| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Biochemistry and Biophysics |
CRC Biomedical Magnetic Resonance Research Group, Department of Biochemistry and Immunology, St. George’s Hospital Medical School, London SW17 0RE [J. R. G., P. M. J. M., S. P. R., H. T., Y-L. C., M. S.]; School of Pharmacy and Pharmaceutical Sciences, The University of Manchester, Manchester M13 9PL [K. J. W., I. J. S.]; and Institute of Molecular Medicine, John Radcliffe Hospital, Oxford OX3 9DU [R. D. L., A. L. H.], United Kingdom
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
5 times less than in the WT tumor [measured in tumor extracts (P < 0.001) and by metabolic imaging (P < 0.05)]. Noninvasive 31P MR spectroscopy showed that the nucleoside triphosphate/Pi ratio of the two tumor types was similar, so the low ATP content of the c4 tumors was not caused by (or a cause of) impaired cellular bioenergetics. Rather, glycine, an essential precursor for de novo purine formation, was significantly lower in the c4 tumors (P < 0.05), suggesting that ATP synthesis was impaired in the mutant tumor cells. Supporting evidence for this hypothesis came from the significantly lower concentrations of betaine, phosphocholine, and choline in the c4 tumors (P < 0.05); these are intermediates in an alternative pathway for glycine synthesis. No significant differences were seen in lactate or glucose content. MR resonances from phosphodiesters, which relate to the metabolic turnover of phospholipid membranes, were significantly lower in the WT tumors than in the c4 tumors, both in vivo (P < 0.05) and in extracts (P < 0.01). We propose that loss of up-regulation of expression of the genes for glucose transporters and glycolytic enzymes in the c4 tumors decreased formation of glycine, an essential precursor of ATP synthesis, and thus caused the low ATP content of the c4 tumors. In summary, these data suggest that disruption of the HIF-1 pathway in these tumor cells impairs the supply of anabolic precursors required for cell synthesis. They suggest potential biochemical targets that may be modified by therapy blocking HIF-1 function. |
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
HIF-1 is a heterodimer, consisting of 
and ß subunits, whose DNA binding activity has been shown to be up-regulated by hypoxia, a common feature of solid tumors because their blood vessels grow in a disorganized manner, limiting the supply of nutrients and oxygen to the tumor cells (6)
. Although hypoxic regulation of HIF-1 occurs principally through the subunit, it has been shown that hepatoma (HEPA-1) cells deficient (c4) in HIF-1ß (also known as the arylhydrocarbon nuclear translocator protein, a component of the xenobiotic response pathway; Ref. 7
) are unable to form a functional HIF-1 complex (8)
. Examination of oxygen-regulated gene expression in these HIF-1ß-deficient cells revealed reduced or absent induction by hypoxia of genes encoding glucose transporters, glycolytic enzymes, and vascular growth factors when compared with WT HEPA-1 cells (9)
. When grown under hypoxic conditions, the c4 cells are less well able to tolerate exposure to hypoxia; Stratford et al. (10)
found a 10-fold differential in survival between WT and c4 cells after 48 h of anoxia. When the cells were grown as solid tumors in vivo, a number of differences were observed, including slower growth and reduced expression of some of the HIF-1 targets (8)
. In situ hybridization showed intense focal induction of VEGF and GLUT-3 in tumors derived from WT cells, but this effect was not seen (GLUT-3) or reduced (VEGF) in tumors derived from c4 cells. Furthermore, immunohistochemical localization of GLUT-1 protein was detected in hypoxic areas in WT tumors, but GLUT-1 was absent in the c4 tumors (11)
.
The HIF-1 pathway is often assumed to be primarily a mechanism for promoting angiogenesis. However, as already mentioned, it also up-regulates glycolysis, and the relative significance of these two mechanisms in terms of tumor growth is of some interest, particularly because the HIF-1 pathway is a prime target in the ongoing search for agents that will inhibit tumor angiogenesis. However, the metabolic consequences and importance of the glycolytic pathway in intermediary metabolism in these tumors in vivo are unknown.
The aim of the present study was to use in vivo MRS and MRI methods, complemented by analysis of tumor extracts by high-resolution MR and classical biochemical methods, to measure physiological and biochemical parameters in wild-type and HIF-1ß-deficient mutant cells grown as solid tumors in vivo in mice, to see how a deficiency in HIF-1ß alters the metabolic phenotype. The measurement of tumor biochemistry in vivo by MRS could be an excellent method for monitoring changes caused by gene modification because changes in protein expression are likely to alter the concentration of cellular metabolites. MRS has the unique ability to measure the concentrations of metabolites in living tissue noninvasively. 31P MRS in vivo allows the assessment of the bioenergetic status of tumors through measurement of NTP relative to Pi as well as pH. 31P MR spectra also contain resonances from PMEs, which are mainly phospholipid precursors, and their glycerol derivatives, PDEs, which are phospholipid breakdown products. The 31P MR spectrum therefore provides information on synthesis and degradation of some components of cellular membranes. High-resolution 1H MRS of tumor extracts gives additional information on a wide range of acid-extractable metabolites. In some cases, we have also used classical enzymatic assays for particular metabolites. Multiple spin echo and multiple GRE MRI give detailed internal topography of the living tumor and also allow measurement of transverse relaxation rates, T2 and T2*, respectively (12) . T2 is influenced by the general tissue microstructure and water content, whereas T2* is additionally sensitive to blood vasculature because the presence of paramagnetic dHb in blood cells causes loss of signal from the adjacent tissue (13) . The tumors used were grown from the HEPA-1 mouse hepatoma lines (14) . The c4 mutant line lacks the aryl hydrocarbon hydroxylase, without which HIF cannot function to activate hypoxia-response elements, but it is otherwise identical to the WT parental line. For a full description, see Maxwell et al. (8) .
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
/6). The tumors reached the size required to give adequate signal-to-noise ratio for 31P MRS (0.5 g; range, 0.4–0.7 g) at 43 days for WT and at 63 days for c4 cells after cell inoculation. Mice were anesthetized with a single injection of a Hypnovel-Hypnorm mixture as described previously (15)
.
In Vivo MR Methods.
Mice were placed in the bore of a 4.7 T horizontal magnet fitted with a 10-G/cm, 12-cm bore high-performance auxiliary gradient insert, interfaced to a Varian Unity Inova spectrometer, so that tumors hung into two-turn 31P or 1H surface coils of 12-mm diameter. Body temperature was measured using a rectally inserted thermocouple before and after the experiment and maintained at 35–37°C with a water-heated pad placed over the mouse.
For 31P MRS, 1.2 g/kg 3-APP in PBS was injected i.p. 20 min prior to data acquisition (16)
. Localized 31P MR spectra were acquired using a modified version of the ISIS technique (17)
to minimize signal contamination from underlying tissues. There is negligible phosphocreatine in hepatomas, so the presence of a phosphocreatine resonance acts as a sensitive index for spectral contamination. Slice selection used adiabatic (sincos) inversion pulses, and spectra were acquired with a hard 90° pulse and repetition time (TR) = 3 s. Because extracellular pH measurements were being made, a double ISIS approach (18)
, which produces two spectra, one centered on 3-APP and one centered on -NTP, was used to minimize the chemical shift artifact introduced by the wide chemical shift range between 3-APP and -NTP. Spectral analysis for determination of peak integrals, and pH was performed using the variable projection (VARPRO) method (19)
, as described by Ojugo et al. (20)
.
1H MR images were acquired from a single 1-mm slice through the center of the tumor using (a) a spin echo sequence with TR = 300 ms and echo times (TEs) of 16–50 ms and (b) a GRE sequence with TR = 80 ms and TEs of 10–50 ms. Each set of images took 3 min to acquire, using 256 phase-encode steps over a 4-cm field-of-view with eight averages. T2 and T2* images were acquired while the mice breathed air and subsequently carbogen (95% O2-5% CO2) for 10 min. Parametric T2 and T2* maps were calculated on a pixel-by-pixel basis from the spin echo and GRE image data sets, respectively, and the average apparent relaxation rates T2 and T2* were calculated for a region of interest that encompassed the whole tumor (12) .
In Vitro Analyses of Tumor Extracts.
Separate cohorts of tumors were grown for these measurements. When the tumors had reached 0.5 g they were freeze-clamped, deproteinized with four volumes of 6% HClO4, and subsequently neutralized (21)
.
Metabolites Measured by Enzymatic Assays.
Glucose, lactate, and ATP were determined on the neutralized extracts according to the method of Bergmeyer (21)
.
Metabolites Measured by 1H MRS.
Neutralized extracts were freeze-dried and reconstituted in 1 ml of D2O, and 0.6 ml of the extracts were placed in 5-mm NMR tubes. 1H MR spectra of the tumor extracts were obtained using a Bruker 500 MHz spectrometer (pulse angle, 45°; repetition time, 3.5 s). The water resonance was suppressed by use of gated irradiation centered on the water frequency. Sodium 3-trimethylsilyl-2,2,3,3-tetradeuteropropionate and fumaric acid were added during the extraction procedure for chemical shift calibration and quantification, respectively. Immediately before the MRS analysis, the pH was readjusted to 7 with HClO4 or KOH.
Metabolic Imaging.
Metabolic imaging of ATP and glucose was assessed using single-photon imaging and bioluminescence as described by Mueller-Kleiser et al. (22)
. Four sections were taken from each of three WT and three c4 tumors. The sections were analyzed and combined to give a mean concentration for each tumor metabolite.
Histology.
After in vivo MR analyses, tumors were excised, fixed in 25% formal-saline, and stained with H&E for histological analysis. The tissue sections were subsequently assessed for gross necrosis and vascularization by an experienced histopathologist. Vessels were identified immunohistochemically, using a rabbit polyclonal antibody raised against mouse PECAM-1 (SC-8306; Santa Cruz Biotechnology, Santa Cruz, CA). The staining was visualized using an antirabbit Horseradish Peroxidase (HRP) Envision kit (K4010; Dako, Ely, United Kingdom) and diaminobenzidine, yielding a dark brown reaction product. Sections were counterstained with hematoxylin and mounted in aqueous medium (Aquamount BDH, Poole, United Kingdom). Vessels were quantitatively determined by the Chalkley point-counting method, producing a Chalkley vascular count. This involved assessment of the three most vascular areas (hot-spots) at x25 magnification (23)
.
Protein.
Protein was measured by the Bradford assay (24)
.
Statistical Analysis.
All data are presented as the mean ± SE, using a two-tailed t test with P < 0.05 for significance.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In Vivo MRS Analysis of NTP, pH, and Phospholipid Metabolites.
31P MRS in vivo showed that in both tumor types, the -, ß-, and 
-phosphates of NTP were clearly discernible in addition to the pH-sensitive Pi signal at 4–5 ppm (see Fig. 1
). Because the ß-NTP signal is unique to NTP (both - and -NTP peaks have contributions from other resonances), the ratio of the ß-NTP peak to that of Pi was used to express the bioenergetic state of the tumors. There were no significant differences in the ß-NTP/Pi ratio between the mutant c4 and WT tumors (Table 1)
. In 31P MR spectra, although there are contributions from other nucleoside triphosphate resonances, the majority (>80%) of the signal comes from ATP. It should also be noted that the signal-to-noise ratio was poorer and the peaks were slightly broader in the c4 tumor (see "Discussion").
|
|
and barely greater than the error of the method (26)
. The extracellular pH (pHe) was acidic in comparison with the pHi, as has been found in many tumor types (20)
, and the transmembrane pH gradient (
pH) of the two tumor types was not markedly different.
Signals from PDEs, catabolites of phospholipid membrane components, were present in all of the c4 tumors examined, but this resonance was scarcely detectable in the WT tumors (compare panels a and b in Fig. 1
). When PDE levels were expressed as a ratio compared with total Pi (
Pi) to normalize the data and allow comparison of the tumors as a group, PDE was 3-fold higher in the mutant compared with the WT (see Table 1
) tumors. PDE resonances were defined as the sum of the glycerophosphocholine and glycerophosphoethanolamine peaks, which are not resolved in the in vivo spectra. As can also be seen from Table 1
, there was no significant difference in the PME/Pi ratio between c4 and WT tumors. PME resonances, thought to be precursors of phospholipid membranes, were defined as the sum of phosphocholine plus phosphoethanolamine; again these peaks were not resolved in the in vivo spectra.
MRS in Vitro Analysis of Phospholipid Metabolites.
Representative 1H spectra of acid extracts of the c4 (Fig. 2a)
and WT (Fig. 2b)
tumors showed that free choline, phosphocholine, and glycine were lower in the c4 than the WT tumors, whereas taurine was similar in both tumor types. The betaine resonance was not very well resolved at pH 7, but when we analyzed the metabolites again at pH 9 (Fig. 2
, inset), we found that betaine was also lower in the c4 than in the WT tumors. A summary of the quantitative 1H spectroscopy data with statistical analysis is shown in Table 1
. This confirmed significantly lower concentrations of betaine (P < 0.05), phosphocholine (P < 0.05), free choline (P < 0.05), total choline (P < 0.05), and glycine (P < 0.05) in the c4 tumor extracts. The phosphocholine/glycerophosphocholine ratio, which most closely represents PME/PDE in vivo, was significantly higher in the WT tumors than in c4 tumors (P < 0.01). The acid extract values would not be expected to represent exactly the in vivo findings because the PME and PDE peaks in the in vivo spectra also had contributions from phosphoethanolamine and glycerophosphoethanolamine, respectively. They may also have included signals from mobile groups in membrane lipids that were not extracted into acid. However, the trend of high PDE relative to PME in the c4 tumor seen in vivo was confirmed in the 1H analyses of the extracts.
|
, along with microregional concentrations measured by the metabolic imaging method, which provided quantitative information on ATP and lactate (Table 1)
. The main finding by both methods was that the ATP content of the c4 tumors was 5 times lower than that of the WT tumors (P < 0.001 for enzymatic analysis; P < 0.05 for metabolic imaging). There was good qualitative concordance between the lactate measurements in the extracts (by 1H NMR and enzymatic analysis) and the data from metabolic imaging: all three methods showed no significant differences between the WT and c4 tumors. However, the absolute lactate concentrations measured enzymatically were substantially lower than those reported by the other two methods. We have no explanation for this finding, although it should be noted that the c4 tumors used in the metabolic imaging study were significantly smaller than the ones used for the MR and extract studies (see below for further discussion).
Tumor Vascularity.
The mean transverse relaxation rate, T2, which is calculated from spin echo images and reflects aspects of the tissue microstructure and water content, was significantly (P < 0.01) slower in the c4 tumors compared with WT tumors (Table 2)
. The relaxation rate T2* is related to the spin-spin MR relaxation time in the presence of magnetic field inhomogeneities. dHb, which is paramagnetic, creates large field inhomogeneities around blood vessels, shortening T2* (27)
. As seen in the WT and c4 tumor T2* maps (Fig. 3)
, the contrast was heterogeneous, reflecting the heterogeneous nature of tumor vascularity; similar images have been seen in several other tumor types (28)
. However, there was no significant difference in the average T2* between c4 and WT tumors. The hyperoxia induced by carbogen breathing may significantly increase T2* by decreasing the dHb concentration and thus cause an increase in signal intensity (28)
. T2*-weighted images of tumors can thus be used to monitor changes in the dHb concentration and give information on changes in blood oxygenation. However, when the host mice were given carbogen to breathe, neither the WT nor the c4 tumors showed any significant (P > 0.1) changes in T2* (Table 2)
.
|
|
0.5 g), analysis of H&E-stained tumor sections showed that both tumor types had abundant mitoses and extensive areas of necrosis, but consistent with the MR images, both tumor types showed no significant differences in gross vascularity or in the geometric pattern of vascularity. Chalkley vascular counts showed a mean of 9.05 (median, 8.33; range, 4–12.33) for the c4 tumor and a mean of 7 (median, 6.66; range, 5–11.66) for the WT tumors, showing that there was no significant difference in vascular density between the two tumor types (P = 0.23, unpaired t test; n = 5). In addition, there were no significant differences in the amount of protein per gram of wet weight of tumor between the c4 and WT tumors: 59.5 ± 6.1 mg protein/g wet weight (n = 5) and 57.0 ± 6.11 mg protein/g wet weight (n = 5), respectively. |
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Another possibly relevant effect of HIF-1 activation is up-regulation of glucose uptake and glycolysis, involving enhanced expression of the GLUT-1 and GLUT-3 glucose transporters and of many glycolytic enzymes. This effect is usually assumed to be another feature of the cell’s response to hypoxia because glycolysis, an anaerobic pathway, would be expected to be more active if the oxygen supply was low. However, it has been known for many years that cancer cells perform aerobic glycolysis, i.e., glycolysis in the presence of adequate oxygen (29) . This suggests a constitutive alteration of the tumor phenotype, perhaps for some other metabolic purpose. Furthermore, HIF-1 is also constitutively expressed as a result of mutations in genes encoding oncoproteins and tumor suppressors (30) and thus may up-regulate tumor glycolysis without the stimulation of hypoxia. Indeed, it was shown many years ago by Weber (31) that several of the enzymes of glycolysis (including hexokinase, phosphofructokinase, and pyruvate kinase) were manyfold more active in fast-growing hepatomas than they were in normal liver, predicting the more recent observations of overexpression of these glycolytic enzymes. We therefore hypothesized that the consequences of the failure of the HIF-1ß-deficient c4 cells to up-regulate the expression of glycolytic enzymes and glucose transporters would be altered levels of downstream metabolites, which could contribute to the observed differences in growth rate.
One surprising feature of the tumors from the mutant c4 cells was the absence of any detectable difference in vasculature. Although the c4 tumors grew significantly more slowly than the WT tumors, partly because of a longer lag phase, by the time they reached the chosen size (0.5 g), there were no obvious histological differences between them, either in vascularity or necrosis. In MRI studies, we also failed to see the expected difference in T2* between the mutant c4 and the control WT tumors (Table 2)
. A difference in T2* would have implied that HIF-1ß expression alters the vascular density (and thus the amount of dHb). The absence of such an effect confirms the histological evidence that deficiency in HIF-1ß causes no gross change in tumor vascularity. Maxwell et al. (8)
reported that determination of the microvessel density by immunoperoxidase labeling for the vascular endothelial marker CD31 during the initial growth phase of c4 tumors in vivo revealed focal differences in vascular development. One possible explanation for this could be the size of the tumors in the present study. To perform in vivo MR it was necessary to use c4 and WT tumors that had reached a larger (0.5 g) size than those used by Maxwell et al. (8)
, which were 0.05–0.2 g. In the larger tumors we studied, any differences in vascularity may have become negligible. Others have also found few vascular differences between these tumors when size matched, but have found an increased radiation responsiveness (32)
.
The loss of HIF-1 (33)
has also been reported to impair growth, but here too similar vascular densities were seen in WT and HIF-1-null tumors despite differences in VEGF secretion. The implication of that report, as well as the present study, is that HIF-1 is an important factor in solid tumor growth, which it affects in ways unrelated to its regulation of angiogenesis. It has recently been reported that pyruvate itself is angiogenic (34)
. We did not measure the pyruvate content of the tumors, and their lactate content was normal. All other things being equal, however, one would expect that the lack of up-regulation of glycolysis in c4 tumors would decrease their rate of pyruvate production, so that mechanism is unlikely to account for their unexpectedly normal angiogenesis.
Perhaps the most surprising metabolic difference found in the HIF-1ß-deficient tumors was their much lower ATP content: 20% of that in WT cells (see Table 2
). This ATP deficiency was demonstrated in vitro by two methods: classical enzymatic assays of tumor extracts and metabolic imaging studies of fresh-frozen histological sections. The low ATP concentration we observed could, in principle, be attributable to one of two effects. If the cells were in a low energy state (as occurs, e.g., in acute hypoxia), the ATP concentration would fall. In that case, the ATP lost from the c4 tumors would have formed an equivalent amount of ADP and Pi. Alternatively, there could have been an absolute reduction in the total adenine nucleotide pool. (The adenine nucleotide pool consists of ATP, ADP, and a much smaller concentration of AMP.)
We were able to eliminate the hypothesis that the c4 tumors are in a low energetic state on the basis of noninvasive 31P MRS studies in vivo. The 31P MRS method cannot distinguish between the nucleotide triphosphate signals (i.e., ATP, GTP, CTP, and UTP), so its results are reported in terms of NTPs, which in most tumors are mainly ATP and GTP (35)
. Furthermore, the relative proportions of ATP to ADP and GTP to GDP in the cytosol are maintained near chemical equilibrium by the enzyme nucleoside diphosphate kinase. For these reasons, tissue NTP is generally accepted as a surrogate for ATP when interpreting 31P MRS spectra. The results in Fig. 1
and Table 1
show that the NTP/Pi ratio, a measure of the energetic state of the tumor, was similar for both WT and c4 tumors. Note, however, that the overall signal-to-noise ratio was lower in the c4 spectrum (Fig. 1)
, which was consistent with a low absolute ATP concentration: less ATP, less signal. These normal in vivo 31P MRS spectra suggest that the HIF-1ß-deficient cells were able to phosphorylate ADP to ATP adequately. Their failure to up-regulate glucose transport or glycolysis did not have any major effect on energy metabolism and, thus, could not account for their low ATP content.
Other possible explanations for the low ATP concentration would be low cellularity (i.e., sparse tumor cells in an acellular matrix) or extensive necrosis in the HIF-1ß-deficient c4 tumors. However, no histological difference was apparent in the tumors we studied (note that low cellularity or necrosis would have to be very marked to account for the 5-fold lower ATP concentration), and the protein content per gram of wet weight was also similar in both tumors, so explanations based on low cellularity or necrosis must also be discarded.
The most plausible hypothesis would seem to be that low expression of the gene for HIF-1ß disrupts ATP synthesis, probably by preventing the synthesis of adequate amounts of a crucial intermediate. A likely candidate would be glycine, which was present at less than half the concentration in the HIF-1ß-deficient c4 cells compared with the WT control tumors (1.90 ± 0.38 µmol/g wet weight in c4 versus 3.93 ± 0.65 µmol/g wet weight in WT; P < 0.05). Glycine, formed (via serine, which we were unable to quantify in the 1H spectra) from 3-phosphoglycerate in the glycolytic pathway, is an essential precursor for purine synthesis. Thus, the inability of HIF-1ß-deficient tumors to up-regulate glycolysis could prevent newly formed cells from synthesizing sufficient purine moieties to produce the required quantity of ATP.
Glycine can also be formed from choline via betaine, and the main choline-related peaks in the 1H NMR spectra of tumor extracts were significantly smaller in the extracts from the c4 tumors, as was that of betaine itself (0.40 ± 0.06 µmol/g wet weight in c4 versus 1.10 ± 0.20 µmol/g wet weight in WT tumors; P < 0.05). The choline pool measured by 1H MRS included the small amount of free choline (3.21 ppm) and larger amounts of phosphocholine (3.22 ppm) and glycerophosphocholine (3.23 ppm). The phosphocholine peak, which includes a contribution from a poorly resolved doublet attributable to phosphoethanolamine, was significantly lower in the c4 tumor extracts (1.22 ± 0.24 µmol/g wet weight in c4 versus 3.78 ± 0.17 µmol/g wet weight in WT tumors; P < 0.05). Total choline, the sum of the choline peaks, was also lower in the c4 tumors (2.93 ± 0.44 µmol/g wet weight in c4 versus 6.79 ± 1.23 µmol/g wet weight in WT tumors; P < 0.05). These pathways are shown diagrammatically in Fig. 4
. Phosphocholine is a precursor of phosphatidylcholine, a membrane phospholipid. Thus, our observations would be consistent with a model in which the choline pool was bled off to create glycine and thus ATP, instead of forming phosphocholine and membrane phospholipids.
|
Another source of glycine for the tumor cells that we did not investigate would be uptake from the host. Presumably this source was not adequate to supply the needs of purine synthesis, perhaps because the supply was limited or the there was insufficient transport capacity across the cell membrane.
A deficiency of 50% in ATP content of cultured hypoxic HIF-1-deficient cells has recently been reported by Seagroves et al. (37)
, but only under hypoxic conditions. They attributed this to a defect in energy metabolism in their HIF-1-null cells, which they showed were unable to up-regulate glycolysis during hypoxia, arguing that because anaerobic glycolysis produces far less ATP than oxidative phosphorylation per molecule of glucose, increased glycolytic activity is required to maintain free ATP levels in hypoxic cancer cells. For reasons explained above, the normal energetics we observed in the c4 tumors argue against this interpretation for the present data. Indeed, the data reported by Seagroves et al. (37)
would also be consistent with the hypothesis we have advanced, in which the low ATP content of HIF-1ß-deficient tumors is attributable to a deficiency in glycine production by the glycolytic pathway. Their observation that ATP was depleted only in HIF-1-null cells cultured under hypoxic conditions could be explained if glycolytic intermediates formed by the non-up-regulated glycolytic pathway could be bled off for purine synthesis under oxic conditions, whereas under hypoxic conditions they were redirected into anaerobic energy metabolism and lactate formation. Seagroves et al. (37)
cultured their cells in DMEM, which contains 0.40 mmol/liter glycine (38)
, whereas the glycine concentration in normal rat arterial blood plasma is 0.216 mmol/liter (39)
. The marginally higher glycine content available to the cultured cells might explain why they are able to synthesize ATP under oxic but not hypoxic conditions.
It is unlikely that the ATP concentration of 20% that we observed in the c4 tumors could be caused by cellular hypoxia because they showed little necrosis and the hypoxic fraction measured using the bioreductive drug marker 2-nitroimidazole linked to theophylline was 7% in both c4 and WT tumors.5
Furthermore, the NTP/Pi ratio of 0.93 ± 0.17 (compared with 1.06 ± 0.07 in WT cells) suggests that the majority of the cells were not severely hypoxic. Lastly, because the tumor as a whole has an ATP concentration of 20%, one would have to hypothesize a very large hypoxic fraction (e.g., 50% of the cells having an ATP concentration of 10%). It is more likely, therefore, that the c4 tumors we studied had a constitutive change in ATP synthesis in all or a large majority of their cells.
Elevated aerobic glycolysis (i.e., glycolysis in normoxic conditions) and up-regulated expression of glycolytic enzymes are seen in many proliferating cell types, both normal and cancerous. Eigenbrodt et al. (40) , who reviewed much of the early work, predicted that purine synthesis via glycine would be one of a number of up-regulated anabolic pathways in proliferating cells that could be supplied by bleeding off the intermediates from this up-regulated glycolytic pathway. They also pointed out that all cells with a high level of aerobic glycolysis (i.e., both proliferating normal cells and cancer cells) have an unusual pyruvate kinase isoenzyme (type M2) rather than isoenzymes such as the L1 found in normal liver cells (40) . The M2 pyruvate kinase is easily inhibited because it has a lower affinity for phosphoenolpyruvate and a stronger inhibition by alanine; it also is inactivated by a cAMP-independent protein kinase. In more recent studies, Mazurek et al. (41) have reported that reduced flux through the lower part of the glycolytic pathway is attributable to dissociation of the M2 pyruvate kinase from a complex of glycolytic enzymes. All of these factors would tend to reduce formation of pyruvate or lactate as the end-products of glycolysis and channel glycolytic intermediates into anabolic pathways, including purine synthesis, rather than the usual aerobic or anaerobic ATP phosphorylation. Thus, cancer cells that express pyruvate kinase M2 have a mechanism for blocking the normal efflux of carbon from the glycolytic pathway and channeling it into anabolic synthesis. We did not measure the pyruvate kinase isoenzymes, but we have incorporated the hypothesis of Eigenbrodt et al. (40) into the following summary, assuming that the WT tumors express the M2 pyruvate kinase, whereas the c4 tumors do not. It should be noted that all tumor cells thus far studied express the M2 pyruvate kinase (41) , so if our assumption is correct, it implies that this expression is mediated through the HIF-1 pathway.
The mechanisms described in the preceding paragraphs are illustrated diagrammatically in Fig. 4
. Fig. 4a
shows the metabolic situation in WT tumors: HIF-1 overexpression has induced expression of glucose transporters and of the enzymes of the glycolytic pathway, including the inhibitable M2 pyruvate kinase isoenzyme. Because formation of pyruvate, and thus lactate, is reduced, a substantial proportion of the glycolytic intermediates are probably channeled into ATP synthesis; this pathway is shown with bold arrows. Phosphocholine is shown as a precursor of membrane phospholipids. Fig. 4b
shows the situation that we hypothesize in the HIF-1ß-deficient c4 tumors, with the concentrations of the measured metabolites shown as percentages relative to those in Fig. 4a
. Because expression of the glucose transporters and the enzymes of the glycolytic pathway are not up-regulated in the c4 tumors, there is a reduced supply of glycolytic intermediates for the synthesis of ATP. This is partially compensated by conversion of phosphocholine, choline, and betaine to glycine, and all of these intermediates are shown to be present at low concentrations. Membrane phospholipids are shown breaking down to PDEs. The glycolytic pathway, although not up-regulated, is not inhibited at the pyruvate kinase step, probably because these tumors do not express the M2 isoenzyme, so synthesis of pyruvate and lactate would not be impaired. Indeed, we found that lactate levels in extracts of the two tumor types were not significantly different. However, the interpretation of this observation is difficult because the concentration measured would have been a mixture of intra- and extracellular lactate, and the latter could include lactate taken up from the host as well as lactate produced by the tumor cells.
These studies have confirmed that HIF-1ß deficiency slows tumor growth and reduces focal expression of VEGF and glucose transporters, but has little effect on the vasculature. The most prominent abnormality was that the ATP content of the c4 tumors was very low. We hypothesis that because the tumor cells were unable to up-regulate glucose uptake or glycolysis, they were unable to synthesize the glycine necessary for de novo purine formation. Evidence in favor of this hypothesis came from the low concentration of glycine (an essential purine precursor, formed from an intermediate in the glycolytic pathway) in the tumors and the low concentrations of metabolites (phosphocholine, total choline, and betaine) in an alternative pathway for glycine synthesis.
These results suggest that a major role of up-regulated glycolysis in growing cancer cells is to provide anabolic precursor molecules for the synthesis of crucial metabolites, specifically ATP. Remarkably, however, the cells seem to be able to grow at a rate not much lower than normal, despite having only 20% of the normal ATP concentration. Furthermore, the ATP bioenergetic parameters, as judged from 31P MRS studies, were also the same as in the WT tumors. This suggests that the normal ATP concentration is much higher than is strictly necessary for cellular energy requirements: there is a substantial degree of redundancy in the system, which the mutant cells exploit.
The HIF-1 pathway is of much interest at present in the development of therapy of cancer and other diseases, and attention has focused on methods associated with hypoxia and angiogenesis. The present results emphasize that HIF-1 has pleiotropic actions and that suppressing it can affect a quite different anabolic pathway that is likely to be of direct relevance for therapeutic effects.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
1 Supported by the Cancer Research Campaign under Program Grant SP1971/0405 (to J. R. G., P. M. J. M., S. P. R., H. T., Y-L. C., and M. S.), the Medical Research Council (to I. J. S. and K. J. W.), and the Imperial Cancer Research Fund (to A. L. H. and R. L.). 
2 Present address: Novartis Pharma AG, BU Oncology, WKL-125.205, CH-4002 Basel, Switzerland.
3 To whom requests for reprints should be addressed, at CRC Biomedical Magnetic Resonance Research Group, Department of Biochemistry and Immunology, St. George’s Hospital Medical School, Cranmer Terrace, London SW17 ORE, United Kingdom. Phone: 02-08-72-55-85-2; Fax: 02-08-72-52-99-2; E-mail: mstubbs{at}sghms.ac.uk
4 The abbreviations used are: HIF-1, hypoxia-inducible factor-1; VEGF, vascular endothelial growth factor; WT, wild-type; MRS, magnetic resonance spectroscopy; MRI, magnetic resonance imaging; NTP, nucleoside triphosphate; PME, Kphosphomonoester; PDE, phosphodiester; GRE, gradient recalled echo; dHb, deoxyhemoglobin; 3-APP, 3-aminopropylphosphonate; ISIS, image-selected in vivo spectroscopy.
5 K. J. Williams and I. J. Stratford, unpublished data.
Received 9/10/01. Accepted 11/30/01.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
in common human cancers and their metastases. Cancer Res., 59: 5830-5835, 1999. is a positive factor in tumor growth. Cancer Res., 60: 4010-4015, 2000.This article has been cited by other articles:
|
|
 |
|
  C. Calabrese, A. Pisi, G. Di Febo, G. Liguori, G. Filippini, M. Cervellera, V. Righi, P. Lucchi, A. Mucci, L. Schenetti, et al. Biochemical Alterations from Normal Mucosa to Gastric Cancer by Ex vivo Magnetic Resonance Spectroscopy Cancer Epidemiol. Biomarkers Prev., June 1, 2008; 17(6): 1386 - 1395. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Denkert, J. Budczies, T. Kind, W. Weichert, P. Tablack, J. Sehouli, S. Niesporek, D. Konsgen, M. Dietel, and O. Fiehn Mass Spectrometry-Based Metabolic Profiling Reveals Different Metabolite Patterns in Invasive Ovarian Carcinomas and Ovarian Borderline Tumors. Cancer Res., November 15, 2006; 66(22): 10795 - 10804. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. G. Bundy, N. G. Iyer, M. S. Gentile, D.-E. Hu, M. Kettunen, A.-T. Maia, N. P. Thorne, J. D. Brenton, C. Caldas, and K. M. Brindle Metabolic Consequences of p300 Gene Deletion in Human Colon Cancer Cells. Cancer Res., August 1, 2006; 66(15): 7606 - 7614. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. M. Brown, R. L. Cowen, C. Debray, A. Eustace, J. T. Erler, F. C. D. Sheppard, C. A. Parker, I. J. Stratford, and K. J. Williams Reversing Hypoxic Cell Chemoresistance in Vitro Using Genetic and Small Molecule Approaches Targeting Hypoxia Inducible Factor-1 Mol. Pharmacol., February 1, 2006; 69(2): 411 - 418. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Vengellur, J. M. Phillips, J. B. Hogenesch, and J. J. LaPres Gene expression profiling of hypoxia signaling in human hepatocellular carcinoma cells Physiol Genomics, August 11, 2005; 22(3): 308 - 318. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. M. Arsham, D. R. Plas, C. B. Thompson, and M. C. Simon Akt and Hypoxia-Inducible Factor-1 Independently Enhance Tumor Growth and Angiogenesis Cancer Res., May 15, 2004; 64(10): 3500 - 3507. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Hopfl, O. Ogunshola, and M. Gassmann HIFs and tumors--causes and consequences Am J Physiol Regulatory Integrative Comp Physiol, April 1, 2004; 286(4): R608 - R623. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y.-L. Chung, H. Troy, U. Banerji, L. E. Jackson, M. I. Walton, M. Stubbs, J. R. Griffiths, I. R. Judson, M. O. Leach, P. Workman, et al. Magnetic Resonance Spectroscopic Pharmacodynamic Markers of the Heat Shock Protein 90 Inhibitor 17-Allylamino,17-Demethoxygeldanamycin (17AAG) in Human Colon Cancer Models J Natl Cancer Inst, November 5, 2003; 95(21): 1624 - 1633. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. S. Isaacs, Y.-J. Jung, E. G. Mimnaugh, A. Martinez, F. Cuttitta, and L. M. Neckers Hsp90 Regulates a von Hippel Lindau-independent Hypoxia-inducible Factor-1alpha -degradative Pathway J. Biol. Chem., August 9, 2002; 277(33): 29936 - 29944. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||
