Expression of HIF-1α and Robo4 in the retinas of streptozotocin-induced diabetic mice

Abstract

Purpose: This study investigated the expression of HIF-1α and Robo4 in the retinas of streptozotocin (STZ)-induced mice and determined the expression correlation of these two factors in early diabetic retinas in vivo. Methods: A high-fat diet together with STZ stimulated type 2 diabetes mellitus (DM). HE staining was used to observe the morphologic features of the retinas following 4, or 8 weeks of hyperglycemia. Immunofluorescence was carried out to analyze the expression of HIF-1α and Robo4 in the retinas at different time points. HIF-1α and Robo4 mRNA and protein expressions were quantified by real-time PCR and western blot. Results: The arrangements of the retinal nerve fiber layer (NFL) and the ganglion cell layer (GCL) were slightly turbulent in the 4-week old diabetic mice, which became aggravated by NFL edema and cytoplasmic vacuoles in the 8-week old group. In the 4-week old group, HIF-1α was expressed slightly higher in NFL and GCL, and Robo4 expression increased in NFL and GCL. In the 8-week old diabetic retinas, HIF-1α expression was enhanced in NFL, GCL, and the outer plexiform layer (OPL); Robo4 expression increased apparently in NFL and GCL. HIF-1α and Robo4 mRNA and protein expressions were also increased slightly in the 1-week old retinas and significantly after 4 and 8 weeks. Conclusions: With aggravating retina structure turbulence in DM mice, both HIF-1α and Robo4 expressions were increased and mainly concentrated in the GCL, INL, and OPL, suggesting a regulatory role of HIF-1α on Robo4 and their combined effect on DM retina damage in vivo.

Introduction

As the most common microvascular complication of diabetes mellitus (DM), diabetic retinopathy (DR) has been identified as the primary cause of blindness in the working-age population [1]. Several vascular, inflammatory, and neuronal mechanisms are involved in the pathogenesis of DR, making the involved mechanisms complicated and multifactorial, in which hypoxia is often implicated as a key player. As an oxygen-sensitive transcription factor, HIF-1α plays an essential role in systemic responses to hypoxia and targets vascular endothelial growth factor (VEGF), erythropoietin (EPO), heme oxygenase-1 (HO-1), and glucose transporter-1 (Glut-1), which has been found to be associated with angiogenesis and fibrovascular membrane (FVM) development [2-4].

Robo4, which belongs to the Robo family, is specifically expressed in the vasculature and upregulated at sites of angiogenesis [5,6], and is essential in preventing blood vessel leakage [7]. Furthermore, studies have confirmed that Robo4 is expressed on the FVM, indicating that Robo4 may play a part in the development of FVM, as well as in the physiological functions in retina cells [8]. Robo4 expression was upregulated only in endothelial cells when exposed to hypoxia in vitro [9], in accordance with the theory that Robo4 is a gene regulated by hypoxia. In previous studies [15], we found that HIF-1α and Robo4 were colocalized in the FVM of PDR patients, which suggests that HIF-1α and Robo4 may play a role in the formation of the FVM. Furthermore, we found that Robo4 is positively regulated by HIF-1α under normoxic and hypoxic conditions in microvascular endothelial cells in vitro [10]. To our knowledge, the in vivo relationships between HIF-1α and Robo4 have not been reported.

Given our earlier finding that Robo4 is positively regulated by HIF-1α in vitro, the aim of the present study was to investigate the dynamic changes of HIF-1α and Robo4 expressions in the retina of STZ-induced DM mice, in order to probe the correlation between the expression of the two factors and retinal morphological changes in vivo.

Material and methods

Animals

All animal experiments were performed in accordance with the Association for Research in Vision and Ophthalmology’s (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research and approved by the Ethics Committee of the Second Hospital of Jilin University, Jilin University, Changchun, China.

Forty male Bal b-c mice were obtained from the Experimental Animal Center of Jilin University in Changchun, China. The animals were housed individually in standard cages at 24±2°C, with 14-16 air changes per hour, a relative humidity 50-60%, and a 12 h light/dark cycle.

Modeling and grouping

Forty mice were fed adaptively for one week and divided into control and diabetes groups. The control group (CON) was given ordinary feed and free access to water. The DM group was provided with high-lipid pellet feed [11]. After 6 weeks, STZ (Sigma, St. Louis, MO, USA, 60 mg/kg, in 0.1 mol/L pH=4.5 sodium citrate buffer) was injected intraperitoneally to the DM mice once a day for 5 days. The levels of glucose in the blood and urine of the modeling rats were tested 48 hours after injection. The modeling was evaluated as successful when the blood glucose concentration was greater than 16.7 mmol/L. The unsuccessful mice were injected once again and tested with the same method to eliminate the animals which did not meet the standard. The control group (CON) were injected with the same volume of a sodium citrate buffer. After successful modeling, the mice in the two groups were fed a normal diet.

Tissue preparation and (hematoxylin and eosin) HE staining

Five mice in each group were randomly chosen and euthanized with an overdose of sodium pentobarbital after STZ or sodium citrate buffer treatment for 4 weeks and 8 weeks. Both eyes were enucleated and fixed with a Bouin’s solution (75% saturated water solution of picric acid, 20% formaldehyde, 5% acetic acid) for 24 hours at room temperature.

The fixed eyes were dehydrated using graded ethanol and embedded in paraffin. A microtome (RM2245, Leica, Heidelberg, Germany) was used to make 4-µm thick horizontal sections through the optic nerve head. The tissues were stained with H&E and examined for morphometry. Images were taken through a light microscope (IX71, Olympus, Kyoto, Japan).

Immunohistochemistry

The sections were deparaffinized, dehydrated in xylol and ethanol, and immersed in 0.1 M citrate buffer (pH 6) under microwave heating for 18 minutes for antigen recovery. After cooling at room temperature for 20 minutes, the sections were washed in PBS, followed by a 30 minute blockade with a blocking buffer. After overnight incubation at 4°C with primary antibodies in PBS-BSA, the sections were washed in PBS and incubated for 30 minutes in the dark with secondary rabbit antibody in PBS-BSA. After being washed in PBS, the sections were stained with Hoechst 33342 (1:5000 dilutions, Sigma) for 3 minutes and mounted with glycerol. Images of the retinas were taken with a confocal microscope (FV-1000, Olympus, Japan). The primary antibodies were anti-HIF-1α (Abcam, London, UK) or anti-Robo4 (Abcam, London, UK), both in 1:100 dilutions. The secondary antibody was Cy3-conjugated goat anti-rabbit IgG (Invitrogen, Carlsbad, CA, USA) in a 1:200 dilution.

Real-time PCR

Total RNA was extracted from mouse retinal tissues using a RNAiso kit (TakaRa, Dalian, China), according to the manufacturer’s protocol, and the RNA concentration and purity were measured using a NanoDrop 2000c spectrophotometer (Thermo Fisher Scientific, Germany). RNA samples with an A260/A280 value of 1.8~2.0 were used for further analysis and the integrity of the RNA samples was assessed by 1% agarose-gel electrophoresis. Total RNA (2 μg) was reverse transcribed with a first-strand cDNA Synthesis Kit (Thermo Fisher Scientific, China). No template control was included in either set of samples. The real-time quantitative polymerase chain reaction (PCR) assays were performed in triplicate on a Light Cycler 480II (Roche Diagnostics, Basel, Switzerland) using a reaction mixture that contained 1 μL of the cDNA, 10 μmol/L gene-specific primers and 10 μL of 2× Fast SYBR Green Master Mix (Roche Diagnostics). All primers were synthesized by Shanghai Sangon Biological Engineering Technology & Services Corporation and the sequences are listed in Table 1. Both HIF-1α and Robo4 were normalized to GAPDH expression using the 2^-ΔΔCt method.

Table 1.

Gene subtype oligonucleotide primers

Gene subtype	Oligonucleotide primers (5’-3’)
Mouse HIF-1α	Forward:	GTATTATTCAGCACGACTT
	Reverse:	GACATTGCCAGGTTTAT
Mouse Robo4	Forward:	GTGGAAAGACGGGAAACC
	Reverse:	AATGCGAACAGCCAGAAG
GAPDH	Forward:	TACCCCCAATGTGTCCGTC
	Reverse:	GGTCCTCAGTGTAGCCCAAG

Western blot analysis

Protein extracts were collected with a tissue-lysis buffer (Dingguo Changsheng, Beijing, China) and the concentrations were determined using a bicinchoninic acid protein assay kit (Beyotime, Jiangsu, China). The proteins were electrophoresed on 8% SDS polyacrylamide gels, and then transferred onto polyvinylidene difluoride membranes (Invitrogen, USA) and blocked in 5% skim milk for 1 h. Primary antibodies were incubated at 4°C overnight at the following dilutions: anti-Robo4 (1:1000, Abcam, UK), anti-HIF-1α (1:1000; Abcam, UK) and β-actin (1:1000; Hangzhou Goodhere Bio Co., China), and then incubated with secondary antibodies (1:5000; Boster, Wuhan, China) for 40 min. Immunoreactive bands were visualized with an enhanced chemiluminescence plus kit (Millipore, Billerica, MA, USA), and the densities of the grey bands were determined using ImageJ software (National Institutes of Health, Bethesda, MD, USA). Western blots were repeated three times and qualitatively similar results were obtained.

Statistics

All statistical analyses were performed using SPSS 13.0 software (Chicago, IL, USA). Numerical data were presented as a mean and standard deviation (mean ± SD). Differences between two groups were analyzed using Student’s t test with P<0.05 considered a significant difference.

Results

Diabetic mouse model induced by STZ

Of the 20 mice in the DM group, 16 reached the standards, while 4 were injected once. Finally, two mice were removed because of failed induction and one because of recovery from hyperglycemia after 4 weeks.

All the diabetic mice had the typical symptoms of DM: polyphagia, polyuria, polydipsia, loss of body weight, and growth retardation at the same time. Five mice from each group were randomly chosen to compare body weight and the concentrations of urine glucose and blood glucose at 4 and 8 weeks. The diabetic mice had a significantly lower weight than the control group (t=4.4609, P<0.01 and t=6.385, P<0.01 at 4 and 8 weeks, respectively). At the two time points, the levels of blood glucose were higher than 16.7 mmol/L; both were statistically higher than those of the control mice (P<0.005) (Table 2 and Figure 1).

Table 2.

Body weight and blood glucose of mice in DM and CON

	Course of disease	DM	CON	t	P
Body weight (g)	4 w	22.224±2.384	28.812±2.285	4.4609	<0.01
	8 w	22.745±3.384	33.452±1.615	6.385	<0.01
Blood glucose (mmol/L)	4 w	24.62±8.319	9.38±0.563	4.087	<0.005
	8 w	23.15±8.339	7.1±1.508	4.235	<0.005

DM: diabetes group; CON: control group; n=5.

Figure 1.

Body weight and blood glucose of mice in DM and CON. The diabetic mice had a significantly lower weight than the control group (t=4.4609, P<0.01 and t=6.385, P<0.01 at 4 and 8 weeks, respectively). At the two time points, the results of blood glucose were higher than 16.7 mmol/L; both were statistically higher than those of the control mice (P<0.005). DM: diabetes group; CON: control group.

Morphology features of the retinas of the diabetic mice

In the control group, each layer of the retina was shown to have a morphologically normal and complete structure (Figure 2 CON). Four weeks after diabetes was induced, the DM retina structure remained complete, while the arrangements of the retinal nerve fiber layer (NFL) and the ganglion cell layer (GCL) were slightly disordered; the disarrangement of the NFL and GCL was associated with worsening NFL edema and cytoplasmic vacuoles 8 weeks later (Figure 2 DM).

Figure 2.

H&E staining of paraffin sections of the retinas in the DM and CON (×40). CON: control group, each layer of the retina was shown to have a morphologically normal and complete structure. DM (4 w): 4 weeks later, the retina structure remained complete, while the arrangements of retinal NFL and GCL were slightly turbulent; DM (8 w): disarrangement of the NFL and GCL were aggravated by NFL edema and cytoplasmic vacuole (black arrow). NFL-nerve fiber layer, GCL-ganglion cell layer, IPL-inner plexiform layer, INL-inner nuclear layer, OPL-outer plexiform layer, ONL-outer nuclear layer, IS-inner segments, OS-outer segments, RPE-retinal pigment epithelium, CON: control group, DM: diabetes group.

Expression of HIF-1α and Robo4 in the STZ-induced diabetic mice

With an immunofluorescent assay, HIF-1α was found in all the layers of the control retinas at different levels. Some were at a high level, and some were at a low level, which indicated that HIF-1α was expressed in all the layers of the retina (Figure 3A-C). In the retinas of the DM group at 1 week, the expression of HIF-1α and Robo4 was not obviously changed (the data are not shown). In the 4-week diabetic mice, the expression of HIF-1α was nearly unchanged except for a higher expression in NFL and GCL; as time passed, fluorescence of HIF-1α was enhanced in the two layers (Figure 3D-I). Robo4 was weakly positive in the NFL, GCL, and OPL of the normal retinas (Figure 4A-C). Four weeks after DM was induced, expression increased in NFL and GCL; in the 8-week diabetic retinas, the fluorescence had apparently increased (Figure 4D-I).

Figure 3.

Expression of HIF-1α in the control and diabetic retina (×40). (A-C) CON, HIF-1α positively expressed in all layers of the control retina (B and C); (D-F) in the 4-week diabetic mice, expression of HIF-1α was nearly unchanged except a higher expression in NFL (arrow in E and F); (G-I) fluorescence of HIF-1α enhanced in NFL and GCL (arrow in H and I). GCL-ganglion cell layer, INL-inner nuclear layer, ONL-outer nuclear layer, CON: control group, DM: diabetes group.

Figure 4.

Expression of Robo4 in the control and diabetic retina (×40). (A-C) CON, Robo4 was weakly positive in the NFL, GCL, and OPL of normal retinas (B and C, arrow). (D-F) Four weeks after DM was induced, expression increased in the NFL and GLC (E and F, arrow). (G-I) in 8-week diabetic retinas, the fluorescence was apparently increased (H and I, arrow). CON-control group, DM-diabetes group, GCL-ganglion cell layer, INL-inner nuclear layer, ONL-outer nuclear layer.

Based on real-time PCR results, relative levels of HIF-1α and Robo4 mRNA in the DM mice were slightly increased in the 1 week group (1.32±0.07 fold, 1.19±0.06 fold), and markedly increased in the 4 week and 8 week groups (1.43±0.13 fold and 2.38±0.15 fold, 1.51±0.13 fold and 2.37±0.19 fold) when compared with the levels in the CON groups (P<0.005; Figure 5A) and increased during the interval of 1 week to 8 weeks. The protein expressions of HIF-1α and Robo4 were also confirmed by western blot and showed similar trends in the mRNA expression. Both the HIF-1α and Robo4 protein expression levels in the DM group were found to be significantly higher after 4 and 8 weeks induction than the levels in the control groups, and they were slightly increased after 1 week (Figure 5B).

Figure 5.

mRNA and protein expression of HIF-1α and Robo4 in the retina. A: Relative levels of HIF-1α and Robo4 mRNA in the DM mice were slightly increased in the 1 week group (1.32+0.07 fold, 1.19+0.06 fold), and markedly increased in the 4 week and 8 week groups (1.43+0.13 fold and 2.38+0.15 fold, 1.51+0.13 fold and 2.37+0.19 fold) when compared with the levels in the CON groups and increased as time passed by from 1 week to 8 weeks. B: Protein expression of HIF-1α and Robo4 was confirmed by western blot. This showed that both HIF-1α and Robo4 protein expression levels in the DM group were significantly higher after 4 and 8 week induction than in the control groups, and slightly increased after 1 week. **: P<0.05, compared to the CON group of same time period; ##: P<0.05, compared to the DM group of 1 week STZ induction.

Discussion

DM is a group of metabolic diseases characterized by hyperglycemia caused by insufficient insulin secretion and/or insulin resistance. DR is one of the most common microvascular complications of DM and is thought to be closely related to the extended course of the disease and poor glycemic control. For an in vivo study of DR, a high-fat diet together with an intraperitoneal injection of STZ can be applied, and more than six months after induction, diabetic rats had fundus changes similar to DR in humans [11,12]. Therefore, in this study, the STZ-induced mice with different time periods were used to monitor the expression of HIF-1α and Robo4 in the retina and to analyze the role of the two factors as well as the correlation between them in the early stages of DR.

In this study, the arrangement of the NFL and GCL was found to be slightly turbulent after 4 weeks and aggravated at 8 weeks, with NFL edema and cytoplasmic vacuoles in the mouse model of DM. As shown in the previous study [13], the number of apoptotic cells increased in mouse DM induced by STZ for one month; most of the apoptotic cells, which were not endothelial cells, were shown to be neurons or glial cells. In the DM patients, increasing evidence suggests that retinal GCL undergoes functional changes before microvascular shifts are detected, thereby indicating retinal GCL degeneration in the early stages of DR. These changes were also identified in our study on 8-week DM mice, namely that a disarrangement of the NFL and GCL was detected with NFL edema and cytoplasmic vacuoles.

Consistent with recent research [14], HIF-1α was expressed in all the layers of normal mouse retinas. HIF-1α is a transcriptional factor related to hypoxic reaction, that can be widely involved in the regulation of energy metabolism, cell proliferation, angiogenesis, and other processes [15]. Under hypoxic conditions, angiopoietin-related protein-4 can be up-regulated by the HIF-1α signaling pathway, and can promote vascular permeability in retinal cells [16]. In a hyperglycemic environment, it can improve vascular endothelial growth factor (VEGF) expression and promote angiogenesis [17,18]. In this study, the increasing expression trend of HIF-1α, especially in GCL at 4 weeks and in GCL together with OPL at 8 weeks, indicated the hypoxic state of the retinas, and aggravation at 8 weeks led to HIF-1α activation and pathologic changes in those layers of the retinas.

Research reveals that nerve and blood vessels share similar patterns of migration and development guidance mechanisms. Several identified gene families play a role in nerve guidance and are also involved in regulating endothelial cells and the budding of new blood vessels. These nervous system guidance axons such as semaphorins, neuropilin, ephrins/ephs, notch and delta, are also known to regulate vessel divergence [19-21]. Members of the Robo family, such as Robo1, Robo2, and Robo3, are expressed mainly in the nervous system and play important roles in regulating nerve axons extension and nerve cell migration. The new member, Robo4, (discovered in 2002) has been found to be expressed specifically on the surface of the endothelial cells of active angiogenesis. Robo4 has certain inhibitory effects on endothelial cell migration [5]. Except for the surface of endothelial cells, Robo4 is also expressed in fibrous tissue and nerve tissue. In Bedell’s study [22], Robo4 was observed to be expressed in the nerve tissue of zebrafish. Huang et al. [8] found Robo4 expressed in fibrous tissues and neovascular endothelial cells of fibrovascular membranes of PDR patients. Robo4 was partly colocalized with GFAP, indicating that Robo4 expression was not limited to blood vessels, but was also present in nerve tissue. During the development stages of immature rats, Robo4 was expressed not only in retinal blood vessels but also in the RGC layer and photoreceptors. After development, the expression on these layers of the retina is significantly decreased [23]. Consistent with recent studies, we found Robo4 to be weakly expressed in the nerve layers of the normal retina. The low distribution of Robo4 in adult mouse retinas increased at 8 weeks of DM in NFL and GCL, which may be related to the important role of Robo4 in early nerve damage.

More importantly, the overexpression sites of both HIF-1α and Robo4 were on the NFL and GCL of the retinas, which was consistent with the retinal parts of the pathological changes on HE staining. HIF-1α and Robo4 may interact with each other to participate in nerve changes during the early stages of DR, and the up-regulation of Robo4 may result from the hypoxia-activated HIF-1α expression and stabilization. In our previous study [10], Robo4 was shown to be transcriptionally upregulated by HIF-1α under normoxic and hypoxic conditions to promote the invasion and proliferation of human microvascular endothelial cells. In the present study, we found that both of the proteins were concentrated in the GCL, INL, and OPL, suggesting a regulatory role of HIF-1α on Robo4 and their combined effects on retina damage in DM.

Taken together, the comparative analysis of HIF-1α and Robo4 expression changes in the retinas of different courses of the DM model showed these proteins increasing in the early stage of diabetes (4 and 8 weeks) in the NFL and the ganglion cells. In addition, their expressions were in similar parts of the retina, and this may be due to progressive hypoxia of the diabetic retina, which stimulated the transcriptional upregulation of Robo4 by HIF-1α.

Disclosure of conflict of interest

None.