Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The c-myc proto-oncogene is activated in a wide variety of human cancers. Like the related genes N-myc and L-myc, it encodes a transcription factor that dimerizes with the Max protein (1). The c-Myc/Max complex stimulates cell growth and counteracts differentiation. Myc proteins compete for Max with Mad and Mad-related proteins, which form complexes that promote differentiation and antagonize growth. Because Max is an abundant and stable protein, the levels of competing Myc and Mad families of proteins determine the ratio of growth stimulatory to growth inhibitory complexes. This delicate balance can be upset by events that increase the concentration of Myc protein, such as gene amplification, chromosome translocation, viral insertion, messenger RNA (mRNA) stabilization, and protein stabilization (2). These events oncogenically activate genes of the myc family by merely increasing the levels of an otherwise normal protein, thereby leading to excessive proliferation. Amplification of the myc gene is particularly common in breast cancer, neuroblastoma, and lung cancer. In lung cancer, amplification of c-myc, N-myc, or L-myc genes is seen in approximately 33% of cell lines and 15% of all lung tumors and is most often found in cancers from patients who have been treated with chemotherapy, suggesting that this is a late event in the development of the cancer (3). However, increased myc expression is seen in a much higher percentage of cases (over 80% of cell lines and 45% of tumors), often in the absence of gene amplification (3, 4). In the majority of such cases, the mechanism leading to overexpression remains unknown. One possible pathway might be through mRNA stabilization.

The c-Myc protein is unstable and is encoded by very labile mRNA (5), which allows tight regulation of the gene. N-myc mRNA appears to be similarly labile. Not much is known about the half-life of L-myc transcripts. Stabilization of c-myc transcripts has been observed in certain cancers, most notably in cancers of the hematopoetic system such as Burkitt's lymphoma (6). In most cases, the stabilized myc transcripts seen in the cancer cells are abnormal in that they have either lost sequences (from the 5' or 3' untranslated regions) or gained sequences (due to the addition of foreign genetic material, usually to the 5' untranslated region). In contrast, unrearranged myc mRNA does not appear to be frequently stabilized, as determined from the analysis of over 20 tumor cell lines from a variety of tissues, which all showed rapid myc mRNA degradation (7). Alterations in trans-acting factors leading to the stabilization of specific mRNAs has been observed sporadically in certain cancers (11, 12). Most recently, stabilization of N-myc mRNA in a neuroblastoma cell line was shown to correlate with the presence of an RNA-binding protein (13).

Two observations suggest myc mRNA decay may be altered in lung cancer. In small cell lung cancer (SCLC) cell line GLC4, the shut-off of immediate early genes after induction by growth factors was severely delayed. In these cells, c-fos and c-jun mRNAs disappeared slowly after the peak of induction, whereas in colon cancer cell line CX1 these mRNAs decreased rapidly (14). Although the investigators did not determine the mechanism of the abnormal pattern of gene shut-off in GLC4 cells, it could be caused by a reduction in the rate of degradation of these labile mRNAs. In two other SCLC cell lines, abnormal half-lives of c-myc and N-myc mRNAs were seen (15). (L-myc half-life was also measured in the latter study; however, since the normal half-life of L-myc is unknown, it is unclear whether the transcripts were stabilized in these cell lines.) To determine whether abnormal mRNA decay is indeed prevalent in lung cancer, we studied the c-myc and N-myc mRNA decay patterns in 11 lung cancer cell lines. We found that mRNA decay patterns appear to be abnormal in many of the lines analyzed. Our results suggest that abnormal mRNA degradation may be a common phenomenon in lung cancer cell lines.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Cell Culture and RNA Isolation

All cell lines used were of human origin. Cell lines HEL299, NCI-H146, NCI-H446, and A427 were obtained from American Type Culture Collection (ATCC; Rockville, MD). NCI-H60 was provided by Dr. Alan Epstein (University of Southern California). NCI-H69, NCI-H82, NCI-H345, NCI-N417, NCI-H526, NCI-H125, and A549 were provided by Dr. Katherine Rich and Dr. Kristin Skinner (University of Southern California), who had obtained them from ATCC. (The NCI-derived cell line names are shortened in the text by omission of the NCI prefix.) HeLa cells were provided by Dr. Cheryl Wellington (University of British Columbia). Cells were grown in 250-ml flasks or on dishes in Dulbecco's modified Eagle's medium or RPMI 1640 (Life Technologies, Gaithersburg, MD) supplemented with 10% fetal calf serum, depending on each cell line's requirements. RNA was isolated from approximately 10⁶ logarithmically growing cells per RNA sample for cells growing in flasks or from one plate containing ~ 10⁶ cells for adhering cells. Transcription was blocked using the drug actinomycin D (2.5 mg/ml stock in water, used for no more than 14 d and stored at 4°C; Sigma Chemical Co., St. Louis, MO), which was added to plates and flasks to a final concentration of 5 µg/ml. RNA was isolated at times 0.5, 1, 2, and 4 h after actinomycin D addition, using the NP-40 method (16) or using TRIZOL (Life Technologies) for cell lines in which RNA degradation was a problem. For growth factor induction experiments, RNA isolation was preceded by a 48-h growth starvation in medium containing 0.5% newborn calf serum followed by the addition of 20% fetal calf serum (prewarmed to 37°C). RNA was isolated at times 0.25, 0.5, 1, 2, and 4 h after serum addition. Actinomycin D was added 15 min after serum addition to a parallel set of plates.

Northern Blot Analysis

Aliquots of 20 µg of each RNA sample were fractionated on gel, blotted, and hybridized as described previously (17). The probes used were: c-myc, 1.3 kilobase (kb) human exon 3 ClaI-EcoRI fragment; c-fos, 0.25 kb human complete exon 2 fragment generated from a genomic clone by polymerase chain reaction; N-myc, 0.9 kb human 3' untranslated region (UTR) probe, kindly provided by Dr. Robert Ross (Fordham University, NY); elongation factor (hEF), 1.3 kb human complementary DNA (cDNA) (18); histone 2B (H2B), 1 kb Xenopus laevis XhoI-BglII fragment. Membranes were wrapped in thin plastic wrap and exposed to PhosphorImager screens (Molecular Dynamics, Sunnyvale, CA) for at least 18 h. The screens were scanned and the bands quantitated using the Molecular Dynamics ImageQuant Software. Differences in the amount of RNA loaded were corrected by using the signal from the very stable hEF mRNA as a reference (18). The percentage mRNA remaining for each cell line was plotted semilogarithmically against the time in hours, and linear regression analysis was used to determine the mRNA half-life.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

To determine whether deregulation of myc genes in lung cancer might be occurring at the level of mRNA decay, the decay patterns of c-myc and N-myc mRNAs were analyzed in lung cancer cell lines. Lung cancer cell lines were chosen as a model system because metabolically active cells are required to measure mRNA decay rates. Multiple lung cancer cell lines have been established over the years (19, 20), and they have been found to stably maintain many of the features of the original tumor. Clinically, lung cancers are grouped into two categories: SCLC (also called oat cell carcinoma), characterized by its neuroendocrine characteristics and aggressive spread throughout the body, and non-small cell lung cancer (NSCLC), consisting of adenocarcinoma, squamous cell carcinoma, large cell carcinoma, and others. For our studies, eight SCLC lines and three NSCLC lines were used (Table 1). In addition, unimmortalized human embryonic lung (HEL) cells (HEL299) were studied to determine the c-myc mRNA half-life in noncancerous lung cells. Human cervix carcinoma cell line HeLa, a tumor cell line expressing high levels of normal, unstable c-myc mRNA and used previously by us and others for mRNA decay studies (10, 17), was analyzed for comparison. The identity and expression level of the myc family members in each cell line were first determined by Northern blot analysis (Figure 1). Poly(A)⁺ mRNA from adult human brain and lung was included for comparison. Nine of the 11 lung cancer cell lines expressed c-myc, whereas two of them (H69 and H526) expressed only N-myc. No L-myc expression was seen in any of the cell lines (data not shown). The apparent differences in mobility of the bands are an artifact caused by "pouting" of the gel. Note that c-myc mRNA is present in low amounts in adult lung (Figure 1, last lane), whereas no N-myc mRNA is detectable. The same is true for HEL cells. In comparison, c-myc or N-myc expression is elevated in all lung cancer lines.

View larger version (114K): [in this window] [in a new window]   Figure 1.   Northern blot analysis of myc gene expression in lung cancer cell lines and controls. All lanes contain 20 µg RNA except for the last two lanes, which contain 1 µg poly(A)+ RNA from adult human brain and lung (Clontech, Palo Alto, CA). The figure is a composite of two blots (Masterblot 1 and 2) containing RNA from all the cell lines. The two blots were hybridized, washed, and exposed simultaneously with identical probes at all times. The filters were probed consecutively with exon 3 probes for c-myc, N-myc, and L-myc, and with the human cDNA for hEF to control for loading differences. The position of the 18S ribosomal RNA band is indicated by "r."

The half-life of the myc transcripts in the cell lines was determined by blocking transcription with actinomycin D followed by RNA isolation at different times after addition of the drug (Figure 2A). (Although there have been reports of actinomycin D itself causing a retardation in mRNA decay, these effects are seen mainly when transfected constructs encoding chimeric mRNAs are used and when cells are incubated with the drug for many hours. Our previous work [6, 17] and the work of many others [7, 10] show that at least for c-myc and c-fos, actinomycin D can be used reliably to obtain half-life values for endogenous mRNAs and that these values differ very little from values obtained through alternative methods such as approach to steady-state labeling or transient promoter induction.) The half-life of the myc transcripts in each cell was determined by plotting the percentage myc RNA remaining against the time (Figure 2B). The results for three independent experiments are summarized in Table 2. HEL299 cells produced the most unstable c-myc mRNA, with a half-life of 13 ± 6 min, followed by HeLa cells (19 ± 3 min). These values agree with published values of 15 to 20 min for the half-life of normal c-myc mRNA, as measured by a variety of techniques (6, 10). NSCLC cell lines H125 and A549 and SCLC line H345 expressed c-myc transcripts that were almost as unstable as those seen in HeLa cells. SCLC lines H146 and H446 produced transcripts that appeared to be marginally more stable than normal, whereas cell lines H60, H82, and N417 clearly showed a reduced rate of c-myc mRNA decay. Stability of c-myc mRNA had been previously measured in cell line H82 and had also been found to be increased, albeit to a slightly lesser extent (2.5-fold) (15).

View larger version (59K): [in this window] [in a new window]   Figure 2.   Decay analysis of myc mRNA. (A) Cells were treated with actinomycin D to a final concentration of 5 µg/ml for the indicated times. RNA was isolated and subjected to Northern blot analysis. Filters were hybridized with c-myc or N-myc probes followed by an hEF probe to control for RNA loading. All lines were analyzed at least three independent times. (B) Graphic representation of the data in A. The percentage of RNA remaining was plotted semilogarithmically against the time. Half-lives were calculated by linear regression. The c-myc decay patterns in all the cell lines are shown in the top two graphs, with cell lines H125 and A549 included in both graphs as a reference. The bottom graph shows the N-myc decay pattern in cell lines H69 and H526, with cell line H125 (expressing c-myc) included for comparison.

Cell lines H69 and H526 exhibited an N-myc mRNA half-life of approximately 173 and 52 min, respectively. Although the normal half-life of N-myc transcripts is not known (because to our knowledge the decay pattern of this mRNA has only been studied in cancer cells), N-myc mRNA has been observed to be quite unstable, with a half-life of about 30 min in a neuroblastoma and a retinoblastoma cell line (9), and 6 min in a different neuroblastoma cell line (21). Longer half-lives have been reported in two murine embryonic carcinoma cell lines (60 and 130 min) (22), and in SCLC line H249 (containing amplified N-myc; 72 min) (15). However, because these latter cell lines are cancer cell lines, they could be examples of N-myc gene activation through mRNA stabilization. Because N-myc encodes a protein with a half-life similar to that of c-myc, it is most likely that the stability of N-myc mRNA is similar to that of c-myc transcripts (1). If this is the case, H69 and H526 should be considered examples of cells carrying stabilized N-myc mRNA.

To determine whether the observed reduction in c-myc and N-myc mRNA decay in the lung cancer cell lines was unique to myc transcripts, we studied the decay pattern for a second growth factor inducible gene, c-fos, encoding a very labile transcript. Both myc and fos mRNAs carry (A+U)-rich elements (AREs) in their 3' UTRs, which target them for rapid decay (23). This is thought to occur by the binding of factor(s) to the (A+U)-rich sequence, resulting in the inhibition of the mRNA decay pathway. Because c-fos expression was very low in these cell lines (data not shown), the gene was induced by growth factor stimulation. The resulting transient expression from the c-fos promoter provides an opportunity to follow the pattern of mRNA decay in the absence of actinomycin D (24). To control for possible alterations in the transcription machinery of the various cell lines, which could result in an abnormal shut-off of the c-fos promoter, the pattern of mRNA disappearance after the peak of induction was also analyzed in the presence of actinomycin D, added 15 min after the addition of growth factors. Cell lines were serum-starved for 48 h, whereupon c-fos transcription was induced by the addition of medium containing 20% fetal calf serum. The pattern of c-fos mRNA induction was analyzed by isolating RNA at various times and subjecting it to Northern blot analysis.

The 11 lung cancer cell lines fell into three groups based on their pattern of c-fos mRNA induction/decay. H125 and A549 cells showed the common c-fos induction pattern (25), with mRNA levels peaking at 30 min and rapidly declining thereafter, leaving no visible c-fos mRNA at 4 h after induction (Figure 3A). Shortening of the mRNA, which has been demonstrated to be due to the poly(A) tail loss that is the first and rate-limiting step of c-fos mRNA degradation (25), was clearly visible at 30 and 60 min after induction. The pattern of induction was not noticeably affected by the addition of actinomycin D at t = 15 min, except for slightly reduced levels of mRNA, which is to be expected because some transcription of the c-fos gene still occurs at 30 min after induction. HEL299 cells showed a very similar pattern of fos expression (data not shown). In contrast, the pattern of induction was delayed in cell lines H60, H82, H146, H446, H69, and H526 (Figure 3B). Induction levels were lower in most cells, but more importantly, the decline in mRNA levels was obviously slower. In all cases (albeit some quite weakly), c-fos mRNA was visible at 4 h after induction (Figure 3D). Interestingly, mRNA shortening appeared to be delayed or absent in these cell lines. These observations suggest that c-fos mRNA decay is perturbed in H60, H82, H146, H446, H69, and H526. These same cell lines showed myc mRNA half-lives of over 30 min. Our H345 stock was lost at the time of these experiments, but a growth factor induction without actinomycin D, showing abundant c-fos mRNA at 4 h after induction (our data, not shown, and Reference 4), indicated that the induction pattern was abnormal in this cell line as well. The last two cell lines, N417 and A427, did not show any induction of the c-fos transcript by serum stimulation so that no clear conclusion could be drawn about the mRNA decay pattern. The abnormalities in myc and/ or c-fos mRNA decay observed in the lung cancer cell lines (summarized in Table 2) suggest that the mRNA degradation machinery is defective in seven of eight SCLC lines and in one of three NSCLC lines. However, the extent to which the decay of myc and fos mRNAs are affected appears to differ among the various cell lines.

View larger version (72K): [in this window] [in a new window]   Figure 3.   (A-C) Northern blot analyses of c-fos induction by serum stimulation in lung cancer cell lines. For each cell line, the c-fos pattern is shown at the top, and the corresponding hEF signal is shown directly below. The pattern of induction with serum alone (Growth Factor) is shown next to the pattern of induction seen when actinomycin D is added 15 min after the serum addition (GF+Act.D). (A) The two cell lines that showed normal induction patterns (H125 and A549). (B) Cell lines that showed delayed c-fos mRNA induction. (C) Cell lines showing no c-fos induction (A427 and N417). (D) Comparison of the decay patterns of cell lines from A and B (GF+Act.D part of curves).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The experiments described above suggest that the mRNA degradation machinery can be affected in lung cancer cell lines. Whether this phenomenon is also present in tumors is hard to determine because metabolically active cells are required for RNA half-life measurements, making such experiments very difficult. It is, however, clear from the analysis of a variety of cell lines from different organs (see beginning paragraphs) that myc mRNA stabilization is not a natural characteristic of tumor cell lines. The fact that it occurs in most of the lung cancer cell lines analyzed suggests that it could be a common phenomenon in lung cancer. The role of this stabilization in the development of lung cancer in vivo will be more easily clarified when the mechanism of the stabilization becomes clear. This mechanism could be: (1) mutations in the myc mRNAs themselves; (2) a general effect on the decay of all mRNAs; (3) a result of the amplification of the genes encoding the mRNAs (by titration of destabilizing factors); or (4) a specific defect in the degradation pathway of unstable proto-oncogene mRNAs.

The first possibility is unlikely because both fos and myc mRNAs appear to be affected. To address the second possibility, we analyzed the degradation pattern of H2B mRNA. This nonpolyadenylated mRNA is regulated in a cell cycle-specific fashion and is degraded by a mechanism different from that responsible for unstable proto-oncogene mRNA destruction (26). Cell cycle-regulated histone mRNAs such as H2B are most stable in early S-phase (with a half-life of up to 2 h) and most unstable in late S-phase/early G₂ (half-life as low as 10 min) (27). In unsynchronized growing cells, histone mRNA half-life may vary. H2B mRNA half-life was determined by hybridizing the blots used for the actinomycin D experiments (Figure 2) with an H2B probe. The half-life of the transcript varied considerably from blot to blot, even when the same cell line was analyzed. For example, it was found to be 66, 71, and 125 min in HEL299 cells, and 41, 52, and 92 min in H125 cells, both of which showed very reproducible and short c-myc mRNA half-lives. Such variability may have to do with the exact percentage of cells in S-phase in the population at a given time. Overall, although the H2B mRNA varied significantly between different blots, this pattern of variation did not correlate with the myc mRNA half-life, which was much more consistent in each cell line. This suggests that the observed abnormalities in c-fos, c-myc, and N-myc decay are not due to a general defect in the degradation of all mRNAs. The one possible exception could be cell line H69, which showed very slow N-myc decay and very stable c-fos mRNA, as well as an abnormally long H2B half-life of 2.5 to 4 h. It is unlikely that these effects were related to the possible inability of H69 cells to be transcriptionally inhibited by actinomycin D because the growth factor induction pattern (in the absence of actinomycin D) was also abnormal.

The observed stabilization of myc mRNA could also be explained by the fact that the large number of myc transcripts generated by gene amplification exceeds the capacity of the degradation machinery. This is unlikely for two reasons. First, there are several examples of cells containing highly amplified c-myc genes that show normal patterns of mRNA decay. COLO 320 cells, a colon cancer cell line with high c-myc gene amplification and expression, contains a normally labile myc mRNA (8). Similarly, HL60 cells, a human promyelocytic leukemia cell line with 20 to 40 copies of the c-myc gene, express abundant myc transcripts that have a half-life of 15 to 20 min (28). Second, during a normal growth factor stimulation, c-fos transcripts are highly induced, yet they are rapidly degraded. In contrast, in those SCLC cell lines showing abnormal fos induction patterns, fos mRNA levels are lower than normal, whereas decay is also slower.

The only remaining explanation for the observed disruptions in the mRNA decay machinery is that the lung cancer cell lines studied are specifically defective in the degradation of labile proto-oncogene mRNAs. Alterations in trans-acting factors leading to stabilization of growth-promoting mRNAs have previously been observed in cancers expressing cytokines, which are encoded by highly labile mRNAs. For example, certain hematopoetic cancers exhibit an autocrine growth stimulation mediated by a stabilized interleukin-3 transcript (11). Stabilization of the granulocyte-macrophage colony stimulating factor mRNA in a monocytic tumor was also shown to be due to a trans-acting factor (12). There are at least two mechanisms by which trans-acting factors could play a role in the disruption of ARE-mediated mRNA decay (23). One possibility is that the factors which normally bind to these AREs and promote decay are altered or absent, so that ARE-containing mRNAs are not properly targeted to the degradation machinery. Alternatively, the AREs could be occupied by abnormal proteins that block access to the natural ARE-binding factors. Very recently, the abnormal decay of apparently normal N-myc mRNAs was observed in a neuroblastoma cell line (13, 21). A fast growing and a slow growing variant of neuroblastoma cell line NBL-W exist. The fast growing cells express higher levels of N-myc, apparently resulting from a more stable mRNA. c-fos mRNA was also observed to be more stable in these cells (29). Increased N-myc and c-fos mRNA stability was correlated with the presence of RNA-binding proteins of the Hu family in the neuroblastoma cells (13, 29). Interestingly, expression of the normally neuronal Hu proteins is also a hallmark of SCLC (30). Thus, we hypothesize that these proteins might also play a role in abnormal mRNA decay in SCLC. The Hu family consists of four genes in humans (31), three of which are neuronal specific (HuD, Hel-N1/HuB, and HuC/ple21) and one of which is ubiquitously expressed (HuR). All four of these proteins can bind tightly to ARE sequences (31) and could thereby interfere with ARE-mediated mRNA decay. Indeed, overexpression of HuR in tissue culture cells by transient transfection was recently shown to inhibit the decay of an unstable reporter mRNA carrying an ARE (32). These results suggest that increased expression of Hu proteins can interfere with mRNA decay. We hypothesize that myc gene deregulation by Hu proteins could provide an explanation for early proto-oncogene activation in SCLC. myc gene activation is more prevalent in SCLC than in NSCLC and is found in almost all SCLC cell lines and in 65% of tumors (3). One possible model could be as follows: SCLC precursor cells, triggered to differentiate down the neuronal pathway, would begin expressing neuronal Hu proteins, which are very early markers for neuronal differentiation. These proteins could bind to AREs that would normally be occupied by destabilizing proteins, thus preventing the mRNAs from being properly destroyed. This stabilization could then lead to an increased expression of proteins encoded by ARE-containing mRNAs, such as MYC, perhaps providing an early stimulation of proliferation. Such proliferating cells would be targets for additional deregulatory events that could further increase MYC expression, such as gene amplification. Experiments are underway to determine whether Hu proteins might be involved in the abnormal decay patterns observed in the lung cancer cell lines.