Inhibition of T-Cell Acute Lymphoblastic Leukemia Proliferation In Vivo by Re-expression of the p16^INK4a Tumor Suppressor Gene

Abstract

T-cell acute lymphoblastic leukemia (T-ALL) is characterized by the presence of differentiation-inhibited pro- and pre-T-cell blasts. The p16^INK4a tumor suppressor gene has been shown to be frequently deleted in human T-ALL cases. Deletion of p16^INK4a may be associated with poor prognosis and relapse of the disease. Radiation-induced murine T-ALL in C57B1/6 mice shares pathogenetic and molecular characteristics with the human disease. We used the murine disease as a model to study the status of the INK4/ARF gene locus and to examine the effect of p16^INK4a-re-expression in T-ALL cells on their leukemic potential in vivo. In 9 of 17 radiation-induced murine T-ALL cell lines, the p16^INK4a protein was not expressed as determined by immunoblotting. Southern blot analysis revealed homozygous deletions of the p16^INK4a gene locus in three of the nine lines, along with the genes encoding p15^INK4b and p19^ARF. Transduction of p16^INK4a-negative T-ALL lines with retrovirus encoding p16^INK4a significantly inhibited their in vitro proliferation by inducing G₁-arrest. Importantly, re-expression of p16^INK4a in p16^INK4a-negative T-ALL cells obliterated the induction of lethal disseminated leukemia in syngeneic mice. This is the first demonstration that re-establishment of p16^INK4a expression is critical for in vivo growth regulation of T-ALL cells.

Introduction

The tumor suppressor protein p16^INK4a modulates the cyclin-dependent kinase (CDK) 4/6-cyclin D-mediated phosphorylation of the retinoblastoma protein (pRB), thereby regulating the G₁- to S-phase transition of the cell cycle [1]. Inactivation of p16^INK4a is a frequent event in tumors and tumor cell lines and can occur by three known mechanisms: homozygous deletion, point mutations, or transcriptional silencing by promoter hypermethylation [2]. In T-cell acute lymphoblastic leukemia (T-ALL), homozygous deletion of p16^INK4a, located on chromosome 9p21, is found in up to 80% of patient samples, whereas point mutations and promoter methylation are infrequent events [3–7]. Furthermore, homozygous deletion of p16^INK4a may be associated with poor prognosis and relapse of the disease [7–9], often requiring additional therapeutic measures such as bone marrow transplantation (BMT).

In addition to p16^INK4a, two other genes are often lost during deletion of the 9p21 locus: The p15^INK4b gene [10], whose product also functions as an inhibitor of CDK4/6-cyclin D complexes, and p14^ARF [11]. In T-ALL, deletion of the p15^INK4b gene is less frequent than that of p16^INK4a [5,7], although the promoter of the gene encoding p15^INK4b is often methylated [5,12]. ARF, which has a partially overlapping coding sequence with p16^INK4a but encodes a unique protein, increases stability and activity of the p53 protein [13–16] and acts as a tumor suppressor independent of p16^INK4a [17]. Recently, Gardie et al. reported that the p14^ARF gene was deleted or disrupted in all 149 human T-ALL cases with rearrangements of the 9p21 locus studied, whereas the p15^INK4b and the p16^INK4a genes remained intact in 21 and 4 cases, respectively [18].

Analysis of radiation-induced thymic leukemias from C57B1/6 mice has shown that these tumors share many characteristics with human T-ALL. Both are composed of pro- and pre-T-cells blocked in differentiation [19,20]. Another similarity between human and murine T-ALL is the presence of mutations in p53, which occur in 30% to 50% of human T-ALL cases and are associated exclusively with the relapse phase of the disease [21]. Similarly, 60% of murine T-ALL samples tested had mutations in p53. These mutations occurred in leukemic animals in which the disease had spread to peripheral organs, whereas those tumors that were confined to the thymus harbored wild-type p53 [22].

The numerous similarities between human and murine T-ALL make the murine disease a suitable model for studying the pathogenesis of T-ALL and for testing potential therapies. Here we report that, as in the human disease, loss of p16^INK4a is a common event in murine T-ALL. Using retroviral gene transfer, we demonstrate that restoration of p16^INK4a expression in murine T-ALL cells that have lost expression of the endogenous protein dramatically inhibits their growth in vitro by inducing G₁-arrest. Cell-cycle arrest by overexpression of the exogenous protein was also observed in a cell line that retained expression of the endogenous protein. Importantly, we found that restoration of p16^INK4a expression in T-ALL cells significantly inhibited their ability to induce lethal disseminated leukemia in syngeneic mice.

Materials and Methods

Cell Culture

We maintained 293 human embryonic kidney cells, Balb/c3T3, and D384, a human glioblastoma cell line, in Dulbecco-Vogt modified Eagle's medium (DMEM, Sigma, St. Louis, MO) with the addition of 10% fetal calf serum (Hy-Clone, Logan, VT). All T-ALL cell lines were established from thymomas developed in C57B1/6 mice that had been treated at 4 to 6 weeks of age with 4 weekly doses of 170 cGy ionizing radiation [19,20,23]. The T-ALL lines were grown in this medium with the addition of nonessential amino acids (Irvine Scientific, Irvine, CA), 4 mmol/L L-glutamine (Irvine Scientific), 150 µmol/L asparagine, and 100 µmol/L thioglycerol (Sigma).

Southern Blot Analysis

Genomic DNA was extracted from the indicated murine T-ALL cell lines or normal mouse tissue and was digested with either Bam HI for analysis of the p15^INK4b, p16^INK4a, and p19^ARF loci or with Eco RI for analysis of the Rb gene; 10 µg of each sample was fractionated by size on a 0.7% agarose gel, transferred to nylon membrane, and hybridized with a [³²P]-labeled probe of either full-length mouse p15^INK4b (SalI fragment of pBS-p15^INK4b), p16^INK4a (EcoRI-Xho1 fragment of pBSp16^INK4a), or p19^ARF (EcoRI-Xho1 fragment of pcDNA-p19^ARF) cDNA. For Rb, the blots were probed with a 1.2-kb fragment of the 3′ end of the mouse Rb cDNA or a 1.9-kb fragment of the 5′ end isolated from a BamHI and BgIII digestion of the pECE-B/X-HA vector.

Immunoprecipitation and Immunoblot Analysis

For analysis of the endogenous murine p16^INK4a protein, 6x10⁶ cells were resuspended in 1.0-mL lysis buffer (150 mmol/L NaCl, 50 mmol/L Tris-HCl, pH 7.5; 0.5% Nonidet P-40; 500 µmol/L phenylmethylsulfonyl fluoride; and 100 U/mL aprotinin), centrifuged at 12,000 rpm for 10 minutes at 4°C, and precleared with protein A sepharose. One half of each lysate was incubated with the anti-simian virus 40 Large T antigen antibody, pAb 419 [24] as a nonspecific control, or with the monoclonal p16^INK4a antibody M-156 (2 µg per sample; Santa Cruz Biotechnology, Inc) overnight at 4°C. Bound proteins were recovered on protein A sepharose (preblocked with 3% nonfat dry milk), washed, resolved on a 15% sodium dodecyl sulfate (SDS) polyacrylamide gel, and transferred to nitrocellulose membranes, following standard protocols. The membranes were blocked in TTBS (140 mmol/L NaCl, 20 mmol/L Tris, 5 mmol/L HCl, and 0.1% Tween-20, pH 7.6) containing 5% nonfat dry milk, washed with TTBS, and incubated with the M-156 antibody (2 µg/mL) overnight at 4°C. Membranes were then washed as before, incubated with peroxidase-conjugated donkey anti-rabbit IgG (Amersham) for 1 hour at room temperature and washed in TTBS. Bands were visualized with ECL Western blot detector reagents (Amersham Life Science, Arlington Heights, IL).

For analysis of the murine stem cell virus (MSCV)-expressed human p16^INK4a, a polyclonal antihuman p16^INK4a antibody (Pharmingen, San Diego, CA; 2 µg for immunoprecipitation and 1:1000 dilution for immunoblot) was used. Binding of human p16^INK4a to the endogenous mouse CDK4 protein in infected cells was demonstrated by immunoprecipitating human p16^INK4a and immunoblotting with the monoclonal anti-CDK4 antibody ACD1 (1.5 µg/mL; Pharmingen), which recognizes both murine and human CDK4.

pRB was analyzed by lysing cells in SDS-polyacrylamide gel electrophoresis (PAGE) sample buffer (50 mmol/L Tris, pH 6.8; 10% glycerol; 2% SDS; and 1 mg/mL bromphenol blue), boiling 5 minutes, pelleting cell debris by centrifugation for 10 minutes at 20°C, resolving on a 7.5% SDS-polyacrylamide gel, and transferring to nitrocellulose. Membranes were probed with the anti-pRB antibody G3-245 (Pharmingen)and the bands visualized by chemiluminescence.

Viral Vectors, Virus Production, and Infection

MSCV-green fluorescent protein (GFP) was constructed as described previously [25]. MSCV-hp16^INK4a was constructed by removing the human p16^INK4a cDNA from pCLp16^INK4a [26] with Eco RI and XhoI and cloning it into these sites of MSCV 2.1 [27]. Ecotropic virus was produced by cotransfecting late-passage 293 cells with the viral vectors and the pCL-Eco packaging vector [28,29]. Viral titers were determined by limiting dilution of virus on Balb/c3T3 cells in the presence of polybrene and selection for 10 days with 350 µg/mL G418, resulting in titers of 10⁷ cfu/mL (MSCV), 2x10⁶ to 10⁷ cfu/mL (MSCV-GFP) and 2x10⁶ to 6 x 10⁶ colony-forming units (CFU) per milliliter (MSCV-p16^INK4a). Cell lines were infected by culturing three times for 2 to 4 hours each in viral supernatant, yielding a final multiplicity of infection of 5.

Proliferation Assays

T-ALL 4, 5, and 68 were infected with ecotropic MSCV, MSCV-GFP, or MSCV-hp16^INK4a or were incubated in medium containing 8 µg/mL polybrene. At 10 hours after the start of infection, cells were plated in triplicate 96-well tissue culture dishes with 12 wells per infected line and 10⁴ cells per well. At 28, 52, and 76 hours after infection, the cells on one plate were labeled with 50 µCi/mL [³H]-thymidine (6.7 Ci/mmol, ICN Pharmaceuticals Inc., Costa Mesa, CA) for 6 hours and then harvested onto glass fiber filters for scintillation counting.

Cell Cycle Analysis

T-ALL 4, 5, and 68 were infected with ecotropic MSCV, MSCV-GFP, or MSCV-hp16^INK4a or were incubated in medium containing 8 µg/mL polybrene. MSCV-GFP infected cells were used to monitor infection efficiency, which averaged 70% of total cells as determined by fluorescence-activated cell sorter (FACS; Becton Dickinson and Company) analysis. On days 1 through 3 after infection, 10⁶ cells from each infected population were resuspended in medium containing 20 µmol/L bromo-deoxyuridine (BrdU). Cells were incubated for 2 hours at 37°C, fixed in 70% ethanol, and stored at 4°C for at least 24 hours. After denaturation in 2.5 mol/L HCl, FITC-labeled BrdU antibody was added for 30 minutes, and cells were washed, stained in 500 µL propidium iodide (PI) solution (100 µg/mL PI, 0.1% Triton X100, 0.1 mmol/L EDTA in phosphate-buffered saline(PBS), and analyzed on a FACScan (Becton Dickinson).

In Vivo Suppression Experiments

T-ALL 5 cells were infected with ecotropic MSCV-GFP or MSCV-hp16^INK4a virus as described. At 24 hours after infection, cells were selected with G418 (1 mg/mL) for 24 hours, washed, and resuspended in PBS; 10⁵ cells were injected into the tail vein of female C57B1/6 mice. Sixteen mice were injected with MSCV-GFP-infected cells, and 15 mice received MSCV-hp16^INK4a-infected cells. Animals were monitored for 100 days for signs of disseminated leukemia including ruffled fur, weight loss, swollen abdomen, and lethargy, and they were sacrificed at signs of severe illness. Animal studies had been approved by the University of California San Diego Animal Subjects Committee.

Reverse Transcriptase Polymerase Chain Reaction Analysis

Total RNA was isolated from the liver and spleen of four final-stage animals injected with MSCV-hp16^INK4a-infected T-ALL 5 cells by using RNA Stat-60 (Tel-TEST “B,” Friendswood, TX), following the manufacturer's instructions. RT was performed with 1 µg of tumor cell RNA per sample. Thirty nanograms of RNA from MSCV-hp16^INK4a-infected T-ALL 5 cells was used as a positive control and 1 µg of RNA from uninfected T-ALL 5 cells as a negative control. One quarter of the RT reactions was used as a template for polymerase chain reaction (PCR). For amplification of the human p16^INK4a cDNA, the following primers were used: Forward primer—5′-AA CAG TCGACG GGC GGC GGG GAG CAG CAT-3′; and reverse primer—5′-TT GGA TCC GGG ATG TCT GAG GGA CCT TCC-3′. These primers were specific for human p16^INK4a cDNA and did not amplify the mouse homologue. PCR was performed in the presence of 1.5 mmol/L MgCl₂, 200 µmol/L dNTP, and 10% dimethyl sulfoxide (DMSO) by using the following protocol: 95°C, 5 minutes; 35 cycles at 94°C, 1 minute; 69°C, 1 minute; 72°C, 1 minute; 72°C, 5 minutes. RT-PCR of mouse β-actin mRNA served as a control for intact RNA. PCR was performed by using the forward β-actin primer 5′-TT GTA ACC AAC TGG GAC GAT ATG G-3′ and the reverse β-actin primer 5′-GAT CTT GAT CTT CAT GGT GCT AGG-3′ in the presence of 2.2 mmol/L MgCl₂ and 200 µmol/L dNTP and the following protocol: 95°C, 5 minutes; 30 cycles at 94°C, 1 minute; 58°C, 1 minute; 72°C, 1 minute; 72°C, 10 minutes.

Genomic DNA was extracted from leukemic spleens of the same four animals. DNA from normal liver tissue of a C57B1/6 mouse served as negative control. PCR screening for the presence of the provirus was carried out as described by using the same hp16^INK4a-specific primers. Control PCR was performed with primers amplifying a 266-bp fragment of intron6-exon7 of murine p53 (forward primer 5′-TGC CGA ACA GGT GGA ATA T-3′ and reverse primer 5′-AGG GAT CCA GTG TGA TGA TGG TAA G-3′) in the presence of 1.5 mmol/L MgCl₂ and 200 µmol/L dNTP at 95°C for 5 minutes, for 30 cycles at 93°C for 1 minute, at 58°C for 1 minute, at 70°C for 1 minute, and at 72°C for 10 minutes.

For screening for the p19^ARF transcript, total RNA was isolated from the T-ALL cell lines as described; 200-ng RNA was subjected to RT, and 2 µL of the reaction was used as a template for PCR. Forward primer 5′-CTT GGT CAC TGT GAG GAT TC-3′ and reverse primer 5′-TGA GGC CGG ATT TAG CTC TGC TC-3′ were used to amplify templates in the presence of 2 mmol/L MgCl₂ and 200 µmol/L dNTP at 95°C for 5 minutes; 35 cycles at 94°C for 1 minute, at 65°C for 45 seconds, at 72°C for 1.5 minutes, and at 72°C for 10 minutes.

Results

Deletions of the p15^INK4b, p16^INK4a, and 19^ARF Genes in Murine T-ALL

Guided by the loss of p15^INK4b, p16^INK4a, and p14^ARF in human T-ALL [5,18], we have screened a panel of murine T-ALL lines by Southern blot analysis to determine the status of these genes in the murine disease (Table 1). In 17 lines studied, we found that homozygous deletions of the p16^INK4a gene occurred in three cases. Additionally, in the T-ALL 63 line, one allele of the gene had undergone partial deletion. Because the p16^INK4a and p19^ARF genes share common exons, the deletion of the p16^INK4a locus suggested that p19^ARF was also deleted. However, because the translation product of p19^ARF exon1β exhibits the growth arrest-inducing activity of the full-length p19^ARF protein [15], we also assayed the status of that exon by Southern blot analysis. We found that p19^ARF exon1β was deleted in the same three lines that had also lost the p16^INK4a and the p15^INK4b locus. These results indicate that deletions of p15^INK4b, p16^INK4a, and p19^ARF do occur in murine T-ALL but at a lower frequency than that reported for human T-ALL.

Table 1.

Status of Tumor Suppressor Genes in Murine T-ALL Cell Lines.

Cell line	p15 gene	p16 gene	p 16 protein	ARF exon 1β	ARF mRNA	Rb gene	pRb protein	p53b
1	+	+	+	+	+	+	Nd	Mut
4	+	+	-	+	-	+	+	Mut/wt
5	Del	Del	-	Del	ND	+	+	Wt
10	Del	Del	ND	Del	ND	+	ND	Mut
11	+	+	+	+	+	+	ND	Mut
20	+	+	-	+	+	+	ND	Wt
42	+	+	-	+	+	+	ND	Wt
43	+	+	+	+	+	+	ND	Mut
44	+	+	+	+	+	+	ND	Mut
46	Del	Del	-	Del	-	+	ND	Wt
63	+	+/Dela	-	+	+	+	ND	Wt
65	+	+	-	+	+	+	ND	Wt
67	+	+	-	+	+	+	ND	Wt
68	+	+	+	+	+	+	+	Mut
71	+	+	+	+	ND	+	+	Mut
114	+	+	ND	+	ND	+	ND	Mut
121	+	+	ND	+	ND	+	ND	Mut

NOTE. Gene status was determined by Southern blot analysis. Protein expression was assayed by immunoprecipitation and Western blot analysis and mRNA expression by RT-PCR analysis.

^aSouthern blot analysis indicates that one allele of p16 in T-ALL 63 is partially deleted.

^bData from Hsiao et al., 1995 (by sequence analysis).

(+, gene/mRNA/protein detected; del, deletion; -, mRNA/protein not detected; ND, not determined)

Expression of the p16^INK4a Protein and p19^ARF Transcript in Murine T-ALL

To establish whether cells with an intact p16^INK4a gene expressed the p16^INK4a protein, immunoprecipitation and Western blot analysis of selected T-ALL lines were performed. Figure 1A shows that the p16^INK4a protein was not produced in 6 out of 12 lines that retained the gene locus. Thus, although only 3 of 17 murine T-ALL lines have deleted the p16^INK4a gene, at least 9 of the 17 did not express the protein. To determine if this was due to transcriptional silencing associated with hypermethylation of the p16^INK4a promoter, we treated the cell lines with the methyltransferase-inhibitor 5-aza-2′-deoxycytidine (5-aza-CdR) [30]. Nontoxic concentrations of the compound that allowed the cells to survive for more than 3 days were in the range of 20 to 50 nmol/L, a concentration insufficient to demethylate the p16^INK4a promoter. Indeed, treatment of the cells with this low concentration of 5-aza-CdR for 6 days did not induce p16^INK4a expression (data not shown). Hence, the mechanism by which p16^INK4a protein synthesis is inhibited in these T-ALL lines remains unresolved.

Figure 1.

Expression of p16^INK4a (A) and pRB (B) in T-ALL cell lines. (A) Lysates from murine T-ALL cell lines were immunoprecipitated by using a monoclonal antibody that recognizes mouse p16^INK4a or PAb 419 as a nonspecific control. Samples were resolved by SDS-PAGE and subjected to immunoblot analysis using the same p16^INK4a antibody. (B) Lysates from murine T-ALL cells were subjected to immunoblot analysis with the anti-Rb antibody G3-245 (Pharmingen). Lysates from Balb/c 3T3 cells were used as a positive control. Both the hypo- and hyperphosphorylated forms of pRB can be seen in each sample. Size markers are indicated on the left.

Although technical difficulties with the available antibodies against p19^ARF, which often yield uninterpretable results, precluded us from assaying for p19^ARF protein, RT-PCR analysis on mRNA extracted from the T-ALL cell lines revealed that one line with an intact ARF gene (T-ALL 4) did not express the transcript. The rest of these lines expressed p19^ARF mRNA (Table 1).

Status of Rb in Murine T-ALL

It has been reported that some human tumor cells that retain p16^INK4a expression have disrupted the p16^INK4a-Rb pathway through the loss of the Rb gene [1,26,31], whereas this inverse correlation does not apply in mouse [32]. To investigate the status of the Rb gene we screened the 17 T-ALL lines by Southern blot analysis. The results demonstrated that the Rb gene was intact in all lines examined (Table 1). Western blot analysis revealed expression of both hypo- and hyperphosphorylated pRB in four of the lines regardless of their p16^INK4a status (Figure 1B). The two forms of pRB were confirmed in an independent Western blot analysis, including serum-starved mouse embryonic fibroblasts as a control for hypophosphorylated pRB (data not shown).

Inhibition Of T-ALL Cell Proliferation in Vitro by p16^INK4a Gene Transfer

Three cell lines (T-ALL 4, T-ALL 5, and T-ALL 68) were infected with MSCV-p16^INK4a and selected for neomycin resistance. Expression of human p16^INK4a in all three cell lines was demonstrated by immunoprecipitation and Western blot analysis. The level of exogenous hp16^INK4a expression in T-ALL 4 and T-ALL 5 was significantly lower than that observed in T-ALL 68 (Figure 2A), reflecting the negative selective pressure against high levels of hp16^INK4a in the former two cell lines, as described in the following section.

Figure 2.

Expression of hp16^INK4a in MSCV-hp16^INK4a-infected T-ALL cells (A) and binding of human p16^INK4a to murine CDK4 (B). (A) Lysates from MSCV-hp16^INK4a-infected and from uninfected T-ALL 4, T-ALL 5, and T-ALL 68 cells were immunoprecipitated with a polyclonal antibody that specifically recognizes human but not murine p16^INK4a or with pAb 419 as a nonspecific control. Immunoprecipitates were resolved by SDS-PAGE and subjected to immunoblot analysis with the same p16^INK4a antibody. Uninfected cells were used as negative controls and the human glioblastoma cell line D384 was used as a positive control for hp16^INK4a. (B) Lysates from infected and uninfected T-ALL 68 cells were immunoprecipitated as in (A), and samples were subjected to immunoblot analysis with a monoclonal antibody that binds murine and human (weakly) CDK4. D384 was used as a control for binding of human CDK4 by human p16^INK4a. Size markers are indicated on the left.

Because the key biochemical role of p16^INK4a in the cell is to bind to CDK4 and CDK6, preventing these proteins from complexing with cyclin D, we determined that murine CDK4 coprecipitates with exogenous human p16^INK4a, as is shown for infected T-ALL 68 cells in Figure 2B. We also tested MSCV-hp16^INK4a-infected T-ALL 4 and T-ALL 5 for the binding of hp16^INK4a to murine CDK4 but were unable to detect any CDK4, probably due to the low level of hp16^INK4a precipitable from these cells.

To examine the biological effect of hp16^INK4a gene transfer, we assayed the proliferation rate of infected T-ALL 4, T-ALL 5, and T-ALL 68 cells by pulse-labeling with [³H]-thymidine. The fraction of infected cells as measured by GFP expression in MSCV-GFP-infected T-ALL 4, T-ALL 5, and T-ALL 68 cells was 79%, 65%, and 65%, respectively (data not shown). The proliferation of T-ALL 4 and T-ALL 5 was significantly reduced by their infection with MSCV-hp16^INK4a, compared to control- or mock-infected cells. There was also a modest decrease in the proliferation of T-ALL 68 in response to infection with MSCV-hp16^INK4a (Figure 3). These results agree with the lack (T-ALL 4 and 5) or presence (T-ALL 68) of endogenous p16^INK4a. In all three lines, the rate of proliferation increased on the third day after infection with MSCV-hp16^INK4a (Figure 3), probably due to the presence and continuing proliferation of uninfected cells.

Figure 3.

MSCV-hp16^INK4a suppresses the proliferation of infected murine T-ALL cell lines. T-ALL 4 (A), T-ALL 5 (B), and T-ALL 68 (C) were infected as described. The rate of proliferation was determined by labeling with [³H]-thymidine for 6 hours on the indicated days. The amount of [³H]-thymidine incorporated for each infected cell line is plotted versus the number of days after infection. The status of endogenous p16^INK4a is indicated for each line.

p16^INK4a-Gene Transfer Mediates Inhibition of Murine T-ALL Cell Proliferation by Inducing G₁-Arrest

To determine the mechanism by which T-ALL cell proliferation was inhibited by hp16^INK4a expression, we pulse-labeled the infected cells with BrdU and stained with PI. The fraction of infected cells as assayed by GFP expression was 68% of T-ALL 4, 95% of T-ALL 5, and 79% of T-ALL 68 cells(data not shown). The fraction of T-ALL 4 and T-ALL 5 cells in the G₁-phase on days 1 and 2 after infection with MSCV-p16^INK4a was doubled compared with control- and mock-infected cells, whereas the S-phase fraction was reduced from 40% to about 10%, as determined by FACS (Becton Dickinson and Co) analysis (Figure 4A and B). Similarly, infection of T-ALL 68 with MSCV-hp16^INK4a induced G₁-arrest, but to a lesser degree than that observed for T-ALL 4 and T-ALL 5 cells(Figure 4C). The different extent to which these cells were inhibited in proliferation (see Figure 3C) as compared with G₁-arrested cells (Figure 4C) is at least partly due to the higher infection efficiency of the cells used for cell cycle analysis. The reduced S-phase fraction of untreated T-ALL 68 cells on day 3 may reflect overcrowding caused by the rapid proliferation of this cell line.

Figure 4.

Expression of human p16^INK4a in murine T-ALL cell lines induces G₁-arrest. T-ALL 4 (A), T-ALL 5 (B), and T-ALL 68 (C) cells were infected as described. On days 1, 2, and 3 after infection, the fraction of cells in the G₁- , S- , and G₂-phases of the cell cycle was determined by BrdU incorporation and staining with PI. Those cells with a DNA content less than 2N as determined by PI staining were considered to be sub G₁. The status of endogenous p16^INK4a is indicated for each line.

Although G₁-arrest was a mechanism by which p16^INK4a-encoding virus inhibited the growth of T-ALL cells, possible induction of apoptosis [33] was also tested by using two different markers for apoptosis. Comparison of both the sub-G₁ fraction and the fraction of Annexin V-positive cells of the MSCV-hp16^INK4a infected with the control-infected T-ALL 4- and T-ALL 68-cell populations revealed no significant difference in the number of apoptotic cells. Although there was an increase in the sub-G₁ fraction among hp16^INK4a-infected T-ALL 5 cells(Figure 4B), which suggested the induction of apoptosis in 16% of the cells, the results indicate that the inhibition of T-ALL cell proliferation by the expression of exogenous p16^INK4a takes place mainly through the induction of G₁-arrest.

Inhibition of Leukemia-Cell Proliferation In Vivo by Expression of Exogenous hp16^INK4a

Previously, we found that infection of T-ALL 5 cells (p53 wt/wt) with wild-type p53-encoding vectors did not affect their growth in vitro nor reduce their ability to induce leukemia in vivo (Norris and Haas, unpublished). T-ALL 5 cells lack both copies of the gene encoding p16^INK4a. Hence, we tested whether infection with virus encoding p16^INK4a would inhibit the capacity of T-ALL 5 cells to cause leukemia in vivo. As shown in Figure 5A, infection of T-ALL 5 with MSCV-hp16^INK4a virus significantly reduced the lethality of leukemia in the recipients (P<.01, chi-squared test)compared with cells infected with MSCV virus encoding GFP. The fraction of infected cells was 94%, as determined by the fraction of GFP-expressing MSCV-GFP-infected cells.

Figure 5.

Infection of T-ALL 5 with MSCV-hp16^INK4a suppresses lethal leukemia in C57B1/6 mice. (A) Cumulative lethality in mice injected with control- or MSCV-hp16^INK4a-treated T-ALL 5 cells. None of the 16 mice injected with MSCV-GFP-infected cells were alive 30 days after injection, whereas 7 of 15 mice injected with MSCV-hp16^INK4as-infected cells were still alive 100 days after injection and showed no signs of disease. (B) RT-PCR was performed on RNA from infiltrated liver and spleen of mice injected with MSCV-hp16^INK4a-infected cells. Expression of hp16^INK4a was undetectable in all eight tumor samples from four different mice. Thirty nanograms RNA from T-ALL 5 cells infected with MSCV-hp16^INK4a was used as a positive control. Mouse β-actin RT-PCR served as a control for RNA integrity. (C) PCR on DNA extracted from leukemic organs of mice injected with MSCV-hp16^INK4a-infected cells could not detect the provirus encoded hp16^INK4a gene. A 1-kb ladder is used as size marker.

At the time the experiment was terminated after 100 days, 47% of animals injected with T-ALL 5-hp16^INK4a cells appeared healthy, and pathological analysis revealed no signs of disease. Pathological analysis of moribund animals in both the MSCV-hp16^INK4a and the MSCV-GFP-treated groups revealed disseminated leukemia cell invasion involving the liver, spleen, kidney, lung, and lymph nodes. Fluorescent microscopy of nonfixed organ sections of moribund mice in the MSCV-GFP-treated control group showed abundant GFP fluorescence in tissues taken as late as 30 days after leukemia cell injection (data not shown), indicating that MSCV retroviral gene expression was maintained in vivo for the period of the experiment.

Leukemia induced by hp16^INK4a-infected cells might have occurred in spite of the expression of exogenous hp16^INK4a, or it might have resulted from uninfected leukemic cells. To distinguish between these two possibilities, RNA was isolated from the leukemic livers and spleens of four animals injected with MSCV-hp16^INK4a-infected T-ALL 5 cells and subjected to RT-PCR with primers specific for human p16^INK4a cDNA. No expression of hp16^INK4a message could be detected in any of the organs(Figure 5B). PCR-screening of DNA extracted from these leukemic organs for the presence of the MSCV-hp16^INK4a provirus also yielded negative results (Figure 5C). Hence, the leukemias that developed in mice injected with MSCV-p16^INK4a-infected-T-ALL cells originated from cells that had not been productively infected prior to injection or from cells that had lost the MSCV-hp16^INK4a provirus.

Discussion

Many similarities exist between human T-ALL and the murine disease, including the immunophenotype of the leukemic cells and the occurrence of p53 mutations in advanced forms of the disease. Here we demonstrate that deletion of the p15^INK4b, p16^INK4a, and ARF genes, which occurs at a high frequency in human T-ALL, is also common in murine T-ALL, albeit at a lower frequency(55% versus 18%). Additionally, we found that six lines with an intact p16^INK4a gene did not express the p16^INK4a protein. Although we cannot definitively exclude that the lack of p16^INK4a protein expression was due to promoter methylation because the maximum concentrations of 5-aza-CdR used might have been too low to demethylate the promoter of the p16^INK4a gene and induce expression, it has been shown that methylation of the p16^INK4a promoter in human T-ALL is rare [5]. In contrast, silencing by promoter methylation is a common means of inactivation of the adjacent p15^INK4b gene in both human [5] and murine T-ALL [34]. Hence, our results that 3 of 17 cases had homozygously deleted the gene might be an underestimation of the frequency by which p15^INK4b is inactivated because we did not assay for protein expression.

Recently, cases of human T-ALL have been reported in which p14^ARF, the third gene residing at the INK4/ARF locus, was deleted while either the p16^INK4a or the p15^INK4b genes remained unaffected [18]. However, in two of three reported cases in which p14^ARF was deleted but p16^INK4a protein was still produced, the p16^INK4a-pRB pathway appeared to be nonfunctional because these samples displayed high levels of phosphorylated pRB despite the presence of p16^INK4a. In our murine T-ALL cell lines, we found that the p19^ARF gene was deleted in 3 of 17 samples, in parallel with the loss of p16^INK4a and p15^INK4b and that p19^ARF expression was lost in one additional line.

One explanation for the frequent targeting of the INK4/ARF locus in T-ALL is the presence of a break-point cluster region on chromosome 9p21 that predisposes that locus for illegitimate V(D)J recombinational events [18,35]. The resulting growth advantage after deletion of one or more of the genes residing at the 9p21 locus allows for clonal expansion of the affected cell [36]. Eventual disruption of the other suppressor pathways is likely to be required for full tumorigenic transformation.

Deletions of the genes encoding p15^INK4b, p16^INK4a and p19^ARF were found in leukemic cell lines derived from mice carrying either thymic or disseminated T-ALL, which are associated with wild-type p53 or mutant p53, respectively [22]. In the present work we observed a strong correlation between the presence of wild-type p53 and the loss of p16^INK4a expression. Cell lines retaining at least one copy of wild-type p53 did not express p16^INK4a, whereas 6 out of 7 lines that harbor mutant p53 express p16^INK4a. Three of the lines that possess wild-type p53 lack both p16^INK4a and p19^ARF (T-ALL 4, 5, 46), thus disrupting the p19^ARF-p53 pathway and relieving the selective pressure for mutation of p53. However, five of the wild-type p53 lines retained an intact INK4/ARF gene locus and expressed p19^ARF mRNA but not p16^INK4a. The latter may have been deleted during establishment of the cell lines, consistent with its role in replicative senescence [37]. Alternatively, p16^INK4a inactivation could have been an early event of in vivo tumorigenesis. Although loss of p16^INK4a does not destabilize the genome or facilitate dissemination of the cells (as mutant p53 does) [22,38] it may be sufficient to enable local (thymic) tumor cell growth without the need for p53 mutation [39]. Taken together, the results suggest that, as in human T-ALL, the deletion/loss of p16^INK4a in murine T-ALL can be an important step in tumorigenesis.

We have shown that re-expression of p16^INK4a in INK4/ARF-negative murine T-ALL cells was sufficient to suppress the growth of these cells in vitro. On the basis of these findings and the fact that loss of p16^INK4a can be a poor prognosticator, particularly in relapse-T-ALL [9], we examined the feasibility and effect of T-ALL growth suppression by p16^INK4a in an ex-vivo-in-vivo model, a situation designed to simulate purging of tumor cells in autologous BMT. We found that nearly 50% of mice injected with T-ALL cells infected with a recombinant retrovirus expressing hp16^INK4a failed to develop disseminated leukemia, whereas all of the mice injected with control-treated cells developed disseminated leukemia in 30 days or less. Neither transgene expression nor provirus could be detected in any of the tumors that developed in the mice injected with the hp16^INK4a-infected cells, underscoring the selective pressure against hp16^INK4a expression in the leukemic cells. Because p16^INK4a appears to be involved in lymphocyte senescence, and exogenous p16^INK4a expression is capable of inducing a differentiated state in lymphoblasts [40,41], we favor the notion that MSCV-hp16^INK4a-infected T-ALL cells underwent permanent growth arrest.

In light of the apparent role of p14^ARF in human T-ALL, it will be interesting to determine the effect of ARF-expression on in vivo leukemogenicity of T-ALL cells. In vitro, overexpression of ARF suppresses proliferation of T-ALL 5 cells and, surprisingly, to a small but significant degree, the growth of T-ALL 68 cells, which possess only a mutant p53 allele (Schoppmeyer and Haas, unpublished results).

It has been reported that restoration of p16^INK4a in p16^INK4a-negative/pRB-positive human ALL lines inhibits cell growth in culture [42]. Together with our novel finding of p16^INK4a-mediated T-ALL growth suppression in vivo, these results suggest that p16^INK4a gene transfer into residual leukemic bone marrow cells before autologous BMT may be used to inhibit leukemia cell growth and further reduce the incidence of post-transplant leukemia relapse. Future studies demonstrating the suppression of human T-ALL cell growth by p16^INK4a in a SCID mouse model are needed to support this idea.

Abbreviations

BMT

bone marrow transplantation

GFP

green fluorescent protein

MSCV

murine stem cell virus