Fructose Synthesis and Transport at the Uterine-Placental Interface of Pigs: Cell-Specific Localization of SLC2A5, SLC2A8, and Components of the Polyol Pathway¹

Abstract

The fetal fluids and uterine flushings of pigs contain higher concentrations of fructose than glucose, but fructose is not detected in maternal blood. Fructose can be synthesized from glucose via enzymes of the polyol pathway, aldose reductase (AKR1B1) and sorbitol dehydrogenase (SORD), transported across cell membranes by solute carriers SLC2A5 and SLC2A8, and converted to fructose-1-phosphate by ketohexokinase (KHK). SLC2A8, SLC2A5, AKR1B1, SORD, and KHK mRNAs and proteins were analyzed using quantitative PCR and immunohistochemistry or in situ hybridization in endometria and placentae of cyclic and pregnant gilts, cyclic gilts injected with estrogen, and ovariectomized gilts injected with progesterone. Progesterone up-regulated SLC2A8 protein in uterine luminal (LE) and glandular epithelia during the peri-implantation period, and expression became exclusively placental, chorion and blood vessels, after Day 30. P4 up-regulated SLC2A5 mRNA in uterine LE and glandular epithelia after implantation, and the chorion expressed SLC2A5 between Days 30 and 85. AKR1B1 and SORD proteins localized to uterine LE during the peri-implantation period, but expression switched to chorion by Day 20 and was maintained through Day 85. Uterine expression of AKR1B1 mRNA was down-regulated by estrogen. KHK protein localized to trophectoderm/chorion throughout gestation. These results provide evidence that components for the conversion of glucose to fructose and for fructose transport are present at the uterine-placental interface of pigs. The shift in expression from LE to chorion during pregnancy suggests free-floating conceptuses are supported by fructose synthesized by the uterus, but after implantation, the chorion becomes self-sufficient for fructose synthesis and transport.

Introduction

Glucose is present, but fructose is the most abundant hexose sugar in porcine endometria and conceptuses (embryo/fetus and extraembryonic membranes) [1]. However, little is known about the synthesis, transport, or metabolism of fructose at the uterine-placental interface. The presence of fructose in conceptuses is unique to species with epitheliochorial or synepitheliochorial placentae, indicating that fructose may be important for growth and development of fetuses supported by these types of placentation [1]. Fructose plays a minor role as an energy source or substrate for the pentose shunt pathway in the placenta and fetus because fructose is oxidized to CO₂ at only ∼20% the rate of glucose [2]. However, fructose can be used as a substrate in a number of metabolic pathways that could support conceptus development, including biosynthesis of glycosaminoglycans, phospholipids, and nucleic acids [3]. Recent in vitro studies demonstrated that porcine trophectoderm cells can metabolize fructose through the hexosamine biosynthetic pathway to stimulate mechanistic target of rapamycin (mTOR) cell signaling and cell proliferation and for the synthesis of hyaluronic acid, a significant glycosaminoglycan in the placenta [4].

Fructose is detected in uterine flushings of pregnant gilts as early as Day 12 and is present in significantly greater quantities than glucose in allantoic fluid and fetal blood [5–7]. Additionally, fructose is undetectable in blood of pregnant gilts and sows [8]. Studies by White et al. [9] demonstrated that fructose injected intraperitoneally into an in utero fetus cannot cross the placenta or be converted into glucose, but glucose injected in the same manner can cross the placenta and be converted to fructose. Those results suggest the placenta is the site of fructose production, and it is hypothesized that fructose synthesis from glucose is a method for the conceptus to sequester a hexose sugar from the mother [6].

In pigs, glucose can be transported from maternal blood across the endometrial luminal epithelium (LE), conceptus trophectoderm/chorion, and into placental blood by solute carrier transporters SLC2A1, SLC2A3, and SLC2A4 [10]. It is reasonable to hypothesize that fructose is synthesized from glucose via the polyol pathway in the placenta. Glucose and NADPH are first converted to sorbitol and NADP⁺ by aldose reductase (AKR1B1), then sorbitol and NAD⁺ are converted to fructose and NADH, respectively, by sorbitol dehydrogenase (SORD). The intermediate, sorbitol, is found in high concentrations in ovine, bovine, and human placentae, especially during early pregnancy [11–13], and the enzyme that converts sorbitol to fructose, SORD, has been localized to the trophectoderm of bovine conceptuses [14]; however, to our knowledge there is nothing known about the polyol pathway or its components in pigs.

After its synthesis, fructose can be transported across cell membranes by members of the facilitative diffusion solute carrier 2A family of glucose transporters (SLC2A). SLC2A5 is a low-affinity, high-capacity transporter for fructose, but not glucose, whereas SLC2A8 has a high affinity for glucose but can also transport fructose [15, 16]. SLC2A8 is necessary for decidualization of the uterus in mice during implantation of blastocysts, most likely due to its ability to transport glucose into the differentiating stromal cells, but SLC2A5 has not been detected in human or rodent endometria or placentae [17–19]. Because humans and rodents, unlike pigs, do not have fructogenic placentae, differences would be expected in expression of these particular transporters.

Once fructose enters into a cell, it is phosphorylated by either ketohexokinase (KHK) to produce the metabolite fructose-1-phosphate or by hexokinase-I or -II to fructose-6-phosphate [20]. Fructose-1-phosphate can either be catalyzed into glyceraldehyde-3-phosphate and dihydroxyacetone phosphate for entry into the glycolytic pathway or it can act as a signaling molecule to induce NFkB activation and subsequent production of cytokines [20, 21]. In contrast, fructose-6-phosphate may be metabolized via various pathways such as the hexosamine pathway [4].

Our hypothesis is that specific cell types in the endometrium and chorioallantois utilize the polyol pathway to convert glucose to fructose to support growth and development of the conceptus. The aims of this study were to determine the cellular localization of enzymes required for synthesis of fructose, the transport system for fructose, and the cells of the placenta and endometrium of pigs that use fructose.

Materials and Methods

Tissue Collection

Sexually mature, 8-mo-old crossbred gilts were observed daily for estrus (Day 0) and exhibited at least two estrous cycles of normal duration (18 to 21 days) before being used in these studies. All experimental and surgical procedures were in compliance with the Guide for Care and Use of Agricultural Animals in Teaching and Research and approved by the Institutional Animal Care and Use Committee of Texas A&M University.

To evaluate the effects of pregnancy on expression of mRNAs and cell-specific localization of the AKR1B1, SORD, KHK, SLC2A5, and SLC2A8 proteins and SLC2A5 mRNA, gilts were assigned randomly to either cyclic or pregnant status. Cyclic gilts were ovariohysterectomized on either Day 5, 9, 11, 13, 15, or 17 of the estrous cycle. Gilts in the pregnant group were bred and ovariohysterectomized on either Day 9, 11, 13, 15, 17, 20, 30, 40, 50, 60, or 85 of pregnancy (n = 3 or 4 gilts/day/status). The lumens of both uterine horns from each gilt were flushed separately with 20 ml physiological saline for those collected on Days 5 to 17 postestrus. Pregnancy was confirmed in mated gilts by the presence of morphologically normal conceptuses in uterine flushings. The uterine flushings were clarified by centrifugation at 3000 × g for 15 min and stored at −80°C until analyzed for total recoverable fructose. Tissue sections (∼1 cm thick) from the middle of each uterine horn of gilts through Day 17 and from implantation sites beginning Day 20 were fixed in fresh 4% paraformaldehyde in PBS (pH 7.2) and embedded in Paraplast-Plus (Oxford Laboratory, St. Louis, MO). Additionally, endometrium was physically dissected from the myometrium, snap-frozen in liquid nitrogen, and stored at −80°C for RNA extraction. Chorioallantoic tissue was physically dissected from endometrium and frozen in a similar manner.

In Vivo Estrogen and Progesterone Models

To evaluate the effects of estrogen and estrogen-induced pseudopregnancy on endometrial expression of AKR1B1, SORD, KHK, SLC2A5, and SLC2A8 mRNAs and cell-specific localization of SLC2A8 proteins and SLC2A5 mRNA, gilts were detected in Day 0 and assigned randomly to receive daily intramuscular injections of estradiol benzoate (EB) (5 mg in corn oil; n = 4) or corn oil alone (control; n = 4) on Days 11, 12, 13, and 14 of the estrous cycle [22]. All gilts were ovariohysterectomized on Day 15 of pseudopregnancy. Tissues were collected as described previously.

To evaluate the effects of long-term progesterone (P4) treatment in the absence of other ovarian hormones on endometrial expression of AKR1B1, SORD, KHK, SLC2A5, and SLC2A8 mRNAs and cell-specific localization of SLC2A8 proteins and SLC2A5 mRNA, gilts were ovariectomized on Day 12 of the estrous cycle and assigned randomly to receive daily intramuscular injections of either corn oil (4 ml) or P4 (200 mg P4 in 4 ml corn oil) on Days 12 through 39 postestrus (n = 3/treatment) [23]. All gilts were hysterectomized on Day 40 postestrus and tissues collected as previously described.

To evaluate the effects of estrogen, P4, and their interaction independent of ovarian hormones on endometrial expression of SLC2A8 mRNA and cell-specific localization of SLC2A8 protein, gilts were ovariectomized on Day 4 of the estrous cycle and assigned randomly to be treated daily from Day 4 through 12 as follows: 1) 200 mg P4 in corn oil (P4); 2) 5 mg EB in corn oil (EB); 3) 200 mg P4 plus 5 mg EB in corn oil (P4+EB); or 4) corn oil alone (control) [24]. All gilts were hysterectomized on Day 12, and tissues were collected as previously described.

Total Recoverable Fructose

Uterine flushings from Days 11 and 15 of the estrous cycle and pregnancy were analyzed for total recoverable fructose. The High Sensitivity Fructose Assay Kit (Sigma-Aldrich, St. Louis, MO) was used to determine concentrations of fructose in uterine flushings according to the manufacturer's instructions. Briefly, samples were deproteinized using a 10-kDa molecular weight cut-off spin filter at 15 000 × g for 5 min. Sample Clean-up Mix was added to each sample to remove glucose. To account for high background levels, a sample blank without the conversion enzyme was run for each sample and subtracted from the sample readings during calculations. Total recoverable fructose was calculated by multiplying the concentration of fructose (determined with the kit) by total recoverable volume of uterine flushings (recorded at the time of sample collection).

RNA Extraction and cDNA Synthesis

Total RNA was extracted from endometrial and chorioallantoic tissue samples using Trizol reagent (Life Technologies, Carlsbad, CA) according to the manufacturer's recommendations. First-strand cDNA was synthesized using the Superscript III First Strand Kit (Life Technologies) according to the manufacturer's instructions. First-strand cDNA was diluted 10-fold before use in the quantitative PCR (qPCR) reaction.

Primers for qPCR and in situ hybridization were designed using National Center for Biotechnology Information Genbank sequences and Primer-BLAST (http://www.ncbi.nlm.nih.gov/). Primers were submitted to BLAST to test for specificity against the known porcine genome. Primer information can be found in Supplemental Table S1 (all Supplemental Data are available online at www.biolreprod.org).

Quantitative PCR

The qPCR assays were performed using PerfeCta SYBR Green Mastermix (Quanta Biosciences, Gaithersburg, MD) in 10 μl reactions with 2.5 mM of each specific primer, on a Roche 480 Lightcycler (Roche, Basel, Switzerland) with approximately 60 ng of cDNA per reaction. The PCR program began with 5 min at 95°C followed by 40 cycles of 95°C denaturation for 10 sec and 60°C annealing/extension for 30 sec. A melt curve was produced with every run to verify a single gene-specific peak. Standard curves with 2-fold serial dilutions were run to determine primer efficiencies. All primer correlation coefficients were greater than 0.95 and efficiencies were 95%–102%. The geometric mean of TATA-binding protein (TBP), succinate dehydrogenase complex subunit A flavoprotein (SDHA), and beta actin (ACTB) was used to normalize data from endometrial tissue, while the geometric mean of TBP, hypoxanthine phosphoribosyl transferase 1 (HPRT1), and tubulin alpha 1B (TUBA1B) was used to normalize data from chorioallantoic tissue. The 2^−ΔΔCt method was utilized to normalize data and those fold changes were subjected to statistical analyses.

In Situ Hybridization

We were unable to obtain an antibody to successfully localize SLC2A5 protein in pig tissues, therefore cell-specific expression of SLC2A5 mRNA in the porcine endometrium and placenta was determined by radioactive in situ hybridization analysis as described previously [25]. Briefly, a partial cDNA for porcine SLC2A5 mRNA was cloned from total RNA from Day 15 pregnant porcine endometrial tissue using specific primers (Supplemental Table S1) and PCR amplification was conducted as follows: 1) 95°C for 5 min; 2) 95°C for 45 sec, 60°C for 30 sec, and 72°C for 1 min for 35 cycles; and 3) 72°C for 10 min. The partial cDNA of the correct predicted size was cloned into a pCRII plasmid using a T/A Cloning Kit (Life Technologies) and the sequence verified using an ABI PRISM Dye Terminator Cycle Sequencing Kit and ABI 3130xI Genetic Analyzer (Applied Biosystems, Foster City, CA).

Radiolabeled antisense or sense complementary RNA probes were generated by in vitro transcription using linearized plasmid templates, RNA polymerases, and [α-³⁵S]-UTP. Deparaffinized, rehydrated, and deproteinated uterine tissue sections (5 μm thick) were hybridized with radiolabeled antisense or sense complementary RNA probes. After hybridization, washing, and ribonuclease A digestion, slides were dipped in NTB-2 liquid photographic emulsion (Kodak, Rochester, NY) and exposed at 4°C for 10 days. Slides were developed in Kodak D-19 developer, counterstained with Harris modified hematoxylin (Fisher Scientific, Fairlawn, NJ), and coverslips affixed with Permount (Fisher Scientific). Slides were evaluated using an Axioplan 2 microscope (Carl Zeiss, Thornwood, NY) interfaced with an Axioplan HR digital camera and Axiovision 4.3 software. Digital images of representative fields were recorded under brightfield or darkfield illumination. Photographic plates were assembled using Adobe Photoshop (version 6.0, Adobe Systems Inc., San Jose, CA).

Immunohistochemistry

AKR1B1, SORD, KHK, and SLC2A8 proteins were immunolocalized in porcine uterine and conceptus tissues as described previously [26]. A rabbit anti-human AKR1B1 polyclonal immunoglobulin G (IgG) (GTX113381; GeneTex, Inc., Irvine, CA), a rabbit anti-human SORD polyclonal IgG (sc-366370; Santa Cruz Biotechnology, Inc., Dallas, TX), a rabbit anti-human KHK polyclonal IgG (GTX109591; GeneTex, Inc.), and a rabbit anti-mouse SLC2A8 IgG (GTX12394; GeneTex, Inc.) were used at 8.9, 5, 8.7, and 5 μg/ml, respectively. Antigen retrieval was performed using a boiling citrate buffer for anti-SORD and anti-SLC2A8, protease for anti-AKR1B1, and no retrieval for anti-KHK. Additionally, a rabbit anti-pig interferon gamma (IFNG) polyclonal IgG (I7662-16N8; United States Biologicals, Salem, MA) was used at 5 μg/ml with the boiling citrate buffer antigen retrieval for identification of trophectoderm tissue. Purified nonrelevant rabbit IgG, at the same concentration as the corresponding primary IgG, was used as the negative control. Immunoreactive proteins were visualized in paraffin-embedded sections (5 μm thick) using the Vectastain ABC Kit (PK-4001; Vector Laboratories, Inc., Burlingame, CA) according to the kit's instructions with 3,3′-diaminobenzidine tetrahydrochloride (D5637; Sigma-Aldrich) as the color substrate. Sections were counterstained with Harris-modified hematoxylin (Fisher Scientific), and coverslips were affixed using Permount mounting medium (Fisher Scientific). Digital images of representative fields of immune-stained tissue were recorded under bright-field illumination, and photographic plates were assembled as described for in situ hybridization.

Statistical Analyses

The qPCR data for normal tissues were analyzed for effects of day, pregnancy status, and their interaction for Days 9, 11, 13, 15, and 17 by 2-way ANOVA via the general linear models procedures of the Statistical Analysis System (SAS Institute, Cary, NC) with the least significant difference multiple testing method. The qPCR data were also subjected to a sliding time window linear regression analysis to determine if there was an effect of day of gestation on the gene expression for the genes of interest throughout pregnancy (Days 9–85). The sliding time windows of the analysis were designed to detect effect of day on gene expression during three biologically relevant periods of time: the beginning, the middle, and the end of the study period. Each window analyzed contained a minimum of 3 days. Some windows contained greater than 3 days as long as the regression model showed a nondecreased performance in terms of the r² and the P -value of the F-statistics. For this analysis, a low r² indicates no change in mRNA levels over time, while a high r² indicates that there is an increase or decrease in mRNA levels. The P-value indicates the robustness of the r².

The data from the hormone therapy studies were analyzed using the Student t-test. Data on total recoverable fructose in uterine flushing were analyzed for effects of day, pregnancy status, and their interaction for Days 11 and 15 by 2-way ANOVA via the general linear models procedures of SAS with the least significant difference multiple testing method. All data are presented as mean ± standard errors of the mean with significance at P < 0.05.

Results

Total Recoverable Fructose Increases in Uterine Flushings from Pregnant Gilts

Total recoverable fructose was significantly higher in uterine fluids from Day 15 pregnant gilts as compared to Day 11 or 15 (0.72 mg) cycling gilts (Table 1). Furthermore, the total recoverable fructose was significantly higher in uterine fluids from Day 15 pregnant gilts as compared to Day 11 pregnant gilts (Table 1), indicating that free-floating conceptuses are exposed to increasing levels of fructose during the peri-implantation period of pregnancy (day × status, P < 0.01).

Table 1

Total fructose in uterine flushings.

Day	Pregnant	Cyclic
Day 11	0.02 ± 0.01 mg	0.02 ± 0.01 mg
Day 15	10.10 ± 1.46 mg	0.72 ± 0.42 mg

SLC2A8 Is Expressed in Uterine LE During the Peri-Implantation Period and by the Placenta for the Majority of Pregnancy

Expression of SLC2A8 mRNA increased in endometria of cyclic and pregnant gilts between Days 5 and 11, and decreased between Days 11 and 20, with no differences between cyclic and pregnant gilts; however, there was an effect of day (Fig. 1A; day, P < 0.01). There was a decrease in expression of SLC2A8 mRNA in endometria between Days 11 and 20 of pregnancy (Fig. 1A; r² = 0.815; P > 0.05), and then expression remained stable through Day 85 (r² = 0.429; P > 0.05). Expression of SLC2A8 mRNA increased in the placenta between Days 30 and 50 and then decreased between Days 50 and 85 of pregnancy (Fig. 1B; r² = 0.909; P < 0.05). Expression of SLC2A8 mRNA was lower in endometria from gilts treated with exogenous EB than controls (Fig. 1C; P < 0.05), but there was no effect of P4 on endometrial expression of SLC2A8 mRNA (Fig. 1D; P > 0.05).

Fig. 1

Expression of SLC2A8 mRNA in endometria and placentae. Expression of SLC2A8 mRNA in endometria (A) increased from Day 5 to 11 during the estrous cycle and pregnancy, then decreased through Day 20 of gestation (P < 0.01). Placental expression of SLC2A8 mRNA (B) increased from Day 30 to 50, then decreased to Day 85 of pregnancy (r² = 0.909; P < 0.05). Daily intramuscular injections of EB on Days 11–13, and 14 of the estrous cycle decreased SLC2A8 mRNA levels compared to controls (C; P < 0.05), while there was no effect of daily intramuscular injections of P4 to ovariectomized gilts on Days 12–39 postestrus on expression of SLC2A8 mRNA (D; P > 0.05). Values are presented as means ± SEM with P < 0.05 considered significant.

SLC2A8 Is a Fructose and Glucose Transporter That Is Highly Expressed in Uterine and Placental Tissues During the Estrous Cycle and Early Pregnancy

SLC2A8 protein was expressed in uterine GE of cyclic (data not shown) and pregnant gilts on Day 9 after onset of estrus (Fig. 2, A and B). SLC2A8 protein was first observed in the uterine LE of pregnant gilts on Day 13 and was maintained though Day 15. Expression of SLC2A8 protein in uterine LE was limited to Day 15 in cyclic gilts (Fig. 2, A and B). SLC2A8 was detected in uterine LE until Day 30, and expression transitioned from uterine LE to chorion, specifically in the tall columnar cells at the top of the uterine folds by Day 50 of gestation. Expression of SLC2A8 protein in uterine GE decreased from Day 20 to 25 but remained detectable at low abundance through Day 85 of pregnancy. Placental areolae and the smooth muscle surrounding the placental blood vessels expressed SLC2A8 on Days 30 through 85 of gestation (Fig. 2, A and B). SLC2A8 protein was detected in conceptuses beginning on Day 11 (Fig. 2C). On Day 15, SLC2A8 was detected in a similar pattern to IFNG, which is localized only to the trophectoderm cells [27]. The embryo at Day 15 also expressed SLC2A8 in what appear to be the somites. SLC2A8 continued to be expressed in the trophectoderm on Day 17 of pregnancy (Fig. 2C).

Fig. 2

Immunohistochemical localization of SLC2A8 in the uterus and conceptus. A, B) SLC2A8 protein was detected by Day 9 in uterine GE in the estrous cycle (data not shown) and pregnancy, while LE expression was first observed on Day 13 of pregnancy and Day 15 of the estrous cycle. SLC2A8 was detected in LE until Day 30 when localization transitioned to the chorionic epithelium. SLC2A8 was observed in placental blood vessels at Day 30. By Day 60, SLC2A8 was detected in the areolae (data not shown), and this expression was maintained through Day 85. C) SLC2A8 protein was detected in the conceptus trophectoderm on Day 11 of pregnancy. At Days 15 and 17, SLC2A8 was codistributed with IFNG, indicating expression by trophectoderm, but not endoderm. Rabbit IgG (in panel B) was used as a negative control. Width of fields is 5.45 mm (2.5×), 890 μm (10×), and 230 μm (40×). Values in parentheses refer to objective lens magnification used to capture images. Legend: LE, luminal epithelium; GE, glandular epithelium; ST, stroma; CE, chorionic epithelium; BV, blood vessel; Tr, trophectoderm.

Because SLC2A8 mRNA expression peaked on Day 11 of the estrous cycle and pregnancy, an additional hormone replacement model was utilized to elucidate regulation of expression of SLC2A8. Ovariectomized gilts were given exogenous EB to mimic estrogen secreted by elongating conceptuses, P4 to mimic P4 secretion by CL, EB+P4 to mimic the normal exposure of the endometrium to estrogen and P4 during early pregnancy, or corn oil as a control. Uteri were then collected on Day 12. Expression of SLC2A8 mRNA in endometria did not differ between control and EB-treated gilts. However, endometrial expression of SLC2A8 mRNA (P < 0.05) increased in response to P4 (Fig. 3A), while P4 in combination with EB decreased (P < 0.05) the effect of P4 to stimulate expression of SLC2A8 mRNA. SLC2A8 protein localized to the apical and basal cell surfaces of uterine LE and GE from gilts treated with P4 or P4+EB (Fig. 3B). The EB-treated gilts appeared to have a marked decrease in overall expression of SLC2A8 in the endometrium compared to that for gilts treated with P4, but there was detectable SLC2A8 protein in the smooth muscle surrounding the blood vessels. The control gilts had no detectable SLC2A8 protein in uterine LE, sporadic protein in uterine GE, and no detectable protein in any other cell type (Fig. 3B). Results from this experiment suggest that P4, either directly or indirectly, up-regulates expression of SLC2A8 in uterine LE and GE, while estrogen may modulate expression of SLC2A8 by P4 and increase expression of SLC2A8 in the tunica media of blood vessels in the uterus.

Fig. 3

Regulation of endometrial expression of SLC2A8 by P4 and EB. Endometrial expression of SLC2A8 mRNA (A) was greatest in endometria of P4-treated gilts, was suppressed in P4+EB-treated gilts as compared to P4 treatment alone, and was lowest in EB and control gilts. Differing letters represent significant differences (P < 0.05). Values are presented as means ± SEM. Immunohistochemical staining of tissues from gilts given daily intramuscular injections of P4, EB, or P4+EB from Day 4 through 12 (B) detected SLC2A8 protein in the uterine LE and GE of P4-treated gilts and endometria from P4+EB-treated gilts had apical and basal cell surface staining in the uterine LE. In EB-treated gilts, immunostaining was observed in the smooth muscle surrounding blood vessels in the endometrium, while in control gilts, immunostaining was barely detectable in uterine GE. Width of fields is 890 μm except for 63× (140 μm). Rabbit IgG was used as a negative control. Legend: LE, luminal epithelium; GE, glandular epithelium; BV, blood vessel.

Expression of SLC2A5 Increases in Uterine LE and GE and Placental Chorionic Epithelium as Pregnancy Progresses

Expression of mRNA for the fructose transporter SLC2A5 increased in the endometria between Days 25 and 85 of pregnancy (Fig. 4A; r² = 0.869; P < 0.01). On Days 11 through 13, expression of SLC2A5 mRNA was higher in the endometria from cyclic compared to pregnant gilts (Fig. 4A; day × status, P < 0.05). Placental expression of SLC2A5 mRNA did not change between Days 30 and 85 of pregnancy (Fig. 4B; P > 0.05). In ovariectomized gilts, EB did not affect expression of SLC2A5 mRNA in endometria (Fig. 4D; P > 0.05), but P4 increased expression of SLC2A5 mRNA compared to control gilts (Fig. 4E; P < 0.01). In situ hybridization analysis localized SLC2A5 mRNA to the uterine LE and GE at low levels during the estrous cycle and early pregnancy, with expression appearing to increase in LE and GE as well as in the placental chorion through Day 85 of pregnancy (Fig. 4C). Treatment with exogenous P4 appeared to increase the expression of SLC2A5 mRNA in uterine LE and GE compared to controls (Fig. 4C).

Fig. 4

Expression of SLC2A5 mRNA in endometria and placentae. Endometrial expression of SLC2A5 mRNA (A) increased between Days 25 and 85 of pregnancy (r² = 0.869; P < 0.01). Expression of SLC2A5 mRNA in placentae (B) did not change between Days 30 and 85 (P > 0.05). Corresponding brightfield and darkfield images from in situ hybridization analysis (C) indicated that SLC2A5 mRNA was lowly expressed in uterine LE and GE during the estrous cycle and early pregnancy. By Day 25 of pregnancy, expression increased in uterine LE and continued to increase through Day 85 of pregnancy in both uterine LE and GE. SLC2A5 mRNA also localized to the chorion from Day 25 through 85. SLC2A5 mRNA localized to the uterine LE and GE in endometria from P4-treated gilts (see Fig. 1 legend for steroid treatment). The sense panels are the negative control. Values are presented as means ± SEM with P < 0.05 considered significant. EB treatment of gilts (see Fig. 1 legend for steroid treatment) had no effect (D; P > 0.05), but P4 treatment increased SLC2A5 mRNA in endometria compared to control gilts (E; P < 0.01). Width of fields is 890 μm. Legend: LE, luminal epithelium; GE, glandular epithelium; ST, stroma; CE, chorionic epithelium.

AKR1B1 Is Expressed in Uterine LE During the Peri-Implantation Period and in Trophectoderm/Chorion for the Majority of Pregnancy

Aldose reductase (AKR1B1) catalyzes the conversion of glucose to sorbitol, which is the first step in the polyol pathway. Endometrial expression of AKR1B1 mRNA increased between Days 9 and 11 and then decreased through Day 17 with higher expression in the endometria from pregnant compared to cyclic gilts (Fig. 5A; day × status, P < 0.05). Endometrial expression of AKR1B1 mRNA increased between Days 20 and 85 of gestation (Fig. 5A; r² = 0.924; P < 0.01). Placental expression of AKR1B1 mRNA was not different between Days 30 and 85 (Fig. 5B; P > 0.05). Treatment of gilts with EB decreased endometrial expression of AKR1B1 mRNA compared to control gilts (Fig. 5C; P < 0.05), but P4 had no effect on AKR1B1 mRNA expression (Fig. 5D; P > 0.05).

Fig. 5

Expression of AKR1B1 mRNA and localization of AKR1B1 protein in endometria and placentae. Endometrial expression of AKR1B1 mRNA (A) increased between Days (D) 9 and Day 11, decreased until Day 15 (day × status, P < 0.05), then increased between Days 25 and 85 of pregnancy (D25P and D85P; r² = 0.924; P < 0.01). Placental expression of AKR1B1 mRNA (B) did not change between Days 30 and 85 (P > 0.05). EB treatment of gilts (see Fig. 1 legend for steroid treatment) decreased AKR1B1 mRNA compared to controls (C; P < 0.05), while P4 treatment (see Fig. 1 legend for steroid treatment) had no effect (D; P > 0.05). AKR1B1 protein (E) localized to uterine LE from Day 13 through 15 of pregnancy and to trophectoderm at Day 15. Expression of AKR1B1 continued in the chorion through Day 85 of pregnancy. Rabbit IgG was used as a negative control. Width of fields is: 890 μm for D5, D20P, D30P, D40P, D60P, D60P IgG; 445 μm for D9, D11, D13, D15; 572 μm for D15 conceptus; and 1208 μm for D85P. Values are presented as means ± SEM with P < 0.05 considered significant. Legend: LE, luminal epithelium; GE, glandular epithelium; ST, stroma; Tr, trophectoderm; CE, chorionic epithelium.

AKR1B1 protein was localized to the uterine LE of pregnant gilts and to placental trophectoderm/chorion but was not detected in the endometria of cyclic gilts (Fig. 5E). AKR1B1 was first observed in uterine LE on Day 13 of pregnancy, was barely detectable in the endometrium by Day 20, and was not detectable in uterine LE between Days 30 and 85 (Fig. 5E). In contrast, conceptus trophectoderm expressed AKR1B1 on Day 15, and expression continued in the chorion through Day 85 of pregnancy. There was low abundance of AKR1B1 protein in the areolae at Day 40 but not at Day 85 of pregnancy (Fig. 5E).

SORD Is Expressed in Uterine LE and GE During the Peri-Implantation Period and Is Expressed in Trophectoderm/Chorion for the Majority of Pregnancy

Sorbitol is converted to fructose by sorbitol dehydrogenase (SORD). Endometrial expression of SORD mRNA peaked on Day 5 of the estrous cycle, then decreased between Days 9 and 17 of the estrous cycle and pregnancy (P < 0.01) with no significant day × status interaction (Fig. 6A; P > 0.05). Expression of SORD mRNA decreased between Days 9 and 13 (r² = 0.858; P > 0.05) of pregnancy and then increased between Days 15 and 30 (Fig. 6A; r² = 0.828; P < 0.05). Placental expression of SORD mRNA increased between Days 30 and 40 (Fig. 6B, r² = 0.999; P < 0.05). There was no effect on expression of SORD mRNA by exogenous treatment of gilts with either EB or P4 (Fig. 6, C and D; P > 0.05). SORD protein was expressed in the uterine GE on Days 5 through 17 of both the estrous cycle and pregnancy (Fig. 6E). Uterine LE began to express SORD on Day 13 of the estrous cycle and pregnancy, and expression appeared to decrease to low levels by Day 20 of gestation. The trophectoderm expressed SORD as early as Day 15, and the chorion, including the areolae, expressed SORD through Day 85 of gestation. SORD appeared to be more highly expressed in the uterine GE between Days 60 and 80 than between Days 30 and 40 of pregnancy (Fig. 6E).

Fig. 6

Expression of SORD mRNA and localization of SORD protein in endometria and placentae. Endometrial expression of SORD mRNA (A) peaked at Day (D) 5, then decreased through Day 15 (P < 0.01). Placental expression of SORD mRNA (B) increased between Days 30 and 40 of gestation (D30P and D40P; r² = 0.999; P < 0.05). Treatment of gilts with EB or P4 (see Fig. 1 legend for steroid treatment) did not affect endometrial expression of SORD mRNA compared to control gilts (C, D; P > 0.05). SORD protein (E) was detected in uterine GE by Day 5 and in uterine LE by Day 13 of both the estrous cycle and pregnancy. Trophectoderm expression of SORD protein was detected by Day 15 and expression continued in trophectoderm/chorion through Day 85. Areolae also expressed SORD protein. Rabbit IgG was used as the negative control. Width of fields is: 890 μm for D5, D20P, D30P, D40P, D60P, D85P, D60P IgG; 445 μm for D9, D13, D15 conceptus, D17; and 286 μm for D15C, D15P, D15 IgG. Values are presented as means ± SEM with P < 0.05 considered significant. Legend: LE, luminal epithelium; GE, glandular epithelium; ST, stroma; Tr, trophectoderm; CE, chorionic epithelium; AE, allantoic epithelium.

KHK Is Primarily Expressed by Trophectoderm/Chorion

Ketohexokinase (KHK) catalyzes the phosphorylation of fructose to produce fructose-1-phosphate. Endometrial expression of KHK mRNA was greater from Day 9 through 17 of pregnancy compared to the corresponding days of the estrous cycle (Fig. 7A; P < 0.01). There was an increase in expression of KHK mRNA through Day 85, with peaks at Days 30 and 85 (Fig. 7A; r² = 0.478; P < 0.05). Placental expression of KHK mRNA increased from Day 30 to 40 of pregnancy (Fig. 7B; r² = 0.786; P > 0.05).Treatment of gilts with either EB or P4 did not affect expression of KHK mRNA (Fig. 7, C and D; P > 0.05). KHK protein was only detectable in GE of the endometrium at Day 30 of pregnancy (Fig. 7E). KHK protein was detected in the trophectoderm as early as Day 11 and in the chorion, including the areolae, at all stages of gestation studied through Day 85. There was stronger immunostaining for KHK in the tall columnar cells at the tops of the uterine-placental folds (Fig. 7E).

Fig. 7

Expression of KHK mRNA and localization of KHK protein in endometria and placentae. Endometrial expression of KHK mRNA (A) was greater in pregnant than cyclic gilts from Day (D) 9 through 17 (day × status, P < 0.01), and the expression increased continually through Day 85 (r² = 0.478; P < 0.05). Placental expression of KHK mRNA (B) did not change from Day 30 to 85 of pregnancy (r² = 0.433, P > 0.05). Treatment of gilts with either EB or P4 (see Fig. 1 legend for steroid treatment) had no significant effect on expression of KHK mRNA compared to controls (C, D; P > 0.05). KHK protein (E) was detected as early as Day 11 in conceptus trophectoderm and trophectoderm/chorion expression continued through Day 85 of pregnancy with areolae also expressing KHK. Endometrial expression was limited to uterine GE at Day 30 of gestation. Rabbit IgG was used as a negative control. Width of fields is 890 μm for all images except D11, D15, D40, which are 445 μm. Values are presented as means ± SEM with P < 0.05 considered significant. Legend: LE, luminal epithelium; GE, glandular epithelium; Tr, trophectoderm; CE, chorionic epithelium.

Discussion

There is an unusual abundance of fructose in fetal fluids and uterine flushings of pigs [6]. Fructose has long been known to be synthesized in the placenta of ungulates and cetaceans during pregnancy, but the pathway by which fructose is synthesized, the transport mechanisms involved, and the cells that may use fructose have not been described previously [4, 6]. Indeed, fructose has, for the most part, been ignored in research regarding ungulate species, possibly because fructose is not metabolized through the glycolytic pathway in the fetus or placenta [28–30]. Results from this study indicate that fructose increases in uterine flushings of pregnant gilts from Day 11 to 15, that there are two transporters, SLC2A5 and SLC2A8, that can transport fructose from the uterus to the conceptus, and that the components of the polyol pathway, by which fructose can be synthesized from glucose, are present at the uterine-placental interface during pregnancy in pigs (Fig. 8).

The transporters capable of transporting fructose from the uterus to placenta localize to different cell types in a stage of pregnancy-dependent manner. SLC2A8 localizes to uterine LE and GE during the peri-implantation period of pregnancy (Fig. 8A) and to the chorion at Day 30 (Fig. 8B). In contrast, SLC2A5 mRNA increases significantly by Day 30 and remains high through Day 85 of pregnancy in uterine LE and chorionic epithelium at the uterine-placental interface (Fig. 4). Results also indicate an interesting stage of pregnancy-dependent change in the cellular localization of the enzymes involved in the polyol pathway. From Day 13 through 17, AKR1B1 and SORD localize to the uterine LE (Fig. 8A); however, by Day 20, AKR1B1 and SORD were undetectable in uterine LE but abundant in the chorion through Day 85 of pregnancy (Fig. 8B). This intriguing shift in expression of the polyol enzymes from uterine LE to chorion during pregnancy suggests that the free-floating, elongating conceptuses are supported by fructose synthesized by the uterus. However, after implantation is initiated and placentation is established at the maternal/conceptus interface, the placenta becomes self-sufficient for synthesis and transport of fructose.

Pig embryos enter the uterus 60 to 72 h after estrus, develop into blastocysts by Day 5, and 0.5–1mm diameter blastocysts shed the zona pellucida between Days 6 and 7. Blastocysts initially transition from spherical (2 to 6 mm diameter) on Day 10, then to tubular and elongated filamentous forms between Days 10 and 12 of pregnancy. The presumptive placental membranes (trophectoderm and endoderm) then elongate 30–45 mm/h from 10 mm spherical blastocysts to 150–200 mm long filamentous conceptuses, after which further elongation occurs until conceptuses are 800–1000 mm in length by Day 16 of pregnancy. This process provides maximum surface area for contact between trophectoderm and uterine LE to facilitate the uptake of nutrients from uterine LE and GE [31]. Significant energy is required to drive these extreme morphological changes. Glucose is essential to conceptus survival, growth, and development, and substantial amounts of glucose are transported across the placental membranes of pigs and mice [32]. SLC2A8 is an outstanding candidate for fructose and glucose transport during these important events during the peri-implantation period because of its localization to both uterine LE and trophectoderm while the conceptus is free-floating.

SLC2A8 protein localizes to the apical and basal cell surfaces of hepatocytes and uterine epithelia as well as the endoplasmic reticulum and lysosomes in spermatocytes [15, 17, 33]. SLC2A8 can transport glucose and fructose into and out of cells when it is located along the apical or basal plasma membranes, but when SLC2A8 is located in the endoplasmic reticulum, it is hypothesized to be for transporting glucose into the endoplasmic reticulum for glycosylation of new proteins, especially in cells with high secretory activity. In this study, SLC2A8 localized to the apical and basal surfaces of the plasma membrane in uterine LE, GE, and Day 11 trophectoderm, but it also localized intracellularly in Day 15 trophectoderm and in the embryo. The change of localization within the trophectoderm is interesting because the Day 15 trophectoderm is more active in secretion of molecules than the Day 11 trophectoderm, and SLC2A8 may facilitate increased synthesis of proteins for secretion.

The steroid hormones, P4, secreted by CL, and estrogen, secreted by the trophectoderm of conceptuses, regulate expression of multiple genes and proteins at the uterine-placental interface in mice, sheep, and pigs, including those involved in nutrient transport and metabolism [17, 34, 35]. Expression of SLC2A8 protein is suppressed in uterine LE of ovariectomized mice by estrogen, but expression in LE is stimulated by P4 [17]. Those results are similar to our current observation that in ovariectomized pigs, EB alone, or when administered with P4, suppresses endometrial expression of SLC2A8 mRNA compared to treatment with P4 alone, although the cellular localization of SLC2A8 protein to uterine LE and GE is not affected by hormone treatments. Although treatment of gilts with EB alone did not change steady-state levels of endometrial SLC2A8 mRNA compared to control gilts, it did induce expression of SLC2A8 protein in the smooth muscle tunica media of endometrial blood vessels. Interestingly, blood vessels did not express SLC2A8 when EB and P4 were administered together to ovariectomized gilts. Furthermore, SLC2A8 protein was prominent in the smooth muscle and endothelium of placental blood vessels between Days 30 and 85 of pregnancy. Because concentrations of estrogen in both placental blood and allantoic fluid increase as pregnancy progresses, estrogen and the relationship between changing concentrations of estrogen and P4 may be responsible for the expression of SLC2A8 mRNA and protein in placental blood vessels [36].

In this study, SORD is highly expressed in the uterine GE on Day 5 of the estrous cycle, but there is no corresponding expression of AKR1B1, which would indicate that there is no potential for activation of the polyol pathway. But SORD may have a function other than converting sorbitol to fructose, and fructose to sorbitol. Another possibility is that there is a different aldose reductase family member that localizes to the GE to catalyze the first step in the polyol pathway. Indeed, there are more than 190 known enzymes in the aldo/keto reductase family, each with their own specificities for different substrates, and those specifically in the AKR1B family of NADPH-dependent oxidoreductases, other than AKR1B1, may be able to convert glucose to sorbitol [37, 38]. In this study, we analyzed AKR1B1 due to a high affinity for glucose and its activity in the polyol pathway in other tissues [39, 40]. We detected AKR1B1 protein in uterine LE from Day 13 through 17 of gestation (Fig. 5) but did not analyze uteri from Day 12 of pregnancy. It has been reported that initial expression of AKR1B1 in LE is on Day 12 of pregnancy [41], and our present results do not conflict with that finding.

The pig develops a diffuse, epitheliochorial-type placenta to support the growth of each individual fetus. This placentation is characterized as superficial, noninvasive, and having a uterine LE that remains intact. By Day 24 of pregnancy, there is complete attachment between the apical surfaces of the uterine LE and the conceptus trophectoderm, and these adhered epithelia begin to fold around Day 30 to increase surface area for nutrient transport [42]. Interestingly, results of this study showed that multiple proteins, SLC2A8, AKR1B1, and SORD, are expressed in the uterine LE during the peri-implantation period, but expression by uterine LE begins to down-regulate as conceptus attachment to the uterine wall is completed and folding of the uterine-placental interface commences. The placental chorion then assumes expression of SLC2A8, AKR1B, and SORD proteins (see Figs. 2A, 5E, and 6E for examples of the temporal/spatial shift in their expression). Additionally, mRNA for the fructose transporter, SLC2A5, becomes highly expressed in uterine LE and chorion during the same period of gestation. As the placenta develops, there is a transition from uterine LE support of the conceptus to placental chorion support of the conceptus that is exemplified by changes in cell-specific expression of the proteins involved in synthesis and transport of fructose.

Expression of SLC2A5 mRNA was induced in the uterine LE and GE of ovariectomized gilts by P4, and SLC2A5 mRNA increased in uterine LE, GE, and placental chorion between Days 25 and 30 of gestation. Therefore, we hypothesize that P4 increases expression of SLC2A5 mRNA on Day 30 of pregnancy, and this transporter remains a prominent feature of the uterine-placental interface of pigs through Day 85. However, we previously reported that the receptor for P4, PGR, is not present in the uterine LE and GE, but is expressed by the stroma of these P4-treated gilts [23]. Therefore, similar to other genes, it is reasonable to hypothesize that increased expression of SLC2A5 mRNA by uterine LE and GE is potentially in response to down-regulation of PGR by P4 [43]. Alternatively, increased expression of SLC2A5 mRNA in uterine LE and GE is potentially due to paracrine-acting growth factors (progestamedins) produced by the PGR-positive stromal cells in response to P4 [44]. Furthermore, the presence of membrane receptors for P4 has not been conclusively ruled out. The regulation of gene expression by P4 in uterine epithelial cells that do not express PGR remains a fertile area for mechanistic study.

A structurally and functionally unique characteristic of the epitheliochorial placenta is the development of chorionic areolae. Areolae are invaginations of the chorioallantois that form as crypts of columnar epithelium over the mouths of each uterine gland. Areolae are made up of folded chorion containing tall columnar cells with long microvilli that are optimal for transporting histotroph secreted from uterine GE [45, 46]. In addition to transporting histotroph, these cells may also play a role in synthesis of nutrients from precursor molecules in histotroph or from the placental vasculature. SLC2A5 mRNA and SLC2A8 protein were detected in the epithelia of areolae and likely transport glucose and fructose to the placental vasculature. Additionally, AKR1B1, SORD, and KHK proteins localized to areolae on Day 40, indicating that areolae are potential sites for synthesis and phosphorylation of fructose. The tall columnar cells at the tops of uterine folds in the interareolar regions of the chorioallantois are also hypothesized to be highly active in transporting molecules as compared to the more cuboidal cells at the bottom of the uterine folds [47]. The localization of SLC2A8 protein to the tall columnar cells supports this hypothesis. However, the localization of KHK to those cells suggests that it phosphorylates fructose to sequester it for potential use in the areolar epithelium that raises questions regarding the complex roles that fructose may have at the uterine-placental interface.

The polyol pathway is up-regulated by hyperglycemia and, therefore, has been studied intensively in diabetes and cancer where activation of the pathway can result in 1) accumulation of the intermediate sorbitol, which functions as an osmoregulator; 2) reduced ability of cells to modulate oxidative stress due to consumption of NADPH; and 3) production of fructose. Organ systems where the polyol pathway is normally active and functional include the liver, kidney, and male reproductive tract where fructose is produced in abundance by the seminal vesicles and the epididymis [39, 48]. While the sow is not hyperglycemic during pregnancy, glucose accumulates at the uterine-placental interface due to the presence of both sodium-dependent glucose transporters [10] and an appreciably lower concentration of glucose than that in the placental blood. This accumulation of glucose could activate the polyol pathway. Conversion of glucose to fructose would then not only sequester fructose in the conceptus because fructose cannot be transported back to the maternal blood, but it would maintain the disparity in concentrations of glucose between maternal and fetal-placental blood to enhance the gradient for transport of glucose from maternal blood to the conceptus.

Shibutani et al. [49] reported that KHK mRNA is present in parthenogenetic pig embryos as early as the 5- to 8-cell stage and that expression continues through the blastocyst stage. Those embryos are also unable to utilize fructose as the only carbohydrate source before the 5- to 8- cell stage of development. Results of the present study indicate that the trophectoderm of pig conceptuses express KHK protein on Day 11 and then the chorion expresses KHK through Day 85 of gestation (Fig. 7E).

In the liver, fructose is used for glycogen and triglyceride synthesis, neither of which is synthesized in substantial amounts by the pig placenta [20, 50]. Also, sperm use fructose as an energy source, but this too is minimal in the placenta [51]. While recent in vitro data using porcine trophectoderm cells indicate that fructose activates the hexosamine biosynthetic pathway to stimulate mechanistic target of rapamycin (mTOR) cell signaling and cell proliferation, and the synthesis of hyaluronic acid, a significant glycosaminoglycan in the placenta, there are still questions about the role of fructose in the placenta [4]. Results of the present research indicate that the trophectoderm has the potential to utilize fructose due to the presence of KHK in those cells; however, the downstream molecules or regulatory targets remain to be determined.

In conclusion, components necessary for fructose synthesis via the polyol pathway and fructose transport to the fetal/placental vasculature are present at the uterine-placental interface throughout pregnancy in pigs (Fig. 8). Additionally, those components are expressed in a temporal and spatial manner that is, at least in part, regulated by the steroid hormones P4 and estrogen. Early in development, the free-floating conceptus can be supported by fructose synthesized by the uterine LE and transported into the uterine lumen. As pregnancy and placentation progress, the chorion takes over the synthesis of fructose, and the conceptus becomes self-sufficient for fructose production. Additionally, the localization of SLC2A8 to uterine LE and trophectoderm during the peri-implantation period provides a mechanism for transport of fructose between the tissues synthesizing fructose and the tissues metabolizing fructose, particularly the conceptus trophectoderm and chorioallantois.

Fig. 8

Current working model for fructose synthesis and transport at the uterine-placental interface of pigs. Fructose levels are substantially higher than glucose levels in the fetal blood, amniotic fluid, and allantoic fluid of pigs [5]. During the peri-implantation period of pregnancy (A), glucose is transported into uterine LE from the maternal vasculature by SLC2A8. The polyol pathway converts glucose to fructose in uterine LE, then fructose and glucose are transported into the uterine lumen by SLC2A8. Fructose can be utilized in the trophectoderm after being transported in by SLC2A8 or be further transported to other extraembryonic tissues or the embryo proper, as is glucose, by SLC2A8. As placentation progresses, during midpregnancy (B), glucose is transported into and through uterine LE by an unknown transporter and the polyol pathway enzymes do not localize to that cell type. Rather, glucose is transported into the chorion by SLC2A8 where AKR1B1 and SORD are present to mediate conversion to fructose. Fructose can either be used by the chorion or be transported by SLC2A5 and SLC2A8 to the placental vasculature during mid- to late pregnancy.

Author notes

¹

This project was supported by National Research Initiative Competitive Grant no. 2006-35203-17199 from the USDA National Institute of Food and Agriculture to G.A.J., by a Graduate Diversity Fellowship provided to C.B.S. by the Office of Graduate Studies at Texas A&M University, and by a Texas A&M College of Veterinary Medicine Graduate Research Mini-Grant Award to C.B.S. Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture.

³

The U.S. Department of Agriculture (USDA) prohibits discrimination in all its programs and activities on the basis of race, color, national origin, age, disability, and where applicable, sex, marital status, familial status, parental status, religion, sexual orientation, genetic information, political beliefs, reprisal, or because all or part of an individual's income is derived from any public assistance program. (Not all prohibited bases apply to all programs.) Persons with disabilities who require alternative means for communication of program information (Braille, large print, audiotape, etc.) should contact USDA's TARGET Center at [phone] (voice and TDD). To file a complaint of discrimination, write to USDA, Director, Office of Civil Rights, 1400 Independence Avenue, S.W., Washington, D.C. 20250-9410, or call [phone] (voice) or [phone] (TDD). USDA is an equal opportunity provider and employer.