p27^KIP1 Deletions in Childhood Acute Lymphoblastic Leukemia

Abstract

The p27^KIP1 gene, which encodes a cyclin-dependent kinase (CDK) inhibitor, has been assigned to chromosome band 12p12, a region often affected by cytogenetically apparent deletions or translocations in childhood acute lymphoblastic leukemia (ALL). As described here, fluorescence in situ hybridization (FISH) analysis of 35 primary ALL samples with cytogenetic evidence of 12p abnormalities revealed hemizygous deletions of p27^KIP1 in 29 cases. Further analysis of 19 of these cases with two additional gene-specific probes from the 12p region (hematopoietic cell phosphatase, HCP and cyclin D2, CCND2) showed that p27^KIP1 is located more proximally on the short arm of chromosome 12 and is deleted more frequently than either HCP or CCND2. Of 16 of these cases with hemizygous deletion of p27^KIP1, only eight showed loss of HCP or CCND2, whereas loss of either of the latter two loci was uniformly associated with loss of p27^KIP1. Missense mutations or mutations leading to premature termination codons were not detected in the coding sequences of the retained p27^KIP1 alleles in any of the 16 ALL cases examined, indicating a lack of homozygous inactivation. By Southern blot analysis, one case of primary T-cell ALL had hemizygous loss of a single p27^KIP1 allele and a 34.5-kb deletion, including the second coding exon of the other allele. Despite homozygous inactivation of p27^KIP1 in this case, our data suggest that haploinsufficiency for p27^KIP1 is the primary consequence of 12p chromosomal deletions in childhood ALL. The oncogenic role of reduced, but not absent, levels of p27^KIP1 is supported by recent studies in murine models and evidence that this protein not only inhibits the activity of complexes containing CDK2 and cyclin E, but also promotes the assembly and catalytic activity of CDK4 or CDK6 in complexes with cyclin D.

Introduction

Orderly progression of the cell cycle requires numerous cyclin-dependent kinase (CDK)-mediated phosphorylation events, which are under negative regulation by at least two distinct gene families (1). One is the INK4 family, which includes the p16^INK4A gene, whose product specifically inhibits cyclin D-CDK4 and cyclin D-CDK6 complexes. Homozygous inactivation of p16^INK4 occurs in a wide spectrum of human cancers, including childhood acute lymphoblastic leukemia (ALL) (2–4). The second family of universal CDK inhibitors includes p21^CIP1,WAF1, p27^KIP1, and p57^KIP2 (5–12). The p21^CIP2,WAF1 protein is inducible by p53 and functions downstream of this transcription factor to arrest the cell cycle in response to ionizing radiation and other forms of DNA damage (8,13). Thus far, there have been no reports of p21^CIP2,WAF1 abnormalities in cancer cells.

By contrast, expression of p27^KIP1, a recently cloned CDK inhibitor assigned to chromosome band 12p12, is not regulated by p53 and appears to be the CDK inhibitor most directly involved in restriction point control (reviewed in Ref (14)). Deletions and other cytogenetically apparent defects in chromosome band 12p12 have been reported in a variety of leukemias, including childhood ALL. The p27^KIP1 gene may well be the primary target of such deletions, even though p27^KIP1 deletions are hemizygous and inactivating mutations of the retained allele are rare (15–18). Haploinsufficiency leading to a reduction in expression of this protein appears to be a viable mechanism for disrupting tumor suppression by p27^KIP1 in mice (19–22). Mice nullizygous for p27^KIP1 uniformly succumb to pituitary tumors and grow more rapidly with uniform increases in organ size due to increased numbers of cells. Mice heterozygous for the disrupted allele still show a defect in growth control, although it is less pronounced than in nullizygous mice (0.15% increase in size). This suggests that the loss of one allele may influence cell division control in multipotent stem cells and more committed progenitors (19–21). Recently, Fero and colleagues (22) have demonstrated conclusively that the p27^KIP1 gene is haploinsufficient for tumor suppression in that p27^KIP1 heterozygous mice are predisposed to tumors in multiple tissues. Furthermore, molecular analysis of these tumors has shown that the remaining wild-type allele is intact, and that p27^KIP1 mRNA is expressed with no evidence of mutations. In this report, we provide evidence to support frequent allelic loss of p27^KIP1 in leukemic blast cells with 12p chromosomal deletions and demonstrate the existence of an additional gene located within a 34 kb region of homozygous deletion in a single case of T-cell acute lymphoblastic leukemia (ALL).

Materials and Methods

Cell Lines, Patient Samples and Normal Lymphocyte DNAs

The leukemia cell lines used in this study included two sublines of HL60 myeloid leukemia (HL-60M and HL-60W), one erythroleukemia (HEL); three acute myeloid leukemias (KG-1, U937, and ML-1); one chronic myeloid leukemia (K562); five pro- or pre-B lymphoid leukemias (RS4;11, UOCB-1, Reh, NALM-6, and 697); one Burkitt lymphoma (DAUDI); and two T-cell leukemias (Molt-4 and CEM). Three leukemia cell lines with translocation junctions in chromosome arm 12p (SupB2, SupB28, and 920) were also used. Three Epstein Barr Virus transformed human lymphoblastoid cell lines (Deb Cav, ElaineIV, and CJTW) served as controls.

Cryopreserved clinical samples of leukemic cells, representing 36 cases of ALL with cytogenetic alterations of chromosome arm 12p, were obtained from the Pediatric Oncology Group and St. Jude Children's Research Hospital. The presenting clinical features of the patients and their specific chromosomal abnormalities are reported in Table 1 (on disk). Nine normal Caucasian DNAs and six mixed race DNAs (BIOS Laboratories) were amplified by polymerase chain reaction (PCR) and screened for a polymorphism at codon 109 of the p27^KIP1 gene by direct sequence analysis.

Table 1.

Clinical and Genetic Features of 36 ALL Cases with Chromosome 12p Abnormalities.

Case	Age (yr) and Sex	Leukocyte count (x109/L)	Immunophenotype	Karyotypea	FISH analysisb				Southern analysisc		RT-PCRd
					p27c	HCP	Cyclin D2	D12Z3	p27	TEL	TEL-AML1
1	5.3/F	20	T-cell	inv(12)	ND	ND	ND	ND	R	ND	ND
2	7.3/F	13.6	B-lineage	del(12)(p11)	+-	+-	+-	++	G	ND	ND
3	3.9/M	5	B-lineage	del(12)(p12)	+-	++	++	++	G	ND	ND
4	2.2/F	65	B-lineage	t(5;12)(q13;p12)/del(12)(p12)	+-	+-	++	++	G	ND	ND
5	4.1/M	8	B-lineage	del(12)(p12)	+-	+-	+-	++	G	ND	ND
6	1.3/M	107	T-cell	del(12)(p11)	+-	++	++	++	G	ND	ND
7	5.2/M	7	B-lineage	del(12)(p12)	+-	++	++	++	G	ND	ND
8	7.8/F	48	B-lineage	del(12)(p12)	+-	+-	+-	++	G	ND	ND
9	12.2/M	121	T-cell	del(12)(p11)	+-	+-	+-	++	G	ND	ND
10	11.1/M	42.5	T-cell	del(12)(p11)	++	++	++	++	G	ND	ND
11	3.5/F	9	B-lineage	del(12)(p11)	+-	++	++	++	G	ND	ND
12	8.5/M	109	B-lineage	del(12)(p11)	+-	++	++	++	G	ND	ND
13	5.6/F	363	T-cell	del(12)(p12)	+-	+-	+-	++	G	ND	ND
14	8.3/M	2.6	B-lineage	del(12)(p12)	+-	++	++	++	G	ND	ND
15	3.3/M	55	B-lineage	del(12)(p12)	+-	+-	+-	++	G	ND	ND
16	3.2/F	130	B-lineage	del(12)(p11)	++	++	++	++	G	ND	ND
17	4.5/M	3.1	B-lineage	del(12)(p11)	++	++	++	++	G	ND	ND
18	6.7/F	6.9	B-lineage	del(12)(p12)	+-	++	++	++	G	ND	ND
19	8.6/F	22	B-lineage	del(12)(p11)	+-	+-	+-	++	G	ND	ND
20	4.8/M	9.9	B-lineage	del(12)(p13)	+-	++	++	++	G	ND	ND
21	3.2/F	5.3	B-lineage	del(12p)	+-	ND	ND	ND	ND	R	+
22	2.9/M	11.6	B-lineage	t(8;12)(p21;p13)	+-	ND	ND	ND	ND	R	+
23	2.2/M	3.4	B-lineage	del(12)(p11)	+-	ND	ND	ND	ND	R	+
24	4.0/M	48.7	B-lineage	add(12)(p13)	+-	ND	ND	ND	ND	R	+
25	1.6/F	186	B-lineage	der(12)t(7;12)(q11;p13)	+-	ND	ND	ND	ND	G	-
26	4.2/F	2	B-lineage	der(12)t(12;17)(p13;q12)	++	ND	ND	ND	ND	G	-
27	3.8/F	9.3	B-lineage	t(9;12)(q13;p12)	+-	ND	ND	ND	ND	R	+
28	18.2/F	50.3	B-lineage	del(12)(p11)	+-	ND	ND	ND	ND	G	-
29	2.8/F	94.8	B-lineage	del(12)(p12)	+-	ND	ND	ND	ND	R	+
30	3.2/M	38.7	B-lineage	del(12)(p12)	+-	ND	ND	ND	ND	R	+
31	3.3/M	8.9	B-lineage	del(12)(p11)	+-	ND	ND	ND	ND	R	+
32	3.3/M	36.4	B-lineage	dic(9;12)(p11;p12)	+-	ND	ND	ND	ND	R	+
33	12.8/M	16.1	T-cell	del(12)(p12)	++	ND	ND	ND	ND	G	ND
34	2.3/F	210	B-lineage	dic(9;12)(p11;p12)	+-	ND	ND	ND	ND	G	+
35	2.8/M	18.8	B-lineage	add(12)(p13)	+-	ND	ND	ND	ND	G	ND
36	4.9/M	22.9	B-lineage	del(12)(p12)/dic(12;15)(p11;p11)	++	ND	ND	ND	ND	R	+

^aKaryotypic abnormality of chromosome 12 is shown.

^b++ indicates no deletion; +- indicates hemizygous deletion; ND indicates not done.

^cR, rearranged; G, germline.

^d+, expression of TEL-AML1; -, no expression of TEL-AML1.

P1 Clones

Primers SI and AV, which produce a 233-bp fragment, were used to screen a P1 library (Genome Systems Inc, St. Louis, MO) in order to obtain genomic clones containing the p27 gene. Another primer set (5′-AGCAACTGGGAGACTCTGAG-3′ and 5′-GCTGATGAAAACCCAAACGG-3′) was used to obtain P1 clones containing the 3′ end of a rearranged fragment found in one patient sample containing a homozygous deletion. A partial restriction map of the genomic region surrounding the p27 gene was made from P1 clone 2305. This region includes the homozygously deleted segment in case 1 and the 6.5-kb Bg/II fragment, which contains the junction fragment resulting from the deletion. The restriction enzymes chosen for mapping were BamHI, Bg/II, EcoRI, HindIII, SacI and XbaI.

Fluorescence In Situ Hybridization

DNAs from the p27^kip1 P1 clones, cosmid clones corresponding to the CCND2 (cyclin D2) and HCP (hematopoietic cell phosphatase) loci, and an alphoid repeat specific for the centromere of chromosome 12 (D12Z1) were used as probes (23–25). After nick translation with either digoxigenin deoxyuridine triphosphate (dUTP) or biotin dUTP, the probes were hybridized as differentially labeled pairs in a 55% formamide, 10% dextran sulfate, and 2x standard saline citrate solution to fixed interphase nuclei from bone marrow cells with 12p abnormalities. Specific hybridization signals were detected by incubating the hybridized slides in fluorescein-conjugated antidigoxigenin antibodies and Texas red avidin. The slides were then mounted in Vectashield (Vector Laboratories, Burlingame, CA) and analyzed.

Genomic PCR and Reverse Transcriptase-PCR

The oligonucleotides used as primers in PCR experiments were as follows: SI(5′-AACGTGCGAGTGTCTAACGG-3′), SII(5′-TACGAGTGGCAAGAGGTGGA-3′), SIII(5′-AAGCGACCTGCAACCGAC-3′), SIV(5′-TTGGTGGACCCAAAGACTGA-3′), AI(5′-CGTTTGACGTCTTCTGAG-3′), AII(5′-G AACCGTCTGAAACATTTTC-3′), AIII(5′-TCTGTTCTGTTGGCTCTTTT-3′), AIV(5′-GTCGGTTGCAGGTCGCTT-3′), and AV(5′-TCCACCTCTTGCCACTCGTA-3′). Thirty cycles of amplification were performed with 50 to 200 ng of template DNA under the following conditions: 95°C for 30 seconds, 55°C for 30 seconds, and 72°C for 60 seconds. Two sets of primers, SI/AII and SIII/AI, generated products of approximately 1.0 kb and 0.5 kb, respectively, which then were used for sequence analysis.

Total RNA (1–5 µg) was used to synthesize cDNA in a procedure employing the AII oligomer or random hexamers in the presence of RNase inhibitor (80 U) and M-MLV (Moloney-Murine Leukemia Virus) RT (200 U). The reaction mixture was incubated at 37°C for 60 minutes. A 506-bp cDNA fragment (p27^KIP1 cDNA probe 1) was generated from normal human total RNA by nested PCR using the outer primers SI/AII and the inner primers SI/AIII. This fragment was subcloned into EcoRV-digested Bluescript SK(+) T-vector (26), sequenced, and used as a p27^KIP1 cDNA probe. One fifth of the RT reaction mixtures was subjected to 30 cycles of PCR with random hexamers and primers SI/AI to generate a 588-bp p27^KIP1 cDNA to be used as a template for direct sequencing. This fragment was also subcloned into the Bluescript T-vector. A 119-bp fragment containing only the second exon (p27 cDNA probe 2) was obtained by digesting the p27^KIP1 cDNA fragment with Hinf1 and used as a probe.

Anchored PCR

3′-RACE (rapid amplification of cDNA ends) was performed with the SIII/SIV gene-specific primers from the p27^KIP1 coding sequences, a dT₁₇-adapter primer (5′-GAGTCGACTCGAGAATTCdT₁₇-3′) and adapter primer (5′-GAGTCGACTCGAGAATTC-3′), according to a previously described method (27). Briefly, cDNA was made with use of the dT₁₇-adapter primer, and PCR was performed first with SIII and the adapter primer and then with SIV and the adapter primer. 3′-RACE was performed with 1 Fg of total RNA from case 1. PCR analysis with the primer set SIV and the adaptor primer produced two different fragments that were subsequently subcloned into a TA cloning vector (PCR TM II, Invitrogen, Carlsbad, CA).

PCR products were directly sequenced with a cycle sequencing kit (Gibco, Rockville, MD) by using the SI/SII/AI primer set according to the manufacturer's instructions.

Southern Blot Analysis, Restriction Mapping and Subcloning

DNA samples were extracted from 10 patient samples that had been identified as containing translocation or inversion junctions on chromosome arm 12p, as well as from 23 others that also had been analyzed by FISH. DNA (10 µg) from each sample was digested with various restriction enzymes, electrophoresed, and transferred to nylon membranes, which were then hybridized with the p27^KIP1 cDNA probe. The blots were stripped and analyzed with a CDK4 probe, which served as the same-chromosome internal control in these experiments (28). Band intensities were measured with a Phosphorlmager (Molecular Dynamics, Redwood, CA). DNA (20 µg) from case 1 was digested with Bgl/II and electrophoresed in a 0.8% agarose gel. The region of the gel containing fragments of between 4.4 kb to 9.6 kb was excised, electroeluted and purified. A genomic library was constructed from these fragments by using Zap express phage (Stratagene, La Jolla, CA). The library was screened by plaque hybridization with the p27^KIP1 cDNA probe, and the rearranged 6.0-kb fragment was isolated. Sixteen cases were analyzed for TEL gene rearrangement by Southern blot analysis and 14 cases for expression of TEL-AML1 by RT-PCR, as previously described (29).

Results

Restriction Mapping of the Human p27^KIP1 Gene

A contig composed of three P1 clones was produced by PCR-based screening of a human P1 library (Figure 1). One of these clones, 2306, was used for FISH localization and interphase cell deletion studies. A physical map of the P1 contig was constructed by using six different restriction enzymes (Figure 1B), and a Bg/II fragment containing the entire p27^KIP1 coding sequence was subcloned and partially sequenced to identify the location of the coding sequences within the p27^KIP1 cDNA (Figure 1C).

Figure 1.

Restriction map of the p27^KIP1 locus. (A) P1 contig containing fourclones that were identified by screening for either p27^KIP1 or the junction fragment of the deletion found in case 1. (B) Partial restriction map of P1 clone 2305 showing the location of the p27^KIP1 gene and the 34.5-kb homozygously deleted segment found in case 1. (Black boxes, the coding sequences of the p27^KIP1 cDNA; genomic fragments derived from the p27^KIP1 locus; hatched box, the case 1 deletion junction fragment; arrow, the deleted 34.5-kb segment). (C) Restriction map of the 6.5-kb Bg/II genomic subclone containing the p27^Kip1 locus. (D) Restriction map of the rearranged Bg/II genomic fragment isolated from case 1 that contained the deletion junction. (Restriction enzyme abbreviations: B, BamHI; Bg, Bg/II; E, EcoRI; H, HindIII; S, SacI; X, XbaI.

FISH Analysis of the p27^KIP1 Gene in ALL Cases with 12p Alterations

Among 35 cases of ALL with cytogenetically visible alterations of chromosome arm 12p that were examined by FISH (Figure 2; Table 1), 29 had hemizygous deletions of the p27^KIP1 probe. Nineteen of these cases were also analyzed for deletion of the HCP and CCND2 genes, which map telomeric to the p27^KIP1 locus on chromosome arm 12p. p27^KIP1 was deleted in 16 cases, but HCP and CCND2 deletions were found in only 8 and 7 cases, respectively. Each case with either an HCP or CCND2 deletion also had a p27^KIP1 deletion; 8 of the 19 cases had p27^KIP1 deletions exclusively (Figure 2C). These results suggest that p27^KIP1 was the nearest of the three probes to the smallest region of overlap of 12p deletions and thus to the putative ALL tumor suppressor gene which has been assigned to this chromosomal region.

Figure 2.

FISH analysis of p27 copy number in primary leukemic cells. A digoxygenin-labeled P1 clone 2306 (labeled green with FITC) containing the entire p27^KIP1 locus was cohybridized with a biotin-labeled chromosome 12-specific centromere probe D12Z3 (labeled with Texas Red). (A) Hybridization to interphase nuclei from normal peripheral blood lymphocytes demonstrating two green signals corresponding to p27^KIP1 and two red signals corresponding to the chromosome 12 centromere. (B) Hybridization to interphase cells from a bone marrow aspirate from a patient with ALL showing a hemizygous deletion of p27^KIP1 with one green signal and two red signals per cell. (C) Schematic representation showing the locations of four FISH probes (cyclin D2, HCP, p27^KIP1 and D12Z3) on chromosome arm 12p and interphase FISH results in 21 ALL cases with 12p deletions. (++, no deletion; ±, hemizygous deletion).

There was no discernible correlation between the finding of p27^KIP1 hemizygosity and presenting leukocyte counts, age, or sex (Table 1). Indeed, the median leukocyte count in this group (20x10⁹/L) was identical to that in patients who retained both alleles.

Southern Blot Analysis of the p27^KIP1 Gene in ALL Cases with 12p Abnormalities

Hybridization with the p27^KIP1 cDNA probe demonstrated a rearrangement of the p27^KIP1 locus in only 1 of the 20 cases examined. This case (no. 1, Table 1) had a chromosome 12 inversion and insufficient leukemic cells remaining after DNA/RNA extraction for FISH analysis. Rehybridization of the same blots with a CDK4 control probe (28), located at 12q13, indicated a relatively reduced p27^KIP1 signal in 16 of the 19 cases examined by both FISH and Southern blotting, a finding consistent with hemizygous deletion of the p27^KIP1 locus demonstrated by the FISH technique.

Analysis of the TEL Gene by Southern Blotting and TEL-AML1 Expression by RT-PCR

TEL, an ETS-related transcription factor gene rearranged in several types of leukemia, also maps to chromosome band 12p13 and therefore is another excellent tumor suppressor candidate. Of 16 cases analyzed by Southern blotting with the TEL cDNA probe, 10 had detectable rearrangements of the TEL gene, with each expressing the TEL-AML1 fusion transcript by RT-PCR. An additional case (no. 34, Table 1) expressed the TEL-AML1 transcript, but the rearranged TEL allele was insufficiently different from the germline allele in a BamHI restriction digest to be detectable by Southern blotting. Of the five cases lacking detectable rearrangement of the TEL gene, three showed hemizygous deletion of p27^KIP1.

Mutational Analysis of the p27^KIP1 Gene in Cases of Primary ALL and Leukemic Cell Lines

DNA and RNA PCR products of p27^KIP1 coding sequences from 15 leukemic cell lines were sequenced to detect possible p27^KIP1 mutations. Three of these cell lines (Nalm-6, KG-1, and Daudi) showed the same single base change of GTC (Val) to GGC (Gly) of both alleles of codon 109, and another cell line (CEM) showed the same alteration hemizygously, in both PCR-amplified RNA and genomic DNA. Because this base change creates a new Bg/I restriction site, we analyzed 15 normal lymphocyte DNA PCR products and found this polymorphism in three of the DNA samples, (one homozygous, two heterozygous). Others have detected a similar polymorphism as a normal variant (17). No additional sequence anomalies were discovered in any of the cell lines. Sequence analysis of 10 primary ALL cases with hemizygous deletions of p27^KIP1 that were identified by FISH revealed a lack of identifiable mutations, except for a single case with the codon 109 polymorphism and another with a third base change in codon 124 (AAC to AAT [Asn]).

Homozygous Disruption of p27^KIP1

Only one case of primary ALL (no. 1) showed homozygous disruption of the p27^KIP1 gene by Southern blotting. The rearranged p27^KIP1 restriction fragments, obtained with three different enzymes, were accompanied by very faint bands in the normal position Figure 3, as a result of the small percentage of residual normal cells that remained in the sample (15% by Wright-stained differential count). The restriction maps of the subcloned normal and rearranged p27^KIP1 Bg/II fragments are shown in Figure 1C and D, respectively. Sequence analysis of the rearranged fragment identified a divergence from the normal sequence in the intron between the first and second coding exons. Sequences immediately 3′ of the divergence point in the rearranged fragment were contributed from a region on 12p that was 34.5 kb downstream of p27^KIP1 exon 1, indicating that a deletion had occurred that includes exon 2 of the p27^KIP1 gene on the retained allele in this case (Figure 1B and D). Consistent with these results, Southern blots probed with an exon 2-specific p27^KIP1 cDNA probe detected only a faint band contributed by the normal residual cells (Figure 3C). Thus, in this case of T-cell leukemia in a 5-year-old girl with an inv(12) karyotype, there was an entirely deleted p27^KIP1 allele and an interstitial deletion of the remaining allele with homozygous loss of exon 2 coding sequences.

Figure 3.

Southern blot analysis of p27^KIP1 restriction fragments in primary leukemia cells from case 1. Ten micrograms of DNA were digested with (A) HindIII or (B) EcoRI and subjected to Southern blot analysis with a p27^KIP1 cDNA probe 1 (top) or a CDK4 genomic probe (27) as a control for DNA loading (bottom). (C) The same blot used in (B) was reprobed with a p27^KIP1 cDNA probe containing only exon 2 coding sequences, which did not hybridize to the rearranged p27^KIP1 locus retained in case 1, but only to a faint band arising from residual normal cells in the specimen.

Identification of a p27^KIP1 Fusion Transcript in Case 1

Anchored PCR was used to identify and clone two p27^KIP1 fusion transcripts from the case just dUTP described. Sequence analysis revealed that in the longer transcript, exon 1 was fused to a short exon of 47 bp (labeled X in Figure 4) preceding the common 3′ exon identified in both transcripts (labeled Y in Figure 4). Because of the absence of a polyA sequence at the 3′-end of fragment Y, it appeared that the adapter primer used for anchored PCR annealed directly to this mRNA sequence in the PCR reaction. Southern blot analysis with the X and Y sequences used as probes showed that they hybridized to P1 clone 4069, which was identified by screening a P1 library with a site from the genomic sequence located at the deletion junction boundary identified in case 1 (see Figure 1). Genomic fragments that contained the X and Y exons were then isolated from P1 clone 4069 and sequenced. Consensus splice acceptor and donor sites were found to lie adjacent to the exonic sequences encoding ‘X,’ and a consensus splice acceptor site was identified that preceded the 5′-end of the sequence corresponding to the anchored PCR product labeled Y (Figure 4b). We also found genomic sequences that matched 8 bp of the 3′ adapter primer sequence just after the 3′-end of product Y. Thus, in this case, mRNA containing coding exon 1 was expressed as a fusion transcript containing exons from another gene downstream of the p27^KIP1 gene on chromosome arm 12p (Figure 4A). In both fusion RNAs, a stop codon was encountered in the 5′ region of the exon referred to as “Y,” causing a truncation of the p27^KIP1 open reading frame.

Figure 4.

Schematic representation of the p27^KIP1 fusion RNA and sequences contributed by exons from another gene. (A) Diagram of two spliced forms of fusion mRNAs found in case 1, including p27 exon 1 fused to either of two exons labeled X and Y (top) or exon Y only (bottom). TGA and TAA indicate stop codons in sequence Y that are in frame with the p27^KIP1 coding sequences. (B) Genomic sequences of the regions encoding exons X and Y showing splice acceptor and donor sites (Roman type) and the exonic sequences contained in the p27^KIP1 RACE products (bold type). The sequence shown at the 3′ end of exon Y matches the 3′ RACE adapter primer.

Discussion

The properties of p27^KIP1 are most consistent with those of a tumor suppressor. This interpretation is supported by the observation that mice nullizygous for the p27^KIP1 gene show neoplastic growth of pituitary cells of the pars intermedia (19–21), and that mice hemizygous for inactivation of the gene are prone to develop tumors of multiple organs after treatment with ionizing radiation or carcinogens (22). The human gene is located on the short arm of chromosome 12 (15,16) in a region that is frequently deleted in both childhood ALL (30,31) and acute myeloid leukemia (32). With rare exceptions, including the case reported here, mutations of the retained p27^KIP1 allele have not been identified (15–17,33–37), suggesting that p27^KIP1 contributes to ALL on the basis of haploinsufficiency, in which one allele maintains lower-than-normal levels of the CDK inhibitor. Empirical support for this possibility comes from studies of p27^KIP1 hemizygous mice, whose size is larger than normal but smaller than that of nullizygous animals and in which p27^KIP1 has been shown to act as a haploinsufficient tumor suppressor (22).

Our studies indicate the existence of a previously unidentified gene on chromosome arm 12p, which is located within a region approximately 34 kb downstream of the second p27^KIP1 coding exon. This new gene contributes an exon to the p27^KIP1 fusion transcripts found in case 1 of our study. The 5′ amino-terminal portion of the p27 molecule contains a CDK inhibitory domain, which is sufficient to inhibit kinase activity by itself (6), so the potential fusion protein encoded by this chimeric transcript may have some residual activity in cell cycle control. The region of p27^KIP1 encoded by the missing exon 2 has been shown to be responsible for binding of this protein to E1A (38), suggesting that it encodes a protein-protein interaction domain involved in some way in the regulation of p27^KIP1 function. Because of a lack of sufficient primary leukemia cells from case no. 1, we were unable to determine whether the fusion transcript identified by anchored PCR encoded a stably expressed protein in this patient's leukemic cells.

Biochemical studies have suggested a mechanism underlying the in vivo selection of haploinsufficiency for p27 rather than nullizygosity in murine tumors and childhood ALL. CDK inhibitors of the CIP/KIP family, including p21^CIP1, p27^KIP1, and p57^KIP2, are potent inhibitors of cyclin E-CDK2 complexes and induce cell cycle arrest when expressed at high levels (1). However, evidence has accumulated indicating that these proteins promote and are in fact required for the formation of functional cyclin D-CDK4 or -CDK6 complexes (39–43), which act in G₁ phase to integrate cell cycle progression with external mitogenic signals (14). Thus, a single allele of p27^KIP1 may encode high enough levels of the protein to contribute to leukemogenesis by promoting the initiation of cell proliferation by complexes of CDK4 or CDK6 with cyclin D. However, the lower-than-normal levels produced by a single allele may be insufficient to block the essential S-phase promoting activity of CDK2 in complexes with cyclin E and cyclin A. The dual activities of p27^KIP1 in both promoting and inhibiting the activity of cell cycle kinases active in G₁ phase of the cell cycle may explain the observation both in childhood ALL and murine models that nullizygosity for p27^KIP1 is not selectively acquired by cancers arising spontaneously in vivo.

Abbreviations

CDK

cyclin-dependent kinase

ALL

acute lymphoblastic leukemia

FISH

fluorescence in situ hybridization