UHRF1 is a novel diagnostic marker of lung cancer

Background: Lung cancer is the leading cause of cancer deaths worldwide. As the sensitivity and specificity of current diagnostic markers are not perfect, we examined whether ubiquitin-like with PHD and ring finger domains 1 (UHRF1), which is overexpressed in various cancers but not yet examined in lung cancer in large scale, can be a novel diagnostic marker of lung cancer. Methods: Immunohistochemical analysis using surgical specimens obtained from 56 US and 322 Japanese patients with lung cancer was performed. Results: The UHRF1 was stained specifically in the nuclei of cancer cells, but not in the other cells. The UHRF1 expression was observed in all histological types of lung cancer, especially in non-adenocarcinomas (non-ADCs), both in the US and Japanese cases. In 322 Japanese non-small cell lung cancer (NSCLC) cases, UHRF1 expression was associated with the histological type (higher in non-ADCs; P<0.00001), gender (higher in male; P=0.00082), smoking (higher in smokers; P=0.00004), pT factor (higher in advanced stage; P=0.00010), and pN factor (higher in cancers with metastasis in regional lymph nodes; P=0.00018). The UHRF1 expression was also associated with poor prognosis for NSCLC patients (P=0.0364). Although UHRF1 overexpression was associated with these malignant indicators, UHRF1 was detectable in half of lung cancer patients in an early pathological stage. Conclusion: The UHRF1 is overexpressed in various types of lung cancer from early pathological stage. Therefore, detection of UHRF1 expression in tissue specimens by immunohistochemistry can be useful for diagnosis of lung cancer in all pathological stages.

Lung cancer is the leading cause of cancer deaths worldwide including the United States and Japan according to the World Health Organization (WHO) database (http://www.who.int/en/). Among various histological types of lung cancers, non-small cell lung cancer (NSCLC) accounts for approximately 80%, whereas small cell lung cancer (SCLC) accounts for approximately 15%. The two major histological types of NSCLC are adenocarcinoma (ADC) and squamous cell carcinoma (SCC). In Japan, SCC accounts for 34.7% and 16.3% of all lung cancer cases in males and females, respectively. On the other hand, ADC accounts for 42.9% and 67.2% of all lung cancer cases in males and females, respectively (incident year 1999Toyoda et al, 2008). The remaining few per cent of NSCLC are other histological types of carcinomas including large cell carcinoma and adenosquamous cell carcinoma; SCC is more common in males, and smoking dramatically increases the risk of this type of cancer; the relative risks of smoking in males were 11.7 and 2.3 for SCC and ADC, respectively, and were 11.3 and 1.4 correspondingly in females in Japan (Wakai et al, 2006). Smoking rates of Japanese males and females in 2007 are 43.3% and 12.0%, respectively, according to Japanese exposure factors handbook published by National Institute of Advanced Industrial Science and Technology in Japan (http://unit.aist.go.jp/riss/crm/exposurefactors/english_summary.html). The 5-year survival rate based on pathologic (p) stages was analysed using a large population of Japanese lung cancer cases (n ¼ 13 010) in 2005 (Asamura et al, 2008). According to the study, the 5-year survival rates of SCLC patients who underwent plumonary resections were as follows: 58.3%, 60.2%, 40.6%, 41.1%, 28.3%, 34.6%, and 30.8% for IA, IB, IIA, IIB, IIIA, IIIB, and IV, respectively. The 5-year survival rates of NSCLC patients who underwent plumonary resections were as follows: 83.9%, 66.3%, 61.0%, 47.4%, 32.8%, 29.6%, and 23.1% for IA, IB, IIA, IIB, IIIA, IIIB, and IV, respectively. This study also showed that ADC histological type, female gender, and age o50 years were significant favourable prognostic factors.
Although there are several biomarkers in clinical use such as SCC antigen, carcinoembryonic antigen, neuron-specific enolase, and pro-gastrin-releasing peptide, none of them are perfect in terms of sensitivity and/or specificity (Ando et al, 2003;Molina et al, 2005). Thus, we have been seeking additional sensitive and cancer-specific biomarkers detectable in serum and tumour tissues to improve the current high mortality of lung cancer, and found more than 20 candidate diagnostic and/or prognostic biomarkers, which are now under development for clinical use (Daigo and Nakamura, 2008). As many molecular signalling pathways are disrupted during lung carcinogenesis, we believe that combinations of biomarkers are necessary for precise diagnosis and prognosis of lung cancer, and thus we continue to search for superior biomarkers.
Ubiquitin-like with PHD and ring finger domains 1 (UHRF1), also known as ICBP90, was identified as a protein whose expression is only detectable in proliferating cells, not in quiescent cells (Hopfner et al, 2000;Unoki et al, 2004). Recently, it was revealed that UHRF1 has a central function in epigenetic modulation during DNA duplication in the S phase (Sharif et al, 2007;Arita et al, 2008;Avvakumov et al, 2008;Hashimoto et al, 2008). Up-regulation of UHRF1 has been reported in various cancers (Mousli et al, 2003;Crnogorac-Jurcevic et al, 2005;Jenkins et al, 2005;Lorenzato et al, 2005;Oba-Shinjo et al, 2005;Pita et al, 2009;Unoki et al, 2009). As no large-scale study of UHRF1 expression in lung cancers has been performed, we examined whether UHRF1 could be a novel diagnostic marker of lung cancer by immunohistochemical analysis to understand the clinical importance of this protein in lung carcinogenesis.
In this report, we examined UHRF1 expression using 56 US and 322 Japanese lung cancer cases by immunohistochemical analysis and found that expression of UHRF1 was significantly up-regulated in almost all histological types of lung cancers, especially in non-ADCs. The expression of UHRF1 was associated with poor prognosis and several other clinicopathological characteristics of the lung cancer patients. As UHRF1 was specifically expressed in cancer cells and detectable in approximately half of lung cancer cases in an early pathological stage, UHRF1 could be a novel diagnostic marker for lung cancer.

Lung cancer clinical tissue specimens
A total of 56 formalin-fixed primary lung cancer tissues from US patients including SCLC and NSCLC was purchased from BioChain Institute (Table 1; Cat T8235724-5, lot B207162; Cat T2235152, lot A504256; Cat T2235152-9, lot A604318, Hayward, CA, USA). A total of 322 formalin-fixed primary Japanese NSCLCs (Table 2) and adjacent normal lung tissue samples used for immunostaining on tissue microarrays was obtained from patients undergoing curative surgical operation at Saitama Cancer Center (Saitama, Japan) (Ishikawa et al, 2004(Ishikawa et al, , 2007. A total of 23 frozen primary lung cancer tissues for RNA extraction was obtained as published earlier (Kikuchi et al, 2003;Taniwaki et al, 2006). This study as well as the use of all clinical materials described above was approved by individual institutional Ethical Committees (Ishikawa et al, 2004(Ishikawa et al, , 2007. Histological classification of tumours was performed based on the WHO criteria. All tumours were staged based on the pathological tumour-node-metastasis classification of the International Union Against Cancer. Histopathological examination of resected tumours revealed that 192 cases were diagnosed as ADC, and the other 130 cases were classified as non-adenocarcinoma (non-ADC).

RNA extraction and qRT -PCRs
Total RNA was extracted from frozen clinical tissues using the TRIzol reagent (Life Technologies, Inc., Gaithersburg, MD, USA) according to the manufacturer's protocol. Extracted RNAs were treated with DNase I (Nippon Gene) and reversely transcribed using oligo(dT) primer and SuperScript II (Invitrogen, Tokyo, Japan). For real-time TaqMan qRT -PCRs, specific primers and probes, which strictly amplify only cDNA and not genomic DNA, for human UHRF1 and b2-microglobulin were purchased from Applied Biosystems (Carlsbad, CA, USA; ID: Hs00273589_m1 and 4333766F, respectively). The PCRs were performed using the ABI Prism 7700 Sequence Detection System (Applied Biosystems) following the manufacturer's protocol. Amplification conditions were 2 min at 501C, 10 min at 951C, and then 40 cycles each consisting of 15 s at 951C and 1 min at 601C. The CT value obtained by UHRF1 amplification was compared among the samples after normalisation using b2-microglobulin expression levels as an endogenous control.

siRNA experiments
The SBC-5 cells, whose origin was SCLC, were obtained from Japanese Collection of Research Bioresources (Osaka, Japan). The cell line was grown in a monolayer in Eagle's Minimum Essential Medium supplemented with 10% foetal bovine serum, penicillin/streptomycin, and glutamine, at 371C in 5% CO 2 . siRNA oligonucleotide duplexes were purchased from SIGMA Genosys (Sigma Aldrich Japan, Tokyo, Japan) for targeting the human UHRF1 transcript or the EGFP and FFluc transcripts. The siRNA targeting sequences (sense strand) are as follows: UHRF1, 5 0 -CUGC UUUGCUCCCAUCAAU-3 0 ; EGFP, 5 0 -GCAGCACGACUUCUUCAA GTT-3 0 ; FFluc, 5 0 -GUGCGCUGCUGGUGCCAACTT-3 0 . The SBC-5 cells were transfected with EGFP siRNA, FFluc siRNA, or the two different UHRF1 siRNAs at a final concentration of 20 nM using Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer's protocol. The siRNA transfection efficiency under the condition was 100%, as evaluated by a fluorescent siRNA (data not shown). Then the cells were harvested after 48 h of transfection for western blotting analysis. Western blotting was performed using anti-UHRF1 mouse monoclonal antibody (1 : 1000, BD Bioscience, Tokyo, Japan) or anti-b-actin mouse monoclonal AC-15 (Sigma Aldrich Japan).

Tissue microarray construction
Lung cancer tissue microarrays were constructed as published earlier using formalin-fixed lung cancer tissues (Ishikawa et al, 2004;Furukawa et al, 2005;Kato et al, 2005). Tissue areas for sampling were selected based on visual alignment with the corresponding haematoxylin and eosin-stained sections on slides. Three to five tissue cores (diameter ¼ 0.6 mm; height ¼ 3-4 mm), which were taken from donor tumour blocks, were placed into recipient paraffin blocks using a tissue microarrayer (Beecher Instruments, Sun Prairie, WI, USA). A core of normal tissue area was punched from each case. A total of 5-mm sections of the resulting microarray block were used for immunohistochemical analysis.

Immunohistochemical staining analysis
The expression patterns of UHRF1 in lung cancer and normal human lungs were examined by immunohistochemistry as described earlier (Unoki et al, 2004(Unoki et al, , 2009  patients. Tumour-specific survival curves were calculated from the date of surgery to the time of death related to lung cancer, or to the last follow-up observation. The Kaplan -Meier method was applied to generate the survival curves. Survival differences were analysed with the log-rank test based on the status of UHRF1 expression. This analysis was performed using StatView (version 5.0; SAS Institute, Inc., Cary, NC, USA).

UHRF1 was highly expressed in the US lung cancer cases
We screened in-house cDNA microarray database (Kikuchi et al, 2003;Taniwaki et al, 2006), and found that UHRF1 mRNA was overexpressed in 67% of NSCLCs and in 93% of SCLCs compared with their adjacent normal lungs. To validate the microarray data, we examined UHRF1 protein expression levels in the 56 US lung cancer cases by immunohistochemistry with information of age, gender, histological type, and pT and pN factors of their cancers (Table 1). First we evaluated specificity of an anti-UHRF1 antibody by western blotting using cellular lysate from SBC5 cells transfected with two control siRNAs and two UHRF1 siRNAs. The result revealed that the antibody specifically recognises endogenous UHRF1 ( Figure 1A). Using the antibody, we performed immunohistochemistric analysis. The analysis revealed that UHRF1 was not expressed in adjacent normal lungs, stromal cells, and invaded inflammatory cells, but was specifically expressed in the nuclei of cancer cells ( Figure 1B -D). Normal mouse IgG served as a negative control of primary antibody in each case, and no staining was observed (data not shown). We scored the staining levels of UHRF1 as high or low. The UHRF1 was overexpressed in 66% of the overall NSCLCs (Table 1). Interestingly, although expression of UHRF1 was detected in almost all histological types of the lung cancers, its expression was significantly higher in non-ADCs (Table 1); 84% of non-ADCs showed high expression of UHRF1, whereas 32% of ADCs were overexpressed UHRF1 (P ¼ 9 Â 10 À5 , w 2 test). Other clinicopathlogical factors did not show any correlation with UHRF1 expression levels in the US samples. Importantly, UHRF1 expression was observed in 50% of lung cancer cases in the early stage (T0 þ T1) ( Table 1).
Overexpression of UHRF1 was observed in Japanese lung cancer cases and associated with histological type, smoking, and gender As the number of samples of the US cases was relatively small, we analysed an additional 322 Japanese lung cancer cases with information of postoperative clinical course, pT and pN factors of their cancers, and smoking habit, besides information of age, gender, and histological type (Table 2). We performed immunohistochemical analysis and scored the staining levels of UHRF1 as high or low similarly to the US cases (Figure 2A). Overexpression of UHRF1 was observed in 60% of the overall NSCLC cases (Table 2). A w 2 test revealed that UHRF1 overexpression was strongly associated with histological type of the Japanese cases (higher in non-ADC; P ¼ 4.2 Â 10 À11 ). The UHRF1 high expression was also associated with smoking (higher in smokers; P ¼ 0.00004 by w 2 test) and gender (higher in males; P ¼ 0.00082 by w 2 test) ( Table 2). Among these three factors, the histological type showed the strongest correlation with overexpression of UHRF1 (P ¼ 4.2 Â 10 À11 ). We also detected UHRF1 mRNA in Japanese cases by quantitative TaqMan PCR and found that UHRF1 mRNA was up-regulated in the overall lung cancers, especially in non-ADC, as the same as UHRF1 protein detected by immunohistochemistry ( Figure 2B). At mRNA level, overexpression of UHRF1 was observed in the overall NSCLS cases even in ADC; UHRF1 mRNA levels were increased in 60% of ADC and in 77% of non-ADC.  (Table 2); UHRF1 high expression was significantly correlated with pT factor (higher in advanced stage; P ¼ 0.00010 by w 2 test) and pN factor (higher in cancers with metastasis in regional lymph nodes; P ¼ 0.00018 by w 2 test). Although overexpression of UHRF1 associated with lung cancer malignancy, approximately half (47%) of Japanese lung cancer patients in T1 stage showed high expression of UHRF1 similarly to the US cases (Table 2). Kaplan -Meier analysis revealed that the high expression of UHRF1 was associated with poor prognosis (P ¼ 0.0364 by log-rank test) ( Figure 2C).

DISCUSSION
Lung cancer is the most common cancer in terms of both incidence and mortality worldwide (1.35 million new cases per year and 1.18 million deaths) according to Cancer Research UK. Survival rate of the cancer is still poor, partially because more than two-thirds of lung cancers are diagnosed at a late stage. Therefore, diagnostic markers of lung cancer, which can detect early stage lung cancer, are required to improve the current situation. Sensitivity and/or specificity of current diagnostic markers still need to be improved (Ando et al, 2003;Molina et al, 2005). We have been seeking superior diagnostic markers and identified a couple dozen of candidates (Daigo and Nakamura, 2008). However, we are still trying to identify other diagnostic markers to increase accuracy of current diagnosis for responding to individual lung cancers, which are caused by different genetic and epigenetic alterations.
In this report, we performed, for the first time, a large-scale analysis of UHRF1 expression in lung cancer cases with clinical information. We observed overexpression of UHRF1 in lung cancer cases, especially in non-ADC cases, regardless of ethnic groups, indicating that frequent overexpression of UHRF1 in non-ADC is common worldwide. Although UHRF1 is fundamentally important for cell proliferation, UHRF1 'low-expression' population exists. The qRT -PCR, which is more sensitive than immunohistochemistry, revealed that UHRF1 mRNA level was up-regulated in approximately 70% of the overall lung cancer cases. In 30% of the rest of lung cancers, other oncogenic pathways may be predominantly activated. We further found that overexpression of UHRF1 in Japanese lung cancer cases associated with gender (higher in male) and smoking. The major histological type of non-ADCs is SCC, which is more common in males (Toyoda et al, 2008), and smoking is the biggest risk factor for SCC (Wakai et al, 2006). In Japan, the smoking ratio for males is approximately four-fold higher than that of females. Thus, it is reasonable that UHRF1 overexpression associated with these three factors, the histological type, gender, and smoking, although further analysis is required for clarifying biological functions of UHRF1 in carcinogenesis of non-ADCs. In the United States, prevalence of smoking was also higher among males than females (23. 9% and 18%, respectively, in 2006) according to the Centers for Disease Control and prevention (http://www.cdc.gov/), but the difference between the genders in the United States is smaller than that in Japan (1.3-fold vs 4-fold). We think that this fact reflected our result that correlation between gender and UHRF1 expression levels in the US cases was not significant.
Overexpression of UHRF1 also associated with cancer malignancy indicated by pT and pN factors and survival rate in Japanese cases. No correlation between UHRF1 expression and pT-pN factors was observed in the US cases because of a lack of statistical power. However, we would like to emphasise that overexpressed UHRF1 was detectable in approximately half of lung cancer patients in an early pathological stage by immunohistochemistry, indicating that UHRF1 can be a diagnostic marker of lung cancer even in the early stage. Combinations of UHRF1 and several other lung cancer biomarkers including ones originally identified by our group (Daigo and Nakamura, 2008) may yield a better outcome. In conclusion, our data indicate that immunohistochemical staining of UHRF1 may ameliorate the sensitivity and specificity of current sputum examination for cancer detection, and/or to score the malignant potential of lung cancer using biopsy/resected tissue specimens accurately.