Expression of novel ING variants is regulated by thyroid hormone in the Xenopus laevis tadpole.

The candidate tumor suppressor gene, ING1, encodes several protein isoforms as a result of alternative splicing that may possess agonistic and antagonistic roles in the control of cell proliferation and apoptosis. Recently a related gene, ING2, was isolated in human whose expression is increased in adenocarcinomas. Little is known about the cellular function and regulation of these ING family members, but the fact that ING proteins contain a plant homeodomain finger suggests that these proteins may modulate transcription factor-mediated pathways. To elucidate how ING may interact in different tissues to modulate function, we used amphibian metamorphosis as a model system in which a single stimulus, thyroid hormone (TH), initiates tissue-specific proliferation, differentiation, and apoptosis. We have isolated the first Xenopus laevis ING2 and demonstrate that transcript levels increase in response to TH treatment. We provide evidence for the existence of splice variants that are differentially expressed in tissues with different TH-induced fates. Western blots using an antibody directed against the highly conserved C-terminal end of ING proteins reveal a tissue-specific pattern of ING isoform expression in adult Xenopus tissues. Analyses of premetamorphic tadpole tissues show a TH-induced accumulation of ING proteins in tail, whereas the levels in the leg are not affected. This TH-induced accumulation is also observed in serum-free tail organ cultures and is prevented by inhibitors of tail apoptosis. Therefore, this work presents the first link between ING expression and a hormonally regulated nuclear transcription factor-mediated apoptotic response opening the possibility that ING family members may be involved in transducing the signal initiated by TH that determines cell fate.


Introduction
The ING1 (inhibitor of growth 1) gene was first isolated by PCR-mediated subtractive hybridization for the enrichment of transcripts found in non-tumorigenic breast epithelial cells followed by a novel in vivo positive selection procedure for growth inhibitors (1). ING1 is implicated in the control of several key cellular processes (for review, see (2)) including cellular proliferation (1,3,4), apoptosis (5-7), senescence (3), and drug resistance (8). ING1 transcript levels are depressed (1,(9)(10)(11)(12)(13) and the ING1 gene is a target for loss of heterozygosity or rearrangement (1,3,(13)(14)(15)(16) in a variety of cancer cells suggesting that ING1 functions as a tumor suppressor. At least four ING1 transcripts are ubiquitously expressed in adult and fetal tissues with varying levels; products of alternative splicing of a variable first exon and a common second exon (4,9,17,18).
Several known protein products are encoded by these transcripts; however, no systematic analysis of protein expression in different normal tissues has yet been reported. (9). Both ING2 and ING4 transcripts are ubiquitously found in fetal and adult human tissues. The gene structure of ING2 appears to be similar to ING1 with 2 exons (Nagashima et al, unpublished), but no splice variants have been described. ING2, like p33 ING1b , can regulate the expression of genes involved in apoptosis such as p21 and bax (20). ING4 transcript levels are decreased in breast and melanocyte cancer cell lines (9) and ING2 transcripts are elevated in adenocarcinomas compared to adjacent normal tissue (19) suggesting that each ING family member is independently regulated and has its own unique effects.
ING proteins belong to a family of plant homeodomain (PHD) finger-containing proteins that include transcription factors and proteins that regulate chromatin structure (22). Although the mechanisms of action for ING proteins have yet to be fully elucidated, there is evidence that ING proteins may affect the activity of p53 (6,20,23), histone acetyl transferases (HATs) (24,25), and histone deacetylases (HDACs) (25,26). The combination of splice variants, multiple potential protein products and at least three related genes allows for considerable possibilities for ING to modulate cellular effects as both agonists and antagonists. Indeed, recent reports suggest that p33 ING1b and p24 ING1c are functional antagonists with respect to modulation of p53 (4) and HDAC activity (26) and that p33 ING1a and p47 ING1 have opposite effects on HAT activity (25).
In order to elucidate how ING modulates cellular outcome in different tissues, we turned to amphibian metamorphosis as a model system in which a single stimulus, thyroid hormone (TH), initiates tissue-specific proliferation, differentiation and apoptosis. The metamorphosis of the tadpole to a frog is absolutely dependent upon a substantial increase of endogenous levels of 3,5,3'triiodothyronine (T 3 ) from undetectable levels in the plasma (27,28). The other TH, thyroxine (T 4 ), also increases, but it is the predominant form transported to target tissues where it is converted to the more active T 3 form. Virtually every tissue in the tadpole is a target of TH and these changes by guest on March 22, 2020 http://www.jbc.org/ Downloaded from can be precociously induced by exogenous TH administration in vivo and in culture (29-35). TH functions to selectively activate tissue-specific genetic programs by regulating gene transcription via specific nuclear receptors (TRs) (29,35-42). TRs have important roles as repressors and activators of gene transcription during Xenopus development (for review see (43)). In premetamorphic tadpoles, TRs function as repressors of TH-inducible genes in the absence of appreciable levels of TH thereby preventing precocious metamorphosis. When endogenous TH levels rise, they act as activators of these genes thereby initiating metamorphosis. Thus, the presence or absence of ligand plays a critical role in gene regulation. However, what still remains enigmatic is how TRs can promote the development of multiple cell fates such as proliferation, reprogramming and apoptosis during metamorphosis. Several factors that modulate TR activity have been described and include HATs/HDACs and p53 (44)(45)(46)(47)(48)(49)(50)(51)(52)(53)(54)(55)(56) and it is postulated that tissuespecific factors may modulate the TH-induced outcome (for review see (43)).
Herein, we describe the isolation, cloning and initial characterization of the first frog ING2 gene (xING2 for Xenopus ING2) and provide evidence suggesting that xING2 is subject to alternative splicing. We demonstrate that transcript levels differentially increase in response to T 3 treatment in tissues with different metamorphic fates. Western blots using an antibody directed against the highly conserved C-terminal end of ING proteins reveal a complex pattern of expression in adult Xenopus tissues. While premetamorphic tadpole tissues show T 3 -induced accumulation of ING proteins in the tail, the levels in the leg are not affected. This T 3 -induced accumulation is also observed in serum-free tail organ cultures and is abrogated by inhibitors of tail apoptosis. Therefore, this work presents the first link between ING expression and a hormonallyregulated nuclear transcription factor-mediated response. Since ING proteins appear to associate with chromatin (7,25)  These membranes were placed on LB plates with 100 µg/mL ampicillin and grown overnight at 37°C. The bacteria were then lysed on the membrane by placement on filter paper with 0.5M NaOH for 5 min, 1M Tris-HCl pH 8.0 for 5 min, and 0.5M Tris pH 8.0/1.25M NaCl 5 min. The DNA was UV cross-linked to the nitrocellulose as described previously and then washed with 2X SSC/2% SDS followed by 2X SSC. The membranes were hybridized with human ING1 cDNA by guest on March 22, 2020 http://www.jbc.org/ Downloaded from probe as described above but for only 1 h prior to the detection with CDP-Star reagent. Positive clones were used to inoculate 5 mL cultures of LB broth with 100 µg/mL ampicillin which were grown overnight at 37°C with shaking. Plasmids were harvested with a Qiaprep Spin Miniprep kit (Qiagen) and subsequently digested with EcoRI. A 1% agarose gel was used to separate the inserts from plasmid vector and the products were Southern blotted overnight as described previously.
Positive clones were then sequenced. The DNA and derived amino acid (aa) sequences were aligned using Clustal W version 1.8 software (57).

Northern Blot Analyses
Analyses of RNA transcripts was done according to the method described in (58)

RT-PCR Analyses
To determine the relative expression levels of xING2 transcripts, RT-PCR analyses were performed. The primers spanning the putative exon 1/2 boundary (XB5/XB8) and within the putative exon 2 (XB9/XB10) are indicated in Figure 1 and yield amplicons of 635 and 253 bp, respectively. All reactions were determined to be in the linear range of amplification and normalized to L8 ribosomal protein transcript whose expression is not affected by T 3 treatment (59). For amplification of control L8 ribosomal protein transcript, the sense primer For TRα the sense primer (5'CACTACCGCTGTATCACTTG3') and antisense primer (5'GGGTGATTATCTTGGTGAACT3') were used (60). For TRβ the sense primer (5'CCAGTGCCAAGAATGTCG3') and antisense primer (5'GTAAACTGGCTGAAGGCT3') were used (60). All primers were used at 20 pmol in a typical 50 µl reaction containing 1.5 U Taq DNA polymerase (Amhersham Pharmacia), 10 nmol dNTPs (Life Technologies) and 1.5 mM MgCl 2 . The PCR reaction was: 7 min at 94°C, 35 cycles of 60 s at 94°C, 60 s at 55°C (for TRα, TRβ, and L8) or 54°C (for xING2 primers), and 1 min at 72°C. A final 10 min extension at 72°C was done. The L8 reaction was the same except only 30 cycles were used. The amplified products were separated on 2% agarose gels and visualized by ethidium bromide staining.
Western blot analyses were performed using a rabbit polyclonal antibody generated using a human GST-ING1 C-terminal end fusion protein (1) using methods described previously (58) with minor modifications. Briefly, blocking was performed overnight at 4°C in PBS containing 5% skim milk, 2% fetal bovine serum (Life Technologies) and 0.1% v/v Tween-20 in PBS pH 7.2. The polyclonal antibody was used at a 1:10,000 dilution in blocking buffer. The antibody incubation was carried out at room temperature for 1 h. Membranes were washed extensively in PBS with 0.1% v/v Tween-20 for 10 minutes and incubated with goat anti-rabbit IgG polyclonal antibody conjugated to horseradish peroxidase (Calbiochem, La Jolla, CA). Peroxidase activity was detected by using an ECL kit according to the manufacturer's instructions (Amhersham Pharmacia Biotech). Control animals had an equal volume of DMSO added to their water. The tadpoles were sacrificed at the indicated times after treatment for the isolation of RNA or protein as described above.
Organ culture of tadpole tails Tadpoles were anaesthetized in 0.1% MS222 (Syndel Laboratories, Vancouver, BC) and quickly immersed (5 s each) in a series of sterile water, 70% ethanol and two more beakers of sterile water.
The tails were severed under aseptic conditions with a sterile scalpel and placed into culture dishes containing culture medium. The culture medium consisted of a 55% dilution of 1X alpha MEM (ICN Pharmaceuticals, Costa Mesa, CA) pH 7.2 supplemented with 14.5 mM NaCl, 1.1 mM Na 2 HPO 4 , 1.1 mM NaH 2 PO 4 , 2 mM L-glutamine, 1 mM L-methionine, 25 mM HEPES, 10 µg/ml fungizone and 50 µg/ml gentamycin sulfate. The cultures were incubated at 25 o C under air and the medium was changed daily. The tails were allowed to recover overnight before the addition of 100 nM T 3 with or without 2 mM EGTA, pH 8.0 (Sigma-Aldrich) that has previously been shown to inhibit tail regression (62).   (Figure 2). The C-terminal PHD finger domain is completely conserved between ING1 and ING2 proteins and the 90 aa region spanning the PHD finger also displays a high degree of conservation.

Isolation of a novel
Previous work on human tissues indicated that ING2 is ubiquitously expressed as two major transcripts of 1.3 and 1.5 kb with the highest expression in the testis (19). Northern blot analyses of adult Xenopus tissues show a similar trend. xING2 is expressed in all tissues examined with testis having the highest expression levels followed by brain and skin in similar amounts then muscle and liver showing very low levels. A 1.3 kb transcript is found in all tissues with an additional 1.0 kb band found only in testis ( Figure 3A). Finding multiple bands in the Northern blot suggests that, at least in the testis, multiple splice variants or transcripts from genes highly related to ING2 are present. These results were obtained using the entire open reading frame of xING2.
Since ING1 is subject to alternative splicing and since the gene structure of ING2 is similar to ING1, we wanted to test whether different splice variants for xING2 exist as well. To test this, we used differential RT-PCR analyses using two primer sets. One primer set (XB5/XB8; Figure 1) specifically amplifies the transcript we have reported in Figure 1. The resultant amplicon is referred to as xING2(1/2) in Figure 3B. The other primer set (XB9/XB10) amplifies a region in the conserved 3' end of the ORF that should be common to all ING2 variants (assuming that splicing occurs in a manner similar to ING1) and is referred to as xING2(2) in Figure 3B. Neither primer set amplifies ING1 sequences (data not shown). If no splice variants are present, then one would expect that relative levels of amplicons generated using the two primer sets would be equal. The RT-PCR results are consistent for the existence of xING2 splice variants in brain, testis and skin ( Figures 3B and C). TRβ transcript levels showed increases as was previously reported (33). In the leg, the xING2 (2) induction pattern closely resembles that of T 3 -induced TRα expression in these tissues where a biphasic response is observed at 6 h and 24 h with maximal levels at 6 h, rather than that observed for TRβ (single peak at 24 h; Figure 5C). In the tail, the xING2(2) induction pattern is similar to both TRα and TRβ with an initial modest increase at 2 h reaching maximal levels at 48 h ( Figure   5B). In the brain, xING2(2) amplicon levels increase 24 h after T 3 treatment and TRα levels decrease slightly. TRβ levels show a marked increase in this tissue ( Figure 5D).  Figure 5E). xING2 expression gradually increases from low levels at NF stage 58 to maximal levels at NF stage 63, whereas TRα levels are maximal at NF stage 60 and decline by NF stage 63, and TRβ levels reach maximal levels at NF stage 60 and remain high ( Figure 5E). The leg does not show any induction of overall xING2 expression similar to the TRα expression pattern ( Figure 5F). TRβ levels peak at NF stage 58 and decrease thereafter. These observed patterns of TR expression concur with those previously reported (33). These data also show that the relative levels of both TRα and xING2 are much lower in the leg compared to the tail during spontaneous metamorphosis.

RT-PCR analyses using primers
RT-PCR analyses using primers spanning the exon 1/2 splice site show that this presumed splice variant of xING2 increases to maximal levels at NF stage 62 in the leg in contrast to the overall pattern of xING2 expression (compare xING2(1/2) to xING2(2); Figure 5F). At this stage, the relative amount is approximately twice that found at maximal levels in the tail at NF stage 62.
In the tail, this presumed splice xING2 variant exhibits a slight delay in increased expression levels (maximal at NF stage 62; xING2(1/2) in Figure 5E) compared to overall xING2 expression; a result that is reminiscent of the T 3 -induction experiments ( Figure 5A). Similar results are observed using cycloheximide/anisomycin and H7 inhibitors (data not shown).

Specific ING protein levels increase upon T 3 -induced apoptosis of the tail in vivo
Given that the 1.3 kb transcript detected in the tail ( Figure 5) is too short to produce the 90-130 or 60 kDa bands, we suspect that these bands are more likely to represent ING1 gene products or unidentified splice variants of xING2. Northern blot analyses using a heterologous ING1 probe shows that a 3.9 kb transcript is identified that is increased upon T 3 treatment in tail tissue providing support for this idea ( Figure 6D). Together, these data show that accumulation of ING proteins correlates with the ability of the tail to undergo TH-induced apoptosis and provide the first evidence that ING is subject to hormonal control.
In this study, we have investigated the expression patterns of ING proteins in a developmental model system in which a single stimulus (TH) can induce both outcomes. We have isolated the first Xenopus laevis ING2 homolog that has a high degree of identity with human and murine counterparts and we demonstrate for the first time that ING expression is hormone responsive.
Moreover, xING2 was found to be an early response gene along with TRα and TRβ (38,63) placing it in a potentially important role for the control of cell fate during TH-dependent metamorphosis.
The ING2 gene is conserved between frogs and humans and we provide the first evidence of differential regulation of presumed splice variants of this gene. Given the high degree of conservation of gene structure between ING1 and ING2 (4,9,17,18)(our work and Nagashima et al, unpublished), and given that several ING1 splice variants have already been identified (4,9,17,18), it is highly probable that alternative splicing contributes to the tissue-specific regulation of ING2.
We also present the first systematic analysis of ING protein expression in adult and tadpole tissues using an antibody that is capable of recognizing both ING1 and ING2 proteins. We demonstrate that there is a great deal of similarity in expression pattern between brain and testis tissues and that there are distinct tissue-specific isoforms. We were unable to determine which protein bands correspond to ING1 versus ING2 proteins and are currently producing isoform-specific antibodies to address this question.
ING1 plays an important role in apoptosis (4)(5)(6)(7)20). We have shown that ING protein expression is elevated in response to TH-induced tail regression and that the timing of this event corresponds to the point of commitment for cells to complete a TH-induced program (38). This study has provided evidence that the 90-130 kDa ING proteins may be important in regulating hormone-induced apoptosis. The tail and brain, two tissues that undergo extensive apoptosis, show a TH-dependent increase in these proteins whereas the leg, whose main response is proliferation and growth, does not. In addition, induction of these proteins is inhibited by a variety of agents that It is clear that cellular context is an important determinant of the TH-regulated response (33), but the mechanism is still poorly understood. TRs bind predominantly as heterodimers with RXR to accessible stimulatory TREs in the absence of T 3 (66)(67)(68)(69)(70). This receptor complex recruits transcriptional co-repressors such as N-CoR, Sin3 and mRPD3 that associate and form a functionally active histone deacetylase (44)(45)(46)(47)(48)(49)(50). Histones are deacetylated resulting in repression of transcription. In the presence of T 3 , the co-repressors are released and acetyltransferases (p300/CBP, P/CAF, TAFII250) are recruited (46,51,52). Histone acetylation then permits transcription. TR action is further modulated by interaction with many other proteins including auxiliary proteins (TRAPs) and the tumour suppressor p53 (53)(54)(55)(56)(71)(72)(73). Since p53, histone acetylases and deacetylases have been reported to interact with ING1 proteins (23)(24)(25)(26), it is reasonable to speculate that ING protein isoforms may modulate TR activity through affecting HAT/HDAC activity to produce tissue-specific outcomes during tadpole metamorphosis.

29.
Helbing   shown beginning at aa 111. The exon 1/2 boundaries for human (Nagashima et al, unpublished) and Xenopus ING2 and murine and human ING1 (4,9) are indicated by a question mark and an asterisk, respectively. The alignment was done using Clustal W alignment software (57). The Genbank Accession numbers for each of the sequences used are AB012853, NM_02353, AF181849, AF181850, AF078834, NM_011919, AF17775352 and AF149724. or within the putative exon 2 (xING2 (2)). L8 ribosomal protein transcript, known to remain constant between tissues (59), is shown below and was used to normalize the xING2 amplification products. C, Graph comparing the fold differences in normalized xING2(1/2) (hatched bars) and xING2(2) (solid bars) transcript levels relative to the liver.   proteins were transferred to nitrocellulose membrane and probed with antibody that is specific for the common region of ING. C, Western blot analyses of total proteins isolated from cultured premetamorphic tails. Tails were cultured in serum-free medium using the method described in (74). Tails were treated with 100 nM T 3 in the presence or absence of 2 mM EGTA that inhibits tail regression (62). Similar results were obtained by inhibiting tail regression with cycloheximide/anisomycin (38) and the protein kinase C inhibitor, H7 (62) (data not shown). Total protein homogenates were isolated at the indicated times and Western blotted as above. The relative sizes of specific bands were determined by comparison with comigrating standard protein markers and are indicated in kiloDaltons. D, Northern blot analyses of total RNA isolated from the tail of premetamorphic tadpoles immersed in T 3 for the indicated times probed with a human 300 bp ING1 PCR fragment from exon 2 (upper panel). Only one band of 3.9 kb is detected. Relative RNA loading is indicated by the intensity of the 28S rRNA bands as visualized by ethidium bromide staining of the gel (lower panel).