As previous studies do not mention the use of a specific carrier for the gavage of mice with benfotiamine 1823, we first suspended benfotiamine in water and administered it by force-feeding (100 μl containing a dose equal to 100 mg of benfotiamine per kg), using a syringe with a tube that was inserted into the stomach. We sacrificed the animals after 1 hour and we determined thiamine derivatives in the blood. Thiamine levels proved to be highly variable: in some animals thiamine levels were up to 10 times higher than in the controls, while in others they were hardly increased. Benfotiamine is very sparingly soluble in water (at least at pH < 8.0) and it is possible that in some cases, either most of the material remained attached to the syringe or the force-feeding tube or it was eliminated with the bolus. Therefore, we tested different hydrophobic solvents such as oil (Mineral oil, Sigma M5904) and organic solvents (ethanol, octanol, dimethyl sulfoxide), but no significant solubility was observed. Finally, we obtained a homogenous solution of benfotiamine by forming inclusion complexes in a 200 mM hydroxypropyl-β-cyclodextrin (HP-β-CD) solution that was used for force-feeding. This time, the inter-animal variability was reasonable (25%, S.E.M/mean × 100, n = 12) and we found that, one hour after the administration, thiamine, ThMP and ThDP levels in the blood were increased by a factor of respectively 822, 14 and 4.7 (Fig. 2). Highly significant increases of free thiamine and ThMP were also observed in the liver, but not in the brain.

First, we measured tissue levels of thiamine, ThMP and ThDP for up to 6 hours after a single oral administration of benfotiamine (100 mg/kg in a HP-β-CD solution). In the blood, thiamine levels rapidly increased, reaching a maximum after 120 min and slowly decreasing thereafter (Fig. 3). ThDP levels steadily increased before reaching a maximum after 4 hours. For ThMP, the highest level was already reached after 30 min and it slowly decreased during the next hours. In the liver, all derivatives peaked after 1 hour before decreasing to near normal levels after 2 hours. No significant variations were observed in the brain. Total thiamine content (per mg of protein) was about twice higher in the liver than in the brain before benfotiamine administration. One hour after the administration, the ratio was over 10. However, liver thiamine decreased rather fast after 1 – 2 hours. In most samples, we also observed the presence of ThTP and AThTP, but the levels of these compounds in the liver and the brain were not influenced by the treatment (not shown).

As no increase in brain thiamine levels was observed after a single administration of benfotiamine, we decided to test a longer treatment. The animals received a daily oral administration of either 100 μl of a solution of benfotiamine (100 mg/kg) in 200 mM HP-β-CD or 100 μl of a 200 mM HP-β-CD solution for 14 days and were sacrificed 24 hours later. After treatment with benfotiamine, the levels of free thiamine, ThMP and ThDP were increased in the liver but not in the brain (Fig. 4). Again, the small amounts of ThTP and AThTP found did not depend on the conditions of treatment.

As treatment with benfotiamine did not yield elevated brain thiamine levels, we wanted to test whether this molecule is able to easily cross cell membranes in cultured cells. We chose mouse neuroblastoma cells (Neuro 2a) as we have previously characterized these cells with respect to thiamine transport 828, metabolism 29 and deficiency 30. The thiamine concentration in the commercial Dulbecco's modified Eagle's medium (DMEM) is 10 μM 8. Therefore, once the cells were nearly confluent, DMEM was replaced by a saline containing 10 μM of either thiamine or benfotiamine. After incubation for up to 4 h at 37°C, the thiamine content of the cells was determined. We did not observe any increase in intracellular thiamine after incubation with benfotiamine and there were no significant differences between the thiamine and benfotiamine groups (Fig. 5). This is in sharp contrast with previous experiments where incubation of the same cell line with 10 μM sulbutiamine under the same conditions led to a 10-fold increase in intracellular free (unphosphorylated) thiamine within 2 hours 29.

As mentioned above, the thiamine concentration in most commercial culture media is about 10 μM, one or two orders of magnitude higher than the K_m for the high affinity thiamine transport present in most cells 82829. Cells grown under those conditions are thus saturated with thiamine. Therefore, we wanted to test whether benfotiamine had any effect on intracellular thiamine levels in Neuro 2a cells previously thiamine-depleted. The cells were grown for 8 days in a medium containing about 7 nM thiamine. Then either benfotiamine (1 μM) or thiamine (1 μM) was added (Fig. 6). After 2 hours in the presence of thiamine, intracellular thiamine already reached a maximum, while with benfotiamine an important lag period was observed. This experiment confirms that thiamine enters Neuro 2a cells more rapidly than benfotiamine. The lag period can be explained if benfotiamine is first dephosphorylated by ecto-phosphatases to the lipidsoluble S-benzoylthiamine that can then enter the cells (see below).

In 1961, Wada et al. reported the physicochemical properties of benfotiamine and its possible use as a therapeutic agent 31. Benfotiamine is more easily absorbed by the body and oral administration results in higher thiamine and ThDP blood levels in animals than an equivalent dose of thiamine. A few years, later Shindo and coworkers 32333435 studied in more detail the mechanism of absorption and the metabolic fate of benfotiamine in animal tissues. Their results suggested that benfotiamine (given orally) is first dephosphorylated to S-benzoylthiamine by the ecto-alkaline phosphatase present in the brush borders of intestinal mucosal cells. The more lipophilic S-benzoylthiamine then diffuses through the membranes of intestinal and endothelial cells and appears in the venous mesenteric blood. A significant part of S-benzoylthiamine is captured by erythrocytes 34 and converted to free thiamine through a slow non-enzymatic transfer of the S-benzoyl group to SH groups of glutathione. In the liver, the remainder can be enzymatically hydrolyzed to thiamine and benzoic acid by thioesterases present in the hepatocytes. On the other hand, thiamine disulfide derivatives require a reduction either enzymatically in the liver by glutathione or non enzymatically in blood by glutathione and possibly other substrates 36. In the present work, we show that after oral administration of benfotiamine to mice, free thiamine appears in the liver at a fast rate, reaching a maximum after one hour, while in the blood the maximum is reached only after two hours (Fig. 3). We therefore propose that most of the S-benzoylthiamine present in the mesenteric blood is captured by the liver and transformed into thiamine. The excess thiamine formed is then rapidly released into the blood stream, as shown by the fast decrease of thiamine content in the liver after 1 – 2 hours (Fig. 3). Such a scheme is in agreement with an earlier report that, after infusion of benfotiamine to the small intestine of the dog, mainly free thiamine (not S-benzoylthiamine) was detected in the carotid blood 35. Free thiamine is not lipophilic and cannot cross the blood-brain barrier by simple diffusion. Transport of blood thiamine to the brain parenchyma is carrier-mediated and it is a slow process 37. In the present study, we find that blood thiamine concentration in the control animals is approximately 0.4 μmol/l, a value close to the half-maximal activation constant for the high affinity transport of thiamine across the blood-brain barrier 8. Though a second, low affinity, component of thiamine transport was also observed, its contribution was small. Thus, raising free blood thiamine concentrations does not necessarily lead to an important increase in thiamine transport across the blood-brain barrier. It is therefore not very surprising that benfotiamine administration does not lead to an increase in total thiamine content of the brain (Figs 2, 3 and 4). It should be noted however that one study showed a 90% increase of ThDP levels in the brains of rats that received a dose of 1645 mg of benfotiamine/kg of diet for 6 months 27.

Wada et al. already noted that benfotiamine was sparingly soluble in organic solvents such as benzene, chloroform and methanol, but was readily soluble in aqueous media at pH ≥ 8.0 31. This is not surprising as the phosphoryl group of benfotiamine has two negative charges at alkaline pH. Here, we confirm that benfotiamine is sparingly soluble in water at pH ≤ 7.0 and cannot be dissolved in octanol or oils. Thus benfotiamine should not be classified as a "lipophilic" compound as many authors still do 10242538. Indeed, benfotiamine appears unable to diffuse across cell membranes. We have shown here that intracellular thiamine content is not increased in cultured neuroblastoma incubated in the presence of 10 μM benfotiamine, while it was increased ten-fold after incubation with 10 μM sulbutiamine 29. Moreover, after a chronic treatment of rats with sulbutiamine intracellular thiamine derivatives were increased by respectively 250% (thiamine), 40% (ThMP), 25% (ThDP) and 40% (ThTP) 14.

This is in apparent contradiction with results obtained with cultured cells of endothelial origin 18192039, showing that benfotiamine is able to counteract glucose toxicity in these cells by increasing transketolase activity. However, the benfotiamine concentrations used were 50 – 100 μM, much higher than in our study. Hammes et al. even report that there was no effect on transketolase activity in cultured endothelial cells at 10 or 25 μM 18. In any event, this is no proof that benfotiamine is able to cross the membranes: indeed, cultured endothelial cells seem to possess an ecto-alkaline phosphatase 40. It is therefore likely that, in these cells, the added benfotiamine is at least partially dephosphorylated to S-benzoylthiamine that can enter the cells as in the case of the intestinal mucosa. The slow dephosphorylation to S-benzoylthiamine might also explain the lag period observed between the addition of benfotiamine to thiamine-depleted Neuro 2a cells and the increase in intracellular thiamine derivatives (Fig. 6). In erythrocytes, it was shown that fursultiamine, a lipophilic disulfide, is rapidly incorporated into the cells while benfotiamine is not 41. Taken together, these results strongly suggest that benfotiamine is unable to cross plasma membranes unless it is dephosphorylated.

Since the discovery of allithiamine 9, a number of derivatives were synthesized. These showed higher bioavailability than thiamine hydrochloride or mononitrate. Lipid-soluble thiamine derivatives were developed mainly in Japan for the treatment of beriberi. It was therefore surprising that sulbutiamine appeared to exert specific effects on brain function. Indeed, it seems to improve memory in rodents 4243 and, in humans, it seems to be beneficial against functional asthenias 134445. These effects have not been reported with thiamine. This difference might be explained if we assume that sulbutiamine (or its degradation product thiamine disulfide) can cross the blood-brain barrier and have specific actions in neurons. However, there is so far no direct evidence that untransformed sulbutiamine is indeed found in the brain. Concerning benfotiamine, there is no evidence that it has any specific effect on the central nervous system, but during the last few years, there was considerable interest in the therapeutic potential of benfotiamine in peripheral tissues. Indeed, it was found effective for the protection of diabetic complications such as diabetic neuropathy 181920212223 and alcoholic neuropathy 46. Our results are in agreement with the different pharmacological profiles of sulbutiamine and benfotiamine. We previously found that sulbutiamine treatment significantly increases thiamine, ThMP, ThDP and ThTP content of rat brain 14, while the present results show that benfotiamine, at a twice higher dose, is unable to raise the levels of intracerebral thiamine phosphate derivatives (Figs 2, 3 and 4). This is in agreement with a previous study showing that after administration of ³H-benfotiamine, liver and kidney were labeled to a higher degree than brain and muscles 47, but this study did not make a difference between benfotiamine and its labeled metabolites and degradation products. Furthermore our results on cultured neuroblastoma cells show that benfotiamine, in contrast to sulbutiamine, does not easily cross cell membranes (Figs 5 and 6). It would therefore be interesting to test whether a thiamine disulfide compound such as sulbutiamine or fursultiamine, would not be more efficient and act at lower concentrations than benfotiamine in counteracting diabetic complications. A recent study has shown that, at high concentration (300 μM), benfotiamine exerts a direct antioxidant effects in three different kidney cell lines, independently of its transformation in thiamine and increased transketolase activity 48. It is however not yet clear to what extend, if any, this may be involved in the improvement of diabetic complications.

Benfotiamine (Sigma-Aldrich) was dissolved in a 200 mM solution of hydroxypropyl-β-cyclodextrin (HP-β-CD, Roquette, Lestrem, France) at a concentration of 25 mg/ml. The solution was homogenized at 4°C. The mice were force-fed with a dose of 100 mg benfotiamine/kg in approximately 100 μl using a syringe prolonged by a flexible tube that was inserted into the stomach. The mice strains (male, 8 weeks old) were either C57B6 or Sv129. Control animals received an oral administration of 100 μl HP-β-CD solution (200 mM) without benfotiamine. The local committee for animal care and use approved all animal experiments.

The mice were anesthetized with isoflurane and the forebrain and the liver were collected. For collecting blood samples, the animals were anesthetized with isoflurane and heparin sulfate sodium salt (5000 I.U./ml, Leo Pharma, Wilrijk, Belgium) was injected directly into the heart (50 μl). Tissues were stored at -80°C until sample preparation. About 50 mg of tissue were homogenized in 500 μl trichloroacetic acid (TCA, 12% w/v) in a glass-glass homogenizer. The homogenates were centrifuged (5000 × g, 10 min, 4°C) and the TCA was extracted from the supernatant with 3 × 1.5 ml diethylether. The samples were kept at -20°C until use. HPLC analysis was performed exactly as previously described 49. Protein concentrations were determined by the method of Peterson 50.

Mouse neuroblastoma cells (Neuro 2a) were grown as previously described in 100 mm Petri dishes containing 10 ml Dulbecco's modified Eagle's medium (DMEM, N.V. Invitrogen SA, Merelbeke, Belgium), supplemented with 10% fetal calf serum 29. When the dishes were nearly confluent (5 – 10 mg of protein/dish), the culture medium was replaced by 10 ml of saline (145 mM NaCl, 5 mM KCl, 1 mM MgCl₂, 1 mM CaCl₂, 10 mM glucose, 10 mM Hepes-Tris, pH 7.3) containing either 10 μM benfotiamine or 10 μM thiamine. A 10 mM stock solution of benfotiamine was prepared in a mixture of ethanol (50%) and water. HCl was added until a homogenous solution was obtained. After incubation (0 to 4 hours at 37°C), the cells were centrifuged (2000 × g, 2 min) and the pellet was suspended in 500 μl TCA (12%). The proteins were sedimented (5000 × g, 3 min) and thiamine derivatives were determined in the supernatant after extraction of TCA by diethylether as described above.

Neuro 2a cells were thiamine-depleted as previously described 830. Briefly the cells were grown in DMEM medium devoid of thiamine (Invitrogen). The concentration of fetal calf serum that contains about 14 nM of thiamine, was reduced to 5%. The medium was replaced every 2 days and the cells subcultured every 4 days. After 8 days, either thiamine (1 μM) or benfotiamine (1 μM) were added and the intracellular thiamine levels were determined as a function of time (up to 6 h) as described above.

Data were analyzed by multivariate analyses of variance (MANOVA), followed by ANOVA for follow-up comparisons or two-way ANOVA followed by the Bonferroni post-test.