All experiments were conducted following the guidelines of the IACUC of the University of Maryland, College Park. Experiments were performed in 1–3 months old knockout (KO) female and male mice lacking the M1 or the M1 and M3 mAChR (M1^−/− and M1/M3^−/− mice, respectively), generously provided by Dr. Jurgen Wess from the NIH. The respective background WT strains of these KO mice are the C57 BL/6 (The Jackson Laboratories) and CF129 mice (Charles River). All mice were obtained from breeding pairs housed in our animal facility, maintained in separate colonies based on strain, and kept on a 12 h light/dark cycle with ad libitum access to food and water, unless otherwise indicated.

A clean standard mouse cage (15 cm × 30 cm), without bedding, was used for behavioral testing. Mice were placed in the cage and allowed to familiarize with the test environment for 30 min. During the familiarization phase a wooden cube (2 cm³) was placed in the cage in the presence of the mouse. At the conclusion of the familiarization phase, the wooden block was removed for 1 min, and thereafter the mouse was exposed to a wooden block scented with pure water, three times (2 min each), with a 1 min inter-trial interval; these exposures familiarize the mouse to the habituation procedure. The second phase consisted of subsequent exposures to a wooden block scented with 100 μL of the test odor at a 1:1000 dilution (in water) as follows: habituated odor, 3×, dishabituated odor, 1×. Each trial was videotaped and the time the mouse spent investigating the block was quantified offline. The investigation time was determined as the time during which the mouse’s nose was within a 1 cm radius from the block. Unless otherwise indicated, and to minimize differences in behavioral performance, the investigation time in each trial was normalized to the investigation time during the first odor presentation (trial 1). Mice are able to naturally discriminate between the pair of odors when the investigation time is significantly increased during the presentation of the novel odors.

Mice were housed individually and placed on feed restriction for the whole duration of the associative learning trials, with ad libitum access to water. Mice were weighed each day to monitor weight loss and maintained at no less than 85% of their original body weight. For the associative learning task mice were trained to associate an odor with a food reward. A piece of nutter-butter cookie was hidden under a 2 cm² filter paper soaked with 75 μL (1:100 v/v in water) of the associated odor (Figure 2A) both buried under the bedding. A filter paper containing the non-associated odor was buried on the opposite side of the cage, and the cage sides where the filter paper and the reward were hidden was randomized. In these experiments we used the isomers of carvone, which both the CF129 and M1/M3^−/− mice failed to discriminate in the habituation-dishabituation paradigm.

Each day of training consisted of five-trials, in which the time that mice took to retrieve the food reward was recorded. For each animal, the data was normalized to the time the mouse took to retrieve the food reward during the first trial. Under our experimental conditions mice, regardless of the strain, learned the task within 2 days.

At the end of each training day (starting at day 1), mice (CF129 and KO) were tested in the habituation-dishabituation test. In addition, at the end of the day 2 of associative training, mice were presented with both odors simultaneously, in the absence of the reward, for 2 min, and the time they spent sniffing each odor was quantified. For each mouse the time spent investigating each odor was normalized to the total time when both odors are presented together. Mice have learnt to discriminate these odors when they spend a significant greater time investigating the associated odor vs. the non-associated odor. For each strain the associative training was continued until the mice were able to discriminate odors presented simultaneously.

Urine samples were collected from mice using standard protocols (Yang and Crawley, 2009). Evaluation of discrimination of conspecific’s urine was assessed in a similar manner to the procedure described in the habituation-dishabituation paradigm above. Male mice were presented with male urine of the same background strain three times (habituated odor). Following the sequential presentations of the male urine, mice were presented with female urine of the same background strain.

For the aversive odor experiment, mice were consecutively presented for 2 min with either the aversive odor, 2,5-dihydro-2,4,5-trimethylethiazoline (TMT), or a neutral stimulus, water, for 2 min with a 1-min interval between trials. As in the previous experiments, the odor was presented using a wooden cube and was placed on one end of the mouse cage. Odor aversion was quantified based on the amount of time the mouse spends in the corresponding half of the cage in which the odor is presented.

Visual discrimination was assessed using two odorless, but different colored and shaped Lego® blocks using the habituation-dishabituation paradigm (see above). The Lego® blocks were previously wiped down with 70%-volume based ethanol to remove pre-existing scents. The first object was presented three times sequentially for 2-min with a 1-min inter-trial. The novel object was presented during the fourth trial for 2 min.

The odor detection threshold for the different strains was determined using a modified version of the habituation-dishabituation test. Mice were habituated to three presentations of a wooden block scented with distilled water and on the seventh trial the diluted odor in water was presented (1/60,000, 1/40,000, 1/30,000 and 1/20,000). During these trials the overall investigation time for each mouse decreases, therefore, for each mouse we normalized the investigation time for each trial to the first trial (water).

To quantify adult neurogenesis we used 5-bromo-2′-deoxyuridine (BrdU), 10 mg/mL in phosphate-buffered saline (PBS). One-month old mice received an intraperitoneal injection of BrdU (100 mg/kg), 3× every 2 h. After the injections, mice were kept in the animal facility with food and water ad libitum for 1 month. Mice were sacrificed and perfused intracardially with 10 mL of cold 1× PBS, followed by 25 mL of cold 4% paraformaldehyde (PFA; Electron Microscopy Services). Brains were then excised and post-fixed in 4% PFA for 4 h at 4°C, and then immersed in 30% sucrose (Fischer) diluted in 1× PBS overnight at 4°C. Following fixation, the brains were sectioned into 20 μm sections using a cryostat and immunohistochemically stained for BrdU.

Staining required an initial stage of antigen retrieval where slices were incubated in 2 M HCl at 37°C for 1 h. This was followed by a blocking period of 2 h where slices were incubated in 10% donkey serum in PBS with 0.1% Triton X-100 (PBS-T). Slices were then incubated overnight in 1:150 dilutions of rat anti-BrdU (Abcam, ab6325) and mouse anti-NeuN (Chemicon International, MAB377) antibodies in PBS-T containing 2.5% donkey serum. The following day, slices were incubated for 2 h in a secondary solution containing 1:750 donkey anti-rat (Invitrogen, A-21209) and donkey anti-mouse (Invitrogen, A-21202) antibodies, conjugated with Alexa594 and Alexa488 fluorophores, respectively. Secondary antibody dilutions were made in PBS-T containing 2.5% donkey serum. To assess basal level of neurogenesis in both CF129 (WT) and M1/M3^−/− strains, mice were injected with BrdU at postnatal week (PW) 8 or PW14 and then sacrificed at PW12 or PW18, respectively. For BrdU positive neuron quantification, we used stereology for unbiased cell counting described in detailed in Nunez-Parra et al. (2011). For OE (see below), BrdU injection was performed at PW14 and the animals sacrificed at PW18. These mice were tested in the habituation-dishabituation before, and after injection of BrdU followed by OE.

Odor-enriched mice were exposed daily, for 24 h, to different aromatic fragrances placed in a metal tea ball hanging from the cover of standard breeding cages. Control mice were reared under the same conditions except that the tea ball was left empty. OE occurred over a 40-day period. Aromatic fragrances included lavender, garlic, paprika, marjoram, curry, rosemary, nutmeg, thyme, basil leaves, cumin, cardamom, tarragon, whole cloves, chocolate, celery, anise, ginger, lemon, orange and banana. Mice were exposed to each fragrance once over the enrichment period (Rochefort et al., 2002).

We used urine samples from mature and sexually naïve, male and female, mice. A gentle pressure on the mice’ lower back was used to induce urine voiding (Watts, 1971). Collected urine was then pooled by sex and strain and stored at −20°C until needed. Odors used for the habituation-dishabituation and hidden-cookie were obtained from Sigma-Aldrich (St Louis), at the purest grade: Butanol (C4-OH), Pentanol (C5-OH), Ethyl hexanoate (C6), Ethyl heptanoate (C7), Ethyl octanoate (C8), L- Carvone (L-car), D-Carvone (D-car); (+)-Limonene (+)-lim, (−)-Limonene (−)-lim. TMT was obtained from (PheroTech). Odor dilutions were made in water on the day of the experiment. Unless otherwise stated, all comparisons made were evaluated using a Student’s t-test to determine statistically significant differences.

Due to the longitudinal nature of the experiments conducted, we used a paired t-test for determining statistical significance. In addition, we used the GPower software¹ to verify the statistical power for our t-tests. Using this software, we found that our reported significant differences had a power level of 99%–99.99%. The sample size for each experiment ranged from 4 to 23 mice and we only used males for our experiments to avoid incorporating an additional variable (estrous cycle and hormonal fluctuations). All mice were used for one test only, unless otherwise stated (i.e., associative learning). The error bars correspond to the standard error of the mean (SEM).

Previous work has shown that cholinergic modulation of excitation of GCs in the OB occurs mainly through activation of type M1 mAChRs (Pressler et al., 2007; Smith and Araneda, 2010; Smith et al., 2015), however, the contribution of this receptor to odor perception remains unknown. To test the ability of the M1^−/− KO mice to naturally discriminate structurally similar odors, we used the habituation-dishabituation paradigm. We first characterized the behavior of the WT background strain of the M1^−/− KO, the C57/BL6 mice (C57). As shown in Figure 1A, bottom left, WT mice habituated to sequential presentations of an odor-scented wooden block (see “Materials and Methods” Section), and showed a decrease in the investigation time. The investigation time for the C7 ester decreased by 75%, across three odor presentations (trial 1, 9.5 ± 0.6 s; trial 3, 2.4 ± 0.3 s; n = 28; p < 0.0001). WT mice readily discriminated structurally linear esters that differed by one carbon moiety, and accordingly the investigation time increased during the presentation of the C8 ester (trial 4, 7.7 ± 0.5 s; p < 0.0001). In addition, WT mice discriminated the C5/C6 ester pair and esters that differed by two carbons (Table 1).

However, although the WT mice habituated to the “L” isomer of carvone (L-car, trial 1, 10.7 ± 0.5 s; trial 3, 3.5 ± 0.4 s; n = 23; p < 0.0001), the investigation time did not significantly increase during the presentation of D-carvone (D-car, 3.4 ± 0.5 s, p > 0.5), suggesting that the WT are not able to discriminate the “L” and “D” enantiomers. This natural inability to discriminate odors extended to other structurally related molecules such as the C4/C5 alcohol pair and the enantiomers of limonene (Table 1).

Similar to the WT mice, M1^−/− mice engaged normally in the habituation phase of the test and exposure to C7 resulted in a 58% decrease by the third odor presentation (Figure 1A, bottom right). In contrast, unlike the WT, the M1^−/− mice failed to discriminate esters whose structure differed by one carbon moiety. The investigation time for the habituated odor (C8) was not different to the response to the habituated C7 (C7, 3.1 ± 0.6 vs. C8; 2.5 ± 0.6 s; n = 8; p > 0.4; Figure 1A, bottom right). Similar results were obtained with the C5/C6 ester pair, the C4/C5 alcohol pair, and the stereoisomers of Limonene and the “L” and “D” carvone enantiomers (Table 1).

In contrast, the M1^−/− mice could readily discriminate between C6 and C8, an ester pair whose structure differs by two carbons (C6, 3.1 ± 0.5 s; C8, 7.8 ± 0.7 s; n = 10; p < 0.001; Figure 1A, bottom right). In summary, this data indicates that ablation of the M1 mAChR does not affect odor habituation per se and that muscarinic neuromodulation trough M1 mAChR is not required to discriminate between perceptually different odor molecules. However, olfactory discrimination of structurally similar molecules is impaired in the M1^−/− mice.

Surprisingly, in electrophysiological recordings of the accessory OB (AOB), the OB region that processes pheromonal information, we found that GCs still exhibited an excitatory muscarinic response in the M1^−/− mice (not shown). Further pharmacological characterization indicated that the depolarization in AOB was sensitive to M3 mAChRs blockers, suggesting an up-regulation of these receptors in the M1^−/− KO mice (see also Smith et al., 2015). To circumvent this problem, for the next experiments we determined the contribution of M1-mAChRs to olfactory processing using a mouse that lacks both the M1 and M3 receptors (M1/M3^−/− double KO mice). In parallel, we performed the same battery of experimental behavioral paradigms in the M1^−/− for comparison purposes.

As shown in Figure 1B, left, CF129 mice (the WT strain from which the M1/M3^−/− derives) showed robust habituation during the sequential presentations of C7 (trial 1, 9.4 ± 0.6 s; trial 3, 2.6 ± 0.4 s; total decrease 72%; n = 20; p < 0.001) and discriminated the C8 ester (trial 4, 6.6 ± 0.5 s; p < 0.001). Although the CF129 mice habituated to L-car (trial 1, 7.7 ± 0.5 s; trial 3, 1.7 ± 0.3 s; n = 16; p < 0.001), the investigation time did not significantly increase during the presentation of D-car (trial 4, 2.3 ± 0.4 s; p > 0.3), suggesting that the CF129 mice and the C57 exhibited a similar olfactory discrimination profile.

Similar to the WT mice, M1/M3^−/− mice was engaged normally in the habituation-dishabituation test during the presentation of the C6. The M1/M3^−/− readily discriminated between a pair of ethyl esters that differed by two carbons, C6/C8 (C6 and C8; trial 1, 10.9 ± 1.7 s vs. trial 3, 4.1 ± 0.7 s; total decrease 63%; n = 14; p < 0.05; Figure 1B, right). In contrast, unlike the WT and mirroring the M1^−/− behavior, the M1/M3^−/− mice failed to discriminate esters whose structure differed by one carbon moiety. As shown in Figure 1B, right, the investigation time for the novel odor (C8) was not different to the response to the habituated odor (trial 1, 2.8 ± 0.5 vs. trial; 3.7 ± 0.6 s; n = 20; p > 0.6). Also, like the WT, they failed to discriminate the “L” and “D” carvone enantiomers (see Figure 2B). We also tested the detection threshold in the M1/M3^−/− mice using a modified version of the habituation-dishabituation paradigm (see “Materials and Methods” Section). The odor detection threshold for the C7 in the M1/M3^−/− mice was the same as in the WT, with detection of the odor occurring at 1/30,000. Thus, for the CF129 at 1/40,000 the investigation time was not different between water and the diluted odor (0.35 ± 0.03 vs. 0.37 ± 0.03, n = 4, p > 0.07) but significantly increased when the odor was presented at the 1/30,000 dilution (0.24 ± 0.04 vs. 0.75 ± 0.01, n = 4, p < 0.02). Similarly for the M1M3^−/−, the investigation of water and the diluted odor was not different (0.31 ± 0.03 vs. 0.42 ± 0.04, n = 3, p > 0.08) but significantly increased when the odor was presented at the 1/30,000 dilution (0.22 ± 0.04 vs. 4 0.64 ± 0.03, n = 3, p < 0.01).

Muscarinic receptors are known to play a role in associative olfactory memories (De Rosa and Hasselmo, 2000; Saar et al., 2001), not only at the level of the OB, but also in the olfactory cortex (PC), one of the main targets of the OB output neurons (Nagayama et al., 2010). Therefore, we examined whether cholinergic deficits had an effect on long-term memory, specifically odor recognition elicited by an associative-odor learning paradigm. It has been shown that repetitive exposure of a non-discriminant odorant results in an increase of the rodent ability to discriminate them and that this improvement depends on activation of mAChR (Fletcher and Wilson, 2002). Thus, we compared the ability of the CF129 and M1/M3^−/− mice to discriminate the isomers of carvone, after associative training with a food reward. For this, mice were trained to associate L-car with a cookie reward and D-car was used as the non-associated odor (see “Materials and Methods” Section, Figure 2A left). As shown in Figure 2A right, by the end of the second day of training both the WT and M1/M3^−/− mice learned the task; however, during training the learning curve for the M1/M3^−/− mice was significantly slower than the WT mice. By the end of the first day of training, the M1/M3^−/− mice take a significantly longer time to find the food reward than the WT (normalized retrieval time, 0.66 ± 0.06 vs. 0.17 ± 0.04; n = 4; p < 0.004). Similar results were obtained with the M1^−/− mice (normalized retrieval time, 0.94 ± 0.07 vs. 0.17 ± 0.04, n = 4, p < 0.05).

To test the ability of the mice to discriminate between the associated and non-associated odors, each odor was presented on a different side of the cage in the absence of the reward. We found that WT but not M1/M3^−/− mice were able discriminate between the carvone isomers by the end of the second day of training (L-car, 0.87 ± 0.03 s; D-car, 0.13 ± 0.03 s; n = 4; p < 0.01; Figure 2C). In contrast, the M1/M3^−/− mice required a total of 9 days of associative learning trials before they could perform successfully in this task (L-car, 0.82 ± 0.04 vs. D-car; 0.18 ± 0.04; n = 4; p < 0.05; Figure 2C). Experiments performed with the M1^−/− showed a similar pattern requiring 8 days of associative training to discriminate the isomers (L-car, 0.77 ± 0.05 vs. D-car 0.23 ± 0.05; n = 4, p < 0.05).

In addition, at the end of each training day we examined the performance of the mice in the habituation-dishabituation test (L-car was used as the habituated odor and D-car as the dishabituated odor). As shown in Figure 2B, by the end of the second day of training, the WT but not the M1/M3^−/− mice were able to discriminate the stereoisomers of carvone (WT, L-car, 0.24 ± 0.05; D-car, 0.82 ± 0.11; n = 4; p < 0.05: M1/M3^−/−, L-car, 0.34 ± 0.07; D-car, 0.42 ± 0.13; n = 4, p < 0.5). However, after 2 additional days of associative training the M1/M3^−/− mice were able to discriminate the carvone isomers (L-car, 0.29 ± 0.07; D-car, 0.77 ± 0.03; n = 4; p < 0.05; Figure 2B). Therefore, in addition to their difficulty in naturally discriminating structurally similar odors, the M1/M3^−/− mice show impaired associative learning capability showing a much slower learning rate than WT mice.

Odors found in nature are complex mixtures of odorants, which animals use to elicit social behaviors critical for survival and reproduction. Therefore, we tested the ability of these mice to engage in the investigation of a more complex odor mixture; for this we used urine, which is a social odor containing several semiochemicals (Tirindelli et al., 2009). As seen in Figure 3A left, both WT and M1/M3^−/− mice habituated to consecutive presentations of male urine and increased the investigation time in the presence of the novel odor (female urine; CF29, 0.34 ± 0.01 vs. 1.23 ± 0.05; n = 4; p < 0.01; M1/M3^−/− 0.24 ± 0.03 vs. 1.47 ± 0.21; n = 4; p < 0.02). We also exposed the animals to an innate aversive odor, TMT, which is a component of fox feces, a natural mice predator that elicits a variety of fear-like behaviors in rodents (Rosen et al., 2015). Similarly, the response to an aversive odor was not different between the WT and the M1/M3^−/− mice (CF29, 0.69 ± 0.04 vs. 0.31 ± 0.04; n = 4; p < 0.01; M1/M3^−/− 0.68 ± 0.02 vs. 0.32 ± 0.02; n = 4; p < 0.02; Figure 3A right). Furthermore, to control for the possibility that our mouse model could have impaired vision, we exposed them to legos of different colors (see “Materials and Methods”). We found that M1/M3^−/− mice are not different than WT (CF29, 0.34 ± 0.01 vs. 1.23 ± 0.05; n = 4; p < 0.01: M1/M3^−/− 0.24 ± 0.03 vs. 1.47 ± 0.21; n = 4; p < 0.02; Figure 3B). For comparison purposes we tested the same odors and novel objects in the M1^−/− and found a normal sensory discrimination profile (Table 1).

Previous studies have indicated a link between cholinergic modulation and adult neurogenesis (Cooper-Kuhn et al., 2004; Kaneko et al., 2006; Paez-Gonzalez et al., 2014; Asrican et al., 2016), a postnatal brain process that is required to maintain adequate olfactory discrimination of odorants (Moreno et al., 2009). Therefore, we hypothesized that the deficit in discrimination of perceptually similar odors in the M1/M3^−/− mice might result from a deficiency in the generation of adult-born neurons. We labeled the newly generated neurons by injecting the animals with BrdU, which is incorporated into the DNA of dividing cells. Accordingly, we quantified the density of BrdU+ neurons in the granule cell layer (GCL) of WT and M1/M3^−/− mice at 12 and 18 weeks postnatal (Figure 4A) using immunofluorescence against BrdU and NeuN (a marker of mature neurons, Figure 4B). As shown in Figure 4C, we found that M1/M3^−/− mice at 3 months of age show a significantly lower density of BrdU+ neurons in the GCL in comparison to the WT mice (2624 ± 429 vs. 5348 ± 723 cells/mm³; n = 4; p < 0.05). The total average density of adult generated neurons across the MOB of M1/M3^−/− mice was also significantly lower compared to the WT (1213 ± 87 vs. 2655 ± 234 cells/mm³; p < 0.05). This suggests that at this age, a decrease in the number of newborn neurons integrated in the OB in the KO mice could play an additional role in the olfactory deficits they exhibit.

Surprisingly, when we quantified BrdU+ neuron density at four and half months of age, we did not observe a difference in adult neurogenesis between M1/M3^−/− and WT mice (Figure 4C). As previously shown, adult neurogenesis decreased with age (Nunez-Parra et al., 2011); however, M1/M3^−/− mice display surprisingly similar levels of BrdU+ cells in the GCL in comparison to their WT counterparts (2084 ± 233 vs. 2196 ± 302 cells/mm³; n = 5). Together these findings suggest that cholinergic excitation plays a diminished role in regulating adult born neuron survival in mice at later ages. However, we note that the natural discrimination of closely related odors in M1/M3^−/− mice is age-independent, implying that adult-born neuron density in these mice is not closely linked with discrimination ability.

Previous work has shown that OE is a potent stimulator of adult neurogenesis (Rochefort et al., 2002). Furthermore, OE is known to increase perceptual learning, whereby mice are able to naturally discriminate closely related odors that were not discriminated previously (Mandairon et al., 2006b; Escanilla et al., 2008). We therefore examined the possibility that OE, and increased adult neurogenesis, could compensate for the olfactory discrimination deficit observed in the M1/M3^−/− mice. To this extent, mice were exposed to different natural odors in their home-cage that were changed daily for a total period of 40 days. Three weeks into the enriched paradigm, animals were injected with BrdU and the number of OB-integrated neurons was quantified 4 weeks after injection (Figure 4A, bottom). As shown in Figure 4C, OE significantly increased the density of BrdU+ neurons in the GCL of both M1/M3^−/− (3600 ± 125 cells/mm³; n = 7) and WT mice (3596 ± 460 cells/mm³; n = 8), in comparison to mice exposed to empty odor containers (M1/M3^−/−: 2084 ± 233 cells/mm³, n = 5; WT: 2196 ± 302 cells/mm³; n = 4; p < 0.05).

To assess the influence of OE on odor discrimination, we performed habituation-dishabituation tests on M1/M3^−/−. Before the exposure, natural odors, M1/M3^−/− mice could not discriminate between C7-C8 odor pair (C7, 3.2 ± 0.7 s; C8, 3.3 ± 0.7 s; n = 13; p = 0.94; Figure 4D); however, after exposure to an odor-enriched environment these mice discriminated the odor pair (C7, 3.5 ± 0.5 s; C8, 9.0 ± 0.6 s; n = 6; p < 0.001). M1/M3^−/− mice exposed to empty odor containers still displayed a deficiency in naturally discriminating these odors (C7, 3.4 ± 0.7 s; C8, 4.1 ± 1.0 s; n = 5; p = 0.72, not shown).

These findings suggest that OE produces changes within the OB circuit necessary for the proper discrimination of perceptually similar odors. Additionally, because M1/M3^−/− mice reared in conditions with empty odor containers have similar levels of neurogenesis in comparison to their WT counterparts, but still fail to discriminate odors that differ by a single carbon, neurogenesis and odor discrimination may not be as tightly coupled as thought (Mandairon and Linster, 2009). In contrast they could be two alternative means to increase olfactory discrimination at different time-scales at later ages.