Idiopathic Parkinson's disease is just one member of a broader family of motor disorders known as parkinsonism (Alvarez et al., 2007), a group that also includes encephalitic, pugilistic, and drug-induced parkinsonism. Regardless of subtype, all forms of parkinsonism share in common four core symptoms: bradykinesia, postural instability, rigidity, and tremor (Hoehn and Yahr, 1967). Tremor, defined as “a rhythmic and involuntary oscillation of a body part, caused by reciprocal innervations of a muscle, which leads to repetitive, stereotyped contractions with regular frequency and amplitude,” is sometimes considered to be the cardinal symptom of parkinsonism (Deuschl et al., 2000; Mansur et al., 2007). This tremor, present predominantly at rest, appears early on in the disease course and causes a significant amount of life distress (Mansur et al., 2007). It presents in up to 75% of individuals with PD, and has been shown to occur in a frequency range of 3–7 Hz, a range distinct from that seen in dyskinesias (1–2 Hz), essential tremor (8 Hz), and postural tremors (8–12 Hz; Findley et al., 1981; Findley and Capildeo, 1984; Marsden, 1984; Elbe and Koller, 1990; Hunker and Abbs, 1990; Deuschl et al., 1996, 2000; Spieker et al., 1997).

Usually beginning unilaterally in the distal segments of the extremities, parkinsonian tremor frequently spreads bilaterally, often affecting both upper and lower limbs, as well as facial muscles, as the disease progresses (Adams and Victor, 1993; Koster et al., 1997; Salamone et al., 1998; Deuschl et al., 2000; Morrison et al., 2008). While the most common form of parkinsonian resting tremor in humans is a “pill rolling” tremor of the hands, tremulous “up-and-down” movements of the jaw in parkinsonian patients have also been well documented (Barbeau, 1986; Selby, 1986; Young, 1986; Hunker and Abbs, 1990; Adams and Victor, 1993; Ben-Pazi et al., 2001; Schneider et al., 2006; Sheffield and Jankovic, 2007; Leventoglu and Baysal, 2008; Jankovic, 2009). Oral tremor also is induced by long-term administration of DA antagonists, a condition known as “rabbit syndrome” and defined as “rhythmic jaw tremor” (Weiss et al., 1980). This type of tremor is alleviated by common antiparkinsonian drugs, but exacerbated by cholinomimetic administration (Sovner and Dimascio, 1977; Weiss et al., 1980; Yassa and Lal, 1986). Thus, it is considered to be a form of drug-induced parkinsonian tremor (Sovner and Dimascio, 1977; Weiss et al., 1980; Tarsy, 1983).

The neural network thought to underlie parkinsonian tremor is complex, consisting of changes to the open and closed loop connections between neocortex, basal ganglia, and thalamus. According to current models of basal ganglia function, reductions of the DA input from substantia nigra pars compacta (particularly from the retrorubral area A8 portion of the SNc in tremor-dominant PD) to the striatum leads to an increase in output from the subthalamic nucleus (STN), a nucleus consisting of neurons and circuits that are particularly correlated with tremor (Reck et al., 2009). This, in turn, produces an overstimulation of GABAergic neurons in substantia nigra pars reticulata (SNr) and globus pallidus internal (GPi), and a subsequent inhibition of the ventral lateral and ventral anterior nuclei of the thalamus, leading ultimately to the reduced output of the sensorimotor cortex and the motor impairments observed in parkinsonism. Most recently, hypotheses of parkinsonism have focused not just on these changes in firing rates, but also on alterations in firing patterns (see Obeso et al., 2000, 2008; Bevan et al., 2002; Brown, 2003 for review). Specifically, increases in the activity of subthalamic and internal globus pallidus neurons are thought to cause these nuclei to enter a regime of rhythmic firing, resulting in an excessive synchrony (particularly in the Beta band; Bergman et al., 1998; Obeso et al., 2000, 2008; Brown et al., 2001; Levy et al., 2002). This excessive synchrony may be particularly important for the generation and maintenance of parkinsonian tremor.

The neurochemical cascade thought to underlie parkinsonian tremor is similarly multifaceted, with multiple neurotransmitters, including GABA, serotonin, adenosine, and acetylcholine (ACh), interacting with DA in the regulation of basal ganglia motor functions related to parkinsonism (Hornykiewicz, 1972, 1973; DeLong, 1990; Young and Penney, 1993; McSwain and Forman, 1995; Finn et al., 1997b; Hauber et al., 1998; Salamone et al., 1998; Mayorga et al., 1999a; Wichmann et al., 2001; Trevitt et al., 2002; Carlson et al., 2003a,b; Ishiwari et al., 2004; Obeso et al., 2008; Vanover et al., 2008). Interestingly, these other neurotransmitters offer additional avenues for therapeutic intervention. In addition to DAergic treatments for parkinsonism, muscarinic antagonists often are used as antiparkinsonian agents, and they have been shown in clinical studies to suppress tremor (McEvoy, 1983; Schrag et al., 1999; Milanov, 2001; Koller, 2002). More recently, antagonists of adenosine A_2A receptors have been shown to produce antitremor effects in animal models (Correa et al., 2004; Salamone et al., 2008a,b) and in human patients with Parkinson's disease (Bara-Jimenez et al., 2003; Pinna, 2009).

The study of parkinsonian tremor has proven to be something of an enigma in the field. Although tremor is considered to be the most specific marker of parkinsonism, with estimates of 95% probability that the presence of classical resting tremor indicates idiopathic PD, tremor severity has not been shown to parallel disease progression (Deuschl et al., 2000). Additionally, the system dynamics of parkinsonian tremor have shown that, despite the nearly uniform frequencies associated with tremor-related oscillations (i.e., 3–7 Hz; Hunker and Abbs, 1990), these oscillations occur independently of one another, with electromyographic (EMG) activity of tremor in one limb being uncorrelated to the EMG activity of another limb exhibiting tremor (O'Suilleabhain and Matsumoto, 1998; Hurtado et al., 2000, 2004; Raethjen et al., 2000, 2009; Ben-Pazi et al., 2001). Even within the tremor circuitry itself, tremors have not been shown to be tightly correlated with tremor-related neural activity in the nuclei of the basal ganglia and thalamus: tremor-related neural activity may wax and wane while limb tremor persists; conversely, tremor-related oscillations may exist in the neurocircuitry while no limb tremor is apparent (Hurtado et al., 2004). These seemingly paradoxical findings have led to a variety of questions about the pathophysiology underlying tremorogenesis in parkinsonism. Yet despite these lingering questions, relatively few clinical studies have specifically emphasized the pharmacology of tremor (e.g., Schrag et al., 2002; Navan et al., 2003, 2005; Sung et al., 2008; Binder et al., 2009) and there is considerable uncertainty about the neurochemical mechanisms that underlie the pathophysiology of parkinsonian tremor (Wilms et al., 1999; Deuschl et al., 2000, 2001; Bergman and Deuschl, 2002; Sung et al., 2008). While parkinsonian tremor seems to result from oscillating activity in the central nervous system (and not due to reflexes in the periphery; see Hunker and Abbs, 1990 for review), it appears to be generated by multiple oscillators, and it is unclear where these oscillators are located (though the basal ganglia loops are a likely candidate, oscillators in the thalamus have also been proposed) or by what mechanism these oscillators produce tremor (O'Suilleabhain and Matsumoto, 1998; Deuschl et al., 2000). Given the uncertainty surrounding this cardinal symptom of parkinsonism, studies employing animal models of parkinsonian resting tremor are imperative (Muthuraman et al., 2008). Basic research using animal models of drug-induced tremor enables researchers to investigate the neurochemical mechanisms that regulate tremor and to study the development of oscillatory patterns of activity in the brain that are thought to underlie tremorogenesis. Eventually, this research could lead to the development of novel treatments.

While a variety of animal models of parkinsonian tremor exist, most of these are models of postural or action tremors, and not of resting tremor (see Wilms et al., 1999). Although a few models of resting tremor have been produced, these have mainly been in primates and have involved the destruction of multiple nuclei or pathways, a paradigm that is expensive and impractical for most preclinical pharmaceutical testing (Wilms et al., 1999). The tremulous jaw movement (TJM) model may provide a resolution for many of these difficulties. TJMs, defined as rapid vertical deflections of the lower jaw that are not directed at any stimulus, are a rodent model of parkinsonian resting tremor (for reviews, see Salamone et al., 1998, 2001, 2005; see also Rodriguez Diaz et al., 2001; Cenci et al., 2002). TJMs generally occur in phasic bursts of repetitive jaw movement activity, with multiple movements within each burst. Willner (1990) cautions that animal models should undergo an evaluation for their validity using criteria similar to those used in the rigorous validation of psychological tests. Thus, a great deal of effort has been expended in the analysis of the pharmacological, temporal, and anatomical characteristics of the TJM model in order to evaluate whether these tremulous movements in rats are sufficiently similar to human parkinsonian tremor to allow for their validation as an animal model of tremor. TJMs have been shown to possess many of the neurochemical, anatomical, and pharmacological characteristics of parkinsonism, and thus meet a reasonable set of validation criteria for use as an animal model of parkinsonian tremor (Salamone et al., 1998; Cenci et al., 2002). Animal models in psychopharmacology are evaluated for their validity based upon a number of criteria (Willner, 1990). These including face validity (the phenomenological similarity between the model and the disorder, including etiology and symptomatology), construct validity (the degree of similarity between the identified variable (e.g., tremor), and the behavior being studied in the model), and predictive validity (primarily, the positive response to therapeutic drugs). As described below, the TJM model of parkinsonism is one of the few models of experimental tremor that meets these validation criteria.

Since oral movements in rats have been observed to result from chronic administration of DA antagonists, it could be suggested that TJMs represent an animal model of tardive dyskinesia (Ellison et al., 1987; Ellison and See, 1989; Creed et al., 2010). However, several lines of evidence conflict with this view. First, the fact that a tremor is orofacial does not make it a model of tardive dyskinesia, per se. As reviewed above, parkinsonian tremors can include up-and-down movements of the jaw (Barbeau, 1986; Hunker and Abbs, 1990), and “rabbit syndrome” is a drug-induced parkinsonian tremor that is characterized by chewing-like movements (Sovner and Dimascio, 1977; Weiss et al., 1980; Tarsy, 1983). Differences between rodents and humans in the relative prevalence of hand versus jaw tremor may simply be due to species differences in motor system anatomy and physiology (i.e., the relative proportion of the motor system dedicated to orofacial versus limb control in rodents and humans; Salamone et al., 1998).

By definition, tardive dyskinesia is produced by chronic administration of DA antagonists. In contrast, TJMs in rats can be produced by acute or sub-chronic administration of DA antagonists (Glassman and Glassman, 1980; Rupniak et al., 1985, 1986; Jicha and Salamone, 1991; Steinpreis and Salamone, 1993; Steinpreis et al., 1993; Egan et al., 1996; Ishiwari et al., 2005; Salamone et al., 2008a; Collins et al., 2010a; Galtieri et al., 2010) or by acute DA depletion with reserpine (Baskin and Salamone, 1993; Steinpreis and Salamone, 1993; Salamone et al., 2008b). Furthermore, while muscarinic ACh antagonists reduce both parkinsonian symptoms in humans and TJMs in rats (Betz et al., 2007; Collins et al., 2011), they worsen tardive dyskinesia (Fahn et al., 1974; Burnett et al., 1980; Noring et al., 1984). Similarly, while L-DOPA reduces TJMs (Cousins et al., 1997), it is well known to induce dyskinesias. Finally, as described earlier, the jaw movements induced by DA depletion or cholinomimetics have frequency characteristics that are quite different from tardive dyskinesia. While TJMs show maximal activity in the 3 to 7-Hz frequency range (See and Chapman, 1991; Salamone and Baskin, 1996; Finn et al., 1997b; Mayorga et al., 1997; Cousins et al., 1998; Salamone et al., 1998; Ishiwari et al., 2005; Collins et al., 2010a; Galtieri et al., 2010), tardive dyskinesia is typically in the range of 1–2 Hz (Alpert et al., 1976; Wirshing et al., 1989a,b).

As described above, basal ganglia regulation of motor function involves interactions between a wide array of transmitters and neuromodulators, including DA, ACh, adenosine, 5-HT, GLU, and GABA, as well as circuits that interconnect striatal and non-striatal regions. The complexity of these interactions provides a daunting challenge to the investigator, but it also provides a framework for understanding how a symptom such as tremor can be induced or ameliorated by a variety of different conditions. Moreover, it provides avenues for the development of novel treatments for tremor.

In view of the continued uncertainty about the pathophysiology of parkinsonian tremor (Deuschl et al., 2000; Sung et al., 2008), it is important to identify the physiological conditions that are correlated with the generation of tremulous movements in animal models. As described above, pharmacological evidence links the VLS subregion of the neostriatum to the production of TJMs. In a recent preliminary study, increased tremor-related oscillatory activity in local field potential (LFP) signals was observed within the VLS subregion of the neostriatum during the occurrence of pilocarpine-induced TJMs (Collins et al., 2010c). During tremor activity, there was a sharp increase in LFP power in the VLS in the 5 to 8-Hz band (Figure 4). This increase in power was noticeably absent in VLS when tremor activity was not occurring. This is consistent with findings from the spectral analyses of EMG traces recorded from the temporalis muscle during tremor and non-tremor epochs (see above). Furthermore, during periods of tremor activity, there was a broad increase in beta band (15–30 Hz) activity (Figure 4). This increase in beta band power was not present during periods without TJM activity. These findings are consistent with studies obtained from parkinsonian patients. Both single-unit and LFP recordings in patients with Parkinson's disease have shown evidence of tremor-related frequencies (∼4–8 Hz) and increased “beta” (∼8–30 Hz) frequency activity in basal ganglia structures, typically globus pallidus, and STN (Hutchison et al., 1997; Weinberger et al., 2006; see also Hutchison et al., 2004 or Hammond et al., 2007 for reviews). Additionally, increases at both ∼6 Hz (tremor frequency) and ∼20 Hz (beta frequency) can be observed in LFP signals in globus pallidus and STN of Parkinson's disease patients off L-DOPA, while after L-DOPA treatment is reinstated, both the tremor and beta frequency LFPs diminish (Brown et al., 2001). The relationship between a specific neurophysiological phenomenon (e.g., altered output in specific structures and changes in the frequency of output) and particular Parkinsonian symptoms is unclear, although some evidence suggests that the low frequency increase at 4–8 Hz is related to tremor, while the increase in “beta” (15–30 Hz) interferes with the initiation and maintenance of movements (Levy et al., 2002; see however Weinberger et al., 2006). It has been proposed that the focal (spatial-limited) occurrence of gamma (∼40–80 Hz) is associated with movement initiation (Courtenmance et al., 2003; Masimore et al., 2005) and that changes in striatal neurochemistry that lead to parkinsonism transform basal ganglia circuits from participating in focal (spatially limited) gamma ensembles toward more global beta oscillations (e.g., Pessiglione et al., 2005). In light of this, the findings of increases in VLS LFP power at 4–8 Hz and in the beta band during TJMs may provide important information about the physiological mechanisms underlying tremorogenesis.

Tremulous jaw movements have also been examined in relation to hippocampal theta and epileptiform activity. Doses of the cholinergic agonist pilocarpine can induce increases in theta (6–10 Hz) during awake-immobility in the rat, but there was no obvious temporal relationship between theta and the presence/absence of TJMs (Collins et al., 2010c; Figure 5). In addition, although very high doses of pilocarpine such as 50.0–100 mg/kg IP can produce epileptiform activity, no epileptiform activity is seen under pharmacological conditions that produce TJMs (data not shown). In conjunction with the earlier findings presented, these recent data provide further evidence that TJMs are being induced via actions within the VLS itself and not being driven by other factors, such as hippocampal theta or epileptiform activity.

Tremulous jaw movements have become a useful model for experimental assessment of diverse tremorogenic and tremorolytic conditions. As reviewed above, adenosine A_2A antagonists can suppress TJMs; consistent with this observation, we recently found that systemic administration of non-sedative doses of the A_2A agonist CGS 21680 can induce TJMs (Figure 6). Moreover, TJMs have been shown to be attenuated by several serotonergic drugs, including the serotonin 5-HT₂ family antagonist mianserin (Carlson et al., 2003b), the 5-HT_2A receptor inverse agonist, ACP-103 (Vanover et al., 2008), and the 5-HT(1A) agonists 8-OH-DPAT, buspirone, and F-97013-GD (Zazpe et al., 2006). The TJM model also has been incorporated into methods used to discriminate between the effects of typical antipsychotics and later-generation compounds such as clozapine, olanzapine, and quetiapine (Chesler and Salamone, 1996; Trevitt et al., 1998, 1999; Betz et al., 2005). Tacrine-induced TJMs also were shown to be suppressed by the T-type calcium channel blocker zonisamide (Miwa et al., 2008, 2009, 2011), which has anticonvulsant and tremorolytic effects in humans. Furthermore, stimulation of GABA transmission in SNr, either by local injection the GABA_A agonist muscimol (Finn et al., 1997a) or by nigral transplantation of engineered GABA-producing cells (Carlson et al., 2003a) reduced cholinomimetic-induced TJMs.

In recent years, high frequency stimulation of the STN has become a standard treatment of tremor used by neurosurgeons (Ashkan et al., 2004). In MPTP-treated primates, high frequency stimulation of the STN has been shown to improve rigidity and motor scores (Benazzouz et al., 1993, 1996; Ashkan et al., 2004). In 1993, the first bilateral STN electrodes were implanted in human patients, and high frequency STN stimulation improved rigidity, tremor, akinesia, postural/gait instabilities, independence, and quality of life, as well as reducing drug-induced dyskinesias (Benabid et al., 1994; Limousin et al., 1995, 1998; Ashkan et al., 2004). Despite its efficacy as a therapeutic measure, however, a clear consensus on the mechanism of action of DBS has yet to be reached, and several competing hypotheses have been put forward. Animal models of tremor could be useful for investigating potential mechanisms of action of DBS. In a recent study by Collins-Praino et al. (2011a), TJMs were induced by a DA D1 antagonist (SCH 39166), a DA D2 antagonist (pimozide), a muscarinic agonist (pilocarpine), and an anticholinesterase (galantamine). Unilateral high frequency stimulation (130 Hz) of the STN, but not of a striatal control site, was shown to be effective at reversing the TJMs induced by all four pharmacological agents (Figure 7). Stimulation at lower frequencies or intensities failed to decrease TJMs compared to baseline “OFF” epochs, indicating that this response is not only dependent upon the brain area stimulated but also upon the frequency and intensity parameters used. When the adenosine A_2A antagonist MSX-3 was co-administered with high frequency stimulation of the STN, both the frequency and intensity parameters necessary to produce a suppression of galantamine-induced TJMs were drastically reduced (Figure 7). These results serve to provide a further validation of the TJM model, and also may have critical significance for the selection of parameters used during deep brain stimulation in parkinsonian patients undergoing this procedure for the treatment of medically refractory tremor. Based on the results of this study, it is possible that the prescription of an adenosine A_2A antagonist to these patients may result in a significant reduction in the frequency and/or intensity parameters needed for therapeutic efficacy and, thereby, a near doubling of the therapeutic window for these parkinsonian patients.

In conclusion, the TJM model shares many of the pharmacological, temporal, and anatomical characteristics of human parkinsonian tremor. Thus, it can be viewed as meeting a reasonable set of validation criteria as a rodent model of parkinsonian resting tremor. The validation of this model represents a significant advance forward in the scientific study of parkinsonian resting tremor, which still has a great deal of uncertainty surrounding its pathophysiology. The TJM model allows for basic research into the neurochemical mechanisms that regulate tremor, and the development of oscillatory patterns of activity in the brain that are thought to underlie tremorogenesis. Eventually, this research may lead to the development of novel treatments for tremor.

The authors declare that, except for income received from my primary employer, no financial support or compensation has been received from any individual or corporate entity over the past three years for research or professional service and there are no personal financial holdings that could be perceived as constituting a potential conflict of interest.