Pthlh, a promising cancer modifier gene in rat tongue carcinogenesis

Susceptibly to the induction of rat tongue cancer (TC) by oral 4-nitroquinoline 1-oxide (4NQO) exposure is a polygenic trait. Among several quantitative trait loci identified by crosses between TC-susceptible Dark Agouti (DA) rats and TC-resistant Wistar-Furth (WF) rats, we focused on tongue cancer susceptibility locus (Tcas3) of chromosome 4. We examined tongue carcinogenesis in the reciprocal congenic strains DA.WF-Tcas3 and WF.DA-Tcas3 and in their parental strains. The Tcas3DA allele, and not the Tcas3WF allele, significantly favored tumor latency, incidence and TC number/size. In genomic DNA of TCs induced in (DA × WF) F1 rats, the resistant Tcas3WF allele was frequently and selectively lost, particularly in larger tumors. Thus, we searched the possible candidate genes in the Tcas3 region using microarray analysis of TCs in F1 rats and revealed significant upregulation of 2 cancer-related genes, parathyroid hormone-like hormone (Pthlh) and Kras2. The relevance of the WF allele of Pthlh as a cancer modifier was indicated by 3 single nucleotide polymorphisms specific to this strain. In contrast, no consistent strain-specific variations were found in Kras2. Moreover, the plasma Ca2+ level was consistently higher in DA rats when compared to the level in WF rats bearing TCs; moreover, the Pthlh-mRNA expression level was >30-fold higher in TCs when compared to this level in the normal tongue mucosa. Immunostaining experiments showed strong PTHrP protein expression in TCs of DA rats, and the signal was intensified in larger TCs. Kras2 was also upregulated in TCs, but to a lesser degree than PTHrP. Thus, Pthlh is a promising candidate modifier gene in the development and progression of rat TCs.

In the present study, we focused on the single-locus effects of the tongue cancer susceptibility QTL 3 (Tcas3) on rat chromosome 4 (RNO4) by constructing reciprocal speed congenic strains; WF rats carrying a DA-derived Tcas3 (Tcas3 DA ) chromosomal segment and DA rats carrying a WF-derived Tcas3 (Tcas3 WF ) chromosomal segment. The modifier effects of Tcas3 on 4NQO-induced tongue carcinogenesis were confirmed by allele type-dependent incidence of TCs and frequent deletion of Tcas3 WF in TCs in (DA x WF) F1 rats. Subsequently, we identified genes responsible for these modifying effects from the Tcas3 region by comparing microarray analyses of TCs and normal tongue tissues, revealing a significant elevation in parathyroid hormone-like hormone (Pthlh) and Kras2 expression. Immunohistochemistry of TCs showed that the Pthlh signal was more intense than the Kras2 signal. Subsequent sequencing of DNA showed 3 unique single-nucleotide polymorphisms (SNP) in the WF Pthlh gene. Rats carrying the Tcas3 DA and 4NQO-induced TCs had elevated serum Pthlh and Ca 2+ levels and their WF Pthlh gene carried 3 single-nucleotide polymorphisms not found in other rat strains. These findings suggest that Pthlh is a promising candidate gene, located at Tcas3, involved in the development and progression of rat TC.

Materials and methods
Animals. The following strains of rats and their F1 (DA x WF) progeny were used in the present study. Dark Agouti/SIc rats were purchased from Japan SLC Inc. (Hamamatsu, Japan); Fisher 344/DuCrj (F344) rats were from Charles River Japan Inc. (Atsugi, Japan); Sprague-Dawley/JcI (SD) rats were from Japan Clea Co. (Tokyo, Japan); and DRH rats were from SEAC Yoshitomi (Fukuoka, Japan). Long Evans/Stm (LE) rats were introduced from the Saitama Cancer Center (Ina, Japan); ACI/Ms (ACI) rats were from the National Institute of Genetics (Mishima, Japan); and Donryu rats were from Osaka University (Osaka, Japan). The Wistar-Furth/Stm (WF) rats were originally from Hiroshima University (Hiroshima, Japan). The present study was carried out according to the Animal Care Guidelines of Asahi University, Gifu, Japan (10-007).
Generation of speed congenic lines. Speed congenic strains for the Tcas3 locus were developed using marker-assisted selection starting from an intercross between DA and WF rats (7,8). The chromosomal segment containing Tcas3 was defined by microsatellite markers D4Rat206 and D4Rat72. A male rat of F1 with DA-derived Tcas3 (Tcas3 DA ) from an F1 x WF cross was selected and mated with WF females (8,9). After serial selective backcrossing with WF rats over 6 generations, heterozygous littermates were mated. Then a pair of littermates that were homozygous for Tcas3 DA were selected and subjected to successive inbreeding. The resulting strain was designated WF.DA-Tcas3. The reciprocal congenic strain DA.WF-Tcas3 was generated using the same mating protocol. All congenic rats used in the present study were of the third or subsequent generation of Tcas3 homozygotes.
Genotype analysis for speed congenic lines. All primers for PCR-based microsatellite analysis were purchased from Research Genetics, Inc. (Huntsville, AL, USA). We used 135 of the 360 polymorphic markers between DA and WF rats to characterize congenic strains. Of these, 40 were found on RNO4 and the others were distributed 10-30 cM apart on each rat chromosome (5,6). Theoretically, the percentage of DA/WF segments was <1%. PCR and agarose electrophoresis of PCR products were performed as described previously (1,2,10).

Tongue cancer induction.
A stock solution of 4NQO (Nacalai Tesque Inc., Kyoto, Japan) at a concentration of 200 mg/l in 5% ethanol was prepared and stored at 4˚C until use. Starting at 6 weeks of age, all rats were given drinking water containing 0.001% 4NQO ad libitum from 5 p.m. to 9 a.m. Rats were inspected once a day for TC development and were weighed once a week. All of the rats given 4NQO (DA, WF, F1 and congenic rats) were sacrificed when they became moribund or on day 180 of the experiment. Full necropsy and histopathological examinations were performed. The diameter of the largest TC tumor (DTCmax) and the number of TCs with a diameter ≥5 mm (TC#5) were recorded at necropsy. Paired samples of the largest TC and kidney or tail from each rat were obtained and stored at -80˚C. High-molecular-weight DNA was extracted from the frozen tissues as previously described (1,2).
Fluorescently tagged primers were purchased from Applied Biosystems (Foster City, CA, USA). Positions of loci were mapped on the Rat Genome Database (6). LOH analysis was performed for tumor samples from the F1 progeny of DA and WF rats as previously described (11,12). Genomic DNA from TCs of 100 reciprocal F1 rats (50 females and 50 males) was used to survey LOH. Samples were scanned using a PRISM 310 Genetic Analyzer, and the data were collected automatically and analyzed using GeneScan software (both from Applied Biosystems). A Genotyper 2.0 was used to score alleles and assess LOH. A difference between the alleles was expressed as the ratio of the tumor signal to the normal signal (T1/T2 over N1/N2). Ratios <0.67 or >1.35 were considered indicative of LOH for that locus (12).
Microarray analysis. TCs >5 mm in diameter and normal tissues from F1 rats were analyzed using microarrays. RNA was isolated from the tissues and subjected to linear amplification by RiboAmp RNA Amplification kit (Arcturus, Mountain View, CA, USA). RNA amplification efficiency was compared with that of control RNA of known quantity (0.1 µg) by running in parallel with the 18 samples. Gene expression analysis was performed using the GeneChip ® Rat Gene 1.0 ST Array (Affymetrix Inc., Santa Clara, CA, USA) technique according to the manufacturer's protocol. The arrays were scanned, and fluorescence intensity was measured using Microarray Suite v5.0 software (Affymetrix Inc.). The arrays were then imported into DNA-Chip analyzer software for normalization and model-based analysis. All statistical analyses were performed using S-Plus 6.0 (Insightful, Seattle, WA, USA) software. Three criteria were applied to determine differentially expressed gene transcripts as follows. First probe sets on the array that were assigned as 'absent' call in all samples. Second, a two-tailed Student's t-test was used for comparison of the average gene expression signal intensity between the F1 TCs (TC, 5-10 mm; n=15) and normal samples (n=3), and differences were considered significant at a critical α level of 0.05. Finally, fold ratios were calculated for gene transcripts that showed a statistically significant difference (P<0.05), and only gene transcripts that exhibited >2-fold changes were included in further analyses.
Single-strand conformation polymorphism (SSCP) analysis and sequencing in F1 TC rats. To detect mutations in genes of the Tcas3 region including Pthlh and Kras2, SSCP analysis was performed in tumor samples from F1 progeny of DA and WF rats, as previously described (12,13). Samples with altered electrophoretic mobility were re-amplified, and direct sequencing of both strands was performed to confirm and characterize mutations.
Quantitative real-time PCR. Total RNA was extracted from the tumors and normal tongue mucosa from 4NQO-induced TCs from 50 F1 rats. Tissue specimens were homogenized using an RNAqueous kit (Ambion, Grand Island, NY, USA). cDNA was generated using a High Capacity cDNA Transcription kit (Ambion) and was amplified by PCR using a TaqMan Universal PCR Master Mix and TaqMan Gene Expression assays with 18S (Rn03928990-g1), Pthlh (Rn00561818-g1) and Kras2 (Rn00580460-m1) (Applied Biosystems). PCR products were analyzed using Applied Biosystems StepOne™ with Gene Amp software and StepOne real-time PCR systems (13).
Direct sequence analysis of Pthlh in 8 rat strains. To detect SNPs in rat Pthlh (NC-005103.3, ID24695, M34108), genomic DNA was obtained from the rat kidney, and the entire sequence of the Pthlh locus was determined in 8 rat strains using an ABI 310 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). Allele calling was performed using the Genotyper 2.0 software (Applied Biosystems), and identified SNPs were compared with those previously reported (12)(13)(14)(15).
To determine whether SNPs in translational regions led to structural or functional amino acid substitutions in the Pthlh protein, we conducted in silico analysis using freely available software, including MutPred (16), Panther (17), PhD-SNP (18), SNAP (19), PolyPhen2 (20) and SIFT (21). In addition, we evaluated the impact of SNPs on transcription factor-binding sites by searching for potential regulatory motifs at the 5' UTR using Matlnspector (22) and TESS (23). The software default values were used in all bioinformatic analyses.
Immunohistochemistry of tongue cancer. All tongue specimens from F1 rats were routinely fixed in buffered 10% formalin and were embedded in paraffin. Serial 4-µm sections were dewaxed in xylene and rehydrated in graded ethanol. Sections were stained with hematoxylin and eosin (H&E) to confirm histological diagnosis, and separate sections were used for immunohistochemical analyses. Sections were incubated with diluted rat polyclonal anti-PTHrP antibody (sc-9685; 1:100) and mouse monoclonal anti-Kras2A antibody (sc-13794; 1:50) (both from Santa Cruz Biotechnology, Inc., USA) as primary antibodies overnight at 4˚C.
Statistical analysis. Numerical data are presented as means ± standard deviation (SD). Statistical analyses were performed using one-way ANOVA or Fisher's test with SPSS 17.0 (SPSS Inc., Chicago, IL, USA) software.

Results
Generation of speed congenic rat strains for Tcas3. To study the phenotypic effects of the single Tcas3 locus on TC development, we established a set of speed congenic strains for Tcas3 using marker-assisted backcrossing of DA and WF rats and generated the WF.DA-Tcas3 and DA.WF-Tcas3 strains, respectively (8,9). WF.DA-Tcas3 rats had an introgressed DA-derived RNO4 segment spanning D4Rat140 to D4Rat70, whereas DA.WF-Tcas3 rats carried a WF-derived RNO4 segment spanning D4Rat112 to D4Rat70 (Fig. 1). Both segments were ~6-cM long, were partly overlapping and contained several cancer-related genes including Pthlh and Kras2 (Fig. 1).
Effects of Tcas3 on tongue carcinogenesis. We previously reported tumor incidence, number and size in descending order in DA, F1 and WF rats (1,2,12). To evaluate the effect of introgressed Tcas3 segments, rats of both congenic strains, parental DA and WF strains and F1, were administered 4NQO, and tongue and oral carcinogenesis was observed. As shown in Fig. 2, DA rats developed TCs with the shortest latency and highest incidence, whereas WF rats developed TCs with the longest latency and lowest incidence. In congenic strains, Tcas3 DA showed modest but significant accelerating effects on TCs in WF rats, whereas Tcas3 WF had TC-inhibitory effects on DA rats. These differences in strains were reflected by  incidence, number and size of tumors (Table I). These observations clearly indicate that Tcas3 alone exhibits significant phenotypic effects on the development and growth of TCs and oral cancers. That is Tcas3 DA augmented grown whereas Tcas3 WF suppressed it.
Selective LOH of Tcas3 WF in larger TCs. To determine whether suppressive Tcas3 WF was selectively deleted in TCs, genomic DNA of TCs induced in F1 rats was examined for LOH. Out of 100 F1 rats, 10 TCs of <5-mm in diameter, 29 TCs of 5-10 mm in diameter and 40 non-TC tissues did not show any LOH. However, in 21 TCs of ≥10-mm diameter, LOH was observed between D4Mgh10 and D4Mgh13 on chromosome 4q44. In the majority of cases, the WF allele was selectively lost (Fig. 3). Among the major cancer-related genes on the Tcas3 segment, the incidence of LOH at Kras2 and Pthlh was 32.7 and 40.8%, respectively in the TC#5 (Table II). Loss of the Tcas3 WF allele only in larger TCs indicates that it represents a late event in tumor progression.

Microarray analysis.
To identify genes that are responsible for the effects of Tcas3, microarray analysis was performed in TCs and normal tongue tissues from F1 rats. According to the Rat Genome Database (6), 27 genes, 6 ESTs and 18 pseudogenes  are present in the Tcas3 region. Table III shows the expression of the 27 genes as determined by microarray analysis. Significantly increased expression was observed for Pthlh (7.33-fold, P<0.001) and slightly less for Kras2 (5.21-fold, P<0.001).
The expression of genes in the Tcas3 region was analyzed in 10 control and 60 TC samples using quantitative real-time RT-PCR. Pthlh mRNA expression in large TCs (>10 mm) was consistently >30-fold higher than that in normal tongue mucosa (Fig. 4), and expression levels were higher in larger tumors.  -Table II. Immunostaining of the PTHrP protein in tongue cancer (TC) samples from F1 rats.
Using computational analyses, we evaluated the impact of the 166th amino acid substitution on the structure and function of the Pthlh protein. MedPred predicted a low impact, with a probability of deleterious mutation score of 0.134 (>0.5 indicates impact). Panther also predicted that the substitution was almost neutral, with a deleterious probability of 0.34315 (>0.5 indicates impact) and PhD-SNP predicted no impact. SNAP also predicted that the impact was neutral with 85% expected accuracy. Albeit with low confidence, SHIFT gave a score of 0 for affected protein function. Polyphen2 did not yield a prediction. Thus, all bioinformatics tools revealed almost no impact of these SNPs on Pthlh protein structure and function.
Subsequently, Kras2 point mutations were examined at codons 12, 13 and 61 in F1 TCs. One of the tumors had a heterozygous CAA→CAT mutation in codon 61 of Kras2, resulting in an amino acid substitution from glutamine to histidine. No other activating mutations were found in the Kras2 gene. Therefore, any involvement of Kras2 may not be predominant in rat TCs.
Plasma electrolyte and antibody levels in the WF, congenic and DA rats. Table IV shows the titer of plasma electrolytes and immunoreactive PTHrP in control and TC-bearing rats. Electrolyte levels in control rats did not vary according to genetic background and Tcas3 genotype. However, plasma Ca 2+ levels were consistently higher in TC-bearing rats than in control rats. We also evaluated levels of PTHrP peptides derived from post-translational cleavage of the whole protein.
The PTHrP-C peptide levels were significantly elevated in TC immunohistochemistry. Among 50 TC#5 and 10 rats with TCs of <5 mm in F1 rats, positive signals for PTHrP were detected in the nucleus and cytoplasm of prickle-type cancer cells (Fig. 6). Staining intensity above or below the cut-off score (10%) was classified as ̔positiveʼ or ̔negativeʼ, respectively, at a magnification of x100 using computer-associated image analyzer software (WinROOF 7.1; Mitani Co., Japan) (26). As shown in Table V, the fraction of positively stained TCs increased with their size. There were significant differences in TCs between the TC#5 and DTCmax groups (P<0.001), and among the TCs <5 mm, 5-10 mm and TC >10 mm (P<1x10 -10 , χ 2 =47.63). The expression of Kras2 was consistently weaker than that of PTHrP (data not shown).

Discussion
Out of 5 QTLs that affect susceptibility to TCs in 4NQO-treated rats, we focused on Tcas3 in the present study. In genome-wide association studies in the F2 rat, the peak logarithm of the odds score 6.88 was observed for Tcas3 at 4q44 (2). This chromosomal region is homologous to human 12p12.1-q11.2 and harbors the cancer-related genes Pthlh and Kras2 (7,12).
To evaluate the effects of a single Tcas3, we generated speed congenic strains for Tcas3 using marker-assisted backcrossing.  As shown in Table I and Fig. 2, the quantitative parameters of carcinogenesis, namely tumor latency, incidence, tumor number and size, were moderately but significantly modified by Tcas3. Hence the Tcas3 DA allele increased susceptibility to tongue carcinogenesis and progression, whereas the Tcas3 WF allele bestowed resistance. Selective loss of the resistance allele Tcas3 WF was frequent in larger TCs induced in F1 rats, and the consensus stretch of LOH contained Pthlh and Kras2. Microarray comparisons of TCs and normal tongue mucosa also indicated that among genes in the Tcas3 region these 2 genes were significantly highly expressed. We focused on Pthlh following the observation that resistant WF Pthlh rats carry 3 unique SNPs. The impact of these SNPs will be discussed in greater detail below. Other observations that suggest the relevance of Pthlh in tongue carcinogenesis include (i) a 30-fold higher expression of Pthlh in TCs than that in normal tongue epithelium, (ii) very strong PTHrP immonostaining in larger TCs from F1 rats and (iii) increased plasma levels of Ca 2+ and PTHrP-C in TC-bearing DA rats when compared with the control rats.
The anabolic effects of Pthlh have been demonstrated in rodents and in humans. Pthlh is expressed in many cell types and usually acts as an autocrine, paracrine, and/or intracrine factor that plays numerous roles in embryonic development and physiology (27,28). PTHrP may also have important  functions in tumor development. PTHrP has been shown to stimulate proliferation and invasiveness of cancer cells as well as protection from apoptosis. PTHrP has the potential to cause malignant hypercalcemia and induce local osteolysis, which facilitates the growth of tumor cells that have metastasized to bone (28,29). Partial homology of PTHrP and parathyroid hormone allows PTHrP to activate parathyroid hormone 1 receptor (PTH1R). PTH1R activation in bone and kidney leads to bone absorption and renal calcium retention, respectively, inducing a rise in blood calcium levels (30). PTHrP consists of 3 molecular domains, the parathyroid hormone-like domain in the N-terminal region, a mid-region and a C-terminal domain. Post-translational cleavage of the PTHrP protein allows these domains to function independently. The parathyroid hormone-like domain stimulates protein kinase A, protein kinase C, and/or the calcium-dependent pathways by activating PTH1R (31,32). The mid-region domain can enter the nucleus since it carries a bipartite nuclear localization sequence at residues 88-91 and 102-106 and an importin β-binding site at residues 66-94 (33). After translocation into the nucleus, the PTHrP mid-region domain modulates gene expression by an as yet undefined mechanism. Nuclear transport of the mid-region domain is regulated by CDK1(CDC2)/CDK2-dependent phosphorylation (34). The C-terminal domain of PTHrP (osteostatin) physically interacts with β-arrestin (35), which regulates internalization and desensitization of ligand-stimulated G-protein-coupled receptors (36). Osteostatin bears a number of phosphorylation sites that are important for the mitogenic activity of PTHrP in vascular smooth muscle cells (37).
To understand the potential impact of SNPs in untranslated regions of PTHrP genes, we performed in silico analyses using computational tools with various prediction algorithms and found several known regulatory motifs on the 5'-UTR of the Pthlh sequence. These analyses identified only 2 transcription factors, TCF-4E and ZFX, in all strains apart from WF, indicating that the SNPs at +2 bp (T→G) and +17 bp (T→C) eliminate the transcriptional factor binding site. In contrast, PLU1-JARID1B.01, SP1 and ROAZ.01 were found only in WF rats, indicating that these 2 SNPs affect transcription factor binding to these motifs. ZFX and ROAZ.01 are zinc finger proteins with a DNA-binding motif, and SP1 is common to many eukaryotic transcriptional regulatory pathways. The transcription factor TCF-4E regulates expression of members of the Wnt pathway and its splicing isoforms are related to the progression of renal cell carcinoma (38). The motif PLU1-JARID1B.01 was previously identified in the development and progression of breast cancer (39). However, no clear relationship between these transcription factors and oral cancer development has been reported to date. Hence, further research is necessary to understand the effect of these motifs and SNPs at +2 and +17 bp. Using several bioinformatics tools with varying predictive algorithms, we also analyzed structural and functional effects of the SNP at +1485 bp, which leads to an amino acid substitution at the 166th amino acid on the Pthlh protein. Notably, none of these analyses suggested that this SNP would have any effect. Potentially, these computational tools, which are generated using characterized protein, may not predict novel functional motifs, even though they yielded consistent predictions. Benelli et al (15) reported an amino acid polymorphism in Pthlh that elicited cancer-modifying effects in a mouse squamous cell carcinoma cell line (15). Importantly, the SNP at position +1485 [+496 bp in the coding sequence (CDS)] bp was common to both mouse and rat models. This SNP can be expected to substitute the polar residue threonine with the hydrophobic residue proline at the 166th amino acid (ACC→CCC) of the precursor protein (position 130th of the mature protein), in which the CDS start is at 527 bp.
The present study revealed that Tcas3 alone contributes moderate but significant phenotypic effects to 4NQO-induced rat tongue carcinogenesis. Several observations suggest that Pthlh is a promising candidate gene in Tcas3, although further studies using appropriate reporter assays and in vivo transgenic analyses are required to provide direct supporting evidence of this. In subsequent studies, we will focus on structural and functional analyses of Pthlh and its products.