway; despite the plethora of physiologic systems were COX-2 appears active, aspirin has a very limited side effect profile. This is likely because in most cases of inhibition of COX-2, the simultaneous inhibition of both positive and negative individual downstream prostaglandin players has a net zero physiologic effect. The most classic opposing prostaglandins are PGE₂ and PGD₂. PGE₂ has been shown to correlate with the oncogenic effects of COX-2, but less work has been done on PGD₂. To better understand the functions of PGD₂ and its metabolites in skin tumors, Mantel et al. evaluated the role of Aldo-keto reductase 1C3 (AKR1C3) expression in proliferation of SCC cell lines.

AKR1C3 is a family of AKR1C isozymes with oxidoreductase activity and can shunt PGD₂ away from its receptors and metabolites through its conversion to F₂ series prostaglandins, although it has many distinct functions. One of those is controlling steroid homeostasis (9) and consistent with this, deregulation of AKR1C3 expression is detected in cancers from steroid hormone-related tissues including breast and prostate (10). Although the localization of AKR1C family has been already described in skin cells including keratinocytes, fibroblasts, and melanocytes (11–13), its function and regulation still remain unclear. Previously, Mantel et al. reported that AKR1C3 is significantly expressed in supra basal layers of human epidermis and is associated with differentiation of keratinocytes (14). In a recent issue (15), the authors demonstrate that PGD₂ and 15d-PGJ₂, its non-enzymatic metabolite, inhibit the proliferation of SCC cells possibly through PPARγ activation. Overexpression of AKR1C3 presumably decreases PGD₂ levels by metabolizing it to detectably higher levels of 9α11β-PGF₂. Although knockdown or overexpression of AKR1C3 had no effect on SCC proliferation, the authors could detect decreased sensitivity to PGD₂ in overexpressing cells.

AKR1C3 is one of four major types (AKR1C1-4) found in human and has stronger catalytic activity and higher affinity for PGD₂ than AKR1C1 and 2 (16, 17). Supporting the authors arguments, the AKR1C3-mediated product PGF₂, binds to its prostanoid receptor and induces MAPK signaling, thereby inactivating PPARγ (17, 18). These results support a role of AKR1C3 to inhibit the effects of PPARγ and might explain the decreased sensitivity to PGD₂ in AKR1C3 expressing cells.

Previous research on PGD₂ has detailed on its role in hair biology. PGD₂ is likely involved in the pathogenesis of androgenetic alopecia (AGA). Also, our group has reported that PGD₂ inhibits hair follicle neogenesis after wounding through the PGD₂ receptor 2 (DP-2/GPR-44). We demonstrated that GPR-44 null mice had increased levels of wound induced hair neogenesis for example (19, 20). In the course of these studies, we tested the effect of PGD₂ on normal human keratinocyte proliferation and were surprised to find that, in contrast to the effects of PGD₂ in SCC cells, the PGD₂ promotes the proliferation of normal keratinocytes (19). This suggests contextual differences between cancer and normal cells. However, more robust dissection of this pathway including PGD₂ and 15d-PGJ₂ treatments of both normal and SCC keratinocytes after siRNA knockdown of PGD₂ receptors (DP-1/DP-2) and PPARγ should clarify these issues. Despite these questions, the anti-proliferative effects of 15d-PGJ₂ via the activation of PPARγ in SCC cells are consistent with the results from previous studies in different cancer cells (21, 22).

Although 15d-PGJ₂ has been known as an endogenous ligand for PPARγ, there exist some skepticism about whether 15d-PGJ₂ is an agonist in vivo for PPARγ. First, there are several PPAR-independent pathways mediated by 15d-PGJ₂ including nuclear factor kB, ROS, and SUMOylation (21, 23, 24). Second, the concentrations of 15d-PGJ₂ to activate PPARγ reported in previous studies (µM range) (25) are greatly higher compared to those of endogenous prostaglandins (pM range) (26, 27) associated with the physiological activity. Finally, the high concentration of 15d-PGJ₂ required for the inhibition of cell proliferation might be due to its strongly electrophilic and oxidative properties. To this end, it may be worth measuring the physiological amounts of PGD₂, 9α11β-PGF₂ and 15d-PGJ₂ in normal and SCC human skin using mass spectrometry (28). Afterwards, further research may be investigated to establish the regulatory mechanisms in the production of these two downstream metabolites from PGD₂ in vivo.

One unresolved paradox for this paper is the context of AKR1C3 expression. In normal tissues, AKR1C3 is expressed in terminally differentiated keratinocytes. However, in this article ARK1C3 is noted to be more highly expressed in SCC cell lines than normal keratinocytes. Also, it is proposed to enhance proliferation in SCC which would not be the case in keratinocytes of the upper layers of the epidermis. Future experiments described above might help clarify some of these questions.

There are many interesting experiments that the present report might suggest. For example, interesting questions are raised by our recent report that PGD₂ might promote AGA, and the present findings that PGD₂ increases the levels of AKR1C3. Given that the function of AKR1C3 to increase the potency of weak androgens (and estrogens) for example from Δ4-androstene-3,17-dione to testosterone, it is very likely that AKR1C3 could be involved in a positive feed-back loop which promotes AGA. Since PGD₂ production itself is commonly promoted by androgens, then the increased levels found in AGA would be thought to promote AKR1C3 which would likely generate additional testosterone which would enhance PGD₂ production further—and exacerbate AGA. Also, since it has been known that AKR1C1 and AKR1C2 are induced in skin by UV light exposure (11), it will be interesting to determine functions of these and AKR1C3 in prostaglandin physiology as it relates to skin cancer and alopecia. As in all science, more questions multiply whenever we try to answer any.