The activity of key metabolic enzymes is regulated by the ubiquitin ligases that control the function of the cyclins; therefore the activity of these ubiquitin ligases explains the coordination of cell-cycle progression with the supply of substrates necessary for cell duplication. APC/C (anaphase-promoting complex/cyclosome)-Cdh1, the ubiquitin ligase that controls G1- to S-phase transition by targeting specific degradation motifs in cell-cycle proteins, also regulates the glycolysis-promoting enzyme PFKFB3 (6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase isoform 3) and GLS1 (glutaminase 1), a critical enzyme in glutaminolysis. A decrease in the activity of APC/C-Cdh1 in mid-to-late G1 releases both proteins, thus explaining the simultaneous increase in the utilization of glucose and glutamine during cell proliferation. This occurs at a time consistent with the point in G1 that has been described as the nutrient-sensitive restriction point and is responsible for the transition from G1 to S. PFKFB3 is also a substrate at the onset of S-phase for the ubiquitin ligase SCF (Skp1/cullin/F-box)-β-TrCP (β-transducin repeat-containing protein), so that the activity of PFKFB3 is short-lasting, coinciding with a peak in glycolysis in mid-to-late G1, whereas the activity of GLS1 remains high throughout S-phase. The differential regulation of the activity of these proteins indicates that a finely-tuned set of mechanisms is activated to fulfil specific metabolic demands at different stages of the cell cycle. These findings have implications for the understanding of cell proliferation in general and, in particular, of cancer, its prevention and treatment.
Rapidly dividing cells, irrespective of whether they are normal or neoplastic, exhibit enhanced glycolysis together with an increase in the rate of utilization of both glucose and glutamine. The partial oxidation of both substrates provides the energy, carbon, nitrogen and reducing equivalents needed to build a new cell. The way in which each substrate contributes to those tasks has been the subject of a great deal of research over many years [1–4]. However, questions remain, the most intriguing of which is the extent to which normal and neoplastic cells share metabolic pathways. This is particularly pertinent since the existence of relevant differences in the metabolic pathways of cancer and normal cells would suggest the possibility of developing therapeutic approaches that selectively interfere with cancer metabolism without side effects related to the mechanism of action. In view of this, it is surprising that systematic comparisons of the metabolism of proliferating normal and cancer cells have not been carried out.
Cell division has been extensively studied over the last 40 years, and great strides have been made in our understanding of the molecular mechanisms that underlie this process (see ). One of the most significant discoveries in this field was the identification of the cyclins  and the understanding of the way in which their activity is regulated by the process of ubiquitination to fulfil particular objectives at specific stages of the cell cycle . The precise way in which a cell provides itself with the necessary nutrients to accomplish the task of dividing has, however, remained unresolved. Research in the 1970s, using fibroblasts stimulated by serum to proliferate, established that there is a period in cell division during G1, a few hours before entry into S-phase, in which cells become committed to DNA replication . These findings were later extended and the term ‘restriction point’ was introduced  to describe the stage in G1 beyond which cells no longer require the presence of mitogenic signals for progression along the cell cycle. This restriction point was also shown to be the stage at which the cells require nutrients . A similar stage in G1, called ‘start’, was also identified in yeast . These studies suggested that the provision of the metabolic substrates required for cell division/proliferation is intimately linked with the progression of the cell from G1- to S-phase, the latter being the period in which most biosynthetic reactions required for cell duplication seem to occur.
REGULATION OF CELL-CYCLE PROGRESSION BY THE UBIQUITIN LIGASES
Cell division is a finely co-ordinated process in which the timely functioning and degradation of cell-cycle proteins play a fundamental role. Two ubiquitin ligase complexes, APC/C (anaphase-promoting complex/cyclosome)  and SCF (Skp1/cullin/F-box) , control the sequential degradation of these proteins .
APC/C is an E3 ubiquitin ligase that plays an essential role in mitosis as well as in G1. In order to initiate degradation it utilizes two activator proteins, Cdc (cell division cycle) 20 and Cdh1, which associate with the enzyme in a cell-cycle-dependent manner and target different substrates for proteasomal degradation . Substrates bind specifically to the APC/C-activator complex through their degradation motifs, the best known of which are the D box (destruction box) [7,15,16] and the KEN box . Although both activator proteins recognize the D box, the KEN box is a preferred targeting signal for APC/C-Cdh1 . The D box, which was originally found in cyclin B, has a short sequence usually represented as a highly conserved RXXL motif, but is better described as RXXLXX[LIVM]. Subsequent studies demonstrated D box-dependent degradation of other key cell-cycle components [18–20]. The proposed D boxes in Cdc6 (i.e. RXXLXXA)  or the modified D box identified in Sgo1 (i.e. HXXLXXI)  do not match the extended motif but are still functional recognition sites for APC/C-Cdh1. The first KEN box to be identified was within Cdc20 itself . Later, functional KEN boxes were also reported within other human proteins, including Cdc6  and securin . Another fundamental difference between APC/C-Cdh1 and APC/C-Cdc20 lies in its regulation. Although the latter complex is activated by the CDK (cyclin-dependent kinase)-dependent phosphorylation of certain APC/C subunits, for example Cdc27 , the former only functions when Cdh1 is in a dephosphorylated state .
The different complexes of the ubiquitin ligase SCF are active during most stages of the cell cycle, in which they carry out a variety of functions, including the regulation of various CDKs . SCF recognizes its substrates via the action of F-box activator proteins such as Skp2, β-TrCP (β-transducin repeat-containing protein) or Fbw7 . Most SCF substrates are recognized and bound by the F-box protein subunit when the substrates are phosphorylated on specific sites . In the case of SCF-β-TrCP, the ubiquitin ligase recognizes its substrate protein via a DSGXXS motif [27,28] in which both serine residues are usually phosphorylated.
APC/C-Cdc20 regulates the proteins responsible for metaphase-to-anaphase transition, whereas APC/C-Cdh1 maintains cells in G1, thus preventing their unco-ordinated entry into a new cell cycle . This is achieved through the degradation of a number of proteins involved in progression into S-phase . Induction of proliferation eventually leads to phosphorylation of Cdh1 , its subsequent degradation by ubiquitination  and a gradual time-dependent decrease in APC/C-Cdh1 activity. This inactivation of APC/C-Cdh1 during mid-to-late G1 results in the appearance of S/M cyclins responsible for entry into S-phase. APC/C-Cdh1 activity has also been reported to be inhibited at G1 to-S transition by binding of Emi1 .
METABOLIC PROTEINS AS SUBSTRATES FOR CELL-CYCLE UBIQUITIN LIGASES
The utilization of glucose
The finding that the activity of the enzyme PFK (phosphofructokinase) 2 is regulated by APC/C-Cdh1  led to the identification of the role of this ubiquitin ligase in the metabolic regulation of the cell cycle and therefore of cell proliferation [34–36]. PFK2 generates fructose-2,6-bisphosphate, the most potent allosteric activator of the key glycolytic enzyme PFK1 . Thus, although PFK2 is not, strictly speaking, a glycolytic enzyme, its activation is one of the most effective ways of stimulating glycolysis (see ). PFK2 has been renamed PFKFB (6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase) in order to describe its role as a family of bifunctional enzymes that catalyse the conversion of fructose 6-phosphate to fructose 2,6-bisphosphate and the reverse reaction . There are four PFKFB isoforms encoded by different genes, of which isoform 3 (PFKFB3) is highly expressed in a variety of different organs and tumour cells [40,41] as well as during development . PFKFB3 has the highest kinase-to-bisphosphatase ratio (740:1) and its main action is therefore the activation of glycolysis (see ). We have demonstrated, using a variety of mammalian cells, that PFKFB3 contains a KEN box motif but the other isoforms do not .
Following the demonstration that PFKFB3 is a substrate for APC/C-Cdh1 in neurons , we suggested that its degradation by the ubiquitin ligase might be an integral part of G1 in the cell cycle and that an increase in PFKFB3 following the decrease in APC/C-Cdh1 might be the mechanism that co-ordinates the provision of glucose with all of the other steps necessary for progression into S-phase . We went on to show in two different cell types, neoplastic and non-neoplastic, that both cell proliferation and glycolysis are prevented by overexpression of Cdh1 and enhanced by its silencing. Furthermore, using a form of PFKFB3 with a mutation in its KEN box rendering it resistant to ubiquitination by APC/C-Cdh1, we demonstrated that, although glycolysis is essential for cell proliferation, glycolysis by itself in the presence of high Cdh1 does not result in cell proliferation . These findings were confirmed in human proliferating T-lymphocytes , thus demonstrating that the mechanism involved in this process is the same for the proliferation of freshly isolated normal cells and is not a phenomenon restricted to cell lines, whether cancerous or normal. These studies also established that, like the cyclins [43,44], the activity of PFKFB3 during the cell cycle is dependent on two stages. The first requires the expression of the gene and consequent synthesis of the protein, both of which occur early on. The enzyme thus generated is continuously ubiquitinated until it becomes active only when APC/C-Cdh1 decreases during mid-to-late G1.
During our studies using human lymphocytes, we came to the conclusion that one of the main difficulties with investigating the metabolism of proliferating cells is that they divide in an asynchronous manner. Because of this, the information obtained at any time point in relation to utilization of substrates, biosynthetic reactions or signalling pathways is the algebraic sum of processes occurring at different stages of cell division. We therefore decided to initiate studies on the metabolism of cells that had been synchronized using standard procedures routinely used by investigators studying the cell cycle. In such cells it is possible to disentangle different events and follow them as they occur in real time. As a result, we have recently correlated the presence of PFKFB3 with the different stages of the cell cycle using HeLa cells treated with either DTB (double thymidine block) or DTB plus nocodazole to synchronize them at the G1/S boundary or in G2/M respectively . PFKFB3 protein levels, which were initially below the limit of detection, rose sharply in mid-to-late G1 following the disappearance of Cdh1. This peak of PFKFB3 was short-lived, lasting only 2–4 h, and correlated with transient enhanced generation of lactate (Figure 1) [45,46]. The disappearance of PFKFB3 after the transition from G1 to S, at a time when APC/C-Cdh1 is no longer active, was shown to be due to the action of SCF-β-TrCP . We found that PFKFB3 also contains a consensus site for β-TrCP recognition (the DSGXXS motif) and that Ser273 is the specific residue whose phosphorylation is required for recognition of the protein by the ubiquitin ligase [45,46] at the onset of S-phase. Thus the presence of PFKFB3 is tightly controlled to ensure the up-regulation of glycolysis at a specific point in G1.
Changes in protein levels of Cdh1, PFKFB3 and GLS1 during the cell cycle and their metabolic consequences
When PFKFB3 was silenced, the synchronized HeLa cells did not progress from G1- to S-phase, but remained predominantly in G1. Furthermore, in cells arrested by glucose deprivation, progression into S-phase after replacement of glucose only occurred if PFKFB3 was present or was substituted by ectopic expression of PFK1, indicating the necessity for up-regulation of glycolysis at this stage of the cell cycle . These results clearly show the intimate link between cell-cycle progression and the provision of glucose, without which cells are unable to divide.
The utilization of glutamine
As early as the 1950s it was known that glutamine is essential for the growth of both normal (mouse fibroblasts) and cancer (HeLa) cells . Later work revealed that glutamine, through its conversion into glutamate and further processing in the TCA (tricarboxylic acid) cycle, is an essential partner of glucose in cell proliferation [3,48]. The first step of glutaminolysis – the conversion of glutamine into glutamate together with the generation of ammonia from the amido group of the substrate – is catalysed by glutaminase. This enzyme, the activity of which is increased in certain tumours  and in proliferating lymphocytes , has recently generated a great deal of interest as a proposed target for anticancer therapy . The two isoforms of glutaminase are known to contain KEN boxes. This background led us to look for, and to identify, glutaminase as a target for APC/C-Cdh1 . We found that the most abundant form in human proliferating lymphocytes is GLS1 (glutaminase 1), the isoenzyme most often associated with proliferating cancer cells [52,53]. Our experiments in proliferating lymphocytes showed that the accumulation of GLS1 occurred as APC/C-Cdh1 decreased and coincided with an increase in the utilization of glutamine (measured as an increase in the generation of ammonia) as well as with the enhanced glycolysis described above (Figure 1). Overexpression of Cdh1 decreased both the proliferative response and the accumulation of ammonia; the concomitant overexpression of a mutated form of GLS1 that was no longer a substrate for APC/C-Cdh1 restored the production of ammonia, but not the proliferative response . Experiments in which the expression of GLS1 was reduced by silencing Gls1, or in which its activity was inhibited pharmacologically using 6-diazo-5-oxo-L-norleucine, confirmed the necessity for this enzyme in the proliferative response of T-lymphocytes .
We also found that, although GLS1 is ubiquitinated by APC/C-Cdh1, it differs from PFKFB3 in that its recognition by this ubiquitin ligase requires the presence of both a KEN and a D box, rather than a KEN box alone . Furthermore, using synchronized HeLa cells, we have shown that GLS1 is not a substrate for SCF-β-TrCP and is not degraded until late mitosis, at which time APC/C-Cdh1 becomes active again. The utilization of glutamine in those experiments correlated with the increase in GLS1 protein at mid-to-late G1 and remained elevated until GLS1 was degraded by APC/C-Cdh1 (Figure 1).
GLUCOSE AND GLUTAMINE: DIFFERENT SUBSTRATES FOR DIFFERENT PURPOSES AND BEYOND
The contrasting post-translational regulation of PFKFB3 and GLS1, which we have verified by studies of ubiquitination and protein stability , suggests that glycolysis and glutaminolysis play different roles at distinct stages in the cell cycle (Figure 2). Indeed, experiments in which HeLa cells were synchronized at either the G1/S boundary or G2/M and then released into medium lacking either glucose or glutamine showed that both glucose and glutamine are required for progression through the restriction point, whereas only glutamine is necessary for progression through S-phase into cell division. Similar results were obtained in experiments in which PFKFB3 or GLS1 was silenced .
Schematic diagram of the way in which the activities of PFKFB3 (controlled by both APC/C-Cdh1 and SCF-β-TrCP) and GLS1 (controlled by APC/C-Cdh1) regulate glycolysis and glutaminolysis at different stages of the cell cycle
Bioinformatic analysis of other enzymes involved in the metabolism of proliferating cells indicates that a number of them are also potential targets of cell-cycle ubiquitin ligases (see ). This suggests that the regulation of metabolic pathways by this system may be more widespread than we have so far established.
Our studies show that the activity of crucial metabolic enzymes ‘oscillates’ during the cell cycle in a similar fashion to that of the cyclins . The cycling of metabolic proteins is regulated by the same ubiquitin ligases that control the function of the cyclins and therefore the activity of these enzymes explains the co-ordination of cell-cycle progression with the supply of substrates necessary for cell duplication. Experimental studies of APC-target proteins have shown that some of them contain only a D box, others contain only a KEN box, and some contain both (Figure 3). The D and the KEN boxes can act as independent or co-ordinated entities for protein degradation. Of the metabolic proteins that we have investigated, PFKFB3 is degraded through an action exclusively via the KEN box whereas GLS1 requires the presence of both KEN and D boxes, as do other proteins such as Sgo1 , Acm1  or FoxM1 .
Location of degradation motifs in substrates for APC/C-Cdh1 and in those that are also substrates for SCF-β-TrCP
PFKFB3 also contains a DSGXXS motif known to be the substrate for SCF-β-TrCP. This box is represented in PFKFB3 by the sequence DSGLSS, but different sequences have been identified in other proteins . SCF-β-TrCP recognizes the DSGXXS motif only when it is phosphorylated, and we have established that Ser273 is the phosphorylation site in PFKFB3 during S-phase. Thus the cycling of the protein is closely controlled by the two ubiquitin ligases acting in concert. Other proteins with consensus recognition motifs for APC/C-Cdh1 and SCF-β-TrCP may oscillate in a similar manner. This might be the case for Cdc25A  – a phosphatase inhibitor of Cdk2, and claspin  – a component of the DNA replication checkpoint, both of which proteins are known to be degraded via these two ubiquitin ligases. The fact that GLS1 persists throughout S-phase, unlike PFKFB3, indicates that glutaminolysis is required beyond G1/S transition.
We have established that the decrease in the activity of APC/C-Cdh1 in mid-to-late G1 releases not only PFKFB3 to activate glycolysis allosterically through the synthesis of fructose 2,6-bisphosphate, but also GLS1 to activate glutaminolysis. This explains the simultaneous increase in the utilization of both substrates during cell proliferation that was reported many years ago . This occurs at a time consistent with the point in G1 that has been described as the nutrient-sensitive restriction point  and which controls transition from G1 to S. It has recently been suggested that the restriction point may have two distinct components, namely a growth factor restriction point that occurs early in G1 and a later point, called the ‘cell growth’ checkpoint, which is responsible for detecting nutritional sufficiency . Our studies suggest that both points might be part of a continuum that is initiated with the early phosphorylation of Cdh1 and the consequent time-dependent decrease in activity of APC/C-Cdh1.
Elucidation of the steps that follow the decrease in activity of this ubiquitin ligase is now of great interest since progression through G1 is not only a central point of co-ordination of cell-cycle progression with metabolism, but also a site at which a great deal of dysregulation occurs in neoplastic transformation [61,62]. Since our studies have shown that Cdh1 lies at the crossroads of two of the hallmarks of cancer, i.e. proliferation and increased cell metabolism , the findings in this area could significantly increase our understanding of the changes occurring during oncogenesis. Interestingly, an alternative splice variant of GLS1 (GAC) that lacks the region containing the motif for recognition by APC/C-Cdh1  is the predominant glutaminase form expressed in certain cancer types [65–67]. Furthermore, PFKFB4, an isoform of PFKFB3 that does not contain a KEN box, has recently been identified as an essential component for the survival of glioma cells and prostate cancer cells [68,69]. Whether these changes provide a metabolic advantage for cancer cells requires further investigation. Recent results have shown that the tumour suppressor PTEN promotes APC/C-Cdh1 activity  and that cells from mice overexpressing PTEN exhibit reduced glucose and glutamine uptake and are resistant to oncogenic transformation . Studies on the possible connection between APC/C-Cdh1 and the metabolic effects of other tumour suppressors such as p53  or known proto-oncogenes such as c-Myc  and Akt  are also likely to be highly relevant.
From G1 to S transition onwards, the activity of PFKFB3 and GLS1 differs (Figure 4), since that of PFKFB3 is short-lasting, coinciding with a peak in glycolysis in mid-to-late G1, while the activity of GLS1 and the uptake of glutamine remain high throughout S-phase until the subsequent increase in APC/C-Cdh1 in late mitosis. This difference, due to the distinct regulation of the activity of each enzyme, suggests that glycolysis is only required by the cell for selective purposes. This needs to be investigated and one possible starting point would be the identification of the kinase responsible for the phosphorylation of the DSGXXS motif in PFKFB3, since that might provide clues to the metabolic significance of this step.
Schematic diagram of metabolic changes at the restriction point following activation by growth factors
It has been known for many years that the rates of utilization of glucose and of glutamine are very high during cell proliferation and that only a small percentage of each precursor (~5%) is used for the synthesis of macromolecules [3,75]. The explanation for this apparently wasteful process, in which the majority of the substrate is excreted as lactate, aspartate, glutamate or alanine , has always been intriguing. The most satisfactory explanation to date was provided in the mid-1980s with the suggestion that the high rates of glycolysis and glutaminolysis are necessary to ensure that the anaplerotic pathways are adequately supplied during the enhanced needs of proliferation without the risk of running out of substrate [3,75].
The glucose that is not converted into lactate during G1- to S-phase could be used in other reactions, including nucleotide biosynthesis via the pentose phosphate pathway, fatty acid synthesis through citrate production from glucose carbons in the TCA cycle, amino acid synthesis, and glycerogenesis. In addition to these well-known pathways, it has recently been suggested that glucose may play a role in cell signalling and gene expression. Thus glucose, together with glutamine, is necessary for the formation of hexosamine, which is important in modifying proteins and lipids for their role in signal transduction . Furthermore, histone acetylation, and therefore gene expression, has been linked to glucose utilization through the actions of ATP citrate lyase .
It is not yet clear whether glucose is used at any other stage in the cell cycle for purposes not associated with the generation of lactate. What our experiments do suggest is that, if glucose does have a role beyond G1/S transition, it can be substituted by glutamine, since removing glucose from the cell medium after this stage has no effect on cell-cycle progression, providing glutamine is present. Glutamine is utilized throughout S-phase and might indeed be the main anaplerotic substrate at this stage, since all cell components can be synthesized from its metabolism via the TCA cycle. Whether this selective use of glutamine in S-phase, which seems to be a common feature of all of the proliferating cells we have studied, is executed by the same mechanism in normal and in cancer cells or whether cancer cells somehow utilize a different mechanism, as has been surmised , needs to be established.
In summary, we have identified the molecular steps underlying the metabolic nature of the restriction point in mid-to-late G1. Glycolysis and glutaminolysis are closely associated with the cell cycle through their control by APC/C-Cdh1. Together they represent the most significant metabolic steps in the process of cell division. Other biochemical and molecular mechanisms associated with these crucial steps now need to be investigated. The use of synchronized cells will provide the methodological tool for these studies. This will result in the generation of a time-dependent flow chart of the metabolic events in the cell cycle that underpin proliferation. Careful comparison between proliferating normal and cancer cells will indicate whether there are relevant differences that justify a ‘cancer-selective’ therapeutic approach.
cell division cycle
box, destruction box
double thymidine block
6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase, isoform 3
β-transducin repeat-containing protein
We thank Professor Jorge Erusalimsky and Dr Slavica Tudzarova for their helpful comments on the paper prior to submission.
Work in the authors' laboratory is supported by the Wellcome Trust [grant number 086729 (to S.M. and S.L.C.)].