Creatine kinase (CK) isoenzymes catalyse the reversible transfer of the phosphate group of phosphocreatine (PCr) to ADP, to yield ATP and creatine (Cr). The CK/PCr/Cr system is present primarily in tissues with high and fluctuating energy demands such as brain, heart and skeletal muscle, and serves as a temporal and spatial "energy buffer" that helps to maintain a high intracellular phosphorylation potential in situations of increased metabolic demand (for reviews, see 12).

In mammals, Cr can be taken up by the intestine from the food, or can be synthesized de novo. The liver is the main site of Cr production in the body (see 2). After its synthesis, Cr is transported through the blood and is taken up by Cr-containing tissues via a specific Cr transporter. Whereas the importance of the liver in Cr biosynthesis is undisputed, some confusion still exists on the CK activity and PCr content in this organ. The majority of findings suggest no or minute levels of CK and PCr in liver tissue and, in particular, in hepatocytes (e.g., 3456). Other studies that used more sensitive experimental approaches provided evidence for low levels of PCr and CK, specifically localized in sinusoidal endothelial cells (789; see also 10). Finally, in a few cases, more extreme findings were made: unusually high levels of CK activity were measured in liver tissue by Shatton et al. 11, Goullé et al. 12, and Wali & Makinde 13.

The majority of studies indicated that the low levels of CK activity in liver are due solely to the brain-type cytosolic CK (BB-CK) isoenzyme. On the other hand, besides BB-CK which was suggested to be present in endothelial and Kupffer cells, Vaubourdolle et al. 14 also provided evidence for the presence of the muscle-type cytosolic (MM-CK) isoenzyme in Ito cells, and for mitochondrial CK (Mi-CK) in hepatocytes. Similarly, Kanemitsu et al. 15 purified Mi-CK from normal human liver, which would imply significant amounts of this isoenzyme in liver tissue. Finally, increases in serum CK activity were frequently observed in cases of severe liver disease, with the most obvious source of CK being the pathological liver tissue itself 161718.

In a number of studies reporting significant levels of CK activity in liver, interference by adenylate kinase (AdK) isoenzymes in the CK activity assays 192021 is very likely (e.g., 13), or can at least not be excluded, thus questioning the validity of these studies. Another possible cause for inconsistent findings might be compensatory up-regulation of CK expression in pathological liver tissue. Two lines of evidence that favour this hypothesis are: (i) partially hepatectomized rat liver was reported to show an increase in BB-CK activity (see 22); and (ii) overexpression of CK isoenzymes in the liver of transgenic mice was shown to stabilize energy metabolism under low-oxygen stress and after a metabolic challenge 2324, to accelerate regeneration of liver mass following major hepatectomy 1025, and to increase endotoxin tolerance 526.

Because of these conflicting data, the goal of the present study was to analyse in detail the CK and AdK activities in pathological liver tissue of patients undergoing orthotopic liver transplantation.

CK and AdK activities were measured in total homogenate (Fig. 1), cytosolic and mitochondrial fractions (data not shown) obtained by differential centrifugation from 25 normal and pathological liver samples. Highest CK activities were observed in the two primary HCCs analysed (liver samples no. 4 and 9), with specific CK activities in the homogenate of 0.36 and 0.21 U·(mg protein)^-1, respectively. In most other liver samples, the specific CK activities in the homogenate were below 0.05 U·(mg protein)^-1. Whereas enzymatic activity measurements revealed low, but consistent CK activity in many of the cytosolic fractions, no CK activity was detected in the mitochondrial fractions, except for HCC sample no. 9 with a specific CK activity of approx. 0.1 U·(mg protein)^-1 (due to limited sample size, subcellular fractionation was not feasible for HCC sample no. 4). These findings were corroborated qualitatively by isoenzyme electrophoresis on cellulose polyacetate strips. Visualization of the different CK isoenzymes by an overlay gel technique revealed that the brain-type cytosolic BB-CK isoenzyme was present in all liver samples. Conversely, bands for the dimeric and octameric forms of Mi-CK were only observed in the two primary HCC samples (nos. 4 and 9; Fig. 2).

The CK/AdK activity ratio in the homogenate was 1.4 and 2.6 for the two primary HCCs (liver samples no. 4 and 9, respectively), 0.5 for liver sample no. 5 (secondary biliary cirrhosis), and < 0.2 for all other liver samples. Similar findings were made for the cytosolic and mitochondrial fractions, with CK/AdK activity ratios of < 0.3 for the cytosolic fraction and < 0.05 for the mitochondrial fraction. For HCC sample no. 9, however, these ratios were significantly higher: 4.8 (cytosolic fraction) and 0.33 (mitochondrial fraction).

CK is an enzyme still widely analysed in clinical diagnostics. Although a wealth of CK measurements have been reported in the scientific literature, there still exist inconsistency and incomplete knowledge on such an apparently simple question as the CK (isoenzyme) content of mammalian liver in both health and disease. In the present study, we detected the presence of BB-CK in all liver samples analysed by using CK activity measurements and cellulose polyacetate electrophoresis. However, in the normal and most pathological liver samples that we analysed, the specific CK activity was very low (< 0.05 U·[mg protein]^-1), levels which are comparable with or lower than data reported for rat and human liver 2930, but much lower than the specific CK activities in skeletal muscle, heart and brain (2–37 U·[mg protein]^-1; 29313233). We additionally observed that (i) the specific AdK activities in these samples were consistently higher than the specific CK activities (on average, > 10-fold), (ii) both activity measurements and cellulose polyacetate electrophoresis revealed similar specific AdK activities in the cytosolic and mitochondrial fractions (although from different AdK isoenzymes; data not shown), and (iii) mitochondrial respiration, in the presence of ATP, could be fully stimulated by AMP, but not by creatine (data not shown). This last observation favours the interpretation that in normal hepatocytes, CK isoenzymes are not expressed, and that the AdK isoenzyme system plays a function in high-energy phosphate buffering and transport, which is similar to the role of CK in brain, skeletal muscle and heart. Although histochemical data are missing, the results obtained here are most consistent with a localization of small amounts of BB-CK in sinusoidal endothelial cells 14.

Interestingly, we observed a strong induction of both BB-CK and Mi-CK expression in two samples of primary HCC. Despite CK/AdK activity ratios in vitro of 1.4–2.6, the specific CK activities were still relatively low (0.21–0.36 U·[mg protein]^-1). Therefore, in the absence of histochemical data, it cannot be concluded with certainty whether the increased levels of CK are due to increased vascularization of the tumour (possibly associated with a higher proportion of CK-containing endothelial cells), or to induction of CK expression in the malignant cells.

Induction of CK expression has been observed previously in many types of tumours (see 2) and may reflect an adaptation of the tumour tissue to the increased energetic demands. Evidence for induction of CK in liver tumours mostly comes from hepatoma cells grown in tissue culture 3435 or, indirectly, from increased amounts of circulating BB-CK and Mi-CK in the blood of patients with liver tumours 36. On the other hand, analysis of the tumour tissue itself, both by classical biochemical methods and by microarray technology, provided inconsistent results. Some authors reported induction of CK expression in liver tumours (373839; and, in part, 40; M. Sakamoto and S. H. Yim, personal communication), others repression 41, and still others observed no statistically significant differences between normal and malignant liver tissue 1130. This may be a reflection of the diverse clinicopathological and biological phenotypes of HCC, with different underlying molecular defects.

A key player in the picture might be the p53 tumour suppressor gene. Mutations in p53 are quite prevalent in HCC, especially in tumours with low cellular differentiation 4243. On the other hand, p53 was shown to control BB-CK expression: transrepression as observed for wild-type p53 is prevented by different mutations in the p53 gene 44. Therefore, it is tempting to speculate that induction of BB-CK in HCCs is caused directly or indirectly by mutations in p53.

Expression of CK in HCC may have therapeutic implications, which is all the more important given (i) the limited responsiveness of HCC to currently available therapeutic approaches and, thus, (ii) the poor prognosis associated with this disease. Cr analogues (cyclocreatine and β-guanidinopropionic acid) and also Cr itself were previously shown to have antitumour activity, both in cell culture and in in vivo models (4546; see also 2). The responsiveness of tumour cells to growth inhibition by cyclocreatine seems to be correlated with their specific CK activity; cell lines with a specific CK activity of > 0.10 U·(mg protein)^-1 were generally sensitive to the drug. As for the liver, β-guanidinopropionic acid and creatine slowed the growth of AS30-D ascites tumour cells in culture (chemically induced rat hepatoma; 34). Similarly, cyclocreatine revealed antitumour effects in a rat model of chemically induced hepatocarcinogenesis 47.

The present findings shed light on some old enigmas and open up fascinating avenues for future research. Our findings do not support significant expression of CK in normal liver and most liver pathologies, but rather indicate that many of the previous misconceptions in this field can be explained by interference from AdK isoenzymes. On the other hand, given the need for improved understanding of the molecular pathogenesis of HCC, and for improved therapies and cures, the induction of CK expression in HCC described here calls for a more in-depth analysis of the interplay between p53 mutations, HCC, CK expression, and the growth-inhibitory effects of cyclocreatine in HCC.

AdK, adenylate kinase; BB-CK, brain-type cytosolic CK isoenzyme; CK, creatine kinase; Cr, creatine; HCC, hepatocellular carcinoma; Mi-CK, mitochondrial CK; MM-CK, muscle-type cytosolic CK isoenzyme; PCr, phosphocreatine.

The authors declare that they have no competing interests.

GM and RM covered the medical part of this study. GM, FNG and MW performed the biochemical experiments. MW drafted the manuscript.