TKTL1 expression in human malign and benign cell lines

The cell lines investigated herein were directly obtained from the companies given in Table 1 to warrant their correct identity. For all experiments described, each cell line was routinely cultured in a 1:1 mixture of DMEM/Ham's F-12 supplemented with 10% (v/v) fetal calf serum (FCS) and 10 ng/ml gentamycine (all reagents from PAA, Coelbe, Germany) at 37°C in the presence of 5% CO₂ and 21% oxygen. Cells were cultured in 75 ml culture flasks (Biochrom, Berlin, Germany) as monolayers and harvested at 80-90% confluence using a cell-scraper (Biochrom) for further experiments. HEK293 cell transfectants stably producing full-length TKTL1 protein (293pCAG TKTL1) were used as positive control cells. HEK293 cells transfected with empty expression vector (293pCAG Δ) do not produce TKTL1 protein and were used as negative control cells. Both transfectants have been previously described in detail [21]. For the present study the transfectants were named as follows: 293pCAG TKTL1 = HEK293-TKTL1 transfectants and 293pCAG Δ = HEK293-control transfectants.

Approximately 1x10⁶ cells per cell line were harvested upon trypsinization, washed twice with PBS and fixed for 1 h in 4% PBS-buffered formalin at room temperature. All centrifugation steps were performed at 280x g at 4°C for 10 min. The pellet was then re-suspended in 150 μl PBS and mixed with the same quantity of pre-cooled 2% high-melting agarose (Fermentas GmbH, St. Leon-Roth, Germany). The mixture was cooled down on ice. The resulting clot was first transferred to a sample-embedding capsule and then to a 4% PBS-buffered formalin solution. The dehydration with graded alcohols and xylene, and the embedding into paraffin (Histosec, Merck, Darmstadt, Germany) were done automatically in parallel to tissue biopsy samples at the Institute of Pathology, University of Würzburg, in a Leica ASP200 S embedding unit. The resulting paraffinized cell clot samples were then set into paraffin blocks. Paraffin blocks with cell samples were cut into 2 μm thick sections and mounted up on aminopropylethoxysilane (APES)-coated slides. Slides for immunohistochemistry were rehydrated in descending concentrations of ethanol before being heated for antigen unmasking in 10 mmol/l sodium citrate buffer (pH 6.0) in a microwave oven at 600 W for 5 min. After rinsing in distilled water, inhibition of endoperoxidase was performed by incubating sections for 10 min in 3% H₂O₂ in methanol. Slides were washed in PBS and incubated with 1% human immunoglobulin (Beriglobin, CSL Behring, Marburg, Germany) in PBS for 15 min to block FC-receptors [39]. Subsequently, slides were incubated with monoclonal mouse anti-TKTL1 antibody (clone JFC12T10, stock solution: 1 mg/ml; Linaris) or polyclonal anti-TKTL-1 antibody (Sigma prestige HPA000505, stock solution: 0.1 mg/ml; Sigma-Aldrich, Deisenhofen, Germany) diluted in antibody dilutent (DAKO, Hamburg, Germany). USB antibody used in Western blotting was not applicable for immunohistochemistry. Optimal antibody concentrations were determined in a series of dilutions with HEK293-TKTL1 transfectants and tumour tissue previously found to be TKTL1 positive [37, 40‐43]. A dilution of 1:200 from the stock solution was optimal for JFC12T10 and 1:20 for the SigmaPrestige antibody. After 60 min of incubation at room temperature in a humidified chamber, slides were washed in PBS and incubated with the horseradish-labelled LSAB2 secondary antibody mixture (DAKO) according to the manufacturer’s protocol. Staining was developed by adding 3,3’diaminobenzidine (DAB ready to use, DAKO) with subsequent counterstaining using haematoxylin. Afterwards, sections were dehydrated by washing in graded ethanol and then embedded in Vitro-Clud (Langenbrink, Germany). To obtain a maximum of homogeneous results all staining procedures were carried out in parallel in one session and repeated twice. Stained cells were photographed at 40x magnification with a KEYENCE Biozero Microscope (Keyence Corporation, Osaka, Japan) applying Z-stack technology.

For protein extraction, 1x10⁶ cells each were lysed in pre-cooled RIPA buffer (Pierce, Rockford, Ilinois) containing phosphatase inhibitors (phosphatase inhibitor cocktails set II, Calbiochem, Germany), proteinase inhibitors (Complete, Roche, Germany) and 2,5 mmol/l Dithiothreitol (DTT) reducing agent (Sigma-Aldrich). The mixture was incubated for 30 min on ice, combined with a thorough shaking every 10 min. Cell lysates were clarified of cell debris by centrifugation at 14,000x g for 5 min through a QIAshredder spin column assembly (Qiagen, Hamburg, Germany) at 4°C. Protein concentration was determined using the Bradford method [44] and coomassie brilliant blue (Roti-Quant; Roth, Karlsruhe, Germany). Afterwards, samples were mixed in 5x loading buffer (Fermentas), denatured at 95°C for 5 min, chilled on ice and stored at −20°C for further analysis. Equal amounts of proteins (20 μg) were loaded on a 10% polyacrylamide gel (SDS-PAGE) and electrophoresed. Proteins were then blotted onto a nitrocellulose membrane (Schleicher & Schuell, Dassel, Germany) for 45 min at 10 V using a semi-dry transfer unit (PeqLab, Erlangen, Germany). The protein transfer was confirmed by reversible membrane staining with Ponceau solution (Sigma-Aldrich). To avoid unspecific binding, the membrane was blocked with 5% non-fat milk (Merck, Darmstadt, Germany) in phosphate buffered saline (PBS)/Tween (0.05%) at room temperature for 1 h. Subsequently, the membrane was incubated with the respective primary anti-TKTL1 antibody at optimal dilution (see below) in PBS/Tween containing 2% non-fat milk at 4°C for 18 h. The following anti-TKTL1 antibodies were used: mouse monoclonal antibody clone JFC12T10 (Linaris GmbH, Wertheim, Germany, stock solution: 1 mg/ml, dilution 1:2,500), polyclonal rabbit Sigma Prestige antibody (HPA000505, Sigma-Aldrich, stock solution: 0.1 mg/ml, dilution 1:1,000), and mouse monoclonal antibody clone 1C10 (US Biological, Biomol GmbH, Hamburg, Germany, stock solution: 1 mg/ml, dilution 1:2,500). After washing with PBS, membranes were incubated with horseradish peroxidase-conjugated secondary antibodies goat-anti-mouse or goat-anti-rabbit, respectively (KPL Gaithersburg, USA, both diluted 1:10,000, stock solution: 0.5 mg/ml) for 60 min at room temperature. A monoclonal mouse anti-β-actin primary antibody, diluted 1:10,000, (Abcam, Cambridge, USA) was used as loading control. Immunoblots were visualized by home-made enhanced chemiluminescence (ECL) consisting of a mixture of 10 ml of solution A (50 mg Luminol (Sigma-Aldrich A4685) dissolved in 200 ml 0,1 mol/l Tris–HCl (pH 8,6)), 1 ml Solution B (11 mg para-hydroxycoumarinacid (Sigma-Aldrich, C9008) dissolved in 10 ml DMSO) and 3 μl of 30% H₂O₂[45] with subsequent exposure on an X-ray film (Fuji Super RX medical X-ray films; Fuji Photo Film, Düsseldorf, Germany) for 30 sec (shown in Figure 1) or for 3 min (USB) and 5 min (SigmaPrestige), respectively (shown in Additional file 1: Figure S1).

1x10⁶ cells each were disintegrated in Trizol reagent (Invitrogen Life Technologies; Darmstadt, Germany). Total RNA was extracted from Trizol as recommended by the manufacturer. RNA integrity was verified using the Experion automated electrophoresis station (Bio-Rad Laboratories Inc.; Munich, Germany) and the RNA concentration was measured at 260 nm. For first strand cDNA synthesis, 1 μg total RNA was employed using the iScriptcDNA synthesis kit from BioRad. The cDNA synthesis of 1:2 diluted cDNA was performed by heating at 25°C for 5 min, at 42°C for 30 min, and at 85°C for 5 min.

Quantitative PCR (qPCR) was performed with MESA Green qPCRMasterMix Kit for SYBR Green containing Meteor Taq hotstart polymerase (Eurogentic GmbH, Köln, Germany). Three sets of TKTL1 primer pairs were used: primer pair 1 (150 bp), forward: TAACACCATGACGCCTACTGC, reverse: CATCCTAACAAGCTTTCGCTG [7]primer pair 2 (80 bp), forward: AAGCCTTTGGGTGGAAC ACTTA, reverse: CTGAGAAGCCTGCCAGAATACC [28]; primer pair 3 (151 bp), forward: GGTATCTGTTGGTGACGATGGT, reverse: GCACATCCCCTTGGCATTGG [35]. TKTL1 primer pair 1 is located in the non-coding region, and primer pairs 2 and 3 are located in the coding region of TKTL1. The following primer pairs were used as reference genes: peptidyl propyl isomerase A (PPIA; access. No. NM_021130.3), forward: TGTCCATGGCAAATGCTGGACCC, reverse: GCGCTCCAT GGCCTCCAC AA (140 bp); β-actin (access. No. NM_001101), forward: CCTTGCCATC CTAAAAGCC, reverse: CACGAAAGCAATGCTATCAC (96 bp). qPCR reactions were performed on a CFX96 real-time PCR system (Bio-Rad) operated by CFX Manager Software (version 3.0). The cycler protocol was 5 min at 95°C, 40 cycles of 15 sec at 95°C, 60 sec at 60°C, and 5 min at 72°C. Gene of interest expression was normalized to the reference genes PPIA and β-actin [46]. Fold expression was calculated with the ∆∆Cq method [47]. Post-amplification melting curves were controlled to exclude primer-dimer artifacts and contaminations. Tktl1 mRNA expression was rated positive if the cycle number of the amplification reaction was obviously below the cycle number for HEK293-control transfectants with the three primer pairs. HEK293 cells were demonstrated to be negative for tktl1 mRNA by RT-qPCR and TKTL1 protein by Western blot [21].

Cells were seeded at 1x10⁵ cells/ml in 48-well flat bottom tissue plates (TPP; Biochrom) with culture medium and were grown for 24 h. Afterwards, medium was completely replaced, and glucose consumption and lactate production at 21% oxygen were analyzed 24 h later. Cell-free supernatant was snap frozen for analysis and cell numbers were calculated with the crystal violet assay [48]. Frozen supernatants were sent to the central laboratory of the University Hospital of Würzburg, thawed and immediately analyzed for glucose and lactate with the Cobas 8000 modular analyzer series (Roche Diagnostics; Mannheim, Germany). Glucose consumption was calculated as the difference between glucose concentration in cell-free control medium (16.4 ± 0.2 mol/l) (mean ± standard deviation) and glucose concentration remaining in the supernatant of cell cultures after an incubation time of 24 h. Lactate production was calculated as the difference between lactate concentration in the supernatant of cell cultures after an incubation time of 24 h and lactate concentration determined in cell-free control medium (0.29 ± 0.04 mol/l) (mean ± standard deviation). Both results were then correlated to the cell number and displayed as consumption/production per 10,000 cells.

Adherently growing cell lines were seeded at 5,000 cells per 100 μl of cell culture medium in a 96 well flat bottom tissue culture plate (NUNC Inc., VWR International, Ismaning, Germany) and grown for 24 h prior to treatment with chemotherapeutic drugs. Then, 100 μl cell culture medium containing paclitaxel or cisplatin (both from Sigma-Aldrich) were added, resulting in final drug concentrations ranging from 1% to 200% of the reported peak plasma concentrations achievable in patients [49]. The maximum plasma concentration was given as 15.9 nmol/ml (13.6 μg/ml) for paclitaxel and 12.7 nmol/ml (3.8 μg/ml) for cisplatin [50]. Cells were incubated for further 24 h and cell viability was measured with Cell Counting Kit-8 (CCK-8; Sigma-Aldrich). In brief, supernatant was carefully removed from the cells and a mixture of 100 μl fresh cell culture medium and 10 μl CCK-8 solution were added per well. Cells were incubated for 1 h at 37°C in a humidified incubator before analyzing the substrate reaction at 450 nm in a plate reader (Tecan GENios plus, Tecan Deutschland GmbH, Crailsheim, Germany). All analyses were performed in triplicates. Results were analyzed with the GraphPad Prism software calculating dose–response curves and IC₅₀ values.

Cell lines were seeded at 100 cells in 200 μl cell culture medium in a 96 well flat bottom tissue culture plate (TPP; Biochrom) 24 h prior to radiation. Graded X-irradiation (2, 4, 6 and 8 Gy) was then performed at room temperature using a 6 MV linear accelerator (Primus, Siemens Concord, CA, USA) at a dose rate of 3 Gy/min. Control cells were treated in a similar way, but without irradiation. Each radiation dose was applied in triplicates. 20 min after irradiation cells were cultivated at 37°C in a humidified incubator at 5% CO₂. Medium was completely replaced after 60 min to start with fresh growth conditions. In addition, cell medium was replaced on day 4 and day 8. Ten days after irradiation, the CCK-8 assay was performed as described above.

So far, TKTL1 protein expression was described primarily in paraffin-embedded tumor tissue. Hence, in order to produce comparable datasets, paraffin-embedded cell pellets of all selected cell lines (n = 20) served to analyze the expression of TKTL1 protein by immunohistochemistry. The cell lines assessed in the present study are summarized in Table 1, including tissue origin, supplier and references. In addition, TKTL1-negative human embryonic kidney cells HEK293 have been stably transfected with the expression plasmid pCAG or with a pCAG-based TKTL1 expression plasmid, which resulted in strong expression of the protein, as described elsewhere [21]. All cell lines have been analyzed by immunohistochemistry with the mouse monoclonal antibody clone JFC12T10 and with TKLT1-specific Sigma Prestige (rabbit polyclonal).

Immunohistochemical staining with JFC12T10 and Sigma Prestige antibodies showed a strong cytoplasmatic staining in HEK293-TKTL1 transfectants, with an additional nuclear staining by the Sigma Prestige polyclonal antibody. JFC12T10, but not the Sigma Prestige antibody also stained the cytoplasm of TKTL1-negative HEK293-control cells. The staining intensity of JFC12T10 in HEK293-control and HEK293-TKTL1 transfectants was comparable indicating an unspecific staining pattern of JFC12T10 in HEK293-control transfectants (Figure 1 and Table 2). Cytoplasmic staining of the 17 malign and the three benign cell lines with JFC12T10 ranged from weak (“+”) to very strong (“+++”) (Figure 1, Table 2). Only HT-29 cells remained unstained by JFC12T10. In contrast, cytoplasmic staining intensity with Sigma Prestige polyclonal antibody was faint (“[+]”) in eight cell lines (only PA1 cells were stained “+”), while no staining was observed in the cytoplasm of the remaining eight cell lines (Figure 1, Table 2). With this antibody, no or only weak background staining was observed in benign cells. Both antibodies stained the nuclei of HepG2 cells, and JFC12T10 additionally stained the nuclei of MCF-7, OVCAR, and U87 cells. While the Sigma Prestige polyclonal antibody stained the surface of PA1 and SiHa cells, a surface staining with JFC12T10 was seen in HUVEC and WS1. The most obvious mismatch of staining results from JFC12T10 and the Sigma Prestige antibody was found for HeLa, 23132/87, OVCAR, U251, WiDr and WS1 cells. These six cell lines demonstrated a strong staining by JFC12T10, but no or faint staining by Sigma Prestige antibody (Figure 1, Table 2).

TKTL1 expression was assessed in whole cell lysates of all 20 cell lines and the two control HEK293 cells by Western blot with the monoclonal anti-TKTL1 antibody clones JFC12T10 and 1C10 as well as with the polyclonal Sigma Prestige. All three antibodies detected TKTL1 protein in HEK293-TKTL1 transfectants (Figure 1) with an apparent molecular weight of 65 kDa, which is the expected molecular weight of TKTL1 protein. In contrast to Sigma Prestige and 1C10, JFC12T10 detected additional proteins with molecular weights ranging from 27 kDa to 70 kDa in HEK293-TKTL1 transfectants (Figure 1, Table 3). These additional proteins were also deteced in TKTL1-negative HEK293-control cells and have previously been reported to be insensitive towards transient reduction of TKTL1 expression using specific siRNA oligonucleotides [21]. The most prominent protein had a molecular weight of 27 kDa and was detected in all cell lines, except for WS1 fibroblast cells and primary fibroblasts. Protein bands with an apparent molecular weight of 37, 40, 45, and 55, 70, 100, and >120 kDa were observed in lysates of the majority of cell lines stained with JFC12T10. These protein bands are interpreted as background signals due to the very weak staining even after long exposure (see Additional file 1: Figure S1). Strikingly, Sigma Prestige and 1C10 antibodies did not detect any TKTL1 protein in 15 of 17 malign cell line studied. Monoclonal antibody clone 1C10, which was generated against full-length TKTL1 protein, produced very faint signals with higher and lower molecular weights than expected for TKTL1 in some of the cell lines tested (Figure 1, Table 3). The signal intensity was not increased by an extended exposure time of 3 min. After this very long exposure time, 1C10 detected a very faint signal at the expected molecular weight of TKTL1 in JAR cells. The polyclonal Sigma Prestige antibody, raised against the N-terminus of TKTL1, did not produce any signals except in the positive control HEK293-TKTL1 cells, even after an extended exposure time of 5 min (Additional file 1: Figure S1). To summarize, all three TKTL1-specific antibodies recognize exogenously expressed recombinant TKTL1 protein in HEK293-TKTL1 transfectants. Monoclonal antibody 1C10 did not show any specific signals in the three benign and in 15 out of 17 malign cell lines tested. Polyclonal Sigma Prestige antibody failed to detect TKTL1 in all cell lines tested, indicating a lack of endogenous TKTL1 protein. In contrast, monoclonal antibody JFC12T10 detected numerous additional protein bands with apparent molecular weights above or below the calculated size of TKTL1 (Figure 1 and Table 3). From our data, we cannot exclude the possibility, that the multiple protein signals detected by JFC12T10 antibody, in addition to the full length signal, represent putative smaller TKTL1 fragments derived from the C-terminus, which were not detected by Sigma Prestige and 1C10 antibodies raised against the N-terminus (Sigma Prestige) or the full-length protein (1C10) respectively.

Quantitative RT-PCR (RT-qPCR) was performed with three different primer pairs specific for the three tktl1 mRNA splice variants available in GenBank (PubMed) and published previously [7, 28, 35]. Results of RT-qPCR for all cell lines investigated are shown in Table 4, while Table 5 summarizes basic quantitative PCR data for all three primer pairs. For this, HEK293-TKTL1 transfectants, HEK293 control transfectants and JAR and U251 cells were analyzed. JAR and U251 were identified to weakly express endogenous tktl1 mRNA with all three primer pairs (Table 4). Primer pair 1 (located in the non-coding region of TKTL1 gene) did not recognize coding TKTL1 mRNA in HEK293-TKTL1 transfectants, whereas primer pairs 2 and 3 did. In comparison to JAR and U251 cells, the abundance of tktl1 mRNA in the other malign and benign cells was comparable to or below levels of the TKTL1-negative HEK293 control cell lines, and thus defined as negative (Table 4). The relative expression levels of tktl1 in JAR and U251 were increased up to 560-fold in comparison to the other cells (example for primer pair 3 and JAR and MDA-MB 231).

High expression of TKTL1 was put into context of enhanced glucose consumption within the pentose-phosphate pathway, resulting in increased lactic acid production [7]. In order to correlate both parameters with TKTL1 expression, glucose usage and lactic acid production were analyzed for confluent cell cultures after 24 h of culture at 21% oxygen. Data are summarized in Additional file 2: Table S1. Mean glucose consumption was 7.4 mmol/l per 10,000 cells (range: 1.9-21.4 mmol/l) after 24 h across all cell lines tested (n = 20) (Figure 2, light grey columns). The corresponding lactic acid production had a mean value of 1.03 mmol/l per 10,000 cells (range: 0.2-4.1 mmol/l) after 24 h (Figure 2, dark grey columns). All cell lines including benign cells showed lactic acid production 21% oxygen, thus demonstrating the “Warburg effect”. Glucose consumption and lactate production were correlated as expected. HCT116, HT-29, 23132/87, SiHa, and SKOV3 cells displayed the highest levels of glucose consumption and lactate production, whereas fibroblasts, HepG2, HUVEC, JEG, JAR, Mel2a, U251, and U87 displayed the lowest levels (Figure 2). JAR and U251 revealed to be the only cell lines within the panel analyzed in the present study to express TKTL1. Both cell lines did not demonstrate an outstanding glucose consumption and production of lactic acid.

Paclitaxel and cisplatin are widely used chemotherapeutic drugs for clinical treatment of solid cancers and are suitable for sensitivity tests. Therefore, all selected cell lines (n = 20) were subjected to treatment with paclitaxel and cisplatin in concentrations ranging from 0.2-31.8 nmol/ml and 0.1-25.4 nmol/ml, respectively. Representative results of dose–response curves after treatment with paclitaxel and cisplatin are shown (Figure 3A, 3B). From the resulting survival curves, IC₅₀ values were calculated (Figure 3C, 3D). HUVEC, HeLa and PA-1 cells were highly sensitive to paclitaxel and cisplatin (IC₅₀ < 5 nmol/ml), whereas U251 and MCF-7 were resistant to paclitaxel and Mel2A, and 23132/87 and HepG2 were resistant to cisplatin (IC₅₀ > 35 nmol/ml). JAR cells displayed intermediate and high sensitivity towards placlitaxel and cisplatin, respectively (Figure 3). Thus, the investigated cell lines exhibited a broad range of resistance and sensitivity to paclitaxel and cisplatin.Next, all selected cell lines were irradiated at doses ranging from 2 to 8 Gy and survival was determined. Representative dose–response curves for radiation are shown for JAR, U251, WiDr, OVCAR, PA1, HeLa, MCF7, and JEG cell lines in Figure 4A. Radiosensitivity was derived from the surviving fraction of cells at 2 Gy. This dose was effective at killing the majority of irradiated HUVEC (surviving fraction 2 ± 1%), and larger doses led to complete eradication of HUVEC (data not shown). All cell lines tested (n = 20) demonstrated a broad range of radiosensitivity. The stomach cancer cell line 23132/87 was highly resistant to radiation and grew to a surviving fraction of 163 ± 1.8% (Figure 4B). WS1, WiDr, HeLa, HCT116, HT29, and PA1 cells displayed strong resistance to radiation, with a survival fraction at 2 Gy of approximately 100% (Figure 4B). In contrast, HepG2, JAR, fibroblasts and HUVEC cells were characterized by radiosensitivity, with survival fractions below 50%. U251 cells revealed intermediate sensitivity towards radiation, with a survival percentage (2Gy) of 64 ± 16%. In summary, the tested cell lines demonstrated broadly varying degrees of radiosensitivity.

Performing RT-qPCR with three different primer pairs, we identified two of 17 malign human cell lines (12%) characterized by weak endogenous tktl1 mRNA expression (JAR and U251). Immunohistochemical proof of TKTL1 protein in cytoplasm, the expected location of TKTL1 as a protein which is supposed to be instrumental in the pentose-phosphate pathway, was successful in these two cell lines with JFC12T10. A very faint staining was seen for JAR cells while U251 cells were negative using Sigma Prestige polyclonal antibody, thus confirming negative Western blot results. All other cell lines demonstrated much lower tktl1 mRNA expression levels and no detection of TKTL1 protein in cell lysates by anti-TKTL1 antibodies 1C10 and Sigma Prestige, but multiple detections of proteins with apparent molecular weights above or below the calculated size of TKTL1 by anti-TKTL1 antibody JFC12T10. A relation of glucose consumption, lactic acid production and chemo- and radioresistance with expression levels of TKTL1 was not observed in any cell line tested in the present study. JAR and U251 cells did not show outstanding “aggressiveness” as measured by their resistance to both chemotherapeutic drugs and radiation or their ability to produce lactate (Figures 2, 3 and 4). While U251 was resistant to paclitaxel treatment and moderately sensitive to cisplatin, JAR cells were sensitive to both chemotherapeutic drugs (Figure 3). In comparison to most other cells tested, both cell lines were sensitive to radiation (Figure 4) and showed a moderate Warburg effect (Figure 2). Therefore, the proposed link between TKTL1 expression and chemo- and/or radioresistance as well as increased lactic acid formation by tumor cell lines was not confirmed by our results. In fact, TKTL1 expression was generally rare among the cell lines tested herein.