l-Carnosine, a Substrate of Carnosinase-1, Influences Glucose Metabolism

The cDNA of hCN1 including the endogenous signal peptide of six leucines was amplified from IMAGE clone acc. no. BX094414 using the primers CN1 forward 5′P-CACCATGGATCCCAAACTCGGGA3′ and CN1 reverse 5′P-TCAATGGAGCTGGGCCATCT3′. The PCR product was ligated into the StuI site of plasmid pTTR1ExV3, containing the transthyretin promoter (TTP) (9). The resulting plasmid, pTTP-hCN1, was sequenced. The transgene was liberated using HindIII and injected into the pronuclei of BKS.Cg-m^+/+ Lepr db/J mice (line 000642; The Jackson Laboratory, Bar Harbor, ME) at the Interfaculty Biomedical Research Institute, University of Heidelberg. All animal procedures were approved by the Regierungspräsidium Karlsruhe (AZ 35-9185.82/A-2/06, AZ53-9185.81/G-32/03).

Genomic DNA was extracted from tail biopsies using the mouse direct PCR Lysis Reagent Kit (Viagen Biotech, Los Angeles, CA) supplemented with proteinase K (Sigma-Aldrich) according to the manufacturer's instructions. The transgene TTP-hCN1 was detected by PCR using the primers CN1Tail forward CCTCGCTTCAGACAAGAGCTCTTCAGAATGA and CN1Tail reverse GCTCTGAAGGCGCTCACAGCATTGATCCAA (at 94°C for 3 min, annealing at 60°C, 2-min extension, 31 cycles) yielding a product of 343 bp (10). The point mutation within the leptin receptor was genotyped by PCR as described by http://jaxmice.jax.org. hCN1 transgenic db/wt mice were bred to nontransgenic db/wt littermates. Diabetic transgenic db/db mice were used as experimental animals (db/db CN1+). Diabetic nontransgenic (db/db water) as well as nondiabetic nontransgenic littermates (db/wt) served as controls. Transgenic mice without the Lepr mutation (wt/wt) were discarded.

A second experimental group of nontransgenic db/db mice (db/db l-carnosine; The Jackson Laboratory) was given drinking water supplemented with 4 mmol/l l-carnosine or water only (db/db water). l-carnosine–supplemented drinking water was replaced three times a week. Direct measurements of l-carnosine within the drinking bottles verified that l-carnosine was stable over a period of at least 5 days at room temperature (data not shown). Daily drinking volumes were determined. Control animals were identical in every aspect (i.e., glucose metabolism and diabetic nephropathy) to control animals derived from our own colony (not shown). Unless stated otherwise, each experimental group contained 8 mice, except for the group of db/db water mice, which contained 16 mice. Experimental mice were housed at 22°C on a 12-h light/dark cycle and fed standard rodent diet (R/M-H diet for mice and rats; Sniff, Soest, Germany). Each group of experimental animals contained the same ratio of females to males.

RT-PCR was performed according to a previously described protocol (11). Photometric RNA concentrations and RNA volumes used for reverse transcription of exactly 5 μg RNA per tissue were as follows: db/db CN1+liver, 1.28 μg/μl per 3.9 μl; db/db water liver, 1.56 μg/μl per 3.2 μl; and db/db CN1+kidney, 1.52 μg/μl per 3.3 μl. cDNA generated from exactly 250 ng tissue RNA or 250 ng hCN1 plasmid were used per PCR. For detection of transgenic hCN1, endogenous mCN1, and β-actin, respectively, 30, 40, and 25 cycles of replication were used. Primers used for specific detection of the hCN1 transgene were forward 5′-cac ctc tgg agg tac cca tc-3′ and reverse gca tcc agg act ttg acc at-3′. Primers used for specific detection of endogenous mouse CN1, forward 5′-tca gga ccc gag tca tct tc-3′ and reverse 5′-ctt gct gat gtc agg gtt ga-3′, were directed against the 3′ untranslated region of the mRNA molecule, which was not present in the transgene.

Molecular biology techniques were performed as described according to standard protocols (12). Immunoblotting or immunohistological staining for CN1 was performed using C17E antiserum (kind gift from Dr. Teufel, Strasbourg, France) (7) and anti-CN1 polyclonal antiserum (R&D Systems, Wiesbaden-Nordenstadt, Germany) with identical results.

Total body weights and fasting plasma glucose (FPG) levels were measured each week at the beginning of the light cycle after an overnight fast until week 22 using Ascensia ELITE XL (Bayer, Leverkusen, Germany). A1C levels were determined using MicromatII (Bio-Rad, Hercules, CA). Insulin levels in mouse serum or INS-1e supernatants were determined using the Mouse Insulin ELISA kit (EZRMI-13K; Linco Research, St. Charles, MO). Glucose and insulin tolerance tests were performed on animals that had been fasted overnight for 16 h as described (13). In brief, animals were injected with either 2 g/kg body wt glucose or 4 IU/kg body wt human regular insulin (Novo Nordisk, Copenhagen, Denmark) into the peritoneal cavity. Glucose levels were determined in blood collected from the tail tip immediately before and 30, 60, and 120 min after the injection. Animals were placed in metabolic cages and received only water supplemented with 4% sucrose. Twenty-four–hour excretion of albumin, urea, and creatinine was measured at 20 weeks of age using Hitachi Analyzer 912 (Roche, Mannheim, Germany).

Measurement of carnosine concentration in mouse EDTA plasma was performed according to the method of Schonherr (14) with modifications. Moreover, 80 μl EDTA plasma was mixed with 20 μl 10% 5-sulfosalicylic acid, 10 μl supernatant was mixed with 20 μl 0.4 mol/l borate buffer (pH 9.5), and 10 μl supernatant was mixed with 10 μmol/l standard carnosine solution. Blank values, carnosine concentration standards, and carnosine composition standards (anserine, histidine) were prepared for each measurement in parallel with the samples.

For derivatization, 30 μl of each sample was mixed with 30 μl of freshly prepared CFC solution (2.5 mmol/l in acetone). After 90 s, 18 μl 1 mmol/l EDTA in 0.1 mol/l NaOH and 0.5 mol/l hydroxylamine hydrochloride in distilled water and 2-methylthioethanol was added. After an additional 3 min, 42 μl quench solution, 20% (vol/vol) acetic acid in acetonitrile, was added. Then 40 μl of diluted sample was injected for separation into the high-performance liquid chromatography system LaChrom Elite HPLC system (Jupiter column, C18, 300 Å, 5 μm particle size; Phenomenex, Aschaffenburg, Germany). The elution was performed with fluorescence detection at 287 nm extinction and 340 nm emission. The profile of the binary gradient was as follows: 0–45 min, 82% A and 18% B; 46–50 min, 18–100% B; 50–55 min, 100% B; 56–60 min, 100–18% B; and 60–80 min, 82% A and 18% B. The retention times strongly depended on the mobile phase composition.

At the end of the observation period, animals were killed at 24 weeks of age and perfused with 3% paraformaldehyde as described (15). Semi-thin kidney sections, 1 μm thick, were evaluated by two renal pathologists in a blinded fashion for mesangial expansion. At least 50 glomeruli were evaluated from each experimental animal; mesangial expansion was classified into three groups: 0, no mesangial expansion; 1, mild; and 2, significant mesangial expansion.

The insulin-producing INS-1e cells were cultured in RPMI-1640 supplemented with HEPES, 10% heat-inactivated fetal bovine serum, 100 units/ml penicillin/streptomycin, 1 mmol/l sodium pyruvate, and 100 μmol/l 2-mercaptoethanol (16). The cells were exposed to various stimuli for 2 h in Krebs-Ringer bicarbonate HEPES buffer (n = 3); 5 mmol/l l-glutamine served as negative control, and l-glutamine and 20 mmol/l l-leucine or 30 mmol/l KCl served as positive control (Sigma-Aldrich, Munich, Germany). Insulin levels were analyzed as a ratio, where basal insulin secretion in the presence of only l-glutamine was normalized to 1. To assess the proliferative index, INS-1e cells were pulsed with BrdU for 2 h, fixed, and stained using an anti-BrdU antiserum (MAB3424; Chemicon International, Hampshire, U.K.).

To lower l-carnosine level within the serum of diabetic mice, the cDNA of hCN1 was placed under the control of the human transthyretin promoter enhancer (TTP promoter) (Fig. 1). The TTP promoter was chosen because it drives expression in hepatocytes as well as the choroid plexus, recapitulating the expression pattern of endogenous hCN1 in humans (9). Low-level expression of transgenic hCN1 was also seen in α-cells of the pancreatic islets (Fig. 2D–F). Rodents do not endogenously express a secreted form of CN1 (7,8). The cDNA of hCN1 contained the endogenous signal peptide so that transgenic hCN1 was secreted into the serum and liquor in vivo as in humans.

The transgene was introduced by pronuclear injection into fertilized eggs derived from heterozygous db/db mice. This mouse line carries a point mutation within the leptin receptor that leads to its functional inactivation resulting in hyperphagia after weaning, obesity, hyperinsulinemia, and elevated plasma glucose levels (type 2 diabetes model) from week 8. The BKS.Cg-m^+/+ Lepr db/J mouse has been shown to be susceptible to diabetic nephropathy (17). Mice were genotyped for the transgene hCN1 as well as the leptin receptor mutation (Lepr) by PCR from genomic DNA preparations of tail biopsies. Of 70 mice, 15 mice were transgenic for hCN1 (21%). All of the results described below were reproduced in at least four different founder lines. A stable line with robust CN1 expression was established from heterozygous db/wt hCN1 transgenic founder mice by crossbreeding with nontransgenic db/wt littermates (line 13).

Messenger RNA was isolated from liver and kidney of hCN1 transgenic db/db mice and from nontransgenic littermates. Nonsaturated RT-PCR analysis using primers specific for transgenic hCN1 demonstrated expression of transgenic mRNA within liver but not kidney (Fig. 2A, transgenic hCN1). TTP-hCN1 plasmid was used as positive control. Expression of endogenous cytosolic CN1 was detected in kidney, as described previously (Fig. 2A, endogenous mCN1). Gel electrophoresis of total RNA as well as nonsaturating RT-PCR specific for endogenous β-actin was used to control for equal loading. Negative controls omitting reverse transcription (Fig. 2, RT−) or omitting mRNA (Fig. 2, water only) were always negative.

To verify that the transgenic hCN1 is efficiently secreted into the serum, 10 μl of serum derived from hCN1 transgenic (db/db CN1+) or nontransgenic (db/db water) controls were subjected to SDS-PAGE and subsequently immunoblotted with various anti-sera directed against CN1. No CN1 protein was detected in nontransgenic controls, as described previously (7). Significant amounts of CN1 protein were detected in hCN1 transgenic animals (arrow).

To verify the functional effects of hCN1 overexpression, l-carnosine levels were measured in EDTA plasma obtained at the beginning of the light cycle of transgenic and control db/db mice (Fig. 2C) at 12 weeks of age. In control mice, endogenous l-carnosine levels averaged ∼1.2 ± 0.5 μmol/l. In hCN1 transgenic mice, l-carnosine levels were significantly lower (P < 0.01), most values below the detection level of 0.02 μmol/l (Fig. 2C, db/db CN1+).

A second experimental group of nontransgenic db/db mice was supplemented with l-carnosine in the drinking water. l-carnosine plasma levels were significantly elevated to 6 ± 1.1 μmol/l in this group (n = 6 for each group) (Fig. 2C, db/db carnosine).

Normal db/db mice (db/db water) developed persistent hyperglycemia between weeks 8 and 10 (Fig. 3A, db/db water), as described (18). hCN1 transgenic db/db mice (db/db CN1+) developed hyperglycemia earlier (weeks 6–7) and reached significantly higher FPG levels than the db/db water group. Conversely, FPG levels were significantly lower in l-carnosine–supplemented mice (Fig. 3A, db/db l-carnosine), and the onset of diabetes was delayed until weeks 17–20. Similarly, A1C levels were elevated in hCN1 transgenic mice (db/db CN1+) and significantly improved in l-carnosine–supplemented mice (db/db carnosine) compared with wild-type diabetic db/db mice (Fig. 3B, db/db water).

Obesity is important for the pathogenesis of altered glucose metabolism in the db/db mouse model. As expected, total body weights increased in parallel in all experimental diabetic groups until week 8. From week 15 onwards, hCN1 transgenic mice started to lose weight until they reached the weight of nondiabetic db/wt control mice by week 22 (Fig. 3C, db/db CN1+). This progressive wasting ultimately led to the death of the mice between 24 and 28 weeks of age, as verified by autopsy; four mice died of dehydration and one of an intra-abdominal abscess. The observation period of this study was therefore limited to 23 weeks (Supplementary Fig. 1 [available in an online appendix at http://dx.doi.org/10.2337/db07-0177]). No ketones were detected within spot-urine samples, and no significant differences in food or water intake were observed until week 20 (not shown). Significant glucosuria of ∼1.7 g/day was observed in the db/db CN1+ group (Fig. 3D).

l-carnosine–supplemented mice continued to gain weight beyond week 15 until the end of the observation period (Fig. 3C, db/db carnosine, week 22), while diabetic control mice (db/db water) reached intermediate body weights by week 15. This data ruled out that glucose metabolism was aggravated in hCN1 transgenic db/db mice as a consequence of increased body mass.

Compared with diabetic controls (db/db water), insulin levels were decreased in hCN1 transgenic mice despite higher FPG levels (Fig. 4A). The opposite pattern was observed in l-carnosine–supplemented mice (db/db l-carnosine), where serum insulin levels were significantly increased with lower FPG levels compared with those of diabetic controls (db/db water). These results suggested that glucose metabolism was primarily affected by altered insulin secretion in hCN1 transgenic and l-carnosine–supplemented mice.

To validate these findings, insulin sensitivity tests were performed (Fig. 4B, age 12 weeks). FPG levels fell proportionally in all experimental groups (20% reduction after 30 min, 30% total reduction after 1 h), indicating that peripheral insulin sensitivity is similar within the experimental groups. Glucose tolerance tests were also performed (Fig. 4C). A similar rise in FPG levels was observed in l-carnosine–supplemented mice (db/db l-carnosine) and diabetic control mice (db/db water). FPG levels in hCN1 transgenic mice (db/db CN1+) exceeded the range of the glucose test (600 mg/dl). In summary, a direct correlation of l-carnosine serum levels with insulin levels was detected in our experimental mice, whereas peripheral insulin sensitivity was unaltered.

To investigate the mechanism by which l-carnosine serum levels affect insulin secretion, pancreatic glands were removed from the experimental animals at 24 weeks of age, paraffin embedded, and subjected to histological analysis (Fig. 5A–E). Islet size was determined by morphometry (Fig. 5E). Overall, pancreatic islets were enlarged in diabetic mice compared with those in controls (db/wt). A striking difference in pancreatic islet size was observed in l-carnosine–supplemented animals (Fig. 5D and E). Hyperplastic islets of l-carnosine–supplemented mice were predominantly composed of insulin-producing β-cells, as demonstrated by immunohistological anti-insulin staining (Fig. 5F).

Next, it was tested whether l-carnosine exerts a direct effect on insulin secretion or proliferation of β-cells in vitro. For this purpose, INS-1e cells were exposed to varying concentrations of l-carnosine. INS-1e cells are derived from β-cells and secrete insulin in a highly regulated glucose-dependent manner (16). In the absence of l-carnosine, insulin secretion increased in the presence of increasing concentrations of l-glucose, as expected (Fig. 5G, open columns). Exposure to various concentrations of l-carnosine did not significantly affect glucose-dependent insulin secretion in INS-1e cells (gray, striped, and filled columns). Statistically significant increments of insulin secretion were detected for higher concentrations of l-carnosine in the absence of l-glucose (Fig. 5G, 0G, 1 mM and 20 mM l-carnosine). However, no significant influence of lower near-physiological levels of l-carnosine on insulin secretion was detected in the absence of l-glucose.

In a second set of experiments, INS-1e cells were cultured at low densities in the presence of varying concentrations of l-carnosine for 3 days. The percentage of BrdU-positive cells in three independent experiments is shown in Fig. 5H. A dose-dependent increase in BrdU-positive cells was observed with increasing l-carnosine concentrations.

These data suggested that insulin secretion was preserved in l-carnosine–supplemented mice because of an increased or preserved mass of β-cells within the pancreas. No significant stimulatory effect of l-carnosine on insulin secretion was observed in vitro, at least near physiological levels.

Because CN1 has been identified as a risk factor for diabetic nephropathy in diabetic humans, it was investigated whether l-carnosine serum levels affect the kidney in our mouse model. Serum creatinine levels were lowest in hCN1 transgenic animals (db/db CN1+) and correlated with the extent of polyuria or glomerular filtration rate as a result of poor glycemic control (Fig. 6A). Serum urea levels showed significantly elevated levels in hCN1 transgenic mice as a consequence of impaired renal function or dehydration (Fig. 6B). Albumin-to-creatinine ratios were significantly increased in all diabetic experimental groups (db/db CN1+, db/db water, and db/db carnosine) compared with nondiabetic controls (db/wt), indicating the presence of diabetic nephropathy in all diabetic groups (Fig. 6E). However, no significant differences between l-carnosine–supplemented and CN1 transgenic animals were observed at 15 or 20 weeks of age.

Renal hypertrophy is an early sign of diabetic nephropathy. At 23 weeks of age, absolute kidney weights were significantly increased in all diabetic animals compared with those of nondiabetic controls (Fig. 6C, db/wt). However, no significant differences were observed among the diabetic experimental groups. The ratio of kidney weight over total body weight (Fig. 6D) suggested significant renal hypertrophy only in transgenic diabetic mice (db/db CN1+). However, the ratio failed to demonstrate renal hypertrophy in normal diabetic db/db mice (db/db water). It was concluded that kidney-to-body weight ratios are not a suitable parameter in this model because body weights were significantly affected by modulation of l-carnosine serum levels, as shown in Fig. 3C.

Finally, mesangial expansion as a sign for diabetic nephropathy was assessed. A representative example for “no mesangial expansion” is shown in Fig. 6G (control) and for “significant mesangial expansion” in Fig. 6H (mes. expansion). In agreement with the results described above, mesangial expansion was present in all three diabetic groups (Fig. 6F) compared with nondiabetic db/wt mice, but no significant differences between any of the diabetic groups could be detected in this mouse model.

In this study, an association of CN1 activity with diabetic disease, previously identified in human diabetic patients (6), was confirmed in a type 2 diabetes mouse model. This association was mediated by the substrate of CN1, l-carnosine. Interestingly, in the type 2 diabetes db/db mouse model described here, l-carnosine primarily affected glucose metabolism. l-carnosine serum levels did not significantly affect diabetic nephropathy, as suggested by an initial genetic study in human patients (6).

The results of this study contribute significantly to our understanding of the genetic association in humans. In humans, the “high-risk” group of patients carried more efficient signal peptides within the hCN1 gene (i.e., with more than five leucines), resulting in higher CN1 activity and lower l-carnosine levels with the serum (6). Consistent with the human data, hCN1 transgenic diabetic mice with lower l-carnosine serum levels were affected by more severe diabetic disease in this study. In a different experimental approach in this study, l-carnosine levels were artificially raised in db/db mice and associated with improved glucose metabolism. This indicates that the human genetic association between CN1 activity and diabetes complications is a consequence of a primary effect on glucose metabolism and that hCN1 is a modifier gene for diabetic disease. A direct effect of the quality of control on glucose metabolism and late diabetes complications has been established in human patients (19,20). In the initial study by Janssen et al. (6), metabolic control of diabetic disease was evaluated during the later stages of diabetes (i.e., after >15 years). It is possible that the higher prevalence of diabetes complications in the high-risk group of human diabetic patients reflected poor metabolic control of diabetes in this group (6).

Further support that hCN1 activity modifies glucose metabolism in human patients comes from a study that performed a genome scan for loci associated with increased FPG levels in families afflicted in part by obesity (21). Significant linkages were found with a trait located at chromosome 18q22 (logarithm of odds 4.4–6.6), the region where CNDP1 (hCN1) is located with ∼150 other genes. No correlation of the identified locus and BMI was observed by Li et al. (21), supporting that decreased body weights in hCN1 transgenic mice were primarily a consequence of calorie loss in the urine in this study.

In humans, the trait on locus 18q22 only affected the FPG levels, not the prevalence of overt diabetic disease, as shown by Li et al. (21). Likewise, l-carnosine levels only affected the extent and severity diabetic disease in our type 2 diabetic mouse model—overt diabetes with FPG levels >200 mg/dl manifested significantly earlier in hCN1 transgenic mice or later in l-carnosine–supplemented mice, but eventually all mice became diabetic. Metabolic control in l-carnosine–supplemented mice eventually became worse, similar to diabetic controls, indicating that l-carnosine is not sufficient to prevent overt diabetes.

Our results are also in agreement with the initial finding by Janssen et al. (6) that the risk for poor control of diabetes (reflected by a higher risk for diabetic nephropathy) is inherited in an autosomal-dominant fashion. A single CNDP1 allele with an efficient leader peptide for secretion (i.e., with more than five leucins) was sufficient to decrease serum l-carnosine levels, which were associated with poor diabetes control.

In this study, l-carnosine levels primarily affected serum insulin secretion, as indicated by the inverse correlation of body weights to FPG levels, insulin levels, and functional tests of glucose metabolism. Our data suggest that insulin secretion is preserved because of an increased β-cell mass as a consequence of increased l-carnosine serum levels. In addition to that, a discrete effect of l-carnosine on glucose-independent insulin secretion was observed in vitro, which can be expected from an amino acid or a small dipeptide such as l-carnosine.

In summary, CN1 is a potential target for pharmacological interventions aimed at manipulating glucose metabolism. Since only approximately one-third of the entire human population is homozygous for the “low-risk” CNDP1 allele (5,6,22), this approach has the potential for a broad applicability to diabetic patients.

This work was supported by a grant from the Else-Kröner Fresenius Stiftung (P05/06/A75/05) and a grant from the German Research Foundation (Deutsche Forschungsgemeinschaft [DFG]) (MO1082/1-2) to M.J.M. M.J.M., W.K., and F.vdW.^† are members of the Forschergruppe 406 of the German Research Foundation (DFG).

We thank Prof. Jens Bruning, Chief of the Institute for Genetics, University Cologne, Cologne, Germany, for his advice and support. We also thank Dr. Rüdiger Waldherr, Prof. of Pathology, University of Heidelberg, Heidelberg, Germany, for assessing the renal biopsies.

This article is dedicated to Prof. Fokko Johannes van der Woude, who passed away on 4 December 2006.