Carnosine Catalyzes the Formation of the Oligo/Polymeric Products of Methylglyoxal

Diabetic nephropathy (DN) is the most common cause of end-stage renal disease in the Western world. Despite increasing knowledge on the pathogenesis of DN and potential therapeutic interventions, progression to end-stage renal disease still occurs in a substantial number of diabetic patients. Tight control of blood glucose continues to be the primary treatment option, but only partially prevents the manifestation of late complications and with large inter-individual variations [1, 2]. Substantial research has linked the highly reactive, glucose-derived metabolite methylglyoxal (MG) to diabetic complications [3, 4]. MG is a potent glycating agent [5] leading to the formation of advanced glycation end products (AGEs) [6] and is formed by the non-enzymatic dephosphorylation of triosephosphates [7, 8]. The formation of the MG-derived AGE and MG-derived hydroimidazolone isomer 1 (MG-H1) correlates with the development of diabetic complications such as DN [9]. Reactive metabolites such as MG are considered secondary messengers of free radicals that act both as signaling molecules and cytotoxic products of lipid peroxidation causing long-lasting biological consequences, in particular by covalent modification of macromolecules [10]. Reducing MG concentrations and antagonizing MG actions is a promising means to prevent diabetic complications [11]. The histidine-containing dipeptide carnosine is able to quench reactive carbonyl species such as MG or the α, β-unsaturated aldehyde 4-hydroxy-trans-2-nonenal (HNE) [12, 13]. Carnosine acts through two reactive moieties, one of the nitrogen atoms of the imidazole ring or the amino group nitrogen of the β-alanine residue. These nucleophilic nitrogens target the two reactive sites of α, β-unsaturated aldehydes, the aldehyde function or the electrophilic C3 atom [14]. Cellular uptake of carnosine occurs by proton-coupled oligopeptide transporters (POTs), membrane proteins that translocate various small peptides and peptide-like drugs across the biological membrane via an inwardly-directed proton gradient and negative membrane potential [15]. At present, four members of the POT family, namely PEPT1 (SLC15A1), PEPT2 (SLC15A2), PHT1 (SLC15A4) and PHT2 (SLC15A3), have been identified in mammals [16]. Unlike PEPT1 and PEPT2, PHT1 [17] and PHT2 also transport a single amino acid, L-histidine, in addition to the proton-stimulated transport of di- and tripeptides. The uptake of β-alanine, the precursor of carnosine, is largely driven by a Na+/Cl- dependent taurine transporter (TauT, SLC6A6) [18].

In diabetic mice, carnosine supplementation mitigates DN, reduces renal vasculopathy [2], normalizes vascular permeability [2] and improves wound healing [19]. In streptozotocin-induced diabetic rats, carnosine treatment prevents apoptosis of glomerular cells, podocyte loss [20, 21] and vascular damage [22]. In patients with type 2 DM we recently demonstrated an association of the susceptibility to develop DN with a polymorphism in the carnosinase 1 gene CNDP1 [23-25]. The kidney has an intrinsic carnosine metabolism with carnosine synthase and carnosinase 1 (CN1) being expressed in glomeruli and tubular cells [26]. CN1 activity is increased under diabetic conditions [27]. Carnosine uptake into renal cells occurs by the proton-coupled oligotransporter PEPT2 [16].

We now provide evidence that carnosine efficiently catalyzes the formation of MG oligo/polymeric products and prevents the formation of AGEs, but at molar ratios to cytotoxic MG concentrations which exceed amounts tolerated by renal cells and which are unable to be reached by cellular carnosine uptake in any case.

Chemicals were purchased from Sigma-Aldrich (Munich, Germany) unless indicated otherwise.

Carnosine, anserine (obtained from Bachem AG, Switzerland) and MG were incubated in a 1: 1 molar ratio in 100 mM sodium phosphate buffer at pH 7.4 with final concentrations of 10 mM. tert-Butyloxycarbonyl lysine (BocLys), lysine with a tert-butyloxycarbonyl protecting group, was used as a control due to being able to rapidly react with MG and form AGEs [28]. The reaction was followed by absorbance measurement at 336 nm for formation of the products for 8 hours with measuring intervals of 5 minutes.

NMR spectra were acquired at 25°C using a Bruker Avance II NMR spectrometer equipped with a 5-mm inverse-configuration probe with triple-axis-gradient capability at a field strength of 14.1 T operating at 600.1 MHz for 1H and 150.9 MHz for ¹³C nuclei. Pulse widths were calibrated following the described protocol [29]. Chemical shifts (δs) are reported relative to the internal standard 3-(trimethylsilyl) propionic-2, 2,3, 3-d4 acid sodium salt (TSP) for which δ_H=0 ppm and d_C=0 ppm. General NMR experimental and acquisition details including the 1D NOESY pulse sequence with pre-saturation for water suppression, selective saturation transfer (saturation time 5 seconds) and T₂ filter with pre-saturation (τ_lock 1.2 seconds) and standard gradient-selected 2D COSY, HSQC and HMBC spectra with pre-saturation have been previously described [30, 31]. For time-course experiments, samples were accurately weighed and made up volumetrically to 500 µl to provide 50 mM solutions of both MG and the amino acid/dipeptide using aqueous sodium phosphate buffer (375 µl, 100 mM, pH 7.4) and D₂O (125 µl) containing TSP while samples for signal assignment were prepared similarly but with nominal dispensation. Following preparation, solutions were transferred to 5-mm NMR tubes for immediate measurement.

MG non-hydrated form (<0.05% relative to total MG). 1H NMR: 9.452 (s, H1), H3 not found. MG monohydrate (61% relative to the dihydrate only). ¹H NMR: 5.284 (s, H1), 2.305 (s, 3H, H3). ¹³C NMR: 211.99 (C2), 92.68 (C1), 27.37 (C3). MG dihydrate (39% relative to the monohydrate only). ¹H NMR: 4.816 (s, H1, δ_H taken from the HSQC spectrum due to overlap with the HDO signal), 1.377 (s, 3H, H3). ¹³C NMR: 98.11 (C2), 94.89 (C1), 24.34 (C3). Oligo/polymeric species were not structurally identified but ¹H methyl signals at 1.472, 1.476, 1.482, 1.507, 1.518, 1.541, 1.544, 1.551 and 1.652 ppm were shown to belong to such species by 1D selective saturation transfer.

Quantification of MG modification of proteins was performed according to Ahmed et al. [32]. In brief, human serum albumin was incubated with MG in a ratio of 1: 10, respectively. Carnosine or anserine were added in ratios of 1: 1; 2: 1; 10: 1; 20: 1; 100: 1; 200: 1 and 1, 000: 1 to MG. Incubation was performed at 37°C with continuous shaking for 24 hours. For purification, the modified proteins were concentrated by centrifugal filtration with a 10 kDa cut-off filter and washed three times with 50 mM ammonium carbonate buffer. The washed proteins were collected and transferred to a glass vial and lyophilized to dryness. Protein samples were analyzed either by SDS-PAGE and Western blot for N-carboxyethyl lysine (CEL) or by ELISA for MG-derived AGEs. For analysis by Western blot, 0.5 µg of protein were denatured in Laemmli buffer, resolved by SDS-PAGE (4 - 15% Mini-PROTEAN TGX, Biorad, Hercules, CA) and transferred onto a nitrocellulose membrane and blocked using 2% milk powder in PBS and 0.05% Tween 20 (PBS-T) at room temperature for 1hr. Membranes were then incubated overnight at 4°C with antibodies against CEL (1: 500 dilution, CEL025, mouse, Biologo) and Albumin (1: 10,000 dilution, 4929, rabbit, Cell Signaling Technology) in 2% milk powder in PBS-T. Blots were washed and then incubated with an appropriate HRP-conjugated antibody (1: 2000 in 2% milk powder in PBS-T) for 1 hour at room temperature. Proteins were visualized on X-ray films using ECL detection reagents (GE Healthcare, Buckinghamshire, UK) with varying exposure times (0.1–2 min). The measurement of the content of the MG-derived AGEs was performed using a competitive ELISA, according to the manufacturer’s instructions (OxiSelect Methylglyoxal Competitive ELISA, STA-811, Cell Biolabs, San Diego, CA).

For measurement of MG-H1 formation by Western blot, tubular cell samples were lysed in RIPA-buffer (radioimmunoprecipitaion assay buffer: 150 mM NaCl, 0.1 % Triton X-100, 0.5 % sodiumdeoxycholate, 0.1 % SDS, 50 mM Tris-HCl (pH 8.0) and one proteaseinhibtor cocktail tablet (cOmplete tablets, Mini EASYpack, Roche Diagnostics, Mannheim, Germany) per 15 ml) and separated by SDS-PAGE in 8% polyacrylamide gels (Twin-Plate Mini Gel Unit, Carl Roth GmbH + Co. KG, Karlsruhe, Germany). Samples were transferred to a nitrocellulose membrane by semi-dry blot. The membrane was blocked with protein-free blocking buffer (Pierce Protein-free Blocking Buffer, Thermo Fisher Scientific, Waltham, MA) for 1 hour at room temperature and incubated with anti-MG-H1 antibody (1: 400 in protein-free blocking buffer) at 4°C over night and after washing incubated with a secondary HRP-conjugated antibody (1: 1000 in protein-free blocking buffer). Western blots were developed with Clarity Western ECL Substrate (Biorad, Hercules, CA) and visualized by a fluourescence imaging system (PEQLAB fusion, PEQLAB, Erlangen, Germany).

Conditionally immortalized murine podocytes (ImmortoMouse, Charles River, Wilmington, MA) were grown in RPMI 1640 medium (Thermo Fisher Scientific, Waltham, MA) with 10% (v/v) fetal calf serum (Biochrom, Berlin, Germany) and 1% penicillin and streptomycin (v/v) (Thermo Fisher Scientific, Waltham, MA) with 10 U/µl γ-interferon (Roche Diagnostics, Mannheim, Germany) at 33°C with 5% CO₂ on collagen type I (BD Biosciences, Bedford, MA) coated flasks. Cells were then detached with trypsin (Thermo Fisher Scientific, Waltham, MA) and seeded at 5, 000 cells per cm² to allow them to differentiate for 12-14 days at 37°C with 5% CO₂ in growth medium without γ-interferon. Immortalized human tubular cells (HK-2; American Type Culture Collection CRL-2190) and immortalized murine mesangial cells (Mes13; American Type Culture Collection CLR-1927) were grown in RPMI 1640 medium with 10 % fetal calf serum (v/v) and 1% penicillin and streptomycin (v/v) at 37°C with 5% CO₂.

Cell viability was measured with MTT assay. In brief, confluent cells were detached with trypsin (Thermo Fisher Scientific, Waltham, MA) and seeded on 96-well plates followed by an additional growth phase for 24 hours. For short-term experiments cells were then treated with MG and carnosine for a further 48 hours, or 7 days (tubular cells) or 14 days (podocytes) for long-term incubations. Media were changed every three days. At the end of the incubation period, 50 µl PBS (Thermo Fisher Scientific, Waltham, MA) with 2 mg/ml 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT) was added to each well. This compound is converted to purple formazan crystals by metabolically active cells. After 4 hours, the medium was discarded and cells were lysed with 200 µl DMSO per well for 1 hour. The absorption of the solution was measured spectrophotometrically at 590 nm, the absorption maximum of formazan crystals.

Cells were harvested by adding 60 µl pre-lysis buffer containing 20 mM Tris/HCl (pH 8.9, 150 mM NaCl, 20 mM NaF and 1% Triton) for detection of CN1 activity and additionally 2 mM EDTA was added for measurement of anserine and carnosine levels.

CN1 activity was assayed as described previously [33]. In brief, the reaction was initiated by the addition of 1 mM carnosine to cell homogenate at pH 7. The reaction was terminated at pre-determined intervals (every 5 minutes) by adding 1% trichloroacetic acid. Liberated histidine was derivatized by adding o-phthalaldehyde and the fluorescence measured using a MicroTek plate reader (λ_Ex:360 nm; λ_Em: 460 nm).

Anserine and carnosine concentrations were measured fluorometrically using HPLC as previously described [34]. The homogenized cells were diluted with sulfosalicylic acid and derivatized using carbozole-9-carbonyl chloride, applied to a Jupiter column (C18, 300 Å, 5 µm particle size, 250 x 4.6mm) and detected by fluorescence detection (Shimadzu RF-20A) at λex 287 nm/λem 340 nm. The mobile phase consisted of solution (A) 50 mM acetate buffer in distilled water, pH 4.37 and solution (B) a mixture of acetonitrile, methanol and tetrahydrofuran 70: 25: 5 (v/v/v). The solutions were degassed prior to use. All samples were measured at least twice and one sample was spiked with standards to definitively identify each analyte. The reliability of the method was 0.91.

Superoxide dismutase activity of all three types of SOD (Cu/Zn, Mn and FeSOD) was measured with renal cell lysate using Superoxide Dismutase Assay Kit (Cayman Chemical Company, Ann Arbor, MI) according to the manufacturer’s manual.

The concentration of reduced glutathione (GSH) was determined in renal cells using a Quantichrome Glutathione Assay kit (Bioassay Systems, Hayward, CA) following the manufacturer’s instructions.

RNA purification was performed with RNeasy Minikit (Quiagen, Venlo, Netherlands) with samples being treated with DNase I (Roche, Basel, Switzerland) for degradation of genomic DNA. Purified RNA was transcribed to cDNA via EpiScript reverse transcriptase (Biozym, Oldendorf, Germany) and non-specific random hexamere primers (Thermo Fisher Scientific, Waltham, MA). Expression of the genes of interest (GPx-1, SOD2 and putative carnosine transporters PEPT1, PEPT2, PHT1 and PHT2) was then analyzed with specific primer pairs via quantitative real time polymerase chain reaction (qPCR) as described before [35]. Table 1 shows the utilized PCR primers.

For the carnosine uptake measurement, renal cells were seeded at densities of 10, 000 cells per well for tubular and mesangial cells and 25, 000 cells per well for podocytes on a 96-well plate in 200 µl of growth medium. Cells were allowed to adhere overnight at 37°C after which medium was removed and the cells washed with PBS. Krebs-Ringer-HEPES (KRH) buffer was added (180 µl per well) and 20 µl of [³H]-L-carnosine (Hartmann Analytic, Braunschweig, Germany), solved in KRH buffer, for 0, 5, 10, 25, 50 and 100 µM final concentrations. The cells were further incubated at 37°C for 20 hours, then washed twice with ice-cold PBS. Either KRH buffer (20 µl per well with cells) or carnosine solution (20 µl per well without cells) was added. After the addition of 200 µl of MicroScint-O scintillation liquid (PerkinElmer, Waltham, MA) to each well and 60 minutes of incubation, the ³H signal was measured with a TopCount NXT microplate scintillation and luminescence counter (Packard, Meriden, CT).

Data were obtained from at least 3 independent experiments while cell viability data were obtained from at least 3 independent experiments comprising at least 5 measurements per data point. Data are given as mean and standard deviation (SD). Effects of MG, carnosine and anserine on cell viability were analyzed with a multiple regression approach. Post-hoc comparisons were computed using least-square means with Dunnett’s contrasts (e.g. the control group were compared to each of the other experimental condition) implemented in Lenth [36]. Effect graphics based on effects-package for R [37]. All statistical analysis were computed with R. A p-value of ≤0.05 was considered significant.

To first evaluate the quenching activity of MG, the absorbance at 336 nm in a cell free environment was measured after co-incubation of MG with carnosine and anserine for 8 hours (ratio 1: 1, 10 mM final concentration). The absorbance at 336 nm increased in a time-dependent manner and was higher for carnosine than for anserine, the latter being comparable with the effect seen for BocLys which served as a quenching control (Fig. 1A). The color of the solutions changed from clear to brownish yellow (Fig. 1B) in each case and the reaction was enhanced by increasing pH. The color change is indicative of the Maillard reaction and the consumption of free MG. The effects were more prominent for carnosine as compared to both BocLys and anserine. To investigate the reactions further, ¹H NMR analysis was conducted.

By ¹H NMR analysis of the co-incubation of MG with either carnosine or anserine, only transient MG-carnosine and MG-anserine adducts (Fig. 1C) were detected and only in very low concentrations prior to the gradual and significant accumulation of oligo/polymeric products of MG evident in the ¹H NMR spectra (Fig. 2) concomitant with the consumption of free MG while the majority of carnosine/anserine on the whole remains unaltered. These observations are consistent with the aforementioned color changes. Since the MG adducts were present in only very low concentrations and are transient species which could not be isolated and they only provided broad signals in the NMR, they could not be fully characterized and their structures determined. Their formation was clear though, by both the appearance of new signals, albeit at low intensities, and by their catalytic effect on the formation of oligo/ polymeric products of MG as well as the observation of saturation transfer from the aromatic signals of the dipeptides. Whether these transient adducts are directly involved in the catalysis as depicted in Fig. 2 is only a presumption but notwithstanding that stable 1: 1 adducts between MG and carnosine or anserine were not formed. The signals attributed to oligo/polymeric products of MG were shown to be high-molecular weight species by the application of a T₂ filter and were similarly shown to be in a dynamic equilibrium with the various free MG species (the non-hydrated form, the monohydrate and the dihydrate) by saturation transfer experiments from the 1H methyl signals of free MG. Co-incubations of MG with BocLys (Fig. 2) and imidazole were also examined for comparison and provided similar results to the co-incubation of MG with carnosine or anserine. Thus, the reaction with BocLys shows that the amine group nitrogen reacts with MG, while reaction with imidazole shows that one of the imidazole nitrogens reacts with MG. Hence carnosine can potentially react with MG either via the amine group nitrogen or one of the imidazole nitrogens. For N-methylated carnosine, i.e. anserine, both imidazole nitrogens are blocked, i.e. non-reactive, and therefore reaction is slower. Thus, the free imidazole nitrogen reacts faster than the amine group nitrogen and hence the reaction order carnosine > anserine > BocLys is rationalized. Times for half of the MG to be consumed in the reactions were 0.6, 2.4 and 5.8 hours for carnosine, anserine and BocLys, respectively (data not shown).

The formation of AGEs derived from human serum albumin in the presence of MG was significantly inhibited by the addition of carnosine at a ratio of 100: 1 to MG (Fig. 3A); for anserine, a higher ratio to MG of 200: 1 was required for significant inhibition (Fig. 3B). To reduce AGE formation by 50%, a carnosine to MG ratio of 400: 1 and an anserine to MG ratio of 700: 1 were required. N-carboxyethyl lysine (CEL) formation from human serum albumin and MG was also reduced dose-dependently by carnosine or anserine. Their reduction of CEL was even more effective with MG to carnosine (Fig. 3C) and MG to anserine (Fig. 3D) ratios of as little as 1: 20 and 1: 1, respectively, decreasing CEL formation significantly.

Quantitative PCR demonstrated that PHT1, one of four known proton-dependent oligopeptide transporters (POTs), together with taurine transporter (TauT) which transports β-alanine, was expressed in tubular, mesangial cells and podocytes. The three other POTs, i.e. PEPT1, PEPT2 and PHT2, were only expressed in tubular cells (Fig. 4A). Uptake of radiolabeled carnosine after 20 hours of incubation was only minor in tubular cells and mesangial cells and almost absent in podocytes (Fig. 4B). Accordingly, intracellular carnosine concentrations hardly increased. Incubation with 0.1 mM radioactive carnosine for 20 hours increased intracellular carnosine concentrations for tubular cells only to 0.07±0.01 µM (p≤0.001) and for mesangial cells only to 0.05±0.03 µM (p≤0.05).

Podocytes, tubular and mesangial cells tolerate carnosine and anserine in high concentrations but to different degrees (Table 2). Carnosine was tolerated best by podocytes whereas the tolerances by tubular (Fig. 5A) and mesangial cells of carnosine were less. Mesangial cells tolerated anserine at higher concentrations as compared to podocytes and tubular cells (Fig. 5B). Exposure to 0.15 mM MG had no effect on cell viability whereas levels above 0.2 mM resulted in a dose-dependent decrease of cell viability (Fig. 5C) for all three renal cell types in comparison to controls. MG was tolerated by podocytes at higher concentrations in comparison to tubular and mesangial cells. Exposure to 0.2 mM MG increased cell viability for tubular cells by 29%±3.4% (p<0.001) and for podocytes by 43%±12.7% (p<0.001), but not for mesangial cells.

Renal cells were then exposed to MG (0.15, 0.3 and 0.5 mM) with increasing concentrations of carnosine and anserine (0.1-0.9 mM). Exposure of tubular cells to 0.15 mM MG (Fig. 5D) had no effect on cell viability independently of the amount of carnosine or anserine added. Exposure to 0.3 mM MG (Fig. 5E) decreased cell viability significantly for carnosine concentrations ranging from 0.2 to 0.5 mM (p<0.01, Dunnet contrasts) while higher carnosine levels had no effect on cell viability. The addition of 0.5 mM MG (Fig. 5F) reduced cell viability to 26.2±16.1% (p<0.05, Dunnet contrasts). The addition of 0.5 and 0.6 mM carnosine increased cell viability at a 6% significance level. In podocytes, no beneficial effect of carnosine (0.1 - 0.9 mM) on MG-induced stress (0.15, 0.3 and 0.5 mM) was detected via Dunnet contrasts. Co-incubation of podocytes with 0.5 mM anserine and 0.15 mM MG even reduced cell viability to 79.6±6.5% compared to the untreated control (p=0.01, Dunnet contrasts). Likewise, extension of MG and carnosine co-incubation to 7 days (tubular cells) and 14 days (podocytes) did not consistently reduce MG-induced cell toxicity. Co-incubation with the carnosine precursor histidine and β-alanine (1: 1 ratio, 0-1.0 mM) with 0.4 mM MG for 48 hours also had no protective effect on MG-induced cell toxicity in mesangial cells, higher levels of β-alanine and histine (1.5 mM/ even decreased cell by proliferation about 7% (66.6±8.9%, p< 0.05).

The incubation of tubular cells with MG increased intracellular CN1 activity. Incubation with 0.5 mM MG increased CN1 activity from 2.1±0.2 to 3.5±0.3 µmol/mg/h (p<0.05). Intracellular carnosine and anserine varied greatly with a mean of 3.6 nmol/mg anserine and a standard deviation of 3.1 [95% CI 1.6-7.5] and a mean of 1.5 nmol/mg carnosine [95% CI 0.9-2.9]. MG-induced increased CN1 activity did not result in a significant decline in intracellular carnosine or anserine levels.

In tubular cells, addition of 0.3 mM and 0.5 mM MG almost halved SOD2-expression, also known as manganese-dependent superoxide dismutase (MnSOD), within 24 hours to 62.3±24.1% (p≤0.05) and 58.8±18.5% (p≤0.01), respectively, compared to untreated controls and was not influenced by the addition of 0.5 mM carnosine. However, activity (7.69±1.44 U/mg) of all three types of SOD (Cu/Zn, Mn and FeSOD) was not influenced by incubation with 0.5 mM MG (7.16±1.28 U/mg) or 0.5 mM carnosine (7.16±0.87 U/mg; both p=ns). SOD activity in podocytes (15.75±2.09 U/mg) was neither significantly altered by incubation with 0.15, 0.3 or 0.5 mM MG (14.7±1.75 U/mg, 10.55±4.15 U/mg and 13.41±1.08 U/mg, respectively; all p=ns) nor by co-incubation with 0.5 mM MG and 0.5 mM carnosine (16.2±2.93 U/mg; p=ns).

Tubular cell glutathione peroxidase gene (GPx-1) expression was unchanged after 24 hours of incubation with 0.3 mM MG (95.8±40.3%), 0.5 mM carnosine (87.6±19.7%) or the combination of both (95.3±17.1% and 98.2±21.8%; all p=ns). GSH concentrations in tubular cells and podocytes remained in the same range with increasing levels of MG or carnosine or the combination of both (data not shown). Also, MG-H1 formation was not reduced by the addition of carnosine (Fig. 6).

Reactive metabolites essentially contribute to the development of DN and the formation of the MG-derived MG-H1 closely correlates with the degree of DN [11, 38]. Recent studies have demonstrated the beneficial effects of carnosine on the development of diabetic complications [2, 19, 22, 27] and it is assumed that those effects are, at least partially, due to the MG-scavenging potential of carnosine. The quenching capacity and structure related reactivity of carnosine for reactive species has already been reported [39-41]. We therefore investigated the scavenging potential of carnosine and anserine on MG-toxicity in renal cells.

In summary, carnosine and anserine both prevented MG-induced AGE and CEL formation in vitro and NMR studies revealed that the two compounds catalyzed the formation of MG oligo/polymeric products, presumably via transient adducts between MG and carnosine/anserine that were observable in only very minor concentrations, rather than forming stable 1: 1 MG adducts. These findings are in line with recent studies demonstrating poor quenching activity of carnosine derivatives towards MG [14, 42]. Quenching capacity of carnosine is higher against other reactive metabolites such as acrolein and HNE [14, 40, 41, 43]. Although both, HNE and MG, belong to the group of reactive carbonyl species reacting with the nucleophilic groups in proteins, the two compounds are chemically distinct with both very different reaction kinetics leading to different stable end products. Dicarbonyl derivatives such as MG are more likely involved in AGE formation, while α, β-unsaturated carbonyls such as HNE generate advanced lipidoxidation end products (ALEs) [41]. Thus, the scavenging ability of carnosine or anserine against HNE, acrolein or other reactive compounds might be of higher relevance.

A protective effect on cell viability or oxidative stress markers of carnosine or anserine addition to MG-stressed renal cell cultures could not be observed, independent of the cell type, incubation times and MG to carnosine/anserine ratios. The different findings in vitro and in vivo might be caused by not achieving the required intracellular carnosine or anserine to MG ratios and/or carnosine is mainly acting extracellularly. In our experiments, in vitro CEL formation was reduced by MG to carnosine ratios of 1: 20 and by MG to anserine ratios of 1: 1. However, carnosine uptake was low in mesangial and tubular cells, almost absent in podocytes and remained below 1 µM for all cells within 20 hours and is probably not crucial for the beneficial effects reported earlier. Although the putative carnosine transporter PHT1 was expressed in all cell types, podocytes showed no carnosine uptake and the cell-specific role of all transporters needs further investigation. On the other hand, the dynamics of MG are faster. The majority of MG is either quickly bound to protein or was detoxified via the glyoxalase pathway involving glyoxylase I and II, metabolizing MG into lactate by using NADPH and GSH [44]. Previous findings showed peak formation of MG-H1 in endothelial cells already after 6 hours and that carnosine fails to prevent the formation of MG-H1 from human serum albumin in vitro [42]. In our experiments, the addition of carnosine did not reduce MG-H1 formation in renal cells.

Since intracellular carnosine levels varied but were hardly influenced by the quantity of carnosine added, carnosine and anserine may act extracellularly. Also the addition of the carnosine precursors β-alanine and histidine to the cells did not increase intracellular carnosine or anserine levels. The putative β-alanine transporter TauT is expressed in all renal cells and we previously demonstrated carnosine synthase (CS) expression in podocytes and tubular cells [26]. On the other hand, increased CN1 activity due to MG-induced carbonylation of the enzyme, as described before [27], also did not alter intracellular carnosine levels. Although we only found low carnosine transport into cells, all three cell types only tolerated a certain amount of carnosine or anserine. Whether the comparatively high tolerance of podocytes against carnosine is due to hardly any carnosine uptake into cells remains speculative. We used both human and mice kidney cells and cannot exclude that our cell lines may have lost parts of their original expression profiles. Reassuringly, CN1 and CS are both expressed in the kidney of human and mouse [26].

Although carnosine showed no protective effect against MG-induced toxicity, low MG levels (0.2 mM) resulted in increased cell viability in podocytes and tubular cells. A protective response in yeast and in mouse cardiac endothelial cells via the initiation of cellular defense mechanism was recently reported [45, 46]. The effectors of MG defense include heat shock proteins and proteins involved in mitochondria-associated protein degradation. No protective effect of MG was detectable for mesangial cells which are particularly sensitive to MG. MG and MG end products CEL and MG-H1 have been reported [9] to be involved in the progression of DN, thus the different sensitivity of renal cell types towards MG might be considered to be a reflection of the pathomechanisms of early DN in vivo. Besides podocyte loss, which seems to be an important pathological feature in DN, hypertrophy of mesangial cells [47] seems to be of importance in the progression of DN and might be caused by MG-induced toxicity.

Taken together, ex vivo carnosine prevents MG-induced CEL formation but in vivo the mechanism of action is unlikely to be through intracellular MG scavenging. Cellular uptake and intracellular synthesis of carnosine and anserine are too low to achieve the concentrations required for efficient quenching and/or MG polymer formation in order to prevent MG-dependent MG-H1 and AGE generation and to preserve renal cell viability. Other mechanisms must be active to explain the repeatedly observed nephroprotective action of carnosine supplementation in diabetic rodents and the observed association of CN1 activity defining CNDP1 genotypes and the risk of nephropathy in diabetic humans. The function of carnosine as an antioxidant, as discussed in the literature [48-52], is controversial and its role in diabetes requires further investigation.

AGE (Advanced glycation end product); CEL (N-carboxyethyl lysine); CN1 (Carnosinase 1); CNDP1 (Carnosinase 1 gene); DN (Diabetic nephropathy); HNE (4-hydroxy-trans-2-nonenal); MG (Methylglyoxal); MG-H1 (MG-derived hydroimidazolone isomer 1); POT (proton-coupled oligopeptide transporter); TauT (Taurine transporter).

This work was supported by the Deutsche Forschungsgemeinschaft (DFG, SFB 1118).

All authors were involved in planning this article, interpretation of literature, and in drafting and critically revising the manuscript. All authors reviewed the final manuscript and gave approval for submission. We thank Kristina Klingbeil for excellent technical assistence.

The authors declare that they have no conflict of interest.

T. Weigand and B. Singler contributed equally to this paper.