Heterogeneity in nuclear transport does not affect the timing of DNA synthesis in quiescent mammalian nuclei induced to replicate in Xenopus egg extracts.

Intact G0 nuclei from quiescent mammalian cells initiate DNA synthesis asynchronously in Xenopus egg extracts, despite exposure to the same concentration of replication factors. This indicates that individual nuclei differ in their ability to respond to the inducers of DNA replication. Since the induction of DNA synthesis requires the accumulation of replication factors by active nuclear transport, any variation in the rate of transport among nuclei could contribute to the variability of DNA replication. Using the naturally fluorescent protein allophycocyanin (APC) coupled with the nuclear localization sequence (NLS) of SV40 T antigen, as a marker of nuclear uptake, we show here that individual G0 nuclei differ in their rate of transport over a range of more than 20-fold. Surprisingly, this variation has no direct influence on the timing or extent of DNA synthesis. Similar results were obtained by monitoring the uptake of nucleoplasmin, a nuclear protein present at high levels in egg extracts. These experiments show that the initiation of DNA synthesis is not driven merely by the accumulation of replication factors to some threshold concentration. Instead, some other explanation is needed to account for the timing of initiation.

Induction of DNA synthesis or apoptosis in mammalian nuclei by Xenopus egg extracts that fail to support the replication of sperm chromatin.

Cell-free extracts of Xenopus eggs are widely used to study DNA replication in higher eukaryotes. However, with Xenopus sperm chromatin as template, it is a common experience that the efficiency of replication varies from extract to extract, for reasons that are unclear. Here we show that the majority of extracts unable to support sperm replication are nevertheless able to induce DNA synthesis in intact mammalian nuclei. Those that do not, induce apoptosis.

The replication capacity of intact mammalian nuclei in Xenopus egg extracts declines with quiescence, but the residual DNA synthesis is independent of Xenopus MCM proteins.

In eukaryotes, the initiation of DNA synthesis requires the assembly of a pre-replicative complex (pre-RC) at origins of replication. This involves the sequential binding of ORC (origin-recognition-complex), Cdc6 and MCM proteins, a process referred to as licensing. After origin firing, the Cdc6 and MCM proteins dissociate from the chromatin, and do not rebind until after the completion of mitosis, thereby restricting replication to a single round in each cell cycle. Although nuclei normally become licensed for replication as they enter G(1), the extent to which the license is retained when cells enter the quiescent state (G(0)) is controversial. Here we show that the replication capacity of nuclei from Swiss 3T3 cells, in Xenopus egg extracts, is not lost abruptly with the onset of quiescence, but instead declines gradually. The decline in replication capacity, which affects both the number of nuclei induced to replicate and their subsequent rate of DNA synthesis, is accompanied by a fall in the level of chromatin-bound MCM2. When quiescent nuclei are incubated in egg extracts, they do not bind further MCMs unless the nuclei are first permeabilized. The residual replication capacity of intact nuclei must therefore be dependent on the remaining endogenous MCMs. Although high levels of Cdk activity are known to block MCM binding, we show that the failure of intact nuclei in egg extracts to increase their bound MCMs is not due to their uptake and accumulation of Cdk complexes. Instead, the failure of binding must be due to exclusion of some other binding factor from the nucleus, or to the presence within nuclei of an inhibitor of binding other than Cdk activity. In contrast to the situation in Xenopus egg extracts, following serum stimulation of intact quiescent cells, the level of bound MCMs does increase before the cells reach S phase, without any disruption of the nuclear envelope.

Individual nuclei differ in their sensitivity to the cytoplasmic inducers of DNA synthesis: implications for the origin of cell cycle variability.

Nuclei of multinucleate cells generally initiate DNA synthesis simultaneously, suggesting that the timing of DNA synthesis depends upon the appearance of a cytoplasmic signal. In contrast, intact nuclei from quiescent mammalian cells initiate DNA synthesis asynchronously in cell-free extracts of Xenopus eggs, despite the common environment. Here we show that the two nuclei of permeabilized binucleate cells enter DNA synthesis coordinately in egg extracts, as they do in vivo, with different pairs of nuclei initiating replication at different times. This indicates that the two nuclei of a binucleate cell are identical in their sensitivity to the inducers of DNA synthesis in egg extracts; this sensitivity varies in general between the nuclei of unrelated cells. The asynchrony of DNA synthesis shown by unrelated nuclei in egg extracts is therefore not an artifact of the cell-free system but a reflection of genuine differences preexisting within the intact cell. Evidence that these differences between nuclei are responsible for a substantial fraction of G1 variability in living cells is presented.

Initiation of DNA synthesis by nuclei from scrape-ruptured quiescent mammalian cells in high-speed supernatants of Xenopus egg extracts.

Demembranated sperm heads, detergent-isolated somatic nuclei and even naked DNA are efficiently replicated in cytoplasmic extracts of activated amphibian eggs, but only after nuclear assembly and the formation of an intact nuclear envelope. DNA synthesis has not previously been shown to be initiated in high-speed (200,000 g) supernatants of egg cytoplasm because they are depleted of the vesicular material required to support nuclear envelope formation. Here we show that mammalian nuclei prepared by scrape-rupture are able to initiate DNA replication in such high-speed supernatants. These nuclei begin DNA synthesis asynchronously. This asynchrony cannot be attributed to differences in the time taken for nuclear assembly. Instead, we suggest that the asynchrony reflects intrinsic differences between nuclei and that these differences are a major cause of cell cycle variability. Our demonstration of initiation in high-speed supernatants now enables the initiation of eukaryotic DNA synthesis to be studied independently of nuclear assembly.

In vitro regulation of HGF-SF expression by epithelial-mesenchymal interactions.

Studies of parameters which affect cellular proliferation, cellular differentiation together with studies of cell-cell interactions which affect cell behavior are particularly interesting in that they can be used to identify and characterise molecules which through changes in gene expression induce/inhibit cell proliferation, differentiation and movement. Such studies are crucial not only in the context of understanding growth and development, but also in understanding the processes of wound healing and regeneration as well as tumor invasion and metastasis. Here we present a summary of some cell culture models which we have developed for the study of the above-mentioned phenomenon, together with their application to the studies of regulation of HGF-SF expression.

Failure of platelet-derived growth factor plus insulin to stimulate sustained proliferation of Swiss 3T3 cells. Requirement for hydrocortisone, prostaglandin E1, lipoproteins, fibronectin and an unidentified component derived from serum.

Platelet-derived growth factor (PDGF) has been reported to be a potent mitogen for Swiss 3T3 cells made quiescent by growth to saturation density in high serum, and its activity is further potentiated by high levels of insulin, which alone have little effect. We show here that this is not the case for sparse 3T3 cells made quiescent by plating in low serum. Under these conditions, insulin alone is at least as effective as PDGF and frequently more so. Together, the response is no more than additive at best, and in many cases less than additive, the combined effect being no greater than for insulin alone. Instead, we find that optimal mitogenic stimulation requires the additional presence, besides PDGF and insulin, of hydrocortisone, prostaglandin E1 and an unidentified, non-dialysable component contained in serum treated with dithiothreitol (DTT) to inactivate endogenous growth factors. Interestingly, overnight pretreatment of the cells with hydrocortisone alone potentiates the subsequent response to PDGF + insulin, i.e. pretreatment induces a long-term memory that persists after the removal of the hydrocortisone from the medium. In short-term (24h) thymidine incorporation assays, the combination of PDGF, insulin, hydrocortisone, prostaglandin E1 and DTT-serum, is as effective as optimal levels of whole serum, but is unable to sustain longer-term proliferation (measured over 6 days). For this, high- and low-density lipoproteins, fibronectin and, to some extent epidermal growth factor (EGF), are also necessary.

Expression of markers of differentiation in a transformed human keratinocyte line induced by coculture with a fibroblast line.

SVK14, an SV40-transformed human keratinocyte line, has previously been reported to be almost completely unable to differentiate, and indeed, to express a set of keratins characteristic of simple epithelia rather than the stratifying epithelium from which they were derived. We have recently shown that IGF I stimulation of SVK14 results in expression of keratin 14, a marker of stratifying epithelia, as well as expression of markers which are characteristic of differentiation in normal human keratinocytes such as involucrin and keratin 10. To study further the capacity of SVK14 to differentiate, we have cocultured SVK14 with a variety of fibroblastic cell lines with a view to examining whether the cocultured partner can promote or interfere with their differentiation. We have observed that SVK14, when cocultured with Swiss 3T3, form organized structures through specific cell-cell interactions in which SVK14 express keratins 14 and 5 and involucrin, while maintaining T-antigen expression. These results are interesting since they show coculturing of a transformed human keratinocyte cell line and a particular fibroblast line can result in induction of characteristics of stratifying epithelia in a cell line with characteristics of simple epithelia. This may be analogous to the epithelial-mesenchymal interactions seen during epithelial development in the very early embryo.

A method for removal of fibroblasts from human tissue culture systems.

The phenomenon of fibroblast overgrowth is one of the major problems encountered during long-term culture of more slowly growing specialized cell types. A cell surface glycoprotein, Thy-1, which was originally found to be present on murine T-lymphocytes and brain cells, is also found to be present on only a few human cell types, mainly fibroblasts and neuronal cells. We have taken advantage of this fact, using a solid-phase immunoadsorption technique termed "panning", to rid our culture system (normal human keratinocytes) of contaminating dermal fibroblasts. A mouse monoclonal antibody raised against human brain Thy-1 was used to attach dermal fibroblasts to a goat anti-mouse immunoglobulin-coated plastic surface. By this method we were able to separate a 1:1 mixture of human dermal fibroblasts and keratinocytes with greater than 97.5% efficiency. Furthermore we have successfully removed dermal fibroblasts from naturally arising contaminated keratinocyte cultures, where the proportion of fibroblasts (less than 10%) was considerably less than that of the artificially mixed populations. These results compare favorably with those expected of the fluorescence-activated cell sorter (FACS) method of cell separation. In addition this technique is comparatively simple and inexpensive and is thought to be of use to other primary tissue culture systems (especially human) where contamination and subsequent overgrowth with fibroblasts remains a problem.

IGF I induces differentiation in a transformed human keratinocyte line.

A comparison of normal epithelial cells with their transformed counterparts could lead to the definition of parameters related to growth and differentiation which are altered by viral transformation and which may be relevant to malignant changes in vivo. Using the SV40-transformed human keratinocyte line, SVK14, which exhibits characteristics of simple, nonkeratinizing epithelia, we have shown that IGF I stimulation of these cells results in extensive multilayering, increased cell size, accumulation of involucrin, modulation of keratin 18 and expression of keratins 14 and 10, whilst T-antigen expression is maintained in the multilayered cells. Since T-antigen expression is correlated directly with impairment of stratification and differentiation, it is interesting that treatment of SVK14 with a single growth factor. IGF I, results in molecular events characteristic of differentiating normal keratinocytes.

The 3T3 cell cycle at low proliferation rates.

When the proliferation rate of Swiss 3T3 cells is decreased by limiting the availability of growth factors, cell cycle variability increases, as predicted by the transition probability model. Nevertheless, the transition probabilities would appear to play a relatively minor role in the regulation of proliferation rate. Instead, at least 40% of the increase in the average cycle time is brought about by an elongation of the minimum cycle time (i.e. the 'deterministic' part of the cycle). In addition, we have found that a substantial proportion of the cells (roughly 20%, in the present experiments, for doubling times of the order of 35-40 h) drop out of cycle in each generation, leading to a growth fraction of less than 1.0. The non-dividing cells, which we have previously shown to remain capable of division, would seem to support the existence of a Go state outside the normal cell cycle, and distinct from the indeterminate states postulated by the transition probability model. Because of the generation of nondividing cells at low proliferation rates, the log alpha and beta plots (distributions of cycle times, and sibling cycle time differences, respectively) are markedly concave, with a continuously decreasing slope. The transition probabilities cannot therefore be estimated directly and it is impossible to determine the extent to which they contribute to the regulation of proliferation rate. Rather, our data suggest that the transition probabilities are not uniform throughout the population under these conditions, but vary substantially from cell to cell. In addition to the changes in cell cycle kinetics, we also report an increased failure rate of cytokinesis, at low proliferation rates, leading initially to the appearance of binucleate cells. Such failures of cytokinesis may be responsible for the well-known rise in the incidence of binucleate and polyploid cells in the liver, with age.

Differences in growth factor sensitivity between individual 3T3 cells arise at high frequency: possible relevance to cell senescence.

At low serum concentrations (3% or less), individual Swiss 3T3 cells display marked heterogeneity in proliferative capacity. Here we show that this heterogeneity arises at extremely high frequency within a clone, often with sister cells showing considerable differences in capacity for further proliferation. The heterogeneity is unlikely to be due to genetic instability or mutation. Instead, it appears to reflect physiological differences between cells in their requirement for serum growth factors. It is suggested that these differences arise because cells are unable to sustain production, at low growth factor concentrations, of some rare component which is itself required for growth factor action. We believe that the generation of heterogeneity in 3T3 cells has much in common with the phenomenon of senescence in diploid cells.

Cell growth, cell division and cell size homeostasis in Swiss 3T3 cells.

By separating large and small 3T3 cells we show here that cell growth (in volume) after stimulation from quiescence is not 'autocatalytic'. Rather, large cells grow significantly more slowly, in relative terms, than small cells. It follows that 3T3 cells do not require a size control mechanism operating at the level of division timing in order to achieve cell size homeostasis.

Apparent desensitization of Swiss 3T3 cells to the mitogens FGF and vasopressin.

The growth factors FGF and vasopressin were found to have only a transient effect on confluent quiescent monolayers of Swiss 3T3 cells. Whether measured as cumulative entry into S-phase by autoradiography, or as cell division by time-lapse filming, the elevated rate of cell proliferation was maintained only over 10-15 hr. Several trivial or artifactual explanations for this transience were ruled out, including toxicity of 3H-thymidine; exhaustion or degradation of medium components, nutrients or growth factors (although some medium depletion was observed); and the generation during quiescence of cells incapable of division. We have also eliminated heritable variation in the capacity to respond to individual growth factors. However, unstable phenotypic heterogeneity in growth factor requirements between cells may play some part, as found elsewhere for the response to low concentrations of serum (Brooks et al, 1984). Cell populations that had ceased to respond to vasopressin recovered their sensitivity after 2-3 days' incubation in conditioned medium lacking vasopressin. The phenomenon thus resembles the mitogen-induced desensitization described by Collins and Rosengurt (1982, 1983). However, in our case, the loss of sensitivity was not selective for vasopressin but applied also to epidermal growth factor (EGF) and to prostaglandin F2 alpha. Furthermore, changes in responsiveness to vasopressin with time were associated with changes in cell density. Although some element of selective desensitization has not been ruled out, the transient response in our experiments can be accounted for in terms of unstable heterogeneity in growth factor requirements and/or in terms of density-dependent regulation of growth.

Apparent heterogeneity in the response of quiescent swiss 3T3 cells to serum growth factors: implications for the transition probability model and parallels with "cellular senescence" and "competence".

When subconfluent, Swiss 3T3 cells made quiescent by serum deprivation are stimulated with low concentrations of serum (ca. 1%), only a proportion of them (roughly 50%) enter S phase despite daily replacement with fresh, low-serum medium. The cells that fail to enter S phase are not incapable of doing so, since most of them initiate DNA synthesis after transfer to 10% serum. It would appear that individual cells vary in their growth factor requirements. Using time-lapse cinemicroscopy a few of the cells that respond to low serum were seen to give rise to several generations of progeny, while the majority of cells failed to divide at all, or divided once at most. Despite this, differences between cells in growth factor requirements do not seem to be heritable in the long term, since attempts to enrich for responding cells by prolonged culture in 1% serum have been unsuccessful. Rather, it would appear that the capacity to respond to low serum is an unstable property lost after a few generations in low serum. The loss of responsiveness shows parallels with "cellular senescence" and could conceivably result from decay of the platelet-derived growth factor-induced state of "competence." But regardless of why some cells respond to low serum while others do not, it is clear that the kinetics of entry into S phase after serum stimulation of quiescent 3T3 cells are not strictly first-order, since the labelling index plateaus after roughly 3 days at values substantially below 100%. As such, the kinetics, though not contradicting the transition probability model, cannot be taken to support it as was previously thought.

The G1 distribution of "G1-less" V79 Chinese hamster cells.

The V79-8 line of Chinese hamster cells has been reported to lack a measurable G1 phase. However, using a combination of time-lapse cinemicroscopy and [3H]thymidine autoradiography, we have found these cells to have a median G1 duration ranging from 1.4 to 2.6 h in different experiments, accounting for more than 15% of the median cycle time. The youngest cell labelled (in seven experiments) was 0.73 h old at the time of fixation suggesting a minimum G1 of between 0.40 and 0.73 h (the duration of the [3H]thymidine pulse being 0.33 h). In those experiments where steady-state proliferation could be established unequivocally, variability in G1 times accounted for all of the variability in cycle times. In addition, the distribution of G1 times (and cycle times) was well described by the two-transition version of the transition probability model. Nevertheless, changes in the average duration of G1 (and hence changes in the transition probabilities) played a comparatively minor role in determining proliferation rate. Instead, the length of S + G2 was markedly influenced by the composition of the culture medium. For purposes of comparison with the 'G1-less' V79-8 line, we have also examined a revertant derived from it (G1+5c) reported to have regained a substantial G1 phase. We are able to confirm that its G1 is indeed longer, the youngest labelled cell being 2.48 h old at the time of fixation. Unlike the parent line, there appeared to be more variability in G1 times than could be explained by two random transitions alone. The proliferation rate of the G1+5c revertant was unusually sensitive to the composition of the culture medium, suggesting the possibility of a metabolic defect.

Microtubule-organizing centres during the cell cycle of 3T3 cells.

When Swiss 3T3 cells are treated with colcemid for 60 min (0.5-2 micrograms/ml), although the cytoplasmic network of microtubules is considerably reduced, a significant but variable number of microtubules remain. These residual microtubules appear to converge on a single, usually ciliated, organizing centre. In contrast, on removal of the colcemid, microtubule regrowth takes place from two organizing centres in approximately 50% of the cells. Surprisingly, whether a cell has one or two organizing centres during recovery does not depend on its position in the cell cycle, though recovery takes place earlier in serum-stimulated cells than in serum-starved quiescent ones. It seems likely that prior to colcemid treatment, cells contain two potential organizing centres only one of which is normally active, the inactive one becoming active after exposure to the drug.