Antihyperalgesic Activity of Atomoxetine on Diabetes-Induced Neuropathic Pain: Contribution of Noradrenergic and Dopaminergic Systems

Atomoxetine is a selective noradrenaline reuptake inhibitor drug. Based on the knowledge that agents increasing monoamine levels in the central nervous system have therapeutic potential for neuropathic pain, it is planned to investigate the possible efficacy of atomoxetine on diabetes-induced hyperalgesia, in this study. Randall-Selitto (mechanical noxious stimuli) and Hargreaves (thermal noxious stimuli) tests were used to evaluate nociceptive perception of rats. Obtained data indicated that streptozotocin-induced diabetes causes significant decreases in the paw withdrawal threshold and paw withdrawal latency values of the animals, respectively. However, atomoxetine administered at 3 mg/kg/day for 7 and 14 days improved these diabetes-induced hyperalgesia responses. Furthermore, antihyperalgesic activity was antagonized with α-methyl-para-tyrosine methyl ester, phentolamine, propranolol, and sulpiride pre-treatments. The same effect was not reversed, however, by SCH 23390. These findings demonstrated, for the first time, that atomoxetine possesses significant antihyperalgesic activity on diabetes-induced neuropathic pain and this effect seems to be mediated by α- and β-adrenergic and D2/D3 dopaminergic receptors. Results of this present study seem to offer a new indication for an old drug; atomoxetine, but these preclinical data should first be confirmed with further well-designed clinical trials.


Introduction
Chronic exposure to hyperglycemia is known to cause serious damage to the nervous system of patients with diabetes mellitus. This damage, which is directly induced by diabetes itself, but not related to any other inherited, traumatic, nutritional, metabolic, infectious, immunological, neoplastic, or toxic reasons, is called diabetic neuropathy (DN). Therefore, DN may simply be defined as the presence of peripheral nerve dysfunction-related symptoms in diabetic patients, after the exclusion of other possible reasons for neuropathy [1,2].
Diabetic neuropathy, as one of the long-term complications of diabetes, eventually develops in up to 50% of the patients and severely restricts their daily life [1]. The most common type of diabetic neuropathy is chronic sensorimotor peripheral polyneuropathy [3]. The clinical appearances of DN alter depending on the type of affected nerve fibres; sensory, motor, or autonomic. DN-induced damage to sensory neurons can cause either a "painful condition" characterized by allodynia (pain perception in response to non-painful stimuli) and hyperalgesia (exaggerated pain sensations to painful stimuli) or a "painless syndrome", with disappearance of sensation to pain, heat, or touch stimulus [4]. Figure 1 presents the effect of atomoxetine (3 mg/kg/day, 2 weeks) treatment on mechanical nociceptive stimulus-induced hyperalgesia responses in diabetic rats. Two-way repeated ANOVA analysis showed that both of the treatment [F (3,28) = 38.73, p < 0.001] and the time factors [F (3,84) = 59.21, p < 0.001] were effective on the paw withdrawal thresholds of rats measured in the Randall-Selitto tests. Besides, there was a significant interaction between the "treatment" and the "time" factors [F (9,84) = 18.14, p < 0.001]. Results acquired from the Bonferroni multiple comparison tests demonstrated that in all of the diabetic groups "paw withdrawal threshold" values of rats measured at 4th weeks were significantly lower than those measured before the induction of diabetes (Week 0) (Figure 1.). On the other hand, administration of atomoxetine at doses of 3 mg/kg for 7 (p < 0.001) and 14 (p < 0.001) days significantly prolonged the reduced paw withdrawal thresholds of diabetic rats. In addition, 14-day administrations of atomoxetine were found to be more effective than Figure 1. Paw withdrawal thresholds of normoglycemic rats administrated physiological saline (Control), and of diabetic rats administrated daily 10 mg/kg pregabalin (Pregabalin + DM) and daily 3 mg/kg atomoxetine (Atomoxetine + DM), in the Randall-Selitto test. Values are given as mean ± S.E.M. Within the groups: significant difference against Week 0, Week 4 and Week 5 groups are *** p < 0.001; c p < 0.001 and &&& p < 0.001, respectively. Between the groups: significant difference against Week 4 control, Week 5 control, Week 6 control, Week 5 DM and Week 6 DM groups are xxx p < 0.001; +++ p < 0.001, ééé p < 0.001, ^^^ p < 0.001 and <<< p < 0.001, respectively. Two-way repeated ANOVA, post hoc Bonferroni test, n = 8.

Atomoxetine Alleviates Diabetes-Induced Thermal Hyperalgesia
The effect of atomoxetine treatment (3 mg/kg/day, 2 weeks) on thermal nociceptive stimulusinduced hyperalgesia responses in diabetic rats was shown in Figure 2. Two-way repeated ANOVA analysis showed that both the treatment [F (3,28) = 62.31, p < 0.001] and the time factors [F (3,84) = 25.07, p < 0.001] were effective on the paw withdrawal latency values of rats measured in the Hargreaves tests. Besides, there was a significant interaction between the "treatment" and the "time" factors [F (9,84) = 11.59, p < 0.001]. Results acquired from the Bonferroni multiple comparison tests demonstrated that in all of the diabetic experimental groups "paw withdrawal latency" values of rats measured at 4th weeks were significantly lower than those measured before the induction of diabetes (Week 0) ( Figure 2). On the other hand, administration of atomoxetine at doses of 3 mg/kg for 7 (p < 0.001) and 14 (p < 0.001) days significantly prolonged the shortened paw withdrawal latency values of diabetic rats. In addition, pregabalin showed the expected antihyperalgesic efficacy following the 7 (p < 0.001) and 14 (p < 0.001) day-long treatments ( Figure 2). Figure 1. Paw withdrawal thresholds of normoglycemic rats administrated physiological saline (Control), and of diabetic rats administrated daily 10 mg/kg pregabalin (Pregabalin + DM) and daily 3 mg/kg atomoxetine (Atomoxetine + DM), in the Randall-Selitto test. Values are given as mean ± S.E.M. Within the groups: significant difference against Week 0, Week 4 and Week 5 groups are *** p < 0.001; c p < 0.001 and &&& p < 0.001, respectively. Between the groups: significant difference against Week 4 control, Week 5 control, Week 6 control, Week 5 DM and Week 6 DM groups are xxx p < 0.001; +++ p < 0.001, ééé p < 0.001,ˆˆˆp < 0.001 and <<< p < 0.001, respectively. Two-way repeated ANOVA, post hoc Bonferroni test, n = 8.

Atomoxetine Alleviates Diabetes-Induced Thermal Hyperalgesia
The effect of atomoxetine treatment (3 mg/kg/day, 2 weeks) on thermal nociceptive stimulus-induced hyperalgesia responses in diabetic rats was shown in Figure 2. Two-way repeated ANOVA analysis showed that both the treatment [F (3,28) = 62.31, p < 0.001] and the time factors [F (3,84) = 25.07, p < 0.001] were effective on the paw withdrawal latency values of rats measured in the Hargreaves tests. Besides, there was a significant interaction between the "treatment" and the "time" factors [F (9,84) = 11.59, p < 0.001]. Results acquired from the Bonferroni multiple comparison tests demonstrated that in all of the diabetic experimental groups "paw withdrawal latency" values of rats measured at 4th weeks were significantly lower than those measured before the induction of diabetes (Week 0) ( Figure 2). On the other hand, administration of atomoxetine at doses of 3 mg/kg for 7 (p < 0.001) and 14 (p < 0.001) days significantly prolonged the shortened paw withdrawal latency values of diabetic rats. In addition, pregabalin showed the expected antihyperalgesic efficacy following the 7 (p < 0.001) and 14 (p < 0.001) day-long treatments ( Figure 2). The effect of α-methyl-para-tyrosine methyl ester (AMPT) pre-treatment on atomoxetineinduced antihyperalgesic response against mechanical nociceptive stimuli applied in the Randall-Selitto test was shown in Figure 3A. The results of two-way ANOVA analysis showed significant effects of both atomoxetine treatment [F (1,28) = 29.70, p < 0.001] and AMPT administration [F (1,28) = 4.22, p < 0.05] on the paw withdrawal thresholds of diabetic rats. In addition, a significant interaction was detected between the "atomoxetine treatment" and "AMPT administration" factors [F (1,28) = 5.94, p < 0.05). Results of the Bonferoni multiple comparisons test showed that AMPT administrations significantly reversed the atomoxetine-induced antihyperalgesic responses in the Randall-Selitto tests (p < 0.01). Figure 3B shows the effect of AMPT pre-treatment on atomoxetine-induced antihyperalgesic responses against thermal nociceptive stimuli applied in the Hargreaves test. The results of two-way ANOVA analysis showed significant effects of both atomoxetine treatment [F (1,28) = 21.45, p < 0.001] and AMPT administration [F (1,28) = 16.54, p < 0.001] on the paw withdrawal latencies of diabetic rats. In addition, a significant interaction was detected between the "atomoxetine treatment" and "AMPT administration" factors [F (1,28) = 25.24, p < 0.001]. Results of the Bonferoni multiple comparisons test showed that AMPT administrations significantly reversed the atomoxetine-induced antihyperalgesic responses assessed in the Hargreaves test (p < 0.001).

Catecholaminergic System Mediates the Beneficial Effect of Atomoxetine on Diabetes-Induced Mechanical and Thermal Hyperalgesia
The effect of α-methyl-para-tyrosine methyl ester (AMPT) pre-treatment on atomoxetine-induced antihyperalgesic response against mechanical nociceptive stimuli applied in the Randall-Selitto test was shown in Figure 3A. The results of two-way ANOVA analysis showed significant effects of both atomoxetine treatment [F (1,28) = 29.70, p < 0.001] and AMPT administration [F (1,28) = 4.22, p < 0.05] on the paw withdrawal thresholds of diabetic rats. In addition, a significant interaction was detected between the "atomoxetine treatment" and "AMPT administration" factors [F (1,28) = 5.94, p < 0.05). Results of the Bonferoni multiple comparisons test showed that AMPT administrations significantly reversed the atomoxetine-induced antihyperalgesic responses in the Randall-Selitto tests (p < 0.01). Figure 3B shows the effect of AMPT pre-treatment on atomoxetine-induced antihyperalgesic responses against thermal nociceptive stimuli applied in the Hargreaves test. The results of two-way ANOVA analysis showed significant effects of both atomoxetine treatment [F (1,28) = 21.45, p < 0.001] and AMPT administration [F (1,28) = 16.54, p < 0.001] on the paw withdrawal latencies of diabetic rats. In addition, a significant interaction was detected between the "atomoxetine treatment" and "AMPT administration" factors [F (1,28) = 25.24, p < 0.001]. Results of the Bonferoni multiple comparisons test showed that AMPT administrations significantly reversed the atomoxetine-induced antihyperalgesic responses assessed in the Hargreaves test (p < 0.001).
The effect of phentolamine (a non-selective α-adrenergic receptor blocker agent) pre-treatments on atomoxetine-induced antihyperalgesic response against mechanical nociceptive stimuli applied in the Randall-Selitto test was shown in Figure 4A. The results of two-way ANOVA analysis showed significant effects of both atomoxetine treatment [F (1,28) = 33.18, p < 0.001] and phentolamine administrations [F (1,28) = 6.34, p < 0.05] on the paw withdrawal thresholds of diabetic rats. In addition, a significant interaction was detected between the "atomoxetine treatment" and "phentolamine administration The effect of phentolamine (a non-selective α-adrenergic receptor blocker agent) pre-treatments on atomoxetine-induced antihyperalgesic response against mechanical nociceptive stimuli applied in the Randall-Selitto test was shown in Figure 4A. The results of two-way ANOVA analysis showed significant effects of both atomoxetine treatment [F (1,28) = 33.18, p < 0.001] and phentolamine administrations [F (1,28) = 6.34, p < 0.05] on the paw withdrawal thresholds of diabetic rats. In addition, a significant interaction was detected between the "atomoxetine treatment" and "phentolamine administration" factors [F (1,28) = 8.66, p < 0.01]. Results of the Bonferoni multiple comparisons test showed that phentolamine administration significantly reversed the atomoxetineinduced antihyperalgesic responses in the Randall-Selitto tests (p < 0.01).    The effect of phentolamine (a non-selective α-adrenergic receptor blocker agent) pre-treatments on atomoxetine-induced antihyperalgesic response against mechanical nociceptive stimuli applied in the Randall-Selitto test was shown in Figure 4A. The results of two-way ANOVA analysis showed significant effects of both atomoxetine treatment [F (1,28) = 33.18, p < 0.001] and phentolamine administrations [F (1,28) = 6.34, p < 0.05] on the paw withdrawal thresholds of diabetic rats. In addition, a significant interaction was detected between the "atomoxetine treatment" and "phentolamine administration" factors [F (1,28) = 8.66, p < 0.01]. Results of the Bonferoni multiple comparisons test showed that phentolamine administration significantly reversed the atomoxetineinduced antihyperalgesic responses in the Randall-Selitto tests (p < 0.01).  Figure 4B presents the effect of phentolamine pre-treatment on atomoxetine-induced antihyperalgesic response against thermal nociceptive stimuli applied in the Hargreaves test. The results of the two-way ANOVA analysis showed that the effect of atomoxetine treatment on the paw withdrawal latencies of diabetic rats was significant [F (1,28) = 55.26, p < 0.001] but the effect of phentolamine was not [F (1,28) = 1.87, p > 0.05]. In addition, a significant interaction was detected between the "atomoxetine treatment" and "phentolamine administration" factors [F (1,28) = 4.84, p < 0.05]. Results of the Bonferoni multiple comparisons test showed that phentolamine administration significantly reversed the atomoxetine-induced antihyperalgesic responses assessed in the Hargreaves test (p < 0.05).  Figure 4B presents the effect of phentolamine pre-treatment on atomoxetine-induced antihyperalgesic response against thermal nociceptive stimuli applied in the Hargreaves test. The results of the two-way ANOVA analysis showed that the effect of atomoxetine treatment on the paw withdrawal latencies of diabetic rats was significant [F (1,28) = 55.26, p < 0.001] but the effect of phentolamine was not [F (1,28) = 1.87, p > 0.05]. In addition, a significant interaction was detected between the "atomoxetine treatment" and "phentolamine administration" factors [F (1,28) = 4.84, p < 0.05]. Results of the Bonferoni multiple comparisons test showed that phentolamine administration significantly reversed the atomoxetine-induced antihyperalgesic responses assessed in the Hargreaves test (p < 0.05).
The effect of propranolol (a non-selective β-adrenergic receptor blocker) pre-treatment on atomoxetine-induced antihyperalgesic response against mechanical nociceptive stimuli applied in the Randall-Selitto test was shown in Figure 5A However, no significant interaction was detected between the "atomoxetine treatment" and the "propranolol administration" factors [F (1,28) = 2.06, p > 0.05]. Results of the Bonferoni multiple comparisons test showed that propranolol administration significantly reversed the atomoxetine-induced antihyperalgesic responses in the Randall-Selitto tests (p < 0.05). Figure 5B displays the effect of propranolol pre-treatment on atomoxetine-induced antihyperalgesic response against thermal nociceptive stimuli applied in the Hargreaves test. The results of two-way ANOVA analysis showed significant effects of both atomoxetine treatment [F (1,28) = 12.49, p < 0.01] and propranolol administration [F (1,28) = 11.11, p < 0.01] on the paw withdrawal latencies of diabetic rats. In addition, a significant interaction was detected between the "atomoxetine treatment" and "propranolol administration" factors [F (1,28) = 8.55, p < 0.01]. Results of the Bonferoni multiple comparisons test showed that propranolol administration significantly reversed the atomoxetine-induced antihyperalgesic responses assessed in the Hargreaves test (p < 0.001).
atomoxetine-induced antihyperalgesic response against mechanical nociceptive stimuli applied in the Randall-Selitto test was shown in Figure 5A. The results of two-way ANOVA analysis showed significant effects of both atomoxetine treatment [F (1,28) = 101.4, p < 0.001] and propranolol administrations [F (1,28) = 4.35, p < 0.05] on the paw withdrawal thresholds of diabetic rats. However, no significant interaction was detected between the "atomoxetine treatment" and the "propranolol administration" factors [F (1,28) = 2.06, p > 0.05]. Results of the Bonferoni multiple comparisons test showed that propranolol administration significantly reversed the atomoxetine-induced antihyperalgesic responses in the Randall-Selitto tests (p < 0.05). Figure 5B displays the effect of propranolol pre-treatment on atomoxetine-induced antihyperalgesic response against thermal nociceptive stimuli applied in the Hargreaves test. The results of two-way ANOVA analysis showed significant effects of both atomoxetine treatment [F (1,28) = 12.49, p < 0.01] and propranolol administration [F (1,28) = 11.11, p < 0.01] on the paw withdrawal latencies of diabetic rats. In addition, a significant interaction was detected between the "atomoxetine treatment" and "propranolol administration" factors [F (1,28) = 8.55, p < 0.01]. Results of the Bonferoni multiple comparisons test showed that propranolol administration significantly reversed the atomoxetine-induced antihyperalgesic responses assessed in the Hargreaves test (p < 0.001).    The effect of sulpiride (D2/D3 dopaminergic receptor blocker) pre-treatment on atomoxetineinduced antihyperalgesic response against mechanical nociceptive stimuli applied in the Randall-Selitto test was shown in Figure 7A. The results of two-way ANOVA analysis showed significant effect of atomoxetine treatment [F (1,28) = 9.36, p < 0.01] on the paw withdrawal thresholds of diabetic rats but sulpiride administrations were ineffective [F (1,28) = 3.4, p > 0.05]. Besides, no significant interaction was detected between the "atomoxetine treatment" and "sulpiride administration" factors [F (1,28) = 4.12, p > 0.05]. Results of the Bonferoni multiple comparisons test showed that sulpiride administration significantly reversed the atomoxetine-induced antihyperalgesic responses in the Randall-Selitto tests (p < 0.05).    The effect of sulpiride (D2/D3 dopaminergic receptor blocker) pre-treatment on atomoxetineinduced antihyperalgesic response against mechanical nociceptive stimuli applied in the Randall-Selitto test was shown in Figure 7A. The results of two-way ANOVA analysis showed significant effect of atomoxetine treatment [F (1,28) = 9.36, p < 0.01] on the paw withdrawal thresholds of diabetic rats but sulpiride administrations were ineffective [F (1,28) = 3.4, p > 0.05]. Besides, no significant interaction was detected between the "atomoxetine treatment" and "sulpiride administration" factors [F (1,28) = 4.12, p > 0.05]. Results of the Bonferoni multiple comparisons test showed that sulpiride administration significantly reversed the atomoxetine-induced antihyperalgesic responses in the Randall-Selitto tests (p < 0.05).
The effect of sulpiride pre-treatment on atomoxetine-induced antihyperalgesic response against thermal nociceptive stimuli applied in the Hargreaves test was shown in Figure 7B. The results of The effect of sulpiride (D 2 /D 3 dopaminergic receptor blocker) pre-treatment on atomoxetine-induced antihyperalgesic response against mechanical nociceptive stimuli applied in the Randall-Selitto test was shown in Figure 7A. The results of two-way ANOVA analysis showed significant effect of atomoxetine treatment [F (1,28) = 9.36, p < 0.01] on the paw withdrawal thresholds of diabetic rats but sulpiride administrations were ineffective [F (1,28) = 3.4, p > 0.05]. Besides, no significant interaction was detected between the "atomoxetine treatment" and "sulpiride administration" factors [F (1,28) = 4.12, p > 0.05]. Results of the Bonferoni multiple comparisons test showed that sulpiride administration significantly reversed the atomoxetine-induced antihyperalgesic responses in the Randall-Selitto tests (p < 0.05).
The effect of sulpiride pre-treatment on atomoxetine-induced antihyperalgesic response against thermal nociceptive stimuli applied in the Hargreaves test was shown in Figure 7B

Antihyperalgesic Efficacy of Atomoxetine is not Related to a Possible Alteration in the Motor Activity of Rats
The effects of atomoxetine on the horizontal and vertical locomotor activities of animals were shown in Figure 8A,B, respectively. diabetic rats. In addition, a significant interaction was detected between the "atomoxetine treatment" and "sulpiride administration" factors [F (1,28) = 20.0, p < 0.001]. Results of the Bonferoni multiple comparisons test showed that sulpiride administration significantly reversed the atomoxetineinduced antihyperalgesic responses assessed in the Hargreaves test (p < 0.001).

Antihyperalgesic Efficacy of Atomoxetine is not Related to a Possible Alteration in the Motor Activity of Rats
The effects of atomoxetine on the horizontal and vertical locomotor activities of animals were shown in Figure 8A,B, respectively. The results of the Bonferroni tests revealed that the number of horizontal and vertical locomotor activities measured at 4th week in all of the diabetic groups were significantly lower than those of measured before the induction of diabetes (Week 0) (p < 0.001). On the other hand, atomoxetine treatment did not induce a further change in the locomotor activities (Figure 8).
Diabetes was established with a single i.v. dose of streptozotocin (STZ). 4 weeks after the induction of diabetes, when neuropathy was developed in diabetic rats, hyperalgesia tests were started [12,13]. Two well-known experimental methods were used to assess antihyperalgesic activity of atomoxetine against mechanical and thermal painful stimuli: Randall-Selitto and Hargreaves tests, respectively. In both of these tests, in all of the diabetic experimental groups, "paw withdrawal threshold" (Figure 1) and "paw withdrawal latency" (Figure 2) values of rats measured at 4 th weeks The results of the Bonferroni tests revealed that the number of horizontal and vertical locomotor activities measured at 4th week in all of the diabetic groups were significantly lower than those of measured before the induction of diabetes (Week 0) (p < 0.001). On the other hand, atomoxetine treatment did not induce a further change in the locomotor activities ( Figure 8).
Diabetes was established with a single i.v. dose of streptozotocin (STZ). 4 weeks after the induction of diabetes, when neuropathy was developed in diabetic rats, hyperalgesia tests were started [12,13]. Two well-known experimental methods were used to assess antihyperalgesic activity of atomoxetine against mechanical and thermal painful stimuli: Randall-Selitto and Hargreaves tests, respectively. In both of these tests, in all of the diabetic experimental groups, "paw withdrawal threshold" (Figure 1) and "paw withdrawal latency" (Figure 2) values of rats measured at 4th weeks were significantly lower than those measured before the induction of diabetes (at week 0). These findings indicated that targeted neuropathy model was successfully established, in the present study.
Following the confirmation of the development of neuropathy at the 4th week, diabetic rats were started to receive atomoxetine at daily dose of 3 mg/kg. Administrations of atomoxetine for 7 (p < 0.001) and 14 (p < 0.001) day-long induced a significant increase in the reduced paw withdrawal thresholds of diabetic rats in the Randall-Selitto tests (Figure 1.). Similarly, in the Hargreaves tests, shortened paw withdrawal latencies of diabetic rats were prolonged with the 7 (p < 0.001) and 14 (p < 0.001) day-long treatments with atomoxetine (Figure 2.). These data indicated that atomoxetine has a notable antihyperalgesic effect in diabetic rats, which is comparable with the efficacy of reference drug pregabalin.
After the antihyperalgesic effect of atomoxetine has been demonstrated, some possible mechanisms underlying this pharmacological effect were investigated by further studies. For this purpose, first, possible contribution of catecholaminergic system to the observed antihyperalgesic effect was investigated by using AMPT, an agent inhibiting the synthesis of catecholamines [14,15]. Obtained data indicated that AMPT pre-treatments reversed the atomoxetine-induced antihyperalgesic responses both in the Randall-Selitto (p < 0.01) and the Hargreaves tests (p < 0.001) (Figure 3).
Alpha-methyl-p-tyrosine is an agent selectively inhibiting tyrosine hydroxylase, a rate-limiting enzyme in the noradrenaline and dopamine synthesis [14,15]. Two consecutive injections of this agent at a dose of 200 mg/kg with 23 h interval, has been shown to cause a 50%-60% reduction in the central noradrenaline levels [14,16]. Therefore, in this study, abolishment of atomoxetine induced antihyperalgesic activity following AMPT pre-treatments suggested that catecholaminergic system could mediate the antihyperalgesic effect of atomoxetine, at least partially. Based on these findings and noradrenaline reuptake inhibitory mechanism of atomoxetine, it can be speculated that this drug might show its pharmacological effect on neuropathic pain by enhancing neurotransmission in the brain stem-spinal descending noradrenaline system, which suppresses nociceptive signaling from primary afferent neurons to the spinal dorsal horn [17]. Moreover, enhanced noradrenaline level in the dorsal horn of the spinal cord might induce antihyperalgesic activity by inhibiting the release of excitatory neurotransmitters from primary afferent fibers or by hyperpolarizing post-synaptic spinal cord dorsal horn cells, via activating spinal α-adrenergic receptors [11]. However, all these possible mechanisms should be clarified by further detailed studies. Actually, contribution of catecholaminergic system to the presented antihyperalgesic activity seem to get additional importance in terms of diabetic neuropathic pain treatment, since diabetes causes serious deficits in the functions of several neurotransmitter systems playing role in pain-suppressing pathways in the central nervous system, including noradrenergic system [18].
As a general knowledge, two main receptor subtypes are known to mediate the physiological effects of noradrenaline: αand β-adrenoceptors. Both of the α-adrenergic [19][20][21] and β-adrenergic receptors [22,23] are known to play active roles in nociception and pain-related processes. Therefore, after demonstrating the participation of catecholaminergic system to the antihyperalgesic activity of atomoxetine, we examined the possible involvement of αand β-adrenergic receptors in the mentioned activity of this drug. We used phentolamine, a non-selective α-adrenergic receptor blocker, and propranolol, a non-selective β-adrenergic receptor blocker, with this aim.
In the Randall-Selitto tests, pre-treatment with phentolamine significantly reversed atomoxetine-induced increase in the paw withdrawal thresholds of diabetic rats (p < 0.01). Moreover, prolonged paw withdrawal latencies of atomoxetine-administrated diabetic rats were significantly shortened in the Hargreaves tests (p < 0.05), following the phentolamine administrations (Figure 4.). Similar to results of the phentolamine studies, pre-treatment with propranolol also reversed the atomoxetine-induced increase in the paw withdrawal thresholds (p < 0.05) and reduced the prolonged paw withdrawal latencies of diabetic rats (p < 0.001) (Figure 5.), indicating the contribution of both of the αand β-adrenoceptors to the anthyperalgesic effect of atomoxetine. These findings are in agreement with the results of the previous studies reporting the critical role of α-adrenergic [19,[24][25][26][27][28] and β-adrenergic [12,13,[29][30][31][32][33] receptors in the therapeutic effects of various drugs on neuropathic pain.
Recent studies indicated that dopaminergic system is also related to the nociceptive transmission in the central nervous system [34]. It has been shown that dopaminergic neurons project from the A11 region in the dorsal posterior hypothalamus innervate the dorsal horn of the medulla spinalis, a critical region related to transmission of painful stimuli [35,36]. Moreover, dopaminergic neurons in the ventral tegmental area (A10) have been demonstrated to release dopamine in the nucleus accumbens and enhance analgesic activity through D 2 -like receptors [37,38]. Based on these facts, we also investigated probable participation of dopaminergic receptors to the presented antihyperalgesic effect of atomoxetine. Possible involvement of D 1 and D 2 /D 3 dopaminergic receptors were examined by SCH 23390 and sulpiride pre-treatments, respectively. Obtained results showed that sulpiride administrations significantly reversed the atomoxetine-induced antihyperalgesic responses both in the Randall-Selitto (p < 0.05) ( Figure 7A) and Hargreaves (p < 0.001) tests ( Figure 7B). However, SCH 23390 administrations did not induce any alteration on the same effect (p > 0.05) ( Figure 6A,B).
These findings, indicating the involvement of D 2 /D 3 , but not D 1 dopaminergic receptors in the antihyperalgesic effect of atomoxetine, are in accordance with the results of the previous studies suggesting that D 2 -like receptors mediate the dopamine-induced blockage of nociceptive transmission at the dorsal horn [35,36,39]. Although atomoxetine has no direct effect on the dopamine transporter, dopaminergic systems might be indirectly involved in the antihyperalgesic effect this drug, similar to some other drugs inhibiting reuptake of monoamines such as tricyclic antidepressants or serotonin and noradrenaline reuptake inhibitors [11,39,40].
It is well described that drug-induced alterations in motor activities of the rodents may affect the results of the experiments performed to evaluate pain. Therefore, monitoring the motor activity of animals throughout the experiments has a critical importance. For this reason, we evaluated the spontaneous locomotor activities of rats that we used in the tests. Data obtained from the activity cage tests designated that diabetes caused significant decrease both in the horizontal ( Figure 8A) and vertical ( Figure 8B) locomotor activities. These findings supported the previous studies reporting the impaired motor activity and motor coordination in diabetic animals [41][42][43]. On the other hand, subacute administration of atomoxetine at doses of 3 mg/kg did not cause any further decrease or increase in the locomotor activity suggesting that antihyperalgesic efficacy of this drug is independent of any probable alteration in the motor activity of the animals.
In this work, some important findings have been revealed about the pharmacological mechanisms underlying the antihyperalgesic effect of atomoxetine, however, it is also possible to conduct further studies to elucidate the exact mechanism of action. For example, specific αand β-adrenergic receptor subtypes contributing to the observed effect can be clarified by using specific αand β-adrenergic receptor subtype blocker agents such as prazosin (α 1 -adrenergic receptor antagonists) [27]; idazoxan, yohimbine (α 2 -adrenergic receptor antagonists) [24,27] or ICI 118,551 (β 2 adrenoceptor antagonist) [44]. Moreover, it will also be useful to investigate possible participation of other neurotransmitter systems associated with nociception and analgesia, such as opioidergic, glutaminergic, nitrergic, GABAergic, or muscarinic systems to the atomoxetine-induced antihyperalgesic effect. Furthermore, investigation of atomoxetine-induced alterations in central and peripheral nervous system at molecular levels, by using immunohistochemical and morphometric methods, may also help to produce new and valuable data related to mode of the presented antihyperalgesic action.
In summary, this is the first study presenting the significant antihyperalgesic activity capacity of atomoxetine against diabetes-induced neuropathic pain and suggesting the involvement of αand β-adrenergic receptors and D 2 /D 3 dopaminergic receptors as underlying pharmacological mechanisms. These results have pharmacological importance since they pointed out a possibility of a new indication for an old drug: atomoxetine. Nevertheless, atomoxetine may be regarded as an alternative drug in the symptomatic treatment of neuropathic pain only if the experimental findings of this preclinical study could be confirmed by well-designed clinical trials.

Animals
Experiments were performed using male Sprague-Dawley rats weighing 250-300 g at the same age. Rats were housed in well-ventilated rooms at 24 ± 1 • C in a 12-h light/12-h dark cycle where the lights were kept open between 8:00-20:00. The experimental procedure of this research was approved by the Anadolu University Animal Experiments Local Ethics Committee (Date: 15 December 2017 and Project code: 2017-04).

Induction of Experimental Diabetes Model in Rats
Animal model of diabetes was established by a single intravenous (i.v.) injection of STZ (50 mg/kg, adjusted to pH 4.5 in 0.1 M citrate buffer) into the tail vein of the rats, as described earlier [12]. An equal volume of citrate buffer was administrated (i.v.) to all healthy rats placed in control groups. 72 h later the injection of STZ, measurements of blood glucose were performed by using a Glukotrend ® device (Roche, Basel, Switzerland). Rats, whose blood glucose levels were higher than 300 mg/dL, were classified as diabetic.

Design of the Experimental Groups
Experimental groups (n = 8) for neuropathic pain experiments were designed as follows (Table 1).  [47] Atomoxetine and pregabalin were administrated at every morning 9:00 a.m. for 2 weeks. In the test days, experiments were performed 30 min after the last i.p. dose of atomoxetine and 60 min after the last oral dose of pregabalin.

Neuropathic Pain Tests
After waiting 4 weeks for the appearance of peripheral neuropathy in diabetic animals [48], hyperalgesia responses were assessed by Randall-Selitto and Hargreaves tests.

Randall-Selitto Test
The mechanical nociceptive stimulus-induced hyperalgesia was assessed by Randall-Selitto device (Ugo-basile, 37215, Verase, Italy) which applies increasing pressure on the dorsal parts of the hindpaws. The threshold of mechanical hyperalgesia was accepted as the force (grams) at which the rat withdrew its paw from the device. The cut off value was determined as 250 g in order to protect the paw from a possible tissue damage [12,13,49].

Hargreaves Test (Plantar Test)
The thermal nociceptive stimulus-induced hyperalgesia was assessed by Hargreaves test device (Ugo-basile, 37370, Verase, Italy), which focuses radiant heat to hind-paws of the rats. The parameter for thermal hyperalgesia was accepted as the paw withdrawal latency (s) of rats, which means the time until the rat withdraws its paw from the hot stimulus. The response time was calculated by taking the average of the three measurements performed at 5 min intervals. The cut off time was determined as 20 s in order to protect the paw from a possible tissue damage [12,50].

Mechanistic studies
Potential participation of catecholaminergic neurotransmission to the antihyperalgesic activity of atomoxetine was examined by AMPT (a catecholamine synthesis inhibitor) pre-treatment studies. For this purpose, AMPT was administrated at dose of 200 mg/kg i.p., twice at 24 h and 1 h before the last dose of atomoxetine [12,13,51].
Similar studies in the literature, as well as the previous studies in our laboratory, have been used to determine the doses of antagonists used in this study.

Activity Cage Tests
An activity cage device (Ugo-basile, 7420, Verase, Italy), which is made of transparent plexiglass material with dimensions of 40 × 40 × 31 cm, was used to evaluate spontaneous locomotor activities of the rats in the experimental groups. Locomotor activities of the animals in the vertical and horizontal directions were recorded for 10 min [54,55].

Statistical Analysis
For statistical calculations, Graphpad Prism. 6.01 package program was used. Data acquired from the Randall-Selitto, Hargreaves and activity cage tests, which were conducted in the same groups for consequent weeks, were analyzed by two-way repeated ANOVA, followed by Bonferroni tests as post hoc. Besides data obtained from the mechanistic studies were analyzed by two-way ANOVA, followed by the same multiple comparison tests.
Results were expressed as mean ± standard error of mean (S.E.M.). p value of 0.05 was considered as significant.