Molecular Vision 2005; 11:775-791 <>
Received 22 March 2005 | Accepted 16 September 2005 | Published 20 September 2005

Gene expression profiles of intact and regenerating zebrafish retina

David A. Cameron, Karen L. Gentile, Frank A. Middleton, Patrick Yurco

Department of Neuroscience and Physiology, SUNY Upstate Medical University, Syracuse, NY

Correspondence to: David A. Cameron, Department of Neuroscience and Physiology, SUNY Upstate Medical University, 750 East Adams Street, Syracuse, NY, 13210; Phone: (315) 464-8149; FAX: (315) 464-7712; email:


Purpose: Investigate the molecular determinants of retinal regeneration in adult vertebrates by analyzing the gene expression of control and post-lesion retina of adult zebrafish, a system that regenerates following injury.

Methods: Gene expression of zebrafish retina and brain were determined with DNA microarray, RT-PCR, and real-time quantitative PCR analyses. Damaged retinas and their corresponding controls were analyzed 2-5 days post-lesion (acute injury condition) or 14 d post-lesion (cell regeneration condition).

Results: Expected similarities and differences in the gene expression profile of zebrafish retina and brain were observed, confirming the applicability of the gene expression techniques. Mechanical lesion of retina triggered significant, time-dependent changes in retinal gene expression. The induced transcriptional changes were consistent with cellular phenomena known to occur, in a time-dependent manner, subsequent to retinal lesion, including cell cycle progression, axonal regeneration, and regenerative cytogenesis.

Conclusions: The results indicate that retinal regeneration in adult zebrafish involves a complex set of induced, targeted changes in gene transcription, and suggest that these molecular changes underlie the ability of the adult vertebrate retina to regenerate.


The CNS of most adult vertebrates, including humans, has a limited ability for cellular repair. This lack of cellular regeneration impacts function because disease- or injury-induced neurological deficits associated with a loss of neurons are not overcome by newly-born, function-restoring cells. In contrast the CNS of some vertebrates, such as adult anamniotes, possesses an intrinsic ability for substantial cellular repair. The retinas of adult fish and amphibians, for example, can regenerate axons and replace cells that have been lost subsequent to trauma [1-4]. The cellular mechanisms that underlie retinal regeneration are the subject of active investigation, with recent evidence suggesting that a type of glial cell, the Müller glia, might function as an inducible stem or precursor cell [5-10].

Little is known about the molecular mechanisms that enable retinal regeneration in adult anamniotes, and by extension our understanding of the molecular attributes that inhibit, retard, or otherwise preclude retinal regeneration in other vertebrates is correspondingly thin. In an effort to identify the molecular determinants of retinal regeneration we investigated the gene expression profile of control and damaged retinas from adult zebrafish (Danio rerio), a model system that is known to regenerate [9,11-13]. DNA microarray, reverse transcriptase polymerase chain reaction (RT-PCR), and quantitative real-time PCR (qPCR) analyses were performed, with gene expression profiles determined and compared across two general experimental time points: acute retinal injury (2-5 days post-lesion and corresponding control retina) and retinal cell regeneration (14 days post-lesion and corresponding control retina). Restoration of lesioned zebrafish retina involves a proliferative response of retinal cells during the first two weeks post-lesion, including Müller glia [9]. This is followed by a regeneration of new retinal cells [11] that leads to a regenerated structure that contains the same repertoire of cell types as that of non-lesioned retina [11,12]. To help establish the sensitivity of the zebrafish microarray to highly predicted changes in expression, gene expression profiles were determined and compared between zebrafish brain and retina tissues.

We report that a retinal lesion induces complex intraretinal changes in gene expression. Individual genes displaying the largest significant changes in expression include those implicated in cellular phenomena that are adaptively significant within the context of retinal repair, including the clearance of cellular debris, progression through the cell cycle, axonal regeneration, and cytogenesis. The observed profiles of lesion-induced changes in retinal gene expression indicate a complex network of molecular mechanisms that enable an adult vertebrate retina to engage in function-restoring regenerative neuronogenesis.


Wildtype adult zebrafish (Danio rerio) of standard length 2.5-3.0 cm were used for all experiments, which were approved by the SUNY Upstate Committee for the Humane Use of Animals. Mechanical excision of a small patch of dorsal retina from the eye of an anesthetized experimental fish was performed as described previously [14,15]; the other eye of the fish served as a control. Fish were housed communally in standard aquaria for either 2, 3, 5, or 14 days post-lesion. Because overt cellular responses to zebrafish retinal injury are evident >1.5 mm away from the lesion site [9] (an estimated 20-35% of the total retinal area), total RNA derived from an entire post-lesion retina was anticipated to contain evidence for a substantial amount of lesion-induced changes in gene expression.

RT and real-time quantitative PCR

Following euthanasia retinas were harvested from at least six fish for each experimental condition, and flash-frozen on dry ice. The frozen material was immediately processed for mRNA extraction and first-strand cDNA synthesis as described previously [16]. The cDNA material was stored at -80 °C, with working aliquots kept at 4 °C. The cDNA material for all PCR analyses was collected from sets of fish different from those used in the gene microarray analyses.

Standard RT-PCR procedures were used with a T-Personal Thermocycler (Biometra; Niedersachsen, Germany). The following single tube master mix was prepared fresh on ice, with the respective volumes scaled linearly for the total number of reaction tubes to be analyzed: 38.25 μl H2O, 5.0 μl 10X Taq buffer with (NH4)2SO4 (Product number B33; Fermentas; Hanover, MD), 3.0 μl MgCl2 (25 mM), 0.5 μl dNTP mixture (10 mM; Eppendorf; Hamburg, Germany), 1.0 μl cDNA sample (at 2-3 μg/μl), and 0.25 μl Taq DNA polymerase (Product number EP0402; Fermentas). The appropriate forward and reverse primers (each at stock concentration of 0.1 nmol/μl) were added to the reaction mixture at a volume of 1.0 μl. A three-step thermocycling regimen was applied for 30 cycles (denature for 30 s at 95 °C, anneal for 15 s at 60 °C, and extend for 30 s at 72 °C). Amplified products were analyzed with standard gel electrophoresis. The amplified products were purified, cloned into pCR4-TOPO (Invitrogen), and their identities confirmed by sequencing (Biotechnology Resource Center, Cornell University). The primer sequences used in these experiments are summarized in Table 1.

Real-time quantitative PCR (qPCR) was performed with a Smart Cycler apparatus (Cepheid; Hamburg, PA). For each combination of primer pair and cDNA sample the following master mix was prepared fresh on ice: 60.5 μl H2O, 9.2 μl 10X Taq buffer, 11.3 μl MgCl2 (25 mM), 1.0 μl dNTP mixture (10 mM Fermentas), 3.0 μl primer (at 0.1 nmol/μl), 1.5 μl Sybr Green solution (Molecular Probes; Eugene, OR), 2.0 μl cDNA sample (at 2-3 μg/μl), and 0.5 μl Taq DNA polymerase (Fermentas). The reaction solution was briefly mixed and 26 μl was added to each of three reaction tubes (Cepheid product number 900-0003). Primer/sample combinations were run in triplicate in each real-time PCR test. The loaded reaction tubes were centrifuged briefly (E & K Scientific; Los Gatos, CA), tapped gently to remove bubbles, and loaded into the Smart Cycler device. The qPCR cycling regimen was 5 min at 95 °C, 35 iterations of a three-step temperature series (15 s at 95 °C, 15 s at the optimal annealing temperature for each pair of primers, 15 s at 72 °C), and a temperature ramp from 60-95 °C in 0.2 °C/s steps. Fluorescence measurements were made and recorded at each 15 s extension step. An optimal annealing temperature was determined empirically for each pair of primers (Table 1). The extremely low level of notch2 expression in control retina precluded its quantitative analysis with qPCR. For each reaction tube a PCR cycle threshold (CT) was defined as the cycle value at which the second derivative of the growth function of Sybr Green fluorescence was maximal [17]. For relative quantification of expression levels, the values of CT for each of the target amplified products in each experimental condition were determined, and referenced to the family of reference CT values from rhodopsin, which was run in parallel with the same master mix. Fold differences in expression level were determined for each amplified product by comparing these relative CT differences, and determining the corresponding linear fold difference, between the control and lesion conditions.

During the optimization procedure it was empirically determined that a satisfactory stock Sybr Green solution (in H2O) had the following spectroscopic absorbance values at 230, 260, 280, and 475 nm, respectively: 0.26, 0.13, 0.16, 0.75. The forward and reverse primer sequences, the optimal values of the annealing temperature (A), and the measured peak of the melt-function's first derivative (the melting temperature) for each amplified product in this investigation are listed in Table 1. Melt curve analysis determined data quality: only qPCR products for which the measured melting temperature was within two standard deviations of the mean melting temperature derived from all tests of that particular primer pair were used for subsequent growth curve analysis. The expected size of all amplified products was confirmed with gel electrophoresis, and the identity of the amplified and cloned products was confirmed by sequencing.

Microarray analysis

Following euthanasia retinas and brains were harvested and frozen on dry ice. Total RNA was immediately collected from each tissue for each experimental condition (control and post-lesion, at 2 d, 3 d, 5 d, or 14 d post-lesion). The material from 8-10 fish was pooled to mitigate potential individual variability or gender-based bias in the samples, and for each group of fish half the animals were dark adapted for at least 2 h to control for potential influences of ambient light conditions across the different time points. Total RNA was collected using the RNeasy Mini and QIAshredder kits (Qiagen; Valencia, CA). Initial determination of total RNA quality and integrity was based upon the detection of 18S and 28S rRNA bands with denaturing gel electrophoresis and ethidium bromide staining. Eight zebrafish eyes typically yielded 3-6 μg of total RNA, which was stored at -80 °C as aqueous stock solutions. All stocks were used within 24 h of collection.

To label the RNA for analysis on the GeneChip, we used the One-Cycle Target Labeling and Control Reagent Kit (Affymetrix; Santa Clara, CA). Briefly, the mRNA fraction from approximately 2 μg of total RNA was reversed transcribed using an oligo-dT primer coupled to a T7 RNA polymerase recognition sequence. After second strand synthesis in the presence of RNase H and subsequent DNA purification, the double stranded cDNA template was used as a template for in vitro transcription (IVT) during which a fixed concentration of biotinylated ribonucleotide was incorporated into the cRNA products. After 4-6 h, the IVT reaction was stopped and DNAase 1 added to the tube to eliminate the template. Each biotin-labeled cRNA sample was then purified, fragmented randomly to 35-200 nucleotides by incubating the sample in a high magnesium buffer at 94 °C for 35 min, added to a hybridization solution that was spiked with positive controls, heated to 95 °C for 5 min, equilibrated to 45 °C for 5 min, and centrifuged at 10,000x g for 5 min. A volume of 250 μl of this labeled solution was added to an Affymetrix GeneChip® Zebrafish Genome Array (the company's first and only version of this GenChip), which contains probes for approximately 14,900 Danio rerio transcripts. The arrays were incubated at 45 °C for 16 h with rotation (60 RPM) in a hybridization oven 640 (Affymetrix), washed and stained with the Fluidics Station 450 using the EukGE-WS2v4 protocol, and scanned once using an Affymetrix GeneChip Scanner. All data were normalized and scaled to a mean target intensity value of 500 using the Gene Chip Operating System software (GCOS; version 1.3).

To screen the data for correlated changes in entire functional gene groups, we used custom-written data mining software (PathStat) to extract all of the gene-by-gene, log normalized differential expression ratios for genes within more than 300 different functional gene groups that were represented by the annotated transcripts on the zebrafish microarray [18-20]. PathStat is available for download; contact Frank A. Middleton (SUNY Upstate Medical University, Syracuse, NY). The mean differential expression log ratios for each functional gene group were converted to z scores by comparison with the mean differential expression ratio for all the transcripts on the array which were classified as Present in either of the two conditions being compared according to the GCOS software. Gene groups were ranked by z scores to determine those with the most significant alteration in expression. In Table 2, Table 3, Table 4, Table 5, Table 6, Table 7, Table 8, and Table 9 the log2 signal ratio values were defined as the log of the post-lesion retina signal divided by the control retina signal, and thus the control-rich values are negative.

Data availabilty

All of the microarray signal data have been deposited in the Gene Expression Omnibus (GEO) respository under the series accession number GSE3303.


Gene Expression: Brain compared to retina

To evaluate the applicability and sensitivity of the Affymetrix Zebrafish Genome Array for studies of retinal gene expression in zebrafish, the array was tested for its ability to characterize and differentiate the gene expression profiles of control brain and retina. Both structures are components of the CNS and therefore were expected to share transcriptional attributes, including high expression levels for genes encoding components of the glycolytic and oxidative phosphorylation machinery. Significant tissue-specific differences in gene expression were also expected, including an abundance of transcripts in retina that encode components of the phototransduction mechanism [21].

Both expectations were verified by the DNA microarray analysis. The absolute signal amplitudes derived from each of the array's oligonucleotides are plotted on Figure 1 for both control brain and retina, and the identity of the fifty most abundant transcripts for brain and retina, based upon raw signal amplitude, are tabulated independently in Table 10 and Table 11, respectively. Both brain and retina contain high transcript levels for components of the glycolytic/oxidative phosphorylation machinery, including cytochrome oxidase, creatine kinase, and glyceraldehyde 3-phosphate dehydrogenase. Based upon raw signal amplitude there was 28% matching identity between the fifty most abundant transcripts of brain and retina. Of these transcripts, 93% encode proteins with known functions. Overall, the metabolic and structural categories represented an aggregate of 40% and 20% of the fifty most abundant transcripts of brain and retina, respectively (Figure 1, Table 10, Table 11). The largest signals in the tissue comparison analysis (defined as log2 of the signal ratios between retina and brain) were observed for retina (Figure 2). Brain, however, exhibited the greater total number of tissue-specific transcripts.

Scatter plots indicated similarity in the gene expression profiles of control retina samples (Figure 3). Additionally, analysis of true control retina indicated that >60% of the fifty most abundant transcripts (approximately 40% of which encode for known components of the phototransduction machinery) were also within the group of fifty most abundant transcripts in every other sample of control retina (n=4). Cross-sample consistency within the ranking of the most abundant retinal transcripts is illustrated on Figure 4. These and other results confirmed the reproducibility and reliability of the DNA microarray technique in the analysis of retinal gene expression in zebrafish.

High abundance brain- or retina-specific transcripts were observed. These transcripts are graphically evident on Figure 1 as points that lie beyond the ten fold expression difference lines, being either brain- or retina-rich. Retina-rich transcripts are particularly evident as a broad cluster of points below the blue line of Figure 1, which denotes transcript levels that are ten fold greater in retina than brain. Of the fifty most retina-rich transcripts compared to brain, approximately 40% encode for known components of the phototransduction mechanism, including opsins, subunits of transducin (a GTP-binding protein), guanylyl cyclase activating proteins, and subunits of cGMP-dependent phosphodiesterase (Table 12). These and other observed retina-rich transcripts were consistent with the retina's specialized function: the transduction of photons into bioelectric signals [22]. Brain-rich transcripts were also observed (compared to control retina; Table 13). These transcripts included several transcription factors (e.g., pitx2, zic1, eng2a, and her4) and structural proteins (e.g., ependymin and isoforms of myosin). A greater number of brain-specific transcripts (3206) were observed relative to retina-specific transcripts (2160). The number of transcripts that were significantly, differentially expressed between zebrafish brain and retina represented over one third of the total number of oligonucleotides represented on the array.

PCR analysis provided qualitative and quantitative confirmation of the presence and tissue specificity of selected transcripts. RT-PCR amplification of the glycolytic enzyme phosphoglucose isomerase 1 (pgi1) was achieved from both brain- and retina-derived material (Figure 5, top; see also Figure 1). Common melt curve profiles of the pgi1 products derived from retina and brain material suggested identity (Figure 5, bottom), and this suggestion was confirmed by sequencing of the amplified product. In contrast the amplification of rhodopsin, which encodes the protein moiety of rod visual pigment, was abundant from retinal material but absent from brain-derived material (Figure 5, top; Figure 1). Melt curve analysis indicated a clear difference in the amplified rhodopsin products from retina and brain-derived material. Because the melt curve profile for brain-derived material was similar to that derived from control experiments in which no cDNA was present in the PCR reaction mixture (Figure 5, bottom), an absence of rhodopsin expression in brain was inferred. In the retinal cDNA samples a greater number of transcripts encoding rhodopsin than pgi1 was suggested from both the DNA microarray (Figure 1) and the RT-PCR analyses (Figure 5, top), with the former indicating a significant difference (respective signal amplitudes of 26900±2400 and 7100±1300 for rhodopsin and pgi1; p<0.001, Student's t-test, n=5). Real-time qPCR analysis supported these results.

Early changes in retinal gene expression

The ability of the Affymetrix Zebrafish Genome Array to detect expected similarities and differences in the gene expression profiles of zebrafish brain and retina motivated its application to the analysis of gene expression subsequent to retinal injury. In order to differentiate gene expression profiles associated with acute injury-response on the one hand, and cellular regenerative phenomena on the other [9,11], samples of post-lesion retinas were collected at 2, 3, and 5 d (acute injury) and 14 d post-lesion (regeneration), along with their corresponding time- and animal-matched control retinas.

For the 2 d post-lesion condition the absolute signal amplitudes for each transcript from control and post-lesion retina are plotted on Figure 6A, and the identity of the top fifty lesion- and non-lesion-rich transcripts, based upon log2 signal ratio analysis of the lesioned and non-lesioned samples, are summarized in Table 2 and Table 3, respectively. For both control and post-lesion retina the top transcripts, based upon signal level, were similar to those of true control retina (Table 11, Figure 3, Figure 4), indicating that retinal lesion does not induce a substantial, short-term override of the retina's primary gene expression profile (the high abundance of transcripts encoding phototransduction components). Retinal lesion did, however, induce statistically significant changes in gene expression within the affected retina, including both up- and downregulation. The tendency of data points in Figure 6 to be located between the green and red lines, as opposed to between the red and blue lines, indicates a general transcriptional enhancement, as opposed to transcriptional suppression, in post-lesion retina (Figure 7, cf. filled and open triangles).

The most significantly changed transcripts were individually analyzed based upon their known functional characteristics. Within the population of up- and downregulated transcripts (Table 2, Table 3) multiple functional categories were evident. Two of these functional categories were judged to be significant in terms of the number of exemplars and their hypothesized relationship to retinal injury and/or regeneration: genes implicated in cellular growth, and genes involved in cellular clearance/removal. Regarding the latter category mechanical lesion of the zebrafish retina inevitably generates cellular debris and damage, and the gene expression profile indicates the presence of active cellular mechanisms for resolving this damage. Specifically, several genes involved in macrophage activation or other immune system mechanisms were observed as targets of expression regulation, including complement C7 precursor [23], chemokine C-X-C motif receptor 4b (cxcr4b) [24], lymphocyte cytosolic plastin 1 (L-plastin) [25], perforin 1 precursor [26], and leukocyte surface antigen CD53 [27]. Furthermore, there was a significant upregulation in matrix metalloproteinase 9, a member of a family of proteins previously implicated in ocular wound healing [28].

The cell growth category of affected genes at 2 d post-lesion was broad. Although transcriptional changes for genes specifically implicated in cellular proliferation were relatively minor, examples of targeted genes within the cell growth category included alpha tubulin 1 [29], the transcription factor c-fos [30], and PHD-finger protein [31]. Members of the granulin family, which has previously been identified as a regulator of cell growth [32], also displayed significant changes in expression level, including hybrid granulin and progranulin. The expression of growth-associated protein 43 (GAP43) and cadherin 2 (cdh2) were also significantly enhanced in post-lesion retina. GAP43 has previously been implicated in retinal development [33] and its enhanced expression has been reported in response to optic nerve injury [34,35]; similarly, enhanced expression of cdh2 has previously been reported in injured retina [36]. The targeted regulation of these and functionally related genes 2 d following retinal lesion indicated the presence of inducible, early-response mechanisms for activating various molecular components of cellular growth phenomena.

RT-PCR and qPCR analyses confirmed lesion-induced changes in gene expression at 2 d post-lesion. Evidence for a lesion-induced increase in the expression of cxcr4b, cdh2, and GAP43 (representative examples of the cell clearance and cell growth categories) was confirmed qualitatively by gel electrophoresis analysis of products amplified by RT-PCR (Figure 8). Transcripts encoding pgi1 and rhodopsin, for which the DNA microarray analysis indicated no condition-dependent changes in expression (Figure 6), were unaffected in the RT-PCR analysis (Figure 8). Real-time qPCR analysis of rhodopsin, pgi1, cxcr4b, cdh2, and GAP43 supported the DNA microarray and RT-PCR results. Specifically, growth curve CT analysis indicated that retinal lesion did not affect pgi1 expression relative to rhodopsin (cf. green and black curves, and the equivalent-length green arrows, of Figure 9A,B). There was evidence, however, for a significant increase in the expression of cxcr4b, cdh2, and GAP43 (p<0.05, Student's t-test; Figure 9C). These increases in expression are evident graphically as lesion-dependent, leftward shifts in the qPCR-derived growth curves (Figure 9A,B, cf. orange, purple, and cyan arrows). The lesion-induced increases in cxcr4b, cdh2, and GAP43 expression inferred from the real-time qPCR analysis indicated substantial increases in the levels of these transcripts in 2 d post-lesion retina (Figure 9D).

Results similar to those of the 2 d post-lesion condition were observed for the 3 d post-lesion (Table 4, Table 5) and 5 d post-lesion (Table 6, Table 7) material. The 3 d and 5 d post-lesion material was operationally anticipated to replicate the 2 d post-lesion material because the degree of lesion-induced cellular activation in the zebrafish retina at these time points is similar [9]. Retinal injury did not change the profile of the most abundantly expressed transcripts at 3 and 5 d post-lesion, but as with the 2 d post-lesion material, genes involved in cellular clearance and growth (e.g., granulins, matrix metalloproteinase 9, GAP43), exhibited enhanced expression (Table 5, Table 7). These results indicated reproducibility in the DNA microarray analysis of gene expression profiles in the zebrafish retina, and reliability in the microarray-derived evidence for transcriptional regulation of cellular clearance and growth mechanisms in the acutely injured retina of adult zebrafish.

Late changes in retinal gene expression

For the 14 d post-lesion condition the absolute signal amplitudes for each transcript from control and post-lesion retina are plotted on Figure 6 (bottom), and the identity of the top condition-rich transcripts, based upon log signal ratio analysis, are summarized in Table 8 and Table 9. Similar to the 2 d post-lesion condition the top transcripts were similar to those of true control retina (Table 11), indicating that retinal lesion does not trigger a delayed, substantial override of the retina's primary gene expression profile. Retinal lesion did, however, induce statistically significant changes in gene expression within the affected retina at 14 d post-lesion. These changes included both an up- and downregulation in transcript abundance, with the former dominating (Figure 7, cf. filled and open circles; Table 8 and Table 9). The total number of significantly affected genes at 2 d post-lesion (1863) was significantly more than at 14 d post-lesion (782; Figure 7, cf. triangles and circles). This observation indicated a time-dependent decrease in lesion-induced gene expression changes that roughly parallels the time course of lesion site being filled with regenerated retina [9,11].

The most significantly changed transcripts at 14 d post-lesion were individually analyzed based upon their known functional characteristics. Within this population of up- and downregulated transcripts (Table 8, Table 9) the dominant functional category, represented by several affected genes, was cell cycle progression. The affected genes within this category include protein regulator of cytokinesis 1 [37], proliferation associated protein 100 (p100) [38], deoxycytidine kinase (dCK) [39], class I γ-tubulin [40], activating transcription factor 3 (ATF3) [41], cyclin B1 [42], and tumor suppressor p53-binding protein [43].

Another functional category that is apparently targeted for transcriptional regulation at 14 d post-lesion is cell growth/differentiation. Within this category are genes known to affect cellular development, such as engrailed 2b (eng2b) [44], zic2 [45], and madh7 (also termed smad7) [46]. Two other targeted genes are noteworthy for their reported importance in aspects of axonogenesis: GAP43, which was also targeted for regulation at 2 d post-lesion, and plasticin, a class III intermediate filament that is expressed by retinal ganglion cells that are regenerating their axons [47]. As in the 2 d post-lesion condition, there was also significant upregulation in the expression of matrix metalloproteinase 9 in 14 d post-lesion retina.

RT-PCR and qPCR analyses confirmed lesion-induced changes in gene expression at 14 d post-lesion. Evidence for a lesion-induced increase in the expression of γ-tubulin, plasticin, and GAP43 (representatives of the cell proliferation and cell growth categories) was confirmed qualitatively by gel electrophoresis analysis of products amplified by RT-PCR (Figure 10). Transcripts encoding pgi1 and rhodopsin, for which the DNA microarray analysis indicated no condition-dependent changes in expression (Figure 6), were similarly unaffected in the RT-PCR analysis (Figure 10). Real-time qPCR analysis of rhodopsin, pgi1, γ-tubulin, plasticin, and GAP43 supported the DNA microarray and RT-PCR results (Figure 11). Specifically, analysis of growth curve CT values (see Methods) indicated, as for the 2 d post-lesion condition, no change in the respective levels of pgi1 and rhodopsin expression (Figure 11A,B, cf. green arrows). There was, however, evidence for an increase in the expression of γ-tubulin, plasticin, and GAP43 in injured retina, evident graphically as a leftward shift in the growth curves for these products in material from post-lesion retina (Figure 11A,B, cf. red, blue, and cyan arrows). These lesion-induced changes in gene expression were all statistically significant (p<0.05; Figure 11C), and indicated substantial increases in the level of these transcripts in post-lesion retina (Figure 11D).

Gene group analysis

Specific gene groups were statistically evaluated for induced transcriptional regulation by retinal lesion. This analysis (which, although objective, was not exhaustive) revealed several gene groups that were apparent targets for transcription regulation at 2 d and/or 14 d post-lesion (Table 14). In the Cellular Group, the synaptic vesicle category was a target for downregulation at both 2 d and 14 d post-lesion. The expression of the RNA polymerase III category of the Molecular Group, in contrast, was differentially regulated: at 2 d the category was significantly upregulated, but this was followed at 14 d by a significant downregulation. These results indicate the operation of complex, lesion-induced gene expression mechanisms that target groups of related genes.

Additional evidence for condition-dependent regulation of specific gene groups was observed. Within the Cellular Group at 2 d post-lesion significant enhancement in gene expression was observed for the ribosome, nucleosome, and heterochromatin categories. Within the Biology Group at 2 d post-lesion a significant downregulation in expression was observed for the guanylyl cyclase category, and within the Molecular Group the hexokinase category was significantly enhanced. These gene categories were not significantly affected at 14 d post-lesion, and no other significant changes in gene group expression were detected at 2 d post-lesion. Several gene categories within the Molecular Group were, however, targeted at 14 d post-lesion: RNA polymerase III (downregulated), TNF receptor ligand (downregulated) cAMP-dependent kinase (downregulated), and Notch receptor ligand (upregulated). Regarding the latter category, which represents ubiquitous signaling mechanisms for controlling cell specification [48-52], an upregulation of notch2 expression was confirmed with RT-PCR analysis at 2 d and 14 d post-lesion (Figure 8, Figure 10). These results indicate that within and across functional gene groups, gene expression within post-lesion retina is differentially regulated in a time-dependent manner.


The current investigation revealed that retinal lesion in the adult zebrafish triggers a complex, but apparently targeted, set of changes in the retina's gene expression profile. The identities and signs of the transcriptional changes are consistent with cellular phenomena known to occur within the retina at specific time points subsequent to lesion, indicating a general applicability of the data to investigations of molecular mechanisms that enable successful retinal regeneration. The large number of significantly affected genes in post-lesion retina precludes a detailed analysis and discussion of each observed transcriptional change and their hypothesized functional significance. Similarly, detailed discussion of the comparative gene expression profiles of zebrafish brain and retina is not possible here, although the observed profiles were as expected [21]. Our discussion is instead focused upon a limited set of overt correlations between injury-induced transcriptional changes and cellular phenomena known to occur during retinal regeneration.

Transcriptional changes in post-lesion retina

Subsequent to retinal lesion in adult zebrafish a complex set of cellular events occur, including the clearance of cellular debris, cellular entry into and progression through the cell cycle, induction of stem/precursor cells, and cellular and axonal regeneration. Although the determinants of these phenomena are not completely understood, the current study provides explicit information about some of the molecular mechanisms that may underlie retinal regeneration.

A substantial number of genes that are transcriptionally enhanced in 2 d post-lesion retina are associated with immune mechanisms (Table 3), and their presence is consistent with an early, immune system-mediated clearance of cellular debris [53]. The relative dearth of changes in proliferation-associated genes at 2 d post-lesion is consistent with the relatively low number of retinal cells progressing through the cell cycle at that time [9]. By 14 d post-lesion, however, there is a roughly 105 fold increase in the number of proliferating cells in post-lesion retina compared to control [9], and the current study suggests that this induced proliferation is substantially driven by enhanced gene transcription (Table 8, Table 9). For example, both proliferation associated protein 100 (p100) and protein regulator of cytokinesis 1 are functionally associated with the mitotic spindle and are present at high levels during phases S, G2, and M [37,38]. Deoxycytidine kinase (dCK) and activating transcription factor 3 (ATF3) are important contributors to DNA synthesis (the former phosphorylates deoxyribonucleosides, whereas the latter is an activator of cyclin D1 and thus enables progression through the G1/S transition [39,41]), consequently their enhanced expression is consistent with cellular proliferation. Lastly, a functional role for class I γ-tubulin in mitosis is known [40] and, like the genes discussed above, its enhanced expression in post-lesion retina is consistent with cellular proliferation. These observations indicate that retinal injury regulates, either directly or indirectly, the expression of multiple proteins that collectively promote cell cycle progression. The injury-induced upstream signal(s) that triggers this targeted regulation is unknown, but could involve mitogenic growth factors [54,55], perhaps including members of the EGF family (Table 3).

Conversely, the targeted transcriptional downregulation of specific genes in post-lesion retina is also consistent with cell cycle progression. For example, the downregulation of tumor suppressor p53-binding protein in 14 d post-lesion retina (Table 8) is likely to be significant because of its known ability to bind to p53 and enhance p53-mediated transcriptional activation [43]. The p53 protein functions as a tumor suppressor by negatively regulating cell cycle progression, and mutations in p53 are known to be associated with a large number of human cancers [56,57]. Suppression of p53-binding protein expression thus represents a potential mechanism for removing cell cycle arrest. Furthermore, because p53 has been implicated in the G0/G1 transition [58,59], we hypothesize that in quiescent retinal cells of the zebrafish lesion-induced negative regulation of p53-mediated signaling directly contributes to these cells' entry into the cell cycle.

Axonal regeneration in the damaged retina and optic nerve of teleosts has received considerable experimental attention [60,61]. Axonal regeneration is an important component of retinal repair, and it encompasses both the re-extension of severed axons from extant ganglion cells and de novo axonogenesis from newly-born cells. The current study supplements earlier investigations by confirming the importance of transcriptional regulation in the triggering and manifestation of axonal regeneration. For example, the intermediate filament protein plasticin is present at high abundance in newly-born and regenerating retinal ganglion cell axons in fish [47], and is upregulated in zebrafish retina following optic nerve injury [62]. The elevated expression of plasticin observed in the current investigation is therefore consistent with a functional importance for axonal regeneration. Additionally, GAP43, which is upregulated in post-lesion zebrafish retina, has previously been implicated in axonal regeneration in several model systems [34,63-65]. These observations indicate that injury-induced axonal regeneration in adult zebrafish involves substantial transcriptional regulation of multiple genes.

Implications for retinal regeneration in higher vertebrates

Retinal lesion in adult mammals leads to reactive gliotic phenomena such as proliferative vitreoretinopathy and subretinal fibrosis [66]. Although not explicitly examined, previous investigations have revealed little evidence for such gliotic phenomena in post-lesion regenerating zebrafish retina [9,11]. This observation is of potential significance because gliotic structures are known to inhibit regeneration in the injured CNS [67,68]. In the adult fish retina a positive correlation might therefore exist between the absence of reactive gliogenic phenomena on the one hand, and the manifestation of neuronal regeneration (both neuronogenesis and axonal regeneration) on the other. Establishing the molecular foundation of this positive correlation is the topic of ongoing investigation.

Lastly, recent evidence from several groups has implicated Müller glia as an important cellular component of regenerative neuronogenesis in the retina. For example, in systems that normally regenerate retinas via injury-induced, proliferative precursors, it is known that Müller glia contribute to the population of cells that enter the cell cycle following injury [5-10]. Even in mammalian systems that typically are not thought to support neuronal regeneration, recent evidence suggests that Müller glia induced to proliferate following injury may function as a source of new retinal neurons [10]. Because Müller glia may therefore be an important cellular substrate for many of the transcriptional events described in this report, targeted manipulations in the gene expression profile of Müller glia might provide a beneficial strategy for inducing mechanisms of neuronal, and potentially functional, restoration in the damaged retinas of adult humans.


The authors thank Michelle Mader, Eduardo Solessio, and Melinda Tyler for discussions. This work was supported in part by the Edward F. MacNichol Memorial Fund, SUNY Upstate.


1. Reh TA, Levine EM. Multipotential stem cells and progenitors in the vertebrate retina. J Neurobiol 1998; 36:206-20.

2. Raymond PA, Hitchcock PF. How the neural retina regenerates. Results Probl Cell Differ 2000; 31:197-218.

3. Stenkamp DL, Cameron DA. Cellular pattern formation in the retina: retinal regeneration as a model system. Mol Vis 2002; 8:280-93 <>.

4. Otteson DC, Hitchcock PF. Stem cells in the teleost retina: persistent neurogenesis and injury-induced regeneration. Vision Res 2003; 43:927-36.

5. Fischer AJ, Reh TA. Muller glia are a potential source of neural regeneration in the postnatal chicken retina. Nat Neurosci 2001; 4:247-52.

6. Wu DM, Schneiderman T, Burgett J, Gokhale P, Barthel L, Raymond PA. Cones regenerate from retinal stem cells sequestered in the inner nuclear layer of adult goldfish retina. Invest Ophthalmol Vis Sci 2001; 42:2115-24.

7. Fischer AJ, Reh TA. Potential of Muller glia to become neurogenic retinal progenitor cells. Glia 2003; 43:70-6.

8. Garcia M, Vecino E. Role of Muller glia in neuroprotection and regeneration in the retina. Histol Histopathol 2003; 18:1205-18.

9. Yurco P, Cameron DA. Responses of Muller glia to retinal injury in adult zebrafish. Vision Res 2005; 45:991-1002.

10. Ooto S, Akagi T, Kageyama R, Akita J, Mandai M, Honda Y, Takahashi M. Potential for neural regeneration after neurotoxic injury in the adult mammalian retina. Proc Natl Acad Sci U S A 2004; 101:13654-9.

11. Cameron DA. Cellular proliferation and neurogenesis in the injured retina of adult zebrafish. Vis Neurosci 2000; 17:789-97.

12. Cameron DA, Carney LH. Cell mosaic patterns in the native and regenerated inner retina of zebrafish: implications for retinal assembly. J Comp Neurol 2000; 416:356-67.

13. Vihtelic TS, Hyde DR. Light-induced rod and cone cell death and regeneration in the adult albino zebrafish (Danio rerio) retina. J Neurobiol 2000; 44:289-307.

14. Hitchcock PF, Lindsey Myhr KJ, Easter SS Jr, Mangione-Smith R, Jones DD. Local regeneration in the retina of the goldfish. J Neurobiol 1992; 23:187-203.

15. Cameron DA, Easter SS Jr. Cone photoreceptor regeneration in adult fish retina: phenotypic determination and mosaic pattern formation. J Neurosci 1995; 15:2255-71.

16. Mader MM, Cameron DA. Photoreceptor differentiation during retinal development, growth, and regeneration in a metamorphic vertebrate. J Neurosci 2004; 24:11463-72.

17. Luu-The V, Paquet N, Calvo E, Cumps J. Improved real-time RT-PCR method for high-throughput measurements using second derivative calculation and double correction. Biotechniques 2005; 38:287-93.

18. Middleton FA, Ramos EJ, Xu Y, Diab H, Zhao X, Das UN, Meguid M. Application of genomic technologies: DNA microarrays and metabolic profiling of obesity in the hypothalamus and in subcutaneous fat. Nutrition 2004; 20:14-25.

19. Middleton FA, Pato CN, Gentile KL, McGann L, Brown AM, Trauzzi M, Diab H, Morley CP, Medeiros H, Macedo A, Azevedo MH, Pato MT. Gene expression analysis of peripheral blood leukocytes from discordant sib-pairs with schizophrenia and bipolar disorder reveals points of convergence between genetic and functional genomic approaches. Am J Med Genet B Neuropsychiatr Genet 2005; 136:12-25.

20. Zhang Y, James M, Middleton FA, Davis RL. Transcriptional analysis of multiple brain regions in Parkinson's disease supports the involvement of specific protein processing, energy metabolism, and signaling pathways, and suggests novel disease mechanisms. Am J Med Genet B Neuropsychiatr Genet 2005; 137:5-16.

21. Hackam AS, Qian J, Liu D, Gunatilaka T, Farkas RH, Chowers I, Kageyama M, Parmigiani G, Zack DJ. Comparative gene expression analysis of murine retina and brain. Mol Vis 2004; 10:637-49 <>.

22. Arshavsky VY, Lamb TD, Pugh EN Jr. G proteins and phototransduction. Annu Rev Physiol 2002; 64:153-87.

23. Hobart MJ. Evolution of the terminal complement genes: ancient and modern. Exp Clin Immunogenet 1998; 15:235-43.

24. Lee C, Liu QH, Tomkowicz B, Yi Y, Freedman BD, Collman RG. Macrophage activation through CCR5- and CXCR4-mediated gp120-elicited signaling pathways. J Leukoc Biol 2003; 74:676-82.

25. Crowhurst MO, Layton JE, Lieschke GJ. Developmental biology of zebrafish myeloid cells. Int J Dev Biol 2002; 46:483-92.

26. Trapani JA, Smyth MJ. Functional significance of the perforin/granzyme cell death pathway. Nat Rev Immunol 2002; 2:735-47.

27. Olweus J, Lund-Johansen F, Horejsi V. CD53, a protein with four membrane-spanning domains, mediates signal transduction in human monocytes and B cells. J Immunol 1993; 151:707-16.

28. Sivak JM, Fini ME. MMPs in the eye: emerging roles for matrix metalloproteinases in ocular physiology. Prog Retin Eye Res 2002; 21:1-14.

29. Hadfield JA, Ducki S, Hirst N, McGown AT. Tubulin and microtubules as targets for anticancer drugs. Prog Cell Cycle Res 2003; 5:309-25.

30. Karin M, Liu Z, Zandi E. AP-1 function and regulation. Curr Opin Cell Biol 1997; 9:240-6.

31. Aasland R, Gibson TJ, Stewart AF. The PHD finger: implications for chromatin-mediated transcriptional regulation. Trends Biochem Sci 1995; 20:56-9.

32. Bateman A, Bennett HP. Granulins: the structure and function of an emerging family of growth factors. J Endocrinol 1998; 158:145-51.

33. Ekstrom P, Johansson K. Differentiation of ganglion cells and amacrine cells in the rat retina: correlation with expression of HuC/D and GAP-43 proteins. Brain Res Dev Brain Res 2003; 145:1-8.

34. Soto I, Marie B, Baro DJ, Blanco RE. FGF-2 modulates expression and distribution of GAP-43 in frog retinal ganglion cells after optic nerve injury. J Neurosci Res 2003; 73:507-17.

35. Udvadia AJ, Koster RW, Skene JH. GAP-43 promoter elements in transgenic zebrafish reveal a difference in signals for axon growth during CNS development and regeneration. Development 2001; 128:1175-82.

36. Liu Q, Londraville RL, Azodi E, Babb SG, Chiappini-Williamson C, Marrs JA, Raymond PA. Up-regulation of cadherin-2 and cadherin-4 in regenerating visual structures of adult zebrafish. Exp Neurol 2002; 177:396-406.

37. Mollinari C, Kleman JP, Jiang W, Schoehn G, Hunter T, Margolis RL. PRC1 is a microtubule binding and bundling protein essential to maintain the mitotic spindle midzone. J Cell Biol 2002; 157:1175-86.

38. Heidebrecht HJ, Buck F, Steinmann J, Sprenger R, Wacker HH, Parwaresch R. p100: a novel proliferation-associated nuclear protein specifically restricted to cell cycle phases S, G2, and M. Blood 1997; 90:226-33.

39. Weber G, Singhal RL, Abonyi M, Prajda N, Hata Y, Szekeres T, Yeh A, Look KY. Regulation of deoxycytidine kinase activity and inhibition by DFDC. Adv Enzyme Regul 1993; 33:39-59.

40. Hoyle HD, Hutchens JA, Turner FR, Raff EC. Regulation of beta-tubulin function and expression in Drosophila spermatogenesis. Dev Genet 1995; 16:148-70.

41. Allan AL, Albanese C, Pestell RG, LaMarre J. Activating transcription factor 3 induces DNA synthesis and expression of cyclin D1 in hepatocytes. J Biol Chem 2001; 276:27272-80.

42. Barnes EA, Kong M, Ollendorff V, Donoghue DJ. Patched1 interacts with cyclin B1 to regulate cell cycle progression. EMBO J 2001; 20:2214-23.

43. Iwabuchi K, Li B, Massa HF, Trask BJ, Date T, Fields S. Stimulation of p53-mediated transcriptional activation by the p53-binding proteins, 53BP1 and 53BP2. J Biol Chem 1998; 273:26061-8.

44. Retaux S, Harris WA. Engrailed and retinotectal topography. Trends Neurosci 1996; 19:542-6.

45. Brewster R, Lee J, Ruiz i Altaba A. Gli/Zic factors pattern the neural plate by defining domains of cell differentiation. Nature 1998; 393:579-83.

46. Zhu HJ, Burgess AW. Regulation of transforming growth factor-beta signaling. Mol Cell Biol Res Commun 2001; 4:321-30.

47. Glasgow E, Druger RK, Levine EM, Fuchs C, Schechter N. Plasticin, a novel type III neurofilament protein from goldfish retina: increased expression during optic nerve regeneration. Neuron 1992; 9:373-81.

48. Lundkvist J, Lendahl U. Notch and the birth of glial cells. Trends Neurosci 2001; 24:492-4.

49. Morrison SJ. Neuronal potential and lineage determination by neural stem cells. Curr Opin Cell Biol 2001; 13:666-72.

50. Gaiano N, Fishell G. The role of notch in promoting glial and neural stem cell fates. Annu Rev Neurosci 2002; 25:471-90.

51. Harper JA, Yuan JS, Tan JB, Visan I, Guidos CJ. Notch signaling in development and disease. Clin Genet 2003; 64:461-72.

52. Voas MG, Rebay I. Signal integration during development: insights from the Drosophila eye. Dev Dyn 2004; 229:162-75.

53. Lillo C, Velasco A, Jimeno D, Cid E, Aijon J, Lara JM. Non-neuronal cells involved in the degeneration and regeneration of the fish retina. J Neurocytol 2001; 30:475-91.

54. Boucher SE, Hitchcock PF. Insulin-related growth factors stimulate proliferation of retinal progenitors in the goldfish. J Comp Neurol 1998; 394:386-94.

55. Fischer AJ, McGuire CR, Dierks BD, Reh TA. Insulin and fibroblast growth factor 2 activate a neurogenic program in Muller glia of the chicken retina. J Neurosci 2002; 22:9387-98.

56. De Vries EM, Ricke DO, De Vries TN, Hartmann A, Blaszyk H, Liao D, Soussi T, Kovach JS, Sommer SS. Database of mutations in the p53 and APC tumor suppressor genes designed to facilitate molecular epidemiological analyses. Hum Mutat 1996; 7:202-13.

57. Hainaut P, Hernandez T, Robinson A, Rodriguez-Tome P, Flores T, Hollstein M, Harris CC, Montesano R. IARC Database of p53 gene mutations in human tumors and cell lines: updated compilation, revised formats and new visualisation tools. Nucleic Acids Res 1998; 26:205-13.

58. Dash BC, El-Deiry WS. Cell cycle checkpoint control mechanisms that can be disrupted in cancer. Methods Mol Biol 2004; 280:99-161.

59. Hofseth LJ, Hussain SP, Harris CC. p53: 25 years after its discovery. Trends Pharmacol Sci 2004; 25:177-81.

60. Stuermer CA, Bastmeyer M, Bahr M, Strobel G, Paschke K. Trying to understand axonal regeneration in the CNS of fish. J Neurobiol 1992; 23:537-50.

61. Bernhardt RR. Cellular and molecular bases of axonal regeneration in the fish central nervous system. Exp Neurol 1999; 157:223-40.

62. Asch WS, Leake D, Canger AK, Passini MA, Argenton F, Schechter N. Cloning of zebrafish neurofilament cDNAs for plasticin and gefiltin: increased mRNA expression in ganglion cells after optic nerve injury. J Neurochem 1998; 71:20-32.

63. Van der Zee CE, Nielander HB, Vos JP, Lopes da Silva S, Verhaagen J, Oestreicher AB, Schrama LH, Schotman P, Gispen WH. Expression of growth-associated protein B-50 (GAP43) in dorsal root ganglia and sciatic nerve during regenerative sprouting. J Neurosci 1989; 9:3505-12.

64. Frey D, Laux T, Xu L, Schneider C, Caroni P. Shared and unique roles of CAP23 and GAP43 in actin regulation, neurite outgrowth, and anatomical plasticity. J Cell Biol 2000; 149:1443-54.

65. Cafferty WB, Gardiner NJ, Das P, Qiu J, McMahon SB, Thompson SW. Conditioning injury-induced spinal axon regeneration fails in interleukin-6 knock-out mice. J Neurosci 2004; 24:4432-43.

66. Fisher SK, Lewis GP. Muller cell and neuronal remodeling in retinal detachment and reattachment and their potential consequences for visual recovery: a review and reconsideration of recent data. Vision Res 2003; 43:887-97.

67. Levine JM, Reynolds R, Fawcett JW. The oligodendrocyte precursor cell in health and disease. Trends Neurosci 2001; 24:39-47.

68. Chen ZJ, Negra M, Levine A, Ughrin Y, Levine JM. Oligodendrocyte precursor cells: reactive cells that inhibit axon growth and regeneration. J Neurocytol 2002; 31:481-95.

Cameron, Mol Vis 2005; 11:775-791 <>
©2005 Molecular Vision <>
ISSN 1090-0535