Aquaporins are upregulated in glandular epithelium at the time of implantation in the rat
- First Online:
- Cite this article as:
- Lindsay, L.A. & Murphy, C.R. J Mol Hist (2007) 38: 87. doi:10.1007/s10735-007-9083-8
- 100 Views
Regulation of luminal fluid is essential for blastocyst implantation. While it has been known for quite some time that there is a reduction in the amount of luminal fluid at the time of implantation, the mechanisms regulating this process are only just emerging. Previous studies have shown an upregulation of aquaporin (AQP) 5 channels in luminal epithelial cells at the time of implantation providing a mechanism for fluid reabsorption across the surface epithelium. However to date the contribution of fluid reabsorption by glandular epithelial cells has not been established. This study using reverse transcriptase polymerase chain reaction demonstrates the presence of several AQP isoforms in the rat uterus at the time of implantation while immunofluorescence data demonstrates an apical distribution of AQPs5 and 9 in the glandular epithelium at the time of implantation. The presence of AQPs5 and 9 in the apical plasma membrane of the glandular epithelium seen in this study provides a mechanism for transcellular fluid transport across these glandular epithelial cells similar to that seen in luminal epithelial cells. The reabsorption of glandular fluid via AQP channels may also regulate luminal fluid volume and be involved in the reduction in luminal fluid seen at the time of implantation.
The uterine luminal fluid which bathes the implanting blastocyst is tightly regulated. There is a dramatic reduction in the amount of luminal fluid at the time of implantation which contributes to the close apposition between the trophoblastic and luminal epithelial cells (Enders and Schlafke 1967), a process which is essential for successful implantation. Previous studies have suggested a role for aquaporin (AQP) water channels in rat uterine luminal epithelial cells in the reduction of luminal fluid at the time of implantation and in positioning of the blastocyst (Lindsay and Murphy 2004b, 2006). The aim of this current study is to identify the contribution of the glandular epithelial cells to the regulation of this luminal fluid and the role that AQPs play in regulation of fluid transport across the glandular epithelium.
The glandular epithelium of the mouse uterus is known to synthesise and secrete proteins during the peri-implantation period (Given and Enders 1981), while ion transport has been observed in human glandular epithelial cells, particularly the absorption of sodium and calcium ions and secretion of potassium ions (Matthews et al.1993). Hence, in addition to regulation by surface epithelial cells, luminal fluid contents and volume could also be regulated by absorption and secretion from glandular epithelial cells, however mechanisms regulating this are currently unknown.
Aquaporins are a family of transmembrane water channels found in a variety of cells and tissues (Borgnia et al.1999; Verkman and Mitra 2000). Members of the AQP family are divided according to permeability characteristics (Verkman and Mitra 2000). Classical AQPs are permeable to water alone and include AQPs0, 1, 2, 4, 5 and 6. The second group, the aquaglyceroporins are permeable to water, glycerol and urea and consist of AQPs3, 7, 8 and 10, while AQP9 is a member of the neutral solute family and is permeable to water, glycerol, urea, purines, pyrimidines and monocarboxylates (Verkman and Mitra 2000).
Aquaporin molecules were originally discovered in the human uterus in 1994 (Li et al.1994) and since then several studies have investigated this family of channels in the uterus during early pregnancy. AQP1 is found in endometrial endothelial cells and in the inner circular layer of myometrium (Richard et al.2003; Lindsay and Murphy 2004a). Furthermore, an increase in AQP1 staining was seen in mesometrial myometrium at the time of implantation, where it may contribute to implantation position of the blastocyst (Lindsay and Murphy 2004a). AQP2 has not been localised in the pregnant mouse uterus (Richard et al.2003) however was seen to be upregulated in response to oestrogen stimulation (Jablonski et al.2003). In humans, AQP2 was found in glandular and luminal epithelial cells with an increase in protein and mRNA during the mid secretory phase, which corresponds with the time of implantation (He et al.2006).
Aquaporin 3 was also only found in the mouse uterus in ovariectomised control and oestrogen treated animals (Jablonski et al.2003) while microarray studies suggests the presence of AQP3 in human uterine luminal epithelial cells (Mobasheri et al.2005). AQP4 could not be localised in the rat uterus using immunohistochemical techniques (Lindsay and Murphy 2004b), however AQP4 was found by in situ hybridisation in the pregnant mouse uterus (Fujita et al.1999).
Another member of the AQP family which appears to play an important role in early pregnancy is AQP5. Initially AQP5 mRNA was localised to mouse uterine luminal and glandular epithelial cells, while immunohistochemical staining revealed basolateral expression in the glandular epithelium only (Richard et al.2003). Further studies demonstrated the presence of AQP5 in the rat uterus (Lindsay and Murphy 2004b) where AQP5 was present in the apical plasma membrane of luminal epithelial cells and an increase in expression seen in mesometrial epithelial cells at the time of implantation, suggesting a role for luminal fluid absorption across the luminal epithelium at this time (Lindsay and Murphy 2004b). Furthermore the appearance of AQP5 in the apical plasma membrane was dependent on progesterone (Lindsay and Murphy 2006), however the asymmetrical expression of AQP5 across the lumen was only seen in the pregnant uterus. Neither AQP6 nor AQP7 has been identified in uterus to date.
In the uterus, AQP8 was localised to placenta (Ma et al.1997) and inner cell mass of the blastocyst as well as in decidualised stroma and receding uterine glands in the mouse uterus (Richard et al.2003). In placenta, AQP9 is highly expressed in syncytiotrophoblastic cells, where it is thought to play a role in fluid and urea transport (Damiano et al.2006). There has however been no evidence of AQP9 in the uterus during early pregnancy.
While the mechanisms of fluid transport across the luminal uterine epithelium are beginning to emerge, there is no such evidence for glandular epithelial cells. In view of the potential role that glandular epithelial cells play in the regulation of luminal fluid it would be interesting to investigate the contribution that AQP channels play in these cells. Hence, the aim of this study is to investigate which members of the AQP family are present in the glandular epithelium during early pregnancy and the change in distribution of these molecules at this time. This work adds to the emerging picture of the role that AQP molecules play in regulation of the luminal fluid environment at the time of implantation.
Materials and methods
Twenty-five female, adult Wistar rats were used in this study and housed at 21°C with 12 h light-darkness cycle and given food and water ad libitum. In order to establish the stage of the oestrous cycle, vaginal smears were taken in the afternoon and those found to be in proestrus were housed in a cage overnight with a male of proven fertility. The following morning females were checked for the presence of spermatozoa in the vagina, if sperm were present this was classified as day 1 of pregnancy. Rats were selected at days 1, 3, 6, 7 and 9 of pregnancy in order to establish changes in the uterus prior to implantation (days 1 and 3), at the time of implantation (days 6 and 7) and post implantation (day 9). At the appropriate day of pregnancy, rats were euthanaised with an intraperitoneal injection of sodium pentobarbitone (Euthal, Delvet, NSW, Australia) and the uterine horn was immediately removed and placed into ice-cold 0.1 M phosphate buffer (pH 7.4).
Polymerase chain reaction (PCR) was performed to identify which of the nine AQP isoforms are present in the rat uterus during early pregnancy. Tissue for PCR was cut into approximately 2–3 mm pieces before being immediately frozen in liquid nitrogen where it was stored until use. Tissue was ground to a powder in a mortar and pestle and then placed in 1 ml of Tri-Reagent (Molecular Research Centre Inc, OH, USA) for approximately 15 m at room temperature (RT). Chloroform (Selby Biolab, VIC, Australia), 200 μl was added, tubes were vigorously shaken and then centrifuged at 12,000 rcf for 10 m at 4°C. The supernatant was placed into a fresh tube containing 0.5 μl glycogen (Sigma, MO, USA) along with 500 μl of isopropyl alcohol (Fluka Biochemica, Buchs, Switzerland) and mixed before incubating at RT for 15 m. Tubes were then centrifuged at 12,000 rcf for 10 m at 4°C, supernatant was discarded and then the pellet was washed in 1 ml 75% ethanol, mixed and centrifuged at 7,400 rcf for 5 m. Ethanol was removed and the pellet was allowed to dry before being dissolved in RNase free water. RNA was then purified using a GenElute Mammalian Total RNA Miniprep kit (Sigma) according to manufacturer’s instructions, quantified via spectrophotometry and stored at −80°C until further use.
Reverse transcription-polymerase chain reaction
RNA samples (1 μg) were incubated in 0.5 μg oligo(dT)15 primer (Promega, WI, USA) at 70°C for 10 m and then reverse transcribed. The reaction buffer contained 50 mM Tris–HCl (pH 8.3), 75 mM KCl, 3 mM MgCl2, 10 mM DTT, 10 mM dNTPs, 25 U recombinant ribonuclease inhibitor (RNasin, Promega) as well as 200 U M-MLV reverse transcriptase (Promega) and was incubated at 42°C for 1 h. The reaction was then incubated at 70°C for 15 m and stored at −20°C until further use.
Polymerase chain reaction was carried out by the use of specific primers to each of the nine isoforms of AQPs. Each reaction contained 20 mM Tris–HCl (pH 7.5), 100 mM NaCl, 0.1 mM EDTA, 2 mM DTT, 1.5–2 mM MgCl2, 200 μM-1 mM dNTPs, 1 μM of gene specific primers, 1 U DNA polymerase (Bioline, London, England) as well as 2 μl of cDNA. The reaction mixture was placed in a Mastercycler Gradient (Eppendorf, Hamburg, Germany) and initiated at 95°C for 7 m, then 35 cycles consisting of denaturation at 95°C for 30 s, annealing of primers to target sequence at 52–58°C for 30 s followed by elongation at 72°C for a further 30 s. Final extension at 72°C for 10 m ended the reaction. Reaction products were mixed with loading dye (Bioline) and ran on a 1.5% agarose gel (Amresco, OH, USA) containing 1 μg/ml ethidium bromide (Amresco) for 30 m at 100 V before being imaged and photographed under UV light. Product size was estimated by running a DNA ladder (Hyperladder IV, Bioline).
Primers for AQPs and β-actin
Ensembl Gene ID
Fwd primer: GCTATTGCAGCGTCATGTCTG
Rev primer: TAGCAGAAGCCACAGCATCAG
Fwd primer: TCCTTCCTTCGAGCTGCCTTC
Rev primer: TGGATTCATGGAGCAGCCGGT
Fwd primer: CCTCTGGACACTTGGATATGG
Rev primer: CAGCTTCACATTCTCTGCCTC
Fwd primer: CTCTGCTTTGGACTCAGCATTG
Rev primer: TTCCTTTAGGCGACGTTTGAG
Fwd primer: ATCTACTTCACCGGCTGTTCC
Rev primer: GTCAGCTCGATGGTCTTCTTC
Fwd primer: AGTCAACGTGGTCCACAACAG
Rev primer: GTTGTAGATCAGCGAAGCCAG
Fwd primer: AACTGTGCACTAGGCCGAATG
Rev primer: GTGATGGCGAAGATACACAGC
Fwd primer: GAACATCAGCGGTGGACACTT
Rev primer: CAATGAAGAGCCTAATGAGCA
Fwd primer: AGCCTGTTGTCATTGGCCTCC
Rev primer: GTTCTCAGATGGCTCTGCCTT
Fwd primer: CATGTACGTAGCCATCCAG
Rev primer: AAACGCAGCTCAGTAACAG
Tissue for immunofluorescence microscopy was cut into 5 mm pieces, immersed in OCT compound (Tissue Tek, CA, USA) and immediately frozen in super cooled isopentane (BDH Laboratory Supplies, Poole, England) before being stored in liquid nitrogen until sectioning. Tissue sections 8 μm thick were then cut using a CM3050 cryostat (Leica, Heerbrugg, Switzerland), placed onto gelatin-chrome alum coated slides and allowed to air dry at room temperature. Two blocks per animal were cut and sections were randomly selected for control and experimental slides. Interimplantation sites were selected particularly on days 7 and 9 of pregnancy to ensure glandular epithelium was present in the section. After being fixed in freshly prepared 4% formaldehyde, sections were washed with PBS and then blocked with PBS containing 5% bovine serum albumin for 30 m. This solution was also used as diluent for primary and secondary antibodies. Sections were then incubated in either anti-AQP5 (3.3 μg/ml) or anti-AQP9 (10 μg/ml) antibodies (Alpha Diagnostics International, TX, USA) for 3 h at room temperature in a humid chamber. After washing in PBS, sections were maintained in limited light and incubated in FITC conjugated goat-anti-rabbit secondary antibodies at a concentration of 7.5 μg/ml for 1 h at room temperature. Sections were washed in PBS, mounted in Vectashield (Vector, CA, USA), coverslipped and viewed immediately with a Diaplan microscope (Leica). Digital micrographs were taken using a Leica DFC480 camera and micrographs produced using Photoshop software (Adobe Systems, CA, USA). Final magnifications were calculated using a stage micrometer.
Ovariectomy and hormonal treatments
A further 20 female Wistar rats were used to investigate hormonal control of changes in the distribution of AQP5 channels in glandular epithelial cells. Animals 12–14 weeks old were bilaterally ovariectomised under isoflurane anaesthesia. After recovery of 4 weeks, rats were randomly divided into four groups and treated with different hormonal regimes. Hormones were dissolved in benzyl alcohol (Sigma) and diluted in peanut oil to achieve appropriate physiological levels (Ljungkvist 1971a, b). All subcutaneous injections were given in the neck. Animals in the control group were given 0.1 ml peanut oil for three consecutive days. Group 2 animals were injected with 0.5 μg 17-β-oestradiol (Sigma) in 0.1 ml peanut oil for three consecutive days. Animals in group 3 were given three consecutive injections of 5 mg progesterone (Sigma) in 0.2 ml peanut oil over 3 days, while animals in the fourth group were given 5 mg progesterone for 2 days and 0.5 μg 17-β-oestradiol as well as 5 mg progesterone on the 3rd day. All animals were euthanaised with an intraperitoneal injection of sodium pentobarbitone (Euthal, Delvet, NSW, Australia) 24 h after the final injection. Uterine tissue was removed and processed for light microscopy as described above. Immunofluorescent staining with anti-AQP5 antibodies was carried out as described earlier.
Randomly selected sections were designated as negative controls and were run in parallel with all experiments. Negative control sections were treated the same as experimental sections except for the absence of primary antibodies and were used to detect the presence of any non-specific secondary antibody binding.
Non-immune controls were also carried out where primary antibodies were replaced with normal purified rabbit IgGs (Sigma).
Controls were carried out with all experimental runs and no staining was evident in any section. Day 7 tissue is shown as a representative example (Fig. 3f).
Ovariectomy and hormonal replacement
The use of RT-PCR and indirect immunofluorescence microscopy on the pregnant and hormonally controlled rat uterus has demonstrated for the first time the presence of AQPs in glandular epithelial cells of the rat uterus. RT-PCR results demonstrated the presence of AQPs1, 5, 7, 8 and 9 in the rat uterus during early pregnancy. Further immunofluorescence studies demonstrated the presence of AQPs5 and 9 in the apical part of the glandular epithelium at the time of implantation and in response to progesterone alone or in combination with oestrogen.
The appearance of AQP1 transcript in the uterus during early pregnancy is not surprising as AQP1 has previously been localised to endometrial blood vessels and myometrium (Richard et al.2003; Lindsay and Murphy 2004a). However, the lack of AQP3 in the rat uterus is unexpected especially due to recent evidence of the involvement of AQP3 in human endometrium (Mobasheri et al.2005). Lack of AQP4 transcript seen in this study supports the absence of AQP4 protein previously described (Lindsay and Murphy 2004b).
The presence of AQP7 and AQP8 transcript but not protein in the rat uterus may be due to a lack of sufficient sensitivity of the immunofluorescence technique. The appearance of AQP8 in the inner cell mass of the blastocyst may provide a source for the positive PCR result seen on day 6 and 7 of pregnancy (Richard et al.2003). The lack of AQP8 staining is in contrast with a mouse study showing the appearance of AQP8 in the decidualised stroma (Richard et al.2003), however species differences between rat and mouse have already been described (Lindsay and Murphy 2004b), particularly in relation to AQPs in the uterus.
The apical localisation of AQP5 in the glandular epithelium during the time of implantation in the rat contrasts with results in mice where AQP5 was localised to the basolateral region of the glandular epithelium (Richard et al.2003), which may reflect a species difference, or a difference in methodology. The same study failed to demonstrate AQP5 expression in the progesterone-primed uterus but does agree with the presence of AQP5 in endometrial glands from animals treated with progesterone in combination with oestrogen, as seen in the present study. In order to strengthen the argument of a species specific distribution of AQP channels, recent studies in humans have identified AQP2 as a potential candidate for uterine water transport (He et al.2006; Hildenbrand et al.2006). Hence it appears that the phenomenon of fluid reabsorption at the time of implantation is common across species, including in humans where an increase in implantation rate is seen with a reduction in luminal fluid (Akman et al.2005), however the AQP isoform present may be species specific.
Morphological studies suggest that the glandular epithelium of the ovariectomised control animal displays both secretory and absorptive functions (Ljungkvist 1971b). Sodium-dependent fluid absorption occurs across the glandular epithelium of the progesterone-primed uterus (Naftalin et al.2002). Other morphological studies demonstrate blind-ending glands, with no evidence of opening onto the luminal surface, in ovariectomised rats treated with progesterone alone and in combination with oestrogen (Ljungkvist 1971a, 1972). Hence, the finding of AQP5 in glandular epithelial cells in the ovariectomised control and progesterone-treated animals may provide a functional mechanism for fluid absorption across the glandular epithelium in response to the osmotic gradient established by sodium ion absorption. If indeed the glands are blind-ending, as previously suggested, then appearance of AQP5 and AQP9 in the apical plasma membrane of the glandular epithelium could also function in reabsorption of glandular contents.
The appearance of AQP5 expression being progesterone dependent as seen in this study is interesting, and seems to contradict recent studies in mice where AQP5 was found to be directly regulated by oestrogen (Kobayashi et al. 2006). There are several possible mechanisms explaining this, such as a species difference, as highlighted above. In addition to this, the study by Kobayashi et al. (2006) used immature mice and a relatively short time course, up to 24 h, where as in this study mature ovariectomised rats were used and were treated with hormones for several days, with tissue removed 3 days after the first injection. Due to the fact that there seems to be two phases to the oestrogen response (Anderson et al.1972), perhaps the results seen by Kobayashi et al. (2006) are indicative of the early phase response to oestrogen. Results presented here also indicate that AQP5 is upregulated in response to progesterone in combination with oestrogen and perhaps in this combination AQP5 may be upregulated via oestrogen response elements found in the promoter region of this gene (Kobayashi et al. 2006). While it was demonstrated in this study that AQP5 was present in the progesterone dominant regimes, both in ovariectomised animals treated with exogenous progesterone and at the time of implantation, a specific hormone dependent response was not examined in this study but future work in this area may reveal the importance of progesterone regulation of AQP5.
The glandular epithelium is continuous with the luminal epithelium and hence it is not surprising that similar mechanisms of fluid transport should occur across these epithelial barriers. Furthermore, the glandular epithelium contributes to the content and volume of luminal fluid. AQPs5 and 9 molecules in the apical plasma membrane of the glandular epithelium demonstrated in this study may contribute to the decrease in uterine fluid volume and hence the closing down of the uterine lumen seen at the time of implantation (Enders and Schlafke 1967) and in response to progesterone alone or in combination with oestrogen (Ljungkvist 1971a, 1972). In addition, the absorption of fluid through AQP5 and AQP9 molecules across the glandular epithelium in the receptive endometrium and in the progesterone-treated animal, could apply a suction force at the endometrial surface (Naftalin et al.2002), which may also contribute to closing down of the uterine lumen. While it is not probable that this force alone would be sufficient for closing down the uterine lumen, this mechanism may play a role in conjunction with reabsorption of water across the luminal epithelium via AQP5 channels (Lindsay and Murphy 2004b, 2006) to assist in the uterine lumen closure seen at the time of implantation. The sodium reabsorption characteristics of human glandular epithelial cells (Matthews et al.1993) provides a mechanism for fluid transport across the glands, similar to that occurring across the surface epithelium.
Thus at the time of implantation there is an upregulation of AQP5 channels in both the luminal and glandular epithelial cells, and an upregulation of AQP9 in glandular epithelium. Previous work has shown a differential gradient in luminal epithelial AQP5 staining with an increase in mesometrial epithelial cells, presumably creating a water gradient across the lumen, thus influencing implantation position (Lindsay and Murphy 2004b). However the expression of AQPs5 and 9 was equal amongst glands from mesometrial and antimesometrial poles of the uterus, indicating a function purely for fluid reabsorption. Hence the reabsorption of luminal fluid seen at the time of implantation, and in response to progesterone administration, is likely to be mediated by absorption of fluid via AQP channels across both the luminal and glandular epithelial surfaces.