Cdkn2a, the cyclin-dependent kinase inhibitor encoding p16^INK4a and p19^ARF, is a candidate for the plasmacytoma susceptibility locus, Pctr1

Abstract

Plasma cell tumor induction in mice by pristane is under multigenic control. BALB/c mice are susceptible to tumor development; whereas DBA/2 mice are resistant. Restriction fragment length polymorphisms between BALB/c and DBA/2 for Cdkn2a(p16) and Cdkn2b(p15), and between BALB/c and Mus spretus for Cdkn2c(p18^INK4c) were used to position these loci with respect to the Pctr1 locus. These cyclin-dependent kinase (CDK) inhibitors mapped to a 6 cM interval of chromosome 4 between Ifna and Tal1. C.D2-Chr 4 congenic strains harboring DBA/2 alleles associated with the Pctr1 locus contained DBA/2 “resistant” alleles of the CDK4/CDK6 inhibitors p16 and p15. On sequencing p16 and p18 cDNAs, two different allelic variants within ankyrin repeat regions of p16 were found between BALB/c and DBA/2 mice. By using an assay involving PCR amplification and restriction enzyme digestion, allelic variants were typed among several inbred strains of mice. One of the variants, G232A, was specific to two inbred strains, BALB/cAn and ABP/Le, of mice and occurred in a highly conserved amino acid in both human and rat p16. When tested with wild-type (DBA/2) p16, both A134C and G232A BALB/c-specific variants of p16 were inefficient in their ability to inhibit the activity of cyclin D2/CDK4 in kinase assays with retinoblastoma protein, suggesting this defective, inherited allele plays an important role in the genetic susceptibility of BALB/c mice for plasmacytoma induction and that p16^INK4a is a strong candidate for the Pctr1 locus.

Mouse plasma cell tumors provide an animal model system relevant to several human B cell malignancies, including human plasma cell tumors, multiple myeloma, Burkitt’s lymphoma, and non-Hodgkin’s lymphomas. The mouse plasma cell tumors are characterized by a hallmark translocation involving the myc oncogene on mouse chromosome (Chr) 15 and one of the Ig loci, either Igh, Igk, or Igl located on Chr 12, 6, and 16, respectively (1). A relatively high incidence (40–60%) of plasma cell tumors can be induced in BALB/cAn mice by the i.p. injection of pristane or silicone gels (2); in contrast, DBA/2N mice do not develop tumors after injection with these agents that induce a state of chronic inflammation in the peritoneal cavity. Backcross (3) and congenic strain analyses (4, 5) have indicated that at least four genes determine susceptibility to mouse plasmacytomagenesis. One of these genes, Pctr1, resides in the mid-portion of mouse Chr 4 near the interferon alpha locus (3, 4).

In the current study, we mapped the genes encoding three cyclin-dependent kinase (CDK) inhibitors, Cdkn2a(p16^INK4a), Cdkn2b(p15^INK4b), and Cdkn2c(p18^INK4c) to the interval of mouse Chr 4 between Ifa and Tal1 where the Pctr1 plasmacytoma susceptibility/resistance locus resides. The protein products of these genes bind to the CDKs CDK4 and CDK6 (6, 7), which, in turn, form complexes with the D-type G₁ cyclins that phosphorylate pRb to control both cell cycle and tumor suppression (8, 9). We compared the sequences of p16 and p18 between susceptible (BALB/c) and resistant (DBA/2N) strains of mice, and found sequence variants in p16 located in two different ankyrin repeat regions of the gene. DBA/2 (wild type) and BALB/c A134C and G232A variants of the p16 genes were fused to the glutathione S-transferase (GST) gene, and purified GST fusion proteins were tested for their ability to inhibit cyclin D2/CDK4 activity in kinase assays with full-length retinoblastoma (Rb) protein. The BALB/c variants were defective in their ability to inhibit Rb phosphorylation, implying that these allelic variants predispose BALB/c mice for the development of plasma cell tumors.

MATERIALS AND METHODS

Mice and Markers.

Backcross mice were bred and maintained in conventional closed barrier conditions at PerImmune (Rockville, MD) under National Cancer Institute Contract NO1-BC-21075. DNA isolation, restriction enzyme digestion, agarose gel electrophoresis, Southern blotting, and restriction fragment length polymorphism analyses were performed as described (3, 10, 11). DNA probes for Ifna, Mtv13, D4Lgm1, D4Rck41, Ccnb1-rs10, Tal2, Gt10, and Nppa were described previously (3, 12). The probes for Cdkn2a(p16) and Cdkn2b(p15) were 1.1-kb and 1.3-kb EcoRI fragments isolated from pBluescript KS (13). The probe for Cdkn2c(p18) was a 510-bp BamHI–EcoRI fragment containing the entire coding sequence of p18 excised from pBluescript SK (14). Mouse map pairs (D4Mit) for microsatellite sequences (Research Genetics, Huntsville, AL) also were typed in the backcross by PCR. PCRs were performed on 500 ng of DNA with either a PTC-100 (MJ Research, Cambridge, MA) or Omnigene (Hybaid, Middlesex, U.K.) thermocycler by using the following cycling conditions: 94°C for 1 min, 55°C for 1 min, 72°C for 30 sec for 28 cycles, and a final extension at 72°C for 10 min.

Tumor Inductions.

Primary tumors were induced with either three 0.5-ml injections of pristane or by a single 0.2-ml injection of pristane followed by inoculation with a retroviral vector containing ras and myc sequences coupled to Ig heavy chain enhancer and promoter sequences (15, 16).

Northern Analyses.

Total RNA and polyadenylated mRNA was isolated from pulverized mouse tissues by using Trizol (GIBCO/BRL) and Fast-Track (Invitrogen), respectively. Fifteen micrograms of total and 5 μg of poly(A)⁺ mRNA were separated on 1% agarose gels containing 1× MOPS and 2% formaldehyde and transferred to positively charged nylon membranes (Hybond N+) in 10× saline sodium citrate. RNA was covalently linked to the membrane with 300 mJoules of UV radiation in a Bio-Rad crosslinker. Probes for full-length p16, exon 1α (bases 32–205), and exon 1β (bases 47–206) were hybridized for 14 hr. Blots were washed with a final stringency of 0.2× saline sodium citrate at 65°C for 20 min.

Reverse Transcription–PCR.

Five micrograms of total or 2 μg of poly(A)⁺ mRNA was coincubated with 500 ng of oligo(dT) cellulose for 10 min at 70°C in a 6-μl volume. After incubation, the following reagents (GIBCO/BRL) were added: 1× reverse transcriptase buffer, 5 units of RNase inhibitor, 5 mM DTT, 1 mM dNTP, and 200 units of Moloney murine leukemia virus–reverse transcriptase. The 20.5-μl reaction was incubated for 45 min at 37°C. PCR with an oil overlay was done in a 50-μl reaction volume containing 5 μl of the reverse transcriptase reaction product, 10 mM Tris (pH 8.3), 50 mM KCl, 1.5 mM MgCl₂, 50 pmols of each primer, 100 μM of each deoxyribonucleotide, and 2.5 units of Taq DNA polymerase (Perkin–Elmer). Each cycle of PCR amplification included 50 sec at 94°C, 45 sec at 62°C, and 1 min at 72°C. Product bands were excised after separation on 2% agarose gels and removed from the gel with Geneclean II (Bio101) NaI reagents. Primers (F, R) for p16 were: ACTGAATCTCCGCGAGGAAAGCGAACT,AGACACGCTAGCATCGCTAGAAGTGAAGC. Primers (F, R) for mP18 were: CACCACTGTGAACAAGGGACCCTAAAGA,CAGTGTGAGGTCAGTGAGAGAGACCTCTACA. Primers (F, R) for mP19 were: GTCGCAGGTTCTTGGTCACTGTGA,AGACACGCTAGCATCGCTAGAAGTGAAGC.

Sequencing.

Reverse transcription–PCR products were cloned into TA vector pCR 2.1 (Invitrogen) according to manufacturer’s recommendations. Dideoxy sequencing (fmol DNA Sequencing System, Promega) was performed on 5 μl of plasmid DNA isolated with Wizard reagents (Promega). The thermocycler repeated extension program was 35 cycles of 95°C for 15 sec, 60°C for 45 sec, and 72°C for 20 sec. Sequence primers included vector M13F and M13R, the reverse transcription–PCR primers, and internally chosen primers. Products were run on denaturing 6% polyacrylamide-urea gels for 2–3 hr, followed by drying and autoradiography.

Allele-Specific PCRs.

Several inbred and wild strains of mice were genotyped for their allelic variants of exon 1 (A134C) and exon 2 (G232A). Exon 1 primers used to amplify the region containing the A134C variant were: (F) ACTGAATCTCCGCGAGGAAAGCGAACT and (R) GAATCGGGGTACGACCGAAAGAGT. Exon 2 primers used to amplify the region containing the G232A variants were: (F) GTGATGATGATGGGCAACGTTCA and (R) GGGCGTGCTTGAGCTGAAGCTA. PCR products containing exons 1 and 2 were digested with NlaIII and BsaAI, respectively, to produce allele-specific bands for detection of the polymorphic variants. Cycling conditions for the amplification of exon 1 were 94°C for 1 min, followed by 35 cycles of 94°C for 1 min, 62°C for 45 sec, 72°C for 2 min, and for exon 2, they were 94°C for 1 min, followed by 35 cycles of 94°C for 30 sec and 71°C for 45 sec and a final extension of 72°C for 10 min for both exon 1 and exon 2 PCRs.

Construction of GST-Fusion Proteins.

DBA/2 (wild type) p16^INK4a was cloned into pBluescript KS. BALB/c allelic variants A134C and G232A p16^INK4a clones were constructed by PCR-directed mutagenesis with Pfu polymerase (Stratagene) (17). Primers used to generate allelic variants were: (A134C F: TGTGCCTGACGTGCGGGCACT and R: AGTGCCCGCACGTCAGGCACA) and (G232A F: GGCAACGTTCACATAGCAGCTCTTC and R: GAAGAGCTGCTATGTGAACGTTGCC). The wild-type and variant genes were fused in-frame to the GST gene in the vector pGEX-5X-3 (Pharmacia), and the sequences of alleles were confirmed by dideoxy sequencing (fmol DNA Sequencing System, Promega). Proteins were isolated with modifications to previously published protocols (18, 19). All proteins were checked by SDS/PAGE gels and quantified by using a Bio-Rad Protein Assay Kit.

In vitro Kinase Assays.

Sf9 insect cell lysates containing CDK4 and cyclinD2 were prepared as described (20, 21). Supernatants were precleared with protein A-Sepharose beads (20 μl/mg of lysate) for 1 hr at 4°C; beads were removed by centrifugation and supernatants incubated with a polyclonal anticyclin D antibody (20 μg/mg lysate) (Upstate Biotechnology) to immunoprecipitate the CDK4-cyclinD2 complex from 200 μg of the lysate. Immunoprecipitated CDK4-cyclinD2 was mixed with increasing amounts of either GST-p16^INK4a wild-type (DBA/2), GST-p16INK4a A134C, G232A, or the combination of A134C and G232A fusion proteins (250–1,500 ng) in a total volume of 40 μl of kinase buffer for either 20 or 30 min at 30°C. After this preincubation, the kinase reactions were begun by adding 200 ng of Rb protein (p110^RB, full length) (QED Bioscience, San Diego, CA) as substrate, 10 μCi of γ-³²P ATP (4,500 Ci/mMol, Amersham), and 50 mM cold ATP to 10 μl of the CDK4-cyclinD2-p16 mixtures and incubating for 20 min at 30°C. Sample preparation, electrophoresis on SDS/PAGE gels, autoradiography and phosphor screen imaging were done as described (20, 21). The kinase assays were repeated four times with the exception that the assay involving 1.5 μg of the various p16 allelic variants was performed once. Densitometric analyses were averaged to calculate the percent of wild-type activity for each of the BALB/c-specific variants.

RESULTS

Cdkn2a, Cdkn2b, and Cdkn2c Map in the Pctr1 Interval.

Restriction fragment length polymorphisms were used to map each of the three CDK inhibitors. For Cdkn2a (p16), PvuII digestion identified 3.4-, 5.1-, and 5.9-kb fragments in BALB/cAnPt and 3.7-, 5.1-, and 6.4-kb fragments in DBA/2N. For Cdkn2b (p15), TaqI digestion identified 400-bp and 1-kb fragments in BALB/cAnPt and 380- and 600-bp fragments in DBA/2N. Cdkn2a and Cdkn2b were hybridized to a panel of 166 (BALB/c × DBA/2)F₁ × BALB/c backcross progeny that had previously been typed for both plasmacytoma susceptibility and a number of other DNA markers distributed throughout the genome (3). The allele distribution patterns were linked to those of several markers on mouse Chr 4. Based on these analyses Cdkn2a and Cdkn2b cosegregated with each other as predicted from their physical locations (13, 22, 23). These two CDK inhibitors mapped distal to the interferon alpha locus on mouse Chr 4 (Fig. 1A). The gene order, recombination fraction (recombinants/total) and distance (in cM) between markers were as follows:

Figure 1.

Cdkn2a, Cdkn2b, and Cdkn2c map to the mid-portion of mouse Chr 4 within the interval harboring the plasmacytoma susceptibility/resistance gene, Pctr1. (A) Haplotype analysis of 166 backcross progeny from the cross (BALB/cAnPt × DBA/2N)F₁ × BALB/cAnPt. The loci genotyped in the cross are indicated on the left. Each column represents the chromosome inherited in the backcross progeny; the number of progeny exhibiting each type of chromosome is listed at the bottom. Empty squares refer to the BALB/cAnPt allele and filled squares to the DBA/2NPt allele. (B) A map of mouse Chr 4 indicating the locations of genes mapped in the cross and derived from the haplotype data in A.

Ifna –(3/166) 1.8 ± 1.03–Cdkn2a, b–(4/166) 2.4 ± 1.2–D4Rck12–(6/166) 3.6 ± 1.4–Mtv13–(1/166) 0.6 ± 0.6–Tal2.

Cdkn2c(p18) was not polymorphic between BALB/c and DBA/2N; however, a PvuII digestion revealed a 3.8-kb fragment in BALB/c and a 2.5-kb fragment in Mus spretus. When Cdkn2c was hybridized to PvuII-digested DNA from 66 (BALB/c × M. spretus)F₁ × BALB/c backcross progeny, its allele distribution patterns cosegregated completely (0/66) with those for Tal2, indicating that the two genes were tightly linked. A genetic map showing the locations of the three CDK inhibitors relative to the Chr 4 markers and the Pctr1 locus is shown in Fig. 1B. In previous studies, C.D2-Lgm1A and C.D2-Lgm1H congenic strains of mice had been shown to carry DBA/2 alleles of genes across this same interval of mouse Chr 4 and were relatively resistant to plasma cell tumor induction (3, 4), thus establishing the fact that the CDK inhibitors, Cdkn2a,b,c, reside in the same interval as the plasma cell tumor susceptibility/resistance gene, Pctr1. Both of these resistant congenic strains of mice carried DBA/2 alleles of p15 and p16; p18 was not polymorphic between BALB/c and DBA/2, but mapped at the border of the Pctr1 interval.

Sequence Variants of Cdkn2a in Susceptible and Resistant Mice.

p16 and p18 cDNAs from BALB/c and DBA/2 were cloned and their sequences compared. The Cdkn2c(p18) cDNAs showed no differences between BALB/c and DBA/2. The Cdkn2a locus encodes two distinct proteins that result from alternative splicing of either exon 1α to make p16^INK4a or exon 1β to make p19^ARF to exon 2. For Cdkn2a, two sense (T75C in exon 1β of p19^ARF and C142A in exon 1α of P16^INK4a and two missense (A134C in exon 1α of p16 and G232A in exon 2 of p16 (= G257A in p19^ARF because p16 and p19 share the same exon 2) allelic variants were found between BALB/c and DBA/2 (Fig. 2). Three additional cDNAs from BALB/c plasmacytoma cell lines TEPC2027, TEPC1165, and X24 also were sequenced and found to contain the BALB/c-specific variants. The two variants of p16 common in BALB/c mice, A134C in exon 1α and G232A in exon 2, produce amino acid changes from histidine to proline and valine to isoleucine, respectively. G257A in p19^ARF results in an amino acid change from arginine to histidine. The A134C change in p16^INK4a occurs within the first ankyrin repeat (Fig. 2). The G232A variant within the second ankyrin repeat (Fig. 2) of p16^INK4a is of interest in that the valine in the DBA allelic form at this position is also present in humans and rats. Two of the tumor cell lines also carried additional somatic mutations within the third ankyrin repeat of p16 (Fig. 2). These constituted single base substitutions: G301A, C304T, and A350G, which produced amino acid changes from valine to methionine, histidine to tyrosine, and histidine to arginine, respectively in p16^INK4a. The sequence variants for p16 observed in BALB/c and tumors are illustrated in Fig. 2. Allele-specific primers were used to amplify the two allelic variants in a number of different inbred strains; these PCR products were digested with enzymes whose restriction sites were destroyed by the base change. The distribution of allele specific variants is shown in Fig. 3. BALB/c and ABP/Le mice were the only inbred strains with the G232A variant of p16^INK4a. This variant also was present in two wild-derived strains of mice.

Figure 2.

Amino acid sequence of the ankyrin repeat regions of DBA/2 (wild-type) p16^INK4a. Amino acid substitutions are indicated above or below the site of the polymorphic variants (ankyrin repeats 1 and 2) and tumor-specific mutations (ankyrin repeat 3). The amino acid sequences between BALB/c and DBA/2 were identical in the third and fourth ankyrin repeats.

Figure 3.

Allele-specific variants of p16^INK4a exon 1 (A134C) and exon 2 (G232A) in a variety of inbred and wild-derived strains of mice. The majority of strains carried the DBA/2N allele and thus are designated as wild type.

BALB/c Allelic Variants of p16^INK4a Are Inefficient Inhibitors of Rb Phosphorylation.

To test whether BALB/c and DBA/2 allelic variant p16 proteins differed in their ability to inactivate the cyclin D2-CDK4 complex, we performed in vitro kinase assays using baculovirus-expressed cyclin D2-CDK4 complexes with full-length Rb as substrate. As shown in Fig. 4 (representative of one of the four experiments), 1 μg of DBA/2 (wild type) p16 was capable of fully inhibiting Rb phosphorylation. In contrast, 1 μg of BALB/c A134C, G232A, and A134C with G232A variant proteins were significantly compromised in their ability to inhibit Rb phosphorylation (Fig. 4). Densitometry revealed the following percent of wild-type activity for 1 μg of each variant tested: A134C (0%), G232A (25%), and the combination of A134C and G232A (0%). In addition, 1.5 μg of each of these variant proteins were also inefficient in their inhibition of the cyclin D2-CDK4 complex.

Figure 4.

Allelic variants of p16^INK4a are inefficient inhibitors of Rb phosphorylation. Extracts from baculovirus-infected Sf9 cells expressing CDK4 and cyclin D2 were mixed with increasing amounts of wild-type or variant A134C (exon 1) and G232A (exon 2) GST-p16^INK4a fusion proteins. The mixtures were incubated for either 20 or 30 min before assaying their ability to phosphorylate pRb as described in Materials and Methods.

Northern Analysis of Cdkn2a.

When the full-length probe for p16 was hybridized to total RNA from BALB/c and DBA/2 kidneys, spleens, livers, and intestines, a normal transcript of approximately 990 bp corresponding to an mRNA that uses exon 1α was evident (Fig. 5). Mature B cells purified from C.B6-interleukin (IL) 6 (N12) transgenic mice also expressed the 990-bp transcript, indicating the use of exon 1α. However, among 10 established plasmacytoma cell lines, only two tumor cell lines expressed the exon 1α transcript (XRPC24 and MOPC265). Eighty percent of the cell lines (TEPC1165, TEPC2027, BPC4, MOPC460, MOPC31C, MOPC315, B9, and SP2/0) expressed a 1.1- to 1.2-kb transcript, typical of the alternatively spliced variant of p16^INK4a that uses the longer exon1 β, yielding the protein referred to as p19^ARF (Fig. 5). In addition, one primary tumor and one established cell line, TEPC 1198, did not express a transcript for exon 1α or β.

Figure 5.

Northern analysis of the Cdkn2a locus reveals that the majority of plasmacytoma cell lines do not express exon 1α transcripts of p16INK4a, but do express a 1-kb transcript corresponding to exon 1β (p19^ARF). (Left to Right) The lanes contain 5 μg of poly(A)⁺ mRNA from BALB/cAn spleen, DBA/2 spleen, a primary plasma cell tumor, and five established plasmacytoma cell lines (in order): MOPC460, TEPC2027, XRPC24, TEPC1165, and MOPC265.

Assessment of Loss of Heterozygosity (LOH) at Cdkn2a.

Second-generation transplanted tumors derived from a series of 16 (BALB/c × C57BL/6)F₁ primary tumors induced with pristane were examined for evidence of LOH at the Cdkn2a locus; only one tumor showed LOH. This tumor has shown LOH at all loci examined from Chr 4 (data not shown). In addition, LOH at Cdkn2a was not seen in a series of 14 primary tumors induced with pristane plus retroviral vectors containing Ha-ras and Myc sequences in (BALB/c × DBA/2)F₁ hybrids.

DISCUSSION

These studies have strongly implicated Cdkn2a as a candidate gene for one of the plasmacytoma tumor susceptibility/resistance loci, Pctr1. Pctr1 maps distal to the Ifa locus (3, 4) in a region of mouse Chr 4 very near the boundary of linkage homology with either human 9p21 or 1p31. In the current study we mapped the Cdkn2a and Cdkn2b loci (human 9p21) 1.8 cM distal to the Ifa locus, placing these two CDK inhibitors near the Jun oncogene (human 1p31) and well within the Pctr1 interval.

Sequence variants in the Cdkn2a gene were found between tumor-resistant DBA/2 mice and tumor-susceptible BALB/cAn mice. The majority of inbred strains were found to carry the DBA/2 allele at both sequence variants. Furthermore, G232A was a rare variant occurring in only two of the 14 inbred strains examined. Both variants, coding for different amino acids at the two positions, are old polymorphisms as indicated by the presence of these variants in wild-derived strains of mice.

The in vitro kinase assays show that BALB/cAn p16 is an inefficient inhibitor of Rb phosphorylation by Cdk4/cyclin D2. When compared with wild-type (DBA/2) p16, the BALB/cAn specific variant proteins (A134C, G232A, and the combination of A134C and G232A) were defective in their ability to inhibit Rb phosphorylation. Because BALB/c mice make a less efficient or defective p16 protein, it is likely that the negative feedback loop between p16 and Rb (24) is ineffective, and transformed B cells are allowed to continue through the cycle instead of becoming growth arrested in G₁.

Several p16 point mutations were found linked to familial melanoma (25). At least five of the point mutations observed in melanoma kindreds also were shown to be less efficient in their inhibition of Rb phosphorylation relative to wild-type p16 (21). Homozygous deletion of p16 is the most common form of alteration seen in human tumors and most mouse tumors examined (26, 27). In general, BALB/c plasma cell tumors do not show evidence of p16 deletions. Only one of the 30 primary tumors induced in F₁ hybrids showed evidence of LOH and only one established plasmacytoma cell line lacked both α and β transcripts.

Northern analyses of primary and established tumor cell lines shows a relatively complex pattern of p16/p19 expression. In normal mouse tissues (spleen, kidney, and purified populations of B cells), p16 mRNA predominated, indicating transcription initiation at exon 1α. In contrast, almost all established plasmacytoma cell lines exhibited expression of p19^ARF, indicating the initiation of transcription at the upstream exon 1β. The majority of plasma cell tumors did not express exon 1α-containing mRNA for p16^INK4a. Preliminary experiments (data not shown) have suggested that p16 is silenced by methylation in a large number of plasmacytomas, similar to that observed in a growing number of human tumors (28–30), including multiple myeloma (31).

In addition to regulation by both deficient (loss of function) allelic variation and methylation, p16^INK4a is likely to be impaired in plasma cell tumors by additional somatic mutation. The Cdkn2a locus was sequenced in three plasmacytoma cell lines; in addition to carrying the BALB/c allelic variants of p16, two of the tumor lines carried additional point mutations. Interestingly, two of the tumor-specific mutations, H75Y and H90R, were observed at two conserved histidines, H83 and H98 in humans. These same histidines also have been mutated in pancreatic adenocarcinomas (H83Y) (32), primary astrocytic tumors (H83Y) (33), primary melanomas (H98P) (34), and gliomas (H98Y) (35). H98P was found to be functionally defective in its ability to inhibit both Rb phosphorylation (34, 36) and promotion of G₁ growth arrest (34); furthermore, protein folding studies revealed that the H98P protein had a disrupted secondary structure and backbone folding (36). On the other hand, H98Y was as effective as wild-type p16 in inhibiting Rb phosphorylation (35). It remains to be tested whether a histidine to arginine change, as seen in plasma cell tumors, inactivates the inhibitor or not. Human wild-type p16 has been found to be a relatively unstable protein with a free energy of unfolding of 1.9 kcal/mol (37). Perhaps this explains why so many mutations (including a sense mutation) disrupt the structure and folding of the molecule. Therefore, it seems plausible to hypothesize that BALB/cAn mice are carrying defective alleles of Cdkn2a (Pctr1) and that tumors arising in these mice are susceptible to somatic mutation that may further alter or inactivate Cdkn2a (p16^INK4a and p19^ARF) function.

Our observations further support a role for Cdkn2a in B cell malignancy. Studies of p16 and p19 knockout mice (with abnormal extramedullary hematopoiesis and development of B cell lymphomas) (38, 57), non-Hodgkin’s lymphomas of B cell origin (14% of 42 patients had p16 mutations) (39), multiple myeloma (hypermethylation observed in 75% of 12 patients) (31), and precursor B acute lymphoblastic leukemias [15% of 81 primary tumors revealed homozygous deletions (40) and a cell line with a translocation between p16 and the Ig heavy chain locus (41)] all point to dysregulation of the Cdkn2a locus in B cell tumors. Our data suggest the possibility that allelic variants, such as A134C and G232A, of Cdkn2a could predispose individuals to the development of B cell tumors.

When wild-type p16^INK4a binds to the cyclin D-dependent kinases CDK4 and CDK6 via its ankyrin repeat regions, it induces G₁ phase arrest by inhibiting Rb phosphorylation (20, 42–44). When Rb is hypophosphorylated, it can bind to and sequester members of the E2f transcription factor family (reviewed in refs. 45 and 46). In this manner Rb acts to inhibit E2f activity, thereby blocking the transactivation of genes (e.g., Myc, Myb, Dhfr, and DNA pol a) essential for S phase progression and cellular proliferation (45–47).

In its hypophosphorylated state, Rb also can bind the nuclear transcription factor, NF-IL-6, albeit, with a 10-fold lower affinity than E2F1 (48, 49). In this situation, Rb acts positively to increase both the DNA binding and transcriptional activities of NF-IL-6 (48). NF-IL-6 (CCAAT/enhancer-binding protein beta, Cebpb) has several roles germane to plasma cell proliferation, growth, and differentiation. It has been shown to promote cellular differentiaton in B cells (48), induce apoptosis in the SP2/0 hybridoma when overexpressed by retroviral transfection (50), and regulate the transcription of IL-6 (51), a potent growth factor required by plasma cell tumors (52–54). Studies suggest that IL-6 and Rb may regulate each other; Rb phosphorylation has been shown to be suppressed in response to IL-6 (55) and the IL-6 promoter may be repressed by Rb (56). Furthermore, IL-6 can activate the synthesis of p18^INK4c and stabilize its interaction with CDK6 in B cells (55). It is intriguing to propose a similar role for IL-6 in the up-regulation of p16^INK4a/p19^ARF, leading to G₁ growth arrest.

One scenario is that an inefficient or nonfunctional p16 allele in BALB/c mice permits hyperphosphorylation of Rb by Cdk4/Cdk6. This could lead to unchecked plasma cell proliferation, further somatic mutation in the Cdkn2a locus, and possibly even to a reduction in apoptosis and IL-6 production by macrophages. The fact that at least two additional genes on Chr 4 also are involved in the development of pristane-induced plasma cell tumors in BALB/c mice underscores the complexity of this genetic disease and its usefulness as a model for human B cell malignancies.

ABBREVIATIONS

CDK

cyclin-dependent kinase

GST

glutathione S-transferase

Rb

retinoblastoma

Chr

chromosome

LOH

loss of heterozygosity

IL

interleukin