| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
From the Department of Genetics, Howard Hughes Medical Institute, Harvard Medical School, Boston, Massachusetts.
 |
Abstract |
|---|
 Top Abstract Material and MethodsResultsDiscussionReferences |
|---|
METHODS. A preliminary microarray analysis led to the selection of 169 candidate genes that might change in level of expression during degeneration. Probes corresponding to these genes were used for in situ hybridization on tissue during the period of cone death. Expression values were assigned to the intensities of in situ hybridization signals and were compared between mutant and wild-type tissue.
RESULTS. During the peak of cone death, the in situ hybridization signals were typically higher in the mutant. This signal change was often true of genes with a wild-type pattern of expression in ganglion cells, bipolar cells, and/or Müller glia. In such cases, the upregulation was highest in bipolar cells and/or Müller glia.
CONCLUSIONS. All retinal cell types responded during the process of retinal degeneration, as revealed by changes in gene expression. Genes that showed changes in the in situ hybridization signals during the period of cone death were typically higher in the mutant, with many of them expressed in both the ganglion cell layer and the inner nuclear layer.
Retinal degenerations (RDs) are retinal diseases often involving the loss of PRs. The disease with the largest number of affected individuals worldwide (20–25 million) is age-related macular degeneration (AMD), with a typical disease onset of >55 years of age.3 Whereas AMD is thought to be caused by the interplay of environmental and genetic factors, other RDs appear to be the result of the simple Mendelian inheritance of a disease-causing allele. To date, 161 loci have been mapped, and for 112 loci, a specific gene has been linked to the disease (RetNet, http://www.sph.uth.tmc.edu/Retnet/ provided in the public domain by the University of Texas Houston Health Science Center, Houston, TX). Interestingly, a high percentage of RD genes are rod-specific, particularly in retinitis pigmentosa (RP), another late-onset RD. In cases in which the disease gene is rod-specific, rods die first, and then cone death follows, with kinetics depending on the particular disease. In humans, there is no RD known in which rods die and cones survive. Mutations in rod-specific genes thus appear to induce cone degeneration by mechanisms that are not intrinsic to cones. There are mutations in cone-specific genes that lead only to cone death.4
Naturally occurring mutations in animals, or experimentally engineered mice with RD, provide valuable models to study nonautonomous cone death. Although rodents have no macula, RD is similar to that in humans. Rod-specific mutations that have been found or introduced lead to rod death and subsequent cone death. One such model of RD is the rd1 mutant mouse,5 which carries a mutation of the rod-specific cGMP phosphodiesterase-ß-subunit (PDEß), which, when mutated, also causes RD in humans.6
The reasons for cone death in rd1, as in other forms of RD, remain a mystery. An important step in approaching this problem is to understand the changes that occur during rod and cone death. A well-described phenomenon is retinal remodeling, which occurs during periods of rod and cone death.7 8 9 10 11 Although axonal arborization is first hypertrophic for certain inner nuclear layer (INL) cells during the period of rod death and bipolar (BP) dendrites start to disappear during the period of cone death, the INL and the ganglion cell layer (GCL) cells seem to survive in significant numbers. Thus the question arises, why do the other neurons in the retina survive when the majority of the PRs die? This question may be particularly apt for the murine models, as rods account for >70% of all retinal cells. It is thus of interest to characterize the response of retinal cells to PR death, to gain some insight into possible protective mechanisms that lead to the survival of cells of the INL and ganglion cell layer (GCL).
Little is known concerning how the different retinal cell types respond to the death of rods and cones in terms of gene expression. Models have been proposed for the cause of nonautonomous cone death, such as the loss of a trophic factor produced normally by rods and required for cone survival. Studies have provided evidence for such a factor, although proof that this accounts entirely for cone death is lacking.12 13 14 15 16 Alternative models, in which dying rods either directly or indirectly lead to a toxic situation for cones, perhaps via the reactions of cells that are intimate with PRs, such as Müller glia (MG), can also be entertained.
Recent studies have applied microarray and proteomics technologies to characterize gene expression during rod and cone death in mouse models with a rod-specific mutation.17 18 19 20 Although these studies provide a catalog of candidate genes, interpretations often remain limited because of the lack of information concerning the cellular locus of expression. To gain more insight into the responses of different retinal cell types in RD, as well as create a list of genes that respond and perhaps influence either the nonautonomous cone death or the survival of non-PR retinal cells, we undertook a screening of the cellular expression patterns of genes that change in rd1. Genes were selected by means of a preliminary screening via microarray analysis, and the gene expression patterns were compared between mutant and wild-type (wt) tissue during cone degeneration. The genes which showed a change in expression intensity during cone death were more likely to be upregulated and were expressed throughout the INL and the GCL of the mutant retina. Such genes often were most highly upregulated in BP cells and/or MG.
|
Material and Methods |
|---|
|
TopAbstractMaterial and MethodsResultsDiscussionReferences |
|---|
Molecular Methods
All experimental procedures performed with ready-to-use reagents were performed according to the manufacturer’s instructions, if not otherwise stated. RNA extractions were performed (TRIzol; Invitrogen, Carlsbad, CA) and quality was assessed (model 2100 Bioanalyzer; Agilent Technologies, Palo Alto, CA).
Real Time PCR
Total RNA was extracted at the time points indicated in the Results section. One microgram of total RNA per time point was used for reverse transcription (Transcriptor; Roche, Indianapolis, IN). RT-PCR was performed (Light Cycler 2000; Roche) with DNA Master SYBR Green I reagents (Roche). The following primers were used for the RT-PCR: Opn1sw forward 5'-AAAGGCTGAACGGGAG-3'; Opn1sw reverse 5'-ACACGTCAGATTCGTCT-3'; rhodopsin forward 5'-TGCCAATATGCCCAC-3'; rhodopsin reverse 5'-GTACCCCATAGTTCCTGT-3'; Gapdh forward 5'-TCGGTGTGAACGGATT-3'; Gapdh reverse 5'-AAGACGCCAGTAGACT-3'. Conditions for the RT-PCR were as follows: Opn1sw 95° for 2 seconds, 51° for 10 seconds, and 72° for 25 seconds and rhodopsin and Gapdh 95° for 0 seconds, 61° for 10 seconds, and 72° for 20 seconds. Each experiment was performed in triplicate, and the data were averaged.
Rd1 Library
The rd1 library was generated from 5-week-old rd1 retinal RNA. The cDNA library was generated and cloned (pBluescript II XR; cDNA Library Construction Kit manual; cat no. 200455; Stratagene, La Jolla, CA).
Microarray Analysis
RNA extractions were performed at postnatal days (P)42 and P63. Hybridizations and slide scanning were performed with a kit according to the manufacturer’s instructions (Genisphere, Hatfield, PA, as described by Corbo and Cepko21 ). Total RNA was extracted from five animals and pooled, and 3 µg of the total RNA was used per probe and hybridization. Experiments were performed in duplicate for each time point and array type. Because only one dye per slide, time point, and array was used, four slides per time point and array type were necessary to perform the experiments in duplicate, two for the wt and two for the rd1 mouse. Thus, a total of eight slides were used per array for the two time points. The microarray analyses were performed on customized cDNA arrays. One microarray contained a retinal EST collection created by Bento Soares (University of Iowa, Ames, IA) encompassing 12,679 cDNAs, referred to here as the 12K array. These ESTs came from wt retinas. Thus, disease-specific genes that are not expressed at significant levels in wt retina are unlikely to be found on such an array. To account for such genes, RNA was extracted from 5-week-old rd1 retinas, and a cDNA library was made. The library was transformed into bacteria to generate a collection of 17,328 clones. Because the library is unsequenced, a prescreening was conducted to select only those clones that change during degeneration. The 17,328 clones were hybridized with 5-week-old rd1 and wt retinal RNA. From the library, 2610 genes were selected that showed potentially significant changes during degeneration. In addition to these genes, 1230 genes from either a laboratory collection of known retinal cDNAs or from an older version of a customized laboratory array (mouse brain ESTs: Bento Soares) were used to generate a minichip of 3840 cDNAs. Both the 12K array and the minichip were then used to select genes for the ISH analysis. A summary of the final 169 genes (123 from the 12K array and 46 from the minichip) selected for the ISH study is presented in Supplementary Table S1 (all supplementary tables are online at http://www.iovs.org/cgi/content/full/48/2/849/DC1). Genes of the 12K array were selected according to the following criteria: First, the average of the signal intensity of the duplicates had to be at least 1000 at one time point, either in the mutant or in the wt. Second, the change had to be more than twofold at both time points examined. This reduced the list to 283 for the genes that were higher in expression in wt and 9 for those that were lower. The list of genes that were higher in the wt comprised primarily PR genes. This list was reduced by removing known PR genes. The list of downregulated genes was increased by clustering (Cluster3 software; Eisen Laboratory: http://rana.lbl.gov/EisenSoftware.htm) the genes that passed the first criteria. This allowed an examination of a more comparable number of up- and downregulated genes. Only genes that clustered closest to the original nine downregulated genes were chosen. Some of the genes chosen through clustering had ratios of >2, but not at both time points, and thus were not picked up by the first selection criterion. A final list of 123 genes was compiled from the 12K array (Supplementary Table S1). Genes from the mini-chip were selected by using a method similar to the one used for the 12K array. The signal intensity was reduced to 500 because of the smaller range between the maximum and minimum signals. Because no sequence information was available at the time the microarray analysis was conducted, only a small number of genes was used for the ISH (n = 46). Meanwhile in a combined effort with the National Eye Institute (NEI) roughly 50% of the 2610 clones from the rd1 library have been sequenced. The sequences are available in the public domain at the National Center for Biotechnology Web page (www.ncbi.nlm.nih.gov) with accession numbers ranging from DW711190 to DW712659.
Histologic Methods
TUNEL.
TUNEL analysis was performed with kit (In Situ Cell Detection Kit; Roche) on 14-µm paraffin-embedded tissue sections. The procedure was the same as the one used for the ISH up to the rehydration in PBS; afterward the manufacturer’s instructions were followed.
Cell Counting.
Cell counting was performed as described (Young and Cepko22 ). Cells were detected with a Cy3-
-mouse antibody (Jackson Immunoresearch Laboratory) and counterstained with 4',6'-diamino-2-phenylindole [DAPI]. A minimum of five retinas was analyzed with a total cell count of 2000 cells.
X-gal Histochemistry.
X-Gal staining was performed according to standard procedures.23 Retinas were fixed for 15 minutes and stained at room temperature overnight.
In Situ Hybridization.
Section ISH was performed on 14-µm paraffin-sections. In brief, the retinas were dissected in PBS, fixed for 30 minutes in 4% PFA/PBS, washed three times for 10 minutes each in PBS, dehydrated in a series of increasing ethanol (EtOH) concentrations (25%, 50%, 75% EtOH/H2O and 2x 100% EtOH 10 minutes each) cleared two times for 5 minutes each in xylene, transferred for 20 minutes in a 1:1 mixture of xylene:paraffin (65°C), washed three times for 20 minutes each in paraffin (65°C), and incubated overnight at 65°C in paraffin. After mounting and sectioning, the slides were dried overnight at 37°C and stored at 4°C until used. Before rehydration of the tissue, the slides were heated at 65°C for 1 hour, cooled quickly, and rehydrated two times for 5 minutes each in xylene, two times for 5 minutes each in EtOH, and then in decreasing concentrations of ethanol to increasing PBS concentrations for 5 minutes each. The sections were fixed (4% PFA/PBS) for 10 minutes, washed three times for 5 minutes each in PBT (PBS 0.1% Tween-20), proteinase K treated for 10 minutes (1µg/mL), washed two times for 5 minutes each in PBT, postfixed in 4% PFA/PBS for 5 minutes, washed three times for 5 minutes each in PBT, acetylated for 10 minutes (0.1 M triethanolamine/25 mM acetic acid anhydrid), washed three times for 5 minutes each in PBT, air dried for 10 minutes, and prehybridized for 1 hour in hybridization buffer (10 mM Tris [pH 7.5], 0.6 M NaCl, 1 mM EDTA, 0.25% SDS, 10% dextran sulfate, 1x Denhardt’s, 200 µg/mL yeast tRNA, and 50% formamide). Hybridizations were performed overnight at 70°C. Riboprobes were synthesized from 300 ng PCR (T3/T7) product of the ESTs described in Supplementary Table S1 and labeled using digoxigenin-tagged UTP (Roche). RNA synthesis was performed in a total of 20 µL. After synthesis, the probe was treated for 15 minutes with DNase (RNase free; Roche), precipitated, washed with 70% EtOH, air dried, and resuspended in 100 µL H2O. Three of 100 µL probes were used in 150-µL hybridization buffer. The following day, cover slips were removed in 5x SSC, and the slides were washed for 30 minutes in 1x SSC/50% formamide at 65°C, 10 minutes in TNE (10 mM Tris [pH 7.5], 500 mM NaCl, and 1 mM EDTA) at 37°C, 30 minutes in TNE+RNaseA (20 µg/mL; Roche) at 37°C, 10 minutes in TNE at 37°C, 20 minutes in 2x SSC at 65°C, and two times for 20 minutes each in 0.2x SSC at 65°C. The slides were then washed in MABT (100 mM maleic acid, 150 mM NaCl, and 0.1% Tween; pH adjusted to pH 7.5) two times for five minutes each, blocked for 30 minutes in MABT+20% heat-inactivated sheep serum, incubated 2 hours in blocking solution+ -digoxigenin-alkaline-phosphatase (1:2500; Roche), washed three times for five minutes each with MABT, and then detected with NBT/BCIP (5-bromo-4-chloro-3-indoyl phosphate–nitroblue tetrazolium; Sigma-Aldrich, St. Louis, MO). The ISH experiments are shown as supplementary figures, available online at http://www.iovs.org/cgi/content/full/48/2/849/DC1, named according to the clusters for their wt expression values (clusters 1–8). The distribution of the 169 genes into their respective clusters is shown in Supplementary Table S2.
|
Results |
|---|
|
TopAbstractMaterial and MethodsResultsDiscussionReferences |
|---|
Second, quantitative real-time PCR was performed to measure the loss of a rod- and cone-specific RNA species. RNA of mutant and wt retina from different time points was analyzed for either rhodopsin or Opn1sw (Opsin1 short wave length is also referred to as blue cone opsin). Since the TUNEL analysis showed no significant number of TUNEL+ cells up to P10, RNA was extracted from P10 and every other day up to P20, and then every week up to postnatal week 11. Primers specific for the housekeeping gene, Gapdh, were used as a control for the amount of total RNA. Concordant with the TUNEL analysis and the DAPI staining of sections (Supplementary Fig. S1), a rapid decrease in rhodopsin transcript was observed from P10 to P20 (Fig. 1A) . After P20, the rhodopsin transcripts reached a low level, which was stable over the remaining period, reflecting a small number of rods still present in the periphery of the retina.25 The initial increase of rhodopsin transcript in wt retinas from P10 to P16 is due to the postnatal maturation of the retina.26 27 The blue opsin transcript levels in the rd1 mutant started to diverge from the wt samples only after P35, and continued to descend until P70 (Fig. 1B) . The initial increase followed by a decrease in wt retina correlates with the fact that the gene is expressed initially throughout the entire retina and is then downregulated in the dorsal retina.28 29
|
-rhodopsin antibody. A linear decrease in rhodopsin-positive cells was observed, suggesting that, at any given time, the percentage of dying rods was the same (Fig. 1C) .30 Because the total percentage of cones is very low in the mouse retina and the changes occur over an extended period, cone death was monitored by means of a cone-specific lacZ line that was crossed into a rd1 mutant background. This line expresses ß-galactosidase under the control of the human red-green opsin promoter but is expressed in all cones of the murine retina.31 This approach allowed a territorial rather than a quantitative assessment of cone death. No difference was observed between wt and rd1 retinas up to P21. A clear loss of X-gal staining was observed at P28, with a central-to-peripheral gradient (Fig. 2) . At P77, cones were mainly present in the periphery, with a few patches in more central regions (data not shown for time points between the 4th and 11th weeks).
|
ISH Analysis
RNA samples from rd1 and wt were applied to cDNA microarrays comprising cDNA collections enriched for expression in the retina and the nervous system. Genes showing changes in levels of expression were selected for analysis by ISH (see the Material and Methods section). The 169 genes chosen from the microarray analysis were almost evenly distributed between genes that were higher in wt versus mutant and those that were higher in mutant versus wt. To minimize the variables of an ISH analysis, and make the results as quantitative as possible, retinas from rd1 and C57Bl/6N were mounted into the same paraffin block. Each slice had two different sections. A ribbon of 8 to 10 slices was transferred onto one slide generating 16 to 20 sections per slide, which were then hybridized with a single probe. A set of two pictures was used for the analysis, and was taken from the same slide with the same exposure for a given probe. If no specific central to peripheral pattern was observed, the picture was taken from a central region. To compare the large amount of ISH data systematically, a method was devised that is similar to that used previously for the developing retina.32 Expression values of 0 to 5 were assigned by visual inspection, as was the location of the ISH signal. The values were assigned for each of the seven retinal cell types; GC, amacrine cell (AC), horizontal cell (HC), BP, MG, rod, and cone) and for the inner segments (IS) of photoreceptors by direct inspection of each panel. A value of 0 was assigned if the cell type was absent. That was the case only for rods in the mutant tissue. The analysis of the two panels for a given probe was performed in parallel in one session. The translation of the ISH patterns into expression values at cellular resolution was based on the cell types that are normally located in the region of the retina where the ISH signal was seen. This was sometimes straightforward, but also presented the following problems. First, RNAs can be located in subcellular regions that are distant from the cell body. This is often seen for rod and cone genes, which show ISH signal in the IS. Nonetheless, this was mostly straightforward to interpret, since the only other cell type with processes in that region are MG cells. Their processes span the entire retina, engulf almost every retinal cell type, and form the outer limiting membrane and inner limiting membrane, confusing assignments not only of rods and cones, but also of GCs and INL neurons. Similarly, subcellular localization of RNA into neuronal processes of INL cells can produce a diffuse pattern in the INL and inner plexiform layer, which complicates the interpretation. In addition, there are several ACs that are displaced into the GCL, so that signal in the GCL could be due to these ACs. Thus, the assignments to a cell type are tentative, but were made to simplify the discussion of the data and provide a starting point for an understanding of cellular responses.
Cellular Gene Expression Changes during RD
Cellular expression levels observed by ISH were compared between wt and mutant by subtracting the mutant expression values for a given cell type and probe from the corresponding wt values (e.g., wt GC – rd1 GC). The advantage of subtracting the values rather than calculating ratios is that large data sets can be manipulated without the problem caused by division by 0. If the difference was >0, the gene was expressed at a higher level in the particular cell type of a wt retina, and if it was <0, it was expressed at a higher level in the cell type of the mutant retina. The analysis of genes expressed in GCL showed that, at P35, during major cone death, two times more genes had values <0 than had values >0 (Fig. 3A) .
|
Although many known PR genes that showed changes on the microarray analysis were not screened by ISH, some genes with microarray changes that were similar to the known PR genes turned out to be enriched in the PR. Genes expressed in rods were always at higher levels in wt retinas (Fig. 3A) as no rods were present in the mutant retinas at P35. Because the expression value assigned for an absent cell type was 0, the difference between wt and mutant values were always >0. In contrast to rods, expression values in cones reflect gene expression changes, as there are still cones present in rd1 retinas at P35. The analysis of cone gene expression levels by ISH showed that 48:1 genes had expression values >0, suggesting that gene expression in cones during this period was globally reduced, perhaps reflecting the fact that many cones are dying (Fig. 3A) . Thus, the observations of the ISH patterns agree with the disease pathology of cone death. (A summary of all the genes that changed and the cells involved is presented in Supplementary Table S3).
As the expression of a given gene was often complex, occurring in more than one cell type, a comparison was made of the levels of expression in the different cell types in wt and rd1. For example, a certain gene could demonstrate higher expression levels in wt GC and MG cells, while at the same time the gene may have higher expression levels in other cell types (e.g., ACs and HCs) of the mutant retina. This analysis showed that of the 125 genes that showed changes by ISH, 53 were expressed exclusively at higher levels in rd1, and 33 were expressed exclusively at a higher level in wt (Fig. 3B) . Twenty-seven genes had exclusively PR-specific changes. Most of these genes are PR-specific, thus their change is due to the lack of rods in the mutant retina at P35. Twelve genes were expressed at higher levels in one cell type and at lower levels in another cell type. Overall, these data show that 90% (113/125) of all the genes that changed have unidirectional changes. Excluding those that were PR-specific, of the genes expressed in the INL and GCL, 87% showed unidirectional changes; they were either solely up- or downregulated in rd1 or wt. Of these, 54% showed an upregulation in the mutant retina, 34% showed a downregulation, and 12% changed both ways. The results may suggest that the disease process similarly regulates the expression of a particular gene, even in different cellular contexts.
The changes in gene expression levels during cone death can be summarized as follows: During cone death, there is a higher level of expression in the GCL and the INL of the rd1 retina, mainly due to an increase in expression levels in GCs, BP cells, and/or MG (Fig. 4) . Changes in the ONL are due to a decrease in the number of rods and cones at the time point of the analysis. The results suggest that the genes assayed by ISH, whose levels are sensitive to RD and that are expressed in the GCL, are more likely to be expressed at higher levels during cone death. Genes that change in RD and are expressed in ACs and HCs during the peak of cone death change almost evenly in both directions. In contrast, among the genes assayed by ISH, BP (and possibly MG) significantly upregulate expression during the peak of cone death. Overall, GCL and INL cells are more likely to upregulate than downregulate gene expression during RD (54% vs. 34%).
|
|
A summary of the ISH analysis is as follows: Gene expression changes observed by ISH during cone cell death in the rd1 mouse model were more likely to be due to upregulation in the mutant tissue. Regardless of whether a gene was up- or downregulated, 90% of the genes that changed did so in only one direction, even though the gene was typically expressed in more than one cell type. The cell types most affected were cells of the GCL, BP and/or MG. There were no specific wt expression patterns for the genes that changed, with the exception of the PR-specific genes.
|
Discussion |
|---|
|
TopAbstractMaterial and MethodsResultsDiscussionReferences |
|---|
ISH as a Comparative Tool
To compare the responses of different cell types to RD in a comprehensive manner, a method that was applicable to many different genes was required, e.g., antibodies are not available to the large number of genes that are expressed in a tissue. ISH on tissue sections was thus chosen for this purpose. However, ISH is not typically a quantitative method. Different probes for the same gene can result in different signal intensities, perhaps due to different mRNA isoforms and/or secondary structures. In addition, some genes that appear to be low abundance by one method can result in strong ISH signal, and vice versa. Thus, expression levels of different genes are difficult to compare by this method. The ISH analysis presented herein neither compared expression levels of different genes to each other, nor compared different probes for the same gene. Rather, the results of hybridizing the same probe on different tissue sections on the same slide were compared. This allowed the ISH signals to be compared since the technical reasons for variability were controlled. In addition, several sections from each type of eye were present on the same slide. To ensure the best possible result, the values for the wt and mutant sections were assigned in the same session. For the reasons just mentioned, only values relative to each other within one slide were compared during the entire analysis.
The study was not performed in a congenic mouse background, and thus strain differences may complicate the interpretation of the data. A new congenic mouse strain generated by Jackson Laboratories was made available to us at the end of the study, allowing us to determine whether the results observed in the noncongenic strains would reproduce in a congenic FVB mouse background without the rd1 allele. We mounted P35 wt (C57Bl/6N), FVB congenic, and rd1 retinas into the same paraffin block, sectioned the retina, and performed ISH with 18 genes. Fourteen of these genes showed the same changes in wt (C57Bl/6N) and the FVB congenic when compared to rd1. Thus, the majority of changes observed by ISH at P35 between wt (C57Bl/6N) and rd1 are likely to reflect the progression of the disease, as opposed to strain differences. Nonetheless, caution is advised in interpreting changes of any individual gene, as roughly 20% of the genes tested in a congenic background failed to show mutation-specific changes. Additional caution is advised when comparing gene expression in noncongenic mouse strains during development. We performed this same study at earlier time points (P10, P12), using both microarrays and ISH to compare expression between C57Bl/6N and FVB. Although many genes changed, we found that a later comparison of these genes in the congenic strain showed that most of the changes were due to strain differences. We presume that slightly different kinetics of development in different strains can lead to such differences.
Cellular Responses to Cone Death in RD
Understanding changes on a cellular level during the cone death phase of PR degeneration can promote our understanding of the disease state in at least two ways. The localization of changes to individual cell types points to which cells are responding during cone death. Identifying the localization may give insight into the stimuli for such changes (i.e., the signals released by compromised cones, and/or protective mechanisms used by cells that survive the large amount of neuronal death in RD). The identification of cellular responses and gene functions may help us understand what metabolic or pathologic processes are going on in different cell types during the cone death phase. This study shows that GCs, BP, and possibly MG react the most during cone death. In many cases, the expression in BP and/or MG showed a greater induction than in the other INL cell types. In age-matched wt retina, the expression pattern of such genes was often a GC, AC, and HC pattern, suggesting that the induction in MG and BP cells may be ectopic, or at least is a very strong induction.
It is fair to question to what extent these findings are representative of the disease process, versus a reflection of our selection process for genes to screen by ISH. The microarray analysis performed was used as a screen to select genes that potentially change, as picking them randomly from a retinal library would probably have yielded fewer changes. If we compare the distribution of the genes in the different ISH categories (wt up; rd up; up-down; PR-only and no change; see Fig. 3B ) to the x-fold change of the microarray study, no correlation could be found except for the PR-only genes. This notion may seem counterintuitive, but may be predicted based on the changes that occur during the process of RD. Most PR-enriched or -specific genes would be at lower levels in the rd1 RNA samples. As these genes are not informative regarding the cellular responses of non-PR cells to the disease process, most of the known PR genes were not screened by ISH. In addition, the RNA preparation of rd1 would be significantly enriched for genes that are expressed selectively in the GCL and/or INL, because 
70% of the cells from the ONL are absent from the rd1 RNA sample. It is thus difficult to recognize the microarray results that point to the genes that are specifically induced in the GCL and or INL. In addition, it is possible that the different time points used for the microarray analysis and the ISH analysis led to some different outcomes. However, we feel that this is unlikely as we scanned over the entire central–peripheral extent of each section, which would reveal temporal differences over a range of ages. Finally, because genes can change in different cell types in different directions, and to different extents in different cell types, changes seen by ISH may not be seen, or may be seen to the same degree, in the microarray data. Nonetheless, by using microarray-detected x-fold changes to guide the selection of genes for ISH analysis, we were able to choose a set of genes in which the majority showed changes by both methods (125/169). The most rigorous way to determine whether the 169 genes chosen for ISH screening in the current study was an unbiased selection of genes would be to compare these data to the results from a random selection of cDNA clones from retinal libraries.
The differences observed in the cellular responses to cone death raise two questions. First, why are the changes in gene expression mainly upregulation in GCs, BP cells and/or MG, whereas in ACs and HCs the changes seem to be more uniformly distributed between up- and downregulated genes? Second, are these observations an rd1-specific phenomenon, or are they common to other RD models? The latter question can be answered by analyzing different mouse models of RD, using probes for the same set of genes.
Types of Genes That Change in RD
A gene ontology analysis of the genes that change during RD has the potential to provide some insight into the nature of the responses in the different cell types. However, we characterized a relatively small number of genes that changed via ISH, and the fact that molecular functions and biological processes are assigned to only roughly 50% of genes make this a speculative exercise. Of the 53 genes that were upregulated in rd1, only 31 were annotated. Similarly, of the 33 genes downregulated in rd1, 19 were annotated, and 5 were annotated of the 12 that were up- and downregulated in different cell types. In total, 57 genes were annotated (Supplementary Table S4) out of the 98 genes that changed by ISH (excluding the PR enriched/specific genes). The class with the highest representation was signal transduction in cell–cell communication, with 19 genes (10 up- and 8 downregulated in rd1 and 1 up- and downregulated). This finding may suggest an increase in cellular crosstalk during cone death. The other classes with high representation were metabolism (8:7:0 respectively), transcriptional regulation (6:2:0, respectively), and transporter proteins (mostly channels; 4:2:2, respectively). The number of genes found in energy metabolism and transporter activities suggests a restructuring of the remaining tissue during cone death and a change in energy supply. Retinal remodeling is a well-studied phenomenon in the rd1 mutant.7 34 BP cells may respond during cone death, as by then they have lost both rod and cone presynaptic partners. MG may be affected, as they normally supply the retinal neurons with energy or growth factors.35 As the energy supply chain via the retinal pigmented epithelium to photoreceptors and the oxygen supply to MG may have been altered by the degeneration of rods and cones, they may need to change their expression profile to adapt to the new environment.
|
Acknowledgements |
|---|
|
Footnotes |
|---|
Submitted for publication December 7, 2005; revised July 20 and September 8, 2006; accepted December 1, 2006.
Disclosure: C. Punzo, None; C. Cepko, None
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked "advertisement" in accordance with 18 U.S.C. 
1734 solely to indicate this fact.
Corresponding author: Constance Cepko, Harvard Medical School, Department of Genetics, Howard Hughes Medical Institute, 77 Avenue Louis Pasteur, Boston, MA 02115; cepko{at}genetics.med.harvard.edu.
|
References |
|---|
|
TopAbstractMaterial and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  D. J. Margolis, G. Newkirk, T. Euler, and P. B. Detwiler Functional Stability of Retinal Ganglion Cells after Degeneration-Induced Changes in Synaptic Input J. Neurosci., June 18, 2008; 28(25): 6526 - 6536. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
