Introduction
Cells divide at a particular size. How is cell growth coordinated with cell division? Biologists in general, and Paul Nurse in particular, have been thinking about this fundamental, deceptively difficult question for decades1. Indeed, Fantes and Nurse2 observed exactly three decades ago that Schizosaccharomyces pombe cells advance into mitosis and thus divide at a smaller size when deprived of nutrients (see Fig. 1). Building on the wealth of knowledge that has accumulated in the intervening 30 years, Petersen and Nurse3 now provide a molecular basis for how cell growth and division can be coordinated.
Figure 1: Control of cell size through the coordination of cell growth and division.
As demonstrated by Fantes and Nurse2 30 years ago, Schizosaccharomyces pombe cells advance into mitosis and thus divide at a smaller size when deprived of nutrients (nitrogen). The current work by Petersen and Nurse3 provides an explanation for this seminal observation by demonstrating that nutrient limitation relieves Tor1-mediated inhibition of mitotic onset.
Full size image (19 KB)TOR (target of rapamycin) is a highly conserved protein kinase that controls cell growth in response to nutrients4. The authors demonstrate that inactivation of TOR signalling, either by shifting cells from a rich to a poor-nitrogen medium or by treating cells with rapamycin, leads to activation of the MAPK (mitogen-activated protein kinase) Sty1 and thereby provokes advancement into mitosis. They further find that negative regulation of Sty1 by TOR involves the translation-regulating kinase Gcn2 and the Sty1 phosphatase Pyp2 (see Fig. 2); in response to nutrient deprivation and TOR inactivation, Sty1 is activated as a result of decreased Pyp2 levels. Although some mechanistic details of this regulation remain to be explained, the important new finding is that the TOR pathway connects to a MAPK pathway to relay a nutrient signal to the cell-cycle machinery.
Figure 2: The signalling pathway by which Tor1 inhibits the onset of mitosis in response to nutrients (nitrogen).
Steps in the pathway do not necessarily reflect direct interactions. The MAPK Sty1 activates Polo kinase Plo1, which in turn activates Cdc25- and Wee1-controlled Cdc2 activity to advance mitosis. Tor1, in response to nutrients, inhibits Sty1 by upregulating the Sty1 phosphatase Pyp2 through negative regulation of the kinase Gcn2. The mechanism by which Gcn2 inhibits Pyp2 is unknown but is distinct from the defined role of Gcn2 as an eIF2
kinase.
Earlier hypotheses of how cell growth is coordinated with cell division focused on the cell cycle machinery, emphasizing alterations of the cyclins and cyclin-dependent protein kinase(s)5. More recently it has become clear that many signalling pathways converge on the cell cycle machinery1. In S. pombe, for example, Sty1 (also known as Spc1) alerts the cell cycle machinery (Cdc2) to changes in the extracellular environment. Sty1 activates Cdc2 via the Polo kinase Plo1 (ref. 6).
Sty1 is closely related by protein alignment to Hog1 in Saccharomyces cerevisiae and to p38 in humans. However, advancement into mitosis in response to nutrient stress has so far been reported only in S. pombe, and Sty1 is the only MAPK reported to advance the cell cycle. Human p38 phosphorylates and inactivates CDC25c through the MAPKAP kinase MK2 (ref. 7) as part of a DNA damage checkpoint, and thereby arrests cell cycle progression. The homologue of MK2 in S. pombe is Srk1. Srk1 phosphorylates Cdc25 and 
srk1 cells have a shorter G2 phase8, suggesting that Srk1 negatively regulates the cell cycle as MK2 does in mammalian cells. Curiously, Sty1 also phosphorylates and activates Srk1. This is a point for future studies.
TOR is found in two functionally and structurally distinct complexes termed TORC1 and TORC2, from yeast to human4. Furthermore, unlike higher eukaryotes that contain a single TOR, both S. pombe and S. cerevisiae have two TOR genes encoding the two proteins TOR1 and TOR2. As determined by sequence alignment and interactions with conserved members of the two TORCs, Tor2 in S. pombe corresponds to TOR1 in S. cerevisiae as the TOR that forms TORC1 (refs 9–12). Conversely, S. pombe Tor1 corresponds to S. cerevisiae TOR2 as the TOR that forms TORC2. In higher eukaryotes, the same, single TOR is found in both TORCs. TORC1 and TORC2 in yeast and mammals are sensitive and insensitive to rapamycin, respectively. Interestingly, whereas only TORC2 was thought to be rapamycin sensitive in S. pombe, Petersen and Nurse3 now present evidence that S. pombe TORC1 is also rapamycin sensitive. S. pombe is so far unique in that both its TORCs appear to be rapamycin sensitive.
At first glance, it might seem surprising that active TOR in S. pombe delays the G2/M transition. TOR is well known for advancing the cell cycle, not for inhibiting it. However, on further reflection, there is no contradiction. Petersen and Nurse show that TOR (Tor1) regulates Sty1 and mitotic onset presumably as part of TORC2. By contrast, it is TOR within TORC1 that stimulates cell cycle progression and it does so at the G1/S transition, in S. pombe and other organisms. Does TORC2 delay G2/M transition in organisms other than S. pombe? This question is also of interest because cell size control in S. pombe is unusual in that it happens mainly in G2, in contrast with most cells, such as S. cerevisiae cells, in which it happens mainly in G1. In S. cerevisiae, genetic inactivation of TORC2 (temperature-sensitive TOR2 mutation) causes cells to accumulate in G2/M13, suggesting that TORC2 induces rather than delays mitotic onset in budding yeast. However, Petersen and Nurse also observe that only decreased levels of Tor1, not an absence of Tor1, results in early mitotic onset and consequently small cells. A complete absence of Tor1, as shown previously14 and as confirmed by Petersen and Nurse, prevents accelerated entry into mitosis on nitrogen starvation and results in elongated cells. The data do not fit into a simple model, and the above result in S. cerevisiae could be analogous to that observed on complete depletion of Tor1 in S. pombe. In mammals, previous studies that examined the role of mTOR in the cell cycle used rapamycin to inhibit mTOR and would therefore not have detected a role for rapamycin-insensitive mTORC2. However, cultured mammalian cells treated with rapamycin have a reduced proliferative rate and are smaller in all phases of the cell cycle15. Clearly, the fascinating observations by Petersen and Nurse will stimulate further investigation into the role of TOR, in particular TORC2, in the cell cycle.
It is important to note that the established role of TOR in controlling cell size is through the regulation of macromolecular synthesis (for example translation) and the consequent accumulation of mass. This is best observed in non-dividing cells such as muscle cells, in which upregulation of mTOR causes upregulation of anabolic processes and thus hypertrophy. Does inhibition of TOR in S. pombe lead to smaller cells not only because cells divide earlier, as now demonstrated by Petersen and Nurse, but also because they accumulate less mass?
Petersen and Nurse also show that shifting cells from glutamate (a good nitrogen source) to proline (a poor nitrogen source) decreases Tor1 kinase activity as measured in vitro. This is a significant finding because it suggests that TORC2, in addition to TORC1, is nutrient sensitive. So far, upstream cues for TORC2 are poorly understood in any organism. In fact, reports of regulated TORC1 or TORC2 activity in vitro are rare. It has been assumed that TOR is regulated by non-covalent interactions that are lost on cell disruption.
The authors show that loss of TOR signalling upon nutrient limitation activates Sty1 by decreasing Pyp2 through Gcn2, but interestingly the mechanism of Gcn2 action seems to be distinct from its known effects on translation. In S. cerevisiae, Gcn2 phosphorylates eukaryotic translation initiation factor 2 (eIF2) on serine 51 (serine 52 in S. pombe) and thereby downregulates general translation while upregulating the translation of Gcn4 by means of a mechanism involving small upstream open reading frames at the 5´ end of the GCN4 messenger RNA. Interestingly, the authors find that Gcn2 decreases the amount of Pyp2 protein but by means of a mechanism independent of serine 52 phosphorylation. In addition, the downstream activation of Sty1 is rapid — more rapid than one would expect from an effect of Gcn2 on eIF2.
The report by Petersen and Nurse provides answers to important and long-standing questions but, like the study by Fantes and Nurse2 three decades ago, it also poses many new and equally fascinating questions. Furthermore, it remains to be determined how decreased TOR signalling establishes the new cell size threshold, although smaller, at which cells divide. And how is this specific cell size threshold then maintained in subsequent cell cycles? Much remains for future studies.