https://orcid.org/0000-0003-1384-8425
Abstract
Growing cells increase multiple biosynthetic processes in response to the high metabolic demands needed to sustain proliferation. The even higher metabolic requirements in the setting of cancer provoke proportionately greater biosynthesis. Underappreciated key aspects of this increased metabolic demand are guanine nucleotides and adaptive mechanisms to regulate their concentration. Using the malignant brain tumour, glioblastoma, as a model, we have demonstrated that one of the rate-limiting enzymes for guanosine triphosphate (GTP) synthesis, inosine monophosphate dehydrogenase-2 (IMPDH2), is increased and IMPDH2 expression is necessary for the activation of de novo GTP biosynthesis. Moreover, increased IMPDH2 enhances RNA polymerase I and III transcription directly linking GTP metabolism to both anabolic capacity as well as nucleolar enlargement historically observed as associated with cancer. In this review, we will review in detail the basis of these new discoveries and, more generally, summarize the current knowledge on the role of GTP metabolism in cancer.
Guanine nucleotides are building blocks for DNA and RNA and are also utilized by a family of guanine binding proteins (G-proteins) for a myriad of cell functions, such as cytoskeletal rearrangements, membrane traffic, protein synthesis and signal transduction (1). Historically, guanine was named after guano because it was one of the first metabolites of nucleotide metabolism discovered in this naturally occurring fertilizer. Indeed, guano is one of the most effective natural fertilizers—extracted from the accumulated excretes of birds or bats—and has been used from at least pre-Inca era (B.C. 500) to today. At the time, fertilizers for agriculture was prime importance not only for farmers but also as a commercial product for government in a similar fashion that to natural gas and oil in the modern era. Like gold, guano mining was widespread in many countries with Peru being the most prominent. Competition to possess its great value led to conflicts including the War of the Pacific (A.D. 1879–1883). In this review, we will introduce the new insights of the guanine nucleotide metabolism—which has such ancient history associated with human civilization and wealthiness—in the context of cancer, regarding how cellular GTP levels in related functionally to ATP levels are regulated through the key enzyme of GTP biosynthesis, inosine monophosphate dehydrogenase (IMPDH).
De Novo and Salvage Biosynthesis of GTP and ATP
Purine nucleotides are biosynthesized through two pathways—salvage and de novo synthesis—in the cell (Fig. 1). The salvage pathway uses the already assembled nucleobase (e.g. hypoxanthine, adenine, guanine) or purine nucleoside (e.g. inosine, adenosine, guanosine) to produce purine mononucleotide [e.g. inosine monophosphate (IMP), GMP, AMP]. There are three salvage ATP biosynthesis pathways: (i) union of adenine and phosphoribosyl pyrophosphate (PRPP) to produce AMP by adenine phosphoribosyl transferase; (ii) direct phosphorylation of adenosine to AMP by adenosine kinase; and (iii) association of hypoxanthine and PRPP by hypoxanthine phosphoribosyl transferase-1 (HPRT1) to produce IMP that is converted to AMP. In contrast, and interestingly, there is a single salvage GTP biosynthesis pathway, in which HRPT1 is the central and essential enzyme. Hypoxanthine or guanine, which can be derived from guanosine cleavage at its N-glycosidic bond, is enzymatically joined with PRPP to produce IMP or GMP, respectively, via HPRT1. HPRT1 mutation in human has been shown to decrease guanine nucleotide levels particularly in the brain, an organ highly dependent on salvage GTP biosynthesis (2). As evidence for this dependence, complete deficiency causes Lesch-Nyhan syndrome, an X-linked recessive genetic disorder with severe abnormal neurological manifestations (3, 4).
Purine nucleotide biosynthesis. Purine nucleotides are synthesized through two pathways, de novo pathway and salvage pathway. Glucose is metabolized to Ribose 5-phosphate through the pentose phosphate pathway. Then, Ribose 5-phosphate is converted to PRPP, which is the first metabolite of de novo purine biosynthesis. IMP can be converted into both ATP and GTP by ADSS and IMPDH, respectively.
In marked contrast to the salvage purine synthesis pathways, de novo purine synthesis is a highly energy-consuming process. For example, in the salvage pathway, GTP synthesis from guanine or hypoxanthine requires 2 or 3 ATPs and 3 or 5 reactions, respectively, whereas de novo GTP synthesis from glucose requires 9 ATPs and 20 reactions (Fig. 1). Although de novo GTP and ATP biosynthesis pathways have in common the metabolic pathway converting glucose to IMP, divergence occurs at the IMP substrate where IMPDH and adenylosuccinate synthase (ADSS), for GTP and ATP synthesis, respectively (Fig. 1). Parenthetically, we appreciate how the de novo purine synthesis pathways were so well-identified and characterized by the early biochemists with limited resources, information and technology of 1950s–1960s. Despite these conditions, their work was so rigorous that the purine metabolism pathways described in biochemistry textbooks remain essentially the same for years. However, what has evolved is the relationship of these definitively defined pathways to normal physiology and disease. A focus of this review is growing appreciation of disordered purine metabolism in the setting of diseases, particularly cancers. For example, Traut (5) reported that the level of GTP in tumour cells increased by nearly 200% compared to that in normal cells, although the level of ATP increased by ∼20%. We re-examined this observation and speculated that this might reflect a property of tumours in which the rate of GTP biosynthesis or consumption, or both, alters more dramatically than that of ATP biosynthesis. Our recent published findings support this speculation and suggest that this is indeed operative under the regulatory actions of IMPDH.
Transcriptional Regulation of IMPDH1 and IMPDH2 Expression
Two IMPDH genes, IMPDH1 in chromosome 7 and IMPDH2 in chromosome 3, have been identified in humans (6, 7). After identification of IMPDH1 and IMPDH2 genes, their 5′ promoter regions were cloned (8–10). Subsequent studies have identified two transcriptional factors, MYC and microphthalmia-associated transcription factor (MITF), as the primary activators of IMPDH transcription (11–17). Both MYC and MITF belong to the same family of basic helix-loop-helix leucine zipper transcription factors, which recognize the canonical E-box sequence (CACGTG) (18, 19). Although the canonical E-box sequence has not yet been identified in the promoter region of IMPDH2 gene (8), MYC and MITF may share the same target sequence in the promoter of IMPDH2. Liu et al. (11) reported that MYC regulates the expression of many nucleotide biosynthetic genes. Their ChIP assay showed MYC directly binds to the promoter of IMPDH2 and the first intron of IMPDH1, then regulating their expression. In haematopoietic stem/progenitor cells, Karigane et al. (16) reported that MITF binds to the promoter region of IMPDH2 gene and upregulates its transcription.
Consistent with these results, our analysis of the TCGA database shows that MYC expression is significantly correlated with IMPDH2 expression (Fig. 2A). Strikingly, overexpression or knockout of MYC further increases or decreases IMPDH2 protein expression levels in human glioblastoma U87MG and LN229 cells, respectively (Fig. 2B and C). Overexpression of MYC also increases IMPDH1 (Fig. 2B and C), which is consistent with the report in glioblastoma (GBM) initiating cells (13). However, according to the TCGA database analysis, correlation coefficient of MYC and IMDPH2 (r = 0.41) is significantly more than that of MYC and IMPDH1 (r = 0.26) (Fig. 2A). Likewise, protein levels of IMPDH2 in glioma far exceed to that of IMPDH1 (20). These data suggest that the presence of additional mechanisms that regulate IMPDH2 via either yet to be identified transcriptional factors or regulation at the protein level.
MYC upregulates the expression of both IMPDH1 and IMPDH2. (A) Correlation of the expression of IMPDH isozymes and MYC in glioblastoma. Each plot indicates the retrieved data from glioblastoma patients (TCGA, HG-U133A platform, n = 538). Pearson’s correlation coefficient (r) values are shown. P-values (two-tailed) were calculated using a t-distribution with n−2 degrees of freedom. (B) MYC gene was knocked out in glioblastoma cell lines, U87MG and LN229, using CRISPR-Cas9 technology. IMPDH1 and IMPDH2 were detected with their specific antibodies. (C) MYC was expressed in a doxycycline-dependent manner in glioblastoma cell lines.
IMPDH1 and IMPDH2 Protein Expression Levels in Tissues and Cancers
Human IMPDH1 and IMPDH2 have 84% identity at protein levels. Consistent with the highly conserved amino acid sequence in the catalytic domain, in vitro analysis using recombinant enzymes show that IMPDH1 and IMPDH2 display very similar kinetics. However, IMPDH2 is slightly more sensitive to the inhibitor, mycophenolic acid (MPA) than IMPDH1 (7), suggesting the possibility of functional differences between IMDPH1 and IMPDH2. In addition, there could be other modes of regulation at the level of individual protein expression and/or at the post-translational as these two IMPDH isozymes show clear differences.
To rigorously address the relative roles of the IMPDH1 and two isozymes and to quantitatively compare IMPDH1 and IMPDH2 expression at protein levels, we set up the quantitative assay using recombinant IMPDH1 and IMPDH2 as standards (20). To the best of our knowledge, this is the first conclusive comparison of the relative abundance of two IMPDH isozymes. Our quantification of cellular levels of the IMPDHs showed 90 times and 7 times lower expression of IMPDH1 than that of IMPDH2 in U87MG cells and LN229 cells, respectively (20). Immunodetection of IMPDH1 in many glioblastoma cell lines, glioblastoma stem cells, glioblastoma patient-derived xenograft models and human glioma specimens showed marginal signal, while that of IMPDH2 showed robust signal.
To examine whether our finding of higher IMPDH2 protein expression than that of IMPDH1 could be found in each tissue, we took advantage of the publicly available proteomic data and compared the abundance of IMPDH1 and IMPDH2 in multiple tissues (https://pax-db.org/) (21). As shown in Fig. 3, the level of IMPDH2 is higher than that of IMPDH1 in most tissues except for the retina, in which mutations of IMPDH1 are identified in patients of the RP10 form of autosomal dominant retinitis pigmentosa as described in the later part of this article.
Comparison of the expression levels of IMPDH isozymes among tissues. Protein abundance of IMPDH1 and IMPDH2 in the selected human tissues is visualized. The level of IMPDH2 is higher than that of IMPDH1 in most tissues. The retina, where IMPDH1 defect is identified, has high expression of IMPDH1. Data were retrieved from PAXdb (https://pax-db.org), which provides quantitative proteomics information (21). Protein abundance is shown in parts per million (ppm), which describes each protein with reference to the entire expressed proteome.
Post-Translational Regulation of IMPDH1 and IMPDH2
Recent single-cell studies show that, even in two-dimensional (2D) cultured cells, cellular activities are heterogeneous and intracellular signalling that could modulate metabolism oscillate in an order of several minutes (22, 23). Cells are exposed to continuous changes in availability of extracellular metabolites, and cellular metabolic demands can be greatly affected by the fluctuating cellular activities on the order of minutes or possibly seconds. In fact, our data show the rapid GTP biosynthesis and consumption rates in glioblastoma cells, where ∼90% of cellular GTP is consumed within 4 h (20). While transcriptional control of IMPDH levels is important for sustained metabolic regulation, cells have developed multiple forms of post-translational regulation of IMPDH.
Accumulating evidence suggests that the activity of both IMPDH1 and 2 are regulated allosterically via guanine nucleotides based on the in vitro experiments (24–30). The cystathionine β-synthetase domain, also called the Bateman domain, is important for this regulation. Binding of GTP or GDP to this domain induces compaction of IMPDH octamer, compromising the activity. Importantly, mutations associated with human retinopathies abrogate this allosteric inhibition, suggesting a physiological implication of the allosteric inhibition.
Some metabolic enzymes including IMPDH are known to form filament-like structures in response to the environmental change (28, 31, 32). Both IMPDHs can form filaments when cellular GTP levels are decreased (32). Filament formation seems to be related in some fashion to cell proliferation (29, 33). However, there is debate as to whether or not filament formation of IMPDHs increases the enzymatic activity (34). Recent studies have more explored the mechanism of this regulation. Johnson and Kollman (35) reported that filament formation of IMPDH2 desensitizes itself to the feedback inhibition of GTP and promotes the activity. Fernández-Justel et al. (30) also showed that filament formation is more resistant to GTP/GDP allosteric inhibition. There are clearly complicated mechanisms regulating IMPDH activity through filament formation and the allosteric inhibition, and in vivo significance of the IMPDH filaments still remain unclear.
Proteasomal degradation of both IMPDH isozymes implicated in GTP regulation were reported (36). ANKRD9, ankyrin repeat domain 9, is a ubiquitin ligase substrate receptor subunit that interacts with IMPDH1 and IMPDH2 with higher affinity binding to IMPDH2, suggesting the regulation of the abundance of IMPDH isozymes by the proteasomal system. Furthermore, ANKRD9 was reported to form rod-like structures with IMPDH2 under nutrient-limited condition (37). The Cys109 and Cys110 residues of ANKRD9 are required for the binding of IMPDH1 and IMPDH2. The mutations in the residues prevent the colocalization of ANKRD9 with IMPDH filaments and increase the abundance of IMPDH2, suggesting the negative regulation of IMPDH2 filament by ANKRD9.
IMPDH was also reported as an insulin-regulated phosphoprotein and was translocated to lipid bodies in a PI 3-kinase-dependent manner (38). Consistent with this, the interaction with and phosphorylation by serine/threonine protein kinase B (PKB), known as AKT, was reported. However, the specific sites of phosphorylation by PKB/AKT have not been identified yet and the role of the phosphorylation is not clear (39). Very recently, IMPDH1 was reported to be phosphorylated at three domains, T159/S160 in the Bateman domain; S416 in the catalytic domain; S477 at the C-terminus (40). Among them, an in vitro kinase assay suggests that protein kinase Cα (PKCα) could be a kinase responsible for T159/S160 phosphorylation. In vitro experiments show that phospho-mimic mutant IMPDH1 at T159/S160 sites desensitizes IMPDH1 to the allosteric inhibition by GTP and GDP. Correspondingly, treatment of bisindolylmaleimide, inhibitor of mixed PKC isotypes, in vivo decreased GTP levels when retinas were exposed to light, suggesting the important role of PKC, presumably through IMPDH1 phosphorylation at T159/S160 site, in the maintenance of GTP levels in the retina. Together, future research exploring the in vivo significance of phosphorylation, or post-translational modification, dependent regulation of IMPDH enzymes has high potential to uncover cross-talk of GTP metabolism and cellular signalling pathways.
Physiological Role of IMPDH1 and IMPDH2
Understanding the roles of IMPDH isozymes in physiological conditions is even less explored. There are two reports in which Impdh2 gene knockout mice were studied. Mitchell’s group has shown that deletion of both alleles of Impdh2 in mice leads to early embryonic lethality despite the existence of intact IMPDH1 (41). Heuckeroth’s group showed the essentiality of IMPDH2 in the early neural crest using Impdh2 conditional knockout (KO) mice (42). These results indicate that IMPDH2 has the essential function in embryogenesis that cannot be compensated by IMPDH1. In contrast, mice with the disruption of Impdh1 gene are born with the expected Mendelian ratio but postnatally develop a progressive defect in the retina (43, 44). Correspondingly, mutations in human IMPDH1 gene were identified in the patients of the RP10 form of autosomal dominant retinitis pigmentosa (45). Furthermore, as Fig. 3 indicates the high level of IMPDH1 expression in the retina, it is likely that there would be a physiologic consequence of disrupted expression. Although the expression of IMPDH2 is not low in the retina, IMPDH2 cannot compensate IMPDH1 deficiency, suggesting the specific role of IMPDH1 in the retina, possibly via the regulation of phosphorylation by PKCα (40).
Historical Connection of IMPDH in the Context of Cancer
An anti-tumour activity of the IMPDH inhibitor, MPA, has been postulated since the 1960s (46–49). In 1975, a landmark report unveiled the first mechanistic connection of IMPDH and cancers (50). The detailed enzymatic analysis showed a significant correlation of IMPDH activity and proliferation rate of liver cancer cells. The association of IMPDH activity and rapid cell proliferation was also demonstrated by solid kinetic studies using liver tissue in which the activity of IMPDH became about five times higher in regenerating liver 24 h after partial hepatectomy.
More recently, upregulation of IMPDH2 has been reported in multiple cancers including glioblastoma (20), leukaemia (51), colorectal cancer (52, 53), nasopharyngeal carcinoma (54), prostate cancer (55), a subset of small cell lung cancers (15) and kidney and bladder cancer (56).
However, questions remain including (i) why IMPDH2 upregulation is important for cancers and (ii) what is the metabolic utility of IMPDH2-synthesized GTP in cancers? Our studies suggest an exquisitely coordinated mechanism of IMPDH, cellular anabolism and nucleolar enlargement; we will first introduce a series of historically important findings providing the basis of these new roles for GTP.
IMPDH-Dependent GTP Biosynthesis Fuels Nucleolar RNA Synthesis Revealed by Stable-Isotope Measure of Influxed Ribonucleic Acid Index
In 1976, Grummt and Grummt (57) reported that intracellular pool of ATP and GTP regulated nucleolar RNA synthesis. At the same time, the Grummt’s group discovered that amino acids availability was a critical factor for nucleolar RNA synthesis. Later on, it was shown that this nucleolar RNA synthesis is driven by the designated polymerase, RNA polymerase I (Pol I). Pol I is a very unique polymerase because it only has a single target gene, ribosome DNA, highly contrasting with other RNA polymerases who have multiple target genes (58). Also, Pol I is the most potent RNA polymerase and its product, rRNA, is the major constitutes (>80%) of cellular total RNAs. Enhanced RNA Pol I activity is directly associated with increased ribosome levels, both a direct reflection of proteins synthesis essential for rapid cell proliferation (59, 60).
The initial finding of the connection between amino acid regulation and nucleolar RNA synthesis is now well appreciated and has been further clarified by the studies of mTOR signalling. mTOR signalling activates RNA Pol I through its components, transcription initiation factor I (TIF-IA) and Upstream binding factor (UBF) (61, 62). Moreover, growth factor signalling through Ras/Extracellular signal-regulated kinase (ERK) and PI3K, in addition to mechanistic target of rapamycin (mTOR), coordinately activate RNA Pol I (63, 64). In addition, upregulation of MYC induces systemic elevation of genes required for ribosome biogenesis (65, 66). Taken together, these extensive studies demonstrate the requirement of RNA Pol I upregulation to foster the highly anabolic rapid cell proliferation characteristic of cancers. In contrast to the results of these signalling studies, the initial finding of purine nucleotides-dependent Pol I activation was relatively unappreciated until very recently.
Our studies are focused on deciphering the nature of the connection between RNA polymerase and nucleolar enlargement and cancer. An attractive model has been proposed regarding TIF-IA, an essential molecule for RNA Pol I transcription by recruiting RNA Pol I complex to ribosomal DNA promoter (67). In support of this model, biochemical studies suggest that TIF-IA has a novel GTP binding capacity and its GTP binding is necessary for TIF-IA’s activity. This would be the first molecular-based model explaining the requirement of GTP for Pol I transcription. Yet to be clarified is how TIF-IA recognizes specifically GTP given that the majority of GTP binding proteins contain four to five functional sequences, called G-box (or G-motif), and together form G-domain (1, 68). In contrast, TIF-IA contains only one G-box, GxxxGKS/T, which is also called a phosphate-binding loop found in many other nucleotide-binding proteins, including kinases. The structural analysis of TIF-IA binding to GTP may provide a concrete mechanistic insight and novel GTP binding module.
To define the metabolic utilization of IMPDH-derived GTP, we developed a method to detect newly synthesized nucleoside incorporation into each RNA, Stable-Isotope Measure of Influxed Ribonucleic Acid Index (SI-MOIRAI). The SI-MOIRAI technique unveiled that IMPDH-derived GTP is preferentially incorporated into rRNA and tRNA (20). Surprisingly, SI-MOIRAI showed the lower utility rate of IMPDH-derived GTP for mRNA synthesis than for rRNA and tRNA, suggesting a possible existence of metabolic channelling shunting the IMPDH-derived GTP towards RNA Pol I and III, which locates in the nucleolus for transcription. Furthermore, in contrast to high GTP incorporation into rRNA and tRNA, de novo synthesized ATP, UTP and CTP showed much less incorporation into the rRNA and tRNA. Consistent with these results, inhibition of RNA Pol I almost completely prevented GTP consumption in glioblastoma cells, while not affecting ATP. Together, the results revealed a surprising and unanticipated role of IMPDH-derived GTP in promoting nucleolar RNA transcriptions by RNA Pol I and III implicating rRNA and tRNA syntheses in enhanced anabolic capacity (Fig. 4).
Increased ribosomal biogenesis activated by de novo GTP biosynthesis. Glioblastoma activates de novo GTP biosynthesis by upregulating IMPDH2 expression. Newly synthesized nucleotides are used for the production of ribosomal and transfer RNA (not shown in the illustration) cooperated with growth factor-mediated signalling pathways such as mTOR and ERK.
Unexpected Role of IMPDH2 in Nucleolar Enlargement Revealed by Two-Dimensional Transcriptome Analysis
Enlarged nucleoli are hallmarks of malignant tumours but beyond the association it has been unclear whether it is an epiphenomenon versus integral to the mechanism of cancer (69). In our course of studies exploring IMPDH2 network, we combined functional IMPDH transcriptome analysis and pathophysiological IMPDH2 transcriptome analysis, and designated this approach as 2D transcriptome analyses (20).
In the first dimension, we sought for the IMPDH-regulated genes by transcriptome analysis identifying downregulated genes upon IMPDH inhibition in glioblastoma cells. In the second dimension, we took advantage of glioma patient databases and screened for genes whose expression levels were correlated with those of IMPDH2 in glioblastoma patients. In theory, the genes overlapping in the first and second dimension analyses should represent the genes regulated by IMPDH2 activity. When the overlapped genes were identified, the results unveiled a significant relevance of IMPDH2 activity and expression in nucleolar morphology. Strikingly, pharmacological inhibition of IMPDH activity as well as IMPDH2 knockout lead to the shrinkage of the nucleolar size, which was reversed by ectopic expression of IMPDH2 (20). We also found that nucleolar proteins, such as nucleostemin and nucleolin, were dispersed from the nucleolus proceeding to the nucleolar shrinkage following IMPDH inhibition. Moreover, this was accompanied by robust p53 activation, or nucleolar stress responses. Thus, continuously high IMPDH activity is critical to prevent malignant tumour cells from the cell suppressive programme of the nucleolus. Furthermore, increased IMPDH2 protein levels in glioma patient specimens correlates with enlarged nucleoli (20).
Collectively, our studies demonstrate that specific upregulation of GTP biosynthesis by IMPDH2 drives rRNA and tRNA transcription. Given that the levels of rRNA transcription are positively correlated to the nucleolar size (70), the finding suggests that IMPDH2-dependent upregulation of rRNA and tRNA transcriptions is the part of the direct mechanism of nucleolar hypertrophy and thereby the malignant growth of glioblastoma. These new results uncover a novel GTP metabolic requirements promoting nucleolar hypertrophy and function.
Concluding Remarks
GTP itself is a well-known biometabolite. However, as described in this review, GTP is not just a component of nucleic acids, but rather it plays a remarkably diverse roles in cells. The biosynthesis of GTP and its intracellular levels are well coordinately regulated, at least in part through IMPDH enzymes. While we and other laboratories have shown the elevated GTP promotes RNA Pol I transcriptions, cells may have a biological sensor of cellular GTP levels. In fact, we have discovered a GTP sensor (71–73). The existence of the GTP sensor implies that cellular functions can be tuned according to the GTP levels in a proactive rather reactive or passive manner. Future research investigating GTP metabolism and GTP-related biological phenomena will provide new insight into the development of novel drugs to intractable diseases including cancer.
Acknowledgements
We thank E.P. Smith for excellent editing of the manuscript.
Funding
This work was supported in part by JSPS KAKENHI (JP18K07233 to S.K.) and MTP UC-Brain Tumor Center grant, Ohio Cancer Research grant and National Institute of Health (NIH) (R21NS100077 and R01NS089815 to A.T.S.).
Conflict of Interest
None declared.
References
Abbreviations
- 2D
two-dimensional
- ADSS
adenylosuccinate synthase
- GTP
guanosine triphosphate
- HPRT1
hypoxanthine phosphoribosyl transferase-1
- IMP
inosine monophosphate
- IMPDH
inosine monophosphate dehydrogenase
- IMPDH2
inosine monophosphate dehydrogenase-2
- MITF
microphthalmia-associated transcription factor
- MPA
mycophenolic acid
- PKB
protein kinase B
- Pol I
polymerase I
- PRPP
phosphoribosyl pyrophosphate
- SI-MOIRAI
Stable-Isotope Measure of Influxed Ribonucleic Acid Index
- TIF-IA
transcription initiation factor I.
