Effects of Gonadal Hormones On Glutamatergic Circuits in the Retina

Gonadal hormones function as neurosteroids in the retina; however, their targets in the retina have not yet been identied. The present study examined the effects of gonadal hormones on glutamatergic circuits in the retina. Extracellular glutamate concentrations, which correspond to the amount of glutamate released, were monitored using an enzyme-linked uorescent assay system. Progesterone and pregnenolone both increased extracellular glutamate concentrations at a physiological concentration in pregnancy, whereas estrogen and testosterone did not. Synaptic level observations using a patch clamp technique revealed that progesterone increased the activity of glutamatergic synapses. We also investigated whether high concentrations of gonadal hormones induced changes in the retina during pregnancy. The present results indicate that progesterone activates glutamatergic circuits as a neurosteroid when its concentration is elevated in pregnancy. The present results indicate that progesterone increased the frequency of glutamate release from bipolar cell terminals without affecting presynaptic vesicle sizes or the properties of postsynaptic glutamate receptors. In addition, the effects of progesterone were distinct, particularly in females, conrming the results of the enzyme-linked uorescent assay system. Two techniques were used in the present study to monitor glutamate release from presynaptic terminals in the animal model: an enzyme-linked uorescent assay system and electrophysiological recordings. We previously demonstrated that the enzyme-linked uorescent assay system was useful for visualizing extracellular glutamate concentrations at individual layers of the retina 35 . Glutamate release from presynaptic terminals was monitored as an increase in extracellular glutamate concentrations using this method. Regarding electrophysiological recordings, we assessed the activity of presynaptic terminals and the biophysical properties of postsynaptic glutamate receptors to monitor excitatory postsynaptic currents (EPSCs).


Introduction
Gonadal hormones synthesized in the central nervous system function as neurosteroids [1][2][3][4] . They modulate GABA A and NMDA receptors and contribute to synaptogenesis in the developing brain as well as the differentiation of oligodendrocytes. These hormones also function as neuroprotective and antianxiolytic agents. The effects of gonadal hormones as neurosteroids have been investigated at multiple levels in the retina. The presence of receptors for and the synthetic ability of gonadal hormones in the retina have been demonstrated using immunohistochemical methods [5][6][7] . Furthermore, sex-related differences in retinal function and diseases 8-10 as well as estrus cycle-dependent changes in ocular function [11][12][13][14] have been reported. Based on these biophysiological ndings, the therapeutic application of gonadal hormones to the treatment of retinal diseases has been examined. The neuroprotective effects of progesterone may be useful for the treatment of retinitis pigmentosa [15][16][17][18][19][20][21][22][23] . Similarly, estrogeninduced increases in retinal blood ow 24 may be bene cial for neuroprotection 25 . Increases in the thickness of the choroid in pregnancy have frequently been reported [26][27][28][29][30] , which may be due to the larger blood volume in pregnancy. In addition, visual changes commonly occur in pregnancy 12-14, 31,32 . The effects of gonadal hormones as neurosteroids in the retina are supported by accumulated evidence; however, their targets have not yet been examined in detail at the cellular level. Furthermore, although changes in retinal thickness in pregnancy have been demonstrated 33,34 , a systematic analysis has not been conducted. Therefore, the present study examined the effects of gonadal hormones in the retina. We investigated whether gonadal hormones modulate the activity of glutamatergic circuits in the retina using an enzymelinked uorescent assay system 35 and patch clamp technique. The results obtained revealed that progesterone increased the activity of glutamatergic circuits, whereas estrogen and testosterone did not. We also attempted to establish whether high concentrations of gonadal hormones in pregnancy induce glutamate toxicity in the human retina. Collectively, the present results indicate that progesterone activates glutamatergic circuits without toxic effects when its concentration is elevated in pregnancy.

Results
Effects of gonadal hormones on extracellular glutamate concentrations Extracellular glutamate concentrations were estimated based on uorescent intensity measured using the enzyme-linked uorescent assay system (Fig. 1A). Since uorescent intensity in the OS contains a signi cant amount of the intrinsic uorescent signal of NADH in this system, the signal level in the OS becomes very high 35 .
In females, the application of 1 µM progesterone increased the intensity of uorescent signals (Fig. 1B).
In most samples, the intensity of uorescent signals peaked within 2-4 min and then gradually decreased.
We then applied a high K solution to the same samples in order to con rm the viability of samples.
The percent change in signal intensity (dF/F) at individual ROI was calculated using our previously described method 35 . We demonstrated that dF/F in individual layers re ected changes in glutamate concentrations and also that these changes were not limited to the synaptic layers (OPL and IPL) 35 . In females, a signi cant increase in dF/F was observed in all layers (Fig. 1C). In males, a signi cant increase in dF/F was not detected in any layers, whereas an increase in dF/F was found in all layers (Fig.   1D).
We then examined the effects of estrogen or testosterone on glutamatergic circuits in females using the enzyme-linked uorescent assay system. The application of 1 µM 17-β-estradiol increased the intensity of uorescent signals in all layers ( Fig. 2A). However, the increase in dF/F was subtle. The application of 1 µM testosterone did not induce any increase in uorescent signals in any layers (Fig. 2B).
Since progesterone is synthesized from pregnenolone, we investigated whether pregnenolone mimics the effects of progesterone. The application of 1 µM pregnenolone induced an increase in uorescent signals in INL, IPL, and GCL (Fig. 2C). We also observed an increase in dF/F in ONL and OPL, whereas that in dF/F was not signi cant.

Effects of progesterone on EPSCs
We investigated whether progesterone increased the release of glutamate in the retina. To monitor the activity of glutamatergic circuits, we recorded EPSCs from retinal ganglion cells that receive glutamatergic inputs from bipolar cells (Fig. 3A, B, and C). We examined the following six parameters.
Frequency and total charge transfer are employed to evaluate the activity of glutamatergic synapses, while the amplitude, decay time, rise time, and charge transfer of individual EPSCs are useful for assessing presynaptic vesicle sizes or the properties of postsynaptic glutamatergic receptors. In the present study, we limited our electrophysiological analysis data to distinguish EPSCs from baseline noise (amplitude >8pA), as described in the Methods section.
In females, progesterone increased the frequency (Fig. 3D, Table 1) and total charge transfer (Fig. 3I, Table 1) of EPSCs. In males, increases in both the frequency (Fig. 3D, Table 1) and total charge transfer  Table 1) were not signi cant; however, progesterone appeared to affect both parameters.
Progesterone increased the charge transfer of individual EPSCs in males (Fig. 3H The present results indicate that progesterone increased the frequency of glutamate release from bipolar cell terminals without affecting presynaptic vesicle sizes or the properties of postsynaptic glutamate receptors. In addition, the effects of progesterone were distinct, particularly in females, con rming the results of the enzyme-linked uorescent assay system.

Changes in retinal thickness during pregnancy
In the present study, we showed that 1 µM progesterone, a nearly maximum concentration in pregnancy 37 , activated the glutamatergic circuit of the retina in vitro. When extracellular glutamate concentrations are high, glutamate may act as a toxin that induces cell death 41 . Therefore, we examined whether the elevated concentration of progesterone in pregnancy is associated any morphological changes in the human retina. Since a change in retinal thickness in pregnancy has been reported in age-matched studies 33,34 , we monitored retinal thickness during pregnancy to assess morphological changes in the same pregnant women.
In the postpartum stage, progesterone concentrations (43 ± 16 nM, mean ± SD) were within normal values for the regular menstrual cycle (Fig. 4). Rapid elevations in progesterone concentrations during pregnancy were observed in the 1st trimester. The mean concentration of progesterone in the 1st trimester (153 ± 3 nM, mean ± SD) was 2.5-fold that of the highest concentration under the regular menstrual cycle. Progesterone concentrations continued to gradually increase in the 2nd trimester (188 ± 50 nM, mean ± SD) and peaked in the 3rd trimester (243 ± 61 nM, mean ± SD).
We used the thickness of the retina in the postpartum stage as the control because the concentration of progesterone returned to the level of the regular menstrual cycle. We measured the thickness of the retina in 9 regions ( Table 2, Fig. 5). Retinal thickness was calculated at the parafovea and perifovea for individuals and averaged (see the Methods section). Signi cant decreases in thickness were observed in the parafovea and perifovea in the 1st trimester ( Fig. 6), but not in the 2nd trimester. In the 3rd trimester, a signi cant decrease in thickness was only detected in the parafovea.

Discussion
In the present study, we investigated the effects of gonadal hormones on glutamatergic circuits at 1 µM. The results obtained demonstrated that progesterone and pregnenolone increased the activity of glutamatergic circuits at physiological concentrations in pregnancy, whereas neither estrogen nor testosterone modulated the activity of these circuits. In women, progesterone and estrogen concentrations in blood increase and decrease during the menstrual cycle and markedly increase in pregnancy 36 . Under a regular menstrual cycle, the concentrations of progesterone and estrogen uctuate between 3 and 60 nM and between 0.3 and 1.2 nM, respectively. In pregnancy, the maximum concentrations of both progesterone and estrogen peak at 1 µM 37 . The concentration of testosterone in blood is <35 nM 36 . Therefore, the effects of 1 µM progesterone on glutamatergic circuits may be of physiological relevance.
Regarding the physiological importance of functional changes in the retina, we herein demonstrated that the modulatory effects of progesterone on glutamatergic circuits were stronger in females than in males. In other words, glutamatergic circuits may be more sensitive to progesterone in females than in males. However, no signi cant differences in the localization or expression of progesterone receptors were reported between male and female mice 23 . Many pregnant women develop physiological alterations in the eye 12-14,30−32 . Among the physiological alterations reported, changes that occur in the visual eld may be related to the present results. A change in the visual eld was detected as an altered mean threshold sensitivity without subjective symptoms. Although increases in progesterone concentrations in pregnancy may induce functional changes in the retina, these changes may be compensated for through plastic changes in the visual system. Further studies are needed to con rm whether subtle subjective symptoms are present in normal pregnant women.
We observed a detectable reduction in retinal thickness during pregnancy and an elevated extracellular glutamate concentration in the enzyme-linked uorescent assay system. However, glutamate toxicity is not likely to induce this reduction in retinal thickness for the following reasons. The increase induced in EPSCs by progesterone may elevate the concentration of glutamate at the synaptic cleft. When this elevation persists for a longer time, it may induce glutamate toxicity. However, it is important to note that the increase observed in the extracellular concentration of glutamate in the enzyme-linked uorescent assay system was detected in the presence of TBOA, an antagonist of glutamate transporters 38 . We previously reported that uorescence did not increase in the absence of TBOA when a high K stimulation was performed 35 , suggesting that glutamate concentrations at the synaptic cleft are tightly controlled by glutamate transporters under physiological conditions. The increase in glutamate release observed in the present study re ects the activation of glutamatergic circuits potentially inducing the subtle functional modulation of visual function 12-14,30−32 , but not glutamate toxicity in pregnancy. In the present study, we used retinal thickness in the postpartum period as the control because the progesterone concentration in blood is close to that in the non-pregnant stage. In a previous age-matched study, foveal thickness in the 1st trimester was similar to that in the non-pregnant stage, whereas a signi cant increase in foveal thickness was detected in the 2nd and 3rd trimesters 33 . If this increase in retinal thickness persists for a few months after delivery, a decrease in retinal thickness may be reconsidered from a different viewpoint.
Alternatively, a temporal decrease in retinal thickness may be explained by increased pressure from the choroid, which becomes thicker due to increased blood ow in pregnancy [26][27][28][29][30] . Further studies are warranted.
We found that progesterone and pregnenolone both increased extracellular glutamate concentrations using the enzyme-linked uorescent assay system. We also demonstrated that progesterone increased glutamatergic inputs from bipolar cells in retinal ganglion cells. This result indicates that bipolar cells are one of the targets of progesterone. At the immunocytochemical level, the expression of progesterone receptors has been reported in both Müller cells and the pigment epithelium 5 of the mouse retina and Müller cells of the pig retina 6 . In addition, at the immunohistochemical level, immunoreactivity for progesterone receptors was found to be widely distributed, including two synaptic layers (OPL and IPL) 23 . In previous studies, progesterone and pregnenolone were shown to act on TRP channels (TRPM1 42 , and TRPM1 has been detected in the dendrites of rod bipolar cells 49 . Pregnenolone has been shown to activate TRPM3 and increase Ca in ux in the mouse retina 45 . However, TRPM3 channels do not appear to be the target of pregnenolone because progesterone was found to inhibit TRPM3 channels via a pregnenolone-independent mechanism 44 . The activation of TRPM1 by pregnenolone has also been supported in the recombinant system 42 . According to this experiment, the current amplitude of homomeric TRPM1 receptors is very small, while that of chimeric receptors of TRPM1 and TRPM3 is large. Therefore, further studies are needed to establish whether TRPM1 is the actual target of progesterone.
In the present study, we showed that an elevated progesterone concentration in pregnancy may activate glutamatergic circuits at the bipolar cell level in the retina and also that the increased release of glutamate from synaptic terminals may induce subtle changes in vision in pregnancy. Since the retina uses many types of neurotransmitters at multiple synaptic levels, the effects of gonadal hormones on other neural circuits need to be clari ed in order to obtain a more detailed understanding of their effects on the retina in pregnancy.

Approval from Ethical Committees
The experimental procedure for enzyme-linked uorescence assays on animals was approved by the Institutional Animal Care and Use Committee of Fujita Health University.
The experimental procedure for electrophysiological recordings on animals was approved by the Animal Experiments Ethical Review Committee of Nippon Medical School.
The experimental procedure for the clinical study was approved by the Ethics Committee at the coordinating center of the University of Tokyo and the Institutional Review Board of the University of Tokyo.
All studies were performed in accordance with the relevant guidelines and regulations (for animal studies), and the Declaration of Helsinki. All animal studies were conducted in accordance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines.
Experimental procedure for animals Two techniques were used in the present study to monitor glutamate release from presynaptic terminals in the animal model: an enzyme-linked uorescent assay system and electrophysiological recordings. We previously demonstrated that the enzyme-linked uorescent assay system was useful for visualizing extracellular glutamate concentrations at individual layers of the retina 35 . Glutamate release from presynaptic terminals was monitored as an increase in extracellular glutamate concentrations using this method. Regarding electrophysiological recordings, we assessed the activity of presynaptic terminals and the biophysical properties of postsynaptic glutamate receptors to monitor excitatory postsynaptic currents (EPSCs).

Measurement of gonadal hormone concentrations for animal experiments
In women, progesterone and estrogen concentrations in blood change during the menstrual cycle and are high in pregnancy. Under a regular menstrual cycle, concentrations change from 3 to 60 nM for progesterone and from 0.3 to 1.2 nM for estrogen 36 . In pregnancy, progesterone and estrogen concentrations in blood both increase to a maximum of 1 µM 37 . The concentration of testosterone in blood is reportedly <35 nM 36 . Therefore, we investigated the effects of gonadal hormones at 1 µM in the present study.

Enzyme-linked uorescent assay
Details on the method used are described in our previous study 35 . In this method, extracellular glutamate concentrations were monitored as the uorescent signal intensity of NADH, which is a product of the catalytic effects of glutamate dehydrogenase (GDH) between glutamate and nicotinamide adenine dinucleotide (oxidized form). Slices were then exposed to HEPES-buffered solution containing one of the gonadal hormones (progesterone, estrogen, or testosterone) or their derivative (pregnenolone sulfate), 5 mM NAD + , and 50 µM TBOA. All solutions were bubbled with 100% O 2 and perfused at a rate of 1.5 ml/min. The high K solution was applied as a control stimulation to assess the viability of samples. All experiments were performed at room temperature. Gonadal hormones or their derivatives (purchased from Sigma) were dissolved in HEPES-buffered solution containing 0.1% dimethyl sulfoxide.

Data analysis
Data were selected using the same criteria reported in our previous study to avoid contamination by artifacts 35 . To assess the effects of progesterone, estrogen, testosterone, and pregnenolone sulfate at the layer level, the intensity of uorescent signals was calculated using uorescent signals for each ROI (Fig. 1A). Statistical analyses were performed using a one-sample t-test. Since samples for progesterone did not show a Gaussian distribution, a non-parametric analysis (the Wilcoxon signed-rank test) was used. All statistical tests were performed using Prism 7.0 (GraphPad Software, La Jolla, CA).

Electrophysiological recordings
We followed the method described in our previous studies 39,40 .

Recordings of EPSCs from retinal ganglion cells
A whole-mount preparation was placed on the chamber vitreous side up and viewed under a uorescent microscope (BX50WI, Olympus, Tokyo, Japan). The input resistance of patch pipettes was 8-12 MΩ when lled with Cs + -based intracellular solution ((in mM) 115 CsCl, 5 QX-314, 0.5 CaCl 2 , 5 HEPES, 5 EGTA, 5 ATP-3Na, and 1 GTP-1Na; pH was adjusted to 7.3 with CsOH). To avoid contaminated recordings from displaced amacrine cells, cells with a membrane capacitance of <14 pF were excluded from the analysis. The average membrane capacitance and input resistance of recorded cells were 26 ± 8 pF and 43 ± 9 MΩ (mean ± SD), respectively. EPSCs were recorded using a patch clamp ampli er (Axopatch-200B; Axon Instruments, Foster city, CA, USA) at a holding potential of -70 mV after blocking IPSCs in Ringer's solution containing 1 µM strychnine and 100 µM picrotoxin. Data were sampled at 10 kHz after passing through a low-pass lter at 5 kHz using a commercially available program (pCLAMP9; Axon Instruments, Foster city, CA, USA). Progesterone was dissolved in dimethyl sulfoxide ( nal concentration in Ringer's solution of 0.1%).

Data analysis of EPSCs
Recorded signals were analyzed off-line. Since the contamination of spontaneous EPSCs made the calculation of mean baseline noise di cult, we were unable to identify EPSCs using the calculated mean baseline noise. Therefore, signals with an amplitude >8pA (approximately 5-fold that of the estimated mean baseline noise when the contamination of spontaneous EPSCs was absent) were automatically detected with the commercial program Minianalysis (Synaptosoft, Decatur, GA, USA) in the present study.
A whole trace was then visualized to check for the over-or under-detection of events. The timing of EPSC events was de ned as the time of an individual EPSC peak. We analyzed the EPSCs of cells when the frequency of EPSCs was >1 Hz. The average frequency, amplitude, rise and decay times, charge transfer of individual EPSCs, and the sum of the charge transfer of EPSCs for 30 seconds in the presence and absence of progesterone were analyzed. Statistical analyses were performed using a paired t-test (twotailed).

Clinical study
Participants and enrollment criteria We followed up twelve pregnant women in the present study. Informed consent was obtained from all participants, and those who did not grant authorization for the use of their medical records in research were excluded from the analysis. The medical histories of participants were reviewed at the outpatient clinic of the University of Tokyo Hospital. Inclusion criteria for pregnant women were as follows: (1) eyes with a spherical equivalent between -6 diopters and +3 diopters, and (2) eyes with clear ocular media.
Exclusion criteria were the presence of other eye diseases (e.g. chorioretinal atrophy in the macula, glaucoma, and any other retinal disorders) and high myopia (-6.0 diopters or less). None of the participants were diagnosed with central serous chorioretinopathy (CSC). All data were fully anonymized before the assessment of data.

Eye examination
Participants underwent a set of a comprehensive ophthalmological examinations in the 1st, 2nd, and 3rd trimesters and the postpartum period. These examinations included the measurement of best-corrected visual acuity, AL (IOL master, Tomey OA-2000, version 5.4.4.0006; Tomey, Nagoya Japan), and refractive error (KR-8900 version 1.0.7; Topcon Corp., Tokyo, Japan). After image capturing in each study visit, exported OCT (HRA spectralis; Heidelberg Engineering GmbH, Dossenheim, Germany) images were analyzed using custom written software. An automated graph-based method was used to measure retinal thickness.

Measurement of progesterone
Progesterone concentrations were measured in blood samples collected at the 4 stages of gestation (the 1st, 2nd, and 3rd trimesters and postpartum period) according to the manufacturer's instructions (progesterone ELISA kit, Cosmo Bio.).

Measurement of retinal thickness
We separated the retina into 9 regions (Fig. 5A) and measured the thickness of each region (for see Table  2) to assess possible glutamate toxicity. Changes in the thickness of the retina during pregnancy were assessed using the thickness of the fovea and two surrounding concentric regions (the parafovea and perifovea) (Fig. 5B). The thicknesses of the parafovea and perifovea were calculated as the average thickness of 4 regions (temporal, nasal, superior, and inferior).

Statistical analysis
All statistical analyses were conducted using statistical programming language "R" (version 3.  Responses to a 17-β-estradiol (A), testosterone (B), or pregnenolone (C) stimulation using the enzymelinked uorescent assay system. The experimental protocol was the same as that shown in Fig. 1. The numbers of samples were 5 for females (estradiol), 5 for females (testosterone), and 3 for females (pregnenolone sulfate). OS, the outer and inner segments of photoreceptors; ONL, the outer nuclear layer; bipolar cells (BCs), EPSCs recorded from RGCs re ected the activity of glutamatergic synapses between BCs and RGCs. B. Example of a whole-cell recording from an RGC. An RGC was held at -70 mV.   Spe, superior perifovea; Spa, superior parafovea; Tpe, temporal perifovea; Tpa, temporal parafovea; Npe, nasal perifovea; Npa, nasal parafovea; Ipe, inferior perifovea; Ipa, inferior parafovea. B. The central region (fovea) and the two concentric regions (the parafovea and perifovea) used for the assessment of retinal thickness. The thickness of the parafovea used was (Spa + Tpa + Npa + Ipa)/4. The thickness of the parafovea used was (Spe + Tpe + Npe + Ipe)/4. Figure 6