Glis3 as a critical regulator of Pit1-lineages and renal functions

Abstract

Abstract

The Krüppel-like GLIS3 transcription factor acts as an effector on the sonic hedgehog (Shh) pathway, regulating several biological processes and the development and postnatal function of various tissues. Given the major role of Shh signaling during the differentiation and anterior/posterior patterning of the adenohypophysis (AH) anlage, we investigated the potential role of glis3 during AH development in the zebrafish model. Glis3KD embryos exhibited increased expression of early AH-inductive genes lim3 and nkx2.2a, along with an expansion of pit1-positive precursors. This shift led to an overproduction of lactotropes, somatolactotropes, and thyrotropes, at the expense of somatotropes, corticotropes, and melanotropes. The most striking difference is the hyperprolactinemia observed in glis3KD larvae, with transcript and protein levels increased by approximately 30-fold and threefold, respectively, compared to controls. As a consequence of the primary role for prolactin in controlling fish osmoregulation, glis3KD larvae exhibited the upregulation of ionocytes expressed in gills and pronephric ducts. Furthermore, glis3KD larvae presented abnormal pronephric primary cilia and glomerular cysts in keeping with the established role of GLIS3 in ciliopathies and polycystic kidney disease. In conclusion, glis3 action appears fundamental to set an adequate number of pit1-precursors and renal function.

Key messages

• Glis3 regulates pituitary cell fate by balancing pit1-positive precursors.

• Glis3 knockdown leads to hyperprolactinemia and altered osmoregulatory mechanisms.

• Glis3 is essential for normal cilia structure and pronephric kidney function.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00109-025-02611-3.

Introduction

The pituitary gland in zebrafish consists of the neurohypophysis (NH), derived from the ventral diencephalon, and the adenohypophysis (AH), the endocrine component from the endodermal compartment [1, 2]. Hormone-producing cells of the AH are crucial regulators of embryonic growth and postnatal development, metabolism, energy homeostasis, and osmoregulation [1–4].

AH development begins during early segmentation (10–16 hpf) when its precursors emerge in the anterior neural ridge (ANR) of the pre-placodal ectoderm (PPE), adjacent to the prospective lens and olfactory placodes [5, 6]. Inductive signals from the neuroectoderm, including Shh, Fgfs, and Bmps, induce the dynamic expression of pitx3, eya1, six1a, dlx3b, and dlx4b from the ANR precursors that will differentiate into adenohypophyseal, lens, or olfactory cells [3, 7]. Signals of the Shh pathways were reported to play a pivotal role during the early steps of AH induction in various species [2, 6, 8, 9]. In zebrafish, Shh appears to be required to restrict the area of ppe that will give rise to AH anlage at the expense of lenses and olfactory placodes [6, 9, 10].

By 18–24 hpf, the spatial arrangement of the horse-shaped AH anlage and the ventral diencephalon results in the medial region of AH receiving a high concentration of Shh, promoting nkx2.2a expression, whereas the lateral sides, receiving less Shh levels, express pax7 [9]. A subset of medial lim3 + cells sequentially express prop1 and pit1, forming precursors for lactotropes, somatolactotropes, thyrotropes, and somatotropes, while pax7 + lateral cells give rise to corticotropes and melanotropes [1, 9, 11].

As the forebrain folds and the oral cavity forms (~ 26 hpf), the AH anlage converges at the midline, defining the anterior (pars distalis, PD) and posterior (pars intermedia, PI) domains. The first endocrine AH cells, lactotropes, corticotropes/melanotropes, and somatolactotropes, begin differentiating at 24 hpf, followed by thyrotropes and somatotropes (36–48 hpf). By 60 hpf, AH differentiation is complete, with distinct regional organization: lactotropes and corticotropes in anterior PD, somatotropes and thyrotropes in posterior PD, and somatolactotropes/melanotropes in PI [2, 3, 7, 12].

Gli-mediated Shh signaling is crucial for the functional patterning of AH. In zebrafish, gli1 mutation causes defects in pituitary patterning and cell-type specification [9, 13], while gli2 mutant displays anterior pituitary hypoplasia accompanied by abnormal ventral CNS patterning [13, 14].

Considering the established role of glis3 in the Shh pathway during endocrine organ development, we sought to determine whether glis3 knockdown (KD) impacts zebrafish AH development and function. To this end, we provide a schematic representation of the multistep development of AH, emphasizing the shha-gli signaling and the potential contribution of glis3 in this (Fig. 1).

Fig. 1.

Contribution of Shh signaling during adenohypophyseal (AH) development. The AH anlage becomes visible around 10 hpf at the level of the anterior neural ridge (anr) within the pre-placodal ectoderm (ppe). Shh–Gli signalling, and potentially Shh–Glis3, originating from the ventral neuroectoderm (ne), is required to induce anr cells to differentiate into ah precursors rather than lens or olfactory cells [6, 9]. By 18 hpf, the ah anlage proliferates and acquires the characteristic horseshoe-shaped structure, located ventro-rostrally to the hypothalamus (hy) and aligned with Shh signalling from the ventral diencephalon. Shh induces the expression of lim3 throughout the placode, while it promotes nkx2.2 specifically in the medial region of the ah anlage; in contrast, the lateral regions exposed to lower Shh levels express pax7 [9, 14–16]. From 24 hpf onward, a subset of lim3-positive cells begins to express pit1, which drives differentiation into lactotropes, somatolactotropes, thyrotropes, and somatotropes (26–60 hpf). Meanwhile, pax7-positive cells express pomc, the precursor giving rise to corticotropes and melanotropes [3, 11]

Methods

Zebrafish husbandry and maintenance

Wild-type AB zebrafish were maintained at 28 °C under a 14:10 light/dark cycle. Embryos from natural spawning were staged using established morphological criteria and incubated in fish water with 0.002% 1-phenyl-2-thiourea (PTU; Sigma-Aldrich, P7629) to block pigmentation and 0.01% methylene blue to prevent fungal growth (Sigma-Aldrich, C.I. 52015).

Morpholino microinjection and rescue experiment

Control and glis3 knockdown (glis3KD) embryos were generated by microinjecting 0.5 ng/embryo of standard control morpholino (stdMO, 5′-CCTCTTACCTCAGTTACAATTTATA-3′; GeneTools) or 0.5 ng/embryo of glis3MO_SPL (5′-TTCTTGTTTTTACCTTTCATACCGC-3′; GeneTools), as described previously [17]. Rescue experiments were performed by co-injecting 0.5 ng/embryo of glis3MO_SPL with 100 pg/embryo of glis3 CDS mRNA [17]. A second splice-blocking morpholino (glis3MO_SPL2, 5′-ACCTGCTGCAAGAGATCAGTTAAAA-3′) has been used to confirm the results on kidney function obtained with the first one [17].

Quantitative real-time PCR (qRT-PCR)

Total RNA was extracted from pools of 30–40 embryos per developmental stage using TRIzol Reagent (ThermoFisher, 15596026). Complementary DNA (cDNA) synthesis was performed using GoScript Reverse Transcription System (Promega, A5003). Quantitative real-time PCR (qRT-PCR) was conducted on an ABI PRISM 7900HT Fast Real-Time PCR System using SYBR Green Master Mix (ThermoFisher, 4309155). Eef1α1 was used as an endogenous control. All experiments were performed in triplicate, and results are presented as mean ± standard deviation (SD). qRT-PCR primer sequences used in this study are provided in Tab. S1.

Whole-mount in situ hybridization (WISH) and fluorescent WISH (FISH)

WISH was performed with DIG-labeled riboprobes as described [18], and primers are listed in Tab. S2. Staining used 0.35 mg/ml NBT and 0.27 mg/ml BCIP (Promega, S3771) in SB9.5 buffer (0.1 M Tris–HCl, pH 9.2, 50 mM MgCl2, 100 mM NaCl, 0.1% Tween-20). Embryos were post-fixed in 4% paraformaldehyde, mounted in 80% glycerol/PBT, and imaged on a Leica M205FA with a DFC450FC camera. Each marker was analyzed in 20 embryos from at least three independent injections. Signal area was measured in Fiji Software using a fixed ROI.

FISH used 0.25 mg/ml Fast Blue BB (Sigma, F3378) and 0.25 mg/ml NAMP (Sigma, N5000) in SB8.2 buffer (0.1 M Tris–HCl, pH 8.2, 50 mM MgCl2, 100 mM NaCl, 0.1% Tween-20). For confocal imaging, 15 dissected heads were mounted ventrally and imaged with a Nikon Eclipse Ti (20 ×, 648 nm laser). Cell counts were done manually from z-stacks, and cell volumes (µm³) were quantified with Volocity Software using a consistent ROI.

Immunofluorescence (IF)

Larvae at 120 hpf were fixed in 4% PFA at 4 °C overnight and immunostained as described [19]. Prolactin was detected using mouse anti-prolactin (1:500, BioSB, PRL02) and Alexa Fluor 488-goat anti-mouse (1:1000, ThermoFisher, A32723). Cilia were labeled with anti-acetylated tubulin (1:250, Sigma, T7451) and Alexa Fluor 555-goat anti-mouse (1:1000, ThermoFisher, A32727).

Embryos and larvae were mounted in 80% glycerol/PBT and imaged with a Leica M205FA using GFP and TRX filters. Prolactin MFI was quantified in Fiji (fixed ROI). Triplicate experiments used 15 embryos/condition from three injections.

Prolactin quantification (ELISA assay)

Prolactin levels (ng/ml) were measured using a Fish PRL ELISA Kit (CusaBio, CSB-E12695Fh). Control and glis3KD larvae were anesthetized with 0.03% tricaine (MS-222, Pentair, TRS2), rinsed in lysis buffer, and disaggregated by pipetting. Standards and triplicate samples (50 µl) were loaded into ELISA plates with anti-PRL and HRP-conjugated antibodies. Absorbance at 450 nm was read on a VictorNivo plate reader (PerkinElmer), and concentrations were calculated with CurveExpert. Experiments were done in triplicate using pools of 20 larvae from at least six independent injections.

Chemical treatments

Hypothyroid larvae were generated by culturing embryos from 3 to 120 hpf in fish water (FW; 0.01 mg Instant Ocean, 0.019 mg MgCl2, 0.01 mg NaCl in ddH2O) supplemented with 0.6% propylthiouracil (Merck, PHR2609). Triiodothyronine (T3, BioTechne, 6666/50) was prepared in DMSO (20 mg/ml), diluted to 20 nM, and administered from 6 hpf.

D-Mannitol 98% (Merck, M425-10MG) at a final concentration of 150 nM was added to fish water (FW) from the 48 hpf onwards and daily replaced until the desired developmental stage.

Analysis of cardiovascular function

Embryos at 48 and 120 hpf (also referred to as 2 and 5 days post-fertilization, dpf) were anesthetized with 0.003% Tricaine (Merck, A5040), photographed under brightfield illumination, and subjected to 16-s video recordings of cardiac activity. Heart rate (HR) was measured manually by counting beats per minute.

Renal clearance assay

Embryos at 72 hpf were anesthetized with 0.03% tricaine, oriented in agarose casts, and microinjected into the Duct of Cuvier with 6 µl of a solution containing 5 mg/ml rhodamine B-dextran (Sigma-Aldrich, R9379) and 2 mg/ml DAPI (Sigma-Aldrich, D9542). Fluorescence was recorded at T0 (15 min post-injection), T1 (+ 24 h), and T2 (+ 48 h) using a Leica M205FA stereomicroscope with UV and TRX filters.

Statistical analysis

Statistical analyses were conducted using PRISM 10.0 (GraphPad Software). Normality was assessed using the Shapiro–Wilk test. Statistical tests were selected based on data distribution, with the unpaired Student’s t-test for normally distributed data or the Mann–Whitney test for non-normally distributed data (significance: ns, not significant P > 0.05, P < 0.05 (*), P < 0.01 (**), P < 0.001 (***)).

Results

Specification of adenohypophyseal placode

We first analyzed shha expression in control and glis3KD embryos during early (12 hpf), mid (16 hpf), and late (19 hpf) somitogenesis. No differences were observed between control and glis3KD embryos at 12 and 16 hpf (Fig. 2a–b’). Consistent with our previous findings describing the appearance of glis3 from the late somitogenesis stages [20], glis3KD embryos at 19 hpf showed loss of glis3 signal in the brain, but increased shha expression in the ventral and dorsal diencephalon and notochord (Fig. 2c–d’ and e, e’).

Fig. 2.

Induction of AH anlage in glis3KD embryos. a–d’ Whole-mount in situ hybridization (WISH) for shha in control (CTRL) and glis3 knockdown (KD) embryos at 12 hpf (a, a’), 16 hpf (b, b’), and 19 hpf (c–d’). Shha expression is detected in the ventral diencephalon (vD, arrowhead), dorsal diencephalon (dD, arrowhead), and notochord (nt, arrowhead). e, e’ WISH for glis3 in CTRL and KD embryos at 19 hpf. Embryos are shown in dorsal view with the head oriented to the top (a–c’, f–j’) or in lateral view with the head to the left (d–e’). Scale bars: a, d, e = 250 µm. f–j’ WISH analysis of early AH-inductive genes at 24 hpf in CTRL and KD embryos: nkx2.2(f, f’), lim3(g, g’), prop1(h, h’), pit1(i, i’), and pax7(j, j’). Scale bar: f = 150 µm. k Quantification of WISH signals (pixel number) in CTRL and KD embryos at 24 hpf. Data are represented as dot plots (n = 7 embryos per group) with mean ± SD. Statistical analysis was performed using the Mann–Whitney test: ***P < 0.001. l Relative mRNA expression of AH-inductive genes in pools of CTRL, KD, and Rescue embryos at 24 hpf, normalized to eef1α. Data are presented as mean ± SD of three independent replicates, each consisting of 30 embryos. Statistical analysis was performed using Student’s t-test: (ns) not significant; ***P < 0.001; ***P < 0.001

At 16 hpf, the expression of early AH-inductive genes, pitx3, eya1, six1a, and dlx3b, expressed in the anterior neural ridge (ANR), was comparable among glis3KD, control, and rescue embryos (Fig. S1). However, by 24 hpf, glis3KD embryos exhibited increased expression of the key AH precursor genes, nkx2.2a, lim3, prop1, and pit1, whereas pax7 was markedly reduced in the pituitary anlage of glis3KD embryos (Fig. 2f–j’, k).

Real-time PCR analysis confirmed the differences detected by WISH, while rescue experiments excluded potential off-target effects on pituitary development. Specifically, co-injection of glis3 morpholino with glis3 mRNA restored normal pituitary gene expression in 24 hpf embryos (Fig. 2l). Notably, glis3KD embryos at 24 hpf did not exhibit major compensatory changes in the expression of other gli (gli1, gli2, gli3) or glis (glis1a, glis1b, glis2) family members at the whole-mount level (Fig. S2).

The loss of glis3, concomitant with elevated Shh levels in the diencephalon, appears to promote upregulation of Shh-responsive genes such as nkx2.2 and lim3, while suppressing pax7, thereby resulting in increased pit1 expression.

Patterning of adenohypophyseal cells

Quantification of pit1 mRNA in pooled embryos at 26, 30, and 33 hpf revealed a 2-, 2.5-, and 5.8-fold increase, respectively, in glis3KD embryos compared to controls, with no significant differences observed between rescue and control embryos (Fig. 3a). Detailed analysis of the pit1 placode indicated a 36–60% increase in cell volume (µm³) in glis3KD embryos compared to controls at these stages (Fig. 3b–e).

Fig. 3.

Patterning of AH cell types in glis3KD embryos. a Relative mRNA expression of pit1 in pools of CTRL, glis3KD, and Rescue embryos at 26, 30, and 33 hpf. b–d’ Representative images of pit1 analyzed by fluorescent in situ hybridization (FISH) in CT and KD embryos at 26 (b, b’), 30 (c, c’), and 33 (d, d’) hpf. e Quantification of total pit1-volume (µm³), using confocal analysis on the same area (ROI) and number of sections for each embryo. Data are displayed as dot plots with Mean ± SD of n = 15 embryos each. Mann–Whitney test was used for statistical analysis, ***P < 0.001. f Relative mRNA expression of AH cell markers in CTRL, glis3KD, and rescue embryos at 36 hpf. g–k’ Representative confocal FISH images of prl (g, g’), smtla (h, h’), smtlb (i, i’), gsua (j, j’), and pomca (k, k’) in CT and KD embryos at 36 hpf. l Relative expression of AH cell markers in pools of CTLR, glis3KD, and rescue embryos at 60 hpf. n–t’ FISH of prl (n, n’), smtla (o, o’), smtlb (p, p’), gsua (q, q’), tshba (r, r’), gh (s, s’), and pomca (t, t’) in CT and KD embryos at 60 hpf. All qRT-PCR data are normalized to eef1α, and the results are shown as mean ± SD of n = 3 independent replicates. Each data point represents a pool of 30 embryos. Student-t-test was used for statistical analysis between groups: *P < 0.05; **P < 0.01; ***P < 0.001. All FISH embryos were acquired by confocal microscopy in ventral view, head to the top. Scale bars in a, b, and g  = 100 µm. m Schematic drawing summarizing the differentiation of AH cell type of controls and glis3KD, showing the posterior expansion of lactotropes, the anterior spread of somatolactotropes and thyrotropes, as well as the reduction of corticotropes/melanotropes and somatotropes. A-P, antero-posterior axis; aPD, anterior pars distalis; pPD, posterior pars distalis; PI, pars intermedia; L, lactotropes; SL, somatolactotropes; T, thyrotropes; S, somatotropes; C/M, corticotropes/melanotropes

At 36 hpf, the mRNA levels of prl (lactotropes), smtla/smtlb (somatolactotropes), and gsuα (thyrotropes and gonadotropes) were significantly elevated in glis3KD embryos compared to both controls and rescue embryos. Conversely, the expression of pomca, a marker for corticotropes and melanotropes, was reduced in glis3KD embryos (Fig. 3f). The average numbers of distinguishable AH cell types at 36 hpf, analyzed by FISH, also differed significantly between glis3KD embryos and controls: prl (13.7 ± 1.49 vs. 7.7 ± 0.73), smtla (9.85 ± 0.75 vs. 5.45 ± 0.60), smtlb (12.15 ± 1.14 vs. 7.75 ± 0.72), gsuα (11.15 ± 1.18 vs. 8.15 ± 0.81), and pomca (7.0 ± 0.79 vs. 16.7 ± 0.98) (Fig. 3g–k’).

By 60 hpf, when all AH cell types are fully differentiated, the transcript levels of prl, smtla/smtlb, gsuα, and tshba were significantly higher in glis3KD embryos compared to both controls and rescue embryos. In contrast, the expression of pomca and the somatotroph gene gh was markedly reduced or nearly undetectable in glis3KD embryos (Fig. 3l). FISH analysis of differentiated AH cell types revealed a 92% increase in the average number of lactotropes in glis3KD embryos compared to controls (21.2 ± 1.94 vs. 10.95 ± 1.32 prl-positive cells). These cells were scattered along both the anterior and posterior pituitary domains, in contrast to their organized distribution within the anterior pituitary in controls (Fig. 3n, n’). Similarly, a number of somatolactotropes positive for smtla and smtlb were increased by 42% (13.9 ± 0.85 vs. 9.75 ± 0.85) and 88% (14.2 ± 1.44 vs. 7.55 ± 0.60), respectively, with a shift along the anterior pituitary domain in glis3KD embryos (Fig. 3o–p’). Thyrotropes marked by gsuα and tshba were also increased by 71% (11.09 ± 1.25 vs. 6.95 ± 0.83) and 66% (13.25 ± 1.52 vs. 7.95 ± 0.60), respectively, invading the anterior pituitary domain (Fig. 3q–r’).

Finally, FISH analysis confirmed a significant reduction in both pomca and gh cell numbers in glis3KD embryos compared to controls, with decreases of 67% (7.5 ± 1.93 vs. 22.4 ± 1.19 pomca-positive cells) and 90% (0.35 ± 0.49 vs. 3.35 ± 0.59 gh-positive cells), respectively (Fig. 3s–t’).

All AH cell counts and associated statistical analyses are provided in Tab. S3 and Tab. S4.

Together, these results indicate that the loss of glis3 skews adenohypophyseal differentiation toward pit1-dependent lineages, particularly lactotropes, somatolactotropes, and thyrotropes, while markedly reducing pit1-independent lineages such as corticotropes and melanotropes (Fig. 3m).

Hyperprolactinemia induced by glis3-deficiency

The most prominent effect of glis3 deficiency was the over-differentiation of lactotropes. At the larval stage, glis3 deficiency resulted in a dramatic elevation of prolactin, with transcript levels increasing ~ 30-fold and protein levels ~ threefold compared to controls (Fig. 4a, b). FISH analysis confirmed a ~ threefold increase in cell volume (Fig. 4c, d) and a significant rise in lactotrope number in glis3KD larvae (prolactin-positive cells, 32.59 ± 3.63 vs. 7.76 ± 0.85 in controls; Tab. S5). Immunolocalization of prolactin revealed comparable GFP signal intensity in the pectoral fins and forebrain nuclei across both groups; however, signal intensity was markedly elevated in the gills, pronephros, and AH of glis3KD larvae (Fig. 4e–g).

Fig. 4.

Hyperprolactinemia of glis3KD larvae. a Relative mRNA expression of prl CTRL, glis3KD larvae at 120 hpf. qRT-PCR data are normalized to eef1α, and the results are shown as mean ± SD of n = 3 independent replicates. Each data point represents a pool of 20 larvae. Student-t-test was used for statistical analysis between groups: ***P < 0.001. b ELISA of prolactin (ng/ml) in control and glis3KD larvae at 120 hpf. Results (mean ± SD of 3 pools, 50 larvae/pool) were analyzed by Student’s t-test: ***P < 0.001. c, c’ Representative FISH images of prl in CT and KD larvae at 120 hpf. All larvae were acquired by confocal microscopy in ventral view, head to the top. Scale bars in c = 100 µm. d Quantification of total cell volume (µm³) of prl using confocal analysis on the same area (ROI) and number of sections for each embryo. Data are displayed as dot plots with mean ± SD of n = 15 embryos each. Mann–Whitney test was used for statistical analysis, ***P < 0.001. e–f’ IF of prolactin in control and glis3KD larvae in lateral (e, e’) or ventral (f, f’) view. Prl localized in gills (g, asterisks), pectoral fins (pf, arrow), pronephric ducts (pd, arrowheads), forebrain (fb, arrow), and adenohypophysis (ah, arrowhead). Scale bars: e = 300 µm; f = 200 µm. g Mean fluorescence intensity (MFI) of prolactin-positive tissues (n = 15) using Fiji Software. Mann–Whitney test: ns, not significant; ***P < 0.001

To determine whether hyperprolactinemia resulted from factors beyond lactotropes’ expansion, we assessed the expression of known regulators of prolactin synthesis. In glis3KD larvae, thyrotropin-releasing hormone (trh) and tyrosine hydroxylase (th) expression levels were significantly elevated (+ 3.71-fold and + 7.09-fold, respectively), whereas prolactin-releasing hormone (prlrh) expression was preserved (Fig. 5a). Correspondingly, the volume of trh- and th-expressing hypothalamic neurons was approximately doubled in glis3KD larvae (Fig. 5b–d).

Fig. 5.

Hypothalamic regulation of prolactin. a Relative mRNA expression of prlrh, trh, and th in controls and glis3KD larvae. qRT-PCR data are normalized to eef1α, and the results are shown as mean ± SD of n = 3 independent replicates. Each data point represents a pool of 20 larvae. Student’s t-test was used for statistical analysis between groups: ns, not significant; **P < 0.01; ***P < 0.001. b–c’ Representative FISH images of trh (b, b’) and th (c, c’) in CT and KD larvae at 120 hpf. All larvae were acquired by confocal microscopy in ventral view, head to the top. Scale bar in b = 100 μm. d Quantification of total cell volume (µm³) confocal analysis on the same area (ROI) and number of sections for each embryo. Data are displayed as dot plots with mean ± SD of n = 15 embryos each. Mann–Whitney test was used for statistical analysis, ***P < 0.001. e Relative mRNA expression of prl, trh, and th in controls, glis3KD larvae, and hypo-PTU larvae in basal condition (1.2% DMSO, gray columns) or upon TH treatment (20 nM T3, light blue columns). All qRT-PCR data are normalized to eef1α, and the results are shown as mean ± SD of n = 3 independent replicates. Each data point represents a pool of 20 larvae. Student’s t-test was used for statistical analysis between CTRL vs glis3KD or PTU (#); 1.2% DMSO vs 20 nM T3 (*)

We previously showed that glis3 is essential for thyroid development, and its silencing induces primary hypothyroidism in zebrafish embryos [17]. The hypothyroid state likely contributes to elevated trh expression, which may in turn promote prolactin synthesis independently of lactotropes’ over-differentiation.

To disentangle these mechanisms, we compared prl, trh, and th expression in glis3KD, hypothyroid (PTU-treated), and control larvae, with and without high-dose T3 (20 nM) exposure from 6 hpf. Under basal conditions, prl expression was significantly higher in glis3KD larvae compared to both PTU-treated (+ sevenfold) and control larvae (+ 25-fold). Notably, trh expression was similarly elevated in glis3KD (+ 2.98-fold) and PTU-treated (+ 2.62-fold) larvae relative to controls. Likewise, th expression increased in both groups (+ 3.85-fold in glis3KD; + 1.85-fold in PTU), reflecting elevated prolactin levels (Fig. 5e).

Following T3 treatment, trh expression was suppressed in all groups by approximately fivefold. In PTU-treated larvae, T3 normalized prl expression (−3.8-fold vs. basal). However, in glis3KD larvae, prl levels remained elevated despite partial reduction (−2.7-fold), persisting at tenfold above control levels. Similarly, th expression decreased in all groups post-T3 (controls, −onefold; glis3KD, −1.8-fold; PTU, −2.3-fold) (Fig. 5e).

In summary, glis3 deficiency causes severe hyperprolactinemia primarily through lactotropes over-differentiation, but also via increased hypothalamic trh expression, with partial thyroid hormone dependence, indicating that glis3 controls prolactin homeostasis through both pituitary-intrinsic and hypothalamic-thyroidal mechanisms.

Renal function of glis3KD larvae

Because prolactin regulates ion and water homeostasis in fish via inducing the expression of ionocytes in the gill and pronephros [20, 21], we examined whether hyperprolactinemia alters renal function in glis3KD larvae. Except for slc9a3.2, the levels of prolactin receptors (prlra and prlrb), and ionocytes ATPase (Na⁺/K⁺) transporters (atp1a1a.5 and atp1b1b), carbonic anhydrase II (ca2), solute carrier (Na⁺/K⁺/Cl⁻) cotransporters (slc12a3 and slc12a10.2), epithelial Ca²⁺channel (trpv6) and aquaporins (aqp1a and aqp3a), were significantly upregulated in glis3KD larvae (Fig. 6a).

Fig. 6.

Renal function in glis3KD larvae. a Relative expression of prolactin receptors and ionocytes in control and glis3KD larvae at 120 hpf. Results (mean ± SD of 3 pools, 20 larvae/pool) analyzed by Student’s t-test: ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001. b Schematic representation of zebrafish pronephric regions: glomerulus (g), neck (n), proximal convoluted tubule (pct), proximal straight tubule (pst), distal tubule, pronephric duct (pd), cloaca (c). Representative morphology of control (CT = 438/451, c–e), glis3kD (KD = 370/566, h–j), glis3KD + 150 nM Mannitol (KD + MANN = 92/123, m–o), and rescue (R = 112/157 r–t) embryos and larvae at 2 and 5 dpf, respectively. Insets showing IF of acetylated tubulin of the pronephric region. Pericardial edema (arrow, pe) and abdominal edema (arrowhead, ae) were reported. WISH of the different pronephric regions of CT (f, g), KD (k, l), KD + MANN (p, q), and R (u, v) larvae using wt1a (purple) and cdh17 (blue) markers, respectively. Depending on the experiment, embryos and larvae were acquired in lateral or dorsal view. Scale bars: c–e = 250 µm; f = 200 µm

Glis3 has been implicated in primary cilium function in animal models [22–24] and in glomerulocystic kidney disease in patients with GLIS3 mutations [25–28]. At 2 days post-fertilization (dpf), glis3KD embryos showed mild morphological abnormalities with disorganized renal cilia but no major cardiac defects (Fig. 6c, c’, h, h’) (Fig. S3, Tab. S6, ESM_1-4.mov). At 5 dpf, compared to controls, glis3KD larvae developed severe abdominal and cardiac edema, accompanied by reduced heart rate, impaired circulation, pronounced ciliary defects, and large glomerular cysts (Fig. 6i–j, k, k’) (Fig. S3, Tab. S6, ESM_7-10.mov). Analysis of the pronephros showed a visible enlargement of the proximal straight tubule (pst) and pronephric duct (pd) of glis3KD larvae compared to controls (Fig. 6g and l). At 2 dpf, glis3 KD embryos did not display changes in pronephric size (Fig. S4), suggesting that the phenotype observed at the larval stage reflects a dilatation due to fluid retention rather than a thickening of pronephric tubules and ducts. Treatment with 150 nM of mannitol prevented abdominal edema and pronephric dilatation but not cilia and glomerular abnormalities (Fig. 6m–p’), indicating that kidney defects are primary and consistent with the essential role of glis3 in ciliary function. Notably, pericardial edema was still present but did not affect cardiovascular functions (Fig. 6q) (Fig. S3, Tab. S6, ESM_11-12.mov). The phenotype was reproduced using an independent splice-blocking MO (glis3KD2, Fig. S5) and largely rescued by wild-type glis3 mRNA (Fig. 6r–v) (Fig. S3 and S4, Tab. S6, ESM_5-6.mov and ESM_13-14.mov). Functional clearance assays further confirmed defective renal excretion in glis3KD embryos, with > 80% retaining fluorescent dextran at 48 h post-injection (hpi), compared to complete clearance in controls (Fig. S6).

Together, these results demonstrate that glis3 is required for renal ciliary integrity and function, and its loss leads to cystic and functional kidney defects in zebrafish larvae.

Discussion

Glis3 is a multifunctional transcription factor with pivotal roles in endocrine development and renal physiology [22]. In humans, recessive GLIS3 mutations underlie the pleiotropic neonatal diabetes-hypothyroidism (NDH) syndrome, characterized by variable severity of diabetes, congenital hypothyroidism (CH), and polycystic kidney disease [23–26]. Mouse models recapitulate several but not all endocrine features of the human disorder, including impaired β-cell development and postnatal insulin synthesis [27–36]. Although Glis3KO mice display normal thyroid morphogenesis, they exhibit defective hormone biosynthesis and follicular proliferation [33]. By contrast, we previously demonstrated that the transient glis3 KD in zebrafish results in defective thyroid primordium specification, thyroid hypoplasia, and reduced thyroid hormone (TH) levels, and elevated TSH, thus reflecting more accurately the human CH due to thyroid dysgenesis [17]. These interspecies differences likely reflect variation in the temporal dynamics of Glis3 activity, redundancy within the Gli/Glis family, and methodological differences between transient knockdown and constitutive knockdown strategies.

Renal involvement is also conserved across species. Patients with GLIS3 mutations often develop polycystic kidney disease, and Glis3^zf/zf mice develop renal cysts by E15.5, frequently lacking primary cilia [37]. It was demonstrated that Glis3 localizes to the primary cilium and interacts with Wwtr1/TAZ, essential for ciliary integrity and renal homeostasis [37, 38]. Consistently, our zebrafish model displayed glomerular cysts and disorganized pronephric cilia, reinforcing the conserved role of glis3 in maintaining renal architecture and function. The analysis of cardiovascular function indicates that fluid retention causes, in turn, an impact on cardiac morphology and blood circulation of glis3KD larvae.

A key novelty of our study is the identification of glis3 as a regulator of AH specification. Acting downstream of Shh, glis3 integrates into the well-established signalling network that patterns the pituitary placode in multiple species [1, 39]. In zebrafish, glis3 shows a “salt and pepper” distribution in the brain, pharyngeal endoderm, and kidney [40]. Glis3 knockdown decreased shha in endoderm but increased expression in diencephalon and notochord, suggesting tissue-dependent roles as activator or repressor, possibly mediated by autoregulatory feedback loops [41, 42]. The interplay with retinoic acid (RA), FGF, and BMP pathways, known to interfere with Shh activity, further highlights glis3 as a potential node integrating multiple morphogenetic cues [8, 15, 43–48]. Although direct interactions between glis3 and these pathways have not been demonstrated, Shh/Gli2 controls the diencephalic Bmp4 and Fgf8 expression in mice during patterning of Rathke’s pouch [15], suggesting that glis3 may also operate within this regulatory network to integrate multiple signals and fine-tune shha expression in a context-dependent manner. An additional layer of complexity is introduced by the fact that GLI and GLIS proteins recognize overlapping, yet distinct, DNA binding sites with different affinities, likely influencing their transcriptional outputs in vivo. Their activity is further shaped by specific cofactors, whose spatial and temporal expression patterns remain poorly defined [13–15, 22, 49–51].

Knockdown of glis3 did not significantly alter the expression of other gli or glis family members in whole-mount analyses, suggesting limited compensatory activity among these factors. However, whole-mount approaches may mask tissue-specific changes, for example, within the developing pituitary, where other gli factors are known to contribute to development, though their interactions with glis3 remain undefined. Future work should address the molecular determinants of glis3 function, its interaction with other GLI/GLIS effectors in shha regulation, and the contribution of binding affinities and cofactors to its tissue-specific activity, particularly in the developing pituitary.

During zebrafish AH development, our data indicate that high shh in the diencephalon of glis3KD embryos triggered aberrant upregulation of nkx2.2a and lim3, the repression of pax7, and the expansion of pit1-positive precursors. Consequently, glis3-deficient AH exhibited disproportionate differentiation toward lactotropes, somatolactotropes, and thyrotropes, with concomitant loss of corticotropes and melanotropes. This scenario is similar to the phenotype of Shh-overexpressing embryos [11], contrasting with the reduction observed upon Shh inhibition [9], and aligns with reported effects of gli2a/b loss of function [14]. Notably, somatotrope numbers were severely reduced, likely reflecting premature exhaustion of the pit1 + precursor pool, given the sequential timing of pituitary lineage differentiation in zebrafish [1, 2, 12, 31]. In fact, mammalian AH placode initially differentiates corticotropes and somatotropes, followed by lactotropes, gonadotropes, and thyrotropes. In contrast, in teleosts, corticotropes and lactotropes emerge as early as 24 hpf, followed sequentially by somatolactotropes around 30–36 hpf, thyrotropes at 42 hpf, and, finally, somatotropes at 48 hpf [1–3, 7, 12, 52].

Among pit1-positive lineages, lactotropes were most affected. By 120 hpf, glis3KD larvae showed > 30-fold prl mRNA upregulation, ~ threefold increased prolactin protein, and a marked rise in lactotrope number. Elevated th and trh expression further implicates compensatory dopaminergic activity and primary hypothyroidism in driving hyperprolactinemia, consistent with mammalian models [34]. Although T3 treatment suppressed trh and reduced prolactin, prl levels remained elevated in glis3KD larvae, suggesting that both hypothyroidism and intrinsic pituitary mispatterning contribute to the phenotype.

Functionally, hyperprolactinemia was associated with upregulation of prolactin receptors, ionocyte transporters, and aquaporins, consistent with prolactin’s role in osmoregulation [20, 35]. Combined with pronephric cystic defects, these alterations led to impaired renal clearance and progressive edema. Thus, glis3 deficiency disrupts endocrine homeostasis and renal physiology through converging mechanisms involving the hypothalamic–pituitary–thyroid axis, prolactin signaling, and ciliary function.

In summary, our findings establish glis3 as an essential determinant of adenohypophyseal patterning, pituitary lineage allocation, and endocrine–renal homeostasis. Beyond clarifying the developmental basis of NDH syndrome, this work positions zebrafish as a powerful model to dissect the context-dependent functions of glis3 and its integration with Shh signaling, transcriptional cofactors, and ciliary pathways in vertebrate organogenesis.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

All authors contributed to the study conception and design. Federica Marelli performed morpholino microinjections, chemical treatments, molecular analysis (WISH, FISH, IF; ELISA assay), image acquisition, and data analysis. Giuditta Rurale did the RNA extraction and RT-qPCR experiments, analysis of renal function, and data analysis. Ilaria Gentile helps with the microinjection and riboprobe preparation. The first draft of the manuscript was written by Federica Marelli and Giuditta Rurale, and all authors commented on previous versions of the manuscript. Luca Persani conceived the study, obtained research funds, and revised and finalized the article. All authors approved the final version of the manuscript.

Funding

This work was supported by the Italian Ministry of Health (Rome, Italy).

Data availability

The datasets generated during the current study are available in the Zenodo repository (https://doi.org/10.5281/zenodo.17258522).

Declarations

Ethics approval

All experiments followed EU regulations on laboratory animals (Directive 2010/63/EU). The zebrafish study was approved by the Animal Welfare Body (OPBA) of the University of Milan, Italy (Protocol 198283). The study was approved by the Institutional Ethical Committee of IRCCS Istituto Auxologico Italiano (code: 05C102_2011).

Competing interests

The authors declare no competing interests.