Differential Gene Expression Profiling after Conditional Müller-Cell Ablation in a Novel Transgenic Model

Abstract

Purpose.: Müller cells, the principal glial cells in the mammalian retina, play an important role in the maintenance of retinal homeostasis. Recent reports suggest that Müller-cell dysfunction may contribute to the pathogenesis of retinal diseases such as idiopathic macular telangiectasia type 2. In the present study, we used microarray to compare retinae isolated from transgenic mice in which the Müller cells of adult mice retinae can be selectively ablated with control mice.

Methods.: Retinae were isolated 1 week, 1 month, and 3 months after tamoxifen-induced selective Müller-cell ablation and microarray were performed with Affymatrix microarrays. Differentially expressed (DE) genes, temporal trends of DE genes, and pathway analysis were conducted. Quantitative real-time polymerase chain reaction (qRT-PCR) was performed to validate the results.

Results.: Strong upregulation of mRNA of proteins involved in gliosis, apoptosis, and neurotrophism was found 1 week after ablation and their related pathways such as the apoptotic and Jak/Stat pathways were identified. Three months after induced Müller-cell ablation, Müller-cell metabolic pathways and vasculopathy-related pathways such as genes involved in glycolysis and tight junctions were downregulated. qRT-PCR analysis showed consistent expression trends of selected genes.

Conclusions.: The results were generally consistent with the previous morphologic findings in this model, in which photoreceptor degeneration soon after Müller-cell ablation, accompanied by blood–retinal barrier breakdown and subsequent retinal neovascularization were reported. These results are consistent with a significant contribution of Müller-cell dysfunction on retinal neuronal injury and vascular pathology at the mRNA level.

Introduction

Müller cells are the principal glial cells of the mammalian retina. They provide metabolic support and nutrition to neurons, ¹ release neuroactive substances through neurotransmitter recycling systems, ² release vasoactive substances such as vascular endothelial growth factor (VEGF) and pigment epithelium derived factor (PEDF), ^3,4 maintain the blood–retina barrier (BRB), ⁵ and regulate ion and water content and pH of the retina. ⁶ Since these functions are essential for retinal homeostasis, dysfunction of Müller cells may play a role in the pathogenesis of retinal diseases. 

Abnormality of Müller cells has been linked to some retinal diseases such as idiopathic macular telangiectasia type 2 (MacTel Type 2). ^7,8 Histopathologic observations on postmortem tissue donated by an affected patient revealed reduced immunoreactivity of Müller-cell markers such as vimentin, retinaldehyde binding protein 1 (RLBP1), and glutamine synthetase (GS) in the central macula. ⁷ Also, proteomic analysis of the fellow eye found that Müller-cell–associated proteins such as glial fibrillary acidic protein (GFAP), vimentin, and GS were reduced in the diseased macula region. ⁸ Interestingly, proteins involved in the glycolytic pathway, one of the most important components of Müller-cell metabolism, ^9–11 were also downregulated in the macula of the diseased retina. ⁸  

To investigate the potential role of Müller-cell dysfunction in the pathogenesis of retinal diseases, we have recently generated an inducible transgenic model using a portion of the regulatory region of the retinaldehyde binding protein 1 (Rlbp1) gene as Müller-cell–specific promoter along with a Cre/Lox-P approach for Müller-cell–specific gene targeting. ¹² In this model, crossing the resultant Rlbp1-Cre Estrogen Receptor (ER) transgenic mice with a transgenic line carrying an attenuated form of the diphtheria toxin fragment A (DTA176) gene (Rosa-DTA176 mice), produced selective, patchy ablation of Müller cells after treatment with tamoxifen (TMX) (Rlbp-CreER-DTA176 mice). ¹² The patchy loss of Müller cells began from 1 day after TMX induction and became relatively stable at 14 days after TMX injection. ¹² The transgenic model developed distinctive morphologic changes including photoreceptor apoptosis, BRB breakdown, and, later, deep retinal neovascularization, which did not occur in the control group (Rlbp1-CreER-LacZ mice). ¹² This transgenic model provides a unique opportunity to study the effect of Müller-cell dysfunction on retinal neuronal damage and vascular pathology. Thus, the aim of the present study was to profile genomic changes after selective Müller-cell ablation, to examine differentially expressed genes and novel pathways that may lead to retinal neuronal damage and vascular pathology at different times after Müller-cell ablation. 

Materials and Methods

Transgenic Mice with Conditional Müller-Cell Ablation

We used the regulatory region of the retinaldehyde binding protein 1 (Rlbp1) gene as a Müller-cell–specific promoter along with the Cre/Lox-P system to produce transgenic mice for Müller-cell–specific gene targeting. ^12,13 Rlbp1-CreER mice were crossed with Rosa-DTA176 mice, which resulted in Rlbp-CreER-DTA176 transgenic mice that were suitable for conditional Müller-cell ablation. ¹² Those crossed with Rosa-LacZ reporter mice (Rlbp-CreER-LacZ) ¹⁴ were used as controls in this study. DTA176 gene expression was induced by daily intraperitoneal injection of TMX (3 mg in 0.2 mL sunflower oil) for 4 consecutive days at around 6 to 8 weeks of age. 

Tissue Collection

All animal experiments adhered to the Association of Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research. The project was approved by the Animal Ethics Committee of the University of Sydney. All animals were kept in 12-hour dark and light cycles and fed with standard chow and water. Twenty-four Rlbp-CreER-DTA176 mice and 24 Rlbp-CreER-LacZ mice were used. Mice were euthanized with carbon dioxide followed by cervical dislocation at 1 week, 1 month, and 3 months after TMX induction (n = 8/group/time point). Their retinas were isolated, snap-frozen in liquid nitrogen immediately, and stored at −80°C until use. Four retinae from Rlbp-CreER-DTA176 mice were of insufficient quality for analysis. Thus, a total of 44 retinae were analyzed: 20 Rlbp-CreER-DTA176 mice and 24 Rlbp-CreER-LacZ mice. 

RNA Extraction and cDNA Preparation

Retinae were thawed in RLT buffer and homogenized with a pellet pestle (Z359971; Sigma-Aldrich; St. Louis, MO). RNA extraction was performed with a commercial kit (RNeasy Mini Kit [catalog 74104]; Qiagen Sciences, Inc., Germantown, MD) according to the manufacturer's instructions. RNA extract (50 μL) was eluted in RNase-free water. The quality and quantity of RNA were assessed with a bioanalyzer (Agilent 2100 Bioanalyzer; Agilent Technologies, Santa Clara, CA) and 200 ng of total RNA was amplified (Applause WT-Amp ST RNA Amplification System [catalog 5500‐24]; NuGEN Technologies, San Carlos, CA) according to the manufacturer's instructions. cDNA was purified with a reaction clean-up kit (MiniElute Reaction Clean-Up Kit [catalog 28204]; Qiagen Sciences), and fragmentation and labeling were performed (Encore Biotin Module [4200‐12]; NuGEN Technologies) according to the manufacturer's instructions. 

Microarray and Data Analysis

DNA microarrays were performed with semiconductor manufacturing techniques (GeneChip Operating System [GCOS]; Affymetrix, Santa Clara, CA). Fragmented and biotin-labeled cDNA samples were mixed with a hybridization master mix (AFX-900454; Affymetrix) and 90 μL of sample and hybridization master mix mixture were injected into gene chips (Genechip Mouse Gene 1.0 ST array [AFX-901169]; Affymetrix) and hybridized for 18 hours at 45°C. Samples were removed from the chips after hybridization, and the chips were washed at a GCOS fluidics station (Affymetrix). With 100 μL of array holding buffer, chips were scanned with GCOS fluidics software. The scanned data were collected for analysis. For quality control analysis, a probe level model (PLM) and a robust multichip average (RMA) were used. Differential expression (DE) probes were selected using a least-squares approach applied by a linear model for microarray data (limma) package. A Bonferroni adjustment of P values was used for a stringent analysis of DE (P < 0.05). Gene set enrichment was used based on the approaches in the package topGO, ¹⁵ and GST from limma to identify differentially expressed gene ontology terms. R package “goTools” was also used to identify DE genes involved in each GO term and pathways identified by KEGG (http://www.genome.jp/kegg/pathway.html). 

Ratio to ratio (RR) statistics were used to estimate differential temporal profiles between Rlbp-CreER-DTA176 and Rlbp-CreER-LacZ groups. This value, which can be thought of as the ratio of the ratio between two factors (time and strain), is produced by RR = (Rlbp1-CreER-DTA176/Rlbp1-CreER-LacZ at 3 months)/(Rlbp1-CreER-DTA176/Rlbp-CreER-LacZ at 1 week). In a log-scale, this equates to the difference of the difference between expression values. This is not the same as examining temporal differences between 3 months and 1 week, but we adjusted the values to the expression levels with Rlbp-CreER-LacZ control mice for each time point. 

Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR)

qRT-PCR was performed to validate microarray results. We chose those highly differentially expressed genes and the genes that may have accounted for the retinal neuronal damage and vascular pathology that we had observed in this transgenic model. ⁹ The primer sequences of these genes are listed in Table 1. Reverse transcription was performed using a cDNA synthesis kit (SuperScript VILO synthesis kit [catalog 11754050]; Invitrogen Corp., Carlsbad, CA) according to the manufacturer's instructions. Diluted cDNA (4 μL) (50 ng cDNA) and 6 μL of primer and supermix mixture (EXPRESS SYBR GreenER qPCR Supermix Universal [catalog A10314]; Invitrogen Corp.) were used in each PCR reaction. A free-download, stand-alone software tool (Relative Expression Software Tool [REST] 2009; M. Pfaffl software; Technical University Munich, Munich, Germany) was used for data analysis. ¹⁶ Expression values were normalized with three reference genes: 18SrRNA, β-actin, and β-tubulin. 

Table 1. 

 

Information on Primers Used for qRT-PCR Analysis

Gene Symbol	Primer Sequence (5′→3′)		Amplicon Size
CXCL13	Forward	TTGTGTAATGGGCTTCCAGA	110
	Reverse	AGGTTGAACTCCACCTCCAG
CXCL10	Forward	CCAAGTGCTGCCGTCATTTTC	133
	Reverse	TCCCTATGGCCCTCATTCTCA
EDN2	Forward	AGACCTCCTCCGAAAGCTG	64
	Reverse	CTGGCTGTAGCTGGCAAAG
A2M	Forward	AGATGGTGAGATTTCGTGTTGTC	220
	Reverse	ACGGTCCTGCCTGATTCTGTA
FGF2	Forward	TGTGTCTATCAAGGGAGTGTGTGC	158
	Reverse	ACCAACTGGAGTATTTCCGTGACCG
KLK22	Forward	GTGTGTGTCCATCAAGCTCCATCC	131
	Reverse	GGCCTCCTGAGTCTCCCTTACAA
LIF	Forward	AATGCCACCTGTGCCATACG	216
	Reverse	CAACTTGGTCTTCTCTGTCCCG
STAT1	Forward	TCACAGTGGTTCGAGCTTCAG	155
	Reverse	GCAAACGAGACATCATAGGCA
STAT3	Forward	TGCGGAGAAGCATTGTGAGTG	159
	Reverse	TTTTCCAGACGGTCCAGGCAGATG
GFAP	Forward	TCCTTCCAAGGTTGTCCATC	198
	Reverse	CCAATCAGCCTCAGAGAAGG
GAPDH	Forward	AAGATGGTGATGGGCTTCCCG	156
	Reverse	TGGCAAAGTGGAGATTGTTGCC
PGM2	Forward	CCGCTTCTACATGACCGAGG	121
	Reverse	GATGATGCAAGATACGGCAGG
ENO1	Forward	GCTGCCTCCGAGTTCTACAG	325
	Reverse	GCAGGGATTCGGTCACAGAG
ENO2	Forward	ATCAGCTCAGGTATCTCCGTG	167
	Reverse	TGGCGATAGAGGGGCAAGT
LDHa	Forward	TGTCTCCAGCAAAGACTACTGT	155
	Reverse	GACTGTACTTGACAATGTTGGGA
LDHb	Forward	AGATCACTGTAGTGGGCGTTG	164
	Reverse	TTTCGGAGTCTGGAGGAACAA
Claudin5	Forward	GCAAGGTGTATGAATCTGTGCT	109
	Reverse	GTCAAGGTAACAAAGAGTGCCA
ZO-1	Forward	GCCGCTAAGAGCACAGCAA	172
	Reverse	GCCCTCCTTTTAACACATCAGA
VE-cad	Forward	AGGACAGCAACTTCACCCTCA	70
	Reverse	AACTGCCCATACTTGACCGTG
Occludin	Forward	TTGAAAGTCCACCTCCTTACAGA	129
	Reverse	CCGGATAAAAAGAGTACGCTGG
β-Actin	Forward	GGCTGTATTCCCCTCCATCG	154
	Reverse	CCAGTTGGTAACAATGCCATGT
β-Tubulin	Forward	GATCGGTGCTAAGTTCTGGGA	154
	Reverse	AGGGACATACTTGCCACCTGT
18S rRNA	Forward	GCAATTATTCCCCATGAACG	139
	Reverse	GGGACTTAATCAACGCAAGC

Western Blot

Proteins were extracted with RIPA buffer (R0278; Sigma-Aldrich) with Complete Mini proteinase inhibitor (11836153001; Roche Applied Science, Basel, Switzerland). Protein quantification was performed with a commercial assay kit (QuantiPro BCA Assay Kit [QPBCA]; Sigma-Aldrich) and a microplate reader (Tecan Group Ltd., Männedorf, Switzerland) was used for analysis of protein concentration. Equal amounts of proteins were loaded into Bis-Tris gels (NuPage, NP0323BOX; Life Technologies, Carlsbad, CA) and transferred to a polyvinylidene difluoride (PVDF) membrane with a dry transfer system (iBlot, BI1001; Invitrogen). Membranes were blocked with 5% BSA in TBST and incubated with primary antibodies overnight at 4°C. Primary antibodies used in the experiment were Enolase 1 (catalog 3810; Cell Signaling Technology, Inc., Danvers, MA), lactate dehydrogenase a (catalog 2012; Cell Signaling Technology, Inc.), and GFAP (NeoMarkers MS-280-P; Thermo Fisher Scientific, Hampton, NH). Horseradish peroxidase conjugated secondary antibodies were used and protein bands were visualized (G:Box BioImaging system). The bands were normalized with α/β tubulin (catalog 2148s; Cell Signaling Technology, Inc.), which served as loading control. GeneTool image scanning and analysis package was used for densitometry and t-test with P < 0.05 were accepted as statistically significant data. 

Results

Summary of Differential Gene Expression, Gene Ontology, and Pathways Analysis at Different Time Points after Selective Müller-Cell Ablation

DE genes were identified at 1 week, 1 month, and 3 months after TMX-induced Müller-cell ablation. We investigated all DE genes and chose for further analysis both highly differentially expressed genes as well as those that were potentially associated with retinal neuronal damage and vascular pathology (Table 2). An overrepresentation test based on gene ontology was performed to provide a survey of various functional categories among the DE genes. The gene ontology analysis with three different GO terms (BP, biological process; MF, molecular function; and CC, cellular component) is listed in Supplementary Table S1 (see Supplementary Material and Supplementary Table S1). Also, the R package “goTool”¹⁷ was used to examine the percentage of the DE genes involved in specific GO terms in each time point (see Supplementary Material and Supplementary Figs. S1–S3). A series of GeneSetTest (GST, implemented in limma) analysis was performed for pathway analysis. We focused on a number of pathways that were closely related to retinal pathology such as glial dysfunction, photoreceptor damage, and the BRB breakdown occurring after Müller-cell ablation (Table 3). In short, we found increased transcription of genes responsible for neuronal apoptosis, neurotrophic factors, and their related pathways in the early stage of ablation, whereas the glycolytic and tight junction pathways were downregulated later. 

Table 2. 

 

Short List of Differentially Expressed Genes 1 Week, 1 Month, and 3 Months after Tamoxifen-Induced Müller-Cell Ablation

Gene Symbol	Fold Change, log2	Gene Name	NCBI Reference Sequence
1 wk after tamoxifen-induced Müller-cell ablation
CXCL10	4.21	Chemokine (C–X–C motif) ligand 10	NM_021274
KLK1B22	4.05	Kallikrein 1-related peptidase b22	NM_010114
LCN2	4.02	Lipocalin 2	NM_008491
SERPINa3n	3.98	Serine (or cysteine) peptidase inhibitor, clade A, member 3N	NM_009252
EDN2	3.06	Endothelin 2	NM_007902
CXCL13	2.95	Chemokine (C–X–C motif) ligand 13	NM_018866
A2M	2.72	Alpha-2-macroglobulin	NM_175628
FGF2	2.33	Fibroblast growth factor 2	NM_008006
GFAP	1.74	Glial fibrillary acidic protein	NM_001131020; NM_010277
KLK1B21	1.2	Kallikrein 1-related peptidase b21	NM_010642
ICAM1	1.18	Intercellular adhesion molecule 1	NM_010493
STAT3	1.17	Signal transducer and activator of transcription 3	NM_011486; NM_213659; NM_213660
CHL1	1.01	Cell adhesion molecule with homology to L1CAM	NM_007697
STAT1	0.89	Signal transducer and activator of transcription 1	NM_009283
VCAM1	0.84	Vascular cell adhesion molecule 1	NM_011693
LIF	0.82	Leukemia inhibitory factor	NM_001039537; NM_008501
CD44	0.78	CD44 antigen	NM_001039150; NM_001039151; NM_009851
SOCS3	0.64	Suppressor of cytokine signaling 3	NM_007707
KCNG4	0.47	Potassium voltage-gated channel, subfamily G, member 4	NM_025734
PIK3AP1	0.46	Phosphoinositide-3-kinase adaptor protein 1	NM_031376
AQP1	−0.44	Aquaporin 1	NM_007472
PGM2	−0.62	Phosphoglucomutase 2	NM_028132
1 mo after tamoxifen-induced Müller-cell ablation
KLK1B22	5	Kallikrein 1-related peptidase b22	NM_010114
CXCL13	1.85	Chemokine (C–X–C motif) ligand 13	NM_018866
KLK1B21	1.71	Kallikrein 1-related peptidase b21	NM_010642
EDN2	1.61	Endothelin 2	NM_007902
FGF2	1.49	Fibroblast growth factor 2	NM_008006
GAPDHS	−0.63	Glyceraldehyde-3-phosphate dehydrogenase, spermatogenic	NM_008085
PGM2	−0.74	Phosphoglucomutase 2	NM_028132
AVEN	−0.56	Apoptosis, caspase activation inhibitor	NM_001165935; NM_028844
3 mo after tamoxfen-induced Müller-cell ablation
KLK1b22	4.34	Kallikrein 1-related peptidase b22	NM_010114
KLK1b21	1.56	Kallikrein 1-related peptidase b21	NM_010642
EDN2	1.42	Endothelin 2	NM_007902
KCNH5	0.27	Potassium voltage-gated channel, subfamily H (eag-related), member 5	NM_172805
KCNJ10	−0.46	Potassium inwardly-rectifying channel, subfamily J, member 10	NM_001039484
KCNB1	−0.49	Potassium voltage-gated channel, Shab-related subfamily, member 1	NM_008420
PGM2	−0.87	Phosphoglucomutase 2	NM_028132

Table 3. 

 

Pathway Analysis Using GeneSetTest 1 Week, 1 Month, and 3 Months after Tamoxifen-Induced Müller-Cell Ablation

Pathways	Gene Set Trend
GeneSetTest (GST) pathway analysis 1 wk after Müller cell ablation
MAPK signaling pathway	Up
Chemokine signaling pathway	Up
Apoptosis	Up
Wnt signaling pathway	Up
TGF-beta signaling pathway	Up
Cell adhesion molecules (CAMs)	Up
Adherence junction	Up
Jak-Stat signaling pathway	Up
T cell receptor signaling pathway	Up
B cell receptor signaling pathway	Up
Insulin signaling pathway	Up
Phototransduction	Down
GeneSetTest (GST) pathway analysis 1 mo after Müller cell ablation
Glycolysis/Gluconeogenesis	Down
Citrate cycle (TCA cycle)	Down
d-glutamine and d-glutamate metabolism	Down
Jak-Stat signaling pathway	Up
Cell adhesion molecules (CAMs)	Up
Tight junction	Down
GeneSetTest (GST) pathway analysis 3 mo after Müller cell ablation
Glycolysis/Gluconeogenesis	Down
Citrate cycle (TCA cycle)	Down
MAPK signaling pathway	Down
Wnt signaling pathway	Down
Cell adhesion molecules (CAMs)	Up
Adherens junction	Down
Tight junction	Down
mTOR pathway	Down

One Week after Müller-Cell Ablation.

In all, 395 genes (309 up; 86 down) were identified 1 week after induced Müller-cell ablation by DE gene analysis. Of these, those related to neurotrophism, including endothelin 2 (EDN2), fibroblast growth factor 2 (FGF2), signal transducer and activator of transcription 1 (STAT1), STAT3, and leukemia inhibitory factor (LIF), were highly upregulated (Table 2). The gliosis marker, GFAP, was also upregulated. Inflammation and cell-adhesion–related genes such as chemokine (C–X–C motif) ligand 10 (CXCL10) and alpha-2-macroglobulin (A2M), intercellular adhesion molecule 1 (ICAM1), vascular cell adhesion molecule 1 (VCAM1), and CD44 were also found. Consistent with individual DE gene analysis, gene ontology analysis revealed that DE genes were mostly related to inflammation (BP), host defense mechanism (BP), cytokine and chemokine receptor interactions (MF), G-protein–coupled receptor binding (MF), protein binding (MF), cell surface (CC), and host cell cytoplasm part (CC) (see Supplementary Material and Supplementary Table S2). Pathway analysis identified upregulation of 77 pathways and 4 downregulated pathways. Among these pathways, inflammation-, apoptosis-, and neuroprotection-related pathways such as Jak/Stat pathway were upregulated, whereas phototransduction was downregulated (Table 3). 

One Month after Müller-Cell Ablation.

DE gene analysis identified 126 genes (58 up; 68 down). Strong upregulation of EDN2-, FGF2-, and KLK-related genes (KLK1B21 and KLK1B22) was observed, whereas a number of genes related to the glycolytic pathway, including spermatogenic glyceraldehyde-3-phosphate dehydrogenase (GAPDHs), glycogen synthase 1 (GYS1), and phosphoglucomutase 2 (PGM2), were downregulated (Table 2). Gene ontology analysis revealed that the DE genes were mostly associated with metabolic process (BP), regulation of mitochondrion organization (BP), mitochondrial outer membrane activity (CC), regulation of neurotransmitter level (BP), synaptic vesicle (CC), and regulation of cell division (BP) (see Supplementary Material and Supplementary Table S2). We also found 15 upregulated and 57 downregulated pathways 1 month after Müller-cell ablation. In particular, pathways likely to be associated with Müller-cell metabolism, such as glycolysis, d-glutamine, and d-glutamate metabolism and TCA cycle, were downregulated along with tight junction molecules, whereas inflammation-related pathways such as cell adhesion molecules were upregulated (Table 3). 

A significant downregulation of Müller-cell metabolism was suggested by this analysis 1 month after induced Müller-cell ablation. Two main functions of Müller cells are (1) providing energy substrate, lactate, to high energy demanding neurons for their own oxidative metabolism (glycolysis); and (2) regulation of neurotransmitters in which converting toxic glutamate to glutamine (d-glutamine and d-glutamate metabolism). The relationship between these two functions was demonstrated by Poitry et al. ¹⁸ who reported that glutamate levels were linked to lactate production by Müller cells. The reduced expression of genes involved in anerobic metabolism that we found in the present study is likely to reflect loss of Müller cells, whereas the reduction in genes involved in oxidative phosphorylation is likely due to loss of photoreceptors. 

Three Months after Müller-Cell Ablation.

In all, 123 genes (46 up; 77 down) were identified by DE analysis 3 months after Müller-cell ablation. EDN2, KLK1b22, and KLK1b21 were upregulated, whereas the glycolysis-related gene, PGM2, was dowregulated (Table 2). Ion channel–, especially potassium channel–, associated genes were differentially expressed 3 months after Müller-cell ablation, with slightly increased transcription of gene encoding potassium voltage-gated channel subfamily H (eag-related) member 5 (KCNB5), and decreased expression of potassium inwardly rectifying channel subfamily J member 10 (KCNJ10) and potassium voltage-gated channel Shab-related subfamily member 1 (KCNH1) (Table 2). Gene ontology analysis revealed that DE genes at 3 months were strongly related to ion channel activity (BP), cellular metabolic process (BP), mitochondrial function (BP), potassium channel activity (MF), mitochondrial activity (CC), and cellular membranes (CC) (see Supplementary Material and Supplementary Table S2). Fifteen pathways were upregulated and 85 were downregulated 3 months after Müller-cell ablation. Pathways associated with glycolysis, TCA cycle, mTOR signaling, Wnt signaling, and tight junction formation were all downregulated, whereas pathways related to cell adhesion were upregulated (Table 3). 

Consistent with our previous morphologic study in which immunostaining revealed deep retinal neovascularization and vascular telangiectasis, ¹² microarray analysis revealed ion channel and blood vessel–related pathways were downregulated 3 months after Müller-cell ablation. Genes associated with the glycolytic pathway remained downregulated. 

Differential Temporal Profile of DE Genes

We related DE genes and their temporal trends to retinal pathology using RR analysis, to understand how Müller-cell disruption may contribute to the progression of retinal diseases. Results were interpreted as follows: RR values > 1 indicate more changes at 3 month, whereas the RR values < 1 indicate more changes at 1 week. RR analysis identified 79 DE genes, of which 75 genes had RR values > 1 and only 4 genes had values < 1. Gene ontology from the Database for Annotation, Visualization, and Integration Discovery (DAVID; http://david.abcc.ncifcrf.gov/; Bioinformatics Resources Microarray Analysis) was applied to the group the genes identified by RR analysis. ^19,20 They were grouped into eight different annotation clusters, but mainly associated with chemotaxis (BP), apoptosis (BP), regulation of cellular biosynthetic process (BP), GTP binding (MF), and immune response (BP) (see Supplementary Material and Supplementary Table S3). 

GST analysis identified 118 pathways; those potentially associated with retinal pathology following Müller-cell ablation are listed in Table 4. Changes in pathways associated with chemokine signaling, apoptosis, and TGF-β signaling were found 1 week after Müller-cell ablation, whereas glycolysis, citrate cycle (TCA cycle), vascular endothelial growth factor (VEGF), tight junction, gap junction, and mTOR pathways were altered significantly 3 months after Müller-cell ablation. Changes in mitogen-activated protein kinase (MAPK) signaling, Wnt signaling, and cell adhesion molecule pathways were found both 1 week and 3 months after Müller-cell ablation. 

Table 4. 

 

Differential Temporal Profile (Ratio-to-Ratio Statistics) of DE Genes and Their Involvement at Different Time Points after Müller-Cell Ablation

GST Pathway Analysis on RR DE Genes	Time Point
Chemokine signaling pathway	1 wk
Apoptosis	1 wk
TGF-β signaling pathway	1 wk
Glycolysis/gluconeogenesis	3 mo
Citrate cycle (TCA cycle)	3 mo
Tight junction molecules	3 mo
mTOR pathway	3 mo
VEGF signaling pathway	3 mo
Cell adhesion molecules (CAMs)	1 and 3 mo
Wnt signaling pathway	1 and 3 mo
MAPK signaling pathway	1 and 3 mo

qRT-PCR

qRT-PCR was performed to validate microarray results (Fig. 1). We chose the genes that were highly differentially expressed or potentially had a significant functional relationship with retinal pathology. These genes are summarized in Table 5. As mentioned earlier, we excluded four samples that showed poor hybridization in the Rlbp-CreER-DTA176 group. To conduct the qRT-PCR with consistent sample sizes in each group, we chose six samples in each group with the highest RNA quality (highest RNA integrity number from the bioanalyzer). We also generated a heat map with those DE genes for a better comparison of expression levels between transgenic and control mice at each time point (Fig. 2). Expression levels of the chosen genes revealed by PCR analysis were consistent with our microarray results. 

Figure 1

 

qRT-PCR validation of DE genes of interest revealed by microarray analysis. A ratio > 1 indicates upregulation, whereas a ratio < 1 indicates downregulation in Rlbp1-CreER-DTA176 mice compared with Rlbp1-CreER-LacZ controls. (A) qRT-PCR analysis 1 week after TMX-induced Müller ablation. (B) qRT-PCR analysis 1 month after TMX induction. (C) qRT-PCR analysis 3 months after TMX induction. P < 0.001; Rlbp1-CreER-DTA176 versus Rlbp1-CreER-LacZ; n = 6 in each group.

Figure 2

 

Heat map illustration of the genes with significant biological functions listed in Table 5. LacZ-7, ‐30, and ‐90 represent Rlbp-CreER-LacZ control mice at 1 week, 1 month, and 3 months after TMX-induced Müller-cell ablation. DTA-7, ‐30, and ‐90 represent Rlbp-CreER-DTA176 transgenic mice at 1 week, 1 month, and 3 months after Müller-cell ablation. Genes associated with reactive neuroprotection and gliosis were significantly upregulated in Rlbp-CreER-DTA176 groups, whereas Pgm2, a gene involved in glycolysis, was dowregulated through all three time points in Rlbp-CreER-DTA176 groups (n = 6–8/group/time point). (Fgf2, fibroblast growth factor 2; A2M, alpha-2-macroglubulin; CXCL13, chemokine (C–X–C motif) ligand 13; Edn2, endothelin 2; Gfap, glial fibrillary acidic protein; Pgm2, phosphoglucomutase 2; Stat1, signal transducer and activator of transcription 1; Stat3, signal transducer and activator of transcription 3; Klk1b22, kallikrein 1-related peptidase b22; Lif, leukemia inhibitory factor).

Table 5. 

 

List of Genes That Are Differentially Expressed or Potentially Have a Significant Functional Relationship with Retinal Pathology after Conditional Müller-Cell Ablation That Have Been Associated with Retinal Pathology

Gene Name	Gene Symbol	Biological Functions	Reference
Chemokine (C–X–C motif) ligand 13	CXCL13	Inflammation process/B cell recruitment	42
Kallikrein 1-related peptidase b22	KLK1B22	Angiogenesis/proteolysis	43
Endothelin 2	EDN2	Stress response/ neuroprotection	29
Alpha 2 macroglobulin	α2M	Inflammation process	44
Fibroblast growth factor 2	FGF2	Neuroprotection/angiogenesis	45
Leukemia inhibitory factor	LIF	Neuroprotection	46
Signal transducer and activator of transcription 1	STAT 1	Apoptosis	47
Signal transducer and activator of transcription 3	STAT 3	Neuroprotection/anti-apoptosis	46
Phosphoglucomutase 2	PGM2	Glycolysis	48

Glycolytic Pathway Analysis with qRT-PCR and Western Blot

We also performed qRT-PCR analysis of genes involved in glycolysis to further investigate changes in Müller-cell metabolism in the later stages after ablation (1 month and 3 months). We selected the genes based on review of the current literature, which identified that GADPH, Enolase 1 (ENO1), ENO2, and lactate dehydrogenase-a and -b (LDHa and LDHb) are considered major components of the glycolytic pathway that may be affected in retinal diseases. ^8,21 The expression of the majority of the genes encoding these proteins was suppressed, although ENO2 and lactate dehydrogenase-b (LDHb) at 1 month and GAPDH and ENO1 at 3 months after Müller-cell ablation were not significantly differentially expressed compared with the Rlbp-CreER-LacZ control group (Fig. 3). To validate changes in molecules involved in glycolysis at the protein level, we further performed Western blot analysis using antibodies against ENO1 and LDHa along with an antibody against GFAP, a marker for reactive gliosis. Protein was extracted from samples collected 3 months after TMX induction. Densitometric analysis revealed that both ENO1 and LDHa were significantly downregulated in Rlbp-CreER-DTA176 mice compared with Rlbp-CreER-LacZ controls, whereas GFAP was significantly upregulated (Fig. 4). 

Figure 3

 

qRT-PCR analysis of changes in genes involved in the glycolytic pathway 1 and 3 months after selective Müller-cell ablation. *P < 0.01, †P < 0.05; Rlbp1-DTA176 versus Rlbp1-LacZ; n = 6 in each group.

Figure 4

 

Western blot analysis of molecules involved in glycolysis 3 months after TMX-induced Müller-cell ablation. DTA 176 represents Rlbp-CreER-DTA176 group and LacZ represents Rlbp-CreER-LacZ control group. (A), Western blot gel images showing reduced expression of ENO1 and LDHa and increased expression of GFAP. (B–D) Analysis of densitometry revealed that ENO1 and LDHa were significantly downregulated, whereas GFAP was significantly upregulated in Rlbp-CreER-DTA176 mice compared with Rlbp-CreER-LacZ controls. *P < 0.05, †P < 0.01; error bars: SEM; n = 6 in each group.

Discussion

In this study we found that induced ablation of Müller cells in adult mice resulted in changes in gene expression that were consistent with the morphologic features we have reported in this model. In short, we found upregulation of neuroprotective and apoptotic genes soon after Müller-cell ablation and downregulation of the glycolysis and pathways related to vascular integrity later. These processes are likely to be involved in the photoreceptor degeneration, vascular leakage, and neovascularization that we observed in this mouse and that are also common features of many retinal diseases such as MacTel Type 2. Thus, the present study provides genomic information that may be useful in understanding the contribution of Müller-cell dysfunction to retinal diseases. 

We believe that the downregulation of transcription of genes encoding proteins of the glycolytic pathway that we found in the later stages after Müller-cell ablation is particularly significant. In the healthy retina, Müller cells continuously provide lactate to high energy demanding photoreceptors through glycolysis. ^1,9 This lactate can be metabolized further into pyruvate by LDH, ²² which eventually serves as a substrate for oxidative metabolism by photoreceptors. ^10,18 The importance of glycolysis in the retina has been emphasized previously. Inhibition of glycolysis by iodoacetate resulted in a dose-dependent loss of all retinal cells, including retinal neurones, in an in vitro study. ²³ When glycolysis was blocked with the same drug in another in vitro study, the ATP and lactate levels in human Müller cells declined to a very low level. ¹¹ Proteomic analysis of a human MacTel Type 2 postmortem sample showed marked downregulation of glycolytic pathway–related proteins as well as Müller-cell–associated proteins in the diseased macula compared with peripheral retina of the same eye and a control eye. ⁸  

We also found reduced activity of GS-related pathways (downregulation of d-glutamine and d-glutamate metabolism) 1 month after Müller-cell ablation, indicating compromised Müller-cell function. GS, a major enzyme that converts the neurotoxic neurotransmitter glutamate to glutamine, ²⁴ is exclusively expressed in Müller cells in the neural retina and its activity is essential for the normal function of excitatory synapses and reduction of neurotoxicity. ^25,26 We found significant reduction of GS immunoreactivity in our morphologic analysis of these mice. ¹² Interestingly, GS uses ATP generated from glycolysis in Müller cells. ¹⁸ Thus, impairment of glycolysis secondary to Müller-cell dysfunction not only fails to provide enough lactate to neurones, it may also affect neurotransmitter metabolism in the retina. 

Induced Müller-cell ablation in this model results in patches of loss alternating with patches of surviving Müller cells that were activated, as reflected by their strong expression of GFAP. ¹² We also found upregulation of GFAP gene, a marker of gliosis, in these transgenic mice 1 week after Müller-cell ablation, ¹² consistent with present analysis. Activated Müller cells expressing GFAP release neurotrophic factors such as ciliary neurotrophic factor, basic FGF, brain-derived neurotrophic factor, LIF, neurotrophin 3 and 4, and nerve growth factor to protect photoreceptors in many retinal diseases. ²⁷ Consistent with this, we found in the present study a significant upregulation of mRNA encoding GFAP, STAT1, neurotrophic factors (FGF2, EDN2, STAT3, and LIF), and their related pathways (apoptosis and the Jak-Stat signaling pathway) soon after Müller-cell ablation. Upregulation of GFAP has also been reported in other retinal diseases such as retinitis pigmentosa in which Roesch et al. ²⁸ performed single-cell microarray analysis. ²⁸ We found strong upregulation of EDN2, a stress response molecule produced by stressed photoreceptors, at all time points after Müller-cell ablation. EDN2 binds to its receptors on Müller cells, causing them to become gliotic and produce neurotrophic factors such as FGF2 and LIF. ^29,30 Jak/Stat pathway related genes such as STAT1 and STAT3, which are pro-apoptotic ³¹ and anti-apoptotic, ^32,33 respectively, were also upregulated soon after Müller-cell ablation. In our previous publication, we observed apoptotic cells in the inner nuclear layer (INL) with TUNEL staining in Müller cell bodies soon after Müller-cell ablation. TUNEL-positive cells were then found in the outer nuclear layer a few days later. ¹² Thus, it appears that patchy loss of Müller cells in this model causes photoreceptor degeneration and death in the affected area, whereas remaining Müller cells mount a protective response. Further studies on DE genes and their encoded proteins in areas of Müller-cell ablation compared with regions with reactive gliosis are warranted to test the hypothesis that stressed photoreceptors in the Müller-cell–ablated areas produce EDN2, whereas remaining Müller cells become gliotic and produce neurotrophic factors in an attempt to prevent photoreceptor death. 

This analysis was generally consistent with the previous morphologic study. Soon after Müller-cell ablation, we found apoptotic cells in the INL with TUNEL staining in Müller cell bodies. TUNEL-positive cells were then found in the outer nuclear layer a few days later. ¹² We also found upregulation of GFAP, a marker of gliosis, in these transgenic mice 1 week after Müller-cell ablation, ¹² consistent with present analysis. 

We have described vascular changes in the later stages after Müller-cell ablation in this model, in which vascular leakage started soon after Müller-cell ablation, whereas severe BRB breakdown and angiogenesis developed later. ¹² Consistent with this, we found that tight junction pathways were downregulated in the later stages after Müller-cell ablation. Tight junctions are critical components of the BRB, which controls the flux of a variety of chemokines, cytokines, and amino acids across the vascular endothelium. ³⁴ Müller cells play an important role in supporting the endothelial cells of the inner BRB, which is mainly responsible for supplying nutrients to the neural retina. ³⁵ The relationship between Müller cells and vascular changes has been reported both in vitro and in vivo. Coculturing Müller cells and primary bovine endothelial cell monolayers improved barrier function in normoxic conditions but led to barrier breakdown under hypoxic conditions. ³⁶ Inducing breakdown of BRB breakdown in rats by disrupting Müller cells with subretinal injection of DL-alpha-AAA ³⁷ resulted in vascular changes in association with reduced expression of claudin 5 that were observed predominantly in areas of Müller-cell disruption. ³⁷ These observations, along with downregulation of tight junction pathways found in the present study support the concept that Müller-cell deficiency disturbs retinal vascular interendothelial tight junctions. 

We found that potassium (K⁺) channel–related genes were suppressed 3 months after selective Müller-cell ablation. Maintenance of retinal K⁺ homeostasis is one of the most important functions of Müller cells. Müller cells take up K⁺ that is released by neurons into the extracellular space by an inwardly rectifying K⁺ channel ³⁸ and release it into extraretinal spaces such as blood and vitreous. ⁶ The osmotic force of the retinal extracellular space increases when K⁺ homeostasis is disturbed, thereby causing retinal edema, ³⁹ and, subsequently, BRB breakdown. ⁴⁰ K⁺ conductance dysregulation and Kir4 channel mislocation with Müller-cell gliosis have been described in some retinal diseases such as diabetic retinopathy. ⁴¹ Thus, we propose that selective Müller-cell ablation causes alterations in K⁺ channel–related genes, which results in dysregulation of osmosis, retinal edema, and BRB breakdown. 

We also analyzed the temporal trend of DE genes over time after induced Müller-cell ablation. RR analysis was consistent with the retinal pathology that developed after Müller-cell ablation: apoptosis and neuronal changes were found 1 week after Müller-cell ablation followed by genes associated with BRB breakdown and the development of intraretinal neovascularization later. Since these features are seen in many retinal diseases, particularly in MacTel Type 2 in which recent clinicopathologic reports have implicated Müller-cell dysfunction, this study may provide insights into the pathologic mechanisms underlying MacTel Type 2, which are currently poorly understood. 

The patchy nature of Müller-cell ablation that occurs in this model means that areas of Müller cell loss alternate with areas of surviving Müller cells, which tend to become gliotic. Since we used whole retinae for this genomic analysis, we are unable to say where the DE genes were expressed. Further research is warranted to perform genomic analysis on Müller-cell ablated areas and on those with reactive Müller-cell gliosis. Also, findings from this animal model should be extrapolated to clinical conditions with care since, unlike humans, mice do not have maculae. 

This global genomic analysis of mouse retinae after induced selective Müller-cell ablation, found genomic changes consistent with photoreceptor degeneration and retinal vasculopathy that have been described in these mice. Increased transcription of genes responsible for neuronal apoptosis, neurotrophic factors, and their related pathways were found early, whereas the glycolytic and tight junction pathways were downregulated later. Since photoreceptor degeneration and vasculopathy are common features of many retinal diseases, further studies in humans and animal models are warranted to identify the specific roles of Müller-cell dysfunction in common conditions such as diabetic retinopathy. Further investigation of the pathways involved in retinal neuronal damage and vasculopathy after selective Müller-cell ablation may lead to novel therapies for retinal diseases that may be associated with Müller-cell dysfunction. 

Supplementary Materials

pdf S1 

doc S1 

Acknowledgments

The authors thank Donna Lai and Ling Zhu for qRT-PCR training and Victoria Wen for microarray technique support. 

Footnotes

 Supported by a Lowy Medical Research Institute grant, National Health and Medical Research Council (NHMRC, Australia) Grant APP1028393, an Ophthalmic Research Institute of Australia grant, and a Rebecca L. Cooper Medical Research Foundation and University of Sydney Bridging grant. Mark Gillies is a fellow of Sydney Medical School Foundation and supported by an NHMRC Practitioner Fellowship.

Footnotes

 Disclosure: S.H. Chung, None; W. Shen, None; K. Jayawardana, None; P. Wang, None; J. Yang, None; N. Shackel, None; M.C. Gillies, None