Binary Switching of Calendar Cells in the Pituitary Defines the Phase of the Circannual Cycle in Mammals

Abstract

Persistent free-running circannual (approximately year-long) rhythms have evolved in animals to regulate hormone cycles, drive metabolic rhythms (including hibernation), and time annual reproduction. Recent studies have defined the photoperiodic input to this rhythm, wherein melatonin acts on thyrotroph cells of the pituitary pars tuberalis (PT), leading to seasonal changes in the control of thyroid hormone metabolism in the hypothalamus. However, seasonal rhythms persist in constant conditions in many species in the absence of a changing photoperiod signal, leading to the generation of circannual cycles. It is not known which cells, tissues, and pathways generate these remarkable long-term rhythmic processes. We show that individual PT thyrotrophs can be in one of two binary states reflecting either a long (EYA3(+)) or short (CHGA(+)) photoperiod, with the relative proportion in each state defining the phase of the circannual cycle. We also show that a morphogenic cycle driven by the PT leads to extensive re-modeling of the PT and hypothalamus over the circannual cycle. We propose that the PT may employ a recapitulated developmental pathway to drive changes in morphology of tissues and cells. Our data are consistent with the hypothesis that the circannual timer may reside within the PT thyrotroph and is encoded by a binary switch timing mechanism, which may regulate the generation of circannual neuroendocrine rhythms, leading to dynamic re-modeling of the hypothalamic interface. In summary, the PT-ventral hypothalamus now appears to be a prime structure involved in long-term rhythm generation.

Graphical abstract

Figure 1

Co-localization of EYA3 and TSHβ to the PT Thyotroph on LPs and Identification of CHGA as a SP Marker (A) Diagram of the photoperiodic treatment for experiment 1. Short photoperiod (SP): 8 hr light, 16 hr dark; long photoperiod (LP), 16 hr light, 8 hr dark. Tissues were collected at 4 and 12 weeks in SPs and LPs, in the mid-light phase, zeitgeber time (ZT; time after lights on): ZT4 in SPs and ZT8 in LPs. Collection points are represented by green arrows. The red line illustrates the natural photoperiod, and the blue lines represent the photoperiod imposed in light-controlled rooms (Figure S1A). (B) Triple immunofluorescence showing expression of αGSU (red), TSHβ (green), and EYA3 (blue) in the PT on SPs and LPs (Figure S2). Scale bars, 20 μm. Quantification of EYA3 and TSHβ co-expression (Table S1 and Figure S3B) is shown as a schematic representing the variety of phenotypes that αGSU-expressing cells show in response to LPs. (C) Triple immunofluorescence showing expression of EYA3 (red), CHGA (yellow), and TSHβ (green) in the PT on SPs and LPs (blue, DAPI) (Figure S3C and Table S3). Scale bars, 50 μm. The white arrow shows a cell co-expressing EYA3, CHGA, and TSHβ; this is one of two cells found in over 17,000 that co-express all three proteins. See also Figures S1–S3.

Figure 2

Binary Cell-Based Timing Mechanism over the Circannual Cycle (A) Prolactin concentrations in plasma for 32 animals in SPs for 12 weeks transitioning into LPs for 4 weeks (red arrow marks this). The first double line in the graph indicates the gap in weeks between sampling during SPs, and the second double line represents a change in the cohort of animals sampled (as the previous group culled at 4 weeks). The dip in prolactin concentration between 13 to 18 weeks may be related to wool shearing (indicated on the graph by a bar spanning the affected period). The percentage of animals that show three consecutive weeks of suppressed prolactin (LP refractory) is shown. Error bars represent the SEM. (B) In situ hybridization and quantification for TSHβ and EYA3 mRNA at SP4, SP12, LP4, LP16, and LP29. Representative images are shown (n = 3). Error bars represent the SD. (C) Triple immunohistochemistry showing protein expression of EYA3 (red), TSHβ (green), CHGA (yellow), and DAPI (blue) at LP29 in three individuals (Figures S3C and S4A and Table S4). Scale bars, 50 μm. See also Figures S2–S4.

Figure 3

Transcriptional Response to Photoperiod and Histogenesis (A) A simplified network of related statistically significant GO terms using the Cytoscape add-on ClueGO [30, 31]. The network comparing SP4 versus LP4 is shown. The filled colored circles (nodes) represent each statistically significant parent GO term. The lines (edges) between the nodes show that there are overlapping genes within each term. The colored ovals group these parent GO terms into more generic functional descriptions (Figure S4B). (B) Graphs of average TMM normalized read counts per million (CPM) [32] from RNA-seq data from SP4, SP16, LP4, and LP22 (experiment 4). KAL1, SEMA3D, SHH, and WIF1 are within the following GO term categories: development, morphogenesis, and differentiation. GPR56 and NCAN are related to cell communication, and DNAH8 and COL2A1 are related to cell movement. False discovery rate (FDR)-corrected p values calculated by EdgeR are indicated as follows: ^∗p < 0.05, ^∗∗p < 0.005, ^∗∗∗p < 0.0005, and ^∗∗∗∗p < 0.0001 (Figures S4C and S5A). (C) Detection of dividing cells by phospho-histone H3 (p-histone3) and hematoxylin staining in the sheep pars tuberalis (PT) and pars nervosa (PN) under SP and LP at days 1, 7, and 28. Arrows show dividing cells. Scale bars, 50 μm. Double immunofluorescence staining of αGSU (green), phospho-histone H3 (red), and DAPI (blue) in ovine PT is shown (Figures S5B and S6). See also Figures S4–S6.

Figure 4

Cellular Remodeling in the PT (A) EM images for LP4/SP 4 and LP29. FS, folliculostellate cells (green); T, thyrotroph (pink). Arrows indicate intercellular junctions. n = 3. Representative images are shown. Quantification of cell contacts is shown as follows: (1) thyrotroph/FS contacts, (2) FS/FS contacts, and (3) thyrotroph/thyrotroph contacts. Cell contacts were identified on the basis of electron dense morphology between cells at the plasma membrane. The morphology represents a potential mix of junctions—zona adherens, desmosomes, and gap junctions. One-way ANOVA was performed with multiple testing corrections, with adjusted p values as follows: ^∗p < 0.05, ^∗∗p < 0.005, ^∗∗∗p < 0.0005, and ^∗∗∗∗p < 0.0001; n = 3. Error bars represent the SEM. (B) EM images of a PT thyrotroph at LP12, SP12, and LP29. Quantification of cell size is shown (μm²). One-way ANOVA was performed with multiple testing corrections, with adjusted p values as follows: ^∗p < 0.05, ^∗∗p < 0.005, ^∗∗∗p < 0.0005, and ^∗∗∗∗p < 0.0001; n = 3. Error bars represent the SEM. (C) Rough ER (RER) and Golgi false colored red and quantification of RER/Golgi. One-way ANOVA was performed with multiple testing corrections, with adjusted p values as follows: ^∗p < 0.05, ^∗∗p < 0.005, ^∗∗∗p < 0.0005, and ^∗∗∗∗p < 0.0001; n = 3. Error bars represent the SEM. (D) Dark spots show the granules present and quantification of secretory areal granule density. In all cases, SP4, SP12, LP4, LP12, and LP29 are presented with n = 3; representative images are shown. One-way ANOVA was performed with multiple testing corrections, with adjusted p values as follows: ^∗p < 0.05, ^∗∗p < 0.005, ^∗∗∗p < 0.0005, and ^∗∗∗∗p < 0.0001. Error bars represent the SEM. See also Figure S5.

Figure 5

Remodeling of the Neural-Glial Interface of the Median Eminence (A) Vimentin immunostaining for tanycytes (brown) of coronal section of the sheep mediobasal hypothalamus (top). Scale bars, 100 μm and 20 μm, respectively. PT, pars tuberalis; Me, median eminence; 3V, third ventricle; HYP, hypothalamus. 3D render series of IHC images showing GnRH (red), vimentin (green), and DAPI (blue) in SPs and LPs are also shown (bottom). Scale bar, 50 μm. (B) EM images of ME nerve terminals (pink) and tanycytes (green) in SP4, SP12, LP4, LP12, and LP29. pv, perivascular space; cap, capillary. Scale bar, 200 nm. n = 3; representative images are shown. (C) Quantification of the (1) percentage area of glial process (tanycytic end foot) within the ME, (2) mean distance of the nerve terminal from the basal lamina (μm), and (3) percentage of nerve terminals in contact with basal lamina. One-way ANOVA was performed with multiple testing corrections, with adjusted p values as follows: ^∗p < 0.05, ^∗∗p < 0.005, ^∗∗∗p < 0.0005, and ^∗∗∗∗p < 0.0001; n = 3. Error bars represent the SEM. See also Figure S1.

Figure 6

Summary of the Changes in the PT and Median Eminence throughout a Circannual Cycle The model proposes that an endogenous timer switches EYA3 expression in the PT thyrotroph cells, driving TSH and hypothalamic thyroid hormone metabolism independently of melatonin. Individual PT thyrotroph cells are either in a long (EYA3⁺) or short (CHGA⁺) state, and the relative proportion of these binary-state cells determines the phase of the circannual cycle. Re-modeling of the morphology of the PT sees changes in cell size and RER. Within the PT, networks of either thyrotrophs (LP) or FS cells (SP) form. Re-modeling of the hypothalamic interface in the ME leads to encasement of neuronal synapses by tanycyte end feet in the non-breeding season (LP), suggesting a physical mechanism for control of GnRH secretion. Collectively, this suggests that the PT thyrotroph operates as a calendar cell, generating long-term neuroendocrine rhythms in both the hypothalamus and pituitary gland.