Early response pattern analysis of the mitogenic pathway in lymphocytes and fibroblasts

When a growth factor or mitogen interacts with its cellular receptors it is assumed that this generates a primary mitogenic signal (P), defined as the first intracellular biochemical response that results directly from the mitogen—receptor interaction. The primary signal initiates the transition from the quiescent state (G₀) to G₁ and commits the cell to metabolic activation by triggering an extensive network of secondary responses that result ultimately in DNA synthesis in S phase. We assume that within this network there is a sequence of specific secondary responses (S ₁ … S_x) that is obligatory for progression through Gi and the subsequent commitment to S phase. An obligatory secondary response is defined by the criterion that specific inhibition of the response aborts the subsequent G ₁ to S transition. These two components, Pand the sequence of responses S₁…S_x, define the mitogenic pathway from G₀ as far as S phase in normal eukaryotic cells (Fig. 1). The general question that concerns us is whether the various mitogens that act on different types of cell operate through distinct mitogenic pathways, or whether there is a common sequence of obligatory responses for different mitogens and cell types that can usefully be defined as a common mitogenic pathway underlying the phenotypic variation in responses in different cell types.

This rather formal statement of the problem serves to define the terms used here in a comparative analysis of early responses to mitogens and of subsequent DNA synthesis in lymphocytes and fibroblasts. Our approach assumes that by defining some of the obligatory responses in G₁, and the causal relationships between them through the use of specific inhibitors, it will be possible to compare the mitogenic pathways, at least in broad outline, for a variety of mitogens acting on two very different types of cell. Interest in the biochemical mechanisms controlling the progression from G₀ to S phase in normal cells has intensified as the proteins encoded by several oncogenes have been identified and shown to be related either to the transduction of the primary signal or to secondary responses in G\ (reviewed by Berridge, 1984). A detailed description of the mitogenic pathway(s) from Go to S phase in normal cells will probably be necessary to understand how control of the pathway is subverted by specific oncogene products.

The simplest version of a common mitogenic pathway would be one in which the different mitogens and growth factors that normally act on cells in vivo (i.e. physiological mitogens) all caused the same primary signal. It is, however, clear from data summarized here, and from previous studies, that the primary mitogenic signal must be different for various mitogens acting on the same type of cell because the patterns of secondary responses generated by the different mitogens can be clearly distinguished. This is a simple example of the use of early response pattern analysis, which can also be applied to obtain detailed information about the relationship between the mitogenic pathways activated by different mitogens, as illustrated below. If the primary signals for different mitogen receptors are distinct, the concept of a common mitogenic pathway will be useful only if there is convergence of the different pathways on an early secondary response from which the subsequent commitment to S phase is driven. A key question, therefore, is whether there is a common early response that is obligatory for mitogenic stimulation.

Some of the secondary responses common to lymphocytes and fibroblasts are summarized in Fig. 2. Among the earliest responses that can be detected, in some instances within 10 s of the addition of mitogens, are ionic signals (increases in the free cytoplasmic Ca²⁺ concentration ([Ca] _i) and cytoplasmic pH (pH,) (Tsien, Pozzan&Rink, 1982; Heskethef al. 1983b, 1985; Moolenaar, Tsien, van der Saag de Laat, 1983; Morris et al. 1984). Some mitogens that cause these ionic signals also stimulate the simultaneous breakdown of phosphatidylinositol (4,5)-bisphosphate (Ptdlns?2) to inositol (l,4,5)-trisphosphate (InsP₃) (Mooreetal. 1984; Taylor etal. 1984; Berridge, Heslop, Irvine & Brown, 1984). At a later stage, approximately 5-to 10-fold longer, there is a general metabolic stimulation of the cells marked by an increase in glycolysis and in the uptake of metabolites such as uridine (reviewed by Hume & Weidemann, 1980). At about the same time mRNA is transcribed from two proto-oncogenes, c-fos and c-myc (Kelly, Cochran, Stiles & Leder, 1983; Greenberg & Ziff, 1984; Moore, Todd, Hesketh & Metcalfe, 1985). This precedes by about another order of magnitude a general increase in RNA and protein synthesis (Hume & Weidemann, 1980). It is clear that these biosynthetic responses late in G ₁ are obligatory (by the criterion of specific block) in both lymphocytes and fibroblasts for progression to S phase. The biochemical mechanisms for the initiation of RNA and protein synthesis are thought to be the same in all eukaryotic cells, so that mitogenic pathways are presumed to be highly convergent for different cells and different mitogens, at least late in G₁. While it is very likely that there are many other specific responses in G ₁ common to both lymphocytes and fibroblasts (e.g. phosphorylation of the S ₆, ribosomal protein etc.) many of these responses remain to be defined in lymphocytes.

To resolve whether there is a common mitogenic pathway in lymphocytes and fibroblasts it is therefore necessary to determine whether any early response common to the cells is obligatory for subsequent DNA synthesis. No clear causal relationships have yet been established between any of the early secondary responses and the general stimulation of RNA and protein synthesis in lymphocytes and fibroblasts later in G ₁. It is only for sea urchin eggs that any early response has been shown to be obligatory for all of the subsequent events in the fertilization pathway that have been examined. When the sea urchin egg is fertilized, there is a transient Ca signal, complete in about 3 min, followed by a sustained increase in pH _i. The general view is that one, or both, of these coupled ionic signals is necessary for all of the subsequent responses up to and including DNA synthesis, which is initiated after about 40 min. The transient Ca signal can be by-passed by raising pH, directly with NH₃, which results in the stimulation of many of the normal metabolic responses and DNA synthesis (Whitaker & Steinhardt, 1982).

Although the activation of sea urchin eggs differs substantially from the activation of somatic cells, notably in that enhanced RNA and protein synthesis are not obligatory for DNA synthesis after fertilization, the ionic signals in the sea urchin egg provide a useful working hypothesis: is the ionic ‘kick-start’ to the fertilization pathway a general mechanism for somatic eukaryotic cells, or is it a peculiarity of the egg system?

T and B lymphocytes of several species show a polyclonal mitogenic response to appropriate antibodies directed against the antigen receptors on the cell surface. B lymphocytes are stimulated by anti-immunoglobulin antibodies (Nash & Ling, 1976), whereas T cells are stimulated by several monoclonal antibodies to components of the T-cell receptor and other surface structures such as the Tn antigen (Beverly & Callard, 1981; Weiss, Daley, Hodgdon & Reinherz, 1984a). A rapid increase in [Ca] _i occurs in both types of cell within a few seconds of addition of the mitogenic antibodies (Pozzan, Arslan, Tsien & Rink, 1982; O’Flynn, Linch & Tatham, 1984; Weiss, Imboden, Shobach & Stobo, 19846), but no corresponding pH; studies have been reported. No mitogenic antibodies for mouse T cells (thymocytes) have been described, but various lectins that act as polyclonal mitogens (e.g. concanavalin A (ConA), phytohaemagglutinin (PHA), etc.) bind to and crosslink components of the T-cell receptor through their carbohydrate moieties and thus mimic the action of specific antibodies (Weiss, Imboden, Shobach & Stobo, 19846; Samelson, Harford, Schwartz & Klausner, 1985). Various studies have shown that receptor cross-linking is obligatory for the mitogenic action of antibodies on T or B cells. The early responses that have been examined so far in thymocytes stimulated with mitogenic antibodies are indistinguishable from those stimulated by lectins, which is consistent with activation of the cells through the same receptor. However, actions of the lectins are distinguishable from those of the mitogenic antibodies in that the lectins inhibit DNA synthesis at supra-optimal concentrations whereas the antibodies do not. Two mechanisms have been proposed for the inhibitory effect of supraoptimal lectin concentrations. Edelman and co-workers showed that the inhibition occurs very late in G ₁ and suggested that the supraoptimal concentrations of lectins bind to proteins on the cell surface, other than the mitogen receptors, that blocked essential surface-modulating activity controlled by the cytoskeleton (McClain & Edelman, 1976). We have shown that supra-optimal lectin concentrations cause rapid capping of the mitogen receptors, thus removing the receptors from the cell surface prematurely before commitment to S phase has occurred (Pozzan, Corps, Hesketh & Metcalfe, 1981). These mechanisms are not mutually exclusive. However, the self-inhibitory effect of the lectins on DNA synthesis complicates correlation with early responses that are not inhibited at supramitogenic lectin concentrations. Thus, although most of the analysis described here relates to ConA as a T-cell polyclonal mitogen, direct comparison with the mitogenic antibodies will be essential. For example, an immediate prediction, based on the assumption of a common mitogen receptor for the mitogenic antibodies and lectins, is that the antibodies will generate a pH; response similar to that induced by the lectins.

The most extensive data for the ionic signals in lymphocytes have been obtained for mitogens that act on mouse thymocytes. We have compared the ionic responses to ConA and to two ‘opportunistic’ co-mitogens, the phorbol ester 12-o-tetradecanoyl phorbol 13-acetate (TPA) and the calcium ionophore A23187. The Ca signal in response to ConA in cells that have been allowed to become quiescent for more than about 8h after isolation by incubation at 37 °C in culture medium consists of two distinguishable components (Hesketh et al. 1985). A rapid, transient increase in [Ca] _i is observed either in normal medium or in the absence of extracellular Ca²⁺, and is therefore attributed to Ca²⁺ release from an intracellular pool (Fig. 3). There is also a sustained increase in [Ca] _i to 150-200nM from approximately 100nM in unstimulated cells (Hesketh et al. 1983b, 1985). This signal declines back to 100 nM with the same time course as cap formation by ConA on the cell surface over approximately 24 h (Hesketh, Bavetta, Smith & Metcalfe, 1983b). It appears, therefore, that the persistent component of the Ca signal requires a continuous interaction of the mitogen with its receptors. This is confirmed by the observation that removal of ConA from its receptors by cr-methyl mannose immediately reverses the Ca signal.

In freshly isolated cells the transient component of the Ca signal is greatly diminished, whereas the persistent signal is very similar to that in cells allowed to become quiescent (Hesketh et al. 1985). The freshly isolated cells are metabolically active (e.g. they have a high rate of lactate production) and contain a small population of larger cells in S phase (Reeves, 1977). The data suggest strongly that the transient release of Ca²⁺ from the intracellular pool occurs only in the quiescent cells in GQ, whereas the persistent Ca signal in response to ConA may also occur in activated cells in the cell cycle. The mechanism by which the persistent Ca signal is generated remains to be established. It is dependent on Ca²⁺ in the medium and presumably requires either a persistent stimulation of Ca²⁺ influx or an inhibition of active Ca²⁺ efflux across the plasma membrane.

Agents that inhibit the Ca signal specifically are essential to establish whether the signal is obligatory for subsequent DNA synthesis, but unfortunately no specific reagents for blocking the signal have been identified. Removal of Ca²⁺ from the medium blocks the persistent Ca signal and eventually abolishes the transient signal by depleting the cells of Ca²⁺ (Hesketh et al. 1985). However, the viability of the cells also decreases in Ca²⁺-free medium and it is clear that Ca²⁺ depletion has extensive effects on the cells in addition to the effects on the Ca signal.

Agents that elevate intracellular cyclic AMP, or membrane-permeant cyclic AMP analogues such as 8-bromo cyclic AMP (8-Br cAMP), antagonize both the transient and persistent components of the Ca signal (Hesketh et al. 1985) (Fig. 3), and block subsequent DNA synthesis in response to ConA (Wang, Sheppard & Foker, 1976). It is very unlikely, however, that the activation of cyclic AMP-dependent protein kinases results in specific block of the Ca signal. It is of interest that the endogenous cyclic AMP concentration is very high in thymocytes immediately after isolation compared with the level in quiescent cells (Moore, Smith, Hesketh & Metcalfe, 1983), and the Ca signal in response to ConA within 30-60 min of isolation is small or undetectable. The persistent Ca signal that is characteristic of freshly isolated cells develops with a time-course similar to that of the decline in the cyclic AMP concentration to the level in quiescent cells (i.e. after about 1-1·5 h; J. P. Moore, unpublished data). Whether the early block on the Ca signal is due solely to the very high cyclic AMP content of freshly prepared thymocytes has not been established, but the observation is consistent with evidence that endogenous cyclic AMP antagonizes the Ca signal. As the cyclic AMP level declines and the [Ca]i response increases, the T cells become refractory to agents that elevate cyclic AMP (e.g. prostaglandin and show no significant response after 2-3 h (J. P. Moore, unpublished data). It has recently been shown that in cultured hepatocytes there is a progressive increase in /3-adrenergic responses mediated by cyclic AMP coinciding with a decline in cr-adrenergic responses mediated by [Ca],. It was proposed that this was due to a decline in the activity of a G protein (N,) which activates the effector system for the [Ca]i response but also has a reciprocal inactivating effect on adenylate cyclase (Itoh, Okajima & Ui, 1984). It is possible that in thymocytes the reverse process occurs in which a G protein progressively inactivates adenylate cyclase, while activating the [Ca]_i response. The effector system for the transient component of the Ca signal in thymocytes is probably coupled to the activation of polyphosphoinositide phosphodiesterase by the release of Ins(l,4,5)P3 (discussed below), and although the mechanism by which the enzyme is activated has not been established, preliminary evidence suggests that polyphosphoinositide phosphodiesterase is coupled functionally to the T-cell receptor through G proteins, as in mast cells and platelets (Haslam & Davidson, 1984; Cockcroft & Gomperts, 1985).

The pH, increase in response to ConA is slower than the Ca signal and is fully developed after 4-5 min (Fig. 4), but the two responses are not directly coupled. Thus, removal of Ca from the medium to abolish the Ca signal did not block the increase in pH; in response to ConA. TPA (10 nM) generated a pH; increase of 0·25-0·30 unit without any significant effect on [Ca]_i, whereas A23187 at mitogenic concentrations (25-50 nM) increased pH; by 0·12-0·18 unit. Unlike the pH; responses to ConA and TPA, the response to mitogenic concentrations of A23187 was dependent on extracellular Ca²⁺ ([Ca]_o) and is directly coupled to the substantial increase in [Ca] _i to 500—750 nM. All three mitogens therefore cause increases in pH, and for each an extracellular Na⁺ concentration greater than 5 mM is required for the maximal response, suggesting that the pH; increase is generated by stimulation of a Na⁺/H⁺ exchanger (Hesketh et al. 1985). The removal of ConA from the cell surface by a-methyl mannose does not immediately affect the pH; response, in contrast to the rapid reversal of the Ca signal, whereas removal of Ca²⁺ from the medium by EGTA immediately reverses the pH_i response to A23187 (J. P. Moore, unpublished data).

Although the Ca signal occurs very rapidly after the addition of ConA, it is a secondary response rather than the primary signal since some of the later secondary responses, including the pH; increase, are generated when the Ca signal is abolished by prolonged incubation at low [Ca]_o (Hesketh et al. 1985). The mechanisms by which the ionic signals are coupled to the primary response to ConA must therefore allow the signals to be generated independently of each other and we have assumed as a working hypothesis that the [Ca] _i and pH_i responses are generated by the breakdown in PtdInsP2, which we have shown to be stimulated by ConA in T cells (Moore et al. 1984). In this ‘dual signal’ hypothesis, derived from the work of Nishizuka, Berridge and others (Nishizuka, 1984; Berridge, 1984; Berridge & Irvine, 1984), the release of Ins(l,4,5)P3 from PtdInsP₂ is assumed to cause the release of Ca²⁺ from the intracellular pool (presumably the endoplasmic reticulum) and to account for the transient component of the Ca signal in thymocytes. Whether the other isomer of inositol trisphosphate, Ins(l,3,4)P₃, which has recently been described (Irvine, Letcher, Lander & Downes, 1984), is responsible for the sustained [Ca] _i increase, for example by inhibiting the plasma membrane Ca²⁺ pump or stimulating Ca²⁺ entry from the medium into the cells, remains to be determined. (It has also been suggested that PtdInsP₂ may regulate the plasma membrane Ca²⁺-ATPase activity directly; Thiyagarajah & Lim (1984).) The other product of PtdInsP₂ breakdown is diacylglycerol, and this is widely assumed to act as an endogenous activator of protein kinase C (PK-C) (Kishimoto et al. 1980). This would account for the increase in pH, in T cells in response to ConA, by analogy with the effect of TPA as an exogenous PK-C activator (Castagna et al. 1982).

An important corollary of this hypothesis, central to the analysis of the mitogenic pathways in T cells, is that it provides a direct explanation for the co-mitogenic action of A23187 with TPA. In this model, the opportunistic mitogens by-pass the physiological pathway acting through PtdInsP₂ breakdown, by increasing [Ca] _i and activating PK-C directly as their respective primary responses, which are assumed to be obligatory secondary responses for mitogenic stimulation by ConA. This would, of course, represent a very early convergence of the physiological and opportunistic mitogenic pathways. Attractive though this is, however, our conclusion from analysis of the responses to the three mitogens is that the secondary responses to ConA cannot be derived solely from the Ca signal it generates together with any activation of PK-C it may cause. The dual-signal hypothesis for thymocytes is considered now in more detail (Fig. 5), to illustrate how this conclusion has been reached from an analysis of the pattern of early responses.

All three mitogens stimulate the synthesis of Ptdins, which can be detected as enhanced [³H]inositol incorporation after 30-60 min (Moore, Smith, Hesketh & Metcalfe, 1982), and stimulate a net increase in Ptdins phosphorylation to PtdlnsP and PtdInsP₂, which is maximal within 10 min (Taylor et al. 1984). However, only ConA also stimulates the breakdown of PtdInsP₂ to release InsPs (Fig. 6). This response occurs faster tlian the stimulation of net PtdInsP₂ synthesis (Fig. 7) and coincides closely in time course with Ca²⁺ release, although analysis of the Ins(l,4,5)P₃ and Ins(l,3,4)P₃ isomers released as a function of time should provide a more critical comparison of this correlation (Irvine et al. 19846). The InsPs response to ConA is substantially inhibited by 8-Br cAMP, which is consistent with the effect of cyclic AMP on the transient Ca signal (Taylor et al. 1984; Hesketh et al. 1985). From these and other data it is entirely plausible that InsP₃ is responsible for the release of intracellular Ca²⁺ in thymocytes. This remains to be demonstrated directly in the permeabilized cells, although the response has been shown in permeabilized fibroblasts (Irvine, Brown & Berridge, 1984). The regeneration of PtdInsP₂ in response to ConA also allows persistent responses to be generated by InsP₂ release from PtdInsP₂.

What is much less certain is whether the activation of PK-C by ConA is sufficient to account for early secondary responses that are common to both ConA and TPA. These include the stimulation of Ptdins synthesis and phosphorylation, glycolysis and uridine uptake, and of c-myc gene activation. There are two mechanisms by which ConA might activate PK-C in the dual-signal hypothesis (Fig. 5): through diacylglycerol or through the increase in [Ca] _i, since the isolated enzyme is coactivated by Ca²⁺ with lipids (Takai et al. 1979). However, detailed analysis of the early responses to A23187 suggests clearly that a rise in [Ca]_; does not activate PK-C. For example, the uptake of uridine into the cells is stimulated by TPA but not by A23187 (A. N. Corps, unpublished data). The uridine response to TPA is not affected by A23187 and therefore a rise in [Ca] _i does not block the uridine response when PK-C is activated. We conclude that the pHi and other early responses to A23187 must be coupled to the [Ca] _i increase by a pathway that does not involve PK-C activation. It also follows that ConA cannot activate PK-C solely through the Ca signal it generates.

Similar analysis of the early metabolic responses suggests that ConA does not cause significant activation of PK-C, for example through released diacylglycerol. ConA stimulates the uridine response, but the response is blocked by prolonged depletion of extracellular Ca²⁺ (Fig. 8). If PK-C activation by ConA depends upon extracellular Ca²⁺, then other responses common to ConA and TPA that depend on PK-C activation should also be dependent on [Ca]_o for ConA. In contrast to this, the stimulation of glycolysis, assayed as lactate production, is independent of [Ca]_o for both ConA and TPA, and the response to the two mitogens together is approximately additive, even at TPA concentrations that cause the maximal response to TPA alone (Fig. 8). This demonstrates that the stimulation of glycolysis by ConA is very unlikely to depend significantly on PK-C activation. Taken together with similar indirect analyses of other responses (e.g. Ptdins synthesis), any activation of PK-C by ConA is estimated to be less than 10% of the maximal activation achieved by TPA. We have been unable to detect an increase in diacylglycerol in mouse thymocytes stimulated by ConA, although there is a substantial amount of the lipid in quiescent cells, which may obscure the release from PtdInsP₂ if this is in a small specific pool. It remains to be demonstrated, however, that a transient increase in diacylglycerol can cause sustained activation of PK-C, and it is therefore necessary to develop a direct assay for PK-C activation in thymocytes by identifying a unique protein substrate for the enzyme, as found in fibroblasts (Rozengurt, Rodriguez-Pena & Smith, 1983a).

The most significant general conclusion from these studies is that when the Ca signal and any PK-C activation generated by ConA are mimicked by appropriate combinations of A23187 and TPA, the early responses to the opportunistic mitogens are small compared with the responses to ConA. We conclude that the lectin must generate an unidentified signal, in addition to the Ca signal and any PK-C activation, to account for both the early responses and for subsequent mitogenic stimulation. The same conclusion was reached recently for another mitogenic lectin, PH A, based on an analysis of mitogenic stimulation of T cells. PH A was required in addition to A23187 with TPA for progression to DNA synthesis when the T cells were depleted of macrophages (Kaibuchi, Takai & Nishizuka, 1985).

The c-myc and c-fos proto-oncogenes are the cellular counterparts of the transforming genes of the avian myelocytomatosis and FBJ murine osteosarcoma viruses (Roussel et al. 1979; Curran & Teich, 1982). It has been reported recently that mitogens activate both the c-myc and c-fos genes within 1 h in various cells in G_o (Kelly et al. 1983; Campisi et al. 1984; Greenberg & Ziff, 1984; Muller, Bravo, Burckhardt & Curran, 1984; Kruijer, Cooper, Hunter & Verma, 1984). The amounts of c-myc mRNA and protein remain elevated in proliferating normal cells and are not thought to be regulated by the cell cycle, other than by entry into GQ (Thompson, Chailoner, Neiman & Groudine, 1985; Hann, Thompson & Eisenman, 1985). The c-myc and c-fos proteins are located in the nucleus, but their functions have not been defined (Abrams, Rohrschneider & Eisenman, 1982; Curran, Miller, Zokas & Verma, 1984). Nevertheless, these specific gene activations establish an early link between the primary signal and nuclear activation and are promising candidates for obligatory secondary responses early in G₁.

In quiescent fibroblasts the c-fos gene is induced more rapidly than c-myc (within 10-20 min) in response to TPA and to some of the growth factors that act on these cells (e.g. platelet-derived growth factor; PDGF), but the expression is transient (Greenberg & Ziff, 1984; Muller et al. 1984; Kruijer et al. 1984). In T and B lymphocytes the c-myc gene has been shown to be activated by ConA and lipopolysaccharide, respectively (Kelly et al. 1983; Moore et al. 1985). The two key questions that concern us are whether the genes are activated by ionic signals in the normal mitogenic pathway, and whether the expression of these genes is obligatory for subsequent DNA synthesis.

We have found that each of the mitogens, ConA, TPA and A23187, activated the c-myc gene in mouse thymocytes within 15-30 min and the amount of c-myc mRNA in the cells increased for at least 4h (Fig. 9). No c-myc mRNA could be detected in unstimulated thymocytes, and the increase in c-myc mRNA in response to ConA is estimated to be at least 10-fold, consistent with the previous report of Kelly et al. (1983). If the amount of c-myc mRNA in cells stimulated with an optimal mitogenic concentration of ConA (approximately 0-80/zgml, see Fig. 10A) for 2h is normalized as 100%, the relative amounts in cells treated under the same conditions with optimal co-mitogenic concentrations of either TPA (10 nM) or A23187 (25— 50 nM) were 280+57 % and 187 + 37 % (S.E.M.; n = 9). Thus, although neither TPA nor A23187 is mitogenic by itself, each caused an increase in c-myc mRNA comparable with that stimulated by ConA. Optimal co-mitogenic concentrations of TPA with A23187 caused larger increases in DNA synthesis than ConA alone (Fig. 10B) and were synergistic in stimulating c-myc mRNA by 850+186% (n = 6) (Fig. 11). Taken together, the data show that while mitogenic stimulation by the three mitogens is always associated with activation of the c-myc gene, this activation is not sufficient to cause commitment to DNA synthesis, which must also depend on and be modulated by other responses.

Activation of the c-fos gene was transient and clearly preceded the increase in c-myc mRNA as found previously in fibroblasts. An increase in c-fos mRNA was stimulated by A23187 within 7·5 min and the amount of mRNA increased by at least 100-fold within 20 min before declining over 3h. The maximal amount of c-fos mRNA was estimated to be at least 30-fold greater than that of c-myc in the cells. The c-fos gene was also activated by ConA and TPA with transient time-courses similar to that shown for A23187. It is assumed that each of the mitogens causes expression of the c-myc and c-fos genes as the proteins, although this has not been demonstrated in thymocytes (but see Persson et al. 1984).

The effects of modulating the Ca and pH signals on the expression of the c-myc gene were examined separately. The pH_i response to each of the three mitogens is abolished by reduction of [Na]_o to less than 1 mM. However, replacing NaCl in the medium with choline chloride did not block the increase in c-myc mRNA stimulated by either ConA or TPA although the response to A23187 was blocked by the removal of Na⁺ (Table 1). We conclude, therefore, that the pH_i increase is not necessary for the activation of the c-myc gene in the normal mitogenic pathway.

The activation of the c-myc gene by mitogenic concentrations of A23187, which raise [Ca] _i from 100 nM in resting cells to approximately 500-750 nM, was abolished completely by addition of EGTA to reduce [Ca]_o to 120nM (Table 1). Under these conditions A23187 causes only a small, transient increase in [Ca] _i that is attributable to the release of Ca²⁺ from intracellular stores. The failure of A23187 to activate the c-myc gene in the absence of extracellular Ca²⁺ is not due to any inhibitory effect of the ionophore under these conditions, since the combination of TPA with A23187 in the presence of EGTA stimulated a large increase in c-myc mRNA, comparable with that induced by TPA alone. The activation of the c-myc gene by the ionophore is therefore directly correlated with the increase in [Ca] _i. In marked contrast to A23187, the increase in c-myc mRNA in mouse thymocytes stimulated by TPA occurs by a mechanism independent of [Ca]_o (Table 1) or of any increase in [Ca] _i.

ConA activates the c-myc gene at low [Ca]_o (120 nM), although to a reduced extent (approximately 50%, see Table 1), and a [Ca]_o greater than approximately 100/ZM was necessary for maximal stimulation of the c-myc mRNA increase. We have noted earlier that ConA causes only a transient increase in [Ca] _i at low [Ca]_o and the data clearly imply that activation of the c-myc gene in response to ConA is not solely a consequence of the increased [Ca];. There may be a contribution from the Ca signal in normal medium, by analogy with the [Ca]_o-dependent activation by A23187, but as emphasized earlier, the removal of Ca²⁺ from the medium may inhibit the activation of the c-myc gene by ConA through mechanisms other than the decrease in the [Ca] _i response.

Agents that elevate cyclic AMP (e.g. prostaglandin E₁), or 8-Br cAMP, inhibit substantially the Ca signal in response to ConA. In contrast, 8-Br cGMP has no effect. Neither cyclic nucleotide analogue (5 IDM) increased c-myc mRNA levels by itself, but 8-Br cAMP was a potent inhibitor of the response to ConA, whereas 8-Br cGMP was relatively ineffective (Table 1). However, the inhibition of c-myc expression by 8-Br cAMP cannot be attributed specifically to its effect on the [Ca] _i response to ConA since 8-Br cAMP also antagonized the activation of the c-myc gene by TPA (Table 1), which is [Ca] _i-independent. The data do not therefore establish a causal relationship between the Ca signal and c-myc activation in response to ConA. However, it may be noted that 8-Br cAMP has consistent effects on the early responses to ConA (Ca signal, InsPs and c-myc expression) and on subsequent DNA synthesis, which are all substantially inhibited, whereas 8-Br cGMP has little effect.

We conclude that expression of the c-myc gene as mRNA can occur without any increase in pH; in thymocytes. The gene can be activated by at least two distinct pathways: by A23187 through elevation of [Ca] _i, independently of PK-C activation; or by TPA through PK-C, independent of [Ca] _i. The response to ConA is, at most, only partially dependent on the increase in [Ca] _i, and it remains to be demonstrated by direct assay whether any activation of PK-C by ConA is a significant component of the [Ca] _i-independent activation of the c-myc gene by this mitogen.

The effects of [Na]_o and [Ca]_o have been determined for all of the early responses examined. None of the responses shown in Fig. 12 were found to depend on [Na]_o except for the pH; signal, although depolarizing media with Na⁺ replaced by K⁺ caused greatly enhanced glycolysis in quiescent cells. Low [Na]₀ inhibited responses to A23187 (e.g. c-myc, see Table 1) but this may be due to toxic effects of the ionophore under these conditions. We conclude that none of the secondary responses to ConA or TPA shown in Fig. 12 depends on the pH_i signal.

Data for the dependence of early responses on [Ca]₀ are summarized in simplified form in Fig. 12, where the relative responses in normal medium ([Ca]_o = 0·43 mM) and in low [Ca]₀ (120 nM) are compared. The main conclusions from the pattern of early responses to each mitogen and the effects of [Ca]_o are summarized.

(1) The responses in normal medium confirm that the primary signals from each mitogen must be different since there are qualitative differences in the pattern of secondary responses.

(2) Neither TPA nor A23187 cross-activate the primary response to each other or to ConA. Thus neither of the opportunistic mitogens causes the full repertoire of responses to ConA, nor do they abolish responses to ConA that they do not activate themselves.

(3) All of the responses to TPA are independent of [Ca]_o, consistent with activation of PK-C independent of [Ca]j as the primary response to this mitogen.

(4) All of the responses to A23187 are abolished at low [Ca]_o, although there is a residual transient [Ca]; increase from Ca²⁺ in the intracellular pool. This is consistent with the [Ca]_; increase in normal medium acting as the primary response to A23187.

(5) The responses to ConA show varied dependence on [Ca]_o confirming that the [Ca]; increase cannot be the primary response to ConA. The mechanisms by which the [Ca]_o-dependent and [Ca]_o-independent responses are generated remain to be established, but they cannot be assumed to be derived solely from the [Ca]i response and activation of PK-C, respectively.

Taken together with the more detailed analyses presented earlier, this summary illustrates the information that can be derived from a comparative analysis of the early responses to the mitogens and the effect of ionic perturbations on them. It provides some insight into the mechanisms through which the responses may be coupled, and evidence for extensive but incomplete convergence of the early secondary responses to mitogens that activate different primary signals.

It was noted earlier that it is difficult to establish whether early responses are obligatory for the mitogenic pathway because completely specific inhibitors of the various responses are not generally available. Mitogenic stimulation by any co-mitogenic combination of the three T-cell mitogens is preceded by an early pH_; increase. Although this response is not obligatory for any of the other early responses examined, it is possible that the pHj increase may be necessary for progression through the later stages of Gp Alternatively, the pH, response may fall entirely within the permissive pHj range for the mitogenic pathway to proceed from Go to S phase, and may therefore be a consequential response rather than obligatory. Many agents that inhibit Na⁺/H⁺ exchange (e.g. amiloride) are not specific and their inhibitory effects on DNA synthesis do not therefore have diagnostic value in determining the role of the pH_i response (Besterman, Tyrey, Cragoe & Cuatrecasas, 1984; Davis & Czech, 1985). However, more specific analogues of amiloride have been reported that block Na⁺/H⁺ exchange in fibroblasts, without inhibiting DNA synthesis (Besterman et al. 1984).

Mitogenic stimulation by the T-cell mitogens is preceded by an early Ca signal, although it may be noted that optimal co-mitogenic combinations of ConA with TPA (Fig. 10A) generate only a very small early Ca signal. The partial dependence of c-myc activation on [Ca]_o in response to ConA is at present the most suggestive evidence that a [Ca] _i increase may normally be required for progression through Gi in the pathway activated by the T-cell receptor. The stimulated uptake of uridine into the cells, although almost entirely dependent on [Ca]_o, is not an obligatory response since progression to DNA synthesis will occur in media without uridine (A. N. Corps, unpublished data).

The inability of A23187 to cause mitogenic stimulation by itself in mouse thymocytes clearly indicates that the early generation of both [Ca] _i and pH_i responses in the cells is not a sufficient stimulus for these cells to reach S phase. The co-mitogenic action of A23187 with TPA is consistent with a mitogenic pathway activated by the two primary signals from the combined mitogens. A23187 with TPA activates all of the early responses that have been reported for ConA, except for the breakdown of PtdInsP₂ to release InsPj (Taylor et al. 1984). This additional response to ConA is consistent with the inference drawn from the early response pattern analysis that there is an unidentified signal derived from the primary response to ConA that is an obligatory component of the mitogenic pathway for ConA. Whether that signal is also activated by A23187 with TP A as a secondary response to their primary signals is of key interest in defining the convergence of the mitogenic pathways. However, the recent observations from Nishizuka’s laboratory suggest that the unidentified signal from the lectin is not activated by TPA with A23187. The requirement for PHA (assumed to act through the T-cell receptor) in addition to TPA and A23187 to cause mitogenic stimulation of mouse thymocytes depleted of accessory cells strongly suggests that PHA (or ConA) generates an independent obligatory response that is not generated by TPA with A23187 (Kaibuchi et al. 1985). Although this inference is clear, the point remains to be demonstrated unambiguously, since the obligatory responses to the mitogens might conceivably differ, depending on the presence or absence of accessory cells (monocytes and macrophages). It should also be noted that although the primary signal for ConA is clearly shown by the degradation of PtdlnsP₂ to be distinct from the responses to TPA with A23187, there is no evidence to indicate whether that particular response is related to the unidentified mitogenic signal from ConA.

In human and pig T lymphocytes, A23187 is able to cause mitogenic stimulation by itself (Maino, Green & Crumpton, 1974; Luckasen, White & Kersey, 1974). This raises the unresolved question as to whether the coupling between the mitogenic pathways activated in the T cells by A23187 or TPA is a function of T-cell species, or whether it is the requirement for activation of the accessory cells that is speciesdependent. There is no evidence at present to suggest that the requirement for interleukin-2 (Gillis & Mizel, 1981) and the expression of its receptors on T cells (Robb, Munck & Smith, 1981) can be by-passed in cell preparations depleted of accessory cells if the T cells are to proceed to S phase. It is clear however that analysis of the later stages in G\ in T cells will be facilitated by the use of pure T-cell preparations.

There are at least two mechanisms by which the transient component of the [Ca] _i response may be involved in cell activation. The rise in [Ca]i may modulate the activity of Ca²⁺-sensitive enzymes (e.g. via calmodulin). Although this may trigger subsequent responses, the requirement for persistent interaction of ConA with the T cells suggests that the transient component of the [Ca] _i increase may be incidental to the release of Ca²⁺ from the endoplasmic reticulum to allow this organelle to function in the activated cells, possibly to allow the synthesis of membrane proteins. If, for example, the synthesis of the receptors for interleukin-2 was inhibited in endoplasmic reticulum loaded with Ca²⁺, it would be expected that the Ca²⁺ concentration inside the endoplasmic reticulum would have to remain low throughout Gi until expression of the interleukin-2 receptor was complete. This may in turn require continuous activation of the T-cell receptor to give sustained release of InsPs isomers and thereby maintain the endoplasmic reticulum at a low Ca²⁺ content. Whether the persistent [Ca]i increase in ConA-stimulated cells, which declines over 24 h back to the level in quiescent cells, is obligatory for progression through Gi is unclear. There are indications that the cells become independent of extracellular Ca²⁺ after the interleukin-2 receptor has been expressed (Weiss et al. 1984a), and it is possible to speculate that both the expulsion of Ca²⁺ from the endoplasmic reticulum to give the transient [Ca] _i response, and the persistent increase in [Ca] _i, are necessary to transit the early Ca²⁺-sensitive phase of G_i.

This hypothesis clearly predicts that no cell could progress to S phase without clearing the Ca²⁺ load from the endoplasmic reticulum and maintaining it in the low Ca²⁺ state for an appropriate period to permit essential membrane protein synthesis. The fibroblast system can provide a critical test of this hypothesis.

A wide range of agents act as mitogens on 3T3 fibroblasts, many of which have been identified and their secondary responses characterized by the work of Rozengurt and colleagues (reviewed by Rozengurt et al. 19836). Here we analyse the relationship between mitogenic stimulation and early responses to seven mitogens that each generate distinct primary signals judged by their patterns of secondary responses (Table 2). Three of the mitogens, platelet-derived growth factor (PDGF), epidermal growth factor (EGF) and insulin, have receptors with intrinsic tyrosine kinase activity that is stimulated by binding the growth factor (Cohen, Carpenter & King, 1980; Ushiro & Cohen, 1980; Ek, Westermark, Wasteson & Heldin, 1982; Cooper et al. 1982; Kasuga et al. 1982; van Obberghen et al. 1983). Nevertheless, their patterns of secondary responses are quite distinct and it remains to be demonstrated formally that activation of the tyrosine kinases is responsible for any or all of the secondary responses. Agents that elevate cyclic AMP or mimic its action are co-mitogenic for 3T3 fibroblasts (Pruss & Herschman, 1979; Rozengurt, Legg, Strang & Courtney-Luck, 1979), presumably through the activation of cyclic AMPdependent kinases, whereas two of the other mitogens, vasopressin (VP) and prostaglandin F_2α. (PGF_2α) generate unidentified primary responses. TPA acts as an opportunistic mitogen on fibroblasts (Dicker & Rozengurt, 1978), but the mitogenic activity of A23187 on these cells is elusive. The Ca²⁺ ionophore generates many of the early secondary responses observed with other mitogens but rarely stimulates significant DNA synthesis as a co-mitogen acting on early-passage fibroblasts. However, in cells that have been passaged for several weeks in culture, A23187 is frequently a potent co-mitogen with insulin. We do not understand the cellular basis of this variability and further analysis of A23187 responses in relation to DNA synthesis has been omitted.

Substantial stimulation of DNA synthesis in quiescent fibroblasts usually requires the simultaneous action of two mitogens or growth factors on the cells. We find that all co-mitogenic combinations that give a synergistic effect on subsequent DNA synthesis initiate the mitogenic pathway from the time that both agents are present: delaying the addition of either agent causes a corresponding delay in DNA synthesis. This observation is not consistent with the proposal that mitogens are either ‘competence’ factors that initiate the mitogenic pathway or ‘progression’ factors that drive the cells through the later stages of Gi. The only exceptions to the requirement for two co-mitogens for DNA synthesis are PDGF and bombesin, which have been reported to cause significant DNA synthesis by themselves (Rozengurt et al. 1983; Rozengurt & Sinnett-Smith, 1983). In our hands, mitogenic stimulation by these agents is variable and less than 30 % of the maximal stimulation achieved with co-mitogenic mixtures of EGF with insulin. Furthermore, the response to either mitogen is enhanced synergistically by the co-addition of insulin. The co-mitogenic effects of the mitogens are summarized in Table 3 and discussed in relation to the ionic signals and other responses below.

The Ca and pH responses for the three growth factors PDGF, EGF and insulin are shown in Fig. 13. Both PDGF and EGF generate a transient Ca signal followed by a sustained increase in pH_;, whereas the responses to insulin are marginal, with no increase in [Ca]j and any increase in pH; limited to less than 0·05. The primary response from the insulin receptor is therefore clearly distinguished from the primary responses to PDGF or EGF. Furthermore, although the ionic responses to PDGF or EGF are similar, the mechanisms by which they are generated are quite distinct: the EGF response is entirely dependent on [Ca]_o (Fig. 14), whereas the response to PDGF is [Ca]_o-independent. The two growth factors may also differ in their effects on PtdInsP₂ metabolism in that EGF does not cause PtdInsP₂ breakdown (J. D. H. Morris and J. P. Moore, unpublished data) but PDGF causes an increase in inositol phosphates (Berridge et al. 1984). Of the other mitogens in Table 3 only VP and PGF_2CV increase inositol phosphates and generate both Ca and pH; responses. TPA causes a substantial pH; increase of approximately 0-20 without a prior Ca signal, similar to its action on thymocytes, and 8-Br cyclic AMP stimulates only a marginal increase in pHj. A23187 generates both a Ca signal and a slower pH; response and both signals are dependent on [Ca]_o as in thymocytes.

An interesting feature of the ionic signals in response to the sequential addition of mitogens is the interactions between the mitogens that they imply. The data in Table 3 show that PDGF blocks the subsequent [Ca] _i response to VP and PGF_2a, but not to EGF. VP blocks only the subsequent [Ca] _i response to PDGF (i.e. there is a reciprocal blocking mechanism), whereas PGF_2a. or EGF do not affect the subsequent [Ca] _i responses to any of the other three mitogens which give Ca signals.

The data suggest that if VP and PGF_2α. both release intracellular Ca²⁺ via InsPs, then neither mitogen depletes the Ca²⁺ pool sufficiently to inhibit the response to the other (Fig. 15). The inability of either VP or PDGF to block the subsequent [Ca] _i response to EGF is also of interest. Both VP and PDGF activate PK-C assayed indirectly by the phosphorylation in intact cells of an 80 kDa polypeptide (Rozengurt et al. 1983"), and TPA blocks the [Ca] _i response to EGF. This suggests that the activation of PK-C by VP or PDGF is small compared with that achieved by TPA. This is also indicated by the additional increases in pH; to TPA when added after VP or PDGF: if PK-C were activated fully by either mitogen there would be no additional pH; response to TPA. In fact, only a small part of the pH; response to PDGF alone can be due to activation of PK-C, since an additional pH; increase is stimulated by PDGF after the response to TPA. Uncoupling of PtdInsP₂ breakdown from PK-C activation is implied by the responses to PGF₂. Activation of PK-C by TPA stimulates an increase in uridine uptake (Dicker & Rozengurt, 1979), but PGF_2α neither causes this uridine response (Jimenez de Asua et al. 1982) nor blocks the uridine response to TPA or VP (J. D. H. Morris, unpublished data). The implication is that any activation of PK-C by PGF_2α is below the threshold detectable by the indirect uridine uptake assay.

The data in Table 3 compare systematically the ionic signals and DNA synthesis in response to all of the pair combinations of the seven mitogens. Of the 21 combinations, 10 stimulate DNA synthesis synergistically to an extent significantly greater than the sum of the responses to the ligands separately, including all six combinations with insulin, which appears to be the most generally effective co-mitogen. (Single mitogens usually give less than 10% of the [³H] thymidine incorporation obtained with optimal combinations of mitogens such as EGF+insulin.) Of the 10 synergistic mitogenic combinations, six stimulate both Ca and pH, responses, and the other four a pH; response only. For one of these combinations (insulin + 8-Br cAMP) the total pHj increase was small (< 0·08) and less than that for the other synergistic combinations (0-20+0-03). The mitogenic combinations that give a pH, increase only as an early ionic response can provide a critical test for the hypothesis that it is necessary to clear Ca²⁺ from the endoplasmic reticulum for progression to S phase. This Ca²⁺ release need not necessarily occur in an early burst, and assays of the releasable Ca²⁺ load in the endoplasmic reticulum throughout Gi will be required to resolve this point.

All of the 11 mitogenic combinations that do not cause synergistic stimulation of DNA synthesis generate both ionic responses, clearly indicating that these signals do not constitute a sufficient stimulus for progression of a substantial proportion of the cells to S phase. A further point is that neither EGF nor insulin (together, the most potent co-mitogenic combination for 3T3 cells) activates c-myc and c-fos expression substantially (Kelly et al. 1983; Greenberg & Ziff, 1984; Kruijer et al. 1984; Muller et al. 1984). The effect of these mitogens in combination on the activation of the c-myc and c-fos genes through G₁ should define limits to which their expression may be necessary for subsequent DNA synthesis.

This work was supported by grants from the SERC and MRC to J.C.M., and from the MRC to J.P.M. We thank Dr Enrique Rozengurt for early passage 3T3 fibroblasts, Dr Georgina Nemecek for highly purified PDGF, Dr Robin Irvine for phosphorylated phosphatidylinositols and inositol (l,4,5)-trisphosphate and Dr John Todd for preparation of c-myc and v-fos probes.