GLUT1-mediated glycolysis supports GnRH-induced secretion of luteinizing hormone from female gonadotropes

The mechanisms mediating suppression of reproduction in response to decreased nutrient availability remain undefined, with studies suggesting regulation occurs within the hypothalamus, pituitary, or gonads. By manipulating glucose utilization and GLUT1 expression in a pituitary gonadotrope cell model and in primary gonadotropes, we show GLUT1-dependent stimulation of glycolysis, but not mitochondrial respiration, by the reproductive neuropeptide GnRH. GnRH stimulation increases gonadotrope GLUT1 expression and translocation to the extracellular membrane. Maximal secretion of the gonadotropin Luteinizing Hormone is supported by GLUT1 expression and activity, and GnRH-induced glycolysis is recapitulated in primary gonadotropes. GLUT1 expression increases in vivo during the GnRH-induced ovulatory LH surge and correlates with GnRHR. We conclude that the gonadotropes of the anterior pituitary sense glucose availability and integrate this status with input from the hypothalamus via GnRH receptor signaling to regulate reproductive hormone synthesis and secretion.


Results
GnRH regulates GLUT1 in gonadotropes. There is evidence that a global metabolic response in gonadotropes is associated with GnRH stimulation and LH secretion. mRNA-seq was performed on sorted pituitaries from female mice in proestrus (the cycle stage in which the LH surge occurs) and diestrus (cycle stage with generally low LH) 19 . Our independent secondary analysis of those data revealed that genes related to cellular catabolism, and therefore generation of energy, were generally increased during proestrus in comparison to diestrus ( Supplementary Fig. S1). These data demonstrate that in vivo physiological changes in LH secretion are likely tied to gonadotrope cellular metabolism and are responsive to changes in upstream GnRH secretion which regulates the LH surge 20 . mRNA-seq analysis of GnRH-treated LβT2 cells 21,22 , a mature C57BL/6 mouse female gonadotrope cell line 23 , indeed demonstrates that GnRH regulates genes associated with gonadotrope cellular metabolism ( Supplementary Fig. S1). LβT2 cells are an excellent model for deciphering mechanisms of GnRH action that can be subsequently validated in vivo, including regulation of LH and FSH secretion by GnRH pulse frequency and amplitude 19,[24][25][26][27] . The gene ontology analysis of mRNA-seq data from LβT2 cells indicating metabolism as the most enriched biological pathway in gonadotropes in response to GnRH corroborates the in vivo observation that metabolic genes are upregulated in gonadotropes during proestrus ( Supplementary  Fig. S1). These findings provide a strong rationale to assess the relationship of cellular metabolism to GnRHinduced secretion of LH from gonadotropes.
GnRH is secreted from hypothalamic neurons in a pulsatile manner, and GnRH pulse frequency and amplitude specifically regulate the downstream gonadotrope response. High frequency GnRH pulses favor LH production while low frequency GnRH pulses favor FSH production 25 . Similar to LH surge-associated genes, we hypothesized that increasing GnRH pulse frequency would increase Slc2a1 mRNA. To test this hypothesis, we analyzed Slc2a1, Slc2a3, Slc2a4, and Slc2a8 mRNA expression data extracted from an mRNA array data set (GSE63251) of LβT2 cells pulsed with an amplitude of 10 or 100 nM GnRH at increasing frequencies 25 . These data showed that Slc2a1 mRNA levels increase with frequency while having no impact on mRNA of other Slc2a family members (Table 1). To confirm that this observation is statistically significant in LβT2 cells, we pulsed these cells either once or twice per hour for 4 h with an amplitude of 10 nM GnRH. We found increased expression of Slc2a1 mRNA in response to increased GnRH frequency (Fig. 1A). However, in contrast to chronic (non-pulsatile) 8 h GnRH treatment 14 , protein levels of GLUT1 were only mildly increased when pulsed once per hour, but not twice per hour (Fig. 1B,C, Supplementary Fig. S2). Because GLUT1 protein levels were only slightly impacted, we surmised that GnRH may be regulating GLUT1 protein translocation to the membrane in addition to gene transcription and translation. To test this possibility, we fractionated LβT2 cells after 30 min of 10 nM GnRH treatment and assessed GLUT1 protein by western blot. We found that GLUT1 in the membrane fraction of LβT2 cells was significantly increased after GnRH treatment (Fig. 1D,E, Supplementary Fig. S2). These data demonstrate that both GLUT1 expression and translocation are increased by physiological GnRH pulse stimulation.
Gonadotropes induce glycolysis in response to GnRH. The function of GLUT1 as a facilitative glucose transporter indicates that GnRH-induced GLUT1 at the cell surface will increase glucose uptake and therefore glycolysis. Using the fluorescent glucose analog 2-NBDG, we confirmed that a 30 min GnRH treatment induces a 1.4-fold increase in glucose uptake in LβT2 cells, similar to chronic GnRH treatment 14 (Fig. 1F). We further tested the impact of GnRH on glycolysis with a glycolytic stress test via extracellular flux analysis which measures the extracellular acidification rate (ECAR, a proxy for acid production from all sources including lactate from glycolysis) in real time. We found that GnRH significantly increases both glycolysis and the maximum Scientific RepoRtS | (2020) 10:13063 | https://doi.org/10.1038/s41598-020-69913-z www.nature.com/scientificreports/ capacity of these gonadotropes to engage glycolysis (Fig. 1G,H). Because ECAR is only an indirect measure of glycolysis, we independently confirmed this result by measuring lactate (a product of anaerobic glycolysis) and found that GnRH increases lactate production by LβT2 cells (Fig. 1I). Together, these data demonstrate that gonadotropes rapidly respond to GnRH by increasing glycolysis.

Glucose metabolism in gonadotropes is mediated by GLUT1. Because GnRH induces GLUT1
translocation and glycolysis in gonadotropes ( Fig. 1), we tested whether GLUT1 mediates GnRH-induced glycolysis. To examine glycolysis and to assess the potential contribution of mitochondria to GnRH-induced metabolic responses, we used extracellular flux (XF) analysis from a mitochondrial stress test to quantify oxygen consumption rate (OCR), a measure of mitochondrial respiration, and ECAR of LβT2 cells. Surprisingly, OCR remained the same in response to GnRH treatment, indicating that mitochondria do not energetically contribute to the GnRH response ( Fig. 2A,B). In contrast, LβT2 ECAR increased and the OCR:ECAR ratio decreased in response to GnRH (Fig. 2C-E), suggesting that gonadotropes elevate glycolysis to support GnRH-induced activity. To determine whether GnRH-induced glycolysis is dependent on GLUT1, we performed a mitochondrial stress test on LβT2 cells with or without pharmacological inhibitors of GLUT1 28,29 . Both GLUT1 inhibitors WZB117 (IC 50 = 10 μM) and BAY-876 (IC 50 = 2 nM, selectivity factor > 100 against GLUT2, GLUT3, and GLUT4) significantly inhibited basal and maximal OCR and the spare respiratory capacity of LβT2 cells ( Fig. 2A,B). Interestingly, only WZB117 reduced OCR and ATP production under GnRH-treated conditions ( Fig. 2A,B). This may be due to the differential impact of these GLUT1 inhibitors on glycolysis in LβT2 cells. WZB117 inhibits GnRH-induced ECAR by ~ 40%, while BAY-876 inhibits basal and GnRH-induced ECAR by ~ 60% and ~ 80% respectively (Fig. 2C,D). Despite the near complete ablation of ECAR by Bay-876, LβT2 cells are able to maintain the majority of their OCR, indicating that GLUT1-mediated glycolysis contributes to, but is not necessary for mitochondrial respiration in these cells. Importantly, both GLUT1 inhibitors prevented a GnRH-induced switch to glycolysis as evidenced by the absence of decreased OCR:ECAR as seen in control LβT2 cells (Fig. 2E). From these data, we conclude that GLUT1 supports GnRH-induced glycolysis in gonadotropes. Our observations demonstrate that a one-time physiological dose of GnRH induces glycolysis in LβT2 cells. In vivo, GnRH is pulsatile and elicits pulsatile secretion of LH 20 . A hallmark of gonadotropes is their ability to resolve signaling input from a GnRH pulse and reset for subsequent stimulation by a following pulse of GnRH 30,31 . Therefore, we examined whether GnRH-induced glycolysis is pulsatile with GnRH. We measured lactate and LH secretion from LβT2 cells in response to hourly GnRH pulses with an amplitude of 10 nM. Lactate peaks had a mean time difference of 3.75 min and a median difference of 2.5 min after LH peaks (Fig. 2F,G). Peak differences were not statistically different as determined by nonparametric testing and were within the 5-min fraction sampling interval (Fig. 2G). The transient increases in lactate associated with LH pulses suggest a temporary activation of glycolysis induced by GnRH. Additionally, the concentration of lactate positively correlated with mean secreted LH (Fig. 2H). These data implicate glycolysis in the regulation of LH secretion.
Glucose supports maximal secretion of LH. Because GnRH-induces lactate production in gonadotropes that correlates with LH secretion, we expect that glycolysis is the process that supports energy-dependent LH transcription, translation, and secretion 26 . Therefore, we used two approaches to test the impact of glucose on pulsatile LH secretion from LβT2 cells. First, we performed pulse experiments with glucose-free media supplemented with pyruvate. The absence of glucose caused a significant reduction in total GnRH-induced LH secretion as quantified by the area under the curve and a ~ 30% decrease in mean LH amplitude ( Fig. 3A-C). Second, we performed pulse experiments with media containing 2-DG, a glucose analog which competitively www.nature.com/scientificreports/ inhibits the production of glucose-6-phosphate from glucose and depletes cellular ATP. We observed a significant inhibition of GnRH induced-LH secretion across all pulses and a ~ 50% reduction in mean amplitude ( Fig. 3D-F). To confirm that glucose metabolism in glucose-free conditions was reduced throughout the 4-h pulse experiment, we assessed expression of a known glucose responsive gene, thioredoxin interacting protein (Txnip) 32,33 , in cells at the completion of the experiment. As expected, LβT2 cells in glucose-free media have significantly reduced Txnip levels compared to control (Fig. 3G). Conversely, cells treated with 2-DG had a threefold increase in Txnip, reflecting the high presence of glucose available in the cells that cannot be metabolized due to the inhibition of glycolysis (Fig. 3G) 32,34,35 . To determine whether the reduction of LH secretion was due to reduced secretion or reduced biosynthesis, we determined LH content of LβT2 cells post pulse. We found that both glucose-free and 2-DG conditions resulted in increased LH stores (Fig. 3H). These data suggest that glucose supports maximal LH secretion. However, it remains unclear whether it is glucose availability or the level of glucose uptake that is important for regulation of LH synthesis and secretion. GnRH were fractionated into cellular components and analyzed by western blot. Each experiment was run on one gel, the membrane cut, then probed simultaneously for all targets. GLUT1 in the membrane fractions was quantified by densitometry and normalized to Na + /K + ATPase, then expressed as fold increase over control for each experiment (N = 5 To resolve the contribution of glucose availability versus glucose uptake, we leveraged the intrinsic heterogeneity of the LβT2 cell line 36 . Based on data showing that LH pulses are correlated with glycolytic bursts (Fig. 2F,G), we hypothesized that cells that take up more glucose will secrete increased amounts of LH. To test  www.nature.com/scientificreports/ this hypothesis, we labeled LβT2 cells with 2-NBDG, a fluorescent glucose analog, and FAC sorted the cell line based on low glucose uptake (Lo, the 1st quartile) or high glucose uptake (Hi, the 4th quartile) (Fig. 3I). These cells were placed back into independent culture and treated ± GnRH for 30 min (Fig. 3J), after which secreted LH was measured (Fig. 3K). Both low and high glucose utilizers responded to GnRH with a significant increase in LH secretion (Fig. 3K). Impressively, high glucose utilizers secreted more basal LH than GnRH-stimulated low glucose utilizers. Coupled with the demonstration that high glucose utilizers had approximately twice as much surface GLUT1 as low glucose utilizers (Fig. 3L), this finding indicates that GLUT1 activity facilitates gonadotrope demand for glucose uptake during GnRH-induced LH secretion.
Glucose support of LH secretion is facilitated by GLUT1. To test the impact of GLUT1-mediated glucose uptake on LH secretion, we inhibited the transporter with a pharmacological and molecular approach. First, we assessed pulsatile LH secretion from LβT2 cells in the presence of the GLUT1 inhibitor WZB117. Both total GnRH-induced LH secretion and the mean amplitude were significantly reduced in the presence of WZB117 ( Fig. 4A-C). Second, we created stably transduced LβT2 cell lines expressing non-targeted shRNA (Ctrl) or shRNA targeted to Slc2a1 (shGlut1). The shGlut1 cell line exhibited ~ 70% reduction in both Slc2a1 mRNA and GLUT1 protein (Fig. 4D, Supplementary Fig. S2). The shGlut1 cell line maintained normal signaling responses to GnRH as measured by immediate early gene induction by GnRH (Egr1 and Fos) ( Supplementary  Fig. S2). Importantly, the knockdown of GLUT1 did not result in compensation by overexpression of other Slc2a family mRNA (Supplementary Fig. S2) but did recapitulate the ablation of GnRH-induced glycolysis observed with GLUT1 inhibitors (Fig. 4E,F). When we pulsed the Ctrl and shGlut1 cell lines with GnRH, we observed significantly lower LH secretion from the shGlut1 cells as compared to Ctrl ( Fig. 4G-I). To ensure that the decrease in LH secretion from the shGlut1 cell line was due to reduced expression of Glut1 and therefore less glucose uptake, we performed qPCR post-pulse to measure Glut1 and Txnip mRNA . We found that the Glut1 knockdown persisted throughout the pulse experiment ( Fig. 4J) and that the shGlut1 cell line had significantly reduced glucose uptake, even less than WZB117 treated cells, as indicated by lower levels of Txnip expression ( Supplementary Fig. S2). In contrast to the shorter-term approaches of glucose starvation and 2-DG treatment ( Fig. 3H), the stable knock down of GLUT1 resulted in significantly lower LH content than control cells ( Fig. 4K), indicating reduced LH synthesis may contribute to the reduced LH secretion in these cells. Together, these data support the conclusion that glucose uptake facilitated through GLUT1, though not necessary for basal or GnRH-induced LH secretion, supports maximal output of LH by gonadotropes.

Primary mouse gonadotropes recapitulate GnRH-induced glycolytic switch observed in LβT2 cells.
Our data in the LβT2 cell line provide evidence that gonadotropes sense glucose and can integrate this information to alter hormone production. Although primary gonadotropes are glucose responsive 13 , to date, no functional protein or metabolic assays have been performed on primary gonadotropes to verify glucose sensing.
To answer the question of whether primary gonadotropes respond to GnRH by inducing glycolysis, we devised a protocol to obtain and culture purified primary gonadotropes, a prerequisite for XF analysis. High GLUT1 surface protein expression 37 , but not mRNA expression 38 , is exclusive to gonadotropes in the pituitary. Using immunocytochemistry (ICC) on dispersed primary pituitary cultures, we confirmed this observation by demonstrating that bright GLUT1 staining almost exclusively colocalizes with LH positive cells (Fig. 5A). Leveraging this pattern of GLUT1 expression in gonadotropes, we developed a FACS staining workflow to isolate gonadotropes from wild-type mice based on the co-expression of GLUT1 and the GnRH Receptor (GnRHR), exclusive to gonadotropes (Fig. 5B, Supplementary Fig. S3). This method captures sufficient cell numbers for XF analysis and eliminates the need for transgenic overexpression of fluorescent proteins, which introduce metabolic artifacts 39 . We validated that this sorting approach yielded viable purified primary gonadotropes. As expected, gonadotropes made up to 10% of pituitary cells with females having ~ 2% more gonadotropes than male mice ( Supplementary Fig. S4). Cells co-expressing GLUT1 and GnRHR secreted basal LH and responded to GnRH in vitro by increasing secretion of LH, while sorted double negative cells (DN) did not ( Supplementary Fig. S4). Lastly, we determined that the sorted gonadotropes were > 99% pure as measured by the percent of cells staining positive for LH by ICC post sort ( Supplementary Fig. S3).
Using primary gonadotropes (GLUT1 hi GnRHR + ) and non-gonadotropes (DN, double negative) in a mitochondrial stress test, we found that gonadotropes are more bioenergetically active than other cells of the pituitary (Fig. 5C,D). In response to GnRH, we observed no change in OCR, just as in the LβT2 cell line (Fig. 5C,E). Further analysis of primary gonadotrope energetics revealed they have little to no spare respiratory capacity (the difference between basal and maximal respiration), indicating that any increase in metabolism must occur outside of the mitochondria (Fig. 5E) (i.e., additional ATP demand in gonadotropes must be met by glycolysis). We found that GnRH treatment appeared to increase ECAR (Fig. 5F), but more importantly, GnRH treatment reduced the OCR:ECAR ratio (Fig. 5G). These data demonstrate that primary gonadotropes, like LβT2 cells, switch from mitochondria-dependent metabolism to glycolysis in response to GnRH (Fig. 5C-G).

Glucose supports basal LH secretion from primary mouse pituitary cultures. To test whether
blocking glycolysis has a similar impact on LH and FSH secretion from primary cells as it does on LβT2 cell LH secretion, we measured secreted LH from mixed primary pituitary cultures, not purified gonadotropes, in control, glucose-free, or 2-DG conditions. After confirming with Txnip expression that glucose-free and 2-DG conditions impacted primary cells similar to the LβT2 cell line (Fig. 6A), we found that primary pituitary cells have significantly reduced secretion of basal LH, but not FSH in glucose-free conditions (Fig. 6B,C). Surprisingly, LH and FSH secretion were not impacted by 2-DG treatment, highlighting a potential difference of ATP production and energy utilization between primary cells and LβT2 cells.
Scientific RepoRtS | (2020) 10:13063 | https://doi.org/10.1038/s41598-020-69913-z www.nature.com/scientificreports/ GLUT1 is only one of a family of glucose transporters that may facilitate glucose uptake in primary gonadotropes. Using mRNA isolated from whole pituitary, primary sorted gonadotropes (Fig. 5B) and LβT2 cells, we performed expression profiling of Slc2a 1-10, and 12 mRNA and demonstrate that like LβT2 cells, Slc2a1 is the most highly expressed glucose transporter in primary gonadotropes in female mice ( Fig. 6D and 13 ). Importantly, we found that primary gonadotropes exhibit the same Slc2a expression profile as LβT2 cells, though at a higher level. These data suggest that GLUT1 is the main glucose transporter in primary gonadotropes, just as in LβT2 cells. Additionally, expression of Slc2a1 is an order of magnitude higher in purified gonadotropes than in whole pituitary in which they comprise approximately 10% of the population of cells, supporting the finding that GLUT1 expression in the pituitary is predominantly from gonadotropes. GLUT1 expression in the pituitary correlates with GnRHR. The ability of GLUT1 to facilitate maximal GnRH-induced LH secretion (Fig. 4) implies that expression of this transporter may be important for LH secretion in vivo. The LH surge is an acute rise of LH to concentrations higher than any other time of the estrous cycle and is the trigger for ovulation in female mice. GnRH binding to gonadotropes increases prior to the LH surge and declines thereafter [40][41][42] . If the expression of glucose transporters is indeed important for the LH surge, we expect that GLUT1 will be increased in the pituitary with the LH surge and will correlate with GnRHR expression. To test our hypothesis, we ovariectomized female mice and treated them with a well-established estrogen protocol to induce an LH surge [43][44][45] . As expected with this LH surge paradigm, approximately 2/3 of the mice treated with estrogen exhibit an LH surge as measured by serum LH (Fig. 7A,B). Partitioning the data by mice with and without a surge, we found that GnRH receptor protein expression in the pituitary is increased with the LH surge when pituitaries are collected at 6 PM, or lights out on the day of proestrus (Fig. 7C, Supplementary Fig. S5). We also found that GLUT1 protein expression in the pituitary is increased (Fig. 7D, Supplementary Fig. S5). Interestingly, these data mirror the impact of increasing frequency or amplitude of GnRH pulses on mRNA expression of Slc2a1 (Table 1) in our initial observations in LβT2 cells. Next, we combined the data from all mice and performed correlation analysis of protein expression of GnRHR and GLUT1. Our analysis found that GLUT1 positively correlates with GnRHR (Fig. 7E). These data further support the importance of GLUT1 in gonadotrope function. We conclude that GLUT1 expression in gonadotropes correlates with GnRHR expression.

Discussion
In this study, we demonstrated that GLUT1 facilitates glucose uptake and glycolysis in support of LH secretion from gonadotropes. Further, we provide evidence to support investigation of gonadotrope GLUT1-mediated glucose metabolism in facilitating the LH surge. These findings implicate gonadotropes in the overall response of the HPG axis to alterations in energy balance such as hypoglycemia or hyperglycemia.
Our data show that GnRH induces translocation of GLUT1 to the cell membrane, which is important for gonadotrope function. This work extends the established central role of GLUT1 in CD4 + T cell activation to another mammalian system 46 , namely gonadotrope control of reproduction. Though the mechanisms of GnRH regulation of GLUT1 translocation are to be determined, we have revealed a potential gonadotrope-specific regulation of GLUT1. GLUT1 expression and localization at the cellular membrane is classically regulated by local glucose concentration through glucose responsive mechanisms 47 . Low glucose availability induces translocation of GLUT1 to the surface to increase cellular glucose uptake while high glucose availability causes the converse through several sensors of glucose metabolites including carbohydrate response element-binding protein (ChREBP), sirtuin 1 (SIRT1), and AMP-dependent protein kinase (AMPK) 47 . Specifically, AMPK www.nature.com/scientificreports/ blocks the internalization of GLUT1 by phosphorylating and degrading TXNIP protein, which induces GLUT1 endocytosis, thereby increasing glucose uptake 32 . Because AMPK is a known signaling intermediate in GnRHinduced activation of Lhb transcription, GnRH degradation of TXNIP protein through activation of AMPK is a possible mechanism by which GnRHR signaling results in increased GLUT1 at the plasma membrane and increased glucose uptake 48 . The ability of GnRH to induce GLUT1 translocation closely ties the regulation of glycolysis to regulation of LH secretion. There are several potential reasons why glucose metabolism may be important for LH synthesis and/or secretion. (1) ATP from glycolysis could support energy-dependent biosynthesis in response to GnRH stimulation. Deriving ATP from glycolysis is an accessible mechanism to control ATP availability to meet acute demand in a pulsatile manner through GnRH-controlled cycling of GLUT1. Gonadotropes express ATP-gated purinergic (P2X2R) Ca 2+ channels 49 . Extracellular ATP induces a transient rise in cytosolic Ca 2+ that stimulates LH release, and GnRH signaling in primary pituitary cultures increases ATP release, potentially activating parallel autocrine signaling that amplifies LH release 50,51 . Though we have not measured the impact of GLUT1 on ATP release from gonadotropes, other work implicates increased glycolysis as a potential source of autocrine ATP activation of P2X2Rs to support increased LH secretion. (2) Another attractive mechanism is the necessity of ATP to fuel parallel GnRH-induced activation of the cAMP pathway 52 . The cAMP/protein kinase A (PKA) pathway enhances LH release and increases the expression of LHβ and GnRHR 53,54 . GnRHR signaling was shown to be coupled to the cAMP pathway in LβT2 cells 55 and in αT3-1 cells, an immature gonadotrope cell line 52 . During the preovulatory surge in female rats, there is a concurrent accumulation of cAMP in the pituitary which can be blocked with a GnRH antagonist, ganirelix 56,57 . Our data show that GLUT1 expression is increased during the LH surge and could be a source of ATP for the coincidental cAMP surge. Analysis of GnRH induced cAMP is warranted in the context of GLUT1 knockdown in LβT2 cells. (3) A third reason why glycolysis may be important for gonadotrope secretion of LH is the generation of nucleic acid precursors and maintenance of redox balance through the pentose phosphate pathway. Regulation of GLUT1 has been shown to suppress reactive oxygen species through its supply of the pentose phosphate pathway metabolite glucose-6-phosphate 58,59 . The pentose phosphate pathway is an important source of NADPH, an electron donor. NADPH protects against redox stress by providing reducing equivalents to antioxidants such as sulfiredoxin, which is necessary to resolve GnRH-induced oxidative stress 31 . In addition to needing reducing equivalents from NADPH, sulfiredoxin, a GnRH-induced redox factor, is also ATP dependent, further highlighting the integral role that glycolysis as an Mechanisms discovered in gonadotrope cell lines are often difficult to validate in primary cells due to the low representation of gonadotropes in the pituitary. Here, we used a FAC sorting approach that allowed us to perform functional studies (i.e. protein and metabolic analysis) on purified gonadotrope cell cultures. Our approach obviates the need for transgenic mice and will make the study of primary gonadotropes broadly accessible. Interestingly, the clear difference in surface protein expression of GLUT1 on gonadotropes compared to other cells of the pituitary is not seen on the mRNA level 38 . In fact, Slc2a1 is ubiquitously expressed throughout the pituitary, and within the gonadotrope population Slc2a1 expression varies. This difference in protein and mRNA expression point to post-transcriptional regulation of Slc2a1 within the pituitary that needs further investigation. Gonadotropes may utilize RNA binding proteins to post-translationally regulate GLUT1 expression. We have previously demonstrated that the RNA binding protein ELAV-like protein 1 (ELAVL1) is important for the stabilization of gonadotrope mRNA including Gnrhr 60 . ELAVL1 stabilizes mRNA and regulates translation of Slc2a1 in 3T3-L1 cells. Data from Terasaka et al. demonstrates that Slc2a1 mRNA is also bound to ELAVL1 in gonadotropes 60 . These data suggest that discrepancies in protein and mRNA expression of GLUT1 may be due to translation control. It remains to be determined whether the surface co-expression of GLUT1 and GnRHR can be translated to other mammalian species. However, acute glucose sensitivity is a common feature of the mammalian HPG axis. In comparison to the LβT2 cell line, our studies in primary mouse gonadotropes revealed differences in the contribution of glucose metabolism to LH secretion. Unlike LβT2 cells, primary gonadotropes are not dividing and have a substantially higher resting OCR:ECAR ratio than LβT2 cells. This indicates that primary gonadotropes may have a higher capacity to rescue LH secretion in glucose-free conditions by relying on mitochondria. Subplasmalemmal mitochondria in gonadotropes contribute to GnRH-induced reactive oxygen species and Ca 2+ flux 61 , a mechanism which supports LH synthesis and secretion. Further, L-type calcium channels are closely associated with these sub-plasmalemmal mitochondria and appear coordinated with ROS generation by GnRH-induced NADPH oxidase activity 61 . This suggests overall that ion channel activation, GLUT1 activity, and ROS generation may be interconnected in eliciting LH secretion in response to GnRH receptor activation.
We observed that glucose-free conditions inhibited basal LH secretion from both LβT2 and primary cells, while GnRH-induced secretion was only inhibited in the cell line. Interestingly, we found no impact on FSH secretion by glucose-free conditions. These results extended the paradigm of differential regulation of LH and FSH at the level of transcription and secretion studied by many research groups including our own. In our recently published studies we show that FSH is indeed co-secreted with LH in primary mouse gonadotropes in response to pulsatile GnRH stimulation 62 . However, LH appears to be the most sensitive to regulation by GnRH and is impaired by the presence of oleate. Al-Safi et al. did show that, in normal weight women, polyunsaturated fatty acids suppress FSH 63 . This suggests, that FSH is not immune to alterations in energy balance, though this www.nature.com/scientificreports/ topic is not fully explored. One can speculate that FSH stimulation preserves the pool of developing follicles and is necessary for long-term viability of the oocyte reserve, while LH operates on a shorter time frame and promotes ovulation rather than maturation. These could be evolutionary pressures which resulted in differential sensitivity of gonadotropin secretion to energy balance. We found that inhibition of glycolysis and ATP by 2-DG had no impact on LH secretion from primary pituitary cultures. This resistance to 2-DG could be due to several factors including the high OCR:ECAR ratio of primary gonadotropes, the support of other pituitary cells in the culture, or the fact that primary LH secretion was tested in response to a single treatment with GnRH as opposed to pulsatile stimulation. Mouse studies with conditional KO of GLUT1 in the gonadotrope to test the dependence of gonadotropes on GLUT1-mediated glucose metabolism will further our knowledge of how the gonadotrope interprets or adapts to energy status to regulate reproduction. Particularly, in vivo studies of gonadotrope GLUT1 KO will help define the role of GLUT1 in the LH surge.
A reduction in reproductive fitness as a physiological response to reduced energy has been long-established. Here we highlight a fundamental cellular and molecular mechanism which may help explain how energy availability can be interpreted by the HPG axis at the level of the pituitary. Though nutrient sensing is known to occur in the hypothalamus, this work makes evident a novel inherent characteristic of gonadotropes, namely regulation of LH secretion through GLUT1-mediated glucose uptake, that serves to integrate energy status signals with reproduction.

Primary mouse pituitary cell culture. Mouse experiments were performed in accordance with UCSD
Institutional Animal Care and Use Committee regulations under an approved protocol. Whole pituitaries were dissected from wild-type C57BL/6 female mice at 9-10 weeks of age. Whole pituitaries were isolated into icecold PBS and then dispersed by incubation with 0.25% collagenase Type IV and 0.25% trypsin-EDTA (1x) (Life Technologies) as previously described 65 . The cells were plated on poly-l-lysine (Sigma-Aldrich Inc.) coated Nunc 96-well plates (Thermo Fisher Scientific) at a density of 1.5 × 10 6 cells per cm 2 . The cells were cultured for 24 h at 37 °C and 5% CO 2 in high-glucose HEPES-buffered DMEM with 10% FBS prior to experimentation.

Lentivirus transduction and puromycin selection.
Lentivirus used for knockdown of GLUT1 was prepared using the Lenti-X packaging system (Takara Bio USA, CA) with pLKO.1 plasmid. Non-targeting shRNA (Sigma-Aldrich, Mission SHC002) and GLUT1 shRNA (Sigma-Aldrich, Mission TRCN0000079328) encoding pLKO.1 plasmids were separately transfected into Lenti-X 293T cells (Takara Bio USA, CA) in complete DMEM with 10% FBS. Lentivirus-containing supernatants were collected 48 and 72 h after transfection, filtered through a 0.45 μM polyethersulfone syringe filter, and concentrated by 10% polyethylene glycol 8,000 incubation for 16 h followed by centrifugation at 1,600 RCF for 1 h at 4 °C. Viral particles were resuspended in 1/10 of the original volume in serum and antibiotic free-DMEM. Lenti-X GoStix (Takara Bio USA, CA), a rapid test to determine p24 levels in supernatant preparations, were used to determine the viral titer. LβT2 cells were cultured in 24-well plates in complete DMEM with 10% FBS. On the next day, lentivirus was added in serial dilutions with 4 µg/ml polybrene (Fisher Scientific) and incubated for 4 h at 37 °C, followed by a change to DMEM with 10% FBS. As previously published 60 , the transduced cells were selected by culture in 0.5 µg/ml puromycin in DMEM with 10% FBS after 48 h incubation at 37 °C. The media with 0.5 µg/ml puromycin was changed every 2-3 days. Surviving cells were maintained in DMEM with 10% FBS supplemented with 0.25 µg/ml puromycin.

Extracellular flux analysis.
For mitochondrial stress tests, LβT2 cells were seeded on collagen IV coated XF96 plates at 6 × 10 4 cells per well in HEPES-buffered DMEM supplemented with 10% FBS and antibiotics as described above. Cells were cultured for 24 h then serum starved overnight. Prior to XF analysis LβT2 cells were washed twice with extracellular flux (XF) assay media (5 mM HEPES buffered-DMEM containing 10 mM glucose, 4 mM l-glutamine, and 1 mM sodium pyruvate). The media was changed to XF assay media for analysis. Primary sorted mouse gonadotropes and non-gonadotropes (DN, double negative) were washed in XF assay media immediately after cell sorting. Primary sorted cells were adhered to Corning Cell-Tak (Fisher Scientific) coated XF 96 plates at 5 × 10 5 per well. Primary cells were equilibrated for 1 h before XF analysis. Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were measured using a modified mitochondrial stress test procedure as follows: basal OCR followed by sequential addition of 10 nM GnRH, 2.5 μM oligomycin (Calbiochem), 500 μM 2,4 Dinotrophenol (DNP) (Sigma-Aldrich) and 1 μM rotenone + 1 μM antimycin A (Enzo) with the XF96 Extracellular Flux Analyzer (Seahorse Bioscience). All extracellular flux mito stress test data analysis was performed using the Seahorse Explorer (SHORE) Analysis program 66 .
For glycolytic (glyco) stress tests, LβT2 cells were seeded as described for mito stress tests. Prior to XF analysis, LβT2 cells were washed twice with glyco stress test (XF) assay media (5 mM HEPES buffered-DMEM containing, 4 mM L-glutamine and 1 mM sodium pyruvate). The media was changed to glyco stress test XF assay media for analysis. Glyco stress test was performed according to standard protocol on an XF96 Extracellular Flux Analyzer (Seahorse Bioscience, Agilent Technologies). Basal ECAR was measured followed by sequential addition of 10 mM glucose, 2.5 μM oligomycin (Calbiochem), and 50 mM 2-DG (Sigma-Aldrich). Data were extracted using the Wave software v2.6.1 (Seahorse Bioscience, https ://www.agile nt.com/en/produ cts/cell-analy sis/cell-analy sis-softw are/data-analy sis/wave-deskt op-2-6) and manually analyzed. Cells were centrifuged for 5 min at 2,655 RCF at 4 °C, resuspended in 250 µl of 1 × hypotonic lysis buffer (20 mM HEPES, pH 7.4, 10 mM NaCl, 3 mM MgCl2, Roche protease cocktail inhibitors and 1 mM PMSF), and incubated for 30 min to allow for cell lysis. Cells were triturated using 10 passes through a 27 ½ G needle. After BCA quantification of the lysates, samples were diluted to protein concentration of the sample with the lowest protein concentration (2-2.5 mg/ml). The lysate was centrifuged at 2,655 RCF for 10 min and the pellet (nuclear "N") was set aside. The supernatant (containing membrane "M" and cytosolic "C" proteins) was centrifuged again at 2,655 RCF for 10 min to pellet residual "N" fraction. The pellets from the two centrifugations were combined, washed twice, and resuspended with 250 µl of hypotonic buffer. The supernatant ("M" + "C") was centrifuged (Rotor TLA120.2 Beckman Coulter Cat. # 343778) at 96,000 RCF for 1 h at 4 °C and the pellet ("M") was set aside. The supernatant was again centrifuged at 96,000 RCF for 1 h at 4 °C. The clarified supernatant after the two ultracentrifugations was reserved as the cytosolic fraction "C". The remaining "M" pellets were combined, resuspended in 250 µl of hypotonic buffer, and triturated using 10 passes through a 27 ½ G syringe. The resuspended "M" fraction was centrifuged for 45 min at 96,000 RCF at 4 °C. The supernatant from this centrifugation was combined with the cytoplasmic fraction "C". The remaining "M" pellet was resuspended in 250 µl of hypotonic buffer, combined with the nuclear "N" fractions, and solubilized by adding NP-40 at a final concentration of 0.5% to create a mixed "M" + "N" fraction. The combined "M" + "N" fraction was centrifuged twice at 2,655 RCF to pellet nuclear membranes. The nuclear pellets were combined in 500 μl of hypotonic buffer and sonicated to solubilize for the final nuclear "N" fraction. The supernatant from this centrifugation was the final membrane "M" fraction. Samples were prepared with Laemmli sample buffer + 0.  Table 2) with cycling conditions recommended by the manufacturer. Reaction efficiency was determined for each primer set. A melt curve was performed after each PCR run to ensure that a single product was amplified. At least three independent determinations of mRNA content were made, and the relative transcript levels were determined with Gapdh mRNA as an endogenous control.
GnRH pulse stimulation and LH secretion assay. For perfusion experiments as previously described 25 , LβT2 cells were seeded on cytodex 3 microcarrier beads (GE Healthcare, Buckinghamshire, UK) at a density of 1.5 × 10 7 cells/ml bead volume. After culture for 24 h, cells were pelleted by spinning at 550 RCF for 1 min, resuspended in 10 ml serum-free DMEM with antibiotics, repelleted, and placed in 10 ml serum-free DMEM for 16 h. Cells were then loaded into perifusion columns and equilibrated for 1 h in serum-free, phenolred free DMEM supplemented with penicillin and streptomycin at a flow rate of 200 µl/min. Subsequently, cells were pulsed for 2 min with 40 nM GnRH for a final amplitude of 10 nM as determined by dilution of phenol red tracer dye at 58 surge was induced through one of two paradigms. For most animals, five days after surgery, animals were given a subcutaneous injection of 0.25 ug of β-estradiol benzoate (EB; Sigma-Aldrich) in 100 μl of sesame oil 4 h after lights on. On the following day, animals were given 1.5 μg of EB in sesame oil (100 μl) 4 h after lights on. The following day, animals were euthanized at lights off and pituitaries were snap frozen on dry ice and blood collected and processed to serum. Sera was processed using a Luminex assay and the Millipore MAGPIX as described, and an LH surge was considered a value greater than 1.5 ng/ml. The LH surge in three animals was induced by ovariectomy and E2 pellets as described 43 . There is no difference in the amplitude of the LH surge between these paradigms 43 . For this paradigm, LH in serum samples was measured via Luminex or via the University of Virginia Ligand Core, and an LH surge was accepted at a concentration of 0.6 ng/ml. Pituitaries where homogenized (PRO Scientific Bio-Gen PRO200 Homogenizer), then lysed in RIPA/NP-40 lysis buffer. Western blots were performed on pituitary lysates.

Statistics.
All experiments were repeated at least three times independently, and reported values are presented as the means ± standard error of the mean. JMP software v14.0 (SAS Institute, https ://www.jmp.com/en_ us/home.html) was used to perform all statistical analysis. Analysis was performed on raw, normalized values or data Box Cox transformed to correct for heteroscedasticity. Data were evaluated by ANOVA and appropriate post hoc testing unless otherwise indicated. A value of p < 0.05 as indicated was considered significant. Asterisks denote significance from the control. Data sharing a letter are not significantly different from each other, while data marked by letters exclusive to another data point are significantly different from each other. Table 2. Primer sequences used for quantitative reverse transcriptase PCR. *All primers were validated by PCR product analysis on an agarose gel and melt curve analysis. Low expression of several Slc2a members precluded accurate determination of reaction efficiency. An assumed efficiency of 100% was used for analysis for these Slc2a members.