The clinical significance of splice variants and subcellular localisation of survivin in non-small cell lung cancers

Survivin is a member of the inhibitor of apoptosis protein family. Survivin has splice variants with different biological functions associated with tumorigenesis. We investigated 134 non-small cell lung cancers (NSCLCs) to study the clinical significance of wild-type survivin, survivin-2B, and survivin-deltaEx3. Real-time PCR analyses were performed for their gene expressions. The subcellular localisation of survivin proteins was evaluated by immunohistochemistry. The Ki-67 proliferation index and the apoptotic index were also evaluated. The survivin-deltaEx3 gene expression was significantly higher in stage II–III than in stage I (P=0.0174), and significantly correlated with the nuclear pan-survivin expression (P<0.0001). The Ki-67 index was significantly higher in wild-type survivin-positive tumours (P<0.0001), survivin-deltaEx3-positive tumours (P<0.0001), and tumours with positive expression of the nuclear pan-survivin (P=0.0047). In contrast, the apoptotic index was significantly lower only in wild-type survivin-positive tumours (P<0.0001). Thus, the wild-type survivin gene expression was associated with apoptotic inhibition and tumour proliferation. Furthermore, the survivin-deltaEx3 gene expression was strongly associated with tumour proliferation, especially in advanced stage NSCLCs. In contrast, the survivin-2B gene expression did not correlate with tumour proliferation or tumour apoptosis.

synthesis was performed with 5 mg of total RNA using a cDNA synthesis kit (Pharmacia, Piscataway, NJ, USA). Because real-time PCR (RT -PCR) is not an appropriate method to discriminate between wild-type survivin and other splice variants, we carried out quantitative RT -PCR assays using densitometric analyses of gel electrophoresis to evaluate the gene expressions of wild-type survivin, survivin-2B, and survivin-deltaEx3. For the gene expressions of wild-type survivin, survivin-2B, and survivin-deltaEx3, oligonucleotides were synthesised as PCR primers based on the published reports (Yamada et al, 2003). The nucleotide sequence 5 0 -CCACCGCATCTCTACATTCA-3 0 was used as the sense primer of wild-type survivin and survivin-deltaEx3, 5 0 -GAGGCTGGCTT CATCCACTG-3 0 was used as the sense primer of survivin-2B, 5 0 -TATGTTCCTCTATGGGGTCG-3 0 as the antisense primer of wild-type survivin, 5 0 -GTTCCTCTCTCGTGATCCG-3 0 as the antisense primer of survivin-2B, and 5 0 -TTTCCTTTGCATGGGGTC-3 0 as the antisense primer of survivin-deltaEx3. All subsequent assays were carried out using the parameters that yielded amplification of wild-type survivin, survivin-2B, survivin-deltaEx3, and b-actin DNA (the internal control) within a linear range. The reaction mixture of wild-type survivin or survivin-2B was subjected to 36 PCR amplification cycles of 60 s at 941C, 60 s at 601C, and 90 s at 721C, and that of survivin-deltaEx3 was subjected to 36 PCR amplification cycles of 60 s at 941C, 60 s at 551C, and 90 s at 721C, and that of b-actin was subjected to 31 PCR amplification cycles of 60 s at 941C, 60 s at 601C, and 90 s at 721C. Preparations of a human adenocarcinoma cell line, A549, were used as positive controls for gene expressions of wild-type survivin, survivin-2B, and survivin-deltaEx3. The amplified DNA samples were run on a 1% agarose gel, and the bands were visualised with ethidium bromide. The densitometric value obtained for a wild-type survivin, survivin-2B, and survivin-deltaEx3 band in a given tumour sample was divided by the value of the b-actin, and the resultant ratio was referred to as the gene expression ratio. Thereafter, the expression ratio for a given tumour sample was divided by the expression ratio of A549 to obtain the standardised gene expression ratio.

Immunohistochemistry
We used a mouse monoclonal antibody against pan-survivin (sc17779; Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) diluted at 1 : 50, and a mouse monoclonal antibody for the Ki-67 antigen (MIB-1; DAKO, Glostrup, Denmark) diluted at 1 : 40. Formalin-fixed paraffin-embedded tissue was cut into 4-mm sections and mounted on poly-L-lysine-coated slides. Sections were deparaffinised and rehydrated. The slides were then heated in a microwave for 10 min in a 10 mmol l À1 citrate buffer solution at pH 6.0. After quenching the endogenous peroxidase activity with 0.3% H 2 O 2 (in absolute methanol) for 30 min, the sections were treated for 2 h with 5% bovine serum albumin to block nonspecific staining. Duplicated sections were incubated overnight with primary antibodies, respectively. Slides were then incubated for 1 h with biotinylated secondary antibodies (Vector Laboratories Inc., Burlingame, CA, USA). The sections were incubated with the avidin -biotin -peroxidase complex (Vector) for 1 h, and antibody binding was visualised with 3,3 0 -diaminobenzidine tetrahydrochloride. Finally, the sections were lightly counterstained with Mayer's haematoxylin.
All immunostained sections were independently evaluated by two authors (JN and MU), without knowledge of the patients' characteristics. At least 200 tumour cells were scored per Â 40 field. Regarding pan-survivin expression, the nuclear staining and the cytoplasmic staining in each section were evaluated, respectively, in a semiquantitative manner that reflected both the intensity and percentage of cells staining at each intensity. The intensity was classified as 0 (no staining), þ 1 (weak staining), þ 2 (distinct staining), or þ 3 (very strong staining). A value designated as the 'HSCORE' thus was obtained for each slide by using the following algorithm: HSCORE ¼ S (I Â PC), where I and PC represent the intensity and the percentage of cells that stain at each intensity, respectively. The percentage of carcinoma cells with positive staining for Ki-67 antigen was scored as the Ki-67 proliferation index.

Detection of apoptosis
The presence of apoptotic cells was detected with the TUNEL method using the In Situ Apoptosis Detection Kit (Takara Biomedicals, Otsu, Japan). After sections were deparaffinised and rehydrated, the slides were treated for 15 min with 20 mg ml À1 Proteinase K. After quenching the endogenous peroxidase activity with 3% H 2 O 2 for 5 min, the sections were then incubated for 90 min at 371C with the TUNEL reaction mixture, including terminal deoxynucleotidyl transferase (TdT). Next, the sections were incubated for 30 min at 371C with anti-FITC horseradish peroxidase conjugate. Staining was developed using 3,3 0 -diaminobenzidine tetrahydrochloride for 15 min. Finally, the sections were lightly counterstained with Mayer's haematoxylin. Sections incubated with the TUNEL reaction mixture without TdT were used as negative control slides. Apoptotic cells were determined based on observations of TUNEL-staining sections and serial HE-staining sections. Cells staining positive for TUNEL, if they represented the histologic features of necrosis in HE-staining sections, were not considered to be apoptotic cells. In each case, a total of 10 000 tumour cells were evaluated at high magnification by two authors (JN and MU) independently, without knowledge of the patients' characteristics. The apoptotic index was defined as the number of apoptotic cells per 1000 tumour cells.

Statistical analysis
The ratio of the Ki-67 proliferation index divided by the apoptotic index was referred to as the growth index for each sample (Tsoli et al, 2002). The statistical significances of wild-type survivin, survivin-2B, and survivin-deltaEx3 gene expression in relation to several clinical and pathological parameters were assessed by t-test or an analysis of variance with Bonferroni/Dunn test. Because the wild-type survivin expression cutoff line of 1.0 demonstrated the most significance in relation to the Ki-67 proliferation index and the apoptotic index, the sample was classified as a wild-type survivin-positive tumour when the standardised wild-type survivin gene expression was 41.0. Because the survivin-deltaEx3 expression cutoff line of 0.5 demonstrated the most significance in relation to the Ki-67 proliferation index, the sample was classified as a survivin-deltaEx3-positive tumour when the standardised survivin-deltaEx3 gene expression was 40.5. The sample was classified as a survivin-2B-positive tumour when the standardised survivin-2B expression was 41.0 (a median value of survivin-2B gene expression). Samples were considered to have a positive expression of nuclear pan-survivin if the HSCORE of the nuclear staining of pan-survivin was more than 10, a positive expression of cytoplasmic pan-survivin if the HSCORE of the cytoplasmic staining of pan-survivin was more than 50. All P-values were based on two-tailed statistical analysis and a P-value of o0.05 was considered to indicate statistical significance.

Wild-type survivin gene expression in NSCLCs
We studied the wild-type survivin gene expression in eight noncancerous samples to evaluate its expression in normal lung tissue. The standardised wild-type survivin gene expression ratio was low in normal lung tissue specimens (mean 0.363±0.135). In contrast, the standardised survivin gene expression ratio varied greatly among the 134 tumour tissue specimens (mean 1.531 ± 0.795, Figure 1A). The standardised wild-type survivin gene expression ratio was significantly higher in tumour tissue specimens than in normal lung tissue specimens (Po0.0001, Figure 2A). Seventy-two carcinomas (53.7%) were wild-type survivin-positive tumours. Regarding tumour histology, the wildtype survivin gene expression was significantly higher in squamous cell carcinomas than in adenocarcinomas (1.690±0.892 vs 1.362 ± 0.664, P ¼ 0.0181, Table 1). In addition, the wild-type survivin gene expression was significantly higher in moderately or poorly differentiated tumours than in well-differentiated tumours (1.614 ± 0.807 vs 1.288 ± 0.712, P ¼ 0.0383).

Survivin-2B gene expression in NSCLCs
The standardised survivin-2B gene expression ratio was low in eight normal lung tissue specimens (mean 0.496±0.246). In contrast, the standardised survivin-2B gene expression ratio also varied greatly among the tumour tissue specimens (mean 1.233±0.830, Figure 1B). The standardised survivin-2B gene expression ratio was significantly higher in tumour tissues than in normal lung tissues (P ¼ 0.0138, Figure 2B). However, no difference was observed in the survivin-2B gene expression regarding tumour histology, tumour differentiation, or any other clinicopathological parameter (Table 1).

Survivin-deltaEx3 gene expression in NSCLCs
The standardised survivin-deltaEx3 gene expression ratio was low in eight normal lung tissue specimens (mean 0.296±0.098). In contrast, the standardised survivin-deltaEx3 gene expression ratio varied greatly among the tumour tissue specimens (mean 0.971 ± 0.610, Figure 1C). The standardised survivin-deltaEx3 gene expression ratio was significantly higher in tumour tissues than in normal lung tissues (P ¼ 0.0022, Figure 2C). One hundred and one carcinomas (75.4%) were survivin-deltaEx3-positive tumours. Regarding tumour histology, the survivin-deltaEx3 gene expression was also significantly higher in squamous cell carcinomas than in adenocarcinomas (1.075 ± 0.616 vs 0.854 ± 0.587, P ¼ 0.0377, Table 1). Regarding tumour differentiation, the survivin-deltaEx3 gene expression was significantly higher in moderately or poorly differentiated tumours than in welldifferentiated tumours (1.049 ± 0.619 vs 0.743 ± 0.526, P ¼ 0.0110). Regarding the clinicopathological characteristics, the survivin-deltaEx3 gene expression was significantly higher in stage II -III NSCLCs than in stage I NSCLCs (1.084 ± 0.661 vs 0.833 ± 0.512, Subcellular localisation of the survivin protein expression in relation to wild-type survivin, survivin-2B, and survivin-deltaEx3 gene expressions. Immunostaining using the antibody against pan-survivin showed various patterns of cytoplasmic staining and nuclear staining ( Figure 3A    cytoplasmic staining, 98 carcinomas (73.1%) had a positive expression of cytoplasmic pan-survivin. However, the HSCORE of the cytoplasmic pan-survivin did not correlate with the gene expression of wild-type survivin, survivin-2B, or survivin-deltaEx3.
On the other hand, regarding the nuclear staining, 88 carcinomas (65.7%) had a positive expression of nuclear pansurvivin. Furthermore, the standardised survivin-deltaEx3 gene expression significantly correlated with the HSCORE for the nuclear staining of pan-survivin (r ¼ 0.642, Po0.0001). However, the HSCORE of the nuclear pan-survivin did not correlate with the gene expression of wild-type survivin or survivin-2B.

Ki-67 proliferation index in NSCLCs
The Ki-67 proliferation index ranged from 2.0 to 95.0 (mean 47.6 ± 27.3, Figure 3B and D). Regarding the gene expressions of splice variants of survivin, the Ki-67 proliferation index was significantly higher in the wild-type survivin-positive tumours than in the wild-type survivin-negative tumours (57.4 ± 22.0 vs 36.2 ± 28.5, Po0.0001, Figure 4A, Table 2). However, there was no difference in the Ki-67 proliferation index according to the survivin-2B gene expression ( Figure 4B). In contrast, the Ki-67 proliferation index was also significantly higher in the survivin-deltaEx3-positive tumours than in the survivin-deltaEx3-negative tumours (53.4±25.5 vs 29.8±25.0, Po0.0001, Figure 4C). Regarding the subcellular localisation of pan-survivin protein, the Ki-67 proliferation index was significantly higher in tumours with a positive expression of nuclear pan-survivin than in tumours with a negative expression of nuclear pan-survivin (52.4 ± 26.5 vs 38.5±26.6, P ¼ 0.0047, Figure 4D). In contrast, there was no difference in the Ki-67 proliferation index according to the cytoplasmic pan-survivin expression ( Figure 4E).

Apoptotic index in NSCLCs
The apoptotic index ranged from 10.0 to 40.0 (mean 20.7±7.4, Figure 3E). Regarding the gene expressions of splice variants of survivin, the apoptotic index was significantly lower in the wildtype survivin-positive tumours than in the wild-type survivinnegative tumours (18.1±5.6 vs 23.1±7.2, Po0.0001, Figure 5A). On the other hand, there was no difference in the apoptotic index according to the survivin-2B expression or survivin-deltaEx3 expression ( Figure 5B and C). Regarding the pan-survivin protein expression, neither the nuclear pan-survivin expression nor the cytoplasmic pan-survivin expression was associated with the apoptotic index ( Figure 5D and E).

Growth index in NSCLCs
The growth index ranged from 0.05 to 9.00 (mean 2.62 ± 1.79). The growth index was significantly higher in the wild-type survivinpositive tumours than in the wild-type survivin-negative tumours (3.47±1.68 vs 1.66±1.37, Po0.0001, Figure 6A). However, there was no difference in the growth index according to the survivin-2B gene expression ( Figure 6B). In contrast, the growth index was also significantly higher in the survivin-deltaEx3-positive tumours than in the survivin-deltaEx3-negative tumours (3.01 ± 1.76 vs 1.48±1.33, Po0.0001, Figure 6C). Regarding the subcellular localisation of pan-survivin protein, the growth index was significantly higher in tumours with a positive expression of  nuclear pan-survivin than in tumours with a negative expression of nuclear pan-survivin (3.02 ± 1.88 vs 1.90 ± 1.33, P ¼ 0.0004, Figure 6D). In contrast, there was no difference in the growth index according to the cytoplasmic pan-survivin expression ( Figure 6E).

DISCUSSION
Survivin is reported to be expressed in most human malignancies (Kawasaki et al, 1998;Monzo et al, 1999;Sui et al, 2002), while its expression is absent in normal differentiated tissues (Salvesen and Duckett, 2002). Although survivin was initially identified as a member of the IAP family (Ambrosini et al, 1997), recent studies have demonstrated that survivin not only inhibits the caspasedependent apoptotic pathway, but also accelerates cell proliferation (Honda et al, 2003). Survivin is known to have at least five spliced variants. These splice variants have different biological functions (Mahotka et al, 2002a). Among them, recent experimental studies demonstrated that survivin-deltaEx3 also inhibits apoptosis and induces cell proliferation, similar to wild-type survivin (Song and Wu, 2005;Takashima et al, 2005). In contrast, survivin-2a and survivin-2B are considered to attenuate the antiapoptotic function of wild-type survivin (Caldas et al, 2005b).
As a result, these splice variants are variably involved in the biological functions associated with carcinogenesis, including cell proliferation, apoptosis, and chemo-radio resistance. Previous clinical studies have reported the survivin expression to be associated with tumour cell proliferation, apoptotic inhibition, and a poor prognosis in various human cancers, including NSCLCs (Martinez et al, 2004;Shinohara et al, 2005). However, only a few clinical reports studied these splice variants of survivin in human cancers (Mahotka et al, 2002a;Meng et al, 2004;Ling et al, 2005;Taubert et al, 2005). Therefore, to clarify the clinical significance of these splice variants of survivin in NSCLCs, we performed the present study on the gene expressions of wild-type survivin, survivin-2B, and survivin-deltaEx3 in relation to the subcellular localisation of survivin proteins, tumour cell proliferation, and apoptosis.
In the present study, regarding the relationship of the expression of these splice variants of survivin, the survivin-deltaEx3 gene expression significantly correlated with the wild-type survivin gene expression (r ¼ 0.646, Po0.001). However, the correlation between the survivin-2B gene expression and the wild-type survivin gene expression was low (r ¼ 0.254). Regarding the biological significance of these splice variants of survivin, the present study also demonstrated the wild-type survivin gene expression to be associated with both tumour cell proliferation and apoptotic inhibition. In contrast, the survivin-deltaEx3 gene expression correlated with the nuclear pan-survivin expression. Furthermore, both the survivin-deltaEx3 gene expression and the nuclear pan-survivin expression were found to be associated with tumour cell proliferation, but not with apoptotic inhibition. However, the survivin-2B gene expression did not correlate with tumour cell proliferation or apoptotic inhibition. In total, the growth index was associated with both the wild-type survivin expression and the survivin-deltaEx3 expression. As a result, the present study revealed wild-type survivin to be associated with both tumour cell proliferation and apoptotic inhibition, while survivin-deltaEx3 was strongly associated with tumour cell proliferation. In fact, previous clinical studies also revealed the survivin-deltaEx3 expression to be associated with a high proliferation rate (Takashima et al, 2005). Recent studies have revealed that the survivin-deltaEx3 overexpression is a significant poor prognostic factor in cancer patients, including NSCLCs (Ling et al, 2005;Taubert et al, 2005). Furthermore, the present study found the survivin-deltaEx3 gene expression to be significantly higher in advanced-stage NSCLCs. Therefore, the survivin-deltaEx3 gene overexpression could produce more aggressive tumours with a high proliferation rate in advanced-stage NSCLCs. Because the biomarkers associated with tumour proliferation are significant factors in advanced-stage NSCLCs (Huang et al, 2005), the level of the survivin-deltaEx3 gene expression could thus affect the clinical impact of NSCLCs, especially in the advanced stages. A recent study on soft tissue sarcoma also reported the overexpression of survivin-deltaEx3 gene to correlate with an advanced clinical stage and a poor prognosis (Taubert et al, 2005).
On the other hand, previous immunohistochemical studies revealed the nuclear expression of survivin proteins also to be associated with tumour cell proliferation and a poor prognosis in cancer patients, as was also shown in the present study, while the cytoplasmic expression of survivin proteins was not (Martinez et al, 2004;Huang et al, 2005;Ling et al, 2005;Shinohara et al, 2005;Taubert et al, 2005). These studies were performed using antibodies against pan-survivin, which cannot distinguish wildtype survivin from other spliced variants. In general, wild-type survivin is considered to be located both in the cytoplasm for apoptotic inhibition, and in the nuclei for cell proliferation. In contrast, survivin-deltaEx3 is probably located in the nucleus because of its nuclear localisation signal (Mahotka et al, 2002b;Song and Wu, 2005). The present study also demonstrated the survivin-deltaEx3 gene expression to correlate with the nuclear pan-survivin expression. As a result of the above findings, the nuclear expression of pan-survivin could be the representation of both wild-type survivin and survivin-deltaEx3, both of which are associated with tumour proliferation. In contrast, the cytoplasmic expression of pan-survivin might indicate the presence of wildtype survivin and survivin-2B (Mahotka et al, 2002b). In fact, the present study revealed that the cytoplasmic pan-survivin expression was not associated with the apoptotic index. This result might be partly because cytoplasmic survivin-2B can inhibit the antiapoptotic function of wild-type survivin. We performed additional immunohistochemistry to clearly identify the subcellular localisation of survivin-deltaEx3 using a commercially available specific antibody against survivin-deltaEx3; however, we were not successful (data not shown).
As a result, the present study demonstrated the wild-type survivin gene expression to be associated with not only apoptotic inhibition but also tumour proliferation. In contrast, the survivin deltaEx3 gene expression significantly correlated with the nuclear expression of pan-survivin and the Ki-67 proliferation index. However, the survivin 2B gene expression did not correlate with tumour proliferation or tumour apoptosis. Considering these results, the combined evaluations of wild-type survivin and survivin-deltaEx3 gene expression may thus be a useful strategy for evaluating NSCLCs. The RNA inhibition of survivin can induce apoptosis in human tumour cell lines Olie et al, 2000). Therefore, the RNA inhibition designed against common regions of both the wild-type survivin and the survivin-deltaEx3 could be a more effective strategy for NSCLCs not only to improve the efficacy of chemo-radio therapy but also to suppress tumour cell proliferation (Li et al, 1998;Lu et al, 2004).