Transketolase proteins or transketolase enzyme activities have been related to neurodegenerative diseases, diabetes, and cancer. Transketolase enzyme variants and reduced transketolase enzyme activities are present in patients with the neurodegenerative disease Wernicke-Korsakoff syndrome. In Alzheimer’s disease patients transketolase protein variants with different isoelectric points or a proteolytic cleavage leading to small transketolase protein isoforms have been identified. In diabetes mellitus patients reduced transketolase enzyme activities have been detected and the lipid-soluble thiamine derivative benfotiamine activates transketolase enzyme reactions, thereby blocking three major pathways of hyperglycemic damage and preventing diabetic retinopathy. In cancer inhibition of transketolase enzyme reactions suppresses tumor growth and metastasis. All the observed phenomena have been interpreted solely on the basis of a single transketolase gene (TKT) encoding a single transketolase enzyme. No mutations have been identified so far in TKT transketolase explaining the altered transketolase proteins or transketolase enzyme activities found in neurodegenerative diseases, diabetes and cancer. We demonstrate the presence of a second transketolase enzyme (TKTL1) in humans. 

 The nonoxidative part of the PPP is controlled by transketolase enzyme reactions. The PPP, also known as the phosphogluconate pathway or the hexose monophosphate shunt, represents a multifunctional pathway. Three main activities have been attributed to this pathway, depending on the cell type and its metabolic state. The first function of the pentose pathway in mammalian cells is the generation of reducing power in the form of NADPH through a carbon flow from hexoses (mainly glucose) to pentoses which are recycled back into glycolysis through the nonoxidative pathway. The second function is the complete oxidative degradation of pentoses by converting them into hexoses, which can enter the glycolytic sequence. The third function of the PPP is to convert hexoses into pentoses, particularly ribose-5-phosphate, required in the synthesis of nucleic acids. Although the PPP represents a basic biochemical pathway, the proposed reactions of the nonoxidative PPP presented in textbooks is still controversial because the degree of 14C isotope labelling and its distribution in carbon atoms of fructose-6-phosphate differed from that predicted by reaction sequences4,5,6,7. The authors themselves detected this disquieting difference between theory and experimental results. Surprisingly, the notable difference between theory and experimental results received little adverse comment and generally uncritical approval from pentose pathway reviewers and textbook authors from that time. To explain the discrepancy between the proposed reaction sequences and the observed labeling results, a mathematical model has been developed8 , but no further experimental analysis of this elementary biochemical reaction sequences has been performed, which could explain the observed results. 

Therefore the current interpretation of results regarding transketolase enzyme reactions are either based on reaction sequences proposed 50 years ago, which are still controversial, or are based on a mathematical model trying to explain these discrepancies between theory and fact. Furthermore, all transketolase related results have been interpreted solely on the basis of a single transketolase gene (TKT), although a transketolase-like gene (TKTL1) has been identified9 . Up to now, one transketolase (TKT) and two transketolase-like genes (TKTL1 and TKTL2) have been identified in the human genome. To evaluate the role of the three transketolase genes in diseases, we analyzed in a first step the expression of all three members of the TKT gene family in cancer. Although the crucial role of transketolase enzyme reactions for tumorigenesis and metastasis is known, no molecular or immunohistochemical analysis of the three candidate genes has been performed previously. We could demonstrate that TKTL1 is specifically upregulated in malignancies at the mRNA level, whereas TKT and TKTL2 are not upregulated. We confirmed the upregulation of the TKTL1 protein in malignancies also at the protein level. We demonstrated the clinical importance of TKTL1 upregulation, since TKTL1 expression was correlated to invasive colonic and urothelial tumors and to poor patient outcome (Langbein et al., submitted). These findings suggest TKTL1 as the relevant target for novel anti-transketolase cancer therapies. There is strong evidence that transketolase proteins or transketolase enzyme reactions are also important for diabetes as well as for neurodegenerative diseases. A high blood glucose level leads to severe chronic complications in a subgroup of diabetes patients.

 Due to a high blood glucose level and a concomitant high glucose concentration in retinal, endothelial and neuronal cells, glucose and other reducing sugars react nonenzymatically with protein amino groups to initiate a post-translational modification process known as nonenzymatic glycation10,11. In diabetes patients such an advanced glycation gives rise to advanced glycation end products (AGE), thereby leading to macro- or microvascular complications, neuropathy and retinopathy. Glucose metabolism leading to AGE formation and retinopathy can be altered by application of vitamin B1 (thiamine), a cofactor of transketolase enzymes. The lipid-soluble thiamine derivative benfotiamine activates transketolase enzyme reactions and blocks three major pathways of hyperglycemic damage and prevents diabetic retinopathy12. Therefore activation of transketolase enzyme reactions represents a major progress in therapy and prevention of chronic diabetes complications. Glucose metabolism and the formation of AGE have also been suggested as a model of aging and was also linked to neurodegenerative diseases like AD. A glycation of proteins in senile plaques of AD patients has been detected. Many senile plaques contained glucoseAGE, indicating that Aβ is glycated by glucose13. Furthermore, a well-defined risk factor for the onset of AD is possession of one or more alleles of the epsilon-4 variant (E4) of the apolipoprotein E (ApoE) gene. Metaanalysis of allele frequencies has found that E4 is rare in populations with long historical exposure to agriculture, suggesting that consumption of a high carbohydrate diet may have selected against E4 carriers14. 

Furthermore, the role of the ApoE4 allele for a reduced cerebral glucose metabolism in aging as well as for a more severe AD phenotype has been shown by PET3,15,16. Alterations of enzymes or enzyme activities involved in glucose metabolism have already been identified in neurodegenerative diseases. Transketolase protein variants and reduced transketolase enzyme activities have been detected in Wernicke-Korsakoff syndrome patients17,18 and in AD patients transketolase protein variants with different isoelectric points or a proteolytic cleavage leading to small transketolase protein isoforms have been identified19,20. Until now, no mutations in Wernicke-Korsakoff syndrome patients have been identified in the TKT transketolase gene21. Here, we characterize the transketolase-like gene TKTL1. Due to a mutation in a putative exon, a stop codon has been detected in the predicted open reading frame. Therefore the TKTL1 gene has been assumed to be a pseudogene; however, we have previously shown that TKTL1 in fact could encode a transketolase-like protein9 . Using a novel monoclonal antibody specifically detecting the TKTL1 protein on paraffin sections, and in ELISA and Western blot format, we were able to identify and isolate native TKTL1 protein(s). In contrast to known transketolase genes/proteins, the TKTL1 gene encodes different tissue-specific transcripts and a full length as well as smaller protein isoforms. 

In contrast to this, the sequence identity between TKTL1 and TKTL2 is 80% on the DNA level and 77% on the protein level. Orthologues of TKT, TKTL1 and TKTL2 were also identified in mouse and rat. In frogs (Xenopus laevis) and fish (Danio rerio) only a single transketolase gene with no exon 3 deletion has been identified. In mouse and rat three transketolase(-like) genes are present and similar to the human TKTL1 transcript, the TKTL1 orthologous transcript also harbors the exon 3 deletion. The genomic structure of the three transketolase genes and the presence of the exon 3 deletion in TKTL1 in humans, mouse and rat indicate that during evolution of the vertebrate genome, this deletion occurred within the TKTL1 gene before the human and murine lineages diverged. The high similarity between TKTL2 and TKTL1 suggests that prior to the deletion of exon 3, an intact copy of TKTL1 was duplicated and integrated into the genome by a reverse transcriptase-mediated event, giving rise to TKTL2. Tissue-specific mRNA expression of TKTL1 To examine the expression level of different TKTL1 transcripts, Northern blot hybridization was performed using multiple human adult tissues (Figure 1A). To avoid the possibility that the detected transcripts are due to cross-hybridization with TKT or TKTL2 gene transcripts, we used a probe from the 3’ untranslated region, with only limited sequence homologies to TKT and TKTL2. Four distinct species were identified: one of 2.5 kb was detected in all tissues examined with the exception of adult heart. A larger transcript of app. 2.7 kb was weakly expressed in lung and pancreas. In skeletal muscle an additional weakly expressed transcript of 1.9 kb was detectable. 

Finally, the smallest transcript of 1.4 kb was the main transcript in heart. This 1.4 kb transcript is also present in skeletal muscle and kidney. By identifying a cDNA clone from heart tissue containing an additional exon with no homology to transketolase sequences, we could isolate part of these alternate transcripts9 . Real-time quantification of expression of transketolase family members in healthy tissues The relative expression level of each member of the transketolase gene family was determined by real-time PCR in the following normal tissues: brain, heart, liver, peripheral blood mononuclear cells, lung, breast, ovary, kidney, testis, spleen, stomach, colon, uterus, esophagus, skin, thymus, bladder, muscle, prostate, and retina. The expression level of the TKT gene in healthy tissues was high compared to the expression of TKTL1 and TKTL2 (not shown). On average the TKT expression level was 60 to 1000-times higher than the TKTL1 or TKTL2 level. The highest level of TKT expression was observed in normal colon, which is 5-10 fold higher compared with the expression level in most other tissues. Under the examination in all other normal tissues, the TKT expression level was very similar with maximal variation of 15-fold differences. In contrast, the expression levels of TKTL1 and TKTL2 were much more tissue-specific and had a greater variation. In testis we observed >12.000-fold higher expression of TKTL1 as compared to lung, ovary, or skin. In testis the TKTL1 expression was even higher than the TKT expression level (twice that of TKT). Strikingly, those healthy tissues (testis, thymus and retina) in which a high aerobic glycolysis has previously been identified22 were precisely the same tissues in which we identified a high level of TKTL1 expression. 

The monoclonal TKTL1 antibody JFC12T10 detects protein isoforms different in size on Western blot To determine the TKTL1 expression at the protein level, we raised antibodies against a C-terminal fragment of the recombinant TKTL1 protein in mice. One out of 32 hybridoma cell lines was selected, producing MAb JFC12T10 specifically detecting the recombinant TKTL1 protein in ELISA experiments or on Western blots, and not cross-reacting with recombinant TKT or TKTL2 protein. Using MAb JFC12T10, protein expression of the TKTL1 gene was examined in five different human cancer cell lines representing four different tumor entities. In the lung carcinoma cell line A-549, the breast carcinoma cell line MCF7, the liver carcinoma cell line HepG2, and the colon carcinoma cell lines HCT116 and HT29, different protein isoforms were detected (Figure 1B). The migration of the largest TKTL1 protein isoform in SDS-polyacrylamide gel electrophoresis indicated a molecular size of 75 kDa, which is significantly larger than the size of the open reading frame of the predicted ORF of a TKTL1-cDNA (acc. no. BC25382) encoding a 65.4 kDa protein. This discrepancy was also observed when the His-tagged recombinant 65.4 kDa protein was expressed in E. coli, affinity purified and separated by SDS-PAGE (Figure 1C). The migration of this N-terminal His-tagged 66 kDa TKTL1- full length protein indicated a size of 75 kDa (largest band in Figure 1C), which was the same as the observed size for the native TKTL1 full length protein (Figure 1B). The smaller bands of the recombinant TKTL1 full length protein are likely due to a C-terminal proteolytic cleavage in E. coli prior to the affinity purification by the N-terminal His-tag. 

As expected from the presence of smaller transcripts in human tissues, smaller protein isoforms were detected in every tested tumor cell line. A weakly expressed protein isoform of app. 58 kDa is present in HCT116 and MCF7 and A549. Even smaller protein isoforms of 40 kDa, 44 kDa and 48 kDa were present in cell line MCF7. Some of these smaller protein isoforms were detectable in the other tested cell lines. TKTL1 is a transketolase To determine whether the TKTL1 protein has transketolase activity, recombinant TKTL1 protein expressed in E. coli as well as the native TKTL1 protein isolated by affinity purification from a tumor cell line were used. The native as well as the recombinant TKTL1 protein performed a transketolase enzyme reaction using the two substrates xylulose-5-phosphate (X5P) and ribose5-phosphate (R5P) (Figure 2). As a first step in characterizing the enzymatic properties of TKTL1, the substrate utilization of TKTL1 was tested. Mutation of His103 in yeast transketolase facilitates a one-substrate reaction, thereby demonstrating that this particular amino acid residue determines substrate specificity and reaction modus in yeast transketolase23. Since the mutation in TKTL1 led to a deletion of 38 amino acid residues including a His residue homologous to yeast His103, we determined whether the TKTL1 transketolase is able to perform a one-substrate reaction with X5P as sole substrate. A one-substrate enzymatic activity of 42% in comparison to the two-substrate reaction was observed for the native TKTL1 protein (Figure 2A). A similar one-substrate enzymatic activity (47%) was also observed for the recombinant TKTL1 (Figure 2B).