Growth Hormone Neuroprotection Against Kainate Excitotoxicity in the Retina is Mediated by Notch/PTEN/Akt Signaling

Abstract

Purpose: In the retina, growth hormone (GH) promotes axonal growth, synaptic restoration, and protective actions against excitotoxicity. Notch signaling pathway is critical for neural development and participates in the retinal neuroregenerative process. We investigated the interaction of GH with Notch signaling pathway during its neuroprotective effect against excitotoxic damage in the chicken retina.

Methods: Kainate (KA) was used as excitotoxic agent and changes in the mRNA expression of several signaling markers were determined by qPCR. Also, changes in phosphorylation and immunoreactivity were determined by Western blotting. Histology and immunohistochemistry were performed for morphometric analysis. Overexpression of GH was performed in the quail neuroretinal-derived immortalized cell line (QNR/D) cell line. Exogenous GH was administered to retinal primary cell cultures to study the activation of signaling pathways.

Results: KA disrupted the retinal cytoarchitecture and induced significant cell loss in several retinal layers, but the coaddition of GH effectively prevented these adverse effects. We showed that GH upregulates the Notch signaling pathway during neuroprotection leading to phosphorylation of the PI3K/Akt signaling pathways through downregulation of PTEN. In contrast, cotreatment of GH with the Notch signaling inhibitor, DAPT, prevented its neuroprotective effect against KA. We identified binding sites in Notch1 and Notch2 genes for STAT5. Also, GH prevented Müller cell transdifferentiation and downregulated Sox2, FGF2, and PCNA after cotreatment with KA. Additionally, GH modified TNF receptors immunoreactivity suggesting anti-inflammatory actions.

Conclusions: Our data indicate that the neuroprotective effects of GH against KA injury in the retina are mediated through the regulation of Notch signaling. Additionally, anti-inflammatory and antiproliferative effects were observed.

It is known that growth hormone (GH) exerts several diverse actions in addition to somatic growth stimulation, which involve metabolic regulation as well as modulation of immune, reproductive, and neural functions, including behavior, among others.^1,2 Regarding the neural effects of GH, it has been shown recently that its clinical application has revealed promising results as a treatment for stroke, spinal cord injury, brain trauma, and neurodegenerative diseases.^3,4 Also, important neurotrophic actions of GH in the neuroretina are related with the modulation of developmental apoptosis and neuroprotection,^5,6 because it has been observed that exogenous GH administration protects against experimental excitotoxic damage.^7,8 In addition, GH acts during neural differentiation and network establishment to promote axon growth^9,10 and synaptogenesis.^11,12 

New strategies to cope with degenerative retinopathies include cotreatments with neurotrophic factors to induce neuroprotection and neuroregeneration. Regenerative neurogenesis in the retina involves Müller glia, which are able to re-enter the cell cycle and transdifferentiate into a progenitor cell phenotype in response to excitotoxic insult.¹³ Müller glia–derived progenitor cells (MGPCs) can subsequently proliferate and redifferentiate into new retinal neurons, although the regenerative capacity of these cells in mammals and birds is rather limited in comparison to the robust regenerative response of fish and other lower vertebrates.^14–18 Formation of MGPCs in the avian retina is controlled by a wide variety of interacting signaling pathways, including Notch,^15,19 MAPK,^20,21 Wnt/β-catenin,²² Shh,²³ BMP/Smad,^24,25 mTOR,²⁶ heparin-binding EGF-like growth factor,²⁷ JAK/STAT,²⁸ glucocorticoid,²⁹ and retinoic acid.³⁰ 

The Notch signaling pathway is important for both retinal development and visual function; and particularly, in lower vertebrates, it is considered to play a pivotal role during retinal regeneration.³¹ In chickens, the blockade of Notch signaling with the γ-secretase inhibitor N-[N-(3,5-Difluorophenacetyl)-L-alanyl]-S-phenylglycine t-butyl ester (DAPT), inhibits the formation of MGPCs after injury.^15,19 Although both, GH and Notch signaling have critical actions in retinal development and neural healing after a neurotoxic injury,^9,15 a direct link between the two has not previously been demonstrated in neural tissue. 

The current study describes the participation of non-canonical signaling pathways in GH-induced neuroprotection after excitotoxic damage in the chicken retina. Here, we demonstrate, for the first time, a relationship between exogenous GH treatment and the Notch signaling pathway in a neural tissue. This newly described signaling pathway for GH action includes phosphatase and tensin homolog (PTEN) and PI3K/Akt signaling as key downstream targets of GH-induced Notch signaling, which promotes retinal neuroprotection after experimentally induced excitotoxity. Surprisingly, GH blocked MGPC transdifferentiation through the downregulation of FGF2 and modulation of the inflammatory response through TNF receptors expression. 

Materials and Methods

Animals

Pathogen-free, fertilized eggs (Gallus gallus, White Leghorn) were kindly donated by Pilgrim's Pride (Querétaro, México) and were incubated at 39°C in a humidified air chamber (IAMEX, México City, México). The eggs were rotated one-quarter of a revolution every 50 minutes until hatch. All experimental animals were killed by decapitation in accordance with the Institute's Bioethical Committee regulations and with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. One-day-old chickens were used for in vivo experiments. Animals were anaesthetized with subcutaneous xylazine (1 mg/kg) and ketamine (2.2 mg/kg) prior to intravitreal injections. The number of animals per experimental group was determined by the principle of 3Rs. In the experiments designed for gene expression analysis (as determined by qPCR), six to eight animals per experimental group were included. In the experiments performed for signaling pathway analysis and proteomics (as determined by Western blotting), six to 10 animals per experimental group were included. In all the experiments, half eye was collected as morphologic control and the other half was used for mRNA or protein analysis. Morphometry and immunohistochemistry (IHC) were performed in four to five animals per experimental condition, and at least three fields in each eye were analyzed. Chicken retinal primary cell cultures were obtained from 10 embryos at embryonic day 10 (ED10) per experiment. In vitro experiments were performed three times including 3- to 4-culture wells per condition. 

Intraocular Injections

For intravitreal injections, kainic acid (KA; 20 μg/dose, Cat. K2389; Sigma-Aldrich Corp., St. Louis, MO, USA), recombinant chicken GH (rcGH; 300 ng/dose; Revholt, PRL, Israel), or DAPT (2 μg/dose; ab120633; Abcam, Cambridge, UK) were diluted in injectable water or DMSO (1 μL of DMSO for 1 μg of DAPT). A total volume of 20 μL was intravitreally injected into the left eye of chickens containing KA, rcGH, and/or DAPT. Right eyes were injected with a corresponding volume of vehicle and used as control (injection water or injection water + DMSO, where appropriate). 

Administration Protocol

The protective effect of GH was determined in a short-term administration protocol that included a single injury and four serial GH injections starting 1 day before the insult.^8,12 Intravitreal injections were administered into the left eye as follows: a prior GH dose (300 ng, 24 hours before damage), coinjection of KA (20 μg) + GH (300 ng), and two subsequent injections at 24 and 48 hours postinjury (GH, 300 ng). A control group receiving only GH was included. The damaged group only received the KA at postnatal day (P) 2; vehicle was administered as a sham. To determine the contribution of Notch signaling to the effects of GH, we coinjected GH with 2 μg of DAPT, a small molecule γ-secretase inhibitor. DAPT is commonly used to inhibit the Notch signaling pathway.^15,19,32 Right eyes were used as negative control (using vehicle as a sham). GH injections pre- and postinjury were used following reports of Fischer et al.,³³ Fischer and Reh,³⁴ Todd et al.,^27,28 and Ritchey et al.,³⁵ in order to induce growth factor expression and neuroprotection in the avian retina. 

Vector Construction and QNR/D Cell Line Transfection

Quail-derived neuroretinal cells (QNR/D)³⁶ acquired from American Type Culture Collection (No. CRL-2532) were cultured in Dulbecco's modified Eagle's medium (DMEM; Gibco, Grand Island, NY, USA) containing D-glucose (4.5 g/L), L-glutamine, and sodium pyruvate (100 mg/L), supplemented with 10% fetal bovine serum (FBS; Gibco) in 5% CO₂/95% O₂ at 39°C in a water-jacketed incubator as previously described.³⁷ For controlled overexpression using coumermycin as plasmid transactivator, we used a regulated expression mammalian system (Promega, Madison, WI, USA), which was fully functional in avian species.⁸ Transfections were carried out according to Martínez-Moreno et al.⁸ with minor modifications according to well-plate format. 

Retinal Primary Cell Culture

Primary cell cultures were performed at ED10 as reported elsewhere.^7,8 Briefly, neuroretinas were dissected in sterile PBS, and dissociated in a cocktail of papain (1 mg/mL), cysteine (0.2 mg/mL), and collagenase (0.18 mg/mL) for 20 minutes at 37°C, shaking gently every 5 minutes. Enzymatic digestion was followed by mechanical disaggregation. Cells were cultured in DMEM supplemented with 10% FBS and stabilized for 24 hours. Afterward, the cultures were stimulated with GH at 1, 10, or 100 nM for 24 hours, and harvested in the presence of TRIzol. 

Quantitative PCR Analysis

RNA was purified from tissue/cellular lysate using the Zymo Direct-zol purification kit and TRIzol (Zymo Research Corp., Irvine, CA, USA). After RNA extraction, DNAase I (Promega) digestion was performed at 37°C for 20 minutes, then the enzyme was inactivated at 72°C for 15 minutes. Complementary (c) DNA was synthesized from 2.0 μg of total RNA using oligo (dT) and random hexamers. Retrotranscription was performed with 100 U of Moloney murine leukemia virus reverse transcriptase (Promega) and 1 mM deoxyribonucleoside triphosphate mix for 60 minutes at 42°C. Target gene expressions (Table 1) were quantified by real-time quantitative PCR (qPCR) in a sequence detection system ABI-PRISM 7900HT (Applied Biosystems, Foster, CA, USA) and using SYBR Green (Maxima; Thermo Fisher Scientific, Waltham, MA, USA) in 10-μL final volume containing 3 μL of diluted cDNA and 0.5 mM of each specific primer. Primer sequences were taken from previous reports.^8,15,26,28 PTEN (97% efficiency) and BMP4 (104% efficiency) primers were designed, including an intron inside the amplification frame to avoid genomic DNA contamination. Reactions were performed under the following conditions: initial denaturation at 95°C for 10 minutes, followed by 45 cycles of 95°C for 15 seconds, 60°C for 15 seconds, and 72°C for 15 seconds. Dissociation curves were included after each qPCR experiment to ensure primer specificity. Relative abundance of mRNA was calculated using the comparative threshold cycle (Ct) method³⁸ and employing the formula 2^−ΔΔCT where the quantification is expressed relative to the geometric mean of 18S mRNA.³⁹ 

Table 1

 

Oligonucleotides

 

Western Blot Analysis

Total proteins were extracted from retinas and QNR/D cultures using homogenization buffer (0.05 M Tris-HCl, pH 9.0) supplemented with protease inhibition cocktail (Mini-complete; Roche, Mannheim, Germany). Equivalent amounts of proteins (40–60 μg) were separated by 12.5% SDS-PAGE under reducing conditions and transferred onto nitrocellulose membranes (Bio-Rad, Hercules, CA, USA). Free binding sites were blocked with 5% non-fat milk (Cat. No. 170-6404; Bio-Rad) in Tris-buffered saline (TBS) for 2 hours at room temperature (RT). Membranes were then incubated overnight at RT with the corresponding antibody (Table 2) in TTBS (1 × TBS with 0.05% tween [vol/vol]). After washing the membranes with TTBS (2 × 5 minutes), they were then incubated for 2 hours with the corresponding horseradish peroxidase conjugated secondary antibody. Bands were visualized using enhanced chemiluminescent Blotting Detection Reagent (Amersham-Pharmacia, Buckinghamshire, UK) on autoradiography film (Fujifilm, Tokyo, Japan). Kaleidoscope molecular weight markers (Bio-Rad) were used as reference for apparent molecular weight determination. Images were captured on Gel Doc EZ Imager (Bio-Rad) and optic density from immunoreactive bands was obtained using Image Lab software (Bio-Rad). Target immunoreactivities were corrected using a loading control (glyceraldehyde 3-phosphate dehydrogenase [GAPDH] or actin). 

Table 2

 

Antibodies

 

Histochemistry and Immunohistochemistry

Enucleated eyes were hemisected longitudinally and the vitreous were removed. Eyes were fixed in 4% paraformaldehyde plus 3% sucrose in phosphate-buffered saline (PBS) for 1 hour. Samples were washed with PBS (3 times), cryoprotected in 30% sucrose/PBS, and freeze mounted onto aluminum sectioning blocks with O.C.T compound (Tissue-Tek, Alphen aan den Rijn, The Netherlands). Sections of 5 μm were cut with a cryotome (Leica CM3050 S; Leica, Wetzlar, Germany) and then mounted on pretreated glass slides (Fisherbrand; Thermo Fisher Scientific). For histologic analysis, slides were stained with hematoxylin. For immunohistochemical analysis, cGH and doublecortin (DCX) were both determined with their respective rabbit polyclonal antibody, diluted 1:500 in PBS plus 0.2% Triton X-100 (Bio-Rad Laboratories, Inc.). Negative controls without primary antibodies were included. Secondary antibodies were diluted 1:2000 in Triton X-100 with 1% non-fat dry milk (Bio-Rad) and applied for 2 hours at RT. Sections were also stained with DAPI (100 ng/mL) and mounted with Vectashield antifade mounting medium (Vector Laboratories, Burlingame, CA, USA).⁸ Images were captured using a confocal microscope Zeiss Axiovert 200 LSM 510 (Carl Zeiss AG, Oberkochen, Germany). 

Identification of STAT5 Binding Domain in Chicken Notch1 and Notch2 Genes

The STAT5 sequences of mouse and chicken (GeneBank accession numbers NP_001157534.1 and NP_990110.1, respectively) were aligned using the Muscle algorithm in the Geneious software (in the public domain, http://www.geneious.com). DNA-binding domains, described by Soldaini et al.,⁴⁰ were realigned and extracted. To predict the STAT5 binding sites on promoters, 1000 bp upstream of the transcription start site (TSS) of chicken Notch1 (ID: 395655) and Notch2 (ID: 374031) genes were analyzed by the JASPAR CORE server (in the public domain, http://jaspar.genereg.net) using the mouse STAT5 matrix profile (ID: MA0519.1). The binding sites matching 80% in the plus strain were considered as positive. 

Statistical Analysis

Gene expression determinations were analyzed by duplicate in qPCR from six to eight animals per experimental condition. Each experimental condition (left eye) used its contralateral eye (right) as a reference for relative change (delta). GH overexpression was corroborated by qPCR from three different transfections, including at least three QNR/D cell cultures per group. Primary cultures included at least four wells per condition in each experiment. Cell number was determined in retinal slices stained with hematoxylin; and cell quantification was performed in at least three microscope fields per eye and four individual eyes per experimental group were analyzed. Cell counting was done in the same area for equivalent group comparisons, the criteria of approximately 1 to 3 mm from the head of the optic nerve was applied for image capturing. Densitometric analysis included at least six to 10 animals per experimental condition. In all the experiments values are expressed as mean ± SEM. Significant differences between groups or treatments were determined by one-way ANOVA, least significant difference (LSD) Fischer provided a post hoc test. P values less than 0.05 were determined to be statistically different (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001). 

Results

DAPT Inhibits GH-Induced Retinal Neuroprotection After KA Excitotoxic Injury

A scheme of the injection and treatment protocols followed in this work is presented in Figure 1A. Histologic analysis showed severe loss of tissue cytoarchitecture in the KA and KA+GH+DAPT groups, while retinal organization was similar to sham control in retinas treated with GH and KA (Fig. 1B). Damage was observed as a degradation of the distinct retinal layers, with spaces and displaced cell bodies found in the outer plexiform layer (OPL) and the inner plexiform layer (IPL). In the KA group (without GH treatment), cell count decreased significantly in the ganglion cell layer (GCL), inner nuclear layer (INL), and outer nuclear layer (ONL) (Figs. 1C–E). Interestingly, GH treatment prevented the cell loss in the three cell layers (GCL, INL, and ONL) and the comparison of KA versus KA+GH groups showed a significant difference in the INL and ONL (P < 0.01 and 0.05, respectively). Coadministration of GH and DAPT in KA-damaged retinas, showed a damage similar to retinas treated with KA only, particularly in the GCL and INL. However, the number of cells in the ONL showed no difference between the KA+GH and KA+GH+DAPT groups (Figs. 1C–E). 

Figure 1

 

Morphometric analysis showing the protective effect of GH against KA excitotoxicity in chicken retinas. (A) Treatments and timeline schematic representation of injection protocols in the experimental groups. Units in nanograms and micrograms. (B) Hematoxylin staining in retinal slices from chickens treated with KA (damage), GH+KA (neuroprotection), and GH+KA+DAPT (Notch inhibition). Sham (vehicle injected) as negative control. Quantification of cell number in retinal layers, including GCL (C), INL (D), and ONL (E). Scale bar: 10 μm. Bars indicate cells per 630 μm² (n = 4–5 animals per group, 3 fields were quantified per retina/animal). Asterisks show significant difference in comparison to control (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001) and number sign (#) shows differences between experimental groups (^#P < 0.05; ^##P < 0.01) as determined by one-way ANOVA for multiple comparisons and LSD Fischer as post hoc test.

Effect of GH Treatment on Notch Signaling

Notch1, Hes5, PTEN, and Ascl1a mRNA expression levels were analyzed in retinas collected from eyes injected with KA and/or GH, in order to study the effects of GH on retinal Notch signaling (Fig. 2). The expression of the receptor Notch1 and its downstream target Hes5 mRNAs were analyzed in retinas treated with KA and/or GH. Notch1 mRNA expression was significantly decreased after GH treatment and increased by KA excitotoxicity (P < 0.001 and 0.05, respectively) (Fig. 2A), while no effect was observed in Hes5 mRNA expression with either GH or KA (Fig. 2B). On the other hand, cotreatment with GH and KA showed a strong significant increase in both Notch1 and Hes5 mRNAs expression in comparison to KA and control groups (Figs. 2A, 2B). Coinjection of KA+GH+DAPT blocked the increase in Hes5 but not Notch1 mRNA expression observed in the GH+KA group and remained at control levels (Fig. 2B). PTEN gene expression increased significantly in KA-damaged retinas (P < 0.001), but the coaddition of GH was able to significantly reverse this effect and decrease (P < 0.01) its expression. DAPT treatment increased PTEN mRNA levels over that of the GH+KA group (P < 0.01), to a midway point between the control and KA groups (but no significant difference was observed with either group). Excitotoxic damage increased Ascl1 mRNA levels (P < 0.05), but its gene expression was not altered by GH alone nor when coinjected with KA (Fig. 2D). Thus, in the absence of KA damage, GH treatment only showed effect upon Notch1 mRNA expression but not in the other genes analyzed. 

Figure 2

 

Expression of Notch1, Hes5, PTEN, and Ascl1a mRNAs in the chicken neuroretina. Left eyes were injected with GH (300 ng), KA (20 μg), GH+KA, or GH+KA+DAPT (2 μg). Relative expression of Notch1 (A), Hes5 (B), PTEN (C), and Ascl1a (D) mRNAs as determined by qPCR. Ribosomal RNA 18s was used as housekeeping gene. Relative mRNA expression values were corrected by the Ct method and employing the formula 2^−ΔΔCT, using the right eye from each animal as a control reference. Bars show mean fold ± SEM (n = 6–8 animals per group). Asterisks indicate significant difference in comparison to control (*P < 0.05; **P < 0.01; ***P < 0.001) and number sign (#) shows difference between experimental groups (^###P < 0.001) as determined by one-way ANOVA for multiple comparisons and LSD Fischer as post hoc test.

GH Treatment and Regenerative Markers Expression

GH treatment prevented the expression of regenerative markers induced by KA neurotoxicity. KA treatment induced a significant increase in Sox2 (Fig. 3A, P < 0.05) and FGF2 (Fig. 3B, P < 0.01) mRNA expression. Conversely, the cotreatment with GH blocked the upregulation of both FGF2 (Fig. 3B, P < 0.01) and Sox2 (Fig. 3A, P < 0.05) induced by KA excitotoxic damage. Unexpectedly, PCNA mRNA was not increased by KA excitotoxicity; however, retinas treated with GH and KA showed a significant decrease (P < 0.01) in comparison to control retinas (Fig. 3C). On the other hand, BMP4 mRNA expression only showed a significant increase (P < 0.01) in the KA+GH group (Fig. 3D, P < 0.01) and no significant differences were observed in the KA or GH groups. 

Figure 3

 

Expression of Sox2, FGF2, PCNA, and BMP4 mRNAs in the chicken neuroretina. Left eyes were injected with GH (300 ng), KA (20 μg), GH+KA, or vehicle from P1 to P4, and eyes were harvested at P5. Relative expression of Sox2 (A), FGF2 (B), PCNA (C), and BMP4 (D) mRNAs as determined by qPCR. Ribosomal RNA 18s was used as housekeeping gene. Relative mRNA expression values were corrected by the Ct method and employing the formula 2^−ΔΔCT, using the right eye from each animal as a control reference. Bars show mean fold ± SEM (n = 6–8 animals per group). Asterisks indicate significant difference in comparison to control (*P < 0.05; **P < 0.01) and number sign (#) shows difference between experimental groups (^##P < 0.01) as determined by one-way ANOVA for multiple comparisons and LSD Fischer as post hoc test.

Antiapoptotic Pathways Involved in GH Neuroprotective Effect

To determine which pathways are involved in GH neuroprotective effect against KA-induced neurotoxicity we analyzed retinal changes in the immunoreactivity (IR) of several antiapoptotic proteins (pAkt, Akt, Bcl-2, Bcl-X, and Mcl-1) by Western blot. Immunoreactive bands for pAkt, Akt, Bcl-2, Bclx, and Mcl-1 showed the corresponding molecular weights (56, 56, 26, 26, and 42 kDa, respectively) previously reported for each; actin immunoreactivity (42 kDa) was used as a loading and normalizing control for Bcl-2, Bcl-X, and Mcl-1 (Fig. 4A), while pAkt was normalized with Akt (Fig. 4A). PAkt-IR was increased in the KA+GH group only, and no significant changes were observed in the GH or KA groups in comparison to the control group. Interestingly, the increase in Akt phosphorylation observed in the GH+KA group was blocked by DAPT coadministration (Fig. 4B). As expected, GH strongly increased Bcl-2-IR (P < 0.001) and it was significantly decreased by KA damage (P < 0.01) in comparison to the control group (Fig. 4C). Significant differences in Bcl-2-IR were found between the KA and KA+GH groups (P < 0.05), noting as well that GH restored Bcl-2-IR to control levels in KA-damaged retinas. However, Bcl-2-IR was not affected by DAPT coadministration. In contrast, Bcl-X showed no significant differences in any of the analyzed groups (Fig. 4D). On the other hand, Mcl-1-IR only showed a significant decrease (P < 0.01) in the group treated simultaneously with GH, KA, and DAPT in comparison to the other groups and the untreated control (Fig. 4E). 

Figure 4

 

Antiapoptotic signaling during GH neuroprotection against KA excitotoxicity. (A) Representative luminograms for pAkt, Akt, Bcl-2, Bcl-X, Mcl-1, and β-actin immunoreactivities in experimental groups treated with GH (300 ng), KA (20 μg), GH+KA, and GH+KA+DAPT (2 μg). β-actin was included as a loading and normalizing control for Bcl-2, Bcl-X, and Mcl-1, while Akt was used to normalize pAkt. Units are expressed in kilodaltons. Left eyes were used for treatments (Tx) and right eyes were used as sham control. Immunoreactive bands were analyzed by densitometry for pAkt (B), Bcl-2 (C), Bcl-x (D), and Mcl-1 (E). Bars represent mean fold ± SEM (n = 6–10 animals per experimental condition). Asterisks indicate significant difference in comparison to control (*P < 0.05; **P < 0.01; ***P < 0.001) and number sign (#) shows difference between experimental groups (^#P < 0.05) as determined by one-way ANOVA for multiple comparisons and LSD Fischer as post hoc test.

GH Increases Doublecortin-IR After KA Damage

As shown by immunofluorescence analysis, DCX-IR was mainly localized, under control conditions, in the OPL, GCL, and optic fiber layer (OFL) of neonatal retina of chickens (Fig. 5B). After an excitotoxic insult with KA, DCX-IR changed its distribution into the outer retina (Fig. 5F); however, it was possible to observe positive cells in other sections of the retina and inside the plexiform layers (arrows). GH-IR was strongly observed in cell bodies within the INL and GCL under control conditions (Fig. 5C). In damaged retinas, GH immunostaining was present with high intensity in the outer retina but also was observed in some retinal ganglion cells (RGCs) and in the OFL (Fig. 5G). Colocalization of GH and DCX under control conditions was only found in the OFL (Fig. 5D), even if both immunoreactivities were present in the OFL, INL, and ONL. KA treatment induced a strong colocalization of both markers, particularly in the outer regions of the retina (Fig. 5H, arrows). Negative controls without primary antibodies showed no immunofluorescence (Figs. 5I–5L). 

Figure 5

 

Effect of KA treatment on doublecortin, NeuN and GH immunoreactivity in the retina. GH and DCX immunofluorescence in control (sham) and KA-treated retina. Sham (vehicle) and KA (20 μg) treated retinas were stained with DAPI (blue [A, E]), and with specific antibodies directed against DCX (green [B, F]) and GH (red [C, G]). Merged images are shown in panels D and H. Arrows denote colocalization of GH and DCX in the same cells (yellow [H]). Negative controls without primary antibodies (I–L). Scale bar: 20 μm. Images are representative of at least four animals per group. Representative immunoblot for doublecortin (M) and NeuN (N) immunoreactivity. Densitometric analysis of immunoblots for DCX (O) and NeuN (P) corrected and normalized with GAPDH immunoreactivity. Left eyes were used for treatments (Tx) and right eyes were used as sham control. Bars represent mean ± SEM (n = 4–6). Asterisks indicate significant differences (**P < 0.01; ***P < 0.001) as determined by one-way ANOVA for multiple comparisons and LSD Fischer as post hoc test.

The effect of GH+KA cotreatment on DCX and NeuN immunoreactivities was determined by Western blot (Figs. 5M, 5N). KA+GH treatment caused a significant increase in DCX-IR over control and KA groups (Fig. 5O, P < 0.01). KA treatment caused a significant downregulation of NeuN-IR, which was not restored by GH cotreatment (Fig. 5P, P < 0.001 and 0.01, respectively). The effect of KA excitotoxicity on DCX and NeuN immunoreactivities after 2, 24, 48, and 96 hours treatment was analyzed by Western blot (Supplementary Fig. SA). It was shown that DCX-IR was significantly upregulated 48 hours postinjection compared with control conditions while NeuN-IR was downregulated 96 hours postinjection (Supplementary Fig. SA). 

Effect of GH on Notch1 and FGF2 Expression In Vitro

The QNR/D was efficiently transfected with a coumermycin-induced cGH overexpression system (Fig. 6). The highest rcGH expression observed in cells transfected with the regulated expression system (pfSPGH) resulted from treatment with coumermycin at 5 and 50 nM (Fig. 6A). GH overexpression resulted in significant decreased FGF2 mRNA levels at a coumermycin concentration of 5 and 50 nM (Fig. 6B; P < 0.05 and 0.01, respectively). In contrast, Notch1 mRNA levels were significantly increased only in transfected cells cultured in the absence of coumermycin (Fig. 6C; P < 0.05), while cell cultures treated with coumermycin showed no significant changes in Notch1 expression at a concentration of either 5 or 50 nM of the antibiotic (Fig. 6C). In primary retinal cell cultures from ED10 chicken embryos, incubations with GH at all doses tested (1, 10, and 100 nM) resulted in a significant decrease in both FGF2 and Notch1 mRNA expression (Figs. 6D, 6E). 

Figure 6

 

In vitro effects of GH upon FGF2 and Notch1 mRNA expression in QNR/D cells and embryonic retinal primary cell cultures. (A) Coumermycin (at 5 and 50 nM) was used to induce GH overexpression in transfected QNR/D cells (pfSPGH; antibiotic-inducible GH expression plasmidic system). Relative expression of FGF2 (B) and Notch1 (C) mRNAs as determined by qPCR in transfected QNR/D cells overexpressing rcGH in a coumermycin-inducible system. Embryonic retinal primary cell cultures were treated with cGH (1, 10, 100 nM) for 24 hours and FGF2 (D) and Notch1 (E) mRNA levels were quantified. qPCR values were corrected by the Ct method and employing the formula 2^−ΔΔCT. Ribosomal RNA 18s was used as housekeeping gene. A single intravitreal dose of GH (300 ng) was used to induce retinal pErk and pAkt immunoreactivity changes after 10, 30, 60, and 120 minutes. (F) Representative luminograms showing pErk, Erk, pAkt, and Akt immunoreactivities. Units in kilodaltons (kDa). Densitometric analysis for pErk (G) and pAkt (H) were corrected and normalized using Erk or Akt, respectively, as a loading control. Units are expressed in kilodaltons. Bars represent mean ± SEM (n = 3; from 3 independent experiments analyzed by duplicate). Asterisks indicate significant difference (*P < 0.05; **P < 0.01; ***P < 0.001) as determined by one-way ANOVA for multiple comparisons and LSD Fischer as post hoc test.

GH Activates pAkt and pErk in Postnatal Chick Retina

As expected, a single intravitreally injection of 300 ng of GH was able to activate the MAPK and IP3/AKT signaling pathways (Figs. 6F–H). Immunoreactive bands for pErk and pAkt were detected by immunoblotting (molecular weight of 42 and 60 kDa, respectively, were confirmed for these proteins) (Fig. 6F). Erk and Akt immunoreactivities were used for either pErk or pAkt normalization, respectively. The activation of MAPK signaling pathway was demonstrated through relative changes in pErk-IR, and a significant increase was observed at 120 minutes after treatment in comparison to the control group (Fig. 6G). GH also induced an increase (P < 0.01) of pAkt immunoreactivity, but only at 60 minutes after its ocular administration, indicating a transient activation of the pathway (Fig. 6H). 

Effect of GH on GFAP, Vimentin, and TNF Receptors After a KA-Induced Excitotoxic Injury

The possibility that GH participates in reactive gliosis was studied by determining the response of vimentin and GFAP by Western blotting (Fig. 7). As shown in Figure 7A, both KA and KA+GH treatments stimulated a significant increase of GFAP (P < 0.05 and 0.01, respectively) in comparison to the control group, however, there was no difference between them. In turn, vimentin was significantly elevated in the KA group in respect to the control, while the KA+GH group showed a tendency to increase but there was not a statistically significant difference (Fig. 7B). 

Figure 7

 

Effect of GH on reactive gliosis markers, GFAP and vimentin, as well as in TNF receptors, after an excitotoxic damage with KA. Controls (sham), KA and KA+GH groups were analyzed by Western blotting and densitometry. Representative luminograms of (A) GFAP, (B) vimentin, (C) TNF-R1, and (D) TNF-R2 immunoreactivities and their corresponding densitometries. In all cases, actin was used a loading and normalizing control. Left eyes were used for treatments (Tx) and right eyes were used as sham control. Bars represent mean ± SEM from four to six animals. Asterisks indicate significant difference (*P < 0.05; **P < 0.01; ***P < 0.001) as determined by one-way ANOVA for multiple comparisons and LSD Fischer as post hoc test.

The proinflammatory-associated TNF receptor 1 (TNF-R1), showed a significant decrease (P < 0.001) in KA-damaged retinas treated with GH, but no substantial effect was observed in damaged retinas without GH treatment (Fig. 7C). On the other hand, the prosurvival-associated TNF-receptor 2 (TNF-R2) was significantly decreased (P < 0.001) in the KA group, and this effect was prevented when GH was coadministered, because the KA+GH group remained similar to the control. 

Identification of Binding Sites for STAT5 in Notch1 and Notch2 Genes

STAT5 is the canonic signaling pathway associated with GH receptor (GHR) activation.⁴¹ The DNA binding domain of STAT5 is highly conserved between mouse and chicken showing 95% of identity and 98% of similarity (Fig. 8A). Mouse STAT5 binds to a DNA motif (Fig. 8B) described by Zhang et al.⁴² Using this profile and the last version of JASPAR,⁴³ STAT5 binding sites were predicted at 258 and 533 bp upstream of chicken Notch1 and Notch2 TSS, respectively (Fig. 8C). 

Figure 8

 

Prediction of STAT5 binding sites on chicken Notch1 and Notch2 promoters. (A) Multiple alignment of chicken and mouse STAT5-DNA–binding domains. Similar residues are colored according to BLOSUM62 as follows: black = identical; gray = similar; white = no similarity. (B) Sequence logo of the matrix profile (ID: MA0519.1) used to predict DNA binding sites. (C) Sequences of chicken Notch1 and Notch2 promoters showing the predicted STAT5 binding sites (red underline), transcription start sites (black arrows), and start codon (blue underline).

Discussion

Notch Signaling Mediates GH Neuroprotective Actions

Our results demonstrate, for the first time, an association between GH treatment and Notch signaling pathway activation in neuroretinal tissues. Notch belongs to a family of transmembrane receptors, which activate their signaling functions by binding to its transmembrane ligands present on an adjacent cell.⁴⁴ Upon binding, the corresponding receptor is cleaved, releasing the Notch intracellular domain (NICD), which is then translocated into the nucleus where it associates with the DNA-binding protein CSL, a transcriptional repressor.⁴⁴ After NICD-CSL association, CSL activity switches to a transcriptional activator, which goes on to induce the expression of several downstream target genes.^44,45 Hes5 is a basic helix-loop helix protein (bHLH) and a transcriptional repressor of proneural genes, such as Ascl1a, and its transcription is promoted by Notch signaling.^15,46 After injury, components of the Notch signaling cascade are activated in the retina during Müller glia transdifferentiation and regeneration.^15,19,47,48 

To test the neuroregenerative potential of GH, we therefore first studied its effects on Notch signaling in an excitotoxic damage model. We have previously designed and validated a short/acute damage protocol, including four serial intravitreal GH injections to induce neuroprotection, in which the excitotoxic injury with KA occurs at the second day.⁸ In this work, we observed that Notch1 and Hes5 shared a mirrored transcriptional response to GH after KA-excitotoxic damage, with a clear upregulation of Notch1 and Hes5 gene expression in chicken retinas when treated with GH and KA. We observed a relative decrease of Notch1 mRNA in retinal cell cultures and in neonatal retinas (in vivo) after an acute GH intraocular administration; however, in the presence of neurotoxic damage this response pattern showed an opposite behavior. This result suggests that the Notch1 response to GH is conditioned by the pathophysiologic status of the retinal tissue. After DAPT administration, however, levels of Notch1 mRNA expression were not significantly changed, whereas Hes5 mRNA expression was returned to control levels, confirming the use of DAPT as a Notch-signaling inhibitor. 

The administration of GH to prevent against KA damage preserved the cytoarchitecture of retina and maintained the cell count in three main retinal layers, as previously reported.^7,8 Coinjection of DAPT and GH in KA-treated retinas resulted in an increase of damage to the structure of the retina and a decrease in cell count in all retinal layers to a level comparable to KA injections alone. Because the decrease in cell number caused by DAPT could either be due to a cell loss or a block of the proliferation/transdifferentiation of Müller cells we thus investigated both possibilities. 

Neuroprotective Action of GH Prevents Müller Cell Transdifferentiation

Surprisingly, GH cotreatment reduced the expression of the proliferation marker PCNA in this short-term protective protocol only when excitotoxic damage was induced; however, contrary to expected, cell proliferation did not increase due KA injury alone. On the other hand, we observed an upregulation of FGF2 gene expression after KA injection. This response is in accord with previously reported effects of NDMA on FGF2 expression.^33,49,50 When GH was coinjected with KA, FGF2 mRNA levels were significantly reduced and returned to levels similar to that of the control group, indicating that GH negatively regulates its expression. We subsequently confirmed this finding in QNR/D cells overexpressing GH and in embryonic primary retinal cell cultures treated with GH. This finding is important as it has been reported that FGF is a potent regulator of Müller cell transdifferentiation and proliferation, with multiple injections of FGF2 being able to cause MGPC formation even in the absence of neurotoxic insult.^{27,29,33,34,50,51} FGF2 acts mainly on the Müller glia by upregulating the complex network of signaling pathways, which promote transdifferentiation, including Notch, MAPK, pCREB, and β-catenin.^19–21,52 In addition, we found that the expression of Sox2 in our model mirrored the expression of FGF2. Sox2 is a transient transdifferentiation marker only expressed by Müller cells in response to damage and/or repeated doses of growth factors.^25,33,53 We have previously proposed a model of GH action on the retina, where it regulates the local expression of endogenous growth factors, which then mediate its neuroprotective activity in this tissue, in addition to its direct effects on RGCs.^8,12 Our current results strengthen and support this model, showing that the negative regulation of FGF2 expression by GH leads to the inhibition of MGPC formation and proliferation. It should be noted that GH has been shown to exert well-established actions promoting regeneration and proliferation of progenitor cells in other models.^54–59 Indeed, it is known that GH is upregulated in proliferating rat hippocampal progenitor cells after KA damage, with the proliferative response further enhanced by GH treatment.⁵⁵ The observation that GH blocked a classical growth factor, such as FGF2, and exerted antiproliferative effects was an unexpected finding; thus, the analysis of other timeframes, treatment protocols, identification of secreted factors, and other intracellular signaling pathways is mandatory to understand this multifactorial neurotrophic effect. The downregulation of FGF2, Sox2, and Ascl1a, together with the upregulation of BMP4, suggests a neuroprotective effect that blocks or delays the proliferative response in order to generate neural progenitors.^33,60 

GH Neuroprotective Actions Involve Notch, Akt, and Bcl-2 Signaling

There are controversial reports about the role of the Notch pathway in neuroprotection. On the one hand, Notch signaling inhibition results in neuroprotection and improved cognitive function in rats with craniocerebral injury and experimental stroke.^61,62 Notch activation is thought to have deleterious effects after ischemic stroke where it increases cell death in hippocampal cells, disrupts glial function, promotes aberrant proliferation, and causes additional damage.^19,63,64 Notch may promote cell death after neural injury through the disruption of Müller cell function, which supports the neural microenvironment.¹⁹ As such, inhibition of Notch signaling in Müller glia prior to excitotoxic injury has been shown to be neuroprotective of retinal amacrine, bipolar, and RGCs.¹⁹ On the other hand, Notch signaling has been shown to promote survival of neural precursor cells through the upregulation of Bcl-2 and Mcl-1⁶⁵ and newly differentiating neurons undergo apoptosis in the absence of Notch signaling in embryonic mice.⁶⁶ Elevated Notch signaling exerts a neuroprotective effect on the RGCs of Sprague-Dawley rats with ocular hypertension, through the upregulation of Bcl-2 and downregulation of Bax and caspase-3.⁶⁷ Additionally, it has been observed that Notch induces activation of the prosurvival Akt signaling pathway through the downregulation of PTEN, an inhibitor of the pathway.⁶³ 

We found that GH cotreatment in the presence of KA promoted the activation of the PI3K/Akt signaling pathway. It is well known that GH is able to act through MAPK and PI3K/Akt in neural tissues, in accordance with previous reports from our group.^8,68–70 Here, we demonstrated that a single dose of GH promoted phosphorylation of Erk and Akt within an hour postinjection. However, activation of the pAkt pathway was transient, and samples injected with GH alone showed no increase in Akt phosphorylation. PTEN is a phosphatase and a negative regulator of cell growth, survival, and proliferation. PTEN negatively regulates the Akt pathway through the dephosphorylation of PIP3 to PIP2, which reduces phosphorylation of Akt through PDK1.^71–73 Notch signaling has been previously shown to reduce PTEN expression in breast cancer cells, which leads to antiapoptotic effects through increased activation of the pErk pathway.⁷⁴ KA injections alone resulted in an increase in PTEN expression, whereas GH was only able to decrease the expression of PTEN below control in damaged tissue. DAPT was able to block the phosphorylation of pAkt by GH treatment, which was correlated with an increase in PTEN expression. Therefore, it is likely that GH is enhancing its own Akt-mediated neuroprotective activity through suppression of PTEN via Notch signaling under KA-damage conditions. 

Antiapoptotic effects observed in this work also involve Bcl-2 actions, because KA injury reduced its immunoreactivity and GH was able to restore to control levels. In discrepancy with other findings,⁶⁷ our results indicated that regulation of Bcl-2 expression was independent of Notch signaling. According to our data Bcl-X and Mcl-1 are not involved in GH-induced neuroprotection against acute KA excitotoxicity in this time frame. 

GH Neuroprotection Include TNF Signaling and Reactive Gliosis Modulation

TNF signaling is a key player in retinal damage, because its activation controls the neuroinflammatory response in neural tissue, as demonstrated by its upregulation after a NDMA damage in the retina.^75–77 Dying retinal neurons release TNF-α, which in turn induces neuroinflammation and subsequently promotes the production of local growth factors required to initiate the retinal transdifferentiation and regeneration process.^78,79 It has been suggested that GH has anti-inflammatory actions and mediates initial processes of inflammatory response; however, there is no direct evidence on neuroinflammation.^80,81 We observed a downregulation of TNF-R1 induced by GH treatment during KA-induced damage and complementarily, TNF-R2 was returned to that in control levels in the presence of KA and GH. Our experimental data suggest that TNF signaling is regulated at the level of the receptor expression, with TNF-R1 having proinflammatory actions and TNF-R2 playing a role in retinal neuroprotection.^82,83 Retinal excitotoxic damage involves the activation and/or dysfunction of glial cells and a neuroinflammatory response; however, its interactions, dynamics, and coordinated responses are still largely unknown. The neuroprotective effects of GH likely hinder the regenerative process through a mechanism involving the modulation of receptors for TNF resulting in anti-inflammatory actions. 

To analyze the effects of GH on reactive gliosis, GFAP and vimentin were used as accepted markers for this process.^33,84,85 As expected, KA induced reactive gliosis as evidenced through increased immunoreactivity of GFAP and vimentin. Interestingly, GH treatment had no effect on GFAP-IR and was able to slightly decrease vimentin-IR. This is likely due to GH anti-inflammatory action through regulation of TNF receptors, as the inflammatory response following neural injury supports deleterious gliosis, resulting in a scar.^86,87 In the chicken model, nonastrocytic inner retinal glia-like cells are known for expressing vimentin and IGF-1 receptor but lower levels of GFAP, so it is possible that GH effect on this gliosis marker is mainly occurring in this cell type.⁸⁸ In higher vertebrates, after damage to the nervous system, the proliferation of glial cells results in the formation of a deleterious scar, rather than partial or fully functional neurogenesis as is found in some lower vertebrates.^29,50,89 Therefore, it is important to investigate the effects of GH in specific glial cell types and the molecular interactions associated to its potential use as a therapeutic agent. 

Notch1 and Notch2 Promoter Regions Include Binding Sites for STAT5

In silico analysis of the Notch receptor family genes revealed conserved DNA binding sites for STAT5 present in the promoter regions for Notch1 and Notch2 genes, in both chickens and mice.^42,43 Because we observed a differential effect of GH on Notch1 expression between damaged and intact tissue, this is likely due to an interaction with other transcriptional factors and signaling pathways, which are coactivated by KA treatment. It therefore seems likely that the functional significance of GH regulating Notch expression is context dependent. We suggest that downregulation of Notch signaling by GH in healthy neural tissue results in neural differentiation. Notch signaling regulates the stemness of progenitors and maintains a proliferative, undifferentiated phenotype.^15,44 Indeed, previous work from our group has demonstrated that GH plays a role in neural differentiation of the retina, promoting axon growth as well as synaptogenesis.^9–12 In damaged neural tissue, GH upregulates components of the Notch signaling pathway leading to the expression of Hes5 and repression of PTEN. Hes5 molecules have been demonstrated to potentiate STAT signaling⁹⁰ and PTEN is a negative regulator of Akt and Erk pathways. In this context, GH-induced Notch signaling could act as a positive feedback mechanism for GH, enhancing its neuroprotective effects. 

Upregulation of Doublecortin by GH in Retinal Neuroprotection

DCX is a neural progenitor cell marker expressed during development and adult neurogenesis, in both cases it is associated with neuronal migration, axon growth, dendritic remodeling, and synapse maturation.^91–93 DCX distribution in retinal tissue was observed in the GCL and in cells within the INL; this confirms that the marker presence is still expressed in newly formed neurons in the neonatal chick eye, which is consistent with the expression in human RGCs.⁹⁴ Excitotoxic damage changed the localization and abundance of DCX in the retina, which was mainly relocated to the OPL. KA treatment induced a DCX increase at 48 hours postinjury, at a time where GHR was also upregulated in response to damage, leading to the idea that an emergency mechanism is activated in response to neurotoxicity.¹² Recently, new actions for the DCX family have been associated to cell survival and regeneration,⁹⁵ therefore it is likely that GH neuroprotective effect include DCX actions. This is also supported by GH and DCX colocalization because in the retina GH binding to GHR produces a complex that is subsequently internalized.^11,12 Rather than generating new neurons, GH may be supporting a neurotrophic action in surviving neurons. 

In summary, actions of GH in the central nervous system involve the participation of several signaling pathways such as PI3K/Akt, MAPK, and as we demonstrated in this work, Notch signaling. Interestingly, GH-induced Notch signaling is involved in neuroprotection through Müller glia. Also, we observed that GH negatively regulated the Müller cell transdifferentiation, which likely preserves glia function to support the damaged retina. Additionally, we found anti-inflammatory effects of GH through modulation of TNF signaling. Thus, GH may support survival of existing retinal neurons rather than promoting the formation of new neurons. This work has corroborated and added additional lines of evidence that may contribute to the potential use of GH as a novel therapy for retinopathies that involve neurodegeneration and excitotoxic damage. 

Acknowledgments

The authors thank Pilgrim's Pride, Querétaro, México, who donated fertilized eggs used during this research. They also thank Gerardo Courtois (lab assistant), Nydia Hernández Rios (confocal microscopy), and Maarten Werdler (histology) at the INB-UNAM Microscopy Unit, and Anaid Antaramián and Adriana González Gallardo at the INB-UNAM Proteogenomic Unit. 

Supported by Programa de Apoyo a Proyectos de Investigación e Innovación Tecnológica de la Dirección General de Asuntos del Personal Académico (PAPIIT-DGAPA) from the Universidad Nacional Autónoma de México (UNAM; México City, México) (IN201817, IN207018, IN208419) and Consejo Nacional de Ciencia y Tecnología (CONACYT, México City, México) (285004) and by a grant from the Natural Sciences and Engineering Research Council (NSERC; Edmonton, Alberta) of Canada. 

Disclosure: T. Fleming, None; J.E. Balderas-Márquez, None; D. Epardo, None; J. Ávila-Mendoza, None; M. Carranza, None; M. Luna, None; S. Harvey, None; C. Arámburo, None; C.G. Martínez-Moreno, None