Activation of the Prostanoid FP Receptor Inhibits Adipogenesis Leading to Deepening of the Upper Eyelid Sulcus in Prostaglandin-Associated Periorbitopathy

Abstract

Purpose.: To investigate the effects of prostaglandin (PG) analogues on adipogenesis so as to clarify the mechanism of a side effect of topical PG analogues: deepening of the upper eyelid sulcus (DUES), which has been reported in this decade.

Methods.: The 3T3-L1 preadipocytes were treated to promote differentiation into mature adipocytes. During the early and late stages of differentiation (days 0, 2, and 7), 1 to 1000 nM latanoprost acid (LAT-A), travoprost acid (TRA-A), tafluprost acid (TAF-A), bimatoprost (BIM), bimatoprost acid (BIM-A), unoprostone (UNO), or prostaglandin F2a (PGF2α) was applied to cells. Oil red O staining was used to detect intracellular lipids on day 10. Stained areas measured on a photograph were compared with those in control cultures. All experiments were performed in a masked manner. Next, similar experiments were performed using primary cultured mouse adipocytes from FP receptor knockout and wild-type mice.

Results.: When PGs were added on day 0 or 2, LAT-A, TAF-A, BIM-A, and PGF2α significantly inhibited adipogenesis (P < 0.01 on day 0, P < 0.05 on day 2) at concentrations of 10 nM and 100 nM, and TRA-A inhibited adipogenesis only at 100 nM. Bimatoprost and UNO did not affect adipogenesis at any concentration. When PGs were added on day 7, 100 nM LAT-A, BIM-A, or PGF2α significantly suppressed adipogenesis (P < 0.05). In mouse primary adipocyte cultures, LAT-A, TAF-A, BIM-A, TRA-A, and PGF2α significantly suppressed adipogenesis in wild-type adipocytes (P < 0.05), but adipogenesis was not suppressed by any of the PG compounds in FP knockout mouse adipocytes.

Conclusions.: Prostaglandin analogues have the potential to inhibit adipogenesis through FP receptor stimulation. Although these findings should be further analyzed in model systems more closely related to orbital fat, PG analogues may directly lead to reduced orbital fat by inhibiting adipogenesis.

Introduction

Prostaglandin analogues (PGs), which lower IOP through the prostanoid FP receptor, ¹ have been widely used as first-line drugs for glaucoma treatment. Prostaglandin analogues are highly effective at lowering IOP when taken once daily, and do not induce systemic side effects. However, in their clinical use over a decade, characteristic local side effects, known as prostaglandin-associated periorbitopathy (PAP), which includes hyperpigmentation of the skin, elongation of eyelashes, and changes in iris color, have been noted. Shah et al. ² reported that some PGs are associated with a risk of dermatochalasis, deepening of the upper eyelid sulcus (DUES), loss of lower lid steatoblepharon, upper lid ptosis, and lower lid retraction. 

In this decade, DUES was reported as a cosmetic side effect of bimatoprost (BIM), travoprost, and latanoprost. ^3–9 A recent clinical prospective study by our group indicated that the incidence of DUES in Japanese patients who switched from latanoprost to BIM for 3 months was 60%, as assessed by objective judgment of photographs. ¹⁰ For these patients with DUES, symptoms abated after switching back to latanoprost from BIM. ¹¹ Shah et al. ² recently reported that BIM and travoprost are associated with a high risk of dermatochalasis or DUES, but no significant change was detected with latanoprost use, and a longer duration of BIM use was associated with a significantly higher risk of other PAP. These case reports of DUES in patients using these newly developed PGs and the rare cases among those using latanoprost suggest that the mechanism of this side effect may be related to the affinity of these drugs for FP receptors. However, the IOP-lowering effects of these PGs appear to be similar based on several meta-analysis studies. ^12–16 Thus, IOP reduction and DUES pathogenesis may occur through different mechanisms. 

A recent anatomical report using magnetic resonance imaging indicated a reduction in orbital adipose tissue after long-term application of PGs. ¹⁷ Moreover, histological analysis of upper eyelid adipose tissue excised during surgery indicated that adipocyte density was higher in eyelids treated with travoprost and BIM than in the untreated contralateral eyes, but that this was not the case for latanoprost-treated eyes. ⁴ This suggests a relative decrease in adipocyte volume in eyelids treated with travoprost or bimatoprost. 

Adipocytes are unique cells involved in the regulation of energy homeostasis. ^18,19 Adipose tissue volumes can be altered under various conditions. ²⁰ Adipocyte differentiation (adipogenesis) is a complex process that involves coordinated changes in hormone sensitivity and gene expression in response to various stimuli, including lipid mediators. Prostaglandins are among the lipid mediators involved in the regulation of adipocyte differentiation. ²¹ Adipogenesis of the preadipocyte cell line 3T3-L1 is inhibited by treatment with prostaglandin F2 alpha (PGF2α) or fluprostenol, a prostanoid FP2 receptor agonist. ²²  

Thus, we hypothesized that orbital fat tissue atrophy due to decreased adipocyte volume contributes to DUES. Investigation of the mechanism of PG-induced DUES may be useful for the future development of IOP-lowering drugs with fewer side effects. In this study, we analyzed the effects of all commercially available PGs on adipogenesis in vitro. Seibold et al. ²³ and Choi et al. ²⁴ previously investigated the effects of PGs on adipocytes, but in both of their studies, the PG agents used were prodrugs, which cannot bind to the FP receptor by themselves without hydrolysis of the terminal carboxyl group. Also, in the former study, cells were treated with PGs that included a preservative agent, benzalconium chloride. Thus, to provide a greater understanding of the effects of PGs on adipogenesis, we investigated the effects of acid-type PGs without any preservatives on adipocyte differentiation. Moreover, to investigate the role of FP receptors in adipogenesis, primary cultured adipocytes from FP receptor knockout mice were used. 

Materials and Methods

Materials

Dulbecco's modified Eagle's medium (DMEM), high-glucose DMEM (HG-DMEM), calcium pantothenate, d-biotin, 3-isobutyl-1-methylzanthine (MIX), insulin, dexamethasone, type 1A collagenase, and oil red O were purchased from Sigma-Aldrich (St. Louis, MO). Fetal bovine serum (FBS), calf serum (CS), and insulin were purchased from Invitrogen (Carlsbad, CA). Penicillin/streptomycin (PS) and trypsin-EDTA were purchased from Gibco (Grand Island, NY). Unoprostone (free acid-type isopropyl unoprostone, UNO), latanoprost acid (LAT-A), travoprost acid (TRA-A), BIM, BIM acid (BIM-A), and prostaglandin F2a (PGF2α) were purchased from Cayman Chemical Co. (Ann Arbor, MI). Tafluprost acid (TAF-A) was provided by Santen Pharmaceutical Co., Ltd. (Osaka, Japan). Pioglitazone was provided by Yoshinori Takeuchi of the Diabetes and Metabolism Department, Tokyo University, Tokyo, Japan. 

Adipocyte Culture and Differentiation of 3T3-L1 Cells

The 3T3-L1 cells were obtained from the Department of Diabetes and Metabolism, University of Tokyo, which is a universally used cell line for lipid studies. The 3T3-L1 preadipocytes were grown until confluence at 37°C in HG-DMEM containing 8 mg/L d-biotin, 4 mg/L calcium pantothenate, 100 U/mL penicillin, 100 μg/mL streptomycin (b.p. HG-DMEM), and 10% CS. 

Two days after cells reached confluence, adipocyte differentiation was initiated by the addition of a cocktail containing 10 μg/mL insulin, 100 μL/mL MIX, and 1 μM DEX, and this day was designated as day 0. Two days later (day 2), culture medium was replaced and 10 μg/mL insulin and 1 μM pioglitazone were added. After day 4, the culture medium was changed to b.p. HG-DMEM + 10% CS and refreshed every 2 days. Undifferentiated cultures were cultivated as a control group. 

Primary Adipocyte Culture From Wild-Type and FP Knockout Mice

Isolation of primary mouse adipocytes was performed using a modified version of previously described methods. ^25,26 Inguinal fat pads were removed from 10-week-old C57BL/6J and FP knockout (FPKO) mice. The fat tissues were minced and digested in 0.1% (wt/vol) collagenase at 37°C for 1 hour. ²⁷ After filtration and centrifugation (135g, for 3 minutes) of the digestion fluid, the floating primary adipocytes in the top layer and the stromal-vascular fraction cells at the bottom of the tube were collected separately and washed. The adipose cells were then placed in a culture flask (Sumiron Flask 50; Sumitomo Bakelite, Tokyo, Japan) filled with DMEM supplemented with 10% FBS. The flask containing mature adipocytes was brimmed with medium, providing the adipocytes with an air-free environment, turned upside down, and then incubated in a humidified 5% CO₂ atmosphere. The cells floated in the medium and adhered to the top inner surface of the flask. When the cells became firmly attached and fibroblast-shaped with no visible fat droplets after a week, the flask was turned upside down so that the cells were on the bottom. The medium was changed every 4 days until the cells were used for experiments. 

Differentiation of Adipocyte Cultures From Wild-Type and FPKO Mice

Preadipocytes were grown to semiconfluence before differentiation. Differentiation was then induced by changing the medium to b.p. HG-DMEM supplemented with 10% FBS, 0.1 mM MIX, 1 μM DEX, and insulin-transferrin-selenium-X supplement containing 10 μg/mL insulin and PS. After 2 days, the differentiation medium was replaced with b.p. HG-DMEM supplemented with 10% FBS, and this day was designated as day 0. Two days later (day 2), the culture medium was changed to b.p. HG-DMEM containing 10% FBS, and PGs were added. Undifferentiated cultures were cultivated as a control group. 

Stimulation With PGs

As described above, the day of initiation of 3T3-L1 adipocyte differentiation was designated as day 0. Cells were treated with LAT-A, TRA-A, TAF-A, BIM, BIM-A, UNO, or PGF2α at a final concentration of 1 to 1000 nM. To observe the effects of PGs on different stages of adipocyte differentiation, cells were divided into three groups, which were treated with PGs on day 0, day 2, or day 7. Day 0 cultures were beginning adipocyte differentiation, day 2 cultures were in the process of differentiation, and day 7 cultures were at the end of differentiation, ²⁸ and 0.01% DMSO was used as a vehicle. Prostaglandin analogues were added every 2 days when the medium was changed. 

Lipid Staining

Oil red O staining was used to detect intracellular lipids on day 10. Cells in culture dishes were washed twice with PBS, and then 10% formamide was added for 10 minutes. After washing with PBS, the medium was replaced with 60% isopropyl alcohol for 1 minute, stained with oil red O for 20 minutes, then washed with 60% isopropyl alcohol for 1 minute and then with PBS. 

Measurement of Lipid Volume by Image Acquisition

Cells stained with oil red O were photographed using a microscope (BZ-9000; KEYENCE, Osaka, Japan) to measure the stained area. In each culture using one agent, 30 areas were counted. The stained area was measured using ImageJ (National Institutes of Health, Bethesda, MD). Stained areas in the treated cultures were expressed as the percentage of the stained area in control cultures. All experiments were performed in a masked manner. 

Statistical Analysis

Data are presented as means ± SDs, and were statistically analyzed using Dunnett's test compared with control cultures. 

Results

Suppression of Adipogenesis by PGs in Differentiated Adipocytes

Figures 1A through 1C show representative images of oil red O staining on day 10 adipocytes that had been treated with PG on day 0, day 2, or day 7, respectively. The relative stained areas of cultures treated with 100 nM LAT-A, TRA-A, TAF-A, BIM, BIM-A, UNO, or PGF2α added on day 0 were 1.2% ± 0.9%, 6.2% ± 4.3%, 0.9% ± 0.9%, 40.7% ± 8.3%, 0.4% ± 0.4%, 109.4% ± 16.5%, and 3.1% ± 1.5%, respectively (n = 10) (Fig. 2A). Relative stained areas of cultures with these PGs added on day 2 were 31.6% ± 13.2%, 38.3% ± 22.8%, 28.8% ± 14.6%, 76.4% ± 39.3%, 23.6% ± 12.4%, 86.5% ± 30.4%, and 36.4% ± 21.8%, respectively (n = 10) (Fig. 2B). Relative stained areas of cultures with these PGs added on day 7 were 26.8% ± 4.2%, 51.0% ± 36.7%, 35.1% ± 17.1%, 38.8% ± 3.2%, 25.2% ± 1.1%, 48.8% ± 27.8%, and 17.8% ± 0.7%, respectively (n = 10) (Fig. 2C). When PGs were added on day 0 or 2, all the acid forms of PGs (LAT-A, TRA-A, TAF-A, BIM-A, and PGF2α) significantly inhibited adipogenesis (P < 0.01 on day 0, P < 0.05 on day 2), but BIM and UNO did not. However, when PGs were added on day 2, the extent of adipogenesis inhibition was less than when added on day 0. LAT-A, BIM-A, and PGF2α significantly suppressed adipogenesis (P < 0.05) when PGs were added on day 7 (Fig. 2C). 

Figure 1

 

Oil red O staining of 3T3-L1 cultures on day 10. Prostaglandin analogues were added on day 0 (A), day 2 (B), or day 7 (C) after initiating differentiation. Scale bars: indicates 100 μm.

Figure 1

 

Continued.

Figure 2

 

Adipogenesis in cultured 3T3-L1 cells with PG treatment. Areas stained by oil red O in 3T3-L1 cultures were measured on day 10 and are indicated as percentage of that in control cultures. Prostaglandin analogues were added on day 0 (A), day 2 (B), or day 7 (C) after initiating differentiation. Asterisks indicate significant differences versus control (P < 0.05, Dunnett's test).

Suppression of Adipogenesis in Primary Cultured Mouse Adipocytes by PGs

Next, we investigated the dose-dependent effects of PGs on adipogenesis. Based on the above results, we applied PGs at a concentration of 10, 100, or 1000 nM to primary cultured mouse adipocytes. We found that TRA-A significantly suppressed adipogenesis at 100 nM, and the other acid forms suppressed adipogenesis at 10 and 100 nM (Fig. 3, P < 0.05 or 0.01). Latanoprost acid, TAF-A, TRA-A, BIM-A, and PGF2α suppressed adipogenesis in a dose-dependent manner, but BIM and UNO showed no effect on adipogenesis even at the highest concentration (1000 nM). 

Figure 3

 

Dose-dependent effects of PGs on adipogenesis in primary cultured mouse adipocytes. Latanoprost acid, TAF-A, TRA-A, BIM-A, and PGF2a suppressed adipogenesis in a dose-dependent manner (P < 0.05 or 0.01), but BIM and UNO did not affect adipogenesis at all even at high concentrations (100 nM and 1000 nM). Travoprost acid affected adipogenesis only at 100 nM (P < 0.05). Latanoprost acid suppressed adipogenesis at 10 and 100 nM concentrations, but not at 1 nM (P < 0.05).

Effects of PGs on Adipogenesis in Primary Cultured Mouse Adipocytes From FP Receptor Knockout Mice

In primary adipocyte cultures from FPKO mice, the relative stained areas of cells treated with 100 nM LAT-A, TRA-A, TAF-A, BIM, BIM-A, UNO, or PGF2α were 97.0% ± 15.4%, 109.4% ± 14.1%, 100.9% ± 16.2%, 92.0% ± 17.9%, 109.1% ± 17.3%, 100.9% ± 8.4%, and 94.6% ± 13.6%, respectively (n = 12) (Fig. 4A). There was no significant difference in these areas among PGs. Relative stained areas of cultures from wild-type mice treated with these PGs were 61.7% ± 29.4%, 69.8% ± 32.3%, 62.8% ± 30.9%, 93.4% ± 29.1%, 52.5% ± 31.5%, 96.2% ± 22.9%, and 70.6% ± 13.6%, respectively (n = 12) (Fig. 4B). LAT-A, TAF-A, and BIM-A strongly inhibited adipogenesis (P < 0.01). TRA-A and PGF2α moderately suppressed adipogenesis (P < 0.05). 

Figure 4

 

Adipogenesis in primary cultured mouse adipocytes lacking FP receptor. Areas stained by oil red O on day 10 of mouse adipocyte primary cultures from FP receptor knockout (A) and wild-type mice (B) were measured and are indicated as percentage of that in control culture; 100-nM PGs were added on day 2 after initiating differentiation. Asterisks indicate significant differences versus control. *P < 0.05; **P < 0.01 (Dunnett's test).

Discussion

Deepening of the upper eyelid sulcus is characterized by deepening of the upper eyelid sulcus. The frequency of this condition was reported in our clinical trial of BIM ¹⁰ and by Shah et al. ² Although our report indicated a high frequency of DUES with BIM treatment, this side effect shows different clinical aspects from injection or hyperpigmentation. In our prospective clinical trial, ¹⁰ only some of the patients with DUES noticed it. One patient stopped BIM treatment for cosmetic reasons, whereas the other patients did not. Interestingly, some patients favored DUES as a cosmetically beneficial effect. Deepening of the upper eyelid sulcus and other side effects of prostaglandin eye drop use have not shown any adverse effects on ocular function, such as dry eye due to opening of the eyelid, incomplete blinking, or restriction of ocular movement, ² but to date, it is not clear whether DUES may gradually induce ocular dysfunction. Thus, it is important to clarify the mechanism of DUES development. 

The current commercially available PGs can be classified into three types. Latanoprost acid, TRA-A, BIM-A, and TAF-A are acid forms of prost-type PG analogues, derived from PGF2α by the addition of a phenyl base at C-17 and conservation of the C-15 hydroxyl base, which is important for binding to the prostanoid FP receptor. Bimatoprost itself is a prostamide type, in which the C-terminal of PGF2α is replaced by ethylamide. BIM directly binds to a dimer prostamide receptor composed of the FP receptor and an FP receptor splice variant. ²⁹ Unoprostone, an acid form of isopropyl unoprostone, is a prostone-type PG analogue that is a type of metabolized PGF2α subjected to oxidation of the C-15 hydroxyl base. Thus, UNO exhibits decreased binding affinity for the FP receptor. ^30,31 In this study, prost-type PGs and PGF2α all suppressed adipogenesis in a dose-dependent manner, but the prostone-type UNO and the prostamide-type BIM did not. This strongly suggests that suppression of adipogenesis is a result of FP receptor stimulation (Figs. 2, 3). Moreover, the inhibitory effect of PGs on adipogenesis was completely abolished in FP receptor–deficient adipocytes (Fig. 4). Deepening of the upper eyelid sulcus may be caused by reduced adipogenesis due to FP receptor stimulation by penetration of the drugs into orbital fat. 

Some other recent publications have investigated the effects of PGs on adipocytes. ^23,24 Seibold et al. ²³ reported that BIM, travoprost, latanoprost, and tafluprost suppress preadipocyte proliferation, whereas timolol and benzalkonium chloride showed significant and dramatic antiproliferative effects and cytotoxicity in adipocytes. However, prodrug-type PGs, and not acid types binding to the FP receptor, were used in this assay. Thus, the results may be quite different from those of our study, in which acid-type PGs were used. Moreover, cytotoxity was evaluated over only a short period by counting cell numbers and not by measuring adipose tissue volume. Thus, this rapid toxic effect in vitro might differ from what occurs with clinical long-term use leading to the development of DUES. Choi et al. ²⁴ also showed inhibition of adipocyte differentiation by prostaglandins using human preadipocytes, a closer model for the effects in humans, but they used prodrug PGs, which do not bind the FP receptor directly, diluted with commercial products. 

One drawback of this study is that we used the 3T3-L1 cell line. The 3T3-L1 is a well-known and very novel preadipocyte cell line that can differentiate to form adipocytes. ³² Our findings in 3T3-L1 cells may not be applicable to human orbital adipocytes. However, the biological assay used requires a large number of cells and excision of human orbital adipocytes presents a significant ethical problem. To clarify the mechanism of adipogenesis, primary cultured mouse adipocytes were also used. 

Our in vitro data were similar to those of other recent clinical reports. Figure 3 indicates that LAT-A, TAF-A, and TRA-A had a similar inhibitory effect to BIM-A at concentrations over 10 nM. Bimatoprost acid tended to suppress adipogenesis even at the lower concentration of 1 nM, but LAT-A did not suppress adipogenesis at a low concentration (P < 0.05). Consistent with these findings, DUES has been most frequently noted with topical use of BIM eyedrop, and rarely with LAT eyedrop treatment. 

Although our results are relevant to clinical DUES, many other factors likely contribute to the mechanism of PG-related DUES, such as tissue penetration, tissue concentration, affinity for the FP receptor, and activation of the FP receptor. Activation of the FP receptor may be necessary for the initial signal to reduce adipogenesis. One observation is that the concentrations of PG eye drops vary, but they appear to have a similar IOP-lowering effect except for UNO. ^12,13 This suggests that the concentration of PG in the ocular adnexal tissues may be different. Second, the concentration of the active acid form PGs may not be similar to the original concentration of the eye drops. There are no reports on the tissue concentration of PGs in the orbita, and the levels of esterase and amidase activity in the ocular adnexal tissue are not known. So, unfortunately, we are not able to measure the actual concentrations of drugs in the orbital tissues. 

In addition to the concentrations of active forms of PG, each PG has a different affinity for and activity at the FP receptor. ^33,34 Among these PGs, tafluprost is known to have high specificity. ³⁵ Bimatoprost may have a dual mechanism and stimulate both FP and prostamide FP receptors. ³⁶ These PGs may also activate different intracellular signaling pathways. However, there have been no in vitro studies comparing the affinity of PGs for the FP receptor or comparing intracellular signaling of all PGs under the same conditions, or in ocular adipocytes. Moreover, in an in vivo situation, drug penetration and concentration in the orbital issues may differ among drugs and individuals. Thus, based on only our results, it would be inappropriate to draw conclusions about the ability of each PG to induce DUES or other PAP, and the fact that the FP receptor contributed the occurrence of DUES by suppression of adipogenesis may be obvious in our study. 

The concentration of BIM eye drops (0.03%) is the highest among PG products. Given that the degradation of BIM into BIM-A by tissue esterase is comparable to that of other PG prodrugs, ^33,37–39 the concentration of BIM-A may be relatively high in orbital tissue. Thus, that the frequency of DUES may be higher with the use of BIM eye drops than with other PGs, may be explained by either the higher concentration of drug, or by the specific pharmacological effects of BIM itself. Because there is a lack of data on tissue penetration and distribution of PGs in the periorbital tissue, further study is needed. 

In this study, we investigated suppression of adipogenesis at three different points during adipocyte differentiation, because expression of the FP receptor and signaling leading to adipogenesis may differ at different stages of adipocyte maturation. As a result, in early adipocytes, all acid-type prost PG analogues inhibited adipogenesis, whereas only BIM-A and PGF2α showed effects on differentiated adipocytes. To date, the cell cycle and turnover rates of adipocytes in the orbital tissue have not been investigated. If orbital fat tissue is composed primarily of differentiated adipocytes, the higher frequency of DUES with BIM may be due to its stronger inhibitory effect on adipogenesis in both premature and differentiated adipocytes. 

These clinical and animal data suggest that the incidence of DUES may be influenced by differences in prostanoid and prostamide agonists or by the pharmacological affinity of these drugs for FP receptors. With regard to clinical aspects, the degradation, penetration, and distribution of PGs into periorbital tissues and the degree of adipocyte differentiation may influence the incidence of DUES. Further studies are needed to clarify the relationship between this basic science research and clinical data on DUES. 

In conclusion, prost-type PG analogues have a potential to inhibit adipogenesis through FP receptor stimulation, which may lead to the reduced orbital fat observed in the clinical side effect of DUES. 

Acknowledgments

Disclosure: Y. Taketani, None; R. Yamagishi, None; T. Fujishiro, None; M. Igarashi, None; R. Sakata, None; M. Aihara, None