Carnosine synthase deficiency is compatible with normal skeletal muscle and olfactory function but causes reduced olfactory sensitivity in aging mice

Carnosine (b-alanyl-L-histidine) and anserine (b-alanyl-3methyl-L-histidine) are abundant peptides in the nervous system and skeletal muscle of many vertebrates. Many in vitro and in vivo studies demonstrated that exogenously added carnosine can improve muscle contraction, has antioxidant activity, and can quench various reactive aldehydes. Some of these functions likely contribute to the proposed anti-aging activity of carnosine. However, the physiological role of carnosine and related histidine-containing dipeptides (HCDs) is not clear. In this study, we generated amouse line deficient in carnosine synthase (Carns1). HCDs were undetectable in the primary olfactory system and skeletal muscle of Carns1-deficient mice. Skeletal muscle contraction in these mice, however, was unaltered, and there was no evidence for reduced pH-buffering capacity in the skeletal muscle. Olfactory tests did not reveal any deterioration in 8month-old mice lacking carnosine. In contrast, aging (18–24month-old) Carns1-deficient mice exhibited olfactory sensitivity impairments that correlated with an age-dependent reduction in the number of olfactory receptor neurons. Whereas we found no evidence for elevated levels of lipoxidation and glycation end products in the primary olfactory system, protein carbonylation was increased in the olfactory bulb of aged Carns1deficient mice. Taken together, these results suggest that carnosine in the olfactory system is not essential for information processing in the olfactory signaling pathway but does have a role in the long-term protection of olfactory receptor neurons, possibly through its antioxidant activity.

Carnosine (b-alanyl-L-histidine) and anserine (b-alanyl-3methyl-L-histidine) are abundant peptides in the nervous system and skeletal muscle of many vertebrates. Many in vitro and in vivo studies demonstrated that exogenously added carnosine can improve muscle contraction, has antioxidant activity, and can quench various reactive aldehydes. Some of these functions likely contribute to the proposed anti-aging activity of carnosine. However, the physiological role of carnosine and related histidine-containing dipeptides (HCDs) is not clear. In this study, we generated a mouse line deficient in carnosine synthase (Carns1). HCDs were undetectable in the primary olfactory system and skeletal muscle of Carns1-deficient mice. Skeletal muscle contraction in these mice, however, was unaltered, and there was no evidence for reduced pH-buffering capacity in the skeletal muscle. Olfactory tests did not reveal any deterioration in 8month-old mice lacking carnosine. In contrast, aging (18-24month-old) Carns1-deficient mice exhibited olfactory sensitivity impairments that correlated with an age-dependent reduction in the number of olfactory receptor neurons. Whereas we found no evidence for elevated levels of lipoxidation and glycation end products in the primary olfactory system, protein carbonylation was increased in the olfactory bulb of aged Carns1deficient mice. Taken together, these results suggest that carnosine in the olfactory system is not essential for information processing in the olfactory signaling pathway but does have a role in the long-term protection of olfactory receptor neurons, possibly through its antioxidant activity.
Carnosine (b-alanyl-L-histidine), the first peptide ever isolated from an animal tissue (1), and related peptides (histidinecontaining dipeptides (HCDs)) are present at exceptional high (millimolar) concentrations in skeletal muscle and the primary olfactory system (2, 3) (see Boldyrev et al. (4) for a comprehensive review). HCDs are also present in various other tissues although at significantly lower concentrations (4,5). Carnosine is synthesized by a cytosolic amino acid ligase, carnosine synthase (Carns1; also known as ATP-grasp domain-containing protein 1, ATPGD1; EC 6.3.2.11) (6). In the presence of GABA (i.e. in the central nervous system), Carns1 also catalyzes the synthesis of homocarnosine (g-aminobutyryl-L-histidine). Carnosine can be modified by N-methyltransferases (7,8) in a tissue-and species-specific manner, forming ophidine (b-alanyl-1-methyl-L-histidine) or anserine (b-alanyl-3-methyl-Lhistidine). The latter is the most abundant HCD in skeletal muscle of mice, rats, and other rodents (4). The highest concentrations of HCDs (.10 g/kg, i.e. .40 mM) have been found in muscles from several birds, especially some Galliformes families, some teleostei (e.g. marlin) (mainly anserine), and cetacea species (mainly ophidine) (4). Carnosine and anserine are degraded by serum carnosinase 1 (EC 3.4.13.20) (9). A homologous enzyme, cytosolic nonspecific dipeptidase or carnosinase 2 (EC 3.4.13.18), exhibiting a much broader substrate specificity also hydrolyzes anserine and carnosine in vitro, but it is not clear to what extent this enzyme is actually involved in carnosine turnover (10,11). In the olfactory system, carnosine is most likely exclusively synthesized in receptor neurons of the olfactory epithelium (OE) (12). Carnosine co-localizes with glutamate in the presynapse of synapses between olfactory receptor neurons and mitral cells (13), and its release from synaptosomes prepared from the olfactory bulb is calcium-dependent and stimulated by depolarization (14). It was therefore proposed that carnosine may act as neurotransmitter or neuromodulator in the primary olfactory system (15). In line with this, carnosine exhibits excitatory activity on mitral cells in the OB (16). In contrast, however, Harding and O'Fallon (17) did not find evidence for the presence of carnosine in synaptic vesicles. It was also proposed that carnosine (and homocarnosine) may be involved in neuron-to-glia signaling (18).
In contrast to the primary olfactory system, carnosine appears to be synthesized by glial cells in other parts of the nervous system (19). There mainly (if not exclusively) oligodendrocytes are responsible for its synthesis (20). The glutamate receptor activation-dependent release of carnosine from oligodendrocytes (21) suggests a possible role of the peptide in glianeuron interaction (22). In heart and skeletal muscle, carnosine affects muscle function via improving pH buffering and/or excitation-contraction coupling by local calcium recruitment (23,24). Besides these potential roles in neuron-glia signaling and muscle excitation-contraction coupling, other less specific functions of carnosine have been documented (4). Particularly, carnosine has antioxidative activity (25) and inhibits protein carbonylation and glycation as well as the formation of advanced glycation end products (AGEs) and advanced lipoxidation end products (ALEs) (4). These processes are likely involved in the anti-aging activity of carnosine (26). Whether the beneficial effects of exogenously supplied carnosine, however, reflect the physiological function of the endogenously synthesized peptide is unclear. To be able to further explore the physiological role of HCDs, we generated Carns1-deficient (Carns1(2/2)) mice. We focused our analysis on the two tissues/systems with by far the highest HCD concentration: skeletal muscle and primary olfactory system. Aging Carns1(2/2) mice exhibited diminished olfactory sensitivity and loss of olfactory receptor neurons, suggesting a role of carnosine in long-term survival of olfactory receptor neurons (ORN).

Carns1(2/2) mice lack carnosine and anserine
Mice deficient in Carns1 were generated by homologous recombination in HM-1 embryonic stem cells (Fig. 1). The targeting construct introduced loxP sites within intron 2 and in exon 10, resulting in deletion of the complete coding region of Carns1 after cre recombinase action (Fig. 1A). Correct targeting was confirmed by Southern blot analysis (Fig. 1B) and PCR (data not shown), and an embryonic stem (ES) cell clone showing homologous recombination in both alleles (f/f in Fig. 1B) was used for generating chimeric and subsequently Carns1 (2/2) mice. Absence of the Carns1 mRNA in Carns1(2/2) mice was confirmed by Northern blot analysis (Fig. 1C). Realtime RT-PCR showed ;50% reduction in the expression of Carns1 in heterozygous mice (Fig. 1D). Western blot analysis using a newly generated Carns1 antiserum confirmed absence of the enzyme in Carns1(2/2) and reduced levels in heterozygous mice compared with WT littermates (Fig. 1E). As expected, carnosine and anserine were undetectable in brain and skeletal muscle of Carns1(2/2) mice (Fig. 2, A and B). Similar levels of the dipeptides were present in Carns1(1/1) and Carns1(1/2) mice (Fig. 2C). This indicates that carnosine synthesis is not limited by Carns1 expression and is in line with previous observations suggesting that b-alanine availability limits carnosine synthesis in vivo, at least in humans and mice (27,28). Absence of carnosine in the OE was further confirmed by immunofluorescence using a carnosine-specific antibody (Fig. 2D). Loss of carnosine-like immunoreactivity in Carns1(2/2) OE was not due to absence of ORN, as indicated by staining of olfactory marker protein (OMP) (Fig. 2D). This confirmed the specificity of the peptide antibody and absence of carnosine in OE of Carns1(2/2) mice. These mice had normal body weight, did not show any obvious abnormal behavior, and had an unaltered survival rate (see the Kaplan-Meier survival curve in Fig. 1F). Because the oldest mice were killed at 24 months, a decrease in the maximal life span in the absence of carnosine cannot be excluded at present. We further focused our analyses on the two tissues with the highest HCD concentration, skeletal muscle and the primary olfactory system.

Absence of carnosine and anserine does not affect excitationcontraction coupling in skeletal muscle
Muscle strength measurements of 3-and 18-month-old male mice using the grip strength test showed an age-dependent decline but no significant effect of the genotype (Fig. 3A). Mean muscle weight and length were normal in Carns1(2/2) at 6-8 months of age (Fig. 3, B and C), although weight and length of the extensor digitorum longus (EDL) muscle (but not soleus (SOL) muscle) showed a higher variance (weight: p = 0.0012 (F-test), length: p = 0.0176 (F-test)). The reason for this difference is unclear. Histological analyses of skeletal muscle did not reveal obvious differences between genotypes with respect to fiber diameter or the number and localization of nuclei (data not shown).
Contractile properties of skeletal muscles were examined using the EDL and SOL muscles isolated from 6-8-month-old mice. No significant shift of the relationship between stimulation frequency and relative force was observed for EDL muscles from Carns1(2/2) mice and WT controls (Fig. 3D). Similar results were obtained with SOL muscle (data not shown). The fatigue index in long fatigue runs using EDL muscle was virtually identical for Carns1(2/2) and Carns1(1/1) mice (Fig.  3E). To examine whether carnosine and anserine contribute substantially to the buffering capacity in skeletal muscle, tetanic stimulation was repeated in the presence of cinnamate (4 mM). It inhibits lactate transport and therefore causes a stronger reduction in intracellular pH during fatigue runs and intensifies the decline in peak tetanic tension (29). We therefore expected a faster decline of force in the case of a stronger acidification in Carns1(2/2) muscle in the fatigue run. Peak tetanic tension was clearly reduced in the presence of cinnamate, as anticipated (Fig. 3E). However, the absence of a significant difference between genotypes, indicates that the buffering capacity of carnosine and anserine in skeletal muscle is negligible or can be efficiently substituted by other compounds.
Absence of carnosine does not impair olfactory function in young Carns1(2/2) mice but does in aging Carns1(2/2) mice Olfaction in Carns1(2/2) mice and controls was examined in young (8 months) and old (18 months) mice using the buried food, preference, and habituation/dishabituation test (Fig. 4). Time to find a buried cookie was not significant different between genotypes in young (one-way ANOVA, F (2,24) = 0.646, p = 0.533) or old mice (F (2,27) = 1.067, p = 0.358) (Fig. 4A). In the preference test, which was only performed with 18-month-old animals, mice could discriminate between the two test odors, as indicated by the significantly longer inspection time for peanut butter (preferred odor) compared with the nonpreferred odor 2-methylbutyric acid (2-MB), irrespective of the genotype (one-tailed t test, p = 0.007 for Carns1(1/1), p = 0.025 for Carns1(1/2), p = 0.001 for Carns1(2/2)) ( Fig. 4B). In addition, we used the habituation-dishabituation test to examine whether Carns1(2/2) mice (18 months) were able to discriminate between different odors assumed to be neutral. This experiment confirmed normal olfactory function in Carns1 (2/2) mice when compared with WT controls (Fig. 4C). We observed, however, that 30% of the Carns1(2/2) mice apparently did not recognize the almond odor (i.e. they showed a reduced inspection time in trial 4 versus trial 3, whereas all control mice (WT and heterozygous) showed an increased inspection time). Therefore, to detect a minor impairment in olfactory function, sensitivity assays were performed using a modified habituation-dishabituation test (Fig. 5). At 8 months of age, identical behavioral detection thresholds were observed for Carns1(2/2) and Carns1(1/1) mice for all test substances (almond, vanilla, lemon). Old (18 months) and aged (24 months) Carns1(2/2) mice, however, exhibited a higher behavioral olfactory threshold compared with WT controls for one (almond: 10-fold increased detection threshold in old mice) or two (almond: 100-fold increased detection threshold; lemon: no detection threshold reached in aged mice) of the three compounds tested (Fig. 5). In summary, the results of the olfactory tests suggest that carnosine does not affect odor perception per se but affects long-term functional integrity of the olfactory system.
Age-dependent decline in the number of olfactory receptor neurons in Carns1(2/2) but not in WT mice The number of ORN in the main olfactory epithelium of endoturbinate I was determined in transverse sections of the nasal cavity obtained from 8-, 18-and 24-month-old mice (Fig.  6A). Sections were stained with antibodies against the ORN marker protein OMP (Fig. 6, B-G). Factorial ANOVA revealed Figure 1. Generation of Carns1(2/2) mice. A, a loxP site was introduced upstream of exon 3, and a loxP-and FRT-flanked neomycin-resistance cassette (neo(R)) was inserted downstream of exon 10. Two loxP sites were linked to MfeI restriction sites to allow for homologous recombination screening by Southern blot analysis. Mice containing loxP/FRT-flanked Carns1 alleles were crossed with deleter-Flp or pgk-cre mice to generate floxed or Carns1 knockout mice.
B, examples of Southern blotting results with selected ES clones showing WT (1/1), heterozygous (f/1), and homozygous floxed Carns1 (f/f) using the different probes indicated in A. The weak signal at 24.3 kb in the f/f cell line likely represents the WT allele of the feeder cells. C, Northern blot analysis of Carns1 expression in the OE. Note that the two separate signals around 6 and 4 kb likely belong to the same mRNA but are displaced by the 28S rRNA. D, quantitative real-time RT-PCR of Carns1 expression in different tissues in WT (1/1) and heterozygous Carns1(1/2) mice (Carns1 mRNA was undetectable in Carns1(2/2) mice) (n = 3-5). E, Western blot analysis of Carns1 in olfactory bulb and skeletal muscle (musculus gastrocnemius) using total tissue homogenates. Blots were reprobed with a-tubulin antibody to control for equal loading. F, Kaplan-Meier survival curve of male Carns1(2/2) (n = 8) and WT (1/1) littermates (n = 7) (mice from five litters). No significant difference by log rank test (x 2 = 0.171; critical value = 3.84). a significant effect of age (F (2,18) = 3.877; p = 0.039) and genotype 3 age interaction (F (2,18) = 8.177; p = 0.003) regarding the number of OMP-positive cells. Post hoc Newman-Keuls test indicated a significant reduction of OMP-positive cells in 24month-old Carns1(2/2) compared with 8-month-old Carns1 (2/2) (p = 0.003) as well as 24-month-old controls (p = 0.036) (Fig. 6H). Thus, whereas the number of mature (OMP-positive) ORN in adult control mice did not change significantly with age, which is in agreement with previous reports (30, 31), we observed a progressive loss of mature ORN in Carns1(2/2) mice.
Apoptotic cells in the OE (determined by TUNEL assay) of 24-month-old mice were rarely detectable in both WT (in line with a previous report (32)) and Carns1(2/2) mice ( Fig. 6, I and J). Moreover, these apoptotic cells were only present in the basal layer of the OE. There was no increase in the number of TUNEL-positive cells in Carns1(2/2) mice, neither in the basal nor in the apical cell layer (Fig. 6K). Ki67 staining did not reveal a decrease in proliferating cells in Carns1(2/2) OE, but also their number was very low in both genotypes (data not shown). Loss of ORN could potentially result from a depletion of the pool of proliferating precursor cells. However, the number of Pax6-positive basal cells in the OE was not significant different between genotypes (Fig. 6, L-N). Carnosine-like immunoreactivity had been observed in the rostral migratory stream, suggesting that carnosine may play a role in migration of neural precursors (33,34). However, size of the olfactory bulb (data not shown) and the number of granule and mitral cells in the mitral cell layer were not significantly changed in Carns1(2/2) mice, suggesting no impairment of neural precursor migration into the olfactory bulb ( Fig. 6, O-Q).
Carns1(2/2) mice show a tendency toward higher protein carbonylation, but ALEs and AGEs were not increased in the primary olfactory system Carnosine has been shown to act as an antioxidant and to reduce protein carbonylation as well as lipoxidation and formation of ALE (4). To examine the possibility of a corresponding reversed effect in the absence of carnosine, protein-carbonyl content (using anti-DNP antibody after treating tissue sections with DNPH) and ALE protein adducts (4-hydroxy-2-nonenal (HNE) and malondialdehyde (MDA)) were determined in OE of 24-month-old mice by immunofluorescence. There were no significant changes in Carns1(2/2) mice when compared with WT controls (Fig. 7, A-I). However, Western blot analysis of the OB showed higher protein-carbonyl content in Carns1 (2/2) mice (Fig. 7, J and K). Total DNP signal was significantly increased in Carns1(2/2) OB (p = 0.008, t test), which was also the case when the most prominent band at 55 kDa was quantified separately (Fig. 7K). Attempts to prove that this band indicates increased carbonylation of tubulin and/or vimentin were unsuccessful because DNP probing of immunoprecipitated tubulin or vimentin resulted in a significant background Extracts corresponding to 0.1 mg of protein of the original tissue homogenate were separated by TLC together with increasing amounts of carnosine and anserine on separate lanes (not shown) and stained with ninhydrin. B, HPTLC analysis of skeletal muscle (musculus gastrocnemius) was done as described in A. C, peptides were quantified by densitometry. Bars and error bars indicate mean and 95% CI (n = 3 for olfactory bulb; n = 4 for skeletal muscle) (nd, not detectable). D, immunofluorescence staining of consecutive sections through the OE indicates the presence of Carns1 and carnosine in the OMP-positive olfactory receptor neurons in WT mice and absence of Carns1 and carnosine in Carns1(2/2) mice. Nuclei were stained with DAPI. Scale bar, 100 mm.
preventing quantification of carbonyl-tubulin or vimentin (data not shown).
In contrast to this partial increase of carbonylated proteins, adducts of the lipoxidation end products HNE and MDA, examined by Western and dot blot, were not significantly increased in the carnosine-deficient OB (Fig. 7, L-N). Similarly, we found no evidence for increased levels of AGEs in Carns1 (2/2) OB by Western blot analysis (Fig. 7O). Corresponding Western and dot blot experiments shown for the OB could not be used to examine OE, because different degrees of blood contaminations (which could not be avoided) caused strong varia-tions in the Western blotting signals that did not allow a reliable quantification (data not shown).

Discussion
Most in vitro studies examining the role of carnosine as an antioxidant, tumor growth inhibitor, anti-glycation, or antiaging compound elicited a significant effect at relatively high carnosine concentrations of 10-20 mM or above (25,(35)(36)(37). Under physiological conditions, a concentration on the same order of magnitude is probably only reached in skeletal muscle and ORN (3,4). Therefore, the possibility to draw conclusions from the above-mentioned studies on the physiological functions of HCDs is limited. Obviously, with a pK a value of about 6.8 (for the imidazole ring), carnosine and related HCDs are efficient buffering substances at cytosolic pH. However, the HCD concentration in most tissues is too low to contribute significantly to the total buffer capacity, with the possible exception of skeletal muscle. Mannion et al. (38) estimated that in human skeletal muscle, carnosine accounts for about 7% of the intracellular buffer capacity, although according to other calculations, the actual contribution might be higher (39). However, whereas b-alanine supply leads to an increase in carnosine concentration that correlated with an improved exercise performance (40), there seems to be no direct evidence for alterations (1/2), and Carns1(1/1) mice (n = 4-8). Main effects ANOVA revealed a significant effect of age (F (1,33) = 11.046; p = 0.0022) but not genotype. B-E, EDL muscle size, contractility, and fatigue were examined with 6-8-month-old mice. Wet weight (B) and length of EDL muscle (C) were not significantly different between genotypes (n = 7-8). D, relationship between relative force (percentage of maximal force) and stimulation frequency for EDL muscle from WT (1/1) and Carns1(2/2) mice (n = 5). No significant difference between genotypes was observed. E, changes of the fatigue index in EDL muscle during long fatigue run in the absence or presence of 4 mM cinnamate (n = 4) revealed a significant effect of cinnamate treatment (p , 0.05; one-way ANOVA with post hoc Bonferroni test), but no significant effect of genotype. Data are shown with mean and 95% CI (in E, symbols indicate the mean, and only the upper or lower half of the 95% CI is shown for every 60 s). . Olfactory function test. A, buried food test was performed with young (7-8 months) and old (18 months) WT (1/1), heterozygous (1/2), and homozygous Carns1(2/2) mice. The time needed to find a buried cookie was determined (cutoff time 15 min). All except two 18-month-old Carns1(1/2) mice found the cookie within 15 min. One-way ANOVA did not reveal significant differences between genotypes (n = 9 for young and n = 10 for old mice). B, the preference test (18-month-old mice) showed a significant difference of inspection times between the preferred (peanut butter) and the nonpreferred odor (2-MB) for all genotypes (t test; *, p , 0.05) (n = 8 for Carns1(1/1), n = 9 for Carns1(1/2), n = 7 for Carns1(2/2)). C, habituationdishabituation tests (at 18 months) showed a significant increase in inspection time when a new odor was presented, indicating that mice of all genotypes could discriminate different odors. Values for trials 3 and 4 as well as trials 6 and 7 were tested for significant differences using Rodger's method (significant differences are indicated by asterisks) (n = 8 for Carns1(1/1), n = 7 for Carns1(1/2), n = 10 for Carns1(2/2)). All data are shown with mean and 95% CI. Data for male and female mice were combined.
in intracellular pH in post-exercise muscle after b-alanine supplementation (39). In line with this, we found that absence of HCDs did not accelerate muscle fatigue under conditions that cause a faster acidification. In principle, because carnosine/anserine are absent from Carns1(2/2) mice throughout postnatal development, loss of the HCD buffer could have been compensated by "up-regulation" of other buffers. An inducible Carns1 knockout or feeding Carns1(2/2) mice carnosine (or b-alanine) during postnatal development could be a way in future experiments to address this issue. An alternative mechanism of carnosine's action in skeletal muscle is that it may act as a shuttle, facilitating calcium supply and thereby improving excitation-contraction coupling (23,41). Accordingly, feeding mice with a high dose of b-alanine that causes a nearly 100% increase in HCDs in EDL muscle resulted in a significant leftward shift of the force-frequency relation (24). In contrast, however, we did not observe a corresponding significant shift toward higher frequencies in the force-frequency relation in Carns1(2/2) EDL or SOL muscle. Type II fibers contain larger amounts of HCD compared with type I fibers (42). Similar to other mammals, the fiber composition in SOL (.30% type I) and EDL (.90% type II) muscle differ considerably in mice (43). The absence of a significant influence of the Carns1 geno-type in both muscles therefore suggests that both fiber types were unaffected by carnosine/anserine deficiency. Thus, whereas elevated levels of carnosine can improve excitation-contraction coupling, it does not seem to play an essential role in this process. However, we did not examine muscle contractility and fatigue in older animals. Similar to the situation in the olfactory system, muscle in aging Carns1(2/2) mice may also become more susceptible to protein carbonylation, which may also impair muscle function.
ORN are possibly the only neurons synthesizing carnosine (22). Carnosine has excitatory activity at mitral (and/or tufted) cells in the OB (16), and a role as neurotransmitter or neuromodulator in the primary olfactory system has been proposed (15), although some experimental data do not support this hypothesis (see Bonfanti et al. (22) and De Marchis et al. (44) for discussion). The normal olfactory behavior and sensitivity of 8month-old Carns1(2/2) mice suggest that carnosine does not play an essential role as a (co)transmitter for primary olfactory signal processing in the primary olfactory system. Because the olfactory discrimination task depends only on short-term memory, we cannot rule out the possibility that carnosine may play a role in long-term synaptic plasticity. On the other hand, long-term plasticity in the OB has been demonstrated for Figure 5. Olfaction sensitivity test. The behavioral detection threshold of WT (1/1) and Carns1(2/2) mice at 8, 18, and 24 months of age for different test substances (almond, vanilla, lemon; diluted 10 27 to 10 22 in water) was determined using a modified habituation-dishabituation protocol. The detection threshold (i.e. the test trial in which the inspection time increased significantly compared with the preceding trial for the first time) is indicated by black (1/1) and red (2/2) arrowheads, respectively. At 18 months, the behavioral olfactory threshold for one odor (almond) was increased 10-fold in Carns1(2/2) mice compared with WT controls (1/1). Detection of two odors (almond and lemon) was impaired in 24-month-old Carns1(2/2) mice (100-fold increased detection threshold for almond). The behavioral detection threshold for lemon odor was not reached in the case of 24-month-old Carns1(2/2) mice; therefore, no red arrowhead is shown (apparently, there is a trend toward longer inspection time). Data are shown with mean and 95% CI (8 months, n = 8; 18 months, n = 7-9; 24 months, n = 8 (2/2), n = 7 (1/1)). Data from male and female mice were combined. nd, not done.
Carnosine synthase-deficient mouse dendro-dendritic synapses between granule and mitral cells (45), but to our knowledge not for the glomeruli-localized (carnosine-positive) synapses between ORN and mitral cells. Nevertheless, partial loss of olfactory sensitivity in aging Carns1(2/2) mice in association with a parallel loss of ORN implies that carnosine is required for long-term maintenance of the functional integrity of the OE. The temporal correlation suggests that loss of ORN is responsible for the increased olfactory behavioral threshold. Reduced sensitivity for some but not all odors tested may be explained by the fact that ORN expressing different receptor genes are lost in aging OE at different rates (46,47). Proliferating basal cells of the OE continuously generate new ORN precursor cells that are able to replace degenerated ORN throughout life, although the turnover rate declines with age (32). The progressive loss of ORN in Carns1 (2/2) mice suggests an imbalance between formation of new immature ORN and loss of mature ORN. Although it appears likely that loss of carnosine affects survival of ORN (i.e. the only cells in the OE that synthesize and contain carnosine), we could not trace their loss to a higher rate of apoptosis in the ORN cell layer. It is known that the apoptosis rate of mature ORN in old OE is very low, and apoptotic cells are mainly found in the basal layer of the OE, containing less mature ORN and their precursors (32). In line with this, we found apoptotic cells only in the basal layer of the OE but not in the apical layer containing mature ORN. Possibly the increase in death rate of Carns1 (2/2) ORN was too low to allow detection of the apoptotic cells with certainty. Loss of ORN was significant only in aged mice, which would be in line with a long-lasting effect of carnosine deficiency affecting primarily long-lived ORN, which are likely more frequent in old OE, as the turnover of ORN appears to be much slower in old compared with young OE (32,48). Because carnosine is not synthesized in the proliferating basal cells, it seems unlikely that Carns1 deficiency directly affects these cells, especially in view of the normal level of Pax6-positive precursor cells in Carns1(2/2) OE.
A property of carnosine that could potentially support survival of ORN is, of course, the general antioxidant activity that has been demonstrated in various studies (25,(49)(50)(51), although the in vivo antioxidative activity of carnosine has been questioned by others (52). Protein carbonylation can disturb functionality of proteins, cause intermolecular cross-linking, and alter protein turnover (53). Besides preventing carbonylation and other oxidative damage of proteins, carnosine may also inhibit cross-linking by reacting with carbonylated proteins (54). Carbonylation of tubulin can impair its polymerization and favor depolymerization of microtubules (55,56). Thus, it is possible that the increase of protein carbonylation in Carns1(2/2) mice may impair the functional properties of ORN axons in the OB, although further studies are necessary to solve this question.
One may presume that increased carbonyl-protein content is a direct consequence of the absence of carnosine's scavenging activity toward methylglyoxal and other carbonyls (57). However, it is also possible that Carns1 deficiency causes increased production of reactive carbonyls or impairs degradation of carbonylated proteins. The absence of carnosine may also alter gene expression as demonstrated by its effect on histone acetylation (58,59). Determination of whether increased protein carbonylation and loss of ORN may be secondary to altered gene expression will require further studies.
We did not observe a significant increase of carbonyl-protein (or HNE and MDA adducts) in the OE. Other antioxidants are thus likely to play a more important role and/or may to a large extent compensate for the loss of carnosine, at least under nonpathological conditions. Although neurons in general contain low levels of GSH (60), ORN contain relatively large amounts of GSH (61), and ALEs like HNE are efficiently quenched by Figure 7. Partial increase of carbonyl-protein content in the OB but normal levels of lipoxidation end products in Carns1(2/2) mice. Sections of OE from 24-month-old WT (A) and Carns1(2/2) mice (B) were treated with DNPH, and DNP adducts were detected by immunofluorescence. Quantification of signals in the ORN layer did not show significant differences between genotypes (C). Sections from the same mice were also stained with antibodies against the ALEs HNE (D and E) and MDA (G and H). Quantification of signal intensities in the ORN layer did not show significant differences between genotypes (F and I). Scale bars, 50 mm. J, carbonyl-protein content in the OB in 22-24-month-old mice was determined by Western blotting of DNPH-treated homogenates. Blots were reprobed with a-tubulin antibody. K, densitometric quantification indicated significant increase in protein carbonylation in Carns1(2/2) mice (area between 130 and 43 kDa; p = 0.008, t test), DNP signal at 55 kDa (p = 0.011), and total DNP signal without 55 kDa band (p = 0.020). L, Western blot analysis of MDA protein adducts in OB. M, Western blot analysis of HNE protein adducts in OB. N, dot blot of MDA and HNE protein adducts in OB. O, Western blot analysis of AGE protein adducts in OB. To control for equal loading, membranes were stained with Ponceau S (PS). Data are shown with mean and 95% CI (n = 4 (C, F, and I) or n = 7 (K and N)).
Carnosine synthase-deficient mouse GSH (see Esterbauer et al. (62) for a review). It remains to be determined whether carnosine deficiency may have a stronger impact under conditions of increased oxidative stress as found in various neurodegenerative diseases (63)(64)(65). This is suggested by the neuroprotective activity of carnosine in different neurodegenerative model systems (66).

Experimental procedures
Generation of Carns1(2/2) mice A conditional Carns1-knockout allele was generated by introducing loxP sites in intron 2 and downstream of the last exon (exon 10) of the Carns1 gene using the recombineering method (67). A corresponding 129AB2.2-BAC clone was obtained from SourceBioScience (Nottigham, UK). The final targeting construct (Fig. 1A) was transfected by electroporation into HM-1 ES cells (68). HM-1 cells are derived from 129/Ola mice. ES cells were grown on SNLP feeder cells. Clones exhibiting homologous recombination were selected by positive/negative selection using 350 mg/ml G418 and 0.5 mg/ml ganciclovir. Genomic DNA from ES cell colony replicates in 96-well plates was isolated as described (69), digested with the restriction enzyme MfeI/MunI (Thermo Fisher Scientific, Waltham, MA, USA). Southern blots were hybridized to 59-, 39-, and internal probes (Fig. 1B), which had been labeled with [ 32 P]dCTP (Hartmann Analytic, Braunschweig, Germany) using Klenow enzyme (Thermo Fisher Scientific). Probes were generated by PCR using the primers listed in Table 1. Oligonucleotides were obtained from Eurofins Genomics (Ebersberg, Germany). Bound probes were detected using a Fujifilm Bioimager. Around 400 G418/ganciclovir-resistant colonies were screened, and the targeting frequency was 1:20. One ES clone showed homologous recombination in both alleles (f/f in Fig. 1B). This clone was used for generating chimeric mice and subsequently heterozygous (Carns1(1/floxFRT)) mice. These mice were bred with Flp deleter (70) and pgk-cre transgenic mice (71) to generate Carns1-heterozygous (Carns1(1/2)) and finally Carns1(2/2) mice. Mice used in the experiments described in this report were backcrossed with C57BL/6J mice for five generations (resulting in a genetic background of about 96% C57BL/6J). Animal experiments were approved by the local and national authorities.

Antibodies
Antibodies used in this study together with supplier information and conditions of use are summarized in Table 2. Antisera directed against murine Carns1 were generated as follows. Murine Carns1 was expressed as two fragments (cDNAs encoding amino acids 1-539 and 540-947 were subcloned into pGEX-5X-1 and pGEX-5X-2, respectively) as GSH-S-transferase fusion proteins in Escherichia coli BL21 (DE3). Inclusion bodies were isolated and used for immunization of rabbits and guinea pigs. Two antisera specifically recognizing Carns1 by immunofluorescence and Western blotting were obtained (see Table 2). Antisera were affinitypurified using full-length hexahistidine-tagged mouse Carns1 expressed in baculovirus-infected HighFive insect cells as described (72).

RNA isolation and Northern blot analysis
RNA was isolated from mouse tissues using TRIzol (Thermo Fisher Scientific), as described (71). Northern blot analysis of mouse tissues was done using digoxigenin-labeled RNA probes, as described (72). A fragment of mouse Carns1 cDNA (nucleotides 424-1254) was amplified by PCR, using primers shown in Table 1, and subcloned into pBluescript SK(2). Digoxigenin-labeled RNA antisense probes were synthesized using T7-RNA polymerase (Roche Diagnostics, Mannheim, Germany) and a DIG RNA Labeling Mix (Roche Diagnostics). Northern blot hybridization was done as described (73).

PCR genotyping
PCR for genotyping was done using standard procedures with REDTaq ® ReadyMix TM PCR Reaction Mix (Sigma-Aldrich, Munich, Germany) and the oligonucleotides shown in Table 1. All oligonucleotides were ordered from Eurofins Genomics (Ebersberg, Germany).
cDNA synthesis and quantitative RT-PCR cDNA was synthesized using superscript II reverse transcriptase (Invitrogen), following the instructions of the manufacturer, using total RNA (1-5 mg) and oligo(dT) primers. Realtime RT-PCR was done as described (73) using the primer pairs listed in Table 1. Expression levels of Carns1 were normalized to ribosomal protein S2 (Rps2) using the 2 2DCt method.

Western blot analysis
Tissues were homogenized in homogenization buffer (10 mM HEPES, 250 mM sucrose, 1 mM EDTA, pH 7.4) with freshly added protease inhibitors (2 mg/ml leupeptin, 2 mg/ml pepstatin, 100 mg/ml Pefabloc (all from Sigma-Aldrich) or HALTprotease inhibitor mixture (Thermo Fisher Scientific)). Protein concentrations were measured using the Bio-Rad DC protein assay with BSA as standard. SDS-PAGE and semi-dry Western blotting were done using the Mini-PROTEAN Tetra cell gel electrophoresis system and Trans-Blot ® SD Semi-Dry Transfer Cell (both from Bio-Rad, Feldkirchen, Germany). The protein marker used was PageRuler Prestained Protein Ladder (Thermo Fisher Scientific; catalog no. 26617; lot 00395064). Antibodies and dilutions used are listed in Table 2. Bound secondary antibodies were detected using Pierce ECL Western blotting substrate (Thermo Fisher Scientific) and a CCD camera (Fusion Solo with FusionCapt Advance Solo 4 software, Vilber Lourmat, Eberhardzell, Germany). Western blot analysis of carbonylated proteins was done as described (74). Quantitative Western blot analyses were done with samples from male and female mice; however, in each independent experiment (analyzing 2-3 biological replicates per genotype), only ageand sex-matched mice were compared, and signal intensities were normalized to the mean of WT controls.

Dot blot analysis
Dot blot analysis of carbonyl-protein content and HNE and MDA adducts was done as described by Tanito et al. (75). Dot intensities were quantified by densitometry using the program AIDA (Elysia-raytest, Straubenhardt, Germany).

High-performance TLC (HPTLC) of peptides
Peptide extracts were prepared by homogenization tissue samples obtained from male mice in 90% methanol using an Ultra-turrax homogenizer (IKA, Staufen, Germany) (three times for 20 s on ice). After centrifugation (10,000 3 g, 10 min, 4°C) to remove precipitated proteins, supernatants were dried in a vacuum centrifuge and finally dissolved in 90% methanol. The 10,000 3 g pellet containing precipitated proteins was dissolved in 1% SDS and used for protein determination using the DC protein assay (Bio-Rad) and BSA as standard. Peptide extracts corresponding to 0.1 or 0.2 mg of protein were applied to HPTLC silica gel plates (Merck, Darmstadt, Germany), together with carnosine, anserine, and amino acid standards (all from Sigma-Aldrich), and chromatograms were developed in 1-butanol/acetic acid/water (50:25:25, v/v/v). Peptides and amino acids were visualized by spraying the TLC plates with 0.2% ninhydrin dissolved in ethanol and heating at 180°C for 2 min. Carnosine and anserine concentrations were determined by densitometry. Carnosine synthase-deficient mouse Histology and immunofluorescence Histological analyses reported were done with male mice. Mice were perfused with 4% paraformaldehyde in phosphate buffer, and tissue samples were postfixed for 16-18 h at 4°C. Decalcification of nasal bones was done with 14% EDTA (pH 7.2). Alternatively, in some experiments, treatment with 6% TCA for up to 1 week was done. Tissue samples were embedded in paraffin and cut at 2 or 4 mm using a microtome (Leica Biosystems, Wetzlar, Germany). Heat-induced epitope retrieval was done by microwave irradiation in 10 mM citrate buffer (pH 6.0). Immunofluorescence staining was done using antibodies and dilutions listed in Table 2. Nuclei were stained with 49,6-diamidine-29-phenylindole dihydrochloride (DAPI) (Sigma-Aldrich). Mature ORN were counted in 2-mm paraffin sections stained with antibodies directed against OMP. For each mouse analyzed, four sections (.20 mm apart) were collected from the same area of the nasal cavity containing the main olfactory epithelium of endoturbinate I (area 23 according to the mouse nasal cavity atlas published by Barrios et al. (76) (accessible at http://www.usc.es/anatembriol/). This was done in three age groups (8,18, and 24 months) using four mice per age cohort and genotype. Data were analyzed by factorial ANOVA with post hoc Newman-Keuls test using the program STATISTICA 6.0 (a p value , 0.05 was considered as significant). Pax6-positive cells in OE from 24-month-old mice were counted after immunofluorescence staining with Pax6 antibody, and data were analyzed by two-tailed t test.
Detection of carbonylated proteins by immunofluorescence was done as described (77) with modifications. Briefly, after deparaffinization, sections were rinsed in PBS (10 min) and incubated in 1 mg/ml DNPH (in 1 M HCl) for 15 min. Control slides were incubated in 1 M HCl. After washing slides three times with PBS, sections were blocked in 10% normal goat serum for 1 h and incubated with rabbit anti-DNP antiserum (diluted 1:1000; Sigma-Aldrich) overnight at 4°C. Slides were washed three times with 0.1% Nonidet P-40 in PBS and stained with anti-rabbit-Cy3 conjugate and DAPI. Signal intensities of DNP, HNE, and MDA immunofluorescence staining were quantified using ImageJ.

TUNEL assay
Apoptotic cells were detected by TUNEL assay using the ApopTag Fluorescein In Situ Apoptosis Detection Kit (Merck), following the instructions of the manufacturer. Negative controls lacked terminal deoxynucleotidyl transferase. Positive controls slides were prepared by treating samples with DNase I (10 units/ml) for 30 min at room temperature.

Isometric force measurements
Isometric force measurements were performed as described previously (78). In brief, SOL and EDL muscles from 6-8month-old male mice were explanted and kept in bicarbonate buffered Tyrode solution containing 118 mM NaCl, 3.4 mM KCl, 0.8 mM MgSO 4 , 1.2 mM KH 2 PO 4 , 11.1 mM glucose, 25 mM NaHCO 3 , and 2.5 mM CaCl 2 gassed with 95% O 2 and 5% CO 2 at room temperature. Muscles were mounted between a force transducer (TR7S) and a fixed hook within the Myostation intact (all from MyoTronic, Heidelberg, Germany). Force was recorded with a Powerlab 8/30 recording system and the Lab-Chart software 7.1 (AD Instruments). Platinum electrodes positioned perpendicular to the muscles and a Myostim stimulator (MyoTronic) were used for electrical stimulation (0.1 ms, biphasic). Muscles were prestretched to the optimal length until single electrical stimuli applied every 10 s evoked maximal force. 2-s-long tetanic stimulation was applied every 5 min with increasing frequencies from 10, 25, 50, 75, 125, and 150 Hz to determine the maximum isometric force. Fatigue protocols consisted of a 10-min-long series of 350-ms-long 100-Hz tetanic pulses every 3.7 s, and the fatigue index was calculated by normalizing all values to the maximal force. After the first fatigue protocol, muscles were incubated with 4 mM cinnamate (Sigma-Aldrich) for 20 min, and the same protocol was repeated. Directly after experiments, muscles were blotted dry on tissue paper and weighted. The cross-sectional area was determined by multiplying the weight with the density (1.06 mg 3 mm 23 ) divided by the optimal length determined during prestretch. Force was normalized to the cross-sectional area of the muscle.

Grip strength test
Grip strength of male mice was measured using a power meter (Sauter, Balingen, Germany) equipped with a grasping grid. The mouse was held at the base of the tail and allowed to grasp the grid with all four paws. The mouse was then gently pulled back, and the maximum force was recorded. The test was repeated during the following 2 days, and the maximal value from the three trials was used for data analysis.

Tests of olfactory function
In preliminary tests using WT mice, we did not observe significant differences between male and female mice concerning their performance in the different olfactory tests applied in this study. Therefore, data for both genders were combined, to minimize the number of animals to be bred. The different olfactory tests were performed with separate cohorts of animals, with the exception of the preference tests, in which all 18-month-old mice had been tested in the buried food test 1 day before. Testing of the different age groups (buried food and sensitivity tests) was always done with separate cohorts of animals.
Buried food test (79)-Mice were food-deprived overnight (18 h). The test cage (area: 30 3 16 cm, height: 13 cm) contained fresh bedding (2-cm filling height), and the mouse was acclimatized to the test cage for 5 min. The mouse was placed back into its home cage for 1 min, and a piece of chocolate cookie (Leibniz Minis BLACK'N WHITE, lot no. 01.01.17/ 6085A; Bahlsen, Hannover, Germany) was buried in one randomly chosen corner of the test area. The mouse was then placed back into the center of the test cage and allowed to inspect it for a maximal time of 15 min. Time was stopped when the mouse grabbed or started to eat the cookie (cut-off time 15 min). Data were tested for significant differences between genotypes using one-way ANOVA.
Preference test (80)-An empty cage (area: 30 3 16 cm, height: 13 cm) was used as test arena, and flavors were presented by soaking Whatman filter paper (2 3 2 cm) with 60 ml of diluted test substance or water. Each experiment was started by acclimatizing the mouse by placing it for 5 min into the empty test cage. Filter paper pieces soaked with water and peanut butter (KG Fancy Food Handels GmbH, Hamburg, Germany) (1 g/100 ml mineral oil (Sigma-Aldrich)), respectively, were placed into the two corners of one short side of the test cage. The mouse was allowed to explore the cage for 3 min, and the times inspecting the water-soaked paper (neutral) and the peanut butter-soaked paper (preferred odor) were measured. After a 1-min brake, the test trial was repeated with water and 2-MB (Sigma-Aldrich; nonpreferred odor) placed into the corners of the opposite short side. The difference between test substance and water inspection times was calculated, and differences between peanut butter and 2-MB were tested for significance using two-tailed t test, separately for each genotype.
Habituation-dishabituation test (modified after Witt et al. (80) and Arbuckle et al. (81))-After acclimatization, as for the preference test (see above), a filter paper soaked in 60 ml of test substance was placed in the center of the test cage. Test substance in trials 1-3 was water, followed by almond (diluted 1:100 in water) in trials 4-6, and vanilla (diluted 1:100 in water) in trials 7-9. Total inspection times in each test trial were determined. Because variances between consecutive trials differed significantly (as determined by F-test), data were logarithmically transformed according to the formula log(1 1 x). Time differences between consecutive trials 3 and 4, as well as trials 6 and 7, were tested for significance using Rodger's method, as described by Rodger and Roberts (82). This test was selected because of its high power, because a statistically significant difference was expected for control (WT) mice.
Sensitivity test-This test was performed and evaluated as described above for the habituation-dishabituation test with the following modifications. After the three test trials with water, six trials with increasing concentrations of the test substance (dilution from 1:10 7 to 1:10 2 ) were conducted. The test was done with three different substances (almond, vanilla, and lemon aroma (all from Dr. August Oetker Nahrungsmittel KG, Bielefeld, Germany)) with the same group of mice on three successive days, with the exception of almond aroma at 18 months, which was tested with a separate cohort. The behavioral olfactory threshold was determined by calculating the test trial that showed the first significant increase in inspection time compared with its preceding trial using Rodger's method (82). Although we did not perform outlier tests, data of one mouse from one experiment (18 months, Carns1(2/2), vanilla aroma) was excluded from the analysis because this mouse did not approach the test substance-soaked filter papers during the whole test session.

Statistics
Data are presented as the mean and 95% CI. Calculations were done using Excel (Microsoft, Redmond, WA, USA), STA-TISTICA 6.0 (Statsoft, Tulsa, OK, USA), or GraphPad Prism (GraphPad Software, San Diego, CA, USA). Data were tested for significant differences (p , 0.05) using ANOVA with appro-priate post hoc tests or two-tailed Student's t test if not otherwise indicated. Correction for multiple comparisons was done according to Hochberg (83) unless otherwise stated.

Data availability
All data are contained within the article.