The effects of different inhibitory pathways of prostaglandin E2 biosynthesis on renomedullary interstitial cells in rats: a multidisciplinary study

Renomedullary interstitial cells (RMICs) are the most dominant cell type in inner renal medulla. Their most distinct characteristic is the presence of multiple lipid droplets in their cytoplasm. These lipid droplets are believed to be the storage units for precursors of prostaglandins (PGs), prostacyclin and medullipin. Especially prostaglandin E2 (PGE2) is synthesized by RMICs in kidney. PGs are produced by three key steps: 1) Arachidonic acid (AA) release from membrane phospholipids by the action of phospholipase A2 (PLA2); 2) Formation of prostaglandin H2 (PGH2) from AA by the action of cyclooxygenases (COXs); 3) Specific PG synthesis metabolism from PGH2. PG biosynthesis can be regulated via activation or inhibition of these steps. In this study, we examined the effects of PGE2 inhibition in different steps on RMIC function, the number of lipid droplets, medullary hyaluronan (HA) content and cell viability. We formed four groups (n=8): First group was control and treated with intraperitoneal (ip) 0.9% saline. Second group in which we inhibited AA release from membrane phospholipids was injected with ip dexamethasone (DEX) (2 mg/kg, 10 days); third group was treated with ip indomethasine (IND) (1 mg/kg, 10 days) to inhibit non-specific COX at the stage of PGH2 formation from AA; and the fourth group was injected with ip celecoxib (CXB) (1 mg/ kg, 10 days) to examine selective cyclooxygenase-2 (COX-2) inhibition. We dissected renal medulla of the sacrificed animals after 10 days to analyze with light and electron microscopy. We counted the lipid droplets in 50 random RIMCs for each animal (x6.000 magnification) in electron microscopy. Our morphometric analysis showed that the number of lipid droplets was significantly decreased in DEX group and was significantly increased in IND and CXB groups when compared to control. In addition, medullary HA content and CD44 immunoreactivity were significantly increased in all groups when compared to control. When we analyzed cell viability, we found that RMIC apoptosis was significantly higher in PGE2 inhibited groups when compared to control. Besides this, 24-hour urine values collected on the 10th day were significantly increased in dexamethasone and indomethacin groups; but in celecoxib group the values were similar to control. These results indicate that lipid granules may be numerical and functionally influenced from PGE2 changes, these granules may be storage units of AA, functional changes in RMICs by PGE2 may influence HA quantity of medullary interstitium and urine volume, and finally PGE2 inhibition may lead to RMIC apoptosis.


Introduction
The kidney has both pro-and anti-hypertensive activities [1,2].
Renomedullary interstitial cells (RMICs) are the most abundant cell type in the inner renal medulla and they have several char-doi: 10.7243/2055-091X-4-13 acteristic features. Their most distinctive characteristic is the presence of multiple lipid droplets in their cytoplasm. The investigators suggest that RMICs are secretory cells because of their well-developed rough endoplasmic reticulum (RER), Golgi complex and the content of multiple lipid droplets [3]. Therefore, they are referred to as the lipid-containing type 1 interstitial cells, which are the principal cells in the inner medulla [4]. In 1969, Bohman and Maunsbach isolated lipid droplets from RMICs and found that they had a vasodepressor activity. The content of these lipid granules are believed to be storage units for precursors of prostaglandins (PGs), prostacyclin and medullipin [1]. The amount and the size of lipid droplets are shown to be involved in the changes of diuretic stages, water-salt balance, renal perfusion pressure (RPP) and medullary blood flow (MBF) [1]. RMICs are often arranged in rows between the loops of Henle and vasa recta, with their long axes being perpendicular to those of adjacent tubules and vessels, thus resembling the rungs of a ladder [3]. This anatomic arrangement suggests that RMICs may play an important role in maintaining urinary concentrating ability via preventing the axial diffusion in concentration gradient [5,6]. Studies demonstrate that the transplants of fragments from the renal medulla or of the cultured RMICs reverse renoprival hypertension [1,[7][8][9][10][11][12][13]. The biochemical analyses of these medullary transplants show the presence of prostaglandins (PGs), triglycerides, cholesterol esters and free fatty acids [14]. At the end of these studies, three types of substances with potential antihypertensive properties are derived from the kidney medulla [12]: 1) Prostaglandins (PGs) 2) Antihypertensive neutral renomedullary lipid (ANRL) 3) Antihypertensive polar renomedullary lipid (APRL). Cell culture studies demonstrate that PGs, especially prostaglandin E 2 (PGE 2 ) are synthesized by RMICs in the kidney [3,7,12,13]. PGs are produced by three key steps: 1) Arachidonic acid (AA) release from membrane phospholipids by the action of phospholipase A 2 (PLA 2 ); 2) Formation of prostaglandin H 2 (PGH 2 ) from AA by the action of cyclooxygenases (COXs); 3) Specific PG synthesis metabolism from PGH 2 . PG biosynthesis can be regulated via activation or inhibition of these steps. Various researchers work with PGE 2 stimulators and inhibitors in their renal medulla studies; however, we are unable to find a research on inhibition before AA release in the literature. We believe that using pre-AA release inhibitors in addition to COX inhibitors better clarify the characteristic of lipid droplets and whether their activity in PGE 2 synthesis continues via other pathways or not. Selective cyclooxygenase-2 (COX-2) inhibitors (coxibs) and several non-selective COX-inhibiting nonsteroidal anti-inflammatory drugs (NSAIDs) block endogen PG synthesis. However, studies suggest that, this inhibition leads to RMIC apoptosis [15][16][17][18][19][20]. Furthermore, hyaluronan (HA) which is a negatively charged glycosaminoglycan (GAG) found in renal interstitium and in the synthesis by RMICs, may play an important role in maintaining of water homeostasis [21,22]. The high concentrations of HA in the kidney medulla support tubules and blood vessels and it is important for urine concentration [22]. HA also has a wondrous renal water handling capacity which is important for maintaining the concentration gradient and affects water transport in renal medulla [21][22][23].
In this study, we examine the effect of inhibition of PGE 2 synthesis in different pathways on RMIC function, the number of lipid droplets, medullary HA content and cell viability. The glucocorticoids inhibit PLA 2 enzyme in AA release stage from membrane phospholipids. Thus, they reduce the formation of all the eicosanoids (PGs, prostacyclins, thromboxanes and leukotrienes). For this reason, we inhibit AA release with PLA 2 inhibition by using dexamethasone (DEX) which is a glucocorticoid. AA released via PLA 2 is converted into PGG 2 by being presented to COX (prostaglandin H synthase, PGHS). Then, COXs generate intermediate endoperoxide PGG 2 which is a substrate for terminal synthases by reducing PGG 2 . We inhibit PGE 2 formation from PGH 2 in two different pathways via COX. For this study, we use indomethacin (IND) which is a non-selective COX inhibitor (COX-1 and COX-2 inhibition), and celecoxib (CXB) which is a selective COX-2 inhibitor.
We aim to analyze the morphological changes and the number of lipid droplets of RMICs in different PGE 2 inhibition groups. Furthermore, we investigate how medullary HA content, amount of urine and cellular viability from PGE 2 inhibition are affected. Finally, we search for an answer: Are the lipid droplets the storage units for PG precursor AA?

Methods
Thirty-two adult male Wistar albino rats weighing 180-200g (Experimental Animals Reproduction and Research Laboratory, Istanbul University Cerrahpasa Medical Faculty, Turkey) were housed in individual cages in a temperature-and humidity-controlled room with a 12-h light/dark cycle. They were fed with standard rat chow and had free access to tap water. Rats were divided into four groups (n=8). Control group was daily injected with 1 ml ip 0.9% NaCl solution during 10 days. Second group was daily injected with ip 2 mg/kg dexamethasone during 10 days. Third group was injected with ip 1 mg/kg indomethasine during 10 days and the last group was injected with ip 1 mg/kg celecoxib during 10 days. All rats were sacrificed under anesthesia at the end of the study (10th day) according to the regulation of animal ethical committee of Istanbul University.

Biochemical analyses
24 h urine samples were collected on the 1st and 10th days from animals. Urine volumes were measured using micropipets and pH values were measured using pH-indicator strips (Merck; pH:0-14). The blood samples were collected on the 10th day from animals via intracardiac-punction method. Samples were taken in heparinisetubes and centrifuged in 1500 rpm for 15 minutes. Serum samples collected and PGE 2 doi: 10.7243/2055-091X-4-13 levels were determined by ELISA method.

Light microscopical analyses
Renal medullar tissues were fixed in 10% neutral buffered formalin dehydrated with graded ethanol and embedded into paraffin. The sections of 5μm thickness were placed onto poly-l-lysine-coated slides. Histological determination of renal medullar HA content was carried out by Alcian blue staining (pH: 2.5). The evaluation of the Alcian blue staining was done utilizing H-SCORE from renal papilla and inner medulla, using randomly selected 5 different fields for each slide, x40 magnification.

Immunohistochemical analyses
The sections of 5 μm thickness were placed onto poly-l-lysine-coated slides. After deparaffinization and rehydration samples were placed in citrate buffer (pH6), antigen retrieval was performed by treating the samples in a microwave oven at 750 W for 5 min three times, and then slides were washed in PBS. Immunohistochemistry procedure was performed using standard streptavidin-biotin-peroxidase method. Endogenous peroxidase activity was blocked by 3% hydrogen peroxide (Dako, Glostrup, Denmark). Each section was then incubated for 15 minutes at room temperature with blocking solution to block cellular peroxidase activity. After blocking, sections were incubated with primary antibodies mouse anti-rat CD44 diluted 1:100 (Life Span Biosciences, USA) and rabbit anti-human, rat, mouse active caspase-3 diluted 1:100 (Millipore) for 1 hour at room temperature, then washed with PBS. Specific staining was performed with biotinlated universal secondary antibody and horse radish peroxidase streptavidin-complex. Immunoreactivity was developed using amino-ethyl-carbazole as chromogen (Invitrogen). The sections were counter-stained with Mayer's hematoxylin and mounted with glycerol gelatin (Sigma). All incubations were performed in a moist chamber at room temperature. As for negative control, distilled water was performed instead of primary antibody.

Electron microscopical analyses
After necropsy, the left kidney medulla was immediately divided into 1mm 3 pieces for transmission electron microscopy. They were firstly fixed in 3% glutaraldehyde in a 0.1 M phosphate buffer solution (PBS), post-fixed by 1% OsO 4 prepared in the same buffer solution, dehydrated and embedded into araldite medium. Semi-thin sections were cut into 1 µm thickness by glass knives to help of the ultra microtome and stained with 1% toluidine blue (prepared with 1% borax in bidistilled water). The sections were examined under a binocular light microscope using immersion objective. Ultrathin sections were obtained in 50 nm thickness onto copper grids (200 meshes) with the same microtome, stained with uranyl acetate and lead citrate and they were investigated by transmission electron microscope (Jeol JEM 1011).

Morphometric studies
We obtained ultra-thin sections from one animal in each experimental group and counted lipid droplets in randomly chosen 50 different RMICs with EM micrographs (x6000 magnification). Data was statistically evaluated.

H-Score and semi-quantitative evaluations
The evaluations of the alcian blue staining and immunohistochemical staining were done utilizing H-SCORE. From each randomly selected slides per animal, five different fields were evaluated. Stainings were scored in a semi-quantitative fashion that included the intensity of specific staining in sections. The evaluations were recorded as values between 0-300 of five intensity categories: 0 (none), 0-75 (weak but detectable above control), 75-150 (distinct), 150-225 (dense), 225-300 (intense). Two observers blinded to the experimental groups performed the H-SCORE. For each tissue, obtained an average H-SCORE value. (0:negative; 1:weak positive; 2:positive; 3:dense positive; 4:very dense positive and presented in Figure 1 and Table 1).

Statistical analyses
Statistical analyses of 1st and 10th days urine volumes, 1st and 10th days urine pH values, PGE 2 values in blood serum, HSCORE values of alcian blue staining and immunohistochemistry analyses and the number of lipid droplets were compared by One-Way ANOVA tests. Statistical calculations were performed using Sigma Stat for Windows, version 3.0 (Jandel Scientific, San Rafael, CA). Comparisons were made versus the inhibition groups as a control. Probability values of less than 0.05 (p<0.05) were considered significant; values are presented as mean ±SEM.

Biochemical findings
In control and experimental groups, urine volumes (ml/24 hours) and pH (0-14) values are shown in Table 1 and Figure 1. 1st day urine volumes were not significantly different from each other (p>0.05). 10th day urine volumes were significantly different between control and experiment groups (p=<0.001). Urine volumes were found to increase in DEX and IND groups, while they were similar in CXB groups when compared with control (Table 1 and Figure 1). The pH values of 1st and 10th day urine samples were not statistically significant (p>0.05) ( Table 1 and Figure 1). We validated our experiment by measuring PGE 2 levels by ELISA, in blood serum taken on the 10th day. In all inhibitions groups, PGE 2 levels were shown to be significantly lower than control (p=<0.05) (Table 1 and Figure 1).

Light microscopic findings
Alcian blue dyes non-sulphatated acid mucopolysaccharides and acetic mucins. In this study, we used this dye to determine HA, a non-sulphatated GAG in interstitial field [24]. HA levels were shown to be significantly elevated in both areas in inhibition groups when compared to control (p<0.05) (Table 1) Figure 1. The graphics of urine volume, urine pH, serum PGE 2 levels and number of lipid droplets. The groups being statistically different from control group remarqued wıth star. doi: 10.7243/2055-091X-4-13 (Figure 1 and 2). Semi-thin sections were dyed with toluidine blue and examined with immersion objective (x100). Increased lipid droplets were observed in groups where PGE 2 cannot be produced because of the COX inhibition after AA step (IND, CXB) (Figure 2).

Immunohistochemical findings
Cells that show a red reaction were accepted as immune positive (Figure 3). In Cas-3 staining there was significant difference between groups (p<0.05) (Table 1 and Figure 1). PGE 2 inhibition was shown to increase apoptosis when compared to control (p<0.001) (Figure 3). H-SCORE values determined for CD44 were significantly different between inhibition and control groups (p<0.05) (Table 1 and Figure 1). PGE 2 inhibition was shown to increase CD44 receptor numbers when compared to control (p<0.001) (Figure 3).

Ultrastructural findings
In control group, RMICs were found to be located between ascending and descending loops of Henle, capillaries and collecting tubules in longitudinal sections (Figure 4). RMICs were shown to have a poligonal shape in this group, with multiple cytoplasmic inclusions (Figure 4). Cytoplasmic density was normal. Most interestingly, osmophylic lipid granules of various sizes were present in the cytoplasm (Figure 4). Nucleus was usually round or eliptoic, with condensed heterochromatine peripherally (Figure 4). In some of the analyzed cells,   outer nuclear membrane was observed to enlarge and form perinuclear cisternas (PNC) (Figure 4). Round shaped mitochondria with less crista and very dense matrix were observed (Figure 4). Lysosomes were consisting of lysosomal material (Figure 4). SER and perinuclear location of Golgi complex was well determined in cells (Figure 4). In some of the samples, RER was also present. Free ribosomes were also observed in some samples (Figure 4).
In DEX group, cellular lipid droplets in RMICs were significantly decreased (Figure 4). Heterochromatin was increased and cells were deprived of organelles (Figure 4). ER cisternas were usually empty (Figure 4). Most specific property of these groups was endothelial hyperthrophy (Figure 4). Lysosomes were filled with granular material (Figure 4). Mitochondrion was rarely observed (Figure 4).
In IND group, lipid droplets in RMICs were significantly increased (Figure 4). Heterochromatin was increased and heterogenly located in the nucleus (Figure 4). ER cisternas were enlarged, containing granular material (Figure 4). PNCs were also enlarged with granular material (Figure 4). Lipid droplets stacked in cytosol, since they were not used, were marked to be degraded, engulfed in lysosomes, and then formed myelin structures (Figure 4). In this group, the frequency of apoptotic cells were highly increased (Figure 4).
In CXB group, lipid droplets in RMICs were significantly increased (Figure 4). Cells were euchromatic and rich in organelle, like an active cell (Figure 4). ER was well developed and enlarged ER cisternas were present (Figure 4). Also developed Golgi system had a perinuclear location. Mitochondria were observed to be hypertrophic (Figure 4). As in the previous group, this group also had increased apoptotic cells (Figure 4).

Morphometric findings
The number of lipid droplets in randomly selected 50 RMICs was counted in ultra-thin sections prepared for electron microscopy analysis, x6000 magnification. Data was analysed by One Way ANOVA (Table 1) (Figure 1). In the DEX group, where AA formation via PLA 2 was inhibited, number of lipid droplets was significantly decreased when compared with control (p=<0.001) (Table 1 and Figure 1). Contrarily, number of lipid droplets were shown to be significantly increased in IND and CXB groups, compared to control (p=<0.001) (Figure 1).

Discussion
Kidneys are known to show pro-and anti-hypertensive activity. The pro-hypertensive effect of kidney is regulated by locally active vasopressor systems, including renin-angiotensin system; while the anti-hypertensive effect is thought to depend on vasodepressor agents [1,13]. Muirhead et al. showed that renal cortical fragment transplantation does not inhibit malign hypertension development, while transplantation of renal medullar fragments of RMICs in culture overcomes renoprival hypertension [7,[9][10][11][12][13]25]. Biochemical analyses revealed that these transplants contain abundantly PGs, triglycerides, cholesterol esters and free fatty acids [14].
Renal medulla is the primary synthesis location for PG synthesis. Tissue culture studies on RMICs revealed that these cells are specialized in PG synthesis [1,13]. It is well known that the most specific property of RMICs is the numerous lipid droplets they contain [1,3,8]. The number and size of these lipid droplets change according to the species and physiological state [1]. The number and size of lipid droplets are reported to decrease in the 24 hours following dehydration [1]. In the water-restored rats, the number and size of lipid droplets increased [1]. Although there is not a clear connection identified between lipid droplets and diuretic state of the animals, further studies should be performed to confirm this. The content of these lipid droplets is also not completely known but they are thought to store pivotal products of PGs [1]. The relation between RMIC lipid droplets and renal medullar antihypertensive effect depends on these findings.
Our study also stems from the hypothesis that lipid droplets could be storage units for AA; and we examined the effect of PGE 2 inhibition at different steps on cellular lipid droplets. It is known that PGs are produced in 3 steps via renal COXs (COX-1 and COX-2) by releasing AA from membrane phospholipids [26]: 1. AA release from membrane phospholipids by PLA 2 .
2. Production of PGH 2 from AA by COX. 3. Specific PG synthesis metabolism. PG production can be regulated via activation and inhibition of these steps. Various researchers used PGE 2 inhibitors and stimulators in their studies on RMICs; however, we could not find any inhibition performed pre-AA-release in our literature studies. We designed this study to clarify how this kind of inhibition together with COX inhibition, and the co-use of specific and non-specific COX inhibitors in experimental groups effect lipid droplets, and whether the activity of these cells during PGE 2 production process continues using other pathways or not. In our study, we used a glucocorticoid DEX, to inhibit PG synthesis in AA release step by PLA 2 ; a non-specific COX inhibitor IND (NSAID), to inhibit production of PGH 2 from AA step; and a specific COX-2 inhibitor CXB (coxib).
In DEX group, cellular lipid droplet numbers were significantly decreased when compared to control. Nucleus was rich in heterochromatin and the cells were deprived of organelles, as in inactive cells. Enlarged ER cisternas were mostly empty. We assume that these cisternas are empty because PGE 2 cannot be produced, since AA synthesis is inhibited. We also observed endothelial hypertrophy and capillary narrowing in this group.
In IND group, the number of cellular lipid droplets was significantly increased. We believe this increase is the result of non-processed (because of the lack of COX) and build-up AA in the cell. Heterochromatin was found to be increased and the cellular structure was diminished and degenerated. In addition, the cells had highly enlarged ER cisternas and doi: 10.7243/2055-091X-4-13 disturbed mitochondria and Golgi. We also observed defected lipid droplets and accumulation of these granules in lysosomes.
The number of lipid droplets showed a smaller increase in CXB group (specific COX-2 inhibition group) when compared to IND group, and that the cells in the CXB group have a richer organelle appearance suggests that PG synthesis continues, although minimally, via COX-1 pathway. The fact that PGE 2 levels measured in blood serum are higher in CXB group compared to IND supports this hypothesis. 10th day urine volumes, showing an increase in DEX and IND groups while CXB values were similar to control, also propose that COX-1 pathway has a role in PGE 2 synthesis in the CXB group. In all three groups where we inhibited PGE 2 , the significant decrease in granule numbers of RMICs in pre-AA-release inhibited group by DEX, and the significant increase observed with COX inhibition after AA release by IND and CXB suggest that these granules may be storage units for AA.
It is proven by in vitro and in vivo studies that RMICs synthesize HA [23,27,28]. HA may be regulated depending on the water balance of the organism. Moreover, it has been reported that AVP decreased HA levels [22] by increasing hyaluronidase activity [29][30][31]. The primary cell receptor for HA is CD44 and it is also identified in RMIHs [28]. It is known that this receptor plays a role in cell-cell clustering, matrix-cell interaction, cell migration and receptor-specific HA internalization/fractionation. CD44 is a transmembrane glycoprotein and consists of 4 functional domains. The change of these domains leads to change in the structure and thus, changes in HA binding and the interaction of CD44 with the cytoskeleton. In our findings, we determined a significant increase in medullar HA and CD44 levels in all PGE 2 inhibited groups compared to control, suggesting an inverse relation between PGE 2 inhibition and levels of renal medullar HA and CD44.
Cell viability is also related to PGE 2 inhibition. Studies showed that COX-2 inhibition results in apoptosis in cultured RMICs [15][16][17][18][19][20]. Aoudjit et al. reported an anti-apoptotic and protective role of COX-2-originated PGE 2 in glomerular epithelial cells, mediated by EP4 [32]. Furthermore, Küper et al. found that PGE 2 inactivates pro-apoptotic protein Bad and protects cell viability in renal medullar epithelial cells [33]. In parallel with these findings, we also determined a significant increase in RMIC apoptosis in PGE 2 inhibited animals when compared to controls. These results suggest that PGE 2 inhibition results in RMIC apoptosis.

Conclusion
We suggest that the changes in PGE 2 affects lipid droplets in RMICs quantitatively and qualitatively; these granules may be the storage units for AA; functional changes in RMICs via PGE 2 affects HA levels in medullar interstitium and urine volume; and finally, PGE 2 inhibition results in RMIC apoptosis.