The following strains of rats and their F1 (DA × 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).

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 × 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.

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).

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).

LOH in the Tcas3 region, including that of Kras2 (RGD ID: 5036392) and Pthlh (RGD ID: 11187) genes of induced TCs in F1 rats was determined using PCR-based microsatellite analysis (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).

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.

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.

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).

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 (12–15).

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.

Plasma electrolytes including Ca²⁺, Na⁺, K⁺, IP and Cl⁻ were determined at Japan SLC, Inc. (24). The serum levels of Pthlh-N (normal <3.9 pmol/l), Pthlh-intact (normal <1.1 pmol/l) and Pthlh-C (normal <55.3 pmol/l), and interleukin (IL)-6, -8 and -11 (SRL Inc., Tokyo, Japan) were also determined (25).

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.

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.

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).

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.

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.

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.

Pthlh and Kras2 were further examined by sequencing germline and/or tumor DNA. Direct sequencing of the germline Pthlh gene was performed for 8 laboratory rat strains: DA, LE, SD, ACI, Fischer 344, Donryu, DRH and WF. The Pthlh gene in resistant WF rats was found to carry 3 SNPs at positions +2 bp (T→G), +17 bp (T→C) and +1485 (A→C) from the 5′ end of exon 1, but these were not observed in the other strains (Fig. 5). The base change at position 2 and 17 does not cause amino acid substitutions, whereas that at position +1485 [496 bp in the coding sequence (CDS)] bp would 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 potential impact of the 3 SNPs was determined using computational analyses. Since 2 SNPs at +2 bp (T→G) and +17 bp (T→C) were found in the untranslated region of the Pthlh gene, we searched for known transcription factor-binding sites within the −950 to 526 bp region. TCF-4E (CTTTGCA) and ZFX (GAGGCCTGGTG) motifs were found at −1 to +6 bp and +11 to +21 bp, respectively, in sequences from DA, LE, SD, ACI, F344, Donryu and DRH rats in which the +2 bp and 17 bp positions were occupied by T. In the sequence from WF rats, PLU1-JARID1B.01 (TGGCTGTGC), SP1 (TGTGC) and ROAZ.01 (GTGCACCCAGAGGCCCG) motifs were found at −4 to +5 bp, +1 to +5 bp and +2 to +8 bp, respectively, with G and C residing at +2 bp and +17 bp, respectively. The agreement rates in matrices of sequence motifs were 100, 98.8, 96.6, 100 and 73.1%, respectively.

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.

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 Ca²⁺ 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-bearing rats, whereas levels of PTHrP-N and PTHrP-intact were unchanged.

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 ‘positive’ or ‘negative’, respectively, at a magnification of ×100 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<1×10⁻¹⁰, χ²=47.63). The expression of Kras2 was consistently weaker than that of PTHrP (data not shown).