Vitiligo is a depigmenting disorder in which the loss of functioning melanocytes causes the appearance of white patches on the skin 1. Melanocytes are pigment-producing cells that provide photoprotection to the skin through the production and distribution of melanin to keratinocytes. Melanocytes are highly dentritic cells, and the presence of dentrites is critically important for melanosome transfer to keratinocytes 2.
Two cyclooxygenase (COX) isoforms are known: COX-1 and COX-2. COX-1 is constitutively expressed in many human tissues. COX-2 occurs constitutively only in a few tissues including the epidermis 3. COX enzymes oxidize arachidonic acid to prostaglandin H2 (PGH2), which is further metabolized by PGE synthase (PGES) into prostaglandin E2 (PGE2) 4.
The role of PGE2 in melanogenesis has been discussed in several studies. PGE2 stimulates dendricity of melanocytes 5, maturation of melanosomes, and orientation of microfilaments in the dendrite process 6. Starner et al.7 demonstrated that nanomolar doses of PGE2 stimulate the high-affinity receptors EP3 and EP4; these receptors, by increasing cAMP release, enhance the tyrosinase activity and proliferation in human melanocytes.
Nordlund et al. 8 applied PGE2 to mice skin and observed an increase in melanocyte density, with increased formation of tonofilaments and keratohyalin granules in keratinocytes. Moreover, Parsad et al.9 and Kapoor et al.10 obtained favorable results when they used topical PGE2 for the treatment of vitiligo, where many of the patients showed marked to complete repigmentation of their lesions. Consequently, they suggested that PGE2 may have a direct stimulating action or an immunomodulatory action.
Naive CD4+ helper T cells develop into four types: T helper1 (Th1), Th2, Th17, and regulatory T cells (Tregs). Th1 cells produce primarily TNF-α and IFN-γ, Th2 secrete IL-4, IL-5, and IL-13, Th17 cells produce IL-17 and IL-6, and Tregs synthesize IL-10 and TGF-β 11. A novel hypothesis has been suggested that skewing of responses toward Th17 or Th1 and away from Tregs and Th2 cells may be responsible for the development and progression of autoimmune diseases including vitiligo 12.
COX-2 has been proposed to favor Th17 responses indirectly by increasing IL-23 and blocking IL-12 release from antigen-presenting cells 13. PGE2 promotes skewing of activated T cells toward the IL-23-induced Th17 cell expansion through the EP2/EP4 cAMP-protein kinase pathway 14, or toward the Th1 phenotype through its action on EP3, promoting the production of IFN-γ from established Th1 cells 15.
On the basis of the above, COX-2 and PGE2 may possibly play a role in the pathogenesis of vitiligo. Therefore, this study was carried out to determine the plasma and tissue levels of COX-2 and PGE2 in patients with vitiligo.
Patients and methods
This analytic cross-sectional study included 22 patients with vitiligo attending the Dermatology Outpatient Clinic of Kasr El Aini University Hospitals of the Faculty of Medicine, Cairo University. The patients recruited had not used any medication in the preceding 4 weeks. Twenty age-matched and sex-matched healthy control individuals were also enrolled. Patients and healthy volunteers signed informed written consent to participate in this study.
A thorough history was obtained from each patient including the duration of disease, age of onset, family history, Koebner phenomenon positivity, relation to stress, and any previous history of systemic or autoimmune diseases. A complete physical and dermatological examination was performed together with estimation of body surface area involvement and disease distribution according to Ortonne 16, who classified vitiligo into three groups, namely, (a) localized: focal, unilateral (segmental), or mucosal, (b) generalized: vulgaris, acrofacial, and mixed, and (c) universal: complete or nearly complete depigmentation. The rule of nines 17 was used to determine the extent of the disease.
The activity of disease was determined by the vitiligo disease activity (VIDA) score, which is an objective criterion proposed by Njoo et al.18 to follow the course of lesions. It is a six-point scale on which the activity of the disease is evaluated by the appearance of new vitiligo lesions or enlargement of pre-existing lesions gauged during a period ranging from less than 6 weeks to 1 year (Table 1).
Two punch biopsies with a diameter of 4 mm were obtained from the depigmented lesions and the clinically normal skin in vitiligo. One skin biopsy was taken from each control participant. All the skin biopsies were taken from hidden areas (nonsun exposed) when possible. The skin biopsies were divided into two parts; the first part was used to determine COX-2 gene expression and the second part was used for PGE2 assessment by the enzyme-linked immunosorbent assay (ELISA) technique. Both parts were kept frozen at −80°C till analysis.
In addition, 5 ml whole-blood samples were collected from the antecubital fossa from patients and control participants in heparinized tubes to assess the plasma level of PGE2 and COX-2 by the ELISA technique. COX-2 activity was assessed using the Cyclooxygenase Enzyme Immunometric Assay Kit. Oxford Biomedical Research Inc., Rochester Hills, Michigan, USA.
Reverse transcriptase-polymerase chain reaction detection of cyclooxygenase-2 gene expression
Total RNA was extracted from skin tissue homogenate using RNeasy purification reagent (Qiagen, Valencia, California, USA) according to the manufacturers’ instruction. cDNA was then generated from 5 μg of total RNA extracted with 1 μl (20 pmol) antisense primer and 0.8 μl superscript avian myeloblastosis virus reverse transcriptase for 60 min at 37°C. For PCR, 4 μl cDNA was incubated with 30.5 μl water, 4 μl MgCl2 (25 mmol/l), 1 μl dNTPs (10 mmol/l), 5 μl 10×PCR buffer, 0.5 μl (2.5 U) Taq polymerase, and 2.5 μl of each primer containing 10 pmol.
The primer sequences were as follows:
Forward 5′-CAAGCACTGTGGGTTTTAAT-3′, reverse 5′-GGTTTTGTCAGCAGATCAAT-3′. The reaction mixture was subjected to 40 cycles of PCR amplification as follows: denaturation at 95°C for 1 min, annealing at 67°C for 1 min, and extension at 72°C for 2 min. PCR products were electrophoresed on 2% agarose stained with ethidium bromide and visualized by an ultraviolet transilluminator. Semiquantification was performed using a gel documentation system (BioDO Analyser; Biometra, Göttingen, Germany). According to the amplification procedure, the relative expression of each studied gene (R) was calculated according to the following formula: densitometrical units of each studied gene/densitometrical units of β-actin.
PCR detection of β-actin: The presence of RNA in all samples was assessed by analysis of the ‘housekeeping’ gene β-actin. Complementary DNA was generated from 1 mg total RNA extracted with avian myeloblastosis virus reverse transcriptase for 60 min at 37°C. For PCR, 4 μl complementary DNA was incubated with 30.5 μl water, 25 mmol/l MgCl2, 1 ml deoxyribonucleotide triphosphates (10 mmol/l), 5 μl 10× PCR buffer, 0.5 μl (2.5 U) Taq polymerase, and 2.5 μl of each primer containing 10 pmol/l. β-Actin primers (forward 5-TGTTGTCCCTGTATGCCTCT-3; reverse 5-TAATGTCACGCACGATTTCC-3). The reaction mixture was subjected to 40 cycles of PCR amplification, denaturation at 95°C for 1 min, annealing at 57°C for 1 min, and extension at 72°C for 2 min.
Measurement of the prostaglandin E2 level by the enzyme-linked immunosorbent assay technique
Skin biopsy was homogenized in phosphate-buffered saline and then centrifuged at 14 000 RPM for 5 min. The supernatant and plasma were then examined for the PGE2 level by the ELISA technique supplied by Oxford Biomedical Research Inc. (Rochester Hills, Michigan, USA) according to the manufacturers’ instructions.
The half-life of PGE2 is short; however, the stability of PGE2 is highly pH dependent. Therefore, we used extraction buffer with an optimum pH to maintain the stability of PGE2 for its measurement 19.
Assessment of the cyclooxygenase-2 level by enzyme-linked immunosorbent assay
The COX-2 level was assessed in plasma samples using the ELISA kit supplied by Alpha Diagnostic International Inc. (San Antonio, Texas, USA) according to the manufacturers’ instruction.
Assessment of cyclooxygenase-2 activity by immunometric assay
COX-2 activity was assessed by measuring the peroxidase activity of the COX enzyme. The peroxidase activity is assayed colorimetrically by monitoring the appearance of oxidized N,N,N′,N′-tetramethyl-p-phenylenediamine (TMPD) at 590 nm. The kit includes isoenzyme-specific inhibitors for distinguishing COX-2 activity from COX-1 activity (Alpha Diagnostic International Inc.).
Data were statistically described in terms of mean±SD, frequencies, and percentages when appropriate. Comparison of quantitative variables between the study groups was carried out using the Mann–Whitney U-test for independent samples. Comparison of sex distribution between the study groups was carried out using the χ2-test. The exact test was used when the expected frequency as less than 5. The correlation between various variables was assessed using the Spearman rank correlation equation for non-normal variables. The significance level was set at (P≤0.05). All statistical calculations were carried out using computer programs Microsoft Excel 2007 (Microsoft Corporation, Redmond, Washington, USA) and SPSS (Statistical Package for the Social Science; SPSS Inc., Chicago, Illinois, USA) version 15 for Microsoft Windows.
Clinical Characteristics of the participants
Twenty-two patients with vitiligo and 20 age-matched and sex-matched healthy control individuals were included in the study. The patient group included 12 women (55.5%) and 10 men (45.5%) and the control group included 11 women (55%) and nine men (45%) (P=0.976). The mean age of the patients and the control individuals was 36±14.259 and 37.6±13.08 years, respectively (P=0.708). Clinically, all patients had generalized disease, 17 (77.27%) patients had vulgaris distribution, and the remaining five (22.73%) patients had vulgaris and acrofacial distribution (mixed type). One patient had VIDA −1 (4.55%), five patients had VIDA 0 (22.7%), one patient had VIDA +1 (4.55%), four patients had VIDA +2 (18.2%), three patients had VIDA +3 (13.6%), and eight patients had VIDA +4 (36.4%) (Table 2).
Tissue levels of cyclooxygenase-2 mRNA and prostaglandin E2 in the studied groups
The mean levels of COX-2 mRNA in both lesional and nonlesional skin biopsies of patients were significantly depressed compared with their corresponding levels in the normal skin of the controls (0.141±0.076, 0.474±0.287, and 0.9130±0.225, respectively, P<0.001). The levels of COX-2 mRNA in lesional skin in patients were lower than COX-2 mRNA in nonlesional skin, with no significant difference (0.141±0.076 and 0.474±0.287, respectively, P=0.619).
Similarly, the mean tissue levels of PGE2 in both lesional and nonlesional skin biopsies were significantly decreased compared with their corresponding levels in normal skin of the controls (1.258±0.572, 2.334±0.641, and 3.896±1.298 ng/mg protein, respectively, P<0.001, P=0.003). The levels of PGE2 in lesional skin in patients were lower compared with nonlesional skin of patients, with no significant difference (1.258±0.572 and 2.334±0.641 ng/mg protein, respectively, P=0.955).
Plasma levels of cyclooxygenase-2, cyclooxygenase-2 enzyme activity, and prostaglandin E2 in the studied groups
In the present study, the mean plasma levels of COX-2 were significantly elevated in patients compared with the controls (1.656±0.502 and 0.399±0.297 ng/ml, respectively, P<0.001). Similarly, the mean levels of COX-2 enzyme activity were significantly elevated in patients’ plasma compared with the controls (28.927±10.603 and 13.590±3.578 U/ml, respectively, P<0.001). Moreover, the mean levels of PGE2 in patients’ plasma were significantly elevated compared with the controls (9.965±2.291 and 3.002±1.5 ng/ml, respectively, P<0.001) (Fig. 1).
Relation between estimated parameters and clinical data and disease characteristics
In patients, the plasma levels and enzyme activity of COX-2 were positively correlated to the VIDA score (r=0.465, P=0.029, and r=0.423, P<0.05, respectively) (Fig. 2). We did not find any statistically significant relationship in the levels of estimated parameters (plasma and tissue COX-2 and PGE2) with age, sex, age of onset, disease duration, family history, koebner phenomenon, clinical disease subtypes, or extent of skin affection.
In patients, no correlation was found between the plasma levels of COX-2 and either its enzyme activity (r=0.111, P=0.622) or plasma PGE2 (r=0.268, P=0.229). In addition, no correlation was found between tissue levels of COX-2 mRNA and tissue PGE2 (r=0.135, P=0.549).
In the controls, the following correlations were detected: the plasma levels of COX-2 were positively correlated to the COX-2 enzyme activity (r=0.709, P=0.022). The plasma levels of COX-2 were negatively correlated to the plasma PGE2 levels (r=−0.754, P=0.012). Moreover, the tissue levels of PGE2 were positively correlated to the tissue levels of COX-2 mRNA (r=0.818, P=0.004).
In the present study, the levels of COX-2 mRNA and PGE2 in normal skin of the controls were significantly higher compared with their corresponding levels in lesional and nonlesional skin of patients. These results are different from the observations made by Hensby et al.20, who detected elevated tissue levels of PGE2 in vitiligo patients when they used quantitative analysis by gas chromatography of exudates from suction blisters.
In the epidermis, both keratinocytes and melanocytes are a major source of the COX-2 enzyme and PGE2 21,22. In active vitiligo, morphological features are reported, including vacuolation/degeneration of basal keratinocytes, melanocytes, and Langerhans cells; melanocytes are decreased, but never completely absent from vitiliginous skin 23. Therefore, decreased expression of COX-2 mRNA and PGE2 in lesional skin of patients could be the result of decreased number and/or impaired function of melanocytes and keratinocytes in vitiliginous skin. COX-2 and PGE2 play a potential role in melanogenesis; therefore, their absence will further contribute to the depigmenting process in vitiligo.
Moreover, the decreased expression of COX-2 mRNA and PGE2 in nonlesional skin in patients compared with controls indicates that clinically pigmented skin may also be somehow immune-committed in the active vitiligo, and some specific cellular event might eventually occur in the epidermis, before the clinically detectable depigmentation 24. An incomplete distribution of pigment along the basal layer in nonlesional skin in vitiligo patients occurs because of either an initial and progressive disappearance of melanocytes or a congenital defective melanocyte distribution 25.
In the present study, the plasma level and enzyme activity of COX-2 and the plasma level of PGE2 were significantly higher in patients than those in the controls (P<0.001). Similar to our findings, Esmaeili et al.26 found increased COX2 gene expression and IL-17 in peripheral blood leukocytes of vitiligo patients compared with the controls, but not reaching statistical significance. In addition, they found a positive correlation between the levels of COX-2 and IL-17. Both Th1 and Th17 are known to play an important role in the pathogenesis of vitiligo 27–29, and elevations in their serum cytokine levels have been detected in patients with active disease 30. On the basis of the above findings, we can conclude that elevated plasma levels of COX-2 and PGE2 could contribute toward the pathogenesis of vitiligo, as COX-2 favors the Th17 phenotype, and PGE2 promotes the differentiation of naive T cells into the Th17 or the Th1 subset, both of which are known to play an important role in the pathogenesis of vitiligo.
In addition, vitiligo is occasionally associated with autoantibodies against the melanocyte-specific proteins 31. PGE2 is involved in B cell activation and immunoglobulin synthesis 32; PGE2 enhances IL-4-induced IgE and IgG1 synthesis in stimulated B lymphocytes 33. Accordingly, the high level of COX-2 and PGE2 in patients’ plasma could be a contributing factor in the development of autoantibodies in vitiligo.
The plasma levels of COX-2 were negatively correlated to the plasma PGE2 level in the controls but not in patients. This finding in the controls is in accordance with an earlier observation 34 suggesting that PGE2 can initiate negative feedback regulation in the induction of COX-2 mediated by cAMP. However, more recent observations 35 demonstrated positive feedback regulation of COX-2 expression by PG metabolites. In the controls, positive correlations were founds between the plasma levels of COX-2 with its enzyme activity and between the tissue levels of PGE2 and COX-2 mRNA. None of these correlations was found in patients. Increased COX-2 expression is usually accompanied by increases in COX-2 enzymatic activity and enhanced capacity of the tissues to synthesize PGs 36.
Therefore, the lack of such correlations in vitiligo patients in the present study could be the result of a functional disturbance of the feedback regulation of PGE2 and COX-2 in plasma and tissue, caused by the disease process.
To our knowledge, this is the first study to correlate the plasma level and enzyme activity of COX-2 with the VIDA score, where strong positive correlations were detected. Therefore, we conclude that estimation of the plasma level or enzyme activity of COX-2 could be used as a marker of disease activity. In addition, this study is the first to evaluate the plasma and tissue levels of COX-2 and PGE2 in vitiligo. We conclude that COX-2 and PGE2 could contribute toward the development of vitiligo through their immunomodulatory role. The high levels of COX-2 and PGE2 in the plasma of patients could contribute to the pathogenesis of vitiligo through shifting of activated T cells toward the Th1 or the Th17 phenotype, both of which are known to play an important role in the pathogenesis of vitiligo.
Further studies examining the effects of topical PGE2 on the tissue levels of PGE2 and COX-2 together with its effects on vitiligo outcome are needed. Further studies examining the effect of phototherapy on the levels of COX-2 and PGE2 are also needed.
Conflicts of interest
There are no conflicts of interest.
1. Whitton ME, Ashcroft DM, González U. Therapeutic interventions for vitiligo
. J Am Acad Dermatol. 2008;59:713–717
2. Scott G, Leopardi S, Printup S, Malhi N, Seiberg M, Lapoint R. Proteinase-activated receptor-2 stimulates prostaglandin production in keratinocytes: analysis of prostaglandin receptors on human melanocytes and effects of PGE2 and PGF2alpha on melanocyte dendricity. J Invest Dermatol. 2004;122:1214–1224
3. Leong J, Hughes-Fulford M, Rakhlin N, Habib A, Maclouf J, Goldyne ME. Cyclooxygenases in human and mouse skin and cultured human keratinocytes: association of COX-2 expression with human keratinocyte differentiation. Exp Cell Res. 1996;224:79–87
4. Ueno N, Takegoshi Y, Kamei D, Kudo I, Murakami M. Coupling between cyclooxygenases and terminal prostanoid synthases. Biochem Biophys Res Commun. 2005;338:70–76
5. Scott G, Fricke A, Fender A, McClelland L, Jacobs S. Prostaglandin E2 regulates melanocyte dendrite formation through activation of PKCzeta. Exp Cell Res. 2007;313:3840–3850
6. Sauk JJ Jr, White JG, Witkop CJ Jr. Influence of prostaglandins E1, E2 and arachidonate on melanosomes in melanocytes and keratinocytes of anagen hair bulbs in vitro. J Invest Dermatol. 1975;64:332–337
7. Starner RJ, McClelland L, Abdel-Malek Z, Fricke A, Scott G. PGE (2) is a UVR-inducible autocrine factor for human melanocytes that stimulates tyrosinase activation. Exp Dermatol. 2010;19:682–684
8. Nordlund JJ, Collins CE, Rheins LA. Prostaglandin E2 and D2 but not MSH stimulate the proliferation of pigment cells in the pinnal epidermis of the DBA/2 mouse. J Invest Dermatol. 1986;86:433–437
9. Parsad D, Pandhi R, Dogra S, Kumar B. Topical prostaglandin analog (PGE2) in vitiligo
– a preliminary study. Int J Dermatol. 2002;41:942–945
10. Kapoor R, Phiske MM, Jerajani HR. Evaluation of safety and efficacy of topical prostaglandin E2 in treatment of vitiligo
. Br J Dermatol. 2009;160:861–863
11. Afzali B, Lombardi G, Lechler RI, Lord GM. The role of T helper 17 (Th17) and regulatory T cells (Treg) in human organ transplantation and autoimmune disease. Clin Exp Immunol. 2007;148:32–46
12. Weaver CT, Harrington LE, Mangan PR, Gavrieli M, Murphy KM. Th17: an effector CD4 T cell lineage with regulatory T cell ties. Immunity. 2006;24:677–688
13. Napolitani G, Acosta-Rodriguez EV, Lanzavecchia A, Sallusto F. Prostaglandin E2 enhances Th17 responses via modulation of IL-17 and IFN-gamma production by memory CD4+ T cells. Eur J Immunol. 2009;39:1301–1312
14. Sakata D, Yao C, Narumiya S. Prostaglandin E2, an immunoactivator. J Pharmacol Sci. 2010;112:1–5
15. Bloom D, Jabrane-Ferrat N, Zeng L, Wu A, Li L, Lo D, et al. Prostaglandin E2 enhancement of interferon-γ production by antigen- stimulated type 1 helper T cells. Cell Immunol. 1999;194:21–27
16. Ortonne JBolognia JL, Jorizzo JL, Rapini RP. Vitiligo
and other disorders of hypopigmentation. Dermatology. 20082nd ed Spain Elsevier:65
17. Kanthraj GR, Srinivas CR, Shenoi SD, Deshmukh RP, Suresh B. Comparison of computer-aided design and rule of nines methods in the evaluation of the extent of body involvement in cutaneous lesions. Arch Dermatol. 1997;133:922–923
18. Njoo MD, Das PK, Bos JD, Westerhof W. Association of the Kobner phenomenon with disease activity and therapeutic responsiveness in vitiligo
vulgaris. Arch Dermatol. 1999;135:407–413
19. Watzer B, Zehbe R, Halstenberg S, James Kirkpatrick C, Brochhausen C. Stability of prostaglandin E2 (PGE2) embedded in poly-D,L-lactide-co-glycolide microspheres: a pre-conditioning approach for tissue engineering applications. J Mater Sci Mater Med. 2009;20:1357–1365
20. Hensby CN, Shroot B, Schaefer H. Prostaglandins in human skin disease. Br J Dermatol. 1983;109:22–25
21. Mempel M, Voelcker V, Köllisch G, Plank C, Rad R, Gerhard M, et al. Toll-like receptor expression in human keratinocytes: nuclear factor kappaB controlled gene activation by Staphylococcus aureus
is toll-like receptor 2 but not toll-like receptor 4 or platelet activating factor receptor dependent. J Invest Dermatol. 2003;121:1389–1396
22. Gledhill K, Rhodes LE, Brownrigg M, Haylett AK, Masoodi M, Thody AJ, et al. Prostaglandin-E2 is produced by adult human epidermal melanocytes in response to UVB in a melanogenesis-independent manner. Pigment Cell Melanoma Res. 2010;23:394–403
23. Tobin DJ, Swanson NN, Pittelkow MR, Peters EM, Schallreuter KU. Melanocytes are not absent in lesional skin of long duration vitiligo
. J Pathol. 2000;191:407–416
24. Aslania FMNP, Noé RAM, Antelo DP, Farias RE, Da PK, Galadari I, et al. Immunohistochemical findings in active vitiligo
including depigmenting lesions and non-lesional skin. Open Dermatol J. 2008;2:105–110
25. Ardigo M, Malizewsky I, Dell’anna ML, Berardesca E, Picardo M. Preliminary evaluation of vitiligo
using in vivo reflectance confocal microscopy. J Eur Acad Dermatol Venereol. 2007;21:1344–1350
26. Esmaeili B, Rezaee SAR, Layegh P, Afshari JT, Dye P, Karimiani EG, et al. Expression of IL-17 and COX2 gene in peripheral blood leukocytes of vitiligo
patients. Iran J Allergy Asthma Immunol. 2011;10:81–89
27. Wankowicz-Kalinska A, van den Wijngaard RM, Tigges BJ, Westerhof W, Ogg GS, Cerundolo V, et al. Immunopolarization of CD4+ and CD8+ T cells to Type-1-like is associated with melanocyte loss in human vitiligo
. Lab Invest. 2003;83:683–695
28. Bassiouny DA, Shaker O. Role of interleukin-17 in the pathogenesis of vitiligo
. Clin Exp Dermatol. 2011;36:292–297
29. Wang CQ, Cruz Inigo AE, Fuentes Duculan J, Moussai D, Gulati N, Sullivan Whalen M, et al. Th17 cells and activated dendritic cells are increased in vitiligo
lesions. PLoS One. 2011;6:e18907
30. Basak PY, Adiloglu AK, Ceyhan AM, Tas T, Akkaya VB. The role of helper and regulatory T cells in the pathogenesis of vitiligo
. J Am Acad Dermatol. 2009;60:256–260
31. Kemp EH, Gawkrodger DJ, Watson PF, Weetman AP. Immunoprecipitation of melanogenic enzyme autoantigens with vitiligo
sera: evidence for cross-reactive autoantibodies to tyrosinase and tyrosinase-related protein-2 (TRP-2). Clin Exp Immunol. 1997;109:495–500
32. Roper RL, Graf B, Phipps RP. Prostaglandin E2 and cAMP promote B lymphocyte class switching to IgG1. Immunol Lett. 2002;84:191–198
33. Roper RL, Phipps RP. Prostaglandin E2 and cAMP inhibit B lymphocyte activation and simultaneously promote IgE and IgG1 synthesis. J Immunol. 1992;149:2984–2991
34. Akarasereenont P, Techatrisak K, Chotewuttakorn S, Thaworn A. The induction of cyclooxygenase-2 in IL-1beta-treated endothelial cells is inhibited by prostaglandin E2 through cAMP. Mediators Inflamm. 1999;8:287–294
35. Vichai V, Suyarnsesthakorn C, Pittayakhajonwut D, Sriklung K, Kirtikara K. Positive feedback regulation of COX-2 expression by prostaglandin metabolites. Inflamm Res. 2005;54:163–172
36. DeWitt DL. Prostaglandin endoperoxide synthase: regulation of enzyme expression. Biochim Biophys Acta. 1991;1083:121–134