Hypoxia, through the transcription factor HIF-1, is responsible for the upregulation of many enzymes involved in glycolysis. Here, we describe the identification of PDK-1 as a HIF-1 target protein in HNSCC using a cDNA microarray in SCC-25 cells and show that PDK-1 not only has an effect on pyruvate and lactate metabolism but also that it is associated with a significant poor prognosis in patients with high PDK-1 expression.
Both Q-PCR and western blotting confirmed that PDK-1 was highly upregulated in hypoxia and, furthermore, its expression was significantly reduced following treatment with RNAi against HIF-1α but not HIF-2α, thereby confirming PDK-1 as a HIF-1-dependent target, which is in agreement with the work by Kim et al (2006)
, and Papandreou et al (2006)
. Several cancer cell lines from common types of cancer were analysed and they showed that PDK-1 expression was increased in hypoxia at both the mRNA and protein level in the majority. These findings suggest that PDK-1 is not tissue- or tumour-specific, in spite of its selective expression in a few normal tissues.
To investigate the function of PDK-1, we used gene silencing using RNAi. Because of the problems with assaying PDH enzyme activity in crude extracts (Korotchkina et al, 2006
), we analysed the function and activity of PDK-1 using the MTS assay. Although this is not a specific assay for PDH, previous studies have shown the utility of analysing the metabolic effects of adding glucose and pyruvate (Segu et al, 1998
) and also shown that it is mainly the extra mitochondrial enzymes that are measured (Berridge and Tan, 1993
). The striking effects in intact cells show the extent of suppression in cancer cells even under basal conditions. Although this assay measures a pool of dehydrogenase activity it is the first demonstration of a change in the activity of the direct target of PDK-1 in vivo
, as opposed to indirect measures of free radical production or oxygen consumption, which are much smaller.
Although hypoxia reduced the growth of cells, there was no effect of PDK-1 activity on cell growth. These conditions represent chronic hypoxia that can occur in tumours. If this reached the extent of anoxia, we reasoned that on recovery there would be reoxygenation and that a burst of free radicals from mitochondria and PDK-1 may protect from that. Under anoxic conditions, there was no difference in the growth of cells treated with PDK-1 RNAi compared with control-treated cells. Under anoxic conditions, there is no mitochondrial respiration, so a difference in growth rate would not be expected. However, we did observe a small increase in growth rate after 48
h reoxygenation in cells treated with PDK-1 RNAi. Thus, PDK-1 suppression may allow cells to produce more ATP per molecule of glucose utilised, increased fatty acid synthesis and quicker balance of the cells, redox state and NAD cycle.
The excreted lactate and pyruvate were measured during cell growth in normoxia and hypoxia. This revealed an effect of PDK-1 suppression. After 16
h hypoxia, there was a significant increase in lactate and pyruvate concentrations, which were reduced in the cells treated with PDK-1 RNAi.
The level of pyruvate at 16
h was reduced to levels seen in normoxia by silencing PDK-1, indicating that PDK-1 isoform is the principal regulator of the PDH complex in these cells. Importantly, in this study, it was found that with PDK-1 RNAi treatment, after 48
h of exposure to hypoxia, the lactate could be reduced to the level seen in normoxia. This would suggest that the prolonged upregulation of PDK-1 in response to hypoxia and HIF-1α is a key factor in maintaining the elevated lactate and lactate to pyruvate ratio. The additional information in our study of the application of this work to clinical tumour samples showed a major prognostic difference in those tumours with PDK-1 expression. PDK-1 is highly expressed in cardiac, brain, lung, and kidney tissue, but clearly it was differentially expressed in malignant tissues. The expression pattern seen was predominantly cytoplasmic, which is similar to the findings that Koukourakis showed in non-small cell lung cancer (Giatromanolaki et al, 2001
; Koukourakis et al, 2005b
). Not previously reported was the finding that PDK-1 demonstrated nuclear expression in a subset of cancers. Although the function of this nuclear fraction is unknown, other glycolytic enzymes have also been reported to show nuclear expression. A high proportion of HNSCC tumours expressed high levels of both PDK-1 and PDH. This is in contrast to a previous study that reported PDH is decreased in epidermal tumours compared to normal epidermis (Eboli and Pasquini, 1994
The striking and adverse outcome of those tumours with highest PDK-1 expression could be related to a survival benefit on the cancer cells in vivo
, perhaps indicating that marginal hypoxic/anoxic cells are important for tumour growth. We have previously measured HIF-1α and HIF-2α, as well as carbonic anhydrase 9, the erythropoietin receptor and erythropoietin, and EGF receptor in this series of cases (Winter et al, 2005
). None of those markers was as strong a factor in predicting outcome as PDK-1. This may indicate the relative importance of one pathway induced by hypoxia vs
another one. We recently carried out a gene array analysis of a series of primary head and neck cancers and showed that the hypoxia gene profile differs in every case (Kong et al, 2006
). Therefore, it is possible that some pathways are biologically more important than others and hence predict outcome better. Another possibility is that it is a robust marker of HIF-1α signalling, not specifically related to its function, the former being shown by several groups to be associated with poor outcome in this and other cancer types. Also, the antigen may be better preserved than HIF and thus more reliable.
We propose a mechanism based on our observation that PDK-1 activity maintains lactate levels in the extracellular medium at about 2-fold higher, most likely by preventing pyruvate metabolism and its entry into the mitochondrial pathway. The high activity of LDHA and monocarboxylate transporters also increased in hypoxia through HIF-1, combined with the inability of cells to convert pyruvate to acetyl CoA by activating PDK-1 results in elevated lactate (Brahimi-Horn et al, 2007
). Lactate can enhance and maintain HIF activation through inhibition of prolyl hydroxylases (Lu et al, 2005
). This would have the effect of amplifying the Warburg effect and, indeed, a role of PDK-1 may be to contribute to the effect.
High levels of lactate have been associated with a poor outcome in a number of tumours, including HNSCC (Walenta et al, 1997
; Brizel et al, 2001
). However, additionally, LDHA has recently been shown to have a critical role in the energy production of cancer cells through glycolysis, and maintaining this pathway may be a more important aspect of inhibition of PDH (Fantin et al, 2006
). Recently, Cairns et al (2007)
have demonstrated that the mitochondrial metabolism of tumour cells is increased by the pharmacologic inhibition of PDK-1. The acute increase in oxygen consumption leads to a corresponding decrease in tumour oxygenation, thereby increasing the effectiveness of some traditional therapies. It will, therefore, be of interest to investigate the relevance of the PDK-1 pathway in in vivo
models to determine whether inhibitors will be worthwhile to develop clinically and also the relative importance of enhancement of lactate production vs
suppression of mitochondrial function.