Thirteen of 40 human myeloma cell lines do not express p18INK4c
Twelve (RPMI-8226; ARP-1; Karpas-620; KMS-12PE; MM-1; OCI-MY1; OCI-MY5; OCI-MY7; OPM-1; OPM-2; SKMM-2; XG-6) of 40 HMCL analyzed have no detectable p18INK4c RNA by an RT.PCR assay (Fig. ), and no detectable p18 exon 3 sequences in genomic DNA (Fig. ). The remaining 28 HMCL express p18 RNA at levels within an approximately twenty-fold range (Table ). Western blots (representative results in Fig. ) detect an apparently normal p18 protein at a level that parallels the level of p18 RNA in 27 of these HMCL, but no p18 protein was detected in KMS-12BM. Sequence analysis of an RT.PCR product showed that KMS-12BM had a homogeneous p18 sequence, with a TT deletion in codon 21 (CTT) that resulted in conversion of Leu to His, which was followed immediately by a TGA stop codon. Additional sequence analyses of RT.PCR products showed no coding mutations in 13 other HMCL (Table ).
Figure 1 Expression of p18INK4c in representative myeloma cell lines. Results of RNA expression, gene content, and protein expression are shown, with experimental details included in Materials and Results. Agarose gels show: A) a 521 bp exon2-exon3 p18 RT.PCR (more ...)
The consensus minimum deletion region in 13 HMCL selectively targets the p18INK4c gene
The p18INK4c gene is located on chromosome band 1p32.3, 9 kb centromeric to the Fas associated factor 1 (FAF1
) gene, which encodes a receptor (TNFRSF6) that is able to mediate apoptosis initiated by interaction with external FAS ligand (TNFSF6) [11
]. Since bi-allelic deletion of p18 might affect FAF1, the minimum deletion region was determined for each of the 12 HMCL in which bi-allelic deletion of p18 had been identified (Fig. ). The minimum deletion region did not involve FAF1 exons in three HMCL (KMS12PE; OCI-MY7: MM.1), but included variable portions of the FAF1 gene in the other nine HMCL. Consistent with these results, FAF1 RNA was not expressed in the nine HMCL that had bi-allelic deletions of some portion of the FAF1 gene. However, with the exception of a very low level of expression in MM-1, FAF1 RNA was expressed in all other HMCL, including KMS-12PE and OCI-MY7 (Table ). Therefore, p18INK4c is the uniform target of deletion, although it is possible that the associated inactivation of FAF1 could affect the tumor cell phenotype
Figure 2 Minimum deletion regions on 1p32.3 for HMCL with bi-allelic deletion of p18INK4c. Pairs of oligonucleotides at the indicated positions (relative to the 5' end of the p18 gene) were used in PCR reactions to identify regions of bi-allelic deletion (black (more ...)
We found no evidence for deletion "hotspots" since each of the twelve HMCL derived from independent tumors appeared to have unique telomeric and centromeric deletion breakpoint regions (Fig. ).
The OPM-1 and OPM-2 HMCL, which were derived independently from the same tumor specimen [12
], had identical deletion profiles (Fig. ). Therefore, the deletion most likely occurred in tumor cells in the patient and not during generation or propagation of the cell lines.
Bi-allelic deletion of p18INK4c is much less prevalent in MM tumors than in HMCL
The expression of p18INK4c RNA was determined for 231 untreated MM tumors, 30 relapsed MM tumors, 33 HMCL, and 16 normal bone marrow plasma cell samples using Affymetrix HG-U95A Gene Chips. The normalized expression of p18 RNA for the 310 samples is summarized in Table . Ten of 33 (30%) HMCL have deletion of the p18 gene and normalized p18 RNA values < 0.55, thereby defining a maximum background signal. In contrast, only 28 of 261 (11%) MM samples, and none of 16 normal PC samples have p18 RNA values < 0.55. To directly assess the prevalence of bi-allelic p18 deletion in MM tumors, we did quantitative real time PCR assays to determine the content of p18 genomic DNA in 31 MM tumors, including twelve tumors with p18 RNA values less than the background level of 0.55 (representative results in Table ). Compared to a reference CYCLIN D3 gene, only three tumors (P051, P057-2, P303) had a significant decrease (nine- to fourteen-fold) in p18 gene content compared to placental DNA. None of the other 28 MM tumors, including nine with background values of p18 RNA, had even a twofold decrease in p18 gene content, so that the expression array results alone cannot confirm bi-allelic deletion. Since most purified samples have been estimated to contain greater than 90% tumor cells, it appears that bi-allelic deletion of p18 occurs in no more than 2 to 3% of MM tumors compared to about 30% of HMCL.
p18 RNA and expression-based proliferation index (PI)
The prevalence of bi-allelic p18 deletion is higher in more proliferative MM tumors
In contrast to highly proliferative HMCL, MM tumors generally have a low proliferation index [13
]. To assess proliferation in MM tumors, we determined an expression-based proliferation index (PI) from the median expression of twelve genes that are associated with a proliferation signature (see Methods) [9
]. A context for the PI (which includes Ki67, a standard marker to determine the fraction of cycling cells) is provided by the fact that a PI>2 occurs in 97% of 33 HMCL, 70% of 30 relapsed MM tumors, 13% of 231 untreated tumors, and none of 16 normal plasma cell samples (Table ). As shown in Table , all three MM tumor samples (P051, P057-2, and P303) with bi-allelic deletion of p18 had a PI>2. Only one other tumor sample (P057), derived from the same patient but at an earlier time than sample P057-2, had a PI>2; p18 gene content was nearly two-fold decreased in P057, suggesting the possibility that most tumor cells have mono-allelic deletion of p18, perhaps with some tumor cells have bi-allelic deletion, and also some contamination by normal cells. Significantly, none of eight tumors with a PI <2 and p18 RNA <0.55 had bi-allelic deletion of p18 (Table ). Including both untreated and relapsed MM tumors, there were 50 tumors with a PI>2. Six of these tumors have background levels of p18, and three of the four tumors analyzed have bi-allelic deletion of p18. Thus, the prevalence of bi-allelic deletion of p18 in more proliferative (PI>2) MM tumors is at least 6% but perhaps 10%, a prevalence that still would be lower than the 30% of HMCL.
HMCL and more proliferative MM tumors frequently over-express p18
As summarized in Table , the normalized expression of p18 RNA was >2 in twenty of 33 (60%) HMCL, and in 43 of 261 (16%) MM tumors, but in none of 16 normal PC samples. The increased expression of p18 in MM tumors was associated with increased proliferation since 5 of 151(3%) MM tumors with a PI<1, 8 of 60 (13%) MM tumors with a PI between 1 and 2, and 30 of 50 (60%) MM tumors with a PI>2 had normalized values of p18 >2. Therefore, unless the p18 gene has been deleted, the expression of p18 RNA usually is increased in HMCL and more proliferative MM tumors.
A proliferation signature is associated with a poor prognosis in many kinds of tumors, including MM tumors [15
]. Figure shows a survival curve of 596 individuals with MM tumors, including 559 individuals with untreated tumors and 37 individuals with relapsed tumors[16
]. There is a significantly reduced overall time of survival for individuals whose tumors have an expression based proliferation index greater than two (P < 0.0001). Given the strong correlation of a PI>2 with increased expression of p18INK4c, it is not surprising that the overall survival time of individuals with tumors having a normalized expression of p18>2 is also significantly reduced (P = 0.0005). In fact, we confirmed that PI is more important than p18 in determining survival since PI>2 with p18<2 also is associated with a significantly poorer prognosis, whereas p18>2 with PI<2 is not associated with a significantly poorer prognosis.
Figure 3 Effect of increased proliferation or increased p18INK4c expression on survival of multiple myeloma patients. Survival data and Affymetrix U133_2.0_Plus RNA expression data was obtained for 596 individuals with MM, including 559 newly diagnosed patients (more ...)
Transfected p18 inhibits proliferation of some but not all HMCL
To assess the effect of exogenous p18 on proliferation of HMCL that express different amounts of endogenous p18, we used a retroviral vector that co-expresses p18 from the 5' LTR and a puromycin-EGFP fusion product from an internal CMV promoter[17
]. We retrofected four HMCL: KMS-12PE and OPM-2 have bi-allelic deletion of p18; JIM-3 expresses a low level of p18; and L363 expresses a high level of p18 (Table ). Representative results from these experiments are shown in Figure . As assessed from Western blots, the level of expression of exogenous p18 in unselected OPM-2 (15% EGFP positive cells) indicated that infected cells express exogenous p18 at a level that is comparable to the endogenous level of p18 in L363 (Fig. ). The level of endogenous plus exogenous p18 in puromycin selected (EGFP+) L363 cells appears to be almost twofold higher than the level of endogenous p18 in L363 (Fig. and other results not shown).
Figure 4 Retrofection of p18 into four HMCL. Representative results are shown for retrofection of four HMCL with pVPG-0 or pVPG-p18 vector. A. Western blot containing extracts from uninfected or infected HMCL. Dilutions of L363 are shown to compare endogenous (more ...)
To determine the effect of exogenous p18 on growth, the percentage of EGFP positive cells – for both the control and p18 vectors – was determined the day after infection (0 cell doublings), and at two additional times as the entire cell population went through a total of five doublings in the absence of puromycin selection (Fig. ). For all HMCL retrofected with the control vector, the fraction of EGFP positive cells increased somewhat during the first two doublings but then remained essentially stable (shown only for L363 in Fig. ); this transient period of a relative increase of transfected cells probably is explained by the fact that retroviruses efficiently infect the most healthy, proliferating cells. The growth properties of L363 cells infected with the p18 and control vectors were indistinguishable, indicating no effect of exogenous p18 on cell growth. However, for the other three HMCL that were infected with the p18 vector, the fraction of EGFP positive cells continuously decreased as the population underwent five doublings (Fig ). Assuming comparable viability for the transfected and non-transfected cells in each population, we have calculated (Methods) the approximate fractional growth inhibition by exogenous p18 for each of these three HMCL: 79% in a single experiment with OPM-2; 61 and 69% in two experiments with KMS-12PE; 52 and 54% in two experiments with JIM-3.
For L363 and OPM-2, we also did a FACS analysis to determine the DNA content as a measure of cell cycle distribution of EGFP positive cells. For L363 cells examined two population doublings after infection, the fraction of EGFP positive cells with G0/G1 and S DNA content, respectively, was 0.42 and 0.36 with the control vector, and 0.44 and 0.31 with the p18 vector. In marked contrast, for OPM-2 cells examined two population doublings after infection, the fraction of EGFP positive cells with G0/G1 and S DNA content, respectively, was 0.42 and 0.35 with the control vector, but 0.85 and 0.05 with the p18 vector. Therefore, unlike the high p18 expressing L363 where exogenous p18 had no inhibitory effect, the cell cycle and proliferation analyses of p18 null OPM-2 cells are consistent with exogenous p18 markedly inhibiting progression from G1 to the S phase of the cell cycle, thereby causing a substantial inhibition of cell proliferation.