Sprague-Dawley rats were purchased from Harlan Laboratories (Madison, WI). Female rats were used because menthol cigarette smokers are more likely to be female (Fernander et al., 2010; Lawrence et al., 2010). Similar to our previous studies (Chen et al., 2007, 2012), the rats arrived at our animal facility on approximately postnatal day 31 and received surgery on approximately postnatal day 38. The definition of adolescence in rodents is controversial. According to a conservative perspective on rodents (Spear, 2000), prototypical adolescent changes occur during postnatal days 28–42. Some developmental changes specific to adolescence do persist through postnatal day 55 (Spear, 2000). Thus, the present experiments were performed within the broadly defined age range of adolescence. The rats were group housed in rooms with a reverse light cycle (lights on from 9:00 pm to 9:00 am). Standard rat chow and water were provided ad libitum. All procedures followed the NIH Guidelines Concerning the Care and Use of Laboratory Animals and were approved by the Animal Care and Use Committee of the University of Tennessee.

A sterile Micro-Renathane catheter (OD = 0.94 mm, ID = 0.58 mm, Braintree Scientific Inc., Braintree, MA) was implanted on approximately postnatal day 38 (Chen et al., 2007). The catheter exited from the back of the animal through a polylactic acid implant (described in detail below). Pain medication (Ketofen, 2 mg/kg) was provided immediately after surgery.

The rats received no prior operant training, water or food deprivation, or social isolation before IVSA; they were group housed according to the treatment conditions throughout the experiment. Daily 3-h nicotine IVSA sessions were conducted in operant chambers (Med Associates Inc., St. Albans, VT) equipped with two stainless steel sipping tubes containing metal ball bearings. Both sipping tubes were connected to contact lickometer controllers. Syringes containing nicotine (30μg/kg/injection, free base, pH 7.0–7.4) and flavor cues were filled prior to each session. Polyethylene tubing connected the nicotine syringe to the jugular catheter through a swivel; the flavor cue was delivered near the metal ball bearings through polyethylene tubing. The flavored solutions included menthol (0.01% w/v, in 0.01% Tween 80), vehicle (0.01% Tween 80), a mixture of saccharin (0.125%) and glucose (3%), and unsweetened grape-flavored Kool-Aid® (0.1%). WS-23 (2-isopropyl-N,2,3-trimethylbutyramide; The Ingredient House, Pinehurst, NC. 0.01% and 0.03% in water), which is a TRPM8 channel agonist that has a potency within the same order of magnitude as menthol (Behrendt et al., 2004), was also used as a sensory cue. One additional group received a composite cue of 0.01% WS-23, 0.4% saccharin, and 0.1% Kool-Aid®.

Licking on the active spout meeting a fixed-ratio 10 reinforcement schedule with a 20 s timeout simultaneously activated both syringe pumps (model PHM100 with a 15-rpm motor, Med Associates) to deliver nicotine or saline in 0.8–1.2 s based on body weight and 60 μl flavor cue in 0.76 s. Therefore, licking on the active spout (hereafter referred to as active licks) delivered the flavor cue only when nicotine was also delivered. Licking on the inactive spout (i.e., inactive licks) had no programmed consequence. A contingent audiovisual cue was provided for three additional groups (oral vehicle with i.v. nicotine, oral menthol with i.v. saline, and oral menthol with i.v. nicotine). The audiovisual cue consisted of the onset of a tone generator (Malloary Sonalert, SC628, 80 dB) and a white cue light (Med Associates, ENV-221M, with 100 mA bulb) located above the spouts for the duration of the nicotine infusion. For the yoked control experiment, each yoked rat was paired with a master rat but was placed in its own operant chamber. The yoked rats received a nicotine infusion concurrently with their master's self-administered nicotine. When the yoked rats licked the active spout, they received menthol but not nicotine.

A stock menthol solution (Sigma, St. Louis, MO) (10×, 0.1% menthol, and 0.1% Tween 80) was heated (25 s) in a microwave oven, mixed, and maintained at room temperature for a maximum of 7 days. A fresh menthol working solution (1×) was prepared daily. This concentration of menthol was used because it can unequivocally activate the thermal-sensitive lingual fibers (Lundy and Contreras, 1994). Methohexital (5 mg/kg; JHP Pharmaceuticals, Rochester, MI), a fast-acting anesthetic, was used to verify the patency of the jugular catheters. Rats that failed this test were excluded from the analysis.

One group of rats self-administered nicotine with contingent cold water (60 μl). Several modifications were made to the cue delivery setup to ensure that the water was delivered at a constant low temperature throughout the 3-h IVSA sessions. The end of the stainless steel spout located outside of the operant chamber was sealed and placed in a bottle filled with a nontoxic cooling gel. These bottles were stored at −80°C before use. Water delivered by syringe pumps first entered a ring of plastic tubing (55 cm, ID = 0.76 mm) submerged in ice water. The plastic tubing was then connected to stainless steel tubing that entered a small opening 2 cm from the tip of the spout located inside the operant chamber. Cold water was delivered close to the ball bearing inside the spout. The water temperatures measured at the tip of the drinking spout were 9.3 ± 0.3, 11.2 ± 0.3, 11.3 ± 0.3, and 11.9 ± 0.4°C at 0, 1, 2, and 3 h, respectively. The room temperature was 18.7–20.2°C during the experiment. A picture of the setup is presented in Figure S1.

During the 3-h extinction sessions, licking was recorded but had no programmed consequence. Rats met the extinction criteria when the number of licks on the active spout were less than 150 (i.e., ~20% of those on the first day of extinction in the menthol-nicotine group) for two consecutive days. The reinstatement of nicotine seeking was then tested. During the 3-h reinstatement sessions, active spout licking delivered a menthol solution (60 μl) under a fixed-ratio 10 schedule with a 20 s timeout. Nicotine was not delivered.

A MakerBot® Replicator 2 (Makerbot Industries, Brooklyn, NY) was used to manufacture the implant from polylactic acid. The implant was printed in 0.1 mm layers. Stainless steel tubing (4 mm, 23 gauge) was inserted 15 mm into the distal end of a 12 cm Micro-Renathane tube to provide a strong surface for tying sutures over the jugular vein. The proximal end of this tubing was connected to a 35 mm, 23 gauge stainless steel tube, which was then inserted into the center of the implant and extended 5 mm beyond the implant. One 10 mm to 12 mm stainless steel spring was placed outside the stem of the implant to prevent it from being damaged during group housing. The implant design and an image of the assembled implant are presented in Figure S2 (see also http://www.thingiverse.com/haochen/).

One perforated divider was placed in the middle of the operant chamber with all operant manipulanda removed. The rats and odorants were placed on opposite sides of the divider. Menthol (0.01%, dissolved in 0.01% Tween 80), WS-23 (0.01% or 0.03% in water), or water (1 mL each) was placed in a plastic weighing boat on the chamber floor ~1 cm from the divider. Eleven naive female Sprague-Dawley rats were tested. Each rat was tested with all three odorants daily for two consecutive days. The odorants were always placed in the same test chambers to avoid potential odor contamination. The odorant sequence was counterbalanced among the rats. The number of nose pokes into the divider was recorded for 20 min by infrared sensors embedded in the divider. The rats remained in their home cages for > 1 h between tests.

The timing of licks on the active spout was analyzed for its microstructure. Licks with interlick intervals of less than 0.5 s were treated as one cluster. Clusters with less than two licks were excluded from the analysis. The size of the lick cluster, defined as the number of licks contained within a cluster, corresponded to the appetitive nature of the solution (Davis and Smith, 1992).

The number of licks and the ratio of active/inactive licks were log-transformed to fit a normal distribution prior to statistical analysis. The data were presented as the mean ± SE. Repeated-measures analysis of variance (ANOVA) was used to analyze the number of licks and infusions, with the session and spout treated as within-subject variables and treatment groups, including the cue used (e.g., menthol, vehicle, WS-23, etc.) and i.v. solution (i.e., nicotine or saline), as between-subject variables. Because the number of active licks included those occurred during the 20 s time-out period, which had no behavioral consequence, these data were noisier than the number of infusions. Therefore, with the exception of the yoke experiment, we, in general, analyzed the number of nicotine infusions when different experimental conditions were compared. The number of active licks were compared between the yoked and the the master groups because the number of infusions was the same between these two groups by design. Tukey's HSD test was used for post-hoc analysis when appropriate. All analyses were conducted using the R statistical analysis package.

The rats were trained to self-administer i.v. nicotine with a contingent oral menthol cue. The control groups self-administered i.v. nicotine with a contingent vehicle cue, i.v. saline with menthol cue, or i.v. saline with vehicle cue. The numbers of infusions that these groups obtained are shown in Figure 1A. Repeated-measures ANOVA found significant main effects by session (F_{9, 171} = 3.1, p < 0.01), nicotine (F_{1, 19} = 23.0, p < 0.001), and menthol (F_{1, 19} = 15.4, p < 0.001). There was also a significant interaction between nicotine and menthol (F_{1, 19} = 26.8, p < 0.001). The number of infusions did not significantly change across the sessions in the menthol-saline (F_{9, 36} = 1.2, p > 0.05) or vehicle-nicotine (F_{9, 45} = 0.5, p > 0.05) groups. On average, these control rats obtained < 5 infusions per session. The number of infusions in the vehicle-saline group significantly changed during the 10 daily sessions (F_{9, 45} = 2.6, p < 0.05), peaking in the sixth session (32.6 ± 5.9 infusions) and decreasing to 18.0 ± 2.9 infusions during the tenth session. The rats in the menthol-nicotine group significantly increased the number of infusions (F_{9, 45} = 3.3, p < 0.01) from 6.2 ± 1.0 infusions during the first session 1 to 10.0 ± 1.5 during the sixth session, and the number of infusions remained greater than 10 thereafter. Thus, the vehicle-saline group obtained a significantly greater number of infusions than the menthol-nicotine group (F_{1, 10} = 23.5, p < 0.001), menthol-saline group (F_1,9 = 32.4, p < 0.001), and the vehicle-nicotine group (F_{1, 10} = 39.0, p < 0.001), suggesting that both menthol and nicotine limited the number of infusions. However, the number of infusions obtained by the menthol-nicotine group was significantly greater than that obtained by the menthol-saline (F_1,9 = 12.0, p < 0.01) and vehicle-nicotine control groups (F_{1, 10} = 13.2, p < 0.01), indicating that contingent presentations of menthol with nicotine enhanced the reinforcing effect of nicotine.

Figures 1B,D–F show the numbers of active and inactive licks by each group. We transformed the numbers of licks to a logarithmic scale to fit a normal distribution. The gradual increase in nicotine intake (Figure 1A) in the menthol-nicotine group was driven by the significant increase in the number of licks on the active spout across the sessions (F_{9, 45} = 4.8, p < 0.001). In contrast, the group of rats yoked to these menthol-nicotine rats (Figure 1C) significantly reduced the number of licks on the active spout across the sessions (F_{9, 45} = 3.1, p < 0.01). Consequently, the yoked rats emitted significantly less active licks compared to their masters (F_{1, 10} = 18.1, p < 0.01). In agreement with Figure 1A, none of the control groups exhibited a significant change in the number of licks across the sessions (p > 0.05 for all). With the exception of the vehicle-saline group (F_{1, 50} = 174.3, p < 0.001), none of the other groups showed a preference for the active spout (p > 0.05 for all).

Menthol induces a multimodal sensory stimulation, including strong odor and taste. We were unable to find a chemical that mimics the odor and taste of menthol that does not simultaneously induce a cooling sensation. Assuming that aversive taste or odor is unlikely to support nicotine intake, we examined the general effects of contingent appetitive odor and taste cues on nicotine IVSA. The rats exhibited a strong preference for the active spout when grape odor was paired with an i.v. saline infusion (Figure 2A, F_{1, 60} = 110.6, p < 0.001). On average, 15.8 ± 2.0 infusions were obtained during the 10 daily sessions (effect of session: F_{9, 54} = 1.5, p > 0.05). However, when grape odor was paired with i.v. nicotine infusions, the rats strongly avoided the active spout (Figure 2B, F_{1, 50} = 82.3, p < 0.001). On average, 1.7 ± 0.26 infusions were obtained during the 10 sessions (effect of session: F_{9, 45} = 1.5, p > 0.05).

We then tested a saccharin/glucose mixture, which incites highly appetitive behavior in rodents (Smith et al., 1976). The rats licked the active spout ~10,000 times after five sessions when i.v. saline was delivered (Figure 2C, effect of spout: F_{1, 40} = 466.0, p < 0.001). On average, the rats obtained 152.0 ± 23.3 infusions per session (effect of session: F_{9, 36} = 6.8, p < 0.001). However, the rats did not prefer the active spout when this solution was delivered contingently with nicotine (Figure 2D, F_{1, 40} = 2.5, p > 0.05). On average, the rats obtained 8.5 ± 2.1 infusions. The number of infusions peaked on session three (24.3 ± 13.4) and then significantly decreased (effect of session: F_{9, 45} = 2.1, p < 0.05) to 4.2 ± 0.2 for the last three sessions.

Cooling, the prominent sensory property of menthol, is mediated by the TRPM8 channel (Voets et al., 2004). The WS-23 compound also stimulates the TRPM8 channel and has been reported to have virtually no taste or odor (Gaudin et al., 2008). We nevertheless used an odor habituation test (Inagaki et al., 2010) to examine whether WS-23 has an odor that can be detected by rats. There was a significant reduction in the number of nose pokes observed for 0.01% menthol from day 1 to day 2 (Figure 3, p < 0.01), reflecting habituation of the rats to the odor of menthol. In contrast, the number of nose pokes for water did not change between the two test sessions (p > 0.05). Furthermore, significantly fewer nose pokes were observed for water compared to menthol on day 1 (p < 0.05). These data established the validity of the assay.

The number of nose pokes for 0.03% WS-23 was significantly reduced between the two test sessions (p < 0.05). The number of nose pokes for 0.03% WS-23 was not different from that for menthol (p > 0.05). Although the number of nose pokes for 0.03% WS-23 was not significantly different from that for water (p > 0.05), the overall data suggested that 0.03% WS-23 is likely to emit an odor that can be detected by rats. The number of nose pokes for 0.01% WS-23 was significantly lower than that for menthol (p < 0.01), not different from that for water (p > 0.05), and did not change between the two test sessions (p > 0.05). These data indicated that 0.01% WS-23 had no detectable odor.

We then tested whether WS-23 supports i.v. nicotine intake (Figure 4). The rats that self-administered saline with WS-23 as the cue exhibited a preference for the active spout (F_{1, 90} = 214.7, p < 0.001). The number of infusions did not significantly change across the sessions (F_{9, 81} = 1.6, p > 0.05). The rats that self-administered nicotine with 0.01% WS-23 as the cue exhibited a strong preference for the active spout (Figure 4B. F_{1, 70} = 89.0, p < 0.001). The number of infusions increased from 8.6 ± 1.7 in session 1 to 13.9 ± 1.7 in session 10 (effect of session: F_{9, 63} = 1.7, p > 0.05). The rats that self-administered nicotine with 0.03% WS-23, which had a detectable odor, increased the number of nicotine infusions from 4.0 ± 0.8 in session 1 to 12.4 ± 1.4 in session 10 (effect of session: F_{9, 54} = 11.4, p < 0.001). These two WS-23 groups had similar number of active licks (F_{1, 13} = 3.6, p > 0.05) and nicotine infusions (F_{1, 13} = 1.3, p > 0.05), but the high concentration group had significantly greater inactive licks than the low concentration group (F_{1, 13} = 7.8, p < 0.05).

To further confirm that the cooling sensation is a conditioned cue for nicotine reward, we tested cold water (~11°C) as the sensory cue. Similar to 0.01% WS-23, the use of cold water as a cue for nicotine IVSA resulted in a significantly greater number of active licks compared to inactive licks (Figure 4D, F_{1, 64} = 62.2, p < 0.001). Furthermore, the number of nicotine infusions tended to increase across the sessions (F_{9, 54} = 1.9, p = 0.07).

We then tested the effect of a composite cue that includes both cooling and olfactogustatory stimulations on nicotine IVSA (Figure 5). The rats received a composite cue that contained WS-23 (0.01%), saccharine (0.4%) and grape-flavored Kool-Aid® (0.1%). The rats licked more on the inactive spout compared to the active spout (F_{1, 70} = 39.1, p < 0.001). The numbers of active licks (F_{9, 63} = 4.7, p < 0.001) and nicotine infusions (F_{9, 63} = 5.6, p < 0.001) both significantly increased across the sessions. However, the number of inactive licks did not change (F_{9, 63} = 1.3, p > 0.05). On average, the rats obtained 14.8 ± 0.87 infusions during the last 3 sessions.

Our previous study (Chen et al., 2011) suggested that an increased number of inactive licks is potentially associated with an aversive response to the stimuli. The affective value of the oral stimuli can be measured by the size of the lick clusters, defined as the number of licks within each round of licking (Davis and Smith, 1992). We therefore correlated the ratio of active/inactive licks with the size of the lick clusters and found a strong correlation (Figure 6, r = 0.75, p < 0.0001) among the rats that received i.v. saline with a variety of sensory cues (menthol, WS-23, saccharine + glucose, and grape flavor). One major advantage of the lick ratio over the cluster size is its wide dynamic range (2⁵–2¹⁰), which is several orders of magnitude greater than that of the lick cluster size (2¹–2⁵). The wide dynamic range is especially advantageous when studying aversive stimuli, when the size of lick cluster is compressed to a very narrow range of 2–6 licks per cluster. A moderate, albeit highly significant, correlation (r = 0.40, p < 0.0001) between the lick ratio and lick cluster size was found in rats self-administering i.v. nicotine with these sensory cues. This reduced correlation in rats that self-administered nicotine is likely because nicotine reduced the size of the lick cluster to the lower end of its narrow dynamic range.

We tested whether audiovisual cues could enhance the preference for the active spout when menthol was used as the contingent sensory cue for nicotine. We first tested the effect of an audiovisual cue on nicotine IVSA in rats received oral vehicle cue (Figure 7A). We found a significant interaction between the effect of the spout and that of the sessions (F_{9, 50} = 3.5, p < 0.01). There were fewer active licks than inactive licks for the first five sessions (F_{1, 25} = 19.4, p < 0.001), and the number of active licks was significantly greater than that of inactive licks for the subsequent five sessions (F_{1, 25} = 10.1, p < 0.01). The number of infusions significantly increased across the sessions (F_{9, 45} = 5.4, p < 0.001). On average, 3.7 ± 0.5 and 14.1 ± 1.9 infusions were obtained during the first and last three sessions, respectively. Compared to the group that self-administered nicotine with a vehicle cue but without an audiovisual cue (Figure 1E), the addition of an audiovisual cue had no effect on the number of inactive licks (F_{1, 10} = 2.5, p > 0.05) but significantly increased the numbers of active licks (F_{1, 10} = 6.5, p < 0.05) and nicotine infusions (F_{1, 10} = 8.4, p < 0.05).

A second control group received i.v. saline with a combined audiovisual and menthol cue (Figure 7B). The contingent audiovisual cue resulted in a preference for the active spout (F_{1, 60} = 46.9, p < 0.001). The number of infusions did not significantly change across the sessions (F_{9, 45} = 1.3, p > 0.05). Compared to rats that received the menthol cue without the audiovisual cue (Figure 1D), the audiovisual cue did not have a significant effect on the number of inactive licks (F_1,10 = 2.6, p > 0.05) but significantly increased the numbers of active licks (F_1,10 = 5.4, p < 0.05) and saline infusions (F_{1, 10} = 5.9, p < 0.05).

The rats preferred the active spout when i.v. nicotine was self-administered with a combined audiovisual and menthol cue (Figure 7C, F_{1, 50} = 41.8, p < 0.001). The effect of the session on the number of infusions was statistically significant (F_9,45 = 3.3, p < 0.01). The number of infusions increased from 4.0 ± 0.35 during the first three sessions to 11.8 ± 0.68 during the last three sessions. Compared to the menthol-nicotine group without the audiovisual cue (Figure 1B), the audiovisual cue significantly reduced the number of inactive licks (F_1,10 = 6.7, p < 0.05) but did not significantly change the number of active licks (F_{1, 10} = 0.42, p > 0.05) or nicotine infusions (F_1,10 = 0.1, p > 0.05), indicating that the audiovisual cue enhanced the affective value of the complex stimuli that the rat received but not the amount of nicotine intake.

Figure 8 compares the mean number of nicotine infusions self-administered by rats that received different sensory cues during the last three IVSA sessions. The rats that received menthol, WS-23, or audiovisual cues self-administered similar amounts of nicotine (Tukey's HSD, ps > 0.05). All of these groups obtained a significantly greater number of infusions compared to the groups that received vehicle, saccharin-glucose, or the grape-flavored cues (Tukey's HSD, ps < 0.001 to 0.05). The amount of nicotine intake by the cold water group was significantly greater than that of the grape-flavored group (Tukey's HSD, p < 0.05).

Two additional groups of rats self-administered saline or nicotine with contingent menthol cues for 10 sessions (Figure 9). These groups replicated the majority of the results shown in Figures 1A,B. The rats in the menthol-saline group did not show a preference for either spout (F_{1, 50} = 1.4, p > 0.05), and the number of infusions did not significantly change across the sessions (F_{9, 45} = 1.4, p > 0.05). On average, very few infusions were obtained (4 ± 1.6). In contrast, the menthol-nicotine group increased the number of infusions across sessions (F_{9, 45} = 5.0, p < 0.001). The average number of infusions during the last three sessions was 9.1 ± 0.6. In contrast to the data shown in Figure 1B, in which the numbers of active and inactive licks were not different, the menthol-nicotine group exhibited a significantly greater number of inactive licks compared to active licks (F_{1, 50} = 57.5, p < 0.001).

The average number of sessions required to reach the extinction criteria was 2.8 ± 0.8 for the menthol-saline group, whereas the menthol-nicotine group required 6.2 ± 0.8 sessions (p < 0.05). In the saline group, the number of active spout licks was not significantly different between the extinction and IVSA sessions (F_{3, 15} = 1.7, p > 0.05). In contrast, the nicotine group exhibited a significantly increased number of licks on the active spout during the extinction sessions (F_{3, 15} = 45.0, p < 0.001). The average number of licks on the active spout increased six-fold (from 119.5 ± 21.2 in the last IVSA session to 720.8 ± 63.5 in the first extinction session, p < 0.001, Tukey's HSD). During the second and third extinction sessions, the numbers of active spout licks in the nicotine group were 3.8–(p < 0.01, Tukey's HSD) and 3.5-fold (p < 0.05, Tukey's HSD) higher, respectively, than that in the last IVSA session.

The effect of menthol on drug-seeking behavior was tested over five consecutive sessions. Both groups displayed a preference for the active spout during the reinstatement sessions (F_{1, 25} = 16.4, p < 0.001 for the saline group and F_{1, 25} = 34.8, p < 0.001 for the nicotine group). The number of licks on the active spout during the reinstatement tests was significantly greater than that for the last five IVSA sessions in the nicotine group (F_{9, 45} = 4.5, p < 0.001), but not in the saline (F_{9, 45} = 0.6, p > 0.05) group. The number of active licks during the first reinstatement session was 862.0 ± 214.2, which was 7.2-fold greater than that during the last IVSA session in the nicotine group. On average, the number of active licks stayed within ~5-fold of that observed in the last IVSA session and did not significantly change during the reinstatement sessions (F_{4, 20} = 0.5, p > 0.05). Additionally, the number of active licks by the nicotine group during the reinstatement session was 3.6 ± 0.4-fold greater, on average, than that observed in the saline group (F_{1, 10} = 11.5, p < 0.01).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.