Introduction

The placebo effect is a psychobiological phenomenon that can be attributable to different mechanisms, including expectation of clinical improvement and Pavlovian conditioning (Pollo and Benedetti 2009; Thompson 2000). The neurobiology of placebo, an understanding of the physiological substrate beyond psychological theories, was born with the discovery that naloxone can reverse placebo analgesia (Amanzio and Benedetti 1999; Benedetti et al. 1999; Levine et al. 1978). Neuroimaging studies have also revealed the brain basis of placebo effects on pain and emotion regulation (Petrovic et al. 2005; Wager et al. 2004). But in most studies, the investigators only focused on the placebo effect obtained within a single domain. That is, they either studied the analgesic effect of a pain-alleviating expectation or the ataractic effect of an anxiety-reducing expectation.

For a given placebo effect such as the analgesic effect, however, its effective scope might be more general than pain alleviating. It was not only because of the established psychological expectation that could be easily generalized and transferred to other domains, but also because of the fact that different kinds of stress and their corresponding regulation shared the similar biological mechanism. For example, substantial evidence implicates the endogenous opioid system as playing a crucial role in the mediation of placebo analgesia (Benedetti 2007; Eippert et al. 2009; Wager et al. 2007). The serotonergic system is known to modulate mood and emotion (Maier and Watkins 2005). However, the opioid system has been implicated in mood disorders, and there may be a neurobiological link between opioids and serotonin (Ide et al. 2010). Neurons producing serotonin contribute to descending pain control pathways by morphine (Millan 2002), and the central serotonergic system is a key component of supraspinal pain modulatory circuitry mediating opioid analgesia (Zhao et al. 2007). Thus, it is quite reasonable for us to infer that the placebo responses in one domain (pain) can affect another domain (emotion).

Consistent with this hypothesis, recent studies by our laboratory (Zhang and Luo 2009; Zhang et al. 2011) showed that the placebo effect in humans can be transferred from one domain to the other, namely from pain to emotion. A significant transferable placebo effect that alleviated negative feelings was observed. EEG recordings showed that the transferable placebo treatment induced decreased P2 amplitude and increased N2 amplitude, with a source location near the posterior cingulate, and fMRI results indicated this transferable placebo treatment, relative to the control condition, was associated with reduced activity in the amygdala and insula and increased activity in the subgenual anterior cingulate cortex known to be important in emotional regulation.

Rodents could also display associative learning and associate the cue condition with the placebo effect by special procedures (Fanselow and Poulos 2005; Roberts et al. 2008). In the previous study, we have evoked opioid and non-opioid placebo responses in mice that were either naloxone reversible or naloxone insensitive, depending on the drug used in the conditioning procedure (Guo et al. 2010). These procedures of mice may serve as a model for investigating placebo analgesia. Therefore, we further investigated whether the placebo effect could be transferred from pain (placebo analgesia) to the other domain (depressive-like behaviors) in placebo analgesic mice. Mice were firstly trained to associate cue with elevated paw latency as previously described. Then the forced swim test (FST) and tail suspension test (TST) were used to measure the depressive-like behaviors on the test day. Mice treated with the classical antidepressant, clomipramine, were used as the positive control. Plasma levels of corticosterone and adrenocorticotropic hormone (ACTH) in mice were also measured in response to FST and TST, in order to see whether there are any differences in hypothalamus–pituitary–adrenal axis (HPA) activity.

Materials and method

Animals

Female C57/BL6 mice weighing 18–22 g at the start of the experiment were used. The animals were housed in groups of four in polycarbonate tubs (43 × 23 × 18 cm) with pine sawdust bedding. The housing rooms were kept in a 12:12 h day–night cycle (lights on at 7:00 AM) and at an ambient temperature of 20–22°C. Food and water were available ad libitum. Animals were handled for at least 2–3 days prior to conducting the experiments. All experiments followed the Guidelines on Ethical Standards for Investigation of Experimental Pain in Animals and were approved by the Institutional Animal Care and Use Committee of the Institute of Psychology of the Chinese Academy of Sciences.

Drugs and treatment

All drugs were administered via intraperitoneal injection (ip). The placebo analgesia mice were administrated with morphine hydrochloride (10 mg/kg) at the animal training procedure; the drug (Sigma, USA) was diluted in a sterile solution of NaCl 0.9%. A group of animals received a treatment with clomipramine hydrochloride (1, 5, or 50 mg/kg) as described previously with some modifications (Drago et al. 2001). Clomipramine hydrochloride (Sigma, USA) was freshly diluted in a sterile solution of NaCl 0.9%, and the mice received three such injections 24, 5, and 1 h prior to behavioral testing. A group of mice treated with NaCl 0.9% were used as controls.

Animal training procedure

The animal training procedure was performed as previously described (Guo et al. 2010). Briefly, mice were treated with saline on day 1 AM and day 2 PM and with morphine on day 1 PM and day 2 AM. They were then put into distinct cue chambers for 30 min before receiving the hot plate test. The same procedure was repeated on days 3 and 4. On day 5, they were treated with saline at 8:00 AM but placed into the cue compartments previously paired with morphine for 30 min before the FST and TST. To ensure the placebo analgesia was induced by the present procedure, one group of mice as placebo analgesia were tested with the hot plate after being injected with saline and placed into the cue compartments previously paired with morphine for 30 min on day 5 AM. The number of animals per group was 12, and those mice scoring less than 5 s or more than 60 s in the pretest were rejected. Mice having burnt paws were also rejected for they affect the results of the TST and FST.

Tail suspension test

The total duration of immobility induced by tail suspension was measured according to the methods described by Steru et al. (1985). Briefly, the mice both acoustically and visually isolated were suspended 50 cm above the floor by an adhesive tape placed approximately 1 cm from the tip of the tail. Immobility time was recorded during a 6-min period (Carbajal et al. 2009; Gehlert et al. 2009). The mice were considered immobile only when they hung passively and were completely motionless. The number of animals per group was 12. The test sessions were recorded by a video camera and scored by an observer blind to the treatment.

Forced swimming test

Mice were individually forced to swim in an open cylindrical container (diameter, 10 cm; height, 25 cm), containing 19 cm of water at 25 ± 1°C. The total duration of immobility was recorded during the last 4 min of the 6-min period. Each mouse was judged to be immobile when it ceased struggling and remained floating motionless in the water, making only those movements necessary to keep its head above water. A decrease in the duration of immobility is indicative of an antidepressant-like effect (Porsolt et al. 1978, 1977). The number of animals per group was 12. The test sessions were recorded by a video camera and scored by an observer blind to the treatment.

Corticosterone and ACTH concentrations

After the behavioral procedures were completed, the animals were immediately killed by decapitation. Plasma concentrations of corticosterone and ACTH were measured by ELISA as previously reported (Guo et al. 2009). Briefly, the trunk blood was collected, EDTA was added, and samples were centrifuged at 800×g for 15 min; plasma was collected and stored at −20°C until assay. Corticosterone and ACTH were extracted from the plasma, added to ethanol, and measured by ELISA. The ELISA kits' cross-reactivity with other steroids was below 0.01%. The corticosterone assay sensitivity was 5 ng/ml. Intra- and inter-assay coefficients of variation were lower than 5% and 8%, respectively. The ACTH assay sensitivity was 10 pg/ml. Intra- and inter-assay coefficients of variation were lower than 5% and 8%, respectively.

Statistical analysis

All data are expressed as means ± SEM. Significance of difference between the two groups was evaluated using Student's t test. One-way analysis of variance (ANOVA) followed by Dunnett's test was used for multiple comparisons. A P value of 0.05 or less was considered as indicative of a significant difference.

Results

Placebo analgesia

A significant difference was shown when the paw latency was measured for five consecutive days (F(8,88) = 12.73, P < 0.001). The mean paw latency after saline injection on day 1 at 8:30 AM and day 2 at 8:30 PM was 17.95 ± 1.14 and 17.76 ± 1.88 s, respectively. When morphine was administered on day 1 at 8:00 PM and day 2 at 8:00 AM, a significant increase in pain tolerance was found (33.21 ± 3.65 and 35.54 ± 3.84 s, respectively) (P < 0.01). Similar results were obtained on days 3 and 4 (P < 0.01). After saline treatment on day 5 at 8:00 AM and a 30-min exposure to the cue compartment in group 4 mice, pain tolerance was significantly elevated (27.72 ± 3.62 s) compared with day 1 AM (the baseline, P < 0.01), indicating that the previous morphine conditioning was sufficient to evoke a placebo effect (Fig. 1).

Fig. 1
figure 1

Morphine-induced placebo effect. After the procedure of morphine conditioning on days 1–4, mice were injected with saline and put into the conditioned cue box for 30 min on day 5 at 8:00 AM. Paw withdrawal latency was significantly elevated, which mimics the morphine analgesic response. **P < 0.01, compared with day 1 AM (control condition)

Full size image

Tail suspension test

Figure 2 shows the effects of placebo analgesia in the TST. One-way ANOVA revealed a significant level of variance for treatment (F(4,55) = 10.59, P < 0.01). As expected, clomipramine at the doses of 1, 5, and 50 mg/kg reduced, in a dose-dependent manner, the duration of immobility in the TST, resulting in a 55%, 69%, and 81% immobility reduction compared with that of the control group, respectively. Placebo analgesia also induced a decrease in immobility that was significantly different from that of the control group (P < 0.05). The comparison of the antidepressant-like actions of placebo analgesia with those induced by clomipramine showed that placebo analgesia produced a more pronounced antidepressant-like effect than that achieved with 1 mg/kg clomipramine (80.96 ± 13.81 vs. 62.1 ± 12.5 s), but comparable to that produced by the 5 mg/kg dose of clomipramine (80.96 ± 13.81 vs. 78.90 ± 26.37 s)

Fig. 2
figure 2

Tail suspension test monitored for 6 min. The sum of the total scores of immobility are reported in seconds. Clomipramine was administered 1, 5, and 24 h before the tail suspension test. Data analysis was performed using Dunnett's t test.*P < 0.05, **P < 0.01, compared with control

Full size image

Forced swimming test

One-way ANOVA revealed a significant level of variance for treatment (F(4,55) = 13.70, P < 0.01). Similar results were obtained in the FST; clomipramine also markedly reduced the FST-induced immobility time in a dose-dependent manner (Fig. 3). Placebo analgesia significantly decreased the FST-induced immobility time in mice (P < 0.05). The comparison of the antidepressant-like actions of placebo analgesia with those induced by clomipramine also showed that placebo analgesia produced a more pronounced antidepressant-like effect than that achieved with 1 mg/kg clomipramine (82.96 ± 22.95 vs. 58.94 ± 19.96 s), but comparable to that produced by the 5 mg/kg dose of clomipramine (82.96 ± 22.95 vs. 86.01 ± 20.46 s).

Fig. 3
figure 3

Porsolt's forced swim test for 6 min. Scores of immobility in seconds are represented for the last 4 min of each trial. Clomipramine was administered 1, 5 and 24 h before the forced swim test. Data analysis was performed using Dunnett's t test.*P < 0.05, **P < 0.01, compared with control

Full size image

Corticosterone and ACTH concentrations

A one-way ANOVA indicated significant differences among groups in plasma concentrations of corticosterone and ACTH after TST (F(4,55) = 4.06, P < 0.01; F(4,55) = 4.04, P < 0.01, respectively). Clomipramine significantly reduced the TST-induced increase in the plasma concentrations of corticosterone and ACTH in a dose-dependent manner, as shown in Fig. 4. Placebo analgesia also markedly decreased the TST-induced plasma concentrations of corticosterone and ACTH compared with control group (both P < 0.05). Similar results were obtained in the FST, the main difference being only the higher sensitivity of this test for placebo analgesia effects compared with the TST. The plasma concentration of corticosterone in FST mice was significantly higher than that in TST (P < 0.05). One-way ANOVA indicated that the plasma concentrations of corticosterone and ACTH significantly differed among groups after FST (F(4,55) = 4.58, P < 0.01; F(4,55) = 5.47, P < 0.01, respectively). Thus, in the FST, clomipramine also significantly decreased the FST-induced increases in the plasma concentrations of corticosterone and ACTH in a dose-dependent manner. Placebo analgesia also markedly reduced the FST-induced plasma concentrations of corticosterone and ACTH compared with the control group (both P < 0.05).

Fig. 4
figure 4

Tail suspension test and forced swim test-induced hormonal secretions. After behavioral procedures were completed, mice were immediately killed by decapitation. Plasma concentrations of corticosterone and ACTH were measured by ELISA. a Plasma corticosterone concentration (nanograms per milliliter), b plasma ACTH concentration (picograms per milliliter). Data analysis was performed using Dunnett's t test. *P < 0.05, **P < 0.01, compared with control; # P < 0.05, control in FST vs. control in TST

Full size image

Discussion

The present study confirmed previous observations that mice can learn to associate the context cue with elevated pain tolerance via a set of procedures. After mice were given 4 days of drug conditioning with the conditioned cue stimulus and the unconditioned drug stimulus, saline injection with the contextual cue could produce placebo analgesia at day 5. Then we test whether placebo analgesia might affect the depressive-like behaviors in mice. Consistent with the previous research in humans (Zhang and Luo 2009), this study also showed that the placebo analgesia established from pain alleviation may affect the depressive-like behaviors in mice. The placebo analgesia, which was established by a set of procedures in mice, was transferable and could produce a significant antidepressant effect on the depressive test. Plasma levels of corticosterone and ACTH further proved that the placebo analgesia that was established from the pain-reducing training not only induced a significant placebo effect on pain, but also decreased significantly the HPA response to stress and produced a stress-alleviating effect.

The tail suspension and forced swimming tests are two widely accepted behavioral models for assessing pharmacological antidepressant activity (Bourin et al. 2005; Porsolt 1979). The characteristic behavior scored in these tests is termed immobility, reflecting behavioral despair as seen in human depression. These two despair tests are the most used experimental models of depression both in strategic and routine studies, and it is well known that the antidepressant drugs are able to reduce the immobility time in rodents (Porsolt et al. 1978). In the present experiments, we tested the antidepressant effect of placebo analgesia with TST and FST. Mice treated with clomipramine hydrochloride in three doses were used as the positive controls. Our results showed that the immobility time of placebo analgesia mice were significantly decreased compared with that of the control mice. The immobility times of placebo analgesia mice in the TST and FST were 80.96 ± 13.81 and 82.96 ± 22.95 s, respectively, which were comparable to that produced by the 5 mg/kg dose of clomipramine hydrochloride. These results showed that placebo analgesia could affect the depressive tests and produced a stronger antidepressant effect in mice.

The HPA axis is a three-gland component of the endocrine system that modulates biological responses to acute and chronic stress (Gourley et al. 2008; Johnson et al. 2006). Immediately following stress, the HPA axis activity increases, initiated by the release of corticotropin-releasing hormone (CRH) from neurons of the paraventricular nuclei of the hypothalamus. CRH, in turn, stimulates ACTH release from the anterior pituitary, and ACTH stimulates corticosterone release from the adrenal cortex. Therefore, ACTH and corticosterone are considered as the markers of stress (Tanke et al. 2008). As Cryan and Holmes (2005) have pointed out, there are different neurobiological substrates underlying the behavioral responses of mice subjected to TST and FST. We found that plasma ACTH and corticosterone concentrations in FST mice were higher than those in TST mice, suggesting that FST might be a much stronger stressor than TST. The FST and TST have different sensitivities to antidepressant treatments, and it is thought that the two tests measure changes in different neurotransmitter systems. The FST is primarily associated with the dopaminergic and serotonergic systems, while the TST is associated with the adrenergic system (Martin and Brown 2009). Our findings that the stress-induced increase in plasma corticosterone and ACTH concentrations are suppressed by the administration of clomipramine are in line with other authors' results (Aulakh et al. 1993; Martin and Brown 2009; Marzouk et al. 1991). Placebo analgesia also decreased the stress-induced corticosterone and ACTH concentrations. These data suggest that placebo analgesia might produce an antidepressant effect via decreasing the HPA axis response to stress.

The question whether the placebo effect depends more on classical conditioning or on cognitive processes was directly examined in several studies (Benedetti et al. 2003a, b; Price et al. 1999). Conditioning theorists have proposed that the placebo is a conditioned Pavlovian response, whereas others have advocated that the placebo is driven by expectancy (Kirsch 2004; Klinger et al. 2007; Rescorla 1988; Stewart-Williams and Podd 2004; Wager and Nitschke 2005). The placebo analgesia effect investigated in this study might be brought about mainly through expectation. The typical procedure for inducing classical Pavlovian conditioning involves presentations of a neutral stimulus along with a stimulus of some significance. This procedure of placebo response is the same as the typical experimental setting. Drug conditioning was performed by the combination of the conditioned cue stimulus with the unconditioned drug stimulus. With 4 days of drug conditioning procedure, placebo analgesic responses were evoked by an exposure to a conditioned cue previously paired with drug conditioning. However, we examined the depressive behaviors in FST and TST but not the pain tolerance on the test day. The plasma concentration of ACTH and CORT were also measured. We found that the placebo analgesia not only increases the immobility time but also decreases the hormonal secretions in mice. A decreased HPA axis response to stress might account for this antidepressant effect, which were different from those trained in the previous conditioning procedure. As rodents may have expectation. For example, conditional responses in rodents such as locomotion have been reported for drugs of abuse and similar to the placebo response in humans, were considered to be associated with the expectation of reward (Bryant et al. 2009). Our results suggest that expectation might be involved in this sort of conditioning. Therefore, it is reasonable to conclude that it is not just unconscious conditioning but also conscious expectation that primarily contributes to the placebo analgesia on depressive-like behaviors. This is in accordance with alternative theories of learning which suggest that cognitive elements are involved in Pavlovian conditioning. In other words, conditioning would lead to the expectation that a given event will follow another event, and this occurs on the basis of the information that the conditioned stimulus provides about the unconditioned stimulus (Benedetti et al. 2003a; Rescorla 1988). As Montgomery and Kirsch (1997) suggested, a conditioned placebo response can result from conditioning but is actually mediated by expectancy.

In the present study, we found that a prior positive experience (placebo analgesia induced by expectation) affected the plasma concentrations of ACTH and corticosterone in mice. A previous study by Benedetti et al. (2003a) showed that expectations are capable of influencing conscious physiological processes such as pain and motor performance; they have no effect on unconscious biological functions such as hormonal secretion, which is affected, however, by a conditioning procedure. This might be because the plasma concentrations of the hormone were increased as the mice were tested in FST and TST, while Benedetti et al. (2003a) measured the hormone concentration in the normal condition.

We just investigate the effect of the placebo analgesic on the depressive-like behavior tests in mice. An antidepressive effect of the placebo analgesia was found in this study. However, another factor, such as adaptation or habituation to a stressful procedure, might be involved in the antidepressant effect. Therefore, the placebo analgesic might produce an unspecific and general stress-reduction effect that may include, but not limit to, the antidepressive effect demonstrated in the present study. Other stress-related tests in placebo analgesia mice remain to be studied in the future.

It is proposed that a hyperalgesic conditioned response could be found following pharmacological conditioning with morphine. However, we did not observe this phenomenon in the present experiment. This might be because we alternately injected the morphine or saline to mice once at AM and PM, which is different from previous procedures (Bespalov et al. 2001). In addition, presenting environmental cues previously associated with morphine, but without the drug, could also attenuate established hyperalgesia (Siegel 1977).

A limiting factor in this study is the fact that placebo analgesia induced by one dose of morphine was evaluated. However, it should be considered that the effect of the size of placebo analgesia appears to differ systematically according to the purpose of the placebo condition. We could use a different dose of morphine in the training process; then it might produce different sizes of placebo analgesia. In addition, a different number of trials during the conditioning can actually affect the magnitude of placebo analgesic responses (Colloca et al. 2010). Thus, we might observe the effect of different sizes of placebo analgesia by different doses or number of trials on depressive-like behavior with TST and FST, which might strengthen our present finding.

In summary, the present results showed that placebo analgesia might affect the depressive behavior in mice. This is the first report that shows that the placebo effect can be transferred from one domain to the other in mice. The placebo analgesia induced by analgesic experiences produced an antidepressant effect in TST and FST, and decreased the HPA response to stress with reducing hormonal secretion in mice.