Photoperiodic regulation in a wild-derived mouse strain

Species living in temperate latitudes face challenging seasonal changes in the environment. The annual rhythm in photoperiod serves as a reliable signal to anticipate these changes and thus optimize transitions in physiology and behavior. In mammals, photoperiod perceived by the eyes is transduced by a complex neural pathway into the nocturnal production of melatonin by the pineal gland (Goldman, 2001). The neuroendocrine pathway decoding the melatonin photoperiodic message and its integration on the control of reproduction, metabolism or molt have been elucidated to a certain extent, using species which present robust seasonal changes in physiology, such as sheep or hamsters.

The pars tuberalis (PT) of the pituitary gland, which is melatonin-sensitive in mammals (Morgan et al., 1994; Weaver et al., 1989; Williams et al., 1989), is the critical neuroanatomical site where the photoperiodic melatonin signal is integrated. Exposure to short nocturnal melatonin profile under long days induces thyrotropin (TSH) expression in PT-specific thyrotrope cells, mainly via upregulation of TSHβ subunit expression, whereas exposure to long melatonin profile in short days downregulates TSH expression (Bockmann et al., 1996; Dardente et al., 2003; Wittkowski et al., 1988). PT TSH, in turn, acts on TSH receptor (TSHr)-expressing hypothalamic tanycytes, located on the wall of the third ventricle (3V) (Hanon et al., 2008; Klosen et al., 2013; Ono et al., 2008), where it upregulates type 2 (dio2) and downregulates type 3 (dio3) thyroid hormone deiodinase expression (Barrett et al., 2007; Revel et al., 2006b). dio2 transforms the circulating thyroxine (T4) form of TH into the active form triiodothyronine (T3), whereas dio3 degrades T3 and T4 to diiodothyronine and reverse T3, respectively. Thus, the opposite TSH-driven regulation in dio2/3 leads to higher intrahypothalamic levels of T3 in long (LP) as compared with short (SP) photoperiod. Central hypothalamic delivery of either TSH or T3 in SD-adapted hamsters and sheep restores summer physiology (Barrett et al., 2007; Hanon et al., 2008; Klosen et al., 2013; Murphy et al., 2012), demonstrating a role of the TSH/DIO/T3 neuroendocrine pathway in the photoperiodic control of seasonal metabolism and reproduction.

The photoperiodic TSH/T3 message has been proposed to control the hypothalamic reproductive network via the regulation of the neuropeptides RFamide related peptide 3 (RFRP3) in the dorsomedial ventromedial hypothalamus (DMH/VMH) and kisspeptin (Kp) in the arcuate nucleus (Arc) (Angelopoulou et al., 2019; Simonneaux, 2018). RFRP3 expression shows a conserved melatonin-driven inhibition leading to lower levels in SP as compared with LP in all species investigated (Ancel et al., 2012; Dardente et al., 2008; Janati et al., 2013; Revel et al., 2008; Sáenz de Miera et al., 2014; Smith et al., 2008; Ubuka et al., 2012), but its effect on reproductive activity shows species, sex and reproductive stage differences (Ancel et al., 2012; Henningsen et al., 2017; Ubuka et al., 2012). Kp is a well conserved activator of GnRH neurons, but Arc Kp expression is inhibited by both melatonin in SP and sex steroids in LP, leading to species-dependent differences in seasonal variation (Rasri-Klosen et al., 2017; Revel et al., 2006a; Sáenz de Miera et al., 2014; Smith et al., 2005). The mechanism through which RFRP3 and Kp neuronal activity are regulated by the photoperiodic changes in the TSH/T3 signal is currently unknown. Functional studies using genetic tools are required to fully unravel the molecular networks that integrate the photoperiodic signal into these neuroendocrine systems; such tools are scarce in ‘traditional’ photoperiodic models, but widely available in mice.

In contrast to seasonal rodent species, common laboratory mice strains present no signs of photoperiodic or seasonal transitions in metabolic and reproductive physiology, remaining reproductive all year round (Bronson, 1979; Laurie, 1946). However, studies using melatonin-deficient and -proficient mice have contributed to the functional characterization of the photoperiodic neuroendocrine pathway in mammals. Photoperiodic differences in TSH, dio2 and dio3 gene expression were observed in the melatonin-proficient CBA strain, while these photoperiodic differences were not observed in the melatonin deficient C57BL/6J (C57) strain, except when the latter received long-term melatonin administration (Ono et al., 2008). This study demonstrates that the TSH/dio2/3 pathway is conserved in mice, whether producing melatonin or not, although neither of these mice strains show photoperiodic differences in reproductive parameters (Ono et al., 2008).

MSM/Ms (MSM) is an inbred mouse strain derived from Japanese wild mice, Mus musculus molossinus Temminck 1844, captured in Japan in the 1970s (Goto et al., 1989); these mice maintain the ability to synthetize melatonin (Goto et al., 1989) with longer nighttime melatonin production in SP than in LP (Kasahara et al., 2010). Notably, offspring of C57×MSM backcrosses show a moderate reduction in testis development associated with melatonin productivity (i.e. associated with carrying MSM allelic variants for melatonin synthesis enzyme genes) (Kasahara et al., 2010), suggesting a potential photoperiodic modulation of the central control of reproduction and metabolism in this strain.

The objective of the present study was to explore the effects of different photoperiodic exposure on pituitary–hypothalamic gene expression, reproductive neuropeptides and on physiology in the M. musculus strain MSM. Furthermore, as maternal melatonin impacts the metabolic and reproductive offspring development in the hamster (Sáenz de Miera et al., 2017; Weaver and Reppert, 1986), we also explored whether the gestational photoperiodic environment influences gene expression in newborn MSM offspring.

Two different mouse strains were used: C57BL/6J (C57; purchased from Jackson, Bar Harbor, ME, USA) and Mus musculus molossinus (MSM; bred in-house from couples purchased from Jackson, with initial founders from the National Institute of Genetics, Mishima, Japan). Mice were maintained at the Chronobiotron (CNRS-UMS 3415) at the University of Strasbourg (photoperiodic and melatonin sampling experiment) and at the animal facilities in the Zoology building at the University of Aberdeen (gestational experiment). Upon arrival, in both animal facilities, mice were placed 3–5 per cage at a constant temperature (21±2°C) and humidity (55±5%), with food and water provided ad libitum, and under long photoperiod (LP: 16 h:8 h light:dark) with lights on at zeitgeber time 0 (ZT0). As indicated below, in some experimental paradigms, mice were transferred into short photoperiod (SP: 8 h:16 h light:dark) conditions.

Protocols were submitted to the Comité Régional d'Ethique en Matière d'Expérimentation Animale of the University of Strasbourg and to the Animal Welfare and Ethics Review Board at the University of Aberdeen. All experiments were conducted in accordance with the UK Home Office (Project Licence 60/4504) and the European Directive UE/63/2010 on the protection of animals used for scientific purposes.

A first cohort of C57 and MSM male mice was used to analyze the effect of photoperiod on reproductive physiology and neuroendocrine gene expression. Six-week-old individuals housed in LP were either kept under LP or transferred to SP conditions for approximately 6 weeks (N=22). At the end of the experiment, animals were euthanized by CO₂ inhalation and their body mass, seminal vesicle mass and paired testes mass were measured, and trunk blood was taken for hormone assays. Gonadosomatic index (GSI) was calculated as testis mass (mg)/body mass (g)×10. Animals were then transcardially perfused with periodate–lysine–paraformaldehyde (PFA) fixative. Brains were sampled and dehydrated through an ethanol gradient and embedded in polyethylene glycol, as previously described (Klosen et al., 1993).

As results of the first cohort indicated a robust photoperiodic integration, a second cohort of MSM mice was used to determine the photoperiodic change in the daily profile of melatonin. Adult male mice (3 to 7 months) were either kept under LP or transferred to SP conditions for 6 weeks, then euthanized by CO₂ inhalation at different times of the day (ZT2, ZT8, ZT17, ZT19, ZT21 and ZT23 under LP; ZT4, ZT9, ZT12, ZT16, ZT20 and ZT23 under SP, N=53). Pineal glands were sampled and stored at −20°C for further melatonin assay.

A third cohort of MSM was used to study the effects of photoperiodic exposure in utero. Two-month-old male and female MSM mice were paired in LP, and 3 days after pairing, all males were removed from the cages, and half of the females were maintained under LP while the other half was transferred to SP. All female mice were checked daily for births, and the day a litter was found was assigned as P0 for that litter. At each sampling point, whole litters were culled and each group included animals from at least three different litters. The first sampling was done at P1; at this point, whole litters – including the dams – were culled for the study. Offspring from the remaining litters were kept in the same gestational photoperiod, were weaned at 3 weeks of age and siblings were placed in same-sex cages. At the second sampling point, at P50, whole cages of animals from LP or SP were culled for the study (N=47). At culling, newborns were decapitated and the full head was snap-frozen and stored at −80°C until processed for gene expression analysis. P50 and dams were killed by cervical dislocation, the body and sex organs (testis in males and both ovaries and uterine horns in females) were weighed, and brains were dissected, snap-frozen and stored at −80°C until processed for gene expression analysis.

Gene expression was analyzed by non-radioactive in situ hybridization in the first cohort of C57 and MSM male mice under LP or SP and by radioactive in situ hybridization in the third cohort of MSM mice under gestational LP or SP.

Non-radioactive in situ hybridization was performed with 300 to 400 bp antisense riboprobes selective for the mouse TSHβ (NM_009432.2), dio2 (NM_010050.4), dio3 (NM_172119.2) and rfrp (NM_021892.1) and rat kiss1 (NM_181692.1) genes, labeled with digoxigenin (DIG RNA labeling kit, Roche, Meylan, France) and hybridized on 10 μm-thick coronal brain sections as already described (Klosen et al., 2013). Briefly, sections (one in 10 throughout the hypothalamus), mounted on SuperFrost ultraplus slides (Menzel-Glaser, Brunswick, Germany), were postfixed in PFA 4% in phosphate buffer (PB) for 10 min at room temperature, rinsed with PBS, treated with 0.5 μg ml⁻¹ proteinase K (Roche) for 30 min at 37°C, rinsed with ice-cold PFA 2% in PB, rinsed again in PBS, acetylated twice for 10 min with 0.25% acetic anhydride in 100 mmol l⁻¹ triethanolamine (TEA) and finally equilibrated in 5× saline-sodium citrate (SSC) 0.05% tween 20% (TW20) twice for 5 min at room temperature. Hybridization was performed with 2 μg ml⁻¹ antisense probe in a medium containing 50% formamide, 5× SSC, 5× Denhardt's solution, 0.1% TW20 and 1 mg ml⁻¹ salmon sperm DNA for 40 h at 60°C. High-stringency washes were performed with 0.1× SSC 0.05% TW20 at 72°C to reduce nonspecific labeling. The DIG tag was detected using an alkaline phosphatase-coupled anti-digoxigenin antibody (1:5000, Roche). Alkaline phosphatase activity was visualized with a mixture of Nitro Blue Tetrazolium/Bromo-Chloro-Indolyl Phosphate for 1 to 2 h and stopped before the staining intensity reached saturation. After detection, slides were premounted using Crystalmount aqueous mounting medium (Sigma-Aldrich, Saint-Quentin-Fallavier, France) and mounted with Eukitt^® (Sigma-Aldrich).

For the radioactive in situ hybridization, antisense rat TSHβ (GenBank accession number M10902) and vole dio2 and dio3 (Król et al., 2012) riboprobes were labeled with ³⁵S-UTP, followed by column purification (Illustra probe quant G-50 micro-columns; GE Healthcare, Little Chalfont, UK) and hybridization was performed as already described (Król et al., 2012). Briefly, 16 μm-thick coronal sections, one in eight sections throughout the hypothalamus, were mounted on poly-l-lysine coated slides, fixed in ice-cold 4% PFA, dehydrated in serial alcohol baths and acetylated in 0.25% acetic anhydride in 0.1 mol l⁻¹ TEA. Sections were hybridized overnight at 60°C using approximately 5×10⁵ cpm labeled probe per slide determined by liquid scintillation counting. The next day, slides were subjected to digestion with RNAse-A for 30 min at 37°C and high-stringency washes with 2×, 1×, 0.5× SSC for 10 min at room temperature and 0.1× SSC at 60°C for 30 min to remove non-specific probe binding. Finally, slides were dehydrated in serial alcohols and exposed to film autoradiography (Kodak, Rochester, NY, USA), with exposure for 2–4 weeks according to intensity of the hybridization signal.

Brain sections of all animals of a given experiment were treated and analyzed similarly together, to ensure identical conditions, and a person blind to the animal's experimental setup quantified the in situ hybridization signals.

For the non-radioactive in situ hybridization, photos were taken at 10× magnification using a Leica DMRB microscope (Leica Microsystems, Rueil-Malmaison, France) with an Olympus DP50 digital camera (Olympus France, Rungis, France), and digital images were captured using the camera software ViewfinderLite (Olympus), ensuring that all lighting parameters were consistent for all images. A background image was taken for each slide and subtracted from the labeling measure. All measurements were taken using ImageJ software (National Institutes of Health, Bethesda, MD, USA) according to an earlier published protocol (Sáenz de Miera et al., 2017). TSHβ and dio2 quantification was made using the ‘segmented line’ tool to cover the staining of TSHβ all along the PT and dio2 in the tanycyte cell bodies, on the paraventricular zone (PVZ). Three consecutive sections along the rostro-caudal axis were measured per animal. The mean gray value in arbitrary units for each animal was calculated as the average of the three measurements. kiss1 mRNA was quantified as the mean pixel gray level of staining in ‘elliptical’ sections of the same area for each neuronal cell body. A minimum of 20 neuronal cell bodies were quantified per animal in sections corresponding to the level of the mid ARC region. For rfrp mRNA, the total number of labeled neurons was hand-counted on the Leica DMRB microscope in all sections to cover the rostro-caudal hypothalamic expression of rfrp and given as number of positive neurons per animal.

For radioactive in situ hybridization, autoradiography films were scanned with an Epson 1640XL transmittance scanner at 720 dpi along with a calibrated series of optical density (OD) standards. OD measurements were analyzed for all genes using ImageJ software. The mean OD signal measured on all brain sections for a given gene is the OD value reported for each animal.

Testosterone levels were measure by ELISA using the Parameter Testosterone Assay (R&D Systems, Minneapolis, MN, USA), with a sensitivity of 0.041 ng ml⁻¹. FSH levels were measured by radioimmunoassay (RIA) by Dr Manuel Tena-Sempere (University of Córdoba, Spain) as previously described (García-Galiano et al., 2012). Melatonin levels were measured in individual pineal glands by RIA (Ribelayga et al., 1999).

All data are represented as means±s.e.m. of n animals per experimental group. Sample sizes in the study were based on our previous work looking at similar response variables in other seasonal rodents (Król et al., 2012; Sáenz de Miera et al., 2014, 2017), typically allowing us to resolve differences of 20% or more between means as statistically significant. All data were checked for normality using the Shapiro–Wilk test and homogeneity of variance using the F-test. When data were not normally distributed, they were transformed to fit this distribution and reanalyzed. Data were analyzed for photoperiodic differences by the two-tailed t-test, or Mann–Whitney test when data were not normally distributed. Two-way ANOVA analysis with photoperiod and sex as factors was used to analyze the photoperiodic differences in GSI and to assess a possible effect of sex on gene expression in P50 offspring. Pearson correlation analysis was used to evaluate the correlation between different genes. All statistical analyses were performed using PRISM, version 7 (GraphPad Software Inc., San Diego, CA, USA).

To analyze the effect of photoperiod on reproductive physiology and neuroendocrine gene expression, we kept adult male C57 and MSM mice in LP or transferred them to SP for 6 weeks prior to euthanization.

In adult MSM, we collected pineal glands to explore how melatonin production is affected by photoperiod. Pineal melatonin content was low during the light period and increased during the dark period in both photoperiodic states (Fig. 1A). In LP, a modest increase in average melatonin content was observed at ZT21, but no significant regression of the melatonin peak was observed (r²=0.86, P=0.20); therefore, peak duration could not be calculated. In SP, the averaged melatonin concentration showed a maximum concentration 6 h after lights off. The result of nonlinear regression analysis indicated a significant melatonin rhythm in SP (r²=0.99, P=0.015) of 6.56 h long with a peak onset estimate (φ₁) at ZT13.8 and an offset estimate (φ₂) at ZT20.4. The maximal mean nocturnal melatonin level measured was higher in SP (y_max=78.98±34.15 ng ml⁻¹) as compared with LP (y_max=41.99±20.74 ng ml⁻¹).

In adult male MSM, all of the reproductive parameters varied significantly between photoperiods (Fig. 1B–D), with significantly smaller testis mass (P<0.05, t-test; not shown), testis/body mass ratio (P<0.01, t-test; Fig. 1B), seminal vesicle mass (P<0.01, t-test; Fig. 1C) and testosterone concentration (P<0.01, t-test; Fig. 1D) in SP than in LP. Circulating FSH tended to exhibit lower levels in SP MSM mice as compared with LP (17.17±1.98 ng ml⁻¹ in LP versus 12.51±1.68 ng ml⁻¹ in SP; P=0.11, data not shown). Body mass remained unchanged between LP- and SP-exposed MSM mice (P=0.24, t-test; Fig. 1E). In C57 mice, all the physiological parameters measured showed no photoperiodic variations (LP versus SP, P>0.05; t-tests; Fig. 1B–E).

In MSM mice transferred from LP to SP, PT TSHβ expression (P<0.01, Mann–Whitney test; Fig. 2A) and tanycytic dio2 expression (P<0.001; t-test; Fig. 2B) significantly decreased by 5- and 3-fold, respectively. Furthermore, TSHβ and dio2 mRNA levels were strongly correlated in MSM mice (Pearson r=0.82, P<0.01). dio3 expression was hardly detectable in LP-exposed animals, but showed a large increase in SP-exposed animals (P<0.01, t-test; Fig. 2D). The number of rfrp-expressing neurons was markedly reduced after transfer from LP to SP (P<0.001, t-test; Fig. 2E), but kiss1 expression was not significantly different between LP and SP, albeit a tendency to be lower in SP (P=0.07, t-test; Fig. 2F). In C57 mice, transfer from LP to SP did not affect any of these investigated genes (Fig. 2).

Notably, in MSM mice only, strong correlations were observed between dio2 mRNA levels and the number of rfrp-expressing cells (MSM: Pearson r=0.93, P<0.001; C57: Pearson r=−0.16, P=0.64; Fig. 2G), as well as between the number of rfrp-expressing neurons and testicular size (MSM: Pearson r=−0.77, P<0.01; C57: Pearson r=−0.39, P=0.24; Fig. 2H).

To explore the effects of photoperiodic exposure in utero, pregnant MSM females were either maintained in LP or transferred to SP, and pups were further maintained in the gestational photoperiod until P50. At P50, female and male offspring gestated and raised in LP showed approximately 1.2-fold larger body mass than those kept in SP (P<0.001 for females, Mann–Whitney test; P<0.0001 for males, t-test; Fig. 3A). Reproductive organ mass was also larger in LP- as compared with SP-treated mice, in females showing 3-fold heavier reproductive organs, and in males showing 1.3-fold heavier testis mass in LP as compared with SP (P<0.0001 for uterus mass, P<0.001 for testis mass, t-tests; Fig. 3B). As a result, GSI analysis showed that the effect of SP exposure was more profound in females than in males (P<0.05 for interaction between sex and photoperiod, two-way ANOVA). In LP females, GSI was 2.5-fold larger than in SP females (P<0.001; post hoc Tukey test; Fig. 3C), whereas it was similar in males in both photoperiods (P=0.83, post hoc Tukey test). Although GSI was similar for males and females in LP (P=0.48, post hoc Tukey test), it was 2.6-fold smaller in females as compared with males in SP. Therefore, somatic and reproductive growth is affected by different photoperiodic exposure from gestation in MSM animals in both sexes, although there is a more profound effect on female reproductive organs.

In P50 offspring, no sex differences were found in any of the genes studied at the same photoperiodic state (not shown, two-way ANOVA, factor sex: TSHβ: P=0.13; dio2: P=0.41; dio3: P=0.77; rfrp: P=0.08); therefore, data from male and female offspring were combined to increase the power for the photoperiodic analysis (N=14). TSHβ gene expression in the PT was significantly higher in newborns and P50 offspring, and was higher also in dams exposed to LP as compared with SP (P<0.0001 for P1; P<0.001 for P50; t-tests; Fig. 4A,D). dio2 gene expression in the PVZ of the 3V and the median eminence was significantly higher in LP than in SP at P50 and P1, and was also higher in LP- than in SP-exposed dams (P<0.01 for P1; P<0.001 for P50; t-tests; Fig. 4B,D). dio3 gene expression in the PVZ was higher in dams exposed to SP and barely detectable in those exposed to LP during pregnancy. In contrast, dio3 gene expression was detected at very low levels with no significant photoperiodic differences at P1 (P=0.12, t-test), and was significantly higher in SP at P50 (P<0.05, Mann–Whitney test; Fig. 4C,D). Hence, TSHβ and dio2 expression reflected the photoperiod to which the mice were exposed in all groups, with a strong positive correlation observed between the levels of dio2 and TSHβ expression in all dams and P50 animals combined (Pearson r=0.75, P<0.001) and at P1 (Pearson r=0.67, P<0.01; Fig. 4E).

The number of rfrp-expressing neurons tended to decrease in dams transferred to SP as compared with those maintained in LP during pregnancy (Fig. 4F). At P50, animals gestated and raised in LP had more than two times more rfrp-expressing neurons than those raised in SP (P<0.0001, t-test; Fig. 4F). At this age, dio2 mRNA levels and the number of rfrp-expressing cells showed a positive correlation (Pearson r=0.75, P<0.01; Fig. 4G).

Seasonal rodents, in contrast to most mouse strains, show robust photoperiodic changes in gene expression in the neuroendocrine network comprising PT TSH and hypothalamic deiodinases, which in turn modulate the central regulation of the reproductive and metabolic axes. In the present study, we show that MSM mice exhibit photoperiodic regulation of reproductive organ and body mass, and testosterone levels in males, associated with corresponding changes in the TSH/dio network as well as RFRP expression. Furthermore, when exposure to different photoperiods in males and females starts from gestation, the reproductive and mass effects on offspring are more pronounced. These are the first results in M. musculus to show an integration of the photoperiodic signal on reproductive neuropeptides and gonadal development. Overall, the photoperiodic response observed in MSM mice appears comparable, albeit of lesser magnitude, to that in classical long-day breeders such as hamsters.

A 6 week exposure of male MSM mice to SP was sufficient to reduce testis size by approximately 15%, as well as reduce circulating testosterone and seminal vesicle mass, as compared with mice maintained in LP, indicating that MSM mice integrate photoperiod into the reproductive axis. Indeed, MSM mice exposed to SP showed a large nocturnal peak in pineal melatonin content that was markedly reduced under LP exposure, as previously observed in plasma (Kasahara et al., 2010). The large SP-induced melatonin production was accompanied by a robust decrease in PT TSHβ and tanycytic dio2 expression and an increase in tanycytic dio3 expression, consistent with the conserved photoperiodic response of these genes in vertebrates (Nakane and Yoshimura, 2014). These results are in agreement with those obtained in other melatonin-proficient mouse strains such as CBA or C3H, which display similar photoperiodic changes in melatonin rhythms (von Gall et al., 2000) mediating changes in expression of PT TSHβ and dio2/3 in hypothalamic tanycytes (Ono et al., 2008; Yasuo et al., 2009). Despite the different melatonin profiles in MSM mice, we have not verified whether the changes observed in physiology and gene expression in this study are indeed mediated by melatonin, as confirmed in other photoperiodic mammals. As previously observed, C57 mice that lack a melatonin rhythm and show overall low levels of melatonin production (Ebihara et al., 1986; Vivien-Roels et al., 1998) displayed no photoperiodic changes in TSHβ and dio2/3. However, C57 mice remain responsive to melatonin administration with similar changes in TSH and deiodinase genes (Ono et al., 2008).

Both DMH/VMH RFRP3 and ARC Kp neurons have been proposed to take part in the photoperiodic control of reproduction (Ancel et al., 2012; Caraty et al., 2007; Klosen et al., 2013; Rasri-Klosen et al., 2017; Revel et al., 2006a, 2008; Ubuka et al., 2012). Here, we observed that the number of rfrp-expressing neurons was markedly higher in LP- as compared with SP-exposed MSM mice, but not in C57 mice. In addition, photoperiodic changes in dio2 and rfrp expression were strongly correlated. Based on what has been consistently observed in other seasonal species (Klosen et al., 2013; Lomet et al., 2018; Revel et al., 2006b, 2008; Sáenz de Miera et al., 2014; Ubuka et al., 2012), this result suggests that the photoperiodic signal is transmitted to RFRP-expressing neurons via the melatonin-regulated TSH/dio network also in MSM mice. Interestingly, the level of rfrp expression was also correlated to testicular size in adult male MSM mice, further suggesting a regulatory role of RFRP3 in the photoperiodic changes in gonadal growth. Although photoperiodic changes in RFRP3 expression are highly conserved, the functional role of RFRP3 in seasonal reproduction is still discussed (Angelopoulou et al., 2019). In SP-adapted male Syrian (Ancel et al., 2012) and Siberian hamsters (Ubuka et al., 2012) and in male mice (Ancel et al., 2017), central RFRP3 delivery has been shown to stimulate luteinizing hormone (LH) secretion. In female rodents, however, acute injection of RFRP3 inhibits the LH surge (Ancel et al., 2017; Henningsen et al., 2017), whereas chronic RFRP3 is able to restore ovarian activity in SP-adapted female Syrian hamsters (Henningsen et al., 2017). On the contrary, ARC kiss1 expression was similar in both photoperiods in both MSM and C57 mice, despite a clear tendency to be lower in LP-kept MSM mice, probably reflecting the negative feedback effect of increased testosterone in these mice (Smith et al., 2005). Altogether, these data suggest that in male MSM mice, the TSH-driven photoperiodic change in dio2/3 regulates RFRP3 synthesis, which in turn may impact testicular activity. Further experiments are required to delineate whether the same regulation is observed in female MSM and whether RFRP3 is able to mediate the photoperiodic change in MSM reproduction.

Photoperiodic differences in maternal melatonin during gestation are known to impact the metabolic and reproductive development of Siberian hamster pups (Sáenz de Miera et al., 2017; Stetson et al., 1986). Here, we observed that when different photoperiodic exposure was applied on MSM mice from gestation to P50, male and female offspring exhibited a larger SP-induced reduction in reproductive organs, and a reduction in body mass, as compared with adult-only SP exposure. This suggests that maternal exposure to different photoperiods leads to a profound integration of the photoperiodic message into both the metabolic and the reproductive axes during development. Notably in females, prenatal SP exposure led to a reduction of approximately 60% in reproductive organ size as compared with prenatal LP exposure, possibly leading to an impairment of fertility. Maintenance of female C57 in SP from gestation has also been shown to delay the age of puberty through a decrease in body mass (Bohlen et al., 2018), highlighting that cues alternative to melatonin-mediated photoperiodism may contribute to seasonality under some circumstances.

SP exposure during pregnancy led to a large reduction in TSHβ and dio2 expression in dams and in newborns as compared with LP exposure, suggesting that the pups' photoperiodic system is responsive to maternal melatonin in utero, as we recently observed in Siberian hamsters (Sáenz de Miera et al., 2017). Further, the lower expression of these genes was maintained when the mice were kept in SP until P50. dio3 expression was markedly increased in the dams exposed to SP only during gestation, i.e. for 20–21 days, remained low with no photoperiodic difference at birth, and was slightly higher in SP again at P50. This large increase under short, but not prolonged SP exposure, suggests a transitory dio3 response to SP. The prediction is that dio3 expression would increase through mid-lactation in SP maintained mice, as observed in hamsters (Sáenz de Miera et al., 2017). Such a transitory regulation of dio3 expression is also observed in Siberian hamsters under decreasing natural photoperiod (Petri et al., 2016) or prolonged SP in laboratory settings (Milesi et al., 2017). These results suggest that the dio3 effect in depleting hypothalamic T3 is only temporarily required for the ‘winter’ transition in physiology (Sáenz de Miera, 2019). Alternatively, dio3 expression in MSM mice could be induced only after a transfer to a shorter photoperiod, and thus may not have happened in the mice born and raised in SP. These results do not allow us to distinguish between both possibilities, and further studies are required to discern the need of dio3 expression (i.e. of T3 local depletion) in the transition to winter physiology.

In line with the inhibitory effect of SP exposure from pregnancy to P50 on TSHβ and dio2 expression, the number of rfrp-expressing cells was more than two times lower in SP- as compared with LP-exposed mice. Interestingly, dams exposed to SP for only the duration of pregnancy showed a trend, but not a significant decrease in the number of rfrp-expressing neurons, confirming the slow dynamic of photoperiod-driven change in RFRP3 expression (Milesi et al., 2017; Revel et al., 2008).

Collectively, our results indicate that the TSHβ/DIOs/RFRP pathway is photoperiodically regulated in MSM mice, as observed in classical seasonal rodents. This regulation is associated with significant, albeit minor, changes in gonadal size, notably when the photoperiodic exposure starts in utero. Whether there is a photoperiodic change in hypothalamic T3, consequent to the dio2/dio3 switch, and whether this change by itself or through RFRP3 neurons is critical for the photoperiodic change in gonadal size observed in MSM mice remain to be established. The present results suggest that the ability to show photoperiodic responses in the reproductive axis is present in M. musculus. In most laboratory mouse strains, opportunistic strategies prevail, probably owing to artificial laboratory selection for breeding efficiency in animal facilities under a constant photoperiod or possibly reflecting their recent eco-evolutionary history as commensals to humans (Bronson, 1979).

The existence of a physiological endpoint reflecting the integration of the photoperiodic message highlights the value of these mice as models for genetic manipulations of this system. The use of genetic animal models has generated large advances in our knowledge of neuronal circuits, developmental processes and neuroendocrine integration (Elias, 2014). For the mammalian photoperiodic response, transgenic TSHr and mt1r knockout on a C57 background have served to establish the role of these receptors in the PT/MBH processing of the melatonin message (Ono et al., 2008; Yasuo et al., 2009). But owing to their lack of seasonal phenotype, genetic analysis of downstream neuroendocrine pathways and physiological consequences have been unexplored in these lines. Because crossing of transgenic lines onto an MSM background is readily feasible, it will now be possible to address this gap, and to take genetic approaches to analyze the downstream coupling of the PT/MBH axis.

The authors thank Béatrice Bothorel for her help on the melatonin regression analysis and Clarisse Quignon for her help with the microscope images. The authors thank the National Institute of Genetics, Japan, for providing the MSM mice used in this study.