Antitumour effect of cyclin-dependent kinase inhibitors (p16INK4A, p18INK4C, p19INK4D, p21WAF1/CIP1 and p27KIP1) on malignant glioma cells

Cyclin-dependent kinase inhibitors (CDKIs) are considered as novel anticancer agents because of their ability to induce growth arrest or apoptosis in tumour cells. It has not yet been fully determined, however, which CDKI is the best candidate for the treatment of malignant gliomas and whether normal brain tissues are affected by CDKI expression. Using recombinant adenoviral vectors that express CDKIs (p16INK4A, p18INK4C, p19INK4D, p21WAF1/CIP1 and p27KIP1), we compared the antitumour effect of CDKIs on malignant glioma cell lines (A172, GB-1, T98G, U87-MG, U251-MG and U373-MG). p27KIP1 showed higher ability to suppress the growth of all tumour cells tested than other CDKIs. Interestingly, overexpression of p27KIP1 induced autophagic cell death, but not apoptosis in tumour cells. On the other hand, p27KIP1 overexpression did not inhibit the viability of cultured astrocytes (RNB) nor induced autophagy. Overall, our findings suggest that gene transfer of p27KIP1 may be a promising approach for the therapy of malignant gliomas.

Recent investigations show that exogenous induction of CDKIs induces growth arrest or apoptosis in tumour cells, indicating the potential use of CDKIs as a therapeutic tool (Guan et al, 1994;Craig et al, 1997;Hiromura et al, 1999). However, it has not been fully investigated which CDKI is the most useful for cancer therapy and whether normal tissues are affected by CDKI expression. To address these issues, in the present study, we compared the antitumour effect of CDKIs on malignant glioma cell lines and cultured astrocytes by using recombinant adenoviral vectors that express CDKIs (p16 INK4A , p18 INK4C , p19 INK4D , p21 WAF1/CIP1 and p27 KIP1 ).

Construction of recombinant adenoviral vectors
The recombinant AdMH4p16, AdMHp18, AdMH4p19, AdMH4p21 and AdMH4p27 containing p16 INK4A , p18 INK4C , p19 INK4D , p21 WAF1/CIP1 and p27 KIP1 were kindly supplied from Dr FL Graham (McMaster University, Ontario, Canada). As described previously (Schreiber et al, 1999), human p16 cDNA (pBluescriptSK-p16) was obtained from Dr D Beach (Cold Spring Harbor, New York, USA). pxep21, pAdcp17 and pAdcp19 were obtained from Dr T Thompson (Baylor College of Medicine, Houston, TX, USA), and pSCZhuwtp27 was a gift from Dr J Roberts (Fred Hutchison Cancer Research Center, Seattle, WA, USA). AdBHGD1,3 containing El and E3 deletions is a control recombinant adenovirus that has identical backbone sequences to the adenoviral constructs expressing CDKIs. et al, 2000). To achieve the infectivity of 490%, A172, GB-1, U87-MG, U251-MG, U373-MG and RNB cells were tested at a multiplicity of infection (MOI) of 60 PFU cell À1 . On the other hand, T98G cells were infected with 180 MOI. The percentage of cell viability was calculated from the mean cell viability of treated cells divided by that of cells with control treatment. To detect the expression of CDKI in infected cells, the immunoblotting assays using anti-CDKI antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) were performed as described previously (Kondo et al, 1996).

Detection of apoptotic or autophagic cell death
To detect apoptosis, the terminal deoxynucleotidyl transferase (Tdt)-mediated nick end labelling (TUNEL) analysis was performed as described previously (Komata et al, 2000). To detect autophagic changes in infected cells, cells were stained with acridine orange (Poly-sciences, Warrington, PA, USA) as described previously (Paglin et al, 2001). At 2 days after adenoviral infection, acridine orange was added at a final concentration of 1 mg ml À1 for 15 min. Microphotographs were obtained with a fluorescence microscope.

Statistical analysis
The data were expressed as means 7s.d. Statistical analysis was performed by using Student's t-test (two-tailed). The criterion for statistical significance was taken as Po0.05.

Effect of CDKI overexpression on viability of malignant glioma cells
To investigate the effect of CDKI on malignant glioma cell lines, cell viability was determined 3 or 5 days after adenoviral infection. As shown in Figure 1, the treatment of U373-MG cells with AdMH4p16 INK4A , AdMH4p18 INK4C , AdMH4p19 INK4D , AdMH4p21 WAF1/CIP1 or AdMH4p27 KIP1 for 3 days significantly inhibited the cell viability compared to that with control vector (Po0.002 to Po0.004). In U251-MG, U87-MG, A172 and GB-1, similar antitumour effects were observed with each CDKI 3 days after adenoviral infection (Po0.0002 to Po0.02). On the other hand, the effect of AdMH4p18 INK4C or AdMH4p19 INK4D was not significant for T98G cells (P ¼ 0.17 or P ¼ 0.71), although other CDKIs were effective (Po0.0001 to Po0.03). Among the CDKIs used in the present study, p27 KIP1 was more effective for tumour cells than the other CDKIs. The number of viable cells of U373-MG treated with AdMH4p27 KIP1 decreased below the initial cell number (5000): 413871794 (day 3) or 255571042 (day 5). This indicates that approximately half of p27 KIP1 -infected U373-MG cells underwent cell death by day 5. Additionally, cell number of other tumour cell lines decreased by 30 -60% from the initial cell number like U373-MG cells 5 days after the treatment with AdMH4p27 KIP1 . These results indicate that 27 KIP1 has the highest antitumour effect on all malignant glioma cells tested in the present study.

Effect of p27 KIP1 in RNB cells
To investigate whether normal brain tissues are affected by p27 KIP1 expression, cultured rat astrocyte, RNB cells, were treated with AdMH4p27 KIP1 . As shown in Figure 2, the effect of p27 KIP1 overexpression was not significant for RNB cells, while the viability of U373-MG cells decreased to 15% of the control (Po0.005). This result suggested that p27 KIP1 -based therapy will be selective for tumour cells.

Effect of p27 KIP1 on induction of apoptosis or autophagy in U373-MG cells
To determine the type of cell death induced by AdMH4p27 KIP1 , we first performed the TUNEL staining. The incidence of TUNELpositive cells in U373-MG cells treated with AdBHGD1,3 or AdMH4p27 KIP1 for 3 or 5 days was less than 1%. Next, we investigated the changes in the cellular acidic compartments to detect the occurrence of autophagic cell death. Vital staining of U373-MG cells with acridine orange revealed the appearance of acidic vesicular organelles (AVO) 3 days after AdMH4p27 KIP1 infection (Figure 3). The staining of p27 KIP1 -infected cells clearly showed punctuate acidic vesicles that were diffusely distributed in cytoplasm. In contrast, there was no change of fluorescent signals in RNB cells treated with AdBHGD1,3 or AdMH4p27 KIP1 . These results indicated that the antitumour effect of p27 KIP1 on malignant glioma cells was because of autophagic cell death.

DISCUSSION
In the present study, we have demonstrated that p27 KIPl shows a greater antitumour effect than the other CDKIs (p16 INK4A , p18 INK4C , p19 INK4D or p21 WAF1/CIP1 ) against six human malignant glioma cell lines. p27 KIPl -infected tumour cells undergo autophagy, but not apoptotic cell death, while overexpression of p27 KIP1 does not inhibit viability of cultured astrocytes. Our findings provide new insights into the potential use of p27 KIP1 as a novel therapeutic tool. p27 KIP1 plays a central role in the negative control of cell growth. p27 KIPl is generally expressed at high levels in cells arrested by treatment with transforming growth factor-b, contact inhibition or serum deprivation (Koff et al, 1993;Polyak et al, 1994). In contrast, p27 KIP1 declines in the presence of mitogenic growth factor signalling or interleukin-2 (Nourse et al, 1994;Cheng et al, 1998). p27 KIP1 -deficient mice develop a variety of abnormalities including multiorgan hyperplasia and pituitary tumours (Nakayama et al, 1996). Furthermore, the higher the levels of p27 KIP1 expression, the better the prognosis with regard to human malignant gliomas (Alleyne et al, 1999), breast cancer (Porter et al, 1997) or lung cancer (Esposito et al, 1997). Therefore, introduction of p27 KIP1 gene into tumour cells is expected to be a promising strategy to inhibit their malignant cellular proliferation.
It has been controversial whether p27 KIP1 expression leads tumour cells to growth arrest or cell death. Some groups demonstrate the induction of growth arrest by p27 KIPl (Sherr and Roberts, 1995;Craig et al, 1997). Others show that p27 KIP1 expression induces apoptosis in several cell lines Schreiber et al, 1999), while p27 KIPl protects cells from apoptosis (Hiromura et al, 1999). In the present study, p27 KIP1 induced autophagic cell death, but not apoptosis in malignant glioma cells. Recently, several groups have proposed two types of programmed cell death (Schwartz et al, 1993;Bursch et al, 2000). Type I programmed cell death, or apoptosis, is mediated by caspases/bcl family, and has typical morphological and biochem-ical characteristics such as chromatin condensation or nucleosomal ladder formation. Since TUNEL-positive cells were not detected in p27 KIP1 -infected U373-MG cells in the present study, it is unlikely that apoptosis is involved in the antitumour effect of p27 KIP1 on malignant glioma cells. In contrast, type II programmed cell death is marked morphologically by increased autophagy and early destruction of the cytoplasm that occur either without nuclear collapse or precedes it (Schwartz et al, 1993;Bursch et al, 2000). More recently, Paglin et al (2001) demonstrated that formation of acidic vesicles was detected in radiation-induced autophagy. Since the development of AVO was detected in p27 KIP1infected U373-MG cells, we suggest that the antitumour effect of p27 KIP1 on malignant glioma cells is mainly because of induction of type II programmed cell death, autophagy. Further study is necessary to investigate the molecular mechanisms underlying p27 KIP1 -induced autophagy in malignant glioma cells.
In summary, p27 KIP1 shows the most potent antitumour effect against malignant glioma cells, while cultured astrocytes are insensitive to p27 KIPl expression. The effect is because of autophagic cell death as well as G0/G1 growth arrest. Therefore, we expect that p27 KIP1 -based therapy for malignant gliomas might be a promising approach that is worth exploring further.

ACKNOWLEDGEMENTS
We thank Dr Frank L Graham for the recombinant adenoviruses (AdBHGD1,3, AdMH4p16 INK4A , AdMH4p18 INK4C , AdMH4pl9 INK4D , AdMH4p21 WAF1/CIPl and AdMH4p27 KIP1 ). This study was supported in part by the USPHS Grants 1R01CA80233 and IR0l CA88936 awarded by the National Cancer Institute, in part by a start-up fund from The University of Texas M.D. Anderson Cancer Center, and by a generous donation from the Anthony D Bullock III Foundation (SK).