There is good evidence that quantitative changes in gene expression are important for some regulatory genes. For example, periodic quantitative changes in the levels of cyclin proteins that activate cyclin dependent serine/threonine kinases (cdk's) appear to be the "core" cell cycle machinery in mammalian cells. In this case, cyclin molecular concentration is directly related to cdk activity and activity is directly related to cell cycle transition rates [17
]. When regulatory proteins affect cell processes involved in disease, measurement of molecular concentration takes on added importance. For example, the expression of a number of oncogenes, proto-oncogenes, and reduced or absence of expression of recessive oncogenes has been postulated to provide diagnostic and/or prognostic information in human cancers [e.g., [18
]]. Currently, cancer gene expression research is seldom done in a rigorously quantitative manner, which may contribute to the sometimes conflicting results [21
]. Much evidence leads to the conclusion that the tumorigenic phenotype of mammalian cells is markedly affected by quantitative changes in expression of regulatory genes. In experimental systems, small shifts in expression can have profound consequences. For instance, in NIH 3T3 cells, expression of SV40 large T antigen (Tag) at high levels (~2 × 106
molecules per G1
cell) decreases the G1
phase time by ~18% [3
]. However, ~43% of this effect can be accounted for with the initial expression of ~ 6000–14,000 molecules per G1
cell and ~87% can be accounted for by expression of ~ 103,000–133,000 molecules [3
Previously, we measured the relative dose response for G1
transit for SV40 Large T antigen expressed in NIH 3T3 cells. In this work, the levels of Tag in G1
increased 18% during exponential population growth as the length of G1
was increasing [11
]. Since Tag is rate-limiting for G1
transit in these cells, the increase in transit time as a function of G1
Tag concentration represented an apparent paradox that is resolved (in the simplest and obvious manner) by stating that additional rate-limiting entities operate to set the G1
transit time. One easily measured and pervasive entity that regulates G1
phase time is cell density, which is taken as a surrogate for cell-cell contact. This effect can be separated analytically from growth factor concentration [15
] and therefore, appears to measure independent cell cycle regulatory mechanisms. Our work [6
] and most other work indicates that the cell density dependent mechanisms that negatively regulate cell proliferation are dominant to Tag. Indeed this may be universal for oncogenes, since we are not aware of instances when anchorage dependent cells fail to enter a plateau phase in tissue culture.
The study presented in this paper represents further investigation into the effect of cell density on gene expression. Cell lines were created by incorporation of retroviral vectors into the genome and expression of Tag from either the retroviral LTR (considered to be a strong promoter) or the herpes virus thymidine kinase promoter, internal to the viral LTR [1
]. In our established NIH 3T3 cell line, Tag is expressed from the htk promoter at considerably lower levels than from the LTR for a considerable time after transduction [unpublished observations, and [11
]]. However, when a large number of htk-Tag astrocyte clones were checked 100 days after creation, the expression levels displayed a very large range, and thus we expect that this weak promoter is subject to positional effects on transcription or mutational events that affect gene expression (unpublished data). Polyclonal populations produced by transduction of htk-Tag vectors that were passaged without selection, displayed expression levels equivalent to or higher than LTR-Tag cell lines (Frisa and Jacobberger, unpublished data). This selection for increased Tag expression makes sense given the low growth rate of primary mouse astrocytes and the ability of Tag to decrease the cell cycle time [11
] and increase saturation density [e.g., [4
]]. In this study, the expression potential of cell lines was not related to proliferation related changes in expression of Tag, emphasizing that clonal variation for whatever reason (which includes promoter strength) accounts for the large range of htk related expression.
Making sense of a quantitative description of the role of Tag and other regulatory proteins in cell cycle regulation and cell transformation requires either that all the parameters affecting expression are measured at the same time so that differences can be accounted for, or an ability to compute some base line or "expression potential" statistic with which to evaluate differences (e.g., between cell lines). A practical application of this knowledge is the ability to create biological standards from cell lines that express varying levels of a given protein and then use these in quantitative and/or longitudinal experiments. Clearly, in cases where gene expression varies significantly with cell density (e.g., P0-3DtkT#5), use of these cells as a standard in a longitudinal study would require that the standard was measured at the same cell density for each experimental sampling, or that a derived statistic (Expression/Density or Expression/G1 intercepts) is used to normalize the data. In this regard, the good agreement between the intercepts and non-density corrected Western blot data for cell lines varying over a 20-fold range is supportive.
Originally, we pursued a strategy of reduced Tag expression from the htk promoter as a means of producing immortalized, differentiated cells that may have a reduced transformation phenotype. We noticed however, that initially slow growing cell lines produced with htk driven gene expression needed to be cloned early to prevent overgrowth of more rapidly growing cells that expressed high levels of Tag. This was not true for clonal or polyclonal transductants that expressed Tag from the MoMuLV LTR, which supports a high level of expression (unpublished observations). In the context of this study, we wished to explore density dependent expression as a function of these two types of transformants. The data presented here show that LTR driven Tag expressing cell lines either increase Tag during exponential growth despite a progressively increasing G1
phase and contact inhibition (density) or the levels to not change significantly, whereas those created with the htk promoter show varying degrees of reduced expression, that is consistent in magnitude between experiments, as the cell cycle lengthens. This suggests that the LTR is less density regulated than the htk promoter, which is subject to down regulation as the cell cycle lengthens during contact inhibition of growth. The single exception is the pattern of expression observed for an LTR driven line expressing a form of Tag that is mutant in binding the Rb family proteins [e.g., [22
]]. This density-dependent reduction inTag expression was somewhat surprising. We do not know the half life of this mutant T antigen (K1). Since it is expressed at levels that are 1–2 fold that of wild-type Tag in NIH 3T3 cells [23
], we would not expect that the half life was significantly lower than that of wild type. Alternatively, the cell line used here is a low expressing cell line (<2 fold above the lowest expressing cell line). Thus, in this cell line, for unknown reasons the protein could be less stable. If the idea that transcription strength mainly drives Tag expression is kept as the working hypothesis, then there are at least three possibilities to explain the K1 data: 1) the K1 protein has a mildly shorter half-life and the small difference results in a different phenotype vis-à-vis the density related expression; 2) failure to bind Rb family members significantly affects the half-life during exponential growth; 3) the LTR vectors can occasionally display integration position effects that in this special case was subject to density dependent control that is different from most LTR driven cell populations.
Both the promoters as well as the protein investigated here are viral. The cells are dependent on Tag expression, and so there is significant selective pressure against extinction of Tag expression. Effects of viral promoters may reflect the evolution of the viruses and their hosts. We have additional evidence that the same type of density effects (i.e., changes in gene expression detected continuously during exponential growth) may operate on native proteins. Tag immortalized human tracheal epithelial (HTE) cells grown at different densities and assayed for cytokeratins 6 and 18 showed a density dependent decrease in cytokeratin immunoreactivity and a concomitant increase in %G1
both in normal growth medium and in differentiation medium (Frisa & Jacobberger, unpublished). Additionally, the expression levels of cyclin B1 are lower in replicating Hela cells at high cell density (Soni & Jacobberger, unpublished results). It is conceivable that other proteins that are at high levels in density-arrested cells may be positively regulated by density during exponential growth. There are many papers that describe genes that are up or down regulated at confluence. Several recent papers identify genes that are up-regulated by cell density. These include TEM1/endosialin, IGF-1, IGF-1R, IGFBP-2, Bak, drp, SP1, p27Kip1
]. However, cell context differences are apparent [e.g. IGFR, [29
]], and one study used several cell lines and demonstrated distinct effects of the cell genetic background [25
]. Down-regulated genes include bFGF, FGF-2, Topoisomerase II-α, EGFR,, human HDAC1, as well as cell cycle regulatory genes among others [30
]. Regions in the bFGF promoter have been identified as associated with density-dependent down-regulation of bFGF [31
]. The IGF-II P3 promoter is down-regulated as a function of cell density, and a 7-base pair sequence has been identified that plays a critical role in this down-regulation [32
]. Neither the MoMuLV LTR or the htk promoters encode this sequence, although both encode a 10-base pair sequence that is identical to the P3 sequence with an insertion of 3 base pairs at position -1081. Most studies focus on transcription and/or protein levels. Fewer studies have addressed translation, however, iso-forms of c-myc and FGF-2 are translationally regulated by cell density [34
]. For many of these genes, down regulation at confluence is not surprising (e.g., cell proliferation associated genes), and upregulation of differentiation specific genes at confluence is equally unsurprising. However, a glance at the literature does not suggest that any quick classification or generalizations should be made.
Regulation of protein levels is multi-factorial. Transcription, translation, protein stability and degradation are involved. Tag levels in rapidly growing polyclonal cell lines appear to be set primarily by transcription [11
]. However, as discussed above, Tag levels in clonal cell lines produced with the weak htk promoter and generated from slowly growing primary cells are related to clonality. SV40 Tag is a long-lived protein with a half-life variously measured at 48–90 hrs. The long half-life of Tag would not normally implicate protein stability, however, the magnitude of changes that we discuss here, although biologically relevant, are within two fold and might be affected by small differences in protein stability. Given this line of reasoning, the data presented here can be interpreted as evidence for a cell contact (density) / cell cycle length related effect on transcription that does not affect the retroviral LTR to the extent that it does the htk promoter. This effect is predictive of the plateau growth phase levels of Tag and thus, ultimately should affect, in the case of Tag or similar oncogenes, the fate of clones in polyclonal populations. For example if an LTR Tag line and htk Tag line with equal expression potential were co-cultured, over time, one would expect the LTR driven line to dominate.
Finally, the htk lines here varied significantly in terms of Tag expression potential, and in terms of the rate at which expression decreased as a function of cell density. There was no correlation in the rate of change in density related expression and the density independent expression (here referred to as intrinsic expression potential). This supports the idea that the density independent expression in the htk lines is set by the clone specific effects (e.g., integration position) and the density/cell cycle related expression is set in these cases by the promoter.