| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 280, Issue 11, 10100-10108, March 18, 2005
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
¶||¶¶**
From the
Toronto Western Research Institute, University Health Network Program, Departments of Laboratory Medicine and Pathobiology and **Ophthalmology and Visual Sciences, ¶Vision Science Research Program, University of Toronto, Toronto, Ontario M5T 2S8, Canada
Received for publication, November 9, 2004 , and in revised form, January 7, 2005.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
When the orJ allele is bred onto a Mus musculus castaneus background retinal development is partially rescued (8). Deletion of the cell cycle inhibitor p27Kip1 also partially rescues proliferation (9). The number of cells in newborn orJ retina is 19-fold less than wild type, but in mice lacking both Chx10 and p27Kip1 this difference is only 4-fold (9). Therefore, a mechanism linking CHX10 to cell cycle regulators controls the expansion of RPCs. Interestingly, bipolar neurons are absent in Chx10–/–, p27Kip1–/– retinas, suggesting that CHX10 plays a role in the development of these cells independent of its role in controlling RPC proliferation (9). In mice carrying a reporter gene driven by Chx10 regulatory regions, expression was observed in a subset of retinal Müller glia as well as RPCs and bipolar cells (4). Therefore CHX10 may affect the development of a subset of Müller cells, in addition to bipolar cells (4).
Unlike CHX10, which is expressed in RPCs as early as E9.5 and is present in all mature bipolar cells (2), VSX1 is not induced in the retina until postnatal day 5 and is restricted to the outer tier of the inner nuclear layer (10–14). Double labeling experiments showed that VSX1 is absent in PKC
-positive rod bipolar cells, but is present in recoverin-positive OFF cone bipolar cells (15). Analysis of knock-in mice in which the VSX1 gene was replaced by lacZ confirmed this finding and showed that VSX1 is also expressed in ON cone bipolar cells (16, 17). Inactivation of VSX1 in the mouse retina leads to incomplete differentiation of OFF-cone bipolar cells (16, 17). In humans, VSX1 mutations are linked to inherited corneal dystrophies, with notable defects in bipolar cell function (18, 19).
Given the important roles of CHX10 and VSX1 in retinal development, we sought to determine their transcriptional activity. We found that CHX10 inhibited activated transcription in multiple contexts. The minimal region required for CHX10 repressor function included the HD and CVC domains. However, CHX10 also enhanced the activation of a specific subset of activators in chick neuronal cultures. Interestingly, we found that VSX1 also repressed transcription. The VSX1 HD mutation R166W, linked to keratoconus, impaired VSX1-mediated repression.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-minimal essential medium supplemented with 10% fetal bovine serum. Retinal, brain, and liver cells isolated from E8–E9 chicks (staged according to Hamburger and Hamilton (20)) were dissociated by incubating in 0.25% trypsin at 37 °C for 30–40 min. The cells were plated onto 35-mm wells pre-coated with 50–100 µg/ml PORN in Dulbecco's H21 + 10% fetal bovine serum media at an optimal transfection density of 2–3 x 106 cells/well.
CAT and Luciferase Assays—One day before transfection by the calcium phosphate method, cell lines were trypsinized and plated on 60-mm dishes at 70% confluency. E8 chick cultures were transfected using FuGENE (Roche Diagnostics) after 1–3 days in vitro. 0.3 µg of cytomegalorivus-
-galactosidase was included in each transfection and -galactosidase activity used to normalize for transfection efficiency. CHX10, RB, and VSX1 expression vectors were used at 5 µg unless otherwise stated in the figure legends. All activator and reporter plasmids were used at 0.5 µg. CAT assays were performed 24 h post-transfection as described (21). The amount of lysate used and the incubation time for the CAT assay varied with transcriptional activator, such that sample activity was within the linear range of the assay (i.e. CAT assay reactions for a given activator were allowed to proceed such that the counts extracted were less than or equal to 20,000 cpm). 100% CAT activity is taken as that obtained in the presence of control effector plasmid.
Plasmids—SVLex was built by replacing the N-Myc insert in SVN-Myc2 with a 
700-bp HindIII/XbaI fragment, derived from CDMLex (22). SVLex contains the SV40 enhancer/promoter upstream of the gene coding the bacterial transcription factor, LexA (amino acids 1–202).
LexRB encodes full-length human RB and has been previously described (21). LexCHX10 was generated by inserting a 3-kb BspMI/blunt XbaI fragment from pBS-KS+-CHX10(
57) (a gift from R. McInnes) into SmaI/XbaI-digested SVLex. LexCHX10 encodes a protein in which LexA is fused to the complete human CHX10 protein (amino acids 1–361). The CHX10 upstream sequence adds 7 codons between the SVLex sequence and the CHX10 ATG. The plasmid used to express Chx10 alone was pECE-HACHX10, a gift from P. Hamel. Expression in this plasmid is also driven by the SV40 enhancer/promoter. To generate GAL4-CHX10, Gal-hRb (21) was digested with SmaI/RI to remove RB, then a blunt SphI/RI fragment (full-length human Chx10) from LexCHX10 was cloned in. The Chx-ABCD series of deletion vectors were cloned into a cytomegalorivus-driven vector (pCDNA) that contained two C-terminal tags: SV40NLS (PKKKRKVE) and triple FLAG (DYK-DHDG-DYKDHD-IDYKDDDDK) (3.1NF). CHX-ABCD (amino acids 1–361) was generated from pBS-KS+-CHX10(57) and the Chx10 stop codon was removed by inserting a PCR fragment containing the 3' sequence downstream of the Cfr10I site. CHX-B (amino acids 147–208), CHX-BC (amino acids 147–264), and CHX-BCD (amino acids 147–361) were engineered by PCR to contain a 5' Kozak sequence, which generated a methionine residue upstream of the indicated CHX10 coding sequence. Full-length human VSX1 was subcloned from pBSKS(–)hVSX1 (kind gift of R. McInnes) into 3.1NF at a BamHI/XhoI junction to generate hV3.1NF. The VSXI stop codon was removed by insertion of a synthetic double-stranded oligo containing the VSX1 sequence 3' of the BstXI site. The R166W mutation was introduced into Vsx1 using the Promega site-directed mutagenesis kit.
All activator expression plasmids have been described previously. GAL4-activator fusions express the GAL4 DNA-binding domain fused to the following activation domains: VP16 (amino acids 410–490) (23), HSF1 (amino acids 201 to 529) (22), c-JUN (amino acids 5 to 253) (22), c-MYC (amino acids 1 to 262 of c-Myc) (24), and SP1 (amino acids 83 to 621) (25). A control vector, expressing amino acids 1–147 of GAL4 is pBXG1 (M. Ptashne).
The reporter plasmids, X4G2CAT, X6G2CAT, X300G2CAT, and X1700G2CAT have all been described previously (22). Briefly, X4G2CAT contains two GAL4 binding sites placed upstream of the human hsp70 TATA region (–40 to +160) and the CAT reporter gene. Four LexA binding sites were placed distal to activator binding sites, 20 bp upstream. X6G2CAT is identical to X4G2CAT, except that there are six LexA binding sites distal to the activator binding sites, instead of four. X300G2CAT is based on X6G2CAT and contains a fragment of 300 bp inserted between the LexA binding sites and the GAL4 binding sites. X1700G2CAT is also based on X6G2CAT and contains an 1,700-bp fragment inserted between the LexA binding sites and the GAL4 binding sites. X4G2PCNA CAT was built by inserting a BamHI/ blunt NcoI 0.7-kb fragment from UAS-PCNA-CAT (provided by M. Sopta) into SalI/blunt NcoI-digested X4G2CAT, which replaces the hsp70 promoter in the latter plasmid with the proliferating cell nuclear antigen (PCNA) promoter. pLD85 and pG5SV40CAT have also been described before (21, 26). HD4-pG5EC was built by J. Epstein, and contains four P3 sites (TAATtaaATTA) each separated by 8 bp, and positioned upstream of five GAL4 sites present in the parent vector pG5EC (M. Ptashne).
Gel Shifts—CHX10 plasmids were in vitro translated using rabbit reticulocyte lysates (Promega), in the presence of either 35S-labeled or unlabeled methionine. 2 µl of each in vitro translated sample was analyzed by 10 or 12% SDS-PAGE for quantification of protein levels by phosphorimager analysis (Bio-Rad Molecular Imager FX) of [35S]methionine signal (adjusted for methionine content of each protein) or chemiluminescent densitometry of anti-FLAG Western-blotted proteins using a Bio-Rad Fluor MultiImager with Quantity One version 4.1.0 software. These two methods of normalizing the amount of in vitro translated protein generated similar results in the gel shift assays used to compare different CHX10 proteins, which were carried out a total of six times. 200 ng of double-stranded probe was end-labeled by T4 DNA kinase end labeling with [
-32P] and polynucleotide kinase. Probe sequences were: HD4, top strand: 5'-CTCTAGCTAATTAAATTAGC-3'; HD4 mutant, top strand: 5'-CTCTAGCTCCTTAAAGGAGC-3'. Probes were purified using G25 spin columns (Amersham Biosciences catalog number 27-5325-01) and 40,000 cpm used per reaction (0.2–2 ng). In vitro translated protein was incubated with labeled probe for 15 min at 26–28 °C in a final volume of 15 µl (2 mM Tris, pH 8.0, 0.5 mM EDTA, 10% glycerol, 0.05% Triton, 88 mM KCl, 150 ng/µl poly(dI-dC), 1 mM dithiothreitol, 12.5 ng/µl salmon sperm DNA). For competition reactions, excess unlabeled probe was first incubated with in vitro translated protein for 2–3 min at 26–28 °C, then labeled probe was added and the tubes incubated for 15 min at 26–28 °C. Samples were resolved on a 4.5% non-denaturing acrylamide gel using 0.5x TBE running buffer. Gels were dried and exposed overnight on phosphorimager screens.
Western Blotting—In vitro translated proteins or cell lysates were separated on 12 or 15% SDS-PAGE, transferred to nitrocellulose, and blocked in 1:10 BM Blocking reagent overnight at 4 °C. Blots were washed briefly and incubated with 3 µg/ml mouse anti-FLAG antibody in phosphate-buffered saline (M2-Sigma). Following washing with phosphate-buffered saline, 0.1% Tween, blots were hybridized with rabbit anti-mouse horseradish peroxidase secondary (Jackson Laboratories) at 1:5000 in 1:20 Roche blocking reagent in phosphate-buffered saline. Blots were washed and chemilumiscence was detected with Roche Western blotting kit and captured and quantified using a Bio-Rad Fluor MultiImager with Quantity One version 4.1.0 software.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
When a human cervical carcinoma line (C33A (27)) and a neuronal/glial line (NG108, generated by fusion of mouse neuroblastoma and rat glioma cells (28)), were transfected with X4G2CAT and LexCHX10 plasmids, no transcriptional activation was observed (data not shown). A plasmid expressing LexA alone (SVLex) had no effect (data not shown). To explore whether CHX10 might inhibit transcription, we tested the effect of Chx10 on several activators. GAL4 fusions used in these assays contained the ADs of HSF1, which are acidic (22); and c-JUN and c-MYC ADs, which are prolineand glutamine-rich (24, 29). c-MYC has an additional AD that has no preponderance of specific amino acids (24).
As a positive control for repression, cells were transfected with LexSu(z)2, which encoded LexA fused to the Drosophila polycomb group protein suppressor two of zeste (Fig. 1A), together with X4G2CAT and GAL4-c-MYC (GM1–262). Su(z)2 is a powerful transcriptional repressor responsible for maintaining the proper spatially restricted expression pattern of the homeotic loci during Drosophila development (22). LexCHX10 repressed activated transcription as effectively as LexSu(z)2 in C33A cells (Fig. 1A). LexA alone (SVLex) did not affect transcription. Repression was dependent on CHX10 being targeted to the promoter by LexA because CHX10 alone failed to repress (Fig. 1A). In NG108 cells, LexCHX10 inhibited GAL4-c-JUN and GAL4-HSF1 mediated activation by 80 and 87%, respectively (Fig. 1B). Similar results were obtained in C33A cells (data not shown). In NG108 cells, titration of LexCHX10 demonstrated dose-dependent repression of HSF1-mediated activation (data not shown).
|
CHX10 Also Represses TATA-less Promoters—Because TATA-containing and TATA-less promoters sometimes show differences in response to activators and repressors (31–34), we also tested the properties of CHX10 on a TATA-less promoter, X4G2PCNACAT. Here, the TATA-containing hsp70 promoter in X4G2CAT was replaced by the TATA-less core promoter from the gene for PCNA (35). LexChx10 inhibited GAL4-c-JUN-mediated activation of this reporter by 85% (Fig. 1D), indicating that Chx10 does not require a TATA box to inhibit gene expression.
CHX10 Represses a Complex Enhancer—The above experiments indicate that CHX10 represses individual activators (c-MYC, HSF1, and c-JUN). In vivo, however, these factors operate in the context of multiprotein structures, or enhanceosomes, which form by cooperative binding of activators to enhancers (36). Thus, we assayed the effect of CHX10 on the SV40 enhancer using the reporters pLD85 (21), in which a GAL4 binding site lies upstream of the thymidine kinase promoter, CAT, and the SV40 enhancer, or G5SV40CAT (26), in which five GAL4 sites lie upstream of the SV40 enhancer/promoter (Fig. 1E). GAL4-CHX10, which encodes the GAL4 DNA-binding domain fused to CHX10, repressed both reporters by 85–90%, whereas the GAL4 DNA-binding domain alone, encoded by pBXG1, had no effect (Fig. 1E). Thus, in addition to discrete activators, CHX10 can repress a complex enhancer. The retinoblastoma protein (RB) is another potent transcriptional repressor (21, 37–42). As observed previously (21), RB repressed the pLD85 reporter (Fig. 1E). Surprisingly, however, RB did not inhibit G5SV40CAT, in which the SV40 enhancer is positioned upstream of the CAT gene. These results suggest that RB and CHX10 use different mechanisms to inhibit transcription.
Repression When CHX10 Is Tethered to Promoters by the HD—HDs of the paired or paired-like classes, bind DNA most efficiently as homoor heterodimers to a palindromic TAAT-(N)nATTA motif (43). Spacing between the TAAT half-sites is predicted by the nature of the residue at position 50 of the HD. In CHX10 this residue is a glutamine (Gln50) (2) suggesting that CHX10 may bind a palindromic P3 consensus site in which TAAT half-sites are separated by three nucleotides (43). In vitro translated CHX10 protein retarded the mobility of a 32P-labeled P3 oligonucleotide in a gel shift assay (Fig. 2A, lane 2). CHX10 was competed by excess unlabeled wild-type P3 probe, but not mutated P3 probe (Fig. 2A, lanes 3 and 4). Unprogrammed lysate did not bind the P3 probe (Fig. 2A, lane 5). Similar results were obtained using the bacterially expressed GSTChx10 fusion protein (data not shown).
|
Selective Repression in Primary Chick Cultures—Results described above suggest that CHX10 is a global inhibitor, because it repressed c-MYC, c-JUN, HSF1, and the SV40 enhancer. The cell lines used are immortalized and transformed, so our next goal was to determine whether CHX10 represses transcription in primary cultures, including retina. Dissociated embryonic day 8 (E8) chick retinal cells were co-transfected with different GAL4 DNA-binding domain activators as well as LexCHX10 or LexRB and X4G2CAT. LexRB repressed all the activators tested, whereas LexCHX10 repressed c-JUN and SP1 but slightly augmented the activity of VP16, HSF1, or c-MYC (1.2–1.9-fold, Fig. 3A). We also co-transfected chick retinal cells with LexCHX10, either GAL4-c-JUN or GAL4-HSF1, and HD4-pG5EC, the reporter containing P3 sites (see Fig. 2B). Again, CHX10 repressed c-JUN (6-fold) but weakly augmented HSF1-mediated activation (1.6-fold, Fig. 3B).
|
The CHX10 Repression Domain Maps to the HD-CVC Region—To examine the minimal region necessary for repression, CHX10 was divided into four sections, A–D, corresponding to the N terminus, HD, CVC motif, and C terminus, respectively (Fig. 4A). Vectors were constructed that expressed various combinations of these regions (CHX-ABC, CHX-BCD, CHX-BC, and CHX-B) fused to a nuclear localization signal and triple FLAG tag. Nuclear localization was confirmed by anti-FLAG immunostaining following transfection of three different cell lines (data not shown). An additional vector was generated in which codon 51 was altered from asparagine to alanine (CHX-N51A). Mutation at this invariant amino acid disrupts DNA binding (46). In gel shift assays, in vitro translated CHX-ABC, CHX-BCD, and CHX-BC interacted with the P3 probe as well as wild type CHX10 (Fig. 4B, cf. lanes 2–4 with lanes 8–16), but CHX-B exhibited reduced binding and, as expected, CHX-N51A did not bind at all (Fig. 4B, lanes 5–7 and 17–19). A dose-response titration confirmed DNA binding levels of the CHX10 proteins (data not shown). The amount of in vitro translated protein used in these gel shift assays was normalized by two methods (see "Experimental Procedures" for details) such that any variance between wild type CHX10 and any mutated protein was less than 2-fold. Anti-FLAG Western analysis of normalized lysates confirmed that adjusted in vitro translated protein levels were within this range (Fig. 4C and data not shown, see "Experimental Procedures" for quantification method).
|
10-fold less plasmid was required to generate the same amount of protein observed with other vectors (Fig. 4, D and E). This result suggests that the N terminus may destabilize CHX10, although a simple correlation was not evident because CHX-BC, which also lacks the N terminus, was required in amounts similar to the full-length vectors (Fig. 4, D and E). Further studies are required to determine whether the differences in expression are because of an effect on translation and/or stability. CHX-ABC and CHX-BC repressed transcription almost as well as wild type CHX10 (ABCD) (Fig. 4D), indicating that neither the N nor C termini were required for repression. CHX-BCD also repressed transcription, albeit to a lesser extent than CHX-ABC or CHX-BC, suggesting that, in the absence of the N terminus, the C terminus slightly inhibits repression (Fig. 4D). As expected, the DNA-binding defective protein CHX-N51A did not repress transcription (Fig. 4D). CHX-B, which retains significant DNA binding activity also did not repress transcription (Fig. 4D). These data indicate that the CHX10 HD and CVC domains are sufficient for DNA binding and repression and that removal of the CVC domain slightly impairs DNA binding but ablates repression. VSX1 Can Repress Transcription—It has been suggested that the CHX10-related protein VSX1 may be an activator (12), but our results with CHX10 raised the possibility that VSX1 may repress transcription. Previously we showed that Vsx1 interacts with the P3 motif (18) and that a natural mutation in the VSX1 HD, R166W, which causes keratoconus in humans, impairs binding to this motif (18). To test VSX1 activity, and whether this mutation impairs VSX1 function, NG108 cells were co-transfected with wild type or VSX1R166W expression vectors, together with the HD4-pG5EC reporter plasmid and GAL4-HSF1. Vsx1 repressed HSF-1-mediated activation more effectively than CHX10 (Fig. 5A), although this may be in part because of slightly lower protein expression level of CHX10 compared with Vsx1 as detected by anti-FLAG Western blotting (data not shown). The R166W mutation significantly impaired repression (Fig. 5A). A dose-response curve confirmed that wild type VSX1 repressed transcription more effectively than VSX1R166W over a wide range of concentrations (Fig. 5B). Protein expression was assessed by Western blot analysis (Fig. 5C) and quantification (see "Experimental Procedures") indicated that the mutated protein was expressed at the same level or slightly higher (up to 1.4-fold) than wild type VSX1.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
How Do CHX10 and VSX1 Regulate Transcription?—Only a handful of HD proteins have a defined repression domain. Engrailed contains two distinct repression domains, one type is most effective in transient assays, whereas the other potently represses in vivo with little effect on episomal templates (51). The repression domain for Even-skipped was mapped to a Ala-Pro-rich region, which included the HD (30, 53). Outside of its function as a DNA-binding motif, the HD has been show to mediate protein-protein interactions for transcriptional repression (53–57). For instance, Msx-1 and Even-skipped interact with the general transcription factors TFIID/TBP, whereas Msx2 interacts with TFIIF (RAP47/RAP30). Msx-1 and -2 require residues from the N-terminal arm for this interaction, whereas this region is not necessary for Even-skipped (53, 55, 58). Alternatively, the HD can mediate interactions with co-repressor molecules, as is the case for NK-2 and Groucho (56, 57).
Whereas a truncated version of CHX10 consisting of the HD and CVC regions bound a P3 site as efficiently as full-length CHX10, removal of the CVC impaired DNA binding. This suggested that the HD alone (Chx-B) may require the CVC domain for effective DNA binding. Alternatively, reduced DNA binding of Chx-B may be a result of changes in conformation or stability, indicating a potential need for additional flanking residues. As CHX10 can repress in several different scenarios, it seems reasonable that it could bind and/or interfere with a basal transcription factor. This action may be mediated through the HD and/or CVC motif of CHX10. Intriguingly, mutations in the CVC domain of ceh10 or human VSX1 cause lethality and eye defects, respectively (1, 18). Also, an isoform of CHX10 has been described that contains a 21-amino acid insert in the CVC domain (7, 44). It will be interesting to determine the effect of these variants on DNA binding or repression.
A number of HD proteins contain eh1-like motifs capable of interaction with members of the Groucho/TLE co-repressor family of proteins (59–62). The N termini of CHX10, VSX1, and the related Prd-like HD protein Rx have an eh1-like octapeptide sequence (63). We found that in vitro translated CHX10 did not interact with a subset of mammalian Groucho homologues (data not shown). However, this does not rule out the possibility that Chx10 may interact with other members of the large Groucho/TLE family or utilize the eh1 motif for repression in vivo (51). An eh1-like sequence in D-Gsc, FTIDSILG, promotes heterodimerization and "interactive repression" (64). Thus, the role of the CHX10 or VSX1 eh1 or other functional domains, as well as the co-repressors that mediate repression by this or other motifs remain to be clarified.
Certain Prd-like HD proteins, including Chx10, interact with the RB protein family (65). RB, and its relatives, are potent transcriptional repressors (21, 37–41, 66), and could potentially mediate repression by CHX10. In the context of the SV40 enhancer, CHX10 appears to use a Rb-independent mechanism of repression (Fig. 1E). In addition, repression by CHX10 seems to be unaffected by overexpression of E1A, which binds and antagonizes the function of the RB protein family.3
DNA Binding by Paired-like HD Proteins—DNA binding specificity for different HDs is determined in large part by the two nucleotides downstream of a TAAT core, but the immediate 3'-flanking nucleotide is most important. Several HDs in which the 50th residue of the HD is a Gln interact with a TAATT motif, whereas Lys50 HDs bind TAATC (43, 67–71). However, other amino acids in the N-terminal arm of the HD, the DNA binding helix, and downstream amino acids can alter the specificity of Gln50 HDs (72, 73). A screen for consensus CHX10 binding sites was performed and revealed the sequence TAAT-TAGC (5). A thymidine directly 3' to the TAAT core is in agreement with the P3 sequences we tested for CHX10 (Fig. 2A, data not shown) and VSX1 (18). Interestingly, the Lys50 HD proteins RX and CRX interact with TAATC sequences and activate transcription (74, 75). The repertoire of TAATT and TAATC sites present on different promoters is likely to have a profound effect on the timing and extent of target gene induction in response to the same pool of Gln50 and Lys50 HD proteins.
Biological Role of CHX10 and VSX—Chx10 loss impairs the proliferation of RPCs early in retinal development and blocks the differentiation of bipolar cells in the postnatal rodent retina (6, 9). Our data are consistent with the possibility that CHX10 may facilitate proliferation and bipolar cell formation by repressing the expression of genes that promote growth arrest and the formation of cell types other than bipolar neurons, respectively. Indeed, a variety of cell cycle inhibitors are upregulated in the CHX10 null retina (4) and, in support of the idea that CHX10 may repress non-bipolar cell genes, both CHX10 and VSX1 were isolated in a one-hybrid assay that utilized three tandem copies of the cone opsin locus control region (LCR) (12), a gene that is silent in bipolar cells. CHX10 affects cell fate (6, 9, 76), whereas VSX1 sculpts the terminal differentiation characteristics of cone bipolar cells (16, 17), thus in addition to terminal differentiation markers like cone opsin, CHX10 may also repress fate determinants. Bipolar cell births parallel those of rod photoreceptors and Müller glia, so CHX10 may repress genes that promote the formation of these cell types (77–80). The observation that ectopic CHX10 expression promotes inner nuclear cell fates at the expense of photoreceptors (76, 81) supports the idea that CHX10 may block the expression of rod rather than Müller determinants. Nevertheless, a Chx10 reporter gene is expressed in Müller glia in the orJ retina (4), consistent with the idea that CHX10-deficient cells become Müller rather than rod cells. However, to rule out the possibility that the latter phenomenon is not a secondary effect of the dramatic effects on proliferation, it will be important to determine whether inactivating Chx10 late in retinal development switches cells from the bipolar to the rod or Müller cell fate.
In addition to facilitating proliferation and bipolar cell differentiation through repression, CHX10 may also potentiate activators that drive these processes. Indeed, as well as increased levels of cell cycle inhibitors, the orJ retina also contains reduced levels of cell cycle activators, such as cyclin D1 (4, 9) and CHX10 cooperates with the basic helix loop helix activators Mash1 and Math3 to drive bipolar cell genesis (76). Furthermore, the CHX10 HD augments transcriptional activation mediated by PAX6 (52). We found that CHX10 weakly potentiated a subset of activators in certain contexts. For example, CHX10 consistently repressed c-JUN activity in all these cell types but while Chx10 repressed HSF1 activity in mammalian cell lines and in chick primary liver cells, it enhanced HSF1 activation mildly in primary chick retinal or brain neuronal cultures. Thus, depending on the availability of cell-specific cofactors and the activators bound to the same target locus, CHX10 can act as a mild activator. Elucidating the full complement of direct CHX10 target genes in vivo will be necessary to clarify targets it represses or potentiates to control retinal cell proliferation and differentiation.
|
FOOTNOTES |
|---|
|| Recipient of a Vision Science Research Program Doctoral Research Award (Toronto, Canada) and an E. A. Baker Foundation and The Canadian National Institute for the Blind/Canadian Institute for Health Research Partnership Doctoral Research Fellowship.
To whom correspondence should be addressed: Toronto Western Research Institute, Mc6–424, Cellular and Molecular Division, 399 Bathurst St., Toronto, Ontario M5T 2S8, Canada. Tel.: 416-603-5865; Fax: 416-603-5126; E-mail: rbremner{at}uhnres.utoronto.ca.
1 The abbreviations used are: HD, homeodomain; RPC, retinal progenitor cell; CAT, chloramphenicol acetyltransferase; PCNA, proliferating cell nuclear antigen; E, enbryonic day.
2 R. Bremner, unpublished data.
3 K. F. Ahmad, K. M. Dorval, and R. Bremner, unpublished data.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
