The expression of p73 is increased in lung cancer, independent of p53 gene alteration

p73 gene, a new p53 homologue, has been identified: it supposedly acts as tumour suppressor gene in neuroblastoma. To clarify whether p73 might be involved in lung carcinogenesis, we examined p73 expression in resected lung cancer and paired normal lung in 60 cases using semi-quantitative reverse transcription-polymerase chain reaction (RT-PCR). We also examined p73 gene status in three representative cases using Southern blot, and p53 gene alteration in 49 cases using PCR-single-strand conformation polymorphism (PCR-SSCP) and direct sequence. In 87% of the cases (52/60) p73 expression in tumour was more than twice as high as that in paired normal lung tissues, and the difference between p73 expression in tumour and normal lung tissue was significant (P < 0.0001). However, Southern blot analysis revealed that none of the cases showed p73 gene amplification. Compared with clinicopathological characteristics, p73 expression correlates significantly with histological differences and age of patient, independently (P < 0.05). Concerning p53 gene status, 43% (21/49) showed p53 gene alteration, but there was no correlation between p73 overexpression and p53 gene alteration. Our results suggest that need for further functional analysis of the role of p73 in lung carcinogenesis. © 1999 Cancer Research Campaign

p73, a new candidate tumour suppressor gene, was recently identified at 1p36, the short arm of chromosome 1, by Kaghad et al (1997). The homology between p73 and p53 is extensive within most conserved p53 domains (Zambetti et al, 1993;Ko et al, 1996;Kaghad et al, 1997). Wild p53 works as a so-called security guard to induce cell cycle arrest or apoptosis in response to cellular stresses on DNA damage (Livingstone et al, 1992;Lowe et al, 1993;Dickman, 1997), and loss or inactivation of p53 is thought to contribute to the development of 50% of all human cancers (Levine, 1997). Therefore, according to the known homology between p53 and p73, p73 is also expected to work as a tumour suppressor gene. Deletions of 1p36 are common in neuroblastomas, and extremely low level of p73 mRNA have been found in the majority of neuroblastoma cell lines (Caron et al, 1995;Kaghard et al, 1997), showing that, at least in neuroblastoma, p73 apparently works as a tumour supprressor gene in spite of the lack of definitive evidence.
However, very recently Mai et al (1998) reported that p73 might be overexpressed in lung cancer tissues in comparison with normal lung. Furthermore, p73 genomic mutation has not been found even after intensive research (Mai et al, 1998;Nomoto et al, 1998;Takahashi et al, 1998). Accordingly, it is questionable whether p73 works as a tumour suppressor gene in lung cancer. To verify the role of p73 in lung carcinogenesis, we measured the amount of p73 mRNA expression in 86 primary lung cancer tissues and paired normal lung tissues by semi-quantitative reverse transcriptionpolymerase chain reaction (RT-PCR). We then analysed the relationship between the amount of p73 expression and clinicopathological characteristics. Moreover, we examined for p53 gene alteration to clarify the relationship between p73 mRNA expression and p53 gene status.

Materials and histological classification
Surgically resected tumours and paired corresponding normal tissues were obtained from 86 patients with primary lung cancers at the Cancer Institute Hospital (Tokyo, Japan), from 1990 to 1993. All samples were quickly frozen in liquid nitrogen and stored at -80°C until RNA was extracted. RNA was estimated by β-actin mRNA expression amplified with RT-PCR, and we excluded 26 cases because of the poor quality of the extracted RNA. Distribution of histological types of the 60 lung cancers were: 34 adenocarcinomas, 19 squamous cell carcinomas, three large-cell carcinomas, two small-cell carcinomas and two adenosquamous carcinomas (World Health Organization, 1981) ( Table 1). The median age of the 60 patients was 61 years (range 44-80 years). Forty-five of the 60 patients were men, and all 15 female patients had adenocarcinoma. The stage of each tumour was determined according to the TNM Classification of Malignant Tumours defined by the International Union Against Cancer (UICC, 1992): 19 stage I, 11 stage II, 21 stage IIIA, eight stage IIIB and one stage IV. To evaluate the consumption of cigarettes smoked before lung cancer diagnosis, a smoking index was used: The expression of p73 is increased in lung cancer, independent of p53 gene alteration Y Tokuchi 1 , T Hashimoto 1 , Y Kobayashi 2 , M Hayashi 1 , K Nishida 2 , S Hayashi 1 , K Imai 1 , K Nakachi 1 , Y Ishikawa 3 , K Nakagawa 4 , Y Kawakami 5 and E Tsuchiya 1,2 cigarette consumption per day multiplied by smoking years. Heavy smokers were defined as those with smoking indices over 400.

Preparation of RNA and expression of p73 mRNA using RT-PCR
All tissue samples were frozen in liquid nitrogen immediately after surgery and subjected to isolation of RNA as previously reported (Tokuchi et al, 1999). RNAs were prepared from 0.1-0.2 g of human primary lung cancers and paired non-cancerous parenchyma according to the method of Chomczynski and Sacchi (1987), which was quantified using UV spectrometry at 260 nm. The RNAs (1.0 µg) in a 40 µl reaction volume were reverse transcribed to synthesize cDNA using Takara RNA PCR kit (Takara, Tokyo) with 2.5-µM random hexamers at 42°C. The oligonucleotides used in PCR amplification were as follows: ). Ten microlitres of the amplified product was subjected to 5% polyacrylamide gel electrophoresis, and the radioactivity was evaluated with Bio-Image analyser BAS2000 (Fuji Film, Tokyo). The PCR product size using P1 and P2 is 301 bases; using P3 and P4, 439 bases. No expected PCR products were observed when RT was not performed. To confirm the results with P1 and P2, an additional PCR using P3 and P4 was also performed: it comprised 40 cycles using a GeneAmp PCR System 9600 (Perkin-Elmer Corp.) in the buffer described above, with denaturing at 95°C for 30 s, and annealing and extension at 70°C for 1 min in each cycle. As a control for RNA quality, separate portions of cDNA were amplified using primers specific for human β-actin. PCR was performed using 1 µl of cDNA, 1 U Taq DNA polymerase (Boehringer Mannheim), 0.5 µM P5 and P6, 0.05 MBq of [α-32 P]-dCTP and 0.15 mM magnesium chloride in 25-µl incubation buffers. The reaction (95°C and 63°C for 30 s, and 72°C for 1 min) was performed for 21 cycles.
To quantify the amount of PCR products of p73 (using P1 and P2) and β-actin, 18 matched samples of lung tumours and normal lung tissues were processed simultaneously, and the radioactivity was determined with BAS2000 (Fuji Film, Tokyo, Japan). The amount of PCR product of each sample was expressed as a value relative to the average radioactivity of 18 normal lungs in each assay. Each PCR and electrophoresis procedure was repeated twice, after which we calculated the average p73 and β-actin expression for each sample. Finally, we arrived at the relative p73:β-actin ratio, calculated by dividing the average amount of p73 by that of β-actin, for each sample.

p73 genomic-Southern blot hybridization
Genomic DNAs from tumours and paired normal lung tissues of three representatives cases were extracted using standard methods as previously reported (Tsuchiya et al, 1992;Tokuchi et al, 1999). After each 10-µg DNA were completely digested with HindIII, DNAs were subjected to 0.8% agarose gel electrophoresis. Using vacuum blotter (Bio-Rad), DNA samples were rapidly transferred to the nylon membrane (Hybond-N; Amersham Japan, Tokyo, Japan) in 10 × standard saline citrate (SSC), and then cross-linked under UV light. RT-PCR products amplified with P1 and P2 primers, which contained exon 5, 6 and a part of exon 7 of p73 coding sequence, were cut from the gel and purified using 0.45-µm centrifugal filter (Millipore), phenol-chloroform extraction and subsequent ethanol precipitation. After this, the purified DNAs were radiolabelled with [α-32 P]dCTP and multiprime DNA labelling systems kit (RPN. 1601Y, Amersham). Southern blot hybridization was performed at 59°C in 5 × SSC, 10 × Denhard's solution, 10 mM EDTA, 200 µg ml -1 salmon sperm DNA and 1% sodium dodecyl sulphate (Sambrook et al, 1989). After sufficient washing, the radioactivity was evaluated with Bio-Image analyser BAS2000 (Fuji Film, Tokyo, Japan).

Statistical analysis
All data were analysed using the Statistical Package for Social Sciences (SPSS), and a comparison of relative p73:β-actin expression ratios in tumours and in normal lung tissues was carried out using the Mann-Whitney U-test. The association between relative p73:β-actin expression ratio and clinicopathological parameters (age, sex, smoking habits, histological types and pathological staging) was estimated in 53 adenocarcinomas and squamous cell carcinomas, using the Mann-Whitney U-test and the Spearman rank correlation method. When any parameters significantly correlated with relative p73 expression, we used the partial correlation analysis to exclude the interaction among these parameters. Linear regression was used to test the association between p73:β-actin expression ratio and clinical characteristics; the optimal regression model was chosen using the backward step-wise selection of variables. Moreover, relative p73:β-actin expression ratio was compared with p53 gene alteration in 49 of the above 53 adenocarcinomas and squamous cell carcinomas using the Mann-Whitney U-test. All P-values < 0.05 were considered significant.

RESULTS
To check the linear relationship between amount of RNA and radiolabelled PCR products in the semi-quantitative RT-PCR method, various amounts of total RNA were processed by RT-PCR amplification. The mean value of the duplicate p73 PCR products amplified by using P1 and P2 increased dose-dependently at 35 cycles of amplification ( Figure 1A), and a similar relationship between β-actin PCR products and amounts of RNA was also observed ( Figure 1B). Figure 2 shows the representative p73 RT-PCR products from paired normal lungs and tumours, in which increased p73 expression was observed in tumours compared with normal lung tissues. Use of another set of primers, P3 and P4, confirmed the results of PCR using P1 and P2 (data not shown). In 87% of the cases (52/60), the relative p73:β-actin ratio in tumours is more than twice as high as that in normal tissues, but the reverse is found in only 5% of the cases (3/60, No 85, 130 and 147 in Table 1).
Distribution of the relative p53:β-actin ratio both in normal lung tissues and in tumours in shown in Figure 3. The expression of p73 mRNA in lung tumours is distributed significantly higher than that in normal lungs (P < 0.0001, Mann-Whitney U-test); the means amounts of p73 mRNA in lung tumours and normal tissues were 6.76 and 1.10 respectively.
In order to examine whether the overexpression of p73 exclusively observed in lung tumours was due to the gene amplification, Southern blot analysis was performed using genomic DNAs. Genomic DNAs were extracted from three representative cases in which p73 mRNA expression was markedly higher in tumours than in normal tissues: none of these cases showed the amplification of p73 gene (Figure 4), thus eliminating amplification as the cause of p73 overexpression in these three cases. Next, we analysed the relationship between the relative p73:β-actin ratio and the clinicopatholigical features in 53 lung adenocarcinomas and squamous cell carcinomas, using the Mann-Whitney U-test and the Spearman rank correlation. The relative p73:β-actin ratios in tumours were found to significantly correlate with differences of histology (adenocarcinoma or squamous cell carcinoma), age, sex and smoking index (Tables 2 and  3). However, we could not ignore inter-relationships within these parameters, so we applied the partial correlation analysis. We found that only differences in histology and age significantly correlated with the relative p73:β-actin ratio independently of other factors (Table 4). This result was reconfirmed by the optimal multivariate linear regression model with the backward step-wise selection of variables: the final model obtained after the steps included age and histological types only as significant and independent variables.
Concerning the comparison of p73 expression in normal lung tissues and clinicopathological parameters, the relative p73:β-actin expression ratio in normal tissues is significantly associated with the age of the patients and p73 expression in paired tumours (P<0.05 respectively, the Spearman rank correlation). Accordingly, the relative p73:β-actin expression ratio in normal tissues also increases with age, although it is far lower than in tumours.
We also examined mutational status of p53 gene in 92% of the cases (49/53) with adenocarcinomas or squamous cell carcinomas by PCR-SSCP, in exons 4-8 and exon 10, and subsequent direct sequencing: we found that 43% (21/49) had mutant p53 (Table 1), much the same frequency as in previous reports (Kishimoto et al, 1992). The mean levels of p73 expression in 21 mutant p53 cases and 28 wild p53 cases are 6.19 and 5.17 respectively, and there was no statistically significant correlation between p53 gene alteration and p73 expression (P = 0.11, Table  2). Moreover, even when we divided the subjects into two groups by histological difference, no association was found between p53 gene alteration and p73 expression.

DISCUSSION
The p53 gene and its protein product have been the centre of intensive cancer studies since more than 50% of human cancers contain this gene abnormality (Levine, 1997). Because p73 closely resembles p53 in transactivation (29% identity with p53 amino acids), DNA binding (63% identity with p53), and p53 oligomerization (38% identity with p53) domains, p73 is thought to play an important role in the development and/or progression of various types of human cancers (Dickman, 1997;Kaghard et al, 1997;Oren, 1997). In fact, p73 can enhance levels of endogenous p21/Waf1 protein, the representative target of p53, and p73 can also inhibit cell growth by inducing apoptosis in a p53-like manner (Jost et al, 1997;Kaghard et al, 1997).
In the present study we clearly showed that levels of p73 mRNA expression is higher in tumours than in normal tissues, as two previous papers dealing with only a few samples had reported (Mai et al, 1998;Takahashi et al, 1998). In four reports the authors failed to find p73 genomic mutation despite intensive search (Kaghard et al, 1997;Mai et al, 1998;Nomoto et al, 1998;Takahashi et al, 1998). Judging from these results, it was possible that wild-type p73, not mutant p73, might be overexpressed exclusively in tumours. Furthermore, our results revealed that p73 overexpression was not due to gene amplification but, in all probability, to induction of transcription or to stabilization of p73 mRNA. Moreover, we showed that p73 overexpression is independent of p53 gene alteration.
There are two possible explanations of the relationship between p73 overexpression and lung carcinogenesis. The first is that p73 may work as a security guard (Dickman, 1997): we found no direct relationship between p53 mutation and p73 overexpression in the present study, but the cell cycle in cancer cells is generally faster than in normal cells regardless of p53 gene status, so p73 might work to arrest the cell cycle as a tumour suppressor gene. The   second possibility is that p73 might act as an oncogene in up-regulation of cell growth, as Mai et al (1998) pointed out. Kaghad et al (1997) reported that p73 protein is neither stabilized nor activated by DNA damage, including damage from UV radiation or actinomycin D, so it is clearly different from p53. Because the expressions of p73 and p53 are induced in different manners, it is possible that p73 may have a role different from that of p53 in lung cancer carcinogenesis. And since, in the present study, p73 overexpression was observed exclusively in lung tumour, p73 may possibly work as an oncogene in lung carcinogenesis.
In the present study, our results suggest that p73 overexpression in tumours correlates with histological type and age of patients. Furthermore, p73 expression in normal lung tissues also correlates with age of patients, although p73 expression in normal lung tissues is much lower than that in tumours. Accordingly, p73 expression might be associated with the changes accompanying aging of the host or differences of histological construction in tumors. These observation will provide some clue as to why p73 is overexpressed in lung tumour.
In conclusion, p73 expression in lung tumours is obviously greater than its expression in normal tissues, independent of p53 gene alteration, indicating that the role of p73 in lung carcinogenesis might be different from its role in neuroblastoma. Further study is needed of this possible new function of p73 protein.