Inhibition of growth of human leukemia cell lines by ... - Nature

Leukemia (1997) 11, 1673–1680  1997 Stockton Press All rights reserved 0887-6924/97 $12.00

Inhibition of growth of human leukemia cell lines by retrovirally expressed wild-type p16INK4A AF Gombart, R Yang, MJ Campbell, JD Berman and HP Koeffler Burns and Allen Research Institute, Davis Bldg 5065, Cedars-Sinai Medical Center, Division of Hematology/Oncology, UCLA School of Medicine, Los Angeles, CA 90048, USA

Loss of the p16INK4A gene by homozygous deletions or point mutations is attributed to the development of many types of cancers including leukemia. T cell acute lymphoblastic leukemias (T-ALLs) and B-cell ALLs show a remarkable rate of 75 and 20% homozygous deletion of this gene, respectively. Restoration of p16 expression in p16-deficient solid tumor cell lines results in a dramatic reduction of growth and maligant phenotype. To test the hypothesis that p16INK4A suppresses the growth of p16-deficient leukemias, we utilized a retroviral system to restore wild-type (wt) or mutant p16 protein expression. We tested the efficacy of our system by expressing the wt or mutant p16 genes in the osteosarcoma cell line, U20S, which lacks p16 and retains functional retinoblastoma protein (pRb). The wt p16 protein formed complexes with both cyclin-dependent kinases (CDK) 4 and 6 and inhibited U20S growth by 30fold. The p16 mutants E120K and R144C formed complexes with CDK4 and CDK6 in cells and inhibited cell growth as effectively as wt p16 (20-fold) while the mutant proteins that did not complex with detectable levels of CDK4 or CDK6 only inhibited growth 0.25- and five-fold (G101W and D141, respectively) or not at all (H83Y and DA4). The COOH-terminal ‘tail’ of the wt p16 protein (amino acid residues 141–156), missing in mutant D141, enhanced the growth suppressive capability of p16. The amino acid substitutions in mutants G101W and H83Y not only disrupted CDK4 and CDK6 binding, but decreased the protein half-lives by two- and three-fold, respectively, compared to wt p16. The wt, but not mutant p16 genes, effectively inhibited the growth of T cell acute lymphoblastic (CEM) and myeloid leukemia (NB-4 and K562) cell lines that lacked the p16 gene, but retained functional pRb. Growth of the T-ALL cell line, HSB-2, which lacked both p16 and pRb, was not inhibited, indicating the growth suppression involved the pRb pathway. These results define regions critical for the function of p16 and demonstrate that restoration of wt p16 expression in p16-deficient leukemias significantly reverted their transformed phenotype and inhibited their growth. Keywords: p16INK4A; leukemia; growth suppression; mutations

Introduction Regulation of cell cycle progression through the restriction (R) point in the G1 phase requires the concerted effort of numerous cyclin-dependent kinases (CDKs), their respective cyclins and cyclin-dependent kinase inhibitors (CDKIs). Dysregulation of the R point plays a major role in tumorigenesis.1–3 The tumor suppressor pRb is an important regulator of the transition from the G1 to S phase.4 Various transcription factors that are required to activate S-phase genes are sequestered by underphosphorylated pRb (active). When pRb is phosphorylated (inactive) at the R point in the G1 phase, these factors are released and the cell progresses through the cell cycle.4 The CDK4 and CDK6 proteins are protein kinases that phosphorylate pRb during the G1 phase of the cell cycle. They interact with the D-type cyclins to form an active kinase com-

Correspondence: AF Gombart Received 11 February 1997; accepted 17 July 1997

plex that is regulated negatively by the INK4 family of CDKIs which include the p15, p16, p18 and p19 proteins.1,2 Constitutive inactivation of pRb can result from loss of the gene, increased CDK4 activation or increased D-type cyclin expression. This, in turn, would lead to loss of growth control and neoplasia. Amplification and overexpression of cyclin D1 and CDK4 are described in a variety of tumors.5 Alternatively, loss of CDKI function could lead to increased CDK4 or CDK6 activity. Evidence from studies on both cell lines and human tumors indicates that the CDK4 inhibitor, p16, functions as a tumor suppressor.1,5–11 This is particularly true for familial melanomas, pancreatic carcinomas, and T cell acute lymphoblastic leukemias (T-ALLs).12–17 The predominant mutation in T-ALLs is a homozygous deletion of the p16 gene occurring in up to 80% of all cases.15–22 This occurs in almost 20% of B cell ALL, about 5–10% of lymphomas,15,17,19,23,24 27% of adult T cell leukemias,25 and are extremely rare in chronic myeloid leukemias (CML),21 chronic lymphocytic leukemias (CLL)19 and acute myelogenous leukemias (AML);26 however, p16 mRNA expression was not detected in many AML patient samples. Hypermethylation of normally unmethylated CpG islands27 or mutations in the promoter region of the p16 gene could explain this loss of transcription. These studies implicate p16 as a tumor suppressor in leukemias; however, studies by our laboratory22,28 and others29 suggest that another tumor suppressor may be located on the same chromosomal arm (9p21) as p16.5,10 Therefore, it is important to establish that p16 functions as a tumor suppressor in leukemias. Although the predominant alteration is homozygous deletion of the p16 gene, frequently point mutations are reported in pancreatic carcinomas,14 familial melanomas,12,13 non-small cell lung cancers,30,31 and esophageal squamous cell carcinomas.31 The reported point mutations in the p16 gene lead to amino acid changes, frame shifts and nonsense codons.32 Using several in vitro assays, we previously examined four missense alleles of p16 derived from tumors and three deletion mutants lacking either portions of the COOHterminal tail unique to the p16 protein (D136 and D141) or the fourth ankyrin (DA4) repeat (Figure 1).33 Two of these missense point mutations, H83Y and G101W, functionally compromised p16 and the remaining two, E120K and R144C, had no effects in vitro. This suggested that E120K and R144C were the result of sequence polymorphisms. Of the three deletion mutants, D136 and D141 appeared to function normally, but deletion of the fourth ankyrin repeat in DA4 completely abrogated p16 function.33 This suggested that the COOH-terminal tail was not required for the p16 protein to bind CDK4, but the integrity of the ankyrin repeats was essential. To explore how these p16 gene mutations function in vivo in cells in culture and determine if p16 can inhibit leukemic cell growth, we utilized recombinant retroviruses containing either wild-type (wt) or mutant p16 genes and infected the

p16INK4A growth suppression of leukemias AF Gombart et al

1674

Figure 1 Schematic diagram summarizing the wild-type and mutant forms of p16 protein examined in this study. The conserved ankyrin repeats (Ank1–Ank4) of p16 wt are located between amino acid residues 12 and 141 and are represented by the shaded boxes. The location of each point mutation is indicated below the p16 wt panel. The COOH-terminal tail (amino acid residues 141–156) present in p16 wt is absent in p15 wt and the p16 deletion mutant, D141. The p16 deletion mutant, DA4, lacks the Ank4 repeat (amino acid residues 110–141), but the COOH-terminal tail is fused in-frame with Ank3. The four ankyrin repeats in p15 wt are located between amino acid residues 14 and 137.

osteosarcoma cell line U20S (p16−, pRb+) to test the efficacy of our system. Once we confirmed the utility of this approach, we infected p16-deficient leukemia cell lines and demonstrated that restoration of p16 protein expression inhibits their growth dramatically; thus, supporting the hypothesis that p16 acts as a tumor suppressor in human leukemia. Materials and methods

Cell lines The five human leukemia cell lines HSB-2 and CEM (T-ALLs), NB-4 (myelocytic leukemia), and K562 (erythroleukemia) were cultured in RPMI 1640 medium supplemented with 10% (20% for CEM) fetal calf serum (FCS) and antibiotics (Gibco/BRL, Gaithersburg, MD, USA). COS-1 (SV-40 transformed African Green monkey kidney cells), and SAOS-2 and U2OS (human osteosarcoma cells) were cultured in DMEM supplemented with 10% FCS and antibiotics. All cell lines were obtained from American Type Culture Collection (ATCC, Gaithersburg, MD, USA).

Generation of recombinant p16 retroviruses The coding region of p16INK4A was amplified from a cDNA clone encoding the predicted amino acid sequence34 using the polymerase chain reaction (PCR) with PFU polymerase (Stratagene, San Diego, CA, USA) and two primers, 59GCGAATTCAGCAGCATGGATCCGG-39 and 59-TGCTTGTCATGAAGTCGAC-39. The PCR product was digested with the EcoRI (59 end) restriction enzyme and subcloned into pGEM3Z (Promega, Madison, WI, USA) cut with EcoRI and SmaI to produce pGEM-p16. The nucleotide sequence of the insert was verified by double-stranded DNA sequencing using Sequenase (Amersham Life Sciences, Arlington Heights, IL, USA). To generate the mutant p16 genes for expression in mam-

malian cells, the plasmids containing the mutant GAL4 DNAbinding domain-p16 fusion proteins33 were digested with EagI and SalI. The fragments were subcloned into pGEM-p16 digested with the same restriction enzymes. Then the resulting wild-type and mutant p16 genes (Figure 1) were subcloned from pGEM3Z, using EcoRI and HindIII, into the retrovirus expression vector pSRaMSVtkNeo; and a COS cell hyperexpression system, kindly provided by Charles Sawyers (Department of Hematology/Oncology, University of California, Los Angeles), was used to produce recombinant retrovirus stocks.35 Titers were determined by infecting the osteosarcoma cell line SAOS-2, which lacks a functional pRb, thereby allowing overexpression of p16, and evaluating the number of G418-resistant colony-forming units (G418-R CFU) per milliliter (ml).

Retroviral infection and proliferation assays For U2OS or SAOS-2, approximately 1 × 105 cells were infected with equal amounts of virus in medium containing polybrene (8 mg/ml) for 2 h, the medium was removed and replaced with fresh medium. Approximately 48 h later, cells were cultured in medium containing 500 mg/ml G418. The hematopoietic cells (2 × 105) were infected in a similar manner, but were cultured in two-layer soft agar containing G418 (CEM, 400 mg/ml; HSB-2, 100 mg/ml; NB4, 800 mg/ml; K562, 400 mg/ml) at 48 h post-infection as described previously.36 In both cases, colonies with more than 50 cells were scored 2 weeks later. For selection in liquid culture, HSB-2 and K562 were infected as described above and 2 days post-infection one-half of the cells were plated in 600 and 800 mg/ml G418, respectively, and cultured for 16 days. Total RNA was prepared on days 2 and 16 as described below.

Western blot and radioimmunoprecipitation analyses For Western blot analysis, cells were lysed in NP-40 (50 mM Tris-Cl (pH 7.4), 150 mM NaCl, 0.5 or 1.0% NP-40) or RIPA


lysis buffer (50 mM Tris-Cl (pH 7.4), 150 mM NaCl, 1.0% NP40, 0.5% deoxycholate (DOC), 0.1% sodium dodecyl sulfate (SDS)). Protein content was determined using the Bradford assay as described by the manufacturer (BioRad, Hercules, CA, USA). The cell lysates were electrophoresed through 12.5 or 15% polyacrylamide gels containing SDS and electrophoretically transferred to Immobilon-P membrane as described by the manufacturer (Millipore, Bedford, MA, USA). Membranes were blocked in TBS-T (Tris-buffered saline, 0.1% Tween-20) containing 1% gelatin and incubated with polyclonal rabbit antisera against either a full-length human GST-p16 fusion protein (Pharmingen, San Diego, CA, USA) or amino acid residues 137–156 (C-20; Santa Cruz Biotechnology, Santa Cruz, CA, USA). Antigen–antibody complexes were detected using the ECL system (Amersham Life Sciences). For radioimmunoprecipitation, cells were washed with PBS, incubated 1 h in methionine- and cysteine-free medium and labeled with 100 mCi 35S-methionine and 35S-cysteine (EXPRE35S35S; Dupont NEN, Boston, MA, USA) per 100 mm plate for 4 h. Cells were lysed in NP-40 lysis buffer containing 0.5% NP-40 for 30 min on ice, microfuged 10 min at 4°C and the resulting supernatant recovered and stored at −80°C until analysis. One-fourth of each lysate was mixed with two volumes of RIPA buffer and polyclonal rabbit antiserum against p16 (Pharmigen) and incubated on ice for 2 h. Antibody– protein complexes were recovered by incubation with 15 ml of protein G- protein A-agarose (Oncogene Sciences, Cambridge, MA, USA) or protein-A sepharose (Pharmacia LKB Biotechnology, Piscataway, NJ, USA) followed by three washes with NP-40 lysis buffer. The recovered proteins were mixed with sample buffer, boiled for 5 min, analyzed by SDSPAGE (polyacrylamide gel electrophoresis), treated for fluorography using FluoroEnhance (Research Products, Mount Prospect, IL, USA) and exposed to Kodak XO-MAT AR film (Kodak, Rochester, NY, USA). For half-life experiments, cells were labeled for 30 min as described above and chased with medium containing excess nonradioactive methionine and cysteine for 0, 0.5, 1.0, 2.0, 4.0 and 8.0 h post-labelling. Cell lysates were prepared in RIPA buffer and immunoprecipitations performed as described above. Relative intensity of the bands was determined as described previously37 and half-lives calculated by curve fitting.

Reverse-transcription polymerase chain reaction (RTPCR) analysis Total RNA was isolated from cells using Trizol reagent (Gibco/BRL) as described by the manufacturer. For each sample, 1 mg of total RNA was treated with RQ1 DNase (Promega), phenol/chloroform extracted and ethanol precipitated with 10 mg of glycogen (Boerhinger Mannheim, Indianapolis, IN, USA) as carrier. The RNA was reverse transcribed in the presence of random hexamers (Pharmacia) using MoMLV reverse transcriptase (Gibco/BRL) in a volume of 40 ml as described by the manufacturer. For PCR amplification of p16 and b-actin, 1 and 3 ml, respectively, of each cDNA were used. The primers for amplification were: p16, 59ATGGAGCCTTCGGCTGAC-39 and 59-CTGCCCATCATCATGACCTGGA-39, b-actin, 59-TACATGGCTGGGGTGTTGAA39 and 59-AAGAGAGGCATCCTCACCCT-39. The PCR conditions were 35 cycles of 95°C, 45 s; 54°C, 30 s; and 72°C, 45 s. For Southern blot analysis, the PCR reactions were electrophoresed on a 3% agarose gel and blotted on to Hybond

N+ membrane (Amersham) as described by the manufacturer. The blot was hybridized with internal oligonucleotides against either p16 (59-GCCCAACGCACCGAATAGT-39) or b-actin (59-CTACAATGAGCTGCGTGTGGC-39) that were end-labeled with 32Pg ATP using T4 kinase (Gibco/BRL) as described by the manufacturer. The final wash was in 0.1 × SSC, 0.1% SDS. The blot was exposed to XO-MAT AR film (Kodak). Results

Mutations of conserved amino acid residues in p16INK4A ankyrin repeats produce nonfunctional proteins To test the ability of p16 to inhibit the growth of leukemia cells and elucidate the functional significance, in mammalian cells, of several cancer-associated mutations,33 recombinant retrovirus stocks were generated using a COS cell hyperexpression system.35 We initially tested the efficacy of these retroviruses in an osteosarcoma cell line, U2OS, that lacks p16, but contains functional pRb. Overexpression of wt p16 arrests these cells in the G1 phase of the cell cycle and their growth is significantly inhibited.38 The U2OS cells were infected with either mutant or wt p16 expressing retroviruses (Figure 1) and grown in medium containing G418 (Table 1). The wt p16 protein inhibited colony growth greater than 30-fold compared to the empty virus control. The p16 mutants E120K and R144C inhibited growth approximately 15- to 20-fold. The G101W mutant inhibited growth slightly (0.25-fold) and H83Y was unable to inhibit growth of the cells. Of the three deletion mutants, DA4 and D141 were tested. DA4 was unable to inhibit growth (Table 1). In contrast, D141 inhibited growth approximately five-fold. A parallel infection of another osetosarcoma cell line, SAOS-2, that lacks a functional pRb, demonstrated that the mutant p16 retrovirus stocks contained the same number or slightly more G418-R CFU per infection than the wild-type (data not shown); therefore, differences in growth inhibition were not due to significant differences in the virus stock titers. To ensure that each recombinant retrovirus vector expressed its respective p16 protein, we examined COS-1 cells transiently transfected with the retrovirus expression vectors (Figure 2). The wild-type p16 and mutant proteins were detected using a polyclonal antiserum made against the fulllength p16 protein (Figure 2a). The deletion mutants migrated more rapidly than the wild-type protein as expected (Figure 2a). The endogenous monkey p16 protein was not Table 1 Growth of U2OS (p16−, pRb+) infected with wild-type or mutant p16 expressing retrovirus

Virus empty p16 wt H83Y G101W E120K R144C DA4 D141

Growth (% of empty virus)a 100 3± 102 ± 76 ± 5± 7± 137 ± 22 ±

1.4 0.1 1.0 4.0 1.4 18.0 11.0

a Results represent the mean ± the standard deviation (s.d.), of two separate experiments. wt, wild-type.

1675


1676

Figure 2 Expression of the wild-type and mutant p16 proteins by recombinant retrovirus vectors transfected into COS-1 cells. Protein lysates from cells transfected with the various retrovirus constructs were analyzed by Western blotting. (a) polyclonal rabbit anti-serum against the full-length p16 protein (Pharmingen). (b) polyclonal rabbit anti-serum against the 20 C-terminal amino acid residues, 137–156, of p16 (C-20; Santa Cruz). The arrows at the right of each panel indicate the positions of the wild-type (top) and mutant p16 proteins (DA4 and D136, middle; D141, lower). The positions of the prestained molecular weight standards (Gibco/BRL) are indicated at the left of each panel. PSR, empty vector; wt, wild-type p16.

detected in the empty vector-transfected cells indicating that either the antiserum was specific to the human protein or COS-1 cells expressed very little or no p16. To verify that the faster migrating bands were the truncated p16 proteins, we examined the same samples with a polyclonal antibody made against the last 20 carboxy-terminal amino acid residues (137–156) of p16. This antibody was unable to detect the deletion mutants as expected (Figure 2b; DA4, D136 and D141). The expression levels of the mutant p16 proteins H83Y, G101W, DA4, D141 and D136 were examined in two or three separate transient transfection experiments and were found, in each case, to be expressed at levels lower than the wild-type protein. Also, this was observed in stably transformed COS-1 cells that were analyzed by immunoprecipitation rather than immunoblot analysis (Figure 3). Other antisera made against the p16 protein revealed the same expression patterns (data not shown) indicating that possible antigenic differences between the mutant proteins were not responsible for differences in the levels of protein detected. The levels of H83Y and G101W expression were significantly higher than that of D141, which was very difficult to detect (Figure 2), but D141 expression inhibited growth more efficiently than H83Y and G101W (Table 1). This indicates that the H83Y and G101W mutations completely abrogate p16 function and suggests that their lower level of expression than wt p16 does not account for their inability to inhibit cell growth. These data demonstrated that the recombinant retrovirus vectors expressed the expected proteins and that wild-type, but not mutant p16, inhibited the growth of a pRb-positive cell line.

Mutant p16 proteins do not interact with CDK4 Our previous study showed that H83Y and DA4 were not able to interact with the CDK4 subunit in the yeast two-hybrid sys-

Figure 3 Interaction of the wild-type and mutant p16 proteins with CDK4 and CDK6 in COS-1 cells. Radiolabeled cell lysates from stably expressing cells were prepared as described in Materials and methods. Equal proportions of each lysate were immunoprecipitated with polyclonal rabbit anti-serum made against the entire p16 protein (Pharmingen). (a) Three days exposure. (b) Thirty days exposure. Bands corresponding to the p16 proteins are indicated by the arrows at the lower right of each panel: upper arrow, wt and missense point mutations; lower arrow, DA4 and D136. Bands corresponding to the CDK4 and CDK6 proteins are indicated by the arrows at the upper right of each panel. Positions of the molecular weight standards are indicated at the left of each panel. Abbreviations as indicated in legend for Figure 1.

tem like the wt p16 protein.33 The mutant G101W interacted almost as efficiently as the wt p16, but was 50 times less effective in inhibition of CDK4 kinase activity.33 To determine the ability of these proteins to interact in cells, we immunoprecipitated radiolabeled extracts from COS-1 cells stably expressing each of the p16 proteins using a polyclonal anti-p16 antibody. As observed for transient expression, we found that the mutant proteins were expressed at levels lower than the wt p16 (Figure 3a and b; 3 and 30 day exposures, respectively). The CDK4 and CDK6 proteins co-immunoprecipitated with wt p16, E120K and R144C (Figure 3a and b). These proteins were not present in the control cells stably transformed with the empty vector (Figure 3a and b) indicating they are specifically co-immunoprecipitating with p16. CDK4 and CDK6 were not co-immunoprecipitated with H83Y, DA4 or G101W (Figure 3b). Immunoprecipitation of these lysates with antiserum against the CDK4 protein co-immunoprecipitated wt p16, but not the mutants H83Y, G101W and DA4 (data not shown). Detection of the p16 deletion mutants D136 and D141 was difficult due to their low level of expression and comigration with non-specifically immunoprecipitated proteins (Figure 3a and b). These results strongly suggest that the H83Y, G101W and DA4 mutant proteins cannot inhibit growth of the U2OS cells as effectively as wt p16 because of inefficient binding to the CDK4 and 6 proteins.


Mutant p16 proteins are less stable than wild-type p16

Table 2 Growth of human leukemia cell lines infected with wildtype p16 expressing retrovirus

Both transiently transfected and stably transformed cells expressed the wt p16, E120K, R144C and DA4 proteins at levels several-fold higher than those of the other remaining mutants H83W, G101W, D141, D136 (Figures 2 and 3). This may be due to reduced protein stability for the latter group of mutations. Recently, a study showed that certain p16 mutations in lung cancer cell lines resulted in expression of proteins with reduced half-lives.39 These included H83Y and a splice site mutation that resulted in the loss of the last four amino acids (153–156) of p16. This explains the low level of expression for our carboxy-terminal deletions and mutant H83Y. To determine if the low level of G101W protein expression was due to a reduced half-life, we performed pulse-chase labeling and immunoprecipitation experiments with COS-1 cells stably expressing wt, H83Y and G101W p16 proteins (Figure 4). The wt p16 was very stable with a half-life of approximately 270 min (Figure 4). In contrast, H83Y and G101W were relatively unstable with half-lives of about 85 and 130 min, respectively (Figure 4).

Cell Line

HSB-2 CEM NB-4 K562

Mutational status a p16

pRb

− − − −

− + ND +

Growth (% of empty virus)b

88 ± 16 14 ± 3 4±2 11 ± 9

a p16 status as determined previously: HSB-2 and CEM,18 and K562;26 pRb status as determined previously: HSB-2, CEM and K562.52 b Results are expressed as a percentage of growth of the same cell line infected with empty retrovirus (ie 100%); results represent the mean ± the standard deviation (s.d.) of two independent experiments done in triplicate plates per experimental point. ND, not determined.

To determine if loss of p16 plays a possible role in the development of leukemias, we examined the ability of exogenously expressed p16 to inhibit the growth of several cell lines. Each cell line was infected with either an empty retrovirus or one expressing wt p16, and these cells were plated in soft agar containing G418 to select for infected cells. The growth of the cell lines CEM, K562 and NB-4, which are homozygously deleted for p16, was dramatically reduced (Table 2). The CEM and K562 cell lines have an intact pRb gene (Table 2), and the strong inhibition of NB-4 growth indicates that it must express a functional pRb, although this was not determined previously (Table 2). We found that mutant forms of p16 (H83Y and DA4) did not inhibit the growth of NB-4 as compared to the wild-type and were similar to the empty virus, demonstrating that functional p16 protein must be overexpressed for growth suppression (data not shown). As expected,

the growth of HSB-2 which lacks both p16 and pRb was not affected by overexpression of p16 (Table 2), demonstrating that functional pRb is required for growth inhibition. To verify that the transduced p16 gene was expressed in the leukemia cell lines after retroviral infection, we prepared total RNA from the HSB-2 and K562 cell lines 2 and 16 days postinfection. The 16 day post-infection samples were selected in the presence of G418 in liquid culture. By this time-point, few K562 cells survived, due to the inhibition of growth by p16 expression, whereas the pRb-minus HSB-2 cells grew well. The RNA samples were treated with RNase-free DNase and tested for expression by RT-PCR using primers specific for the human p16 gene. Also, as a control, we examined b-actin expression. The PCR was performed three times for each sample to verify that negative results were not due to failure of the PCR. A representative Southern blot is presented in Figure 5. As expected, cells that were infected with empty retrovirus (−) did not express detectable levels of p16, but were positive for b-actin at 2 and 16 days post-infection (lanes 1, 3, 5 and 7; Figure 5). In contrast, both HSB-2 and K562 cells that were infected with p16 containing retrovirus (+) were positive for the p16 and b-actin expression after 2 days (lanes 2 and 6, Figure 5). After 16 days of selection in G418, the HSB-2 cells

Figure 4 Half-lives of wild-type and mutant p16 proteins. COS-1 cells stably expressing the wild-type, H83Y or G101W p16 proteins were plated at equal numbers and radiolabeled for 30 min and chased with ‘cold’ methionine for 0, 0.5, 1.0, 2.0, 4.0 and 8.0 h. Lysates were immunoprecipitated with polyclonal anti-serum against the full-length p16 protein (Pharmingen); electrophoresed through a 12.5% polyacrylamide gel and fluorographed. Autoradiographs were analyzed by densitometry and the results were used to determine the half-life (T1/2) of each protein by curve fitting. Abbreviations as described in legend for Figure 1.

Figure 5 RT-PCR analysis of p16 expression in leukemia cell lines. The cell lines HSB-2 and K562 were infected with retroviruses either lacking (−) or containing (+) the wild-type p16 gene. Total RNA was prepared on days 2 and 16 post-infection. The day 16 sample was cultured for 14 days in G418 to select for only those cells infected by the retrovirus. RT-PCR analysis was performed for both the p16 and b-actin mRNAs (positions indicated at the right of the figure). The negative (N) and positive (P) controls consisted of PCR reactions either lacking a cDNA template or using a cDNA prepared from the cervical cancer cell line, HeLa, respectively. The products were examined by Southern blot analysis using internal oligonucleotides specific for each product.

Human leukemia cell line growth is inhibited by wildtype p16 INK4A expressed by recombinant retrovirus

1677


1678

expressed easily detected levels of p16, but K562 did not (lanes 4 and 8, respectively; Figure 5). These samples were positive for b-actin expression indicating that failure to observe p16 was not due to problems with the cDNA template. The negative control (N, lane 9; Figure 5) did not contain any cDNA and the positive control (P, lane 10; Figure 5) was a cDNA prepared from the total RNA of the cervical cancer cell line, HeLa. The K562 cells that grew during the selection had apparently eliminated the expression of the p16 gene, but maintained expression of the neomycin gene that confers resistance to G418. These results demonstrated that exogenous expression of wild-type p16 in leukemia cell lines with homozygous deletions of the gene occurred after retroviral infection. Discussion To test the hypothesis that the p16 gene is a growth suppressor in human leukemia cells, we utilized a retroviral expression system to overexpress both wild-type and mutant forms of the p16 protein in various human leukemia cell lines. We first examined the effect of the wt and mutant p16 genes in the osteosarcoma cell line U2OS. The growth of these cells is dramatically inhibited by overexpression of a number of CDKIs including p1638 and allowed us to test the efficacy of the retroviruses. Our initial in vitro study of the mutant p16 proteins suggested that two groups of inhibitors existed: those that bound to CDK4 and inhibited kinase activity, which included wt p16, E120K, R144C, D136 and D141, and those that did not bind to CDK4 or inhibit kinase activity and included H83Y, G101W and DA4 (Figure 1).33 The first group would be predicted to inhibit growth of cells strongly and the second would not. In contrast, the growth suppression data in U2OS cells (Table 1) indicated that the mutants should be grouped into three classes: strong, intermediate and non-functional suppressors of growth. The strong suppressors, wt p16, and mutants E120K and R144C, inhibited growth by 30-, 20- and 15-fold, respectively, and bound efficiently to CDK4 and CDK6 (Figure 3). The intermediate suppressors, G101W and D141, inhibited growth by 0.25- and five-fold, respectively, and did not bind detectable levels of CDK4 or CDK6. The non-functional suppressors, H83Y and DA4, did not inhibit the growth of cells at all and were unable to bind CDK4 or CDK6. Those mutations (G101W, H83Y and DA4) that affected the highly conserved amino acid residues in the ankyrin repeats altered p16 function the most.33,40–46 The E120K and R144C mutants significantly inhibited cell growth when overexpressed, suggesting these changes were not deleterious to the function of p16. It is possible that these changes may affect the function of p16 protein to inhibit cell cycle and growth in the context of normal expression levels and patterns and were not revealed by the assays used in our study. Alternatively, since these mutations either do not involve conserved amino acid residues or are outside of the ankyrin repeats (Figure 1), they may be polymorphisms and are not associated with disease. Recently, a study of lung cancer cell lines H596 and Calu3, that express wild-type or mutant H83Y p16, respectively, showed that H83Y was much less stable than the wt p16 (40 vs 200 min) protein.39 Also, the H83Y protein did not bind to CDK4 or CDK6, suggesting that it may not inhibit cell cycle progression or growth.39 Our results indicated H83Y is unable to inhibit cell growth when exogenously expressed in cells. In

addition, H83Y did not bind CDK4 or CDK6 and its half-life was about three times shorter than wt p16 (Figure 4). This reduction in half-life is similar to that reported previously.39 Our results verify those of the previous study, but eliminate any possible cell line differences, since we compared the H83Y mutant and wt p16 in COS-1 cells. Like H83Y, DA4 was unable to inhibit cell growth due to its failure to interact with CDK4 both in vitro and in cells (Figure 3). Both the H83Y and DA4 mutations affect the integrity of the ankyrin repeats which are critical for protein–protein interactions.33,47 Further studies are necessary to determine if failure to bind to CDK4 or CDK6 results in decreased protein stability of p16. Ambiguities arise when the activity of G101W is examined by various assays. It binds CDK4 in yeast, but does not inhibit CDK4 activity efficiently. Fifty-fold higher molar concentrations were required for wild-type levels of inhibition.33 Forced expression in cells leads to a mild growth inhibition (Table 1); however, it arrests cells in G1 almost as effectively as the wt p16.41 This mutation was reported to disrupt binding to CDK4, but not CDK6;44 however, we did not detect efficient binding to either CDK4 or CDK6 in cells (Figure 3). Structural characterization of G101W by NMR and circular dichroism indicate the mutation may result in a structural perturbation.48 Recent studies demonstrated G101W is a temperature-sensitive mutant that binds to CDK4 and CDK6, similar to wt p16, at 30°C, but is nonfunctional at 42°C in vitro.45 At 34°C, G101W arrests cells in G1 as efficiently as wt p16; but at 40°C, the arrest is substantially reduced.45 In cells at 37°C, G101W shows a two-fold reduction in half-life when compared to wt p16 (Figure 4). The temperature sensitivity, reduced half-life and inefficient binding of CDK4 and CDK6 may explain why G101W only inhibits growth very slightly at 37°C (Table 1). Under normal cellular conditions this mutation impairs p16 function and is strongly implicated in the development of melanoma.12 Further studies are required to determine if lower temperature returns G101W to a normal conformation, thus resulting in both increased protein stability and binding of G101W to CDK4 and CDK6. Our in vitro study indicated that the COOH-terminal deletion mutants retain the ability to bind to CDK4 and inhibit its activity almost as efficiently as the wt p16.33 In contrast, D141 is six times less efficient at inhibiting cell growth (Table 1). We were not able to verify that D141 binds CDK4 and CDK6 in cells, because it was expressed at very low levels (Figures 2 and 3). Although we have not determined the halflife of the D141 deletion mutant, studies of a lung cancer cell line with a splice site mutation indicated that loss of the last four amino acid residues of p16 significantly reduces the protein’s half-life.39 The weaker growth suppression by D141 is most likely a result of a reduced half-life and under normal cellular conditions D141 does not accumulate to levels high enough to inhibit growth as efficiently as wt p16. The 20amino acid COOH-terminal region of p16 is absent in the highly homologous p15 protein (Figure 1) which also inhibits cell growth;49 therefore, this region is probably not essential for the growth inhibitory function of either protein. Of note, the levels of wt p16 protein are relatively constant throughout the cell cycle with a peak at S-phase.50 In contrast, p15 levels are upregulated by either transforming growth factor beta or cell-to-cell contact.34 Currently, we are elucidating the role of the COOH-terminal tail in regulating the stability of these proteins in cells and its biological significance. Loss of p16 is strongly implicated in the development of leukemias, particularly T cell ALLs; however, it has not been demonstrated that exogenous expression of p16 has a growth


suppressive effect in leukemias. Using the recombinant p16expressing retrovirus, we reverted the transformed phenotype of a human T-ALL cell line, CEM, that lacks the p16 gene, and inhibited its growth (Table 2). This growth inhibition was dependent on a functional pRb since the T-ALL cell line, HSB2, was not inhibited (Table 2) and was able to express the p16 gene even after 2 weeks of selection in G418 (Figure 5). Homozygous deletions and point mutations are extremely rare in AMLs, but we have found that many patient samples do not express p16 mRNA.26 This suggests that another mechanism may reduce its expression. Introduction of p16 dramatically reduced the growth of myeloid cell lines K562 and NB4, and those cells that survived 2 weeks of selection in G418 did not express p16 (K562, Figure 5). In addition, the nonfunctional mutants H83Y and DA4 failed to inhibit growth of NB4, indicating that overexpression of wt p16 is required for inhibition (data not shown). Future studies remain to determine if mutations or hypermethylation of the p16 promoter play a role in abrogating expression of p16 in myeloid leukemias. During preparation of our manuscript, Quesnel and colleagues51 reported that retroviral transfer of wt p16INK4a strongly reduced the growth of K562, CEM and Jurkat cell lines in both liquid and soft agar medium. Both our study and theirs support the hypothesis that loss of p16 expression is critical in leukemogenesis.

Acknowledgements This work was supported in part by NIH grants DK41936, CA42710, DK42792, as well as the Parker Hughes Fund and the Concern Foundation. AFG was supported in part by a Cedars-Sinai Research Institute Fellowship. HPK holds the Mark Goodson Chair of Oncology Research. We thank Toshiyasu Hirama for his assistance with densitometry and helpful discussions and Marge Goldberg for secretarial assistance. We are grateful to Charles Sawyers for providing the COS cell hyperexpression system to produce recombinant retroviruses and David Beach for providing the p16INK4A cDNA.

References 1 Hunter T, Pines J. Cyclins and cancer. II: cyclin D and CDK inhibitors come of age. Cell 1994; 79: 573–582. 2 Sherr CJ, Roberts JM. Inhibitors of mammalian G1 cyclin-dependent kinases. Genes Dev 1995; 9: 1149–1163. 3 Strauss M, Lukas J, Bartek J. Unrestricted cell cycling and cancer. Nature Med 1995; 1: 1245–1246. 4 Weinberg RA. The retinoblastoma protein and cell cycle control. Cell 1995; 81: 323–330. 5 Hall M, Peters G. Genetic alterations of cyclins, cyclin-dependent kinases, and CDK inhibitors in human cancer. Adv Cancer Res 1996; 68: 67–108. 6 Serrano M, Hannon GJ, Beach D. A new regulatory motif in cellcycle control causing specific inhibition of cyclin D/CDK4. Nature 1993; 366: 704–707. 7 Kamb A, Gruis NA, Weaver-Feldhaus J, Lin Q, Harshman K, Tavtigian SV, Stockert E, Day RS III, Johnson BE, Skolnick MH. A cell cycle regulator potentially involved in genesis of many tumor types. Science 1994; 264: 436–440. 8 Nobori T, Miura K, Wu DJ, Lois A, Takabayashi K, Carson DA. Deletions of a cyclin-dependent kinase-4 inhibitor gene in multiple human cancers. Nature 1994; 368: 753–756. 9 Hirama T, Koeffler HP. Role of the cyclin-dependent kinase inhibitors in the development of cancer. Blood 1995; 86: 841–854. 10 Kamb A. Cell-cycle regulators and cancer. Trends Genet 1995; 11: 136–140.

11 Shapiro GI, Rollins BS. p16INK4A as a human tumor suppressor. Biochim Biophys Acta 1996; 1242: 165–169. 12 Hussussian CJ, Struewing JP, Goldstein AM, Higgins PAT, Ally DS, Sheahan MD, Clark WH Jr, Tucker MA, Dracopoli NC. Germline p16 mutations in familial melanoma. Nat Genet 1994; 8: 15–21. 13 Kamb A, Shattuck-Eidens D, Eeles R, Liu Q, Gruis NA, Ding W, Hussey C, Tran T, Miki Y, Weaver-Feldhaus J, McClure M, Aitken JF, Anderson DE, Bergman W, Frants R, Goldgar DE, Green A, MacLennan R, Martin NG, Meyer LJ, Youl P, Zone JJ, Skolnick MH, Cannon-Albright LA. Analysis of the p16 gene (CDKN2) as a candidate for the chromosome 9p melanoma susceptibility locus. Nat Genet 1994; 8: 22–26. 14 Caldas C, Hahn SA, da Costa LT, Redston MS, Schutte M, Seymour AB, Weinstein CL, Hruban RH, Yeo CJ, Kern SE. Frequent somatic mutations and homozygous deletions of the p16 (MTS1) gene pancreatic adenocarcinoma. Nat Genet 1994; 8: 27–32. 15 Ogawa S, Hirano N, Sato N, Takahashi T, Hangaishi A, Tanaka K, Kurokawa M, Tanaka T, Mitani K, Yazaki Y, Hirai H. Homozygous loss of the cyclin-dependent kinase 4-inhibitor (p16) gene in human leukemias. Blood 1994; 84: 2431–2435. 16 Hebert J, Cayuela JM, Berkeley J, Sigaux F. Candidate tumor-suppressor genes MTS1 (p16INK4A) and MTS2 (p15INK4B) display frequent homozygous deletions in primary cells from T- but not from B-cell lineage acute lymphoblastic leukemias. Blood 1994; 84: 4038–4044. 17 Quesnel B, Preudhomme C, Philippe N, Vanrumbeke M, Dervite I, Lai JL, Bauters F, Wattel E, Fenaux P. p16 gene homozygous deletions in acute lymphoblastic leukemia. Blood 1995; 85: 657–663. 18 Otsuki T, Clark HM, Wellmann A, Jaffe ES, Raffeld M. Involvement of CDKN2 (p16INK4A/MTS1) and p15INK4B/MTS2 in human leukemias and lymphomas. Cancer Res 1995; 55: 1436–1440. 19 Stranks G, Height SE, Mitchell P, Jadayel D, Yuille MAR, De Lord C, Clutterbuck RD, Treleaven JG, Powles RL, Nacheva E, Oscier DG, Karpas A, Lenoir GM, Smith SD, Millar JL, Catovsky D, Dyer MJS. Deletions and rearrangement of CDKN2 in lymphoid malignancy. Blood 1995; 85: 893–901. 20 Cayuela J-M, Hebert J, Sigaux F. Homozygous MTS1 (p16INK4A) deletion in primary tumor cells of 163 leukemic patients. Blood 1995; 85: 854. 21 Sill H, Goldman JM, Cross JCP. Homozygous deletions of the p16 tumor suppressor gene are associated with lymphoid transformation of chronic myeloid leukemia. Blood 1995; 85: 2013–2016. 22 Takeuchi S, Bartram CR, Miller CW, Tobler A, Seriu T, Janssen JWG, Reiter A, Ludwig WD, Zimmermann M, Schwaller J, Lee E, Miyoshi I, Koeffler HP. Analysis of a family of cyclin-dependent kinase inhibitors: p15/MTS2/INK4B, p16/MTS1/INK4A and p18 genes in acute lymphoblastic leukemia (ALL) of childhood. Blood 1995; 86: 755–760. 23 Duro D, Flexor M-A, Bernard O, d’Agay M-F, Berger R, Larson CJ. Alterations of the putative tumor suppressor gene p16/MTS1 in human hematological malignancies. C-R Acad Sci Paris 1994; 317: 913–919. 24 Gombart AF, Morosetti R, Miller CW, Said JW, Koeffler HP. Deletions of the cyclin-dependent kinase inhibitor genes p16 INK4A and p15INK4B in non-Hodgkin’s lymphomas. Blood 1995; 86: 1534–1539. 25 Hatta Y, Hirama T, Miller CW, Yamada Y, Tomonaga M, Koeffler HP. Homozygous deletions of the p15 (MTS2) and p16 (CDKN2/MTS1) genes in adult T-cell leukemia. Blood 1995; 85: 2699–2704. 26 Nakamaki T, Kawamata N, Schwaller J, Tobler A, Fey M, Pakkala S, Lee YY, Kim BK, Fukuchi K, Tsuruoka N, Kahan J, Miller CW, Koeffler HP. Structural integrity of the cyclin-dependent kinase inhibitor genes, p15, p16 and p18 in myeloid leukemias. Br J Haematol 1995; 91: 139–149. 27 Merlo A, Herman JG, Mao L, Lee DJ, Gabrielson E, Burger PC, Baylin SB, Sidransky D. 59 CpG island methylation is associated with transcriptional silencing of the tumor suppressor p16/CDKN2/MTS1 in human cancers. Nature Med 1995; 1: 686–692. 28 de Vos S, Miller CW, Takeuchi S, Gombart AF, Cho SK, Koeffler HP. Alterations of CDKN2 (p16) in non-small cell lung cancer. Genes Chromosom Cancer 1995; 14: 164–170. 29 Jen J, Harper W, Bigner SH, Bigner DD, Papadopoulos N, Mar-

1679


1680 30

31

32

33 34 35

36

37

38

39

kowitz S, Willson JKV, Kinzler KW, Vogelstein B. Deletion of p16 and p15 genes in brain tumors. Cancer Res 1994; 54: 6353–6358. Hayashi N, Sugimoto Y, Tsuchiya E, Ogawa M, Nakamura Y. Somatic mutations of the MTS (multiple tumor suppressor) 1/CDK4I (cyclin-dependent kinase-4 inhibitor) gene in human primary non-small cell lung carcinomas. Biochem Biophys Res Commun 1994; 202: 1426–1430. Mori T, Miura K, Takahisa A, Nishihira T, Mori S, Nakamura Y. Frequent somatic mutation of the MTS1/CDK4I (multiple tumor suppressor/cyclin-dependent kinase 4 inhibitor) gene in esophageal squamous cell carcinoma. Cancer Res 1994; 54: 3396– 3397. Okamoto A, Demetrick DJ, Spillare EA, Hagiwara K, Hussain SP, Bennet WP, Forrester K, Gerwin B, Serrano M, Beach DH, Harris CC. Mutations and altered expression of p16INK4 in human cancer. Proc Natl Acad Sci USA 1994; 91: 11045–11049. Yang R, Gombart AF, Serrano M, Koeffler HP. Mutational effects on the p16INK4A tumor suppressor protein. Cancer Res 1995; 55: 2503–2506. Hannon GJ, Beach D. p15INK4B is a potential effector of TGF-Binduced cell cycle arrest. Nature 1994; 371: 257–261. Muller AJ, Young JC, Pendergast AM, Pondel M, Landau NR, Littman DR, Witte ON. BCR first exon sequences specifically activate the BCR/ABL tyrosine kinase oncogene of Philadelphia chromosome-positive human leukemias. Mol Cell Biol 1991; 11: 1785– 1792. Elstner E, Linker-Israeli M, Said J, Umiel T, de Vos S, Shintaku IP, Heber D, Binderup L, Uskokovic M, Koeffler HP. 20-epi- vitamin D3 analogues: a novel class of potent inhibitors of proliferation and inducers of differentiation of human breast cancer cell lines. Cancer Res 1995; 55: 2822–2830. Hirama T, Miller CW, Wilczynski SP, Koeffler HP. p16(CDKN2/cyclin-dependent kinase-4 inhibitor/multiple tumor suppressor-1) gene is not altered in uterine cervical carcinomas or cell lines. Mod Pathol 1996; 9: 26–31. Guan K-L, Jenkins CW, Li Y, Nichols MA, Wu X, O’Keefe CL, Matera AG, Xiong Y. Growth suppression by p18, a p16INK4/MTS1and p14INK4B/MTS2 -related CDK6 inhibitor, correlates with wildtype pRb function. Genes Dev 1994; 8: 2939–2952. Shapiro GI, Park JE, Edwards CD, Mao L, Merlo A, Sidransky D, Ewen ME, Rollins BJ. Multiple mechanisms of p16INK4A inactivation in non-small lung cancer cell lines. Cancer Res 1995; 55: 6200–6209.

40 Lukas J, Parry D, Aagaard L, Mann DJ, Bartkova J, Strauss M, Peters G, Bartek J. Retinoblastoma-protein-dependent cell-cycle inhibition by the tumour suppressor p16. Nature 1995; 375: 503–506. 41 Koh J, Enders GH, Dynlacht BD, Harlow E. Tumour-derived p16 alleles encoding proteins defective in cell-cycle inhibition. Nature 1995; 5: 506–510. 42 Ranade K, Hussussian CJ, Sikorski RS, Varmus HE, Goldstein AM, Tucker MA, Serrano M, Hannon GJ, Beach D, Dracopoli NC. Mutations associated with familial melanoma impair p16INK4 function. Nat Genet 1995; 10: 114–116. 43 Wick ST, Dubay MM, Imanil I, Brizuela L. Biochemical and mutagenic analysis of the melanoma tumor suppressor gene product/p16. Oncogene 1995; 11: 2013–2019. 44 Reymond A, Brent R. p16 proteins from melanoma-prone families are deficient in binding to Cdk4. Oncogene 1995; 11: 1173–1178. 45 Parry D, Peters G. Temperature-sensitive mutants of p16 CDKN2 associated with familial melanoma. Mol Cell Biol 1996; 16: 3844–3852. 46 Lilischkis R, Sarcevic B, Kennedy C, Warlters A, Sutherland RL. Cancer-associated mis-sense and deletion mutations impair p16INK4 CDK inhibitory activity. Int J Cancer 1996; 66: 249–254. 47 Lux SE, John KM, Bennett V. Analysis of cDNA for human erythrocyte ankyrin indicates a repeated structure with homology to tissue-differentiation and cell cycle control proteins. Nature 1990; 344: 36–42. 48 Tevelev A, Byeon I-JL, Selby T, Ericson K, Kim H-J, Kraynov V, Tsai M-D. Tumor suppressor p16 INK4A: structural characterization of wild-type and mutant proteins by NMR and circular dichroism. Biochemistry 1996; 35: 9475–9487. 49 Tenan M, Benedetti S, Finocchiaro G. Deletion and transfection analysis of the p15/MTS2 gene in malignant gliomas. Biochem Biophys Res Commun 1995; 217: 195–202. 50 Tam SW, Shay JW, Pagano M. Differential expression and cell cycle regulation of the cyclin-dependent kinase 4 inhibitor p16Ink4. Cancer Res 1994; 54: 5816–5820. 51 Quesnel B, Preudhomme C, Lepelley P, Hetuin D, Vanrumbeke M, Bauters F, Velu T, Fenaux P. Transfer of p16inka/CDK2 gene in leukaemic cell lines inhibits cell proliferation. Br J Haematol 1996; 95: 291–298. 52 Cheng J, Scully P, Shew J-Y, Lee W-H, Vila V, Haas M. Homozygous deletion of the retinoblastoma gene in an acute lymphoblastic leukemia (T) cell line. Blood 1990; 75: 730–735.