Effects of glial cell line-derived neurotrophic factor on cultured murine retinal progenitor cells

Isolation and culture of late retinal progenitor cells

RPCs were previously isolated from the neural retina of postnatal day 1GFP transgenic mice [28]. Briefly, retinas were harvested from newborn GFP transgenic C57BL/6 mice (gift from Masaru Okabe, University of Osaka, Osaka, Japan) [29] and subjected to several cycles of collagenase digestion to dissociate the tissue. Cells were then forced through a nylon mesh of 100 µm pore size, centrifuged, and resuspended in standard culture medium, containing Dulbecco’s modified Eagle’s medium/Ham’s F12 1:1 (Omega Scientific, Tarzana, CA) supplemented containing 1% (by vol.) N2 neural supplement (Gibco, Invitrogen, Carlsbad, CA), 2 mM L-glutamine (Sigma Aldrich, St. Louis, MO), 2,000 U nystatin (Gibco), 50 mg/ml penicillin-streptomycin (Sigma), and 20 ng/ml recombinant human epidermal growth factor (EGF; Invitrogen, Carlsbad, CA). GFP+ neurospheres appeared within the first 2 to 3 days. Culture media were changed every 2 days, and proliferating cells were passaged at regular intervals of 4–5 days. These cells were immunoreactive for nestin (a marker for neural progenitor cells) and K_i-67 (a marker for cell proliferation).

Treatment conditions

To first determine the optimal concentration of GDNF for use in the present study, we tested the effects of four different concentrations of GDNF (2.5, 5, 10, and 20 ng/ml) on the gene expression profile of genetically modified RPCs. Cells were treated with GDNF for 5 days, RNA was extracted, and a quantitative PCR (qPCR) assay was performed. The progenitor markers remained at similar levels to those of untreated controls under the different concentrations of GDNF examined (Figure 1). Given this general equivalence, combined with our interest in looking for a possible physiologic (as opposed to toxic) influence, we chose 10 ng/ml for use in our system.

RPCs were cultured in standard base media with the only difference being the presence or absence of EGF (20 ng/ml) and/or 10 ng/ml recombinant human GDNF (PHC7045; Biosource, Invitrogen) as follows:

1) No recombinant factors (nonmitogenic conditions, negative control),

2) EGF alone (proliferation conditions, positive control, i.e., identical to “standard medium”),

3) GDNF alone, or

4) EGF+GDNF together.

Morphometry

Cellular morphology was recorded on treatment day 1, 3, and 5 using a microscope-mounted camera and imaging system. In addition, cellular proliferation and sphere formation were also recorded and quantified every 45 min using an IncuCyte live-cell imaging system (Essen Instruments, Ann Arbor, MI) located within the incubator.

Cell viability

RPCs were cultured under four different conditions consisting of the base medium and either no added growth factors, EGF alone, GDNF alone, or EGF+GDNF, for 5 days. Cell viability was measured using a Cell Counting Kit-8 assay (Dojindo, Kumamoto, Japan). In brief, cells were suspended at a final concentration of 1×10⁴ cells/well and cultured in 96-well, flat-bottomed microplates. At the end of each treatment day, 10 μl WST-8 [2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium, monosodium salt] was added to each well. The plates were incubated for an additional 4 h at 37 °C to convert WST-8 into formazan. The absorbance was then measured at 450 nm with a spectrophotometer. Cell viability is in direct proportion to the absorbance at 450 nm; therefore, viability was expressed as the A₄₅₀ value. All experiments were performed in 96-well plates (ten wells/condition).

In addition, growth of RPCs in medium with EGF alone, or with EGF and GDNF, was assessed over a one-month time course. Cells were passaged before reaching confluence. At each passage, the cell number was counted using a hemocytometer, and a population doubling level (PDL) was determined as current PDL=Log(final harvest cell number/initial seeded cell number)×3.33.

Caspase-3 activity assay

After 24 h culture, RPCs exposed to the four different conditions mentioned above were collected and washed in cold phosphate buffered saline (PBS; 2.68 mM KCI, 1.47 mM KH₂PO₄, 135.60 mM NaCl, 8.10 mM Na₂HPO₄). Cell pellets were resuspended in 100 μl lysis buffer. After 20 min incubation on ice, each cell lysate was spun at 18,000× g at 4 °C for 15 min. The supernatant was collected and 5 μl of cell lysate from each condition was diluted to 100 μl in the assay buffer consisting of 20 mM Hepes at pH 7.4, 0.1% CHAPS, 5 mM dithiothreitol, and 2 mM EDTA, supplemented with 2 mM of the caspase-3 substrate Ac-DEVD-pNA. The enzymatic activity was determined spectrophotometrically at 405 nm (molar absorptivity of p-nitroaniline, ε^mM=10.5 at 405 nm). Caspase-3 activity was determined as µmol pNA released per minute per milligram of protein.

RNA isolation

RPCs were grown in culture media that contained EGF or EGF+GDNF for 5 days. Total RNA was extracted from each sample using the RNeasy Mini Kit (Cat. No. 74104; Qiagen, Valencia, CA). After RNA isolation, samples were treated by DNaseI (Macherey-Nagel, Düren, Germany) to digest and eliminate any contaminating genomic DNA. RNA concentration was measured for each sample at a wavelength of 260 nm (A₂₆₀), and the purity of extracted total RNA was determined by the A₂₆₀/A₂₈₀ ratio. Quantitative reverse-transcriptase PCR analyses and microarray analysis were only performed on samples with A₂₆₀/A₂₈₀ ratios between 1.9 and 2.1.

Microarray processing and analysis

After RNA isolation (two treatments, with three samples per treatment), 1 μg total RNA from each sample was used to prepare material for hybridization with each of the six Affymetrix mouse gene 1.0 ST arrays at the UCI Genomics High-Throughput Facility at the University of California, Irvine, as recommended by the manufacturer (Affymetrix Genechip Whole Transcript Sense Target Labeling Assay Manual; Affymetrix, Inc., Santa Clara, CA). The integrity and concentration of total RNA were measured, followed by hybridization, scanning, and generation of raw expression data, which were subsequently normalized using a standard technique by the same facility, as follows.

All starting total RNA samples were quality assessed before beginning the target preparation/processing steps by running out a small amount of each sample (typically 25–250 ng/well) onto a RNA 6000 Nano LabChip that was evaluated on an Agilent Bioanalyzer 2100 (Agilent Technologies, Palo Alto, CA). Single-stranded, then double-stranded, cDNA was synthesized from the poly(A)⁺mRNA present in the isolated total RNA (typically 100 ng total RNA starting material for each sample reaction) using the GeneChip WT cDNA Synthesis Kit (Affymetrix, Inc., Santa Clara, CA) and random hexamers tagged with a T7 promoter sequence. The double-stranded cDNA was then used as a template to generate many copies of antisense cRNA from an in vitro transcription reaction for 16 h in the presence of T7 RNA polymerase using the Affymetrix Genechip WT cDNA Amplification Kit. Ten micrograms of cRNA were input into the second-cycle cDNA reaction with random hexamers that were used to reverse-transcribe the cRNA from the first cycle to produce single-stranded DNA in the sense orientation.

The single-stranded DNA sample was fragmented (WT Terminal Labeling Kit, Affymetrix) to an average strand length of 60 bases (range 40–70 bp) following prescribed protocols (Affymetrix GeneChip WT Sense Target Labeling Assay Manual). The fragmented single-stranded DNA was subsequently labeled with recombinant terminal deoxynucleotidyl transferase and the Affymetrix proprietary DNA Labeling Reagent, which is covalently linked to biotin. Following the recommended procedure, 0.54 μg of this fragmented single-stranded target cDNA was hybridized at 45 °C with rotation for 17 h (Affymetrix GeneChip Hybridization Oven 640) to probe sets present on an Affymetrix mouse-gene 1.0 ST array. The GeneChip arrays were washed and then stained (streptavidin-phycoerythrin) on an Affymetrix Fluidics Station 450 (Fluidics protocol FS450_007). Arrays were scanned using the GeneChip Scanner 3000 7G and GeneChip Operating Software v1.4 to produce CEL intensity files.

Normalization was performed using the probe logarithmic intensity error (PLIER) estimation method, which includes a quantile normalization protocol within the associated software algorithm. Briefly, the probe cell intensity files (*.CEL) generated above were analyzed using Affymetrix Expression Console software v1.1 using the PLIER algorithm to generate probe-level summarization files (*.CHP). The algorithm used was from PLIER v2.0 (quantification scale: linear; quantification type: signal and detection p value; background: PM-GCBG; normalization method: sketch-quantile).

Statistical analysis was subsequently conducted using JMP Genomic 3 software (SAS Institute Inc.) The raw data were log-2 transformed and imported into the software for analysis. The strategy used was to compare the gene expression profile of the control group (EGF alone) with the experimental group (EGF+GDNF). Data from the arrays were analyzed by the clustering of differences between treatments and identification of significant changes. Changes were considered statistically significant if the difference in expression between the groups had a p value less than 0.05.

Reverse transcription and quantitative PCR

To further assess expression changes, transcription levels of selected genes were examined by performing semiquantitative real-time PCR using an Applied Biosystems 7500 Fast Real-Time PCR Detection System (Applied Biosystems, Foster, CA). Two micrograms of total RNA in a 20 μl reaction volume were reverse-transcribed using an Omniscript cDNA Synthesis Kit (Qiagen). Oligonucleotide primer sequences (Table 1) were designed using Primer 3 software. The primers were synthesized commercially (Invitrogen) and qPCR was performed in 20 μl total volume containing 10 μl of 2× Power SYBR Green PCR Master Mix (Applied Biosystems), 10 μl of cDNA, and 300 nM of gene-specific primers. Cycling parameters for qPCR were as follows: initial denaturation at 95 °C for 10 min, followed by 40 cycles of 15 s at 95 °C and of 1 min at 60 °C. To normalize template input, the β-actin transcript level was measured for each sample (endogenous control). The efficiency of the PCR reaction was measured with primers using serial dilution of cDNA (1:1, 1:5, 1:25, 1:125, 1:625, and 1:3,125). The relative expression of the gene of interest, (Etarget) ΔCt target (control-treated)/(Eref)ΔCt ref (control-treated) [30], was then evaluated by the Pfaffl method [30]. The value obtained for crossing threshold (Ct) represents the number of PCR cycles at which an increase in fluorescence signal (and therefore cDNA) can be detected above the background. The increase is exponential for the particular gene. Data are expressed as fold-change relative to untreated controls, after normalizing to β-actin (our choice from among a set of “house-keeping” genes).

Western blotting

Samples for protein analyses were homogenized in CelLytic M (C2978; Sigma), and proteins were measured using the Bio-Rad detergent-compatible protein assay (500–0006; Bio-Rad). Proteins (50 µg) were electrophorised using a 3%–8% Tris-acetate gel (EA0375BOX; Invitrogen) for 70 min at 150 V and then transferred from the gel to a polyvinyldifluoride membrane using an iBot Gel Transfer Device (IB1001; Invitrogen). Western blotting was then performed using the WesternBreeze Chromogenic Western Blot Immunodetection Kit (WB7103, Invitrogen) according to the manufacturer’s protocol. After blockade of nonspecific binding sites using Blocker/Diluent solutions provided, the polyvinylfluoride membranes were incubated for 2 h at room temperature with mouse primary antibodies directed against either K_i-67 (556003, 1:200; BD), nestin (611658, 1:200; BD), or β-actin (A5441, 1:200; Sigma). This was followed by washing and 1 h incubation with an alkaline phosphatase-conjugated antimouse IgG secondary antibody provided in the kit. Immunoblots were developed using chromogenic substrate for 10 min. The reactive bands were scanned and quantified by densitometry using UN-SCAN-IT Gel 6.1 (Silk Scientific, Inc. Orem, UT) The density of product from the EGF treatment group was defined as 1, and the data from the EGF+GDNF condition was expressed as the fold-change of this value.

Statistical analysis

The results represent the average of three experiments (±SE). Except where specifically indicated, each experiment was performed in triplicate. For cell viability and caspase-3 activity studies, ten fields per sample were analyzed for each condition. Statistical significance was determined by Student’s two-tailed t-test.

RPC morphology and sphere formation

RPCs were subjected to different treatment conditions for up to 5 days. As a result, the morphology of RPCs clearly differed in response to the presence or absence of EGF (Figure 2). There was increased extension of processes in response to the absence of EGF (Figure 2A-F), whereas the appearance of undifferentiated RPCs cultured in either EGF alone or EGF+GDNF remained stable throughout the treatment period. Undifferentiated RPCs either adhered to the surface of the uncoated flasks or were observed floating in the culture medium, either as single cells or spheres of different sizes. From day 3, short processes with few, if any, branches extended from some of the adherent cells (Figure 2H,K). No obvious morphological differences were observed in response to the presence of 10 ng/ml GDNF (GDNF alone) when compared to culture conditions without EGF (no recombinant factors; Figure 2D-F). However, in the presence of GDNF (EGF+GDNF), more and larger spheres were observed, as compared to EGF alone (Figure 2I,L).

The proliferation of RPCs and formation of spherical cellular aggregates (spheres) were also monitored via the IncuCyte system over the five-day time course, allowing comparison of EGF and EGF+GDNF conditions. By way of the acquired video, dissociated RPCs were seen to exhibit a strong propensity to aggregate in culture, with the formation of spheres being evident as early as 5.5 h after seeding. Aggregates started with the movement of suspended single RPCs within the culture medium and adherence to other cells, as opposed to cell division and clonal expansion (see Appendix 1, Figure 3). Over time, the spheres enlarged and their movement lessened, consistent with an increased contribution of cell proliferation to sphere formation. The number of spheres with a cross-sectional area of >870 µm² increased in both treatment conditions, beginning on day 0 and continuing to day 5, at which time greater numbers of spheres had reached threshold criteria in the GDNF-supplemented medium (p<0.05). In addition, more spheres had attained higher categories of cross-sectional area in this treatment condition as well, consistent with increased cellular proliferation (Figure 4).

Cell viability and proliferation

Murine RPCs exhibited exponential growth in the presence of EGF alone, as well as with EGF+GDNF, but diminished asymptotically in the absence of both factors, or with GDNF alone. Cell viability was compared for the EGF alone versus the EGF+GDNF conditions using the cell-counting kit. By this method, the number of RPCs increased from day 1 to 4, although there was no difference between conditions. By day 5, however, there were significantly more viable cells in the EGF+GDNF compared to those in EGF alone (Figure 5A). Moreover, the growth over a one-month period confirmed that the addition of GDNF to a standard EGF-containing proliferation medium provides sustained augmentation of RPC numbers (Figure 5B).

Caspase-3 activity

We determined the level of caspase-3 activity under conditions with or without EGF or GDNF after 24 h treatment. Caspase-3 activity levels with no growth factors, as well as with GDNF alone, were substantially elevated compared to conditions that included EGF (p<0.05), and activity was significantly higher in the no-growth-factor condition than in GDNF alone (p<0.05). Between EGF and EGF+GDNF, no difference was observed (Figure 6).

Microarray screening for GDNF-induced changes in gene expression

The signal levels of individual cDNA from the experimental (EGF+GDNF) and control (EGF) groups were compared after hybridization on a mouse whole-genome, gene-level microarray that consisted of over 28,853 genes. The fold change and p value of each of the genes were calculated, and log₂ (fold change) and −log₁₀ (p value) were transformed. Genes with a p value of <0.05 and an absolute fold change value of >1 were arbitrarily defined as upregulated in GDNF-treated cells, whereas those with a p value <0.05 and an absolute fold change value of <1 were defined as downregulated. The distribution analysis was done as a quality control step. The overlayed kernel density estimates (Figure 7A) derived from the distribution analysis showed the raw univariate distributions for all arrays, and allowed the visualization of sources of variation attributed to technical procedures. The distributions for these arrays were very similar, indicating that this represented a high-quality data set requiring little if any normalization for further analysis. Principle component analysis revealed a broad degree of overlap in expression patterns between the two treatment conditions, with little suggestion of a treatment effect (Figure 7C). However, ANOVA analysis showed that of the approximately 25,189 genes that were expressed in the cultures treated with EGF+GDNF, 5,250 of these were associated with the presence of GDNF (Figure 7B). A total of 3,865 genes were upregulated and 1,385 genes down regulated following the addition of GDNF to EGF-containing standard medium. In this case, changes in 24 genes were detected in the EGF+GDNF condition (Table 2). Among those were 16 known and eight predicted genes, the latter not yet associated with any biologic functions (data not shown). Of the 16 known genes, six were olfactory receptors with three being upregulated and three down-regulated. There were two genes with possible associations to photoreceptor cilia, namely myosin IXa and a dynein (Dnahc7a); however, changes were slight and in opposite directions.

Effect of GDNF on RPC gene expression determined by qPCR

In addition to checking for global changes, the potential influence of GDNF on a specific subset of genes with known associations to RPCs was also evaluated using quantitative PCR. Following treatment of RPCs with GDNF for up to 5 days, expression of most progenitor-associated markers remained effectively stable (Figure 8). At baseline, in the presence of EGF, cultured RPCs expressed the progenitor markers Nestin, K_i-67, ceh-10 homeo domain containing homolog (C. elegans) [Chx10], sex determining region Y-box 2 (Sox2), v-myc myelocytomatosis viral oncogene homolog (avian) [C-myc], and Vimentin. This pattern was sustained in the EGF+GDNF condition. Similarly, the expression levels of precursor and lineage-related markers including βIII-tubulin, microtubule-associated protein 2 (MAP2), doublecortin (DCX), cellular retinaldehyde binding protein (CRALBP), protein kinase C, alpha (PKC-α), and Recoverin. These were relatively unaffected by the addition of GDNF to the EGF-based proliferation medium (Figure 1). Of the above markers, a subset exhibited marginally increased expression, namely, Hes5 (1.03-fold), K_i-67 (1.07-fold), Mash 1 (1.12-fold) and Vimentin (1.20-fold; Figure 8). Of these, the microarrays showed no significant difference for Hes5, K_i-67, or Vimentin (Mash 1 was not included; Table 3). Taking the comparison further, the qPCR data did not confirm the marginal changes seen with the microarrays for Nanog, Notch1, or GFAP.

Western blot analysis

PCR detects changes in mRNA expression; however, the extent to which these changes are reflected at the level of proteins is also of interest. Western blotting was employed for this purpose on samples from the EGF alone and the EGF+GDNF treatment groups. Nestin was used as a marker of neural progenitor cells and K_i-67 as a marker of cell proliferation.

Anti-nestin identified two bands with the anticipated molecular weight, approximately 220 kDa and 200 kDa, with β-actin labeling used as the loading control. The upper band of nestin (approximately 220 kDa) was strongly labeled in both the EGF and EGF+GDNF conditions and the lower band (approximately 200 kDa) was little changed in the EGF+GDNF condition, compared with EGF alone. Overall, no significant difference between treatment conditions was detected by quantification of the entire blot-lane intensity. Two bands were detected by K_i-67 antibody, approximately 395 kDa and 345 kDa. Quantification analysis showed that the intensity of the entire blot lane was increased in the EGF+GDNF condition by 1.189±0.0095-fold (Figure 9). These results at the protein level were consistent with the mRNA results for Nestin and K_i-67, as determined by qPCR (Figure 8), namely, that the total Nestin mRNA did not change in response to EGF+GDNF treatment over 5 days, whereas K_i-67 was slightly upregulated (Figure 9).

Conclusions

This study shows that GDNF is not a potent inducer of RPC differentiation. Its results support the use of these cells for GDNF delivery to the diseased retina. Because the expression profile of cultured RPCs is not markedly altered by exposure to GDNF, it may be the case that transduced RPCs will retain their ability to repopulate local retinal populations following transplantation.