It is well documented that TS inhibitors such as 5-FU and ZD1694 can promote a 2–4-fold increase in the TS enzyme level in cultured cell lines, tumor models and clinical specimens 
. To determine if the treatment of zebrafish embryos with 5-FU or ZD1694 would also result in the over-expression of TS, Western immunoblot analysis was performed to evaluate the TS expression levels in untreated and 5-FU- or ZD1694-treated embryos. As shown in , both 5-FU and ZD1694 significantly increased TS expression in zebrafish embryos. At a concentration of 0.1–1.0 µM 5-FU, TS expression was increased in a dose-dependent manner (, lane 1–5). The highest induction of TS expression, which was about a 6.5-fold increase over untreated embryos, was found in those treated with 1.0 µM 5-FU ( A, B). Increasing 5-FU concentration to 2 µM did not result in an additional enhancement in TS expression (data not shown). In embryos treated with ZD1694, TS expression was more potently increased; treatment with 0.4 µM ZD1694 resulted in 10 times more TS expression compared to untreated embryos (). Of note, the expression of β-actin, remained unchanged in both untreated and treated embryos. Northern blot analyses were also performed to determine changes in the TS mRNA level in response to 5-FU and ZD1694 treatment. Treatment of zebrafish embryos with 0.1, 0.2, 0.5 or 1.0 µM 5-FU for 24 h did not result in any alteration of the TS mRNA level (, panel 3). Similarly, the mRNA level was also unaffected following treatment with various concentrations of ZD1694 (, panel 3). These findings suggested that the induction of TS protein in zebrafish after drug treatment is dose-dependent and occurs at the post-transcriptional level.
Expression profile of TS mRNA and protein in zebrafish embryos treated with 5-FU or ZD1694.
To more precisely identify the underlying molecular events of the regulation of TS gene expression, the level of TS protein was determined in response to 5-FU drug treatment for varying amounts of time. Densitometry analysis of the Western immunoblot experiments demonstrated a 3.0-, 4.1-, 6.2- and 7.4-fold increase () in TS protein when zebrafish embryos were treated with 1.0 µM 5-FU for 8 (, lane 1), 16 (, lane 2), 24 (, lane 3) or 32 h (, lane 4), respectively. Furthermore, TS expression in untreated embryos remained mostly unchanged from 8 to 32 h (, lane 1–4). These results suggest that 5-FU induced the expression of TS in zebrafish in a time-dependent manner.
Time-dependent effect of 5-FU on TS expression.
To determine if TS could bind to its cognate mRNA in vitro, we performed an electrophoretic mobility shift assay (EMSA) with TS recombinant protein and 32P-labeled full-length TS mRNA. A TS mRNA probe was incubated with TS protein and analyzed using a non-denaturing 5% acrylamide gel. The results demonstrated the presence of a specific RNA-TS complex (, lane 1). To determine the specificity of the TS–mRNA interaction, a 100-fold excess of luciferase mRNA was included in the reaction mixture, and the results demonstrated that luciferase mRNA did not affect the binding of TS protein to its own mRNA (, lane 2). In contrast, no interaction was found when an excess amount of TS mRNA was added to the reaction mixture (, lane 3). Replacing TS protein with BSA also did not lead to a bound complex (, lane 4). These results demonstrated that the TS protein could bind specifically to its cognate mRNA in vitro.
Specific interaction between zebrafish TS and its cognate mRNA in vitro.
In order to more precisely localize the TS binding site in the full-length TS mRNA, we initially performed a series of RNA competition gel mobility-shift experiments. Several TS RNA sequences from zebrafish TS mRNA were synthesized in vitro and used as unlabeled RNA competitors in gel mobility-shift assays to determine the relative binding affinity (IC50
) of zebrafish TS protein for each sequence (). Full-length TS mRNA effectively competed for TS protein binding (, IC50
0.50 nM), while the luciferase mRNA had a 1000-fold lower protein binding affinity than the TS full-length mRNA (). TS mRNA sequences, including TS 1–580, TS 1–290, TS 1–145, TS 1–72 and TS 1–36 displayed similar TS protein binding activity as the full-length TS mRNA in the competition assay (). However, the TS mRNA sequences TS 581–1163, TS 291–580, TS 146–290, TS 73–145 and TS 37–74 were unable to compete for TS protein binding (). The results from these competition experiments suggested the presence of a TS binding site within the first 37 nt of the TS mRNA.
Relative binding affinity of TS RNA constructs.
Previous studies have demonstrated that a stem-loop secondary structure is present in the upstream region of human TS mRNA, which included AUG site, that plays key roles in RNA and human TS protein interaction 
. In the present study, the first 37 nt in the TS RNA sequence were subjected to MFOLD analysis (http://mfold.bioinfo.rpi.edu/cgi-bin/rna-form1.cgi
) and the predicted stable stem-loop structure was found in nt13–32 in TS mRNA, designated TS:N20 (). To more accurately define the sequence and/or structural requirements for the RNA–protein complex formation within this specific region, we synthesized the 20-nt RNA construct (nt 13–32) found in the 5′UTR of the TS RNA sequence. TS:N20 competed for TS protein binding with activity similar to the full-length TS RNA (, IC50
An RNA gel mobility shift assay was used to determine structural requirements for the TS:N20–TS protein interaction. 32
P-labled TS:N20 RNA could form an RNA–protein complex in the presence of 100 ng pure recombinant TS protein (, lane 2). In contrast, deleting of the UGC sequence in the loop structure of theTS:N20, the mutant Δ20-TS resulted in complete loss of this interaction (, lane 3). To provide further support for the importance for maitain an intact stem-loop structure, two other TS RNA variants were synthesized: Δ20–18-TS RNA, which contained a two base substitution, GU→CA, at nt 18 and 19, and Δ20–23-TS RNA, which contained base substitutions, GCU→AAA, at nt 23–25. These two TS RNA variants were unable to form a TS–RNA complex with the zebrafish TS protein (, lanes 4 and 5). Additionally, when 100 ng full-length TS mRNA was included in the reaction mixture, the interaction of TS:N20 with TS protein was abolished completely (, lane 6). To more precisely measure the ability of these three RNAs to interact with TS protein, we determined their relative binding affinity for TS protein compared to full-length RNA. TS:N20 bound TS protein with a relative affinity (, IC50
0.62 nM; , panel 1) similar to the full-length RNA. The three RNA variants, Δ20-TS RNA, Δ20–18-TS RNA and Δ20–23-TS RNA, were unable to compete with 32
P-labeled full-length TS RNA for TS protein binding, even at a high concentration (, panel 2–4). The MFold program was used to predict the stability of the secondary structure of TS:N20 and its mutant constructs (). Like TS:N20, both Δ20–18-TS and Δ20–23-TS RNA could form a stable stem-loop structure. However, the loop of TS:N20 consisted of six nucleotides, while the loop in Δ20–18-TS and Δ20–23-TS RNA contained only five and four nucleotides, respectively. These findings suggested that TS:N20 localized to the 5′UTR of the TS RNA sequence is a critical site for TS protein binding, and that some specific nucleotides and the loop size of the secondary structure formed by the 20-nt sequence play a key role in RNA–TS protein interaction.
Interaction of TS:N20 with zebrafish TS protein in vitro.
Stability of TS:N20 and its mutant constructs.
To further assess the in vivo binding activity of TS protein, zebrafish embryos were homogenized and the TS–mRNA complex was immunoprecipitated using a TS monoclonal antibody. Western blot analysis and RT-PCR experiments were performed to determine if the precipitated complex included TS mRNA and TS protein. The results demonstrated that TS protein could be co-precipitated in the presence of the monoclonal antibody, TS106 (, lane 2). TS mRNA from nt 340–1040 was also amplified by RT-PCR using TS-specific primers (, lane 4 and 5). In contrast, no specific interaction was detected without the TS antibody (, lane 2). Additionally, when an unrelated monoclonal antibody was used, no amplified RNA was detected (, lane 3). These results confirmed that TS interacted with its own RNA in vivo specifically.
Zebrafish TS protein interacted with its own mRNA in vivo.
To determine the biological relevance of the TS protein–mRNA interaction, a rabbit reticulocyte lysate in vitro translation system was used. Incubation of zebrafish TS mRNA with the reticulocyte lysate yielded a protein product with a molecular mass of approximately 35 kDa (, lane 1). To determine if the translation of TS was affected by the presence of other proteins, BSA (, lane 2), luciferase (, lane 3), human p53 protein (, lane 4) or alpha-chymotrypsin (, lane 5) were included in the translation mixture. Our results showed that unrelated proteins did not affect TS translation (, lane 1–5). In contrast, when an excess amount of zebrafish TS protein was added to the reaction mixture, the translation of TS mRNA was significantly inhibited (, lane 1–5). The formation of TS in the translation system was negatively dose-dependent on the amount of TS added to the translation system (, lane 1–5).
TS protein inhibited in vitro translation of zebrafish TS mRNA.
Studies by Berger et al. revealed that treatment of cancer cells with 5-FU resulted in increased stability of the TS protein 
. In order to determine if treatment with 5-FU affected TS stability in zebrafish embryos, we examined the TS remained after the embryos were treated with cycloheximide and 5-FU at different times. As shown in , -FU treatment of the embryos significantly increased TS stability. Compared with the untreated embryos, the half-life of TS increased by about 2.7 times; the half-life of zebrafish TS increased from 7.2 h to around 19.5 h after 5-FU treatment (). These results suggested that treatment of embryos with 5-FU could strongly enhance the stability of TS.
Effect of 5-FU on the stability of TS protein in zebrafish embryos.