Introduction

The use of deep brain stimulation (DBS) as a treatment for neuropsychiatric disorders and symptoms is among the most important recent advances in clinical neuromodulation. DBS is a neurosurgical procedure that involves the implantation of electrodes into specific brain targets to modulate local and widespread connectivity in dysfunctional networks [1]. To date, several thousands of patients have undergone DBS for various neuropsychiatric conditions [2,3,4,5,6]. Notably, patients with Parkinson’s disease (PD) are among the most common candidates for this treatment option [7]. In select cases, DBS can also induce long-term alleviation of symptoms in patients with dystonia [8], Tourette’s syndrome [9], or epilepsy [10]. In terms of psychiatric conditions, clinical DBS studies have shown promise in relieving symptoms of obsessive-compulsive disorder (OCD) [11, 12], major depressive disorder [13], and substance use disorder (SUD) [14, 15]. Despite encouraging reports, the optimal DBS brain targets and underlying mechanisms that lead to benefits and/or side effects in distinct pathologies remain unclear.

Experimental DBS in animal models plays an essential role in our understanding of the multiscale neurobiological mechanisms of DBS, as well as the development of new technologies. Rodents — in particular, the Norway rat (Rattus norvegicus) and the house mouse (Mus musculus) — are especially useful for this purpose. Different models can capture core features observed in neuropsychiatric disorders and can be objectively tested in a series of standardized behavioural tests. In addition, the underlying mechanisms of DBS may be explored using healthy rodent strains [16, 17]. In this article, we present a systematic review of preclinical DBS studies to synthesize the literature on the current landscape of DBS in rodent models.

Methods

A systematic analysis of the international literature was performed in accordance with PRISMA [18] guidelines (See Supplementary Fig. 1 for PRISMA flow diagram). The PubMed MEDLINE (National Library of Medicine) database was used to search articles published between January 2000 and December 2022, using the search terms “deep brain stimulation”, “DBS”, “rodent”, “mouse”, and “rat”. Duplicates and unrelated reports were excluded. The remaining reports were first screened by title and abstract, then selected articles progressed to full-text review based on prespecified inclusion and exclusion criteria (Inclusion: Studies published in English, Original articles with full-text availability, Articles describing studies of DBS in rats or mice, Articles reporting stimulation parameters used; Exclusion: Non-original studies (i.e., conference abstracts, reviews, meta-analyses, commentaries, editorials, and protocols), Articles reporting on clinical population, Articles using other neuromodulatory techniques, Investigations performed in-vitro or using other species) See Supplementary References for full list of included studies. Data extraction was performed by three authors (KZ, RM, FVG) focusing on 5 categories: Bibliographic data, Animal model characteristics (group based on DSM-5 classification [19]; Supplementary Table 1), DBS target and settings, Behavioural outcomes, Mechanisms of action.

Results

Of the 1674 reports screened, 407 original research articles were included in this review (rat: 356, mouse: 51; Supplementary Fig. 1, Supplementary References). Considering the differences in disease models between species (which may involve distinct cellular and molecular targets), the data extracted from mouse and rat studies are presented separately (see Supplementary Table 2 for the summary of combined data).

Rodent models of DBS

Figure 1 shows the proportion of articles published by the disease model and the distribution of these articles over the years. Among rat studies, DBS effects have been investigated in 15 distinct models of disorders and in animals presenting no pathology (i.e., standard healthy strains; Fig. 1A). DBS for movement disorders was the most researched field and included Parkinson’s disease (PD; 24.2%), motor impairments (1.1%), dyskinesia (0.8%), and tremor (0.6%). Depression and epilepsy were the second and third most studied models, followed by SUD (Cocaine 2.8%, Morphine 2.8%, Alcohol 0.8%, Methamphetamine 0.6%, Heroin 0.3%). Models of dementia/cognition focused on cognitive performance (3.8%), Alzheimer’s disease (AD; 1.1%), and dementia (1.1%). Models of physical injury (traumatic brain injury, TBI: 1.4%, pain: 1.1%, spinal cord injury: 1.1%, stroke: 0.6%, ischemia: 0.3%) and eating disorders (obesity: 1.4%, food intake: 0.8%, hedonic feeding: 0.8%, glucose metabolism: 0.3%), occupied smaller portions of the rat DBS landscape. Trauma or stressor-related disorders (i.e., post-traumatic stress disorder, PTSD), obsessive-compulsive disorder, psychosis, anxiety disorders, tinnitus, neurodevelopmental disorders, bladder function (related to neurological disorders), and sleep-wake disorders represented the least studied rat models for DBS.

Fig. 1: Rodent studies by disease model.
figure 1

Total percentage of rodent studies in each disease model, published after the year 2000, for rats (A) and mice (B). Number of published articles by period using rat models (C) and mouse models (D).

Full size image

For mouse studies, DBS effects have been investigated in 12 distinct models of disease and in healthy strains (Fig. 1B). Dementia/cognition (AD: 9.8%, cognitive performance: 3.9%, dementia: 2.0%) and movement disorders (PD: 9.8%, ataxia: 3.9%, tremor: 2.0%) were the most studied, followed by depression models. Neurodevelopmental disorders (Rett syndrome: 5.9%, Autism Spectrum Disorder, ASD: 2.0%), SUD (alcohol: 2.0%, cocaine: 3.9%), epilepsy, anxiety disorders, eating disorder (hedonic feeding: 2.0%, obesity: 2.0%), and physical injury (TBI: 2.0%, stroke: 2.0%) were investigated less frequently. OCD, psychosis, and sleep-wake disorder represented the least studied mouse models for DBS. There has been an upward trend in preclinical DBS publications since the early 2000s (Fig. 1C & D). To our knowledge, no rat DBS study was published between 2000–2002, and no mouse DBS study was published between 2000–2007. In the early stages, these studies focused on PD, epilepsy, or healthy animals; however, the disease models evaluated for DBS effects have substantially diversified over time.

Model characteristics

Although Sprague-Dawley (SD; 52.9%) and Wistar rats (30.1%) were the main healthy strains used (Fig. 2A), a handful of studies used Long Evans or Lewis rats to model psychosis [20], OCD [21,22,23], cocaine [24, 25] or alcohol addiction [26], catalepsy [27,28,29], PD [30,31,32,33,34,35,36,37,38,39,40,41,42,43,44], TBI [45, 46], and spinal cord injury [27, 28]. The unilateral 6-hydroxydopamine (6-OHDA) nigrostriatal lesion PD model was the most commonly used ( > 87% of included studies). While neurodevelopmental disorders were modelled in SD rats by exposure to valproic acid [47, 48], apomorphine [49] or antinuclear antibody [50], the Flinders Sensitive Line and Zucker rats were used exclusively to model depression and obesity, respectively. For modelling epilepsy, most studies used SD and Wistar rats receiving chemoconvulsants (e.g., kainic acid, pilocarpine, pentylenetetrazol, or FeCl3 solution in one study [51]) or chronic electrical stimulation of the temporal lobes, predominantly the amygdala. Transgenic models of absence seizures (i.e., Generalized Absence Epilepsy in Rats from Strasbourg [52], and Wistar Albino Glaxo/Rijwijk [52, 53]), AD [54, 55] and Huntington’s disease [56] were less frequently used. Conversely, almost half of mouse models (44.2%) were generated using transgenic mice (Fig. 2B), especially for studying neurodevelopmental disorders (i.e., MECP2 [57,58,59,60], Shank3 [57], or CDKL5 [61]), OCD (i.e., Sapap3 [62]) and sleep-wake disorders (i.e., Tg(HCRT-MJD)1Stak [63]) which used transgenic animals exclusively. The C57BL/6J mouse (46.2%) was the most common healthy strain used, with animals being exposed to external factors to induce symptoms. The high percentage of studies using transgenic mouse models may help explain the great difference between the rat and mouse literature reporting on using both males and females in the same study (Rat: 1.9%, Mouse: 19.2%). Nevertheless, most studies utilized only male animals for both species (Fig. 2). Supplementary Table 3 describes the validity of rodent models of disease.

Fig. 2: Rodent strain by disease model, and percentage of animals used by sex.
figure 2

A Rat studies. B Mouse studies. Abbreviations: C57: C57BL/6, CD-1: Cluster of Differentiation 1, GAERS: Genetic Absence Epilepsy Rat, P-rat: alcohol-preferring rat, WAG/Rij: Wistar Albino Glaxo from Rijswijk.

Full size image

Stimulation parameters

DBS parameters are generally programmed by amplitude (Amperes [A] or Volt [V]), pulse width (seconds), and frequency (Hertz [Hz]). Among the included studies, DBS was predominantly administered at high frequency (130 Hz), short pulse duration (60 µs), and amplitudes of 100 µA or 2.7 V. Although the average parameters used in rat and mouse models fell within similar ranges, the average stimulation parameters were higher in rats. Understanding the pattern of stimulation is fundamental for advancing the preclinical field of research and interpreting the results reported. See Supplementary Table 4 for a detailed description of the stimulation parameters reported. In rats, the highest average amplitude was observed in healthy strains, neurodevelopmental disorders and epilepsy, and the lowest average amplitude was reported in models of OCD and tinnitus. However, studies using healthy strains also showed the largest amplitude range, followed by models of movement disorders and dementia/cognition. For mouse studies, the largest average amplitude was observed in models of OCD and physical injury, and the lowest was reported in models of neurodevelopmental disorders. Notably, a few studies reported stimulation intensities above 1000 µA [64,65,66], resulting in a large overall range. Most studies used high-frequency stimulation, with 1 kHz being the max for healthy rats and mice. The lowest frequency was observed in rat and mouse models of epilepsy, as well as rat studies of dementia/cognition and movement disorders. On average, studies in rats employed longer pulse widths; however, both species showed the same mode (60 µs).

There was large heterogeneity surrounding stimulation duration and administration pattern among the included studies. While acute stimulation (i.e., stimulation ≤ 1 day) was more commonly observed in rat studies, chronic stimulation (i.e., stimulation ≥ 2 days) was more frequently applied in mouse studies (Fig. 3). DBS has also been applied intermittently (series of sessions with DBS-on separated by periods of DBS-off), continuously (DBS-on only), or during behavioural testing (the least explored option in both species). Although rat studies similarly used continuous or intermittent stimulation, mouse studies focused on intermittent stimulation, performed twice as often as continuous stimulation.

Fig. 3
figure 3

Stimulation settings for rat and mouse studies.

Full size image

Brain targets

The majority of studies targeted telencephalic brain areas (rat: 43.7%, mouse: 47.9%) and diencephalic areas (rat: 39.7%, mouse: 29.2%). While white matter structures were more commonly targeted in mouse studies (rat: 5.2%, mouse: 16.7%), mesencephalic areas were more frequently targeted in rat studies (rat: 7.1%, mouse 4.2%). The metencephalon was the target of DBS in 4% of rat studies and 2.1% of mouse studies, and the olfactory bulb was targeted in rats only (0.2%; Fig. 4). Across rat studies, the subthalamic nucleus (STN) was the most common target (22.2%), followed by the thalamus (13.5%), nucleus accumbens (nAcc; 11.7%), and frontal cortex (FC; 11.2%, Fig. 4A). Although the majority of studies targeting the STN investigated models of PD, a few articles have also explored this target in models of SUD [24, 67], depression [68, 69], epilepsy [70, 71], OCD [23], and psychosis [72]. Thalamic targets have been studied in various models, with the anterior nucleus of the thalamus (ANT) being predominantly studied in epilepsy [73,74,75,76,77,78,79,80,81,82,83]. The nAcc was targeted in over half of studies investigating SUD and has also been explored in models of eating disorders, depression, OCD, psychosis, and anxiety disorder. Overall, 17% of the studies using healthy strains targeted the core and shell aspects of the nAcc. FC structures have been extensively targeted in healthy strains [84,85,86,87,88,89,90] and in models of depression, SUD [91,92,93], psychosis [72, 94, 95], cognitive performance [96, 97], anxiety [98], ASD [48], eating disorder [99], and PTSD [100], with most studies applying chronic stimulation, including chronic continuous stimulation for several days [68]. Among mouse studies (Fig. 4B), the thalamus and nAcc were the most commonly targeted brain structures (both 14.8%), followed by the STN (13%) and fornix (10.9%). The central nucleus of the thalamus was the thalamic region most commonly targeted (40%), mainly in studies investigating underlying mechanisms of DBS in healthy strains. Stimulation of the nAcc was investigated in models of depression [101, 102], anxiety disorder [103], eating disorder [104], and SUD [105, 106] however, the number of articles on each model is limited. The STN was predominantly explored in the context of movement disorders [107,108,109] (71%), and forniceal stimulation has been investigated exclusively in models of neurodevelopmental disorders [58,59,60] and dementia/cognition [61, 110, 111].

Fig. 4: Brain targets for neurostimulation.
figure 4

The proportion of studies targeting diverse brain areas and the number of articles published by disease model colour-coded for the brain target region. A Rat studies. B Mouse studies.

Full size image

Mechanisms of action

DBS effects on neurochemistry and electrophysiology were the most commonly investigated mechanisms. These studies focused on neuronal activity (26%), gene/protein expression (25.3%), neurotransmitter levels (23%), neuroinflammation and neuroprotection (15.5%), and neural/synaptic plasticity (6.6%). Electrophysiological studies accounted for 31.7% of published articles (Table 1). In models of depression, anxiety disorder, OCD, and PTSD, the DBS-induced antidepressant- and/or anxiolytic effects were associated with changes in serotonin and dopamine levels, as well as modulation of neuronal activity along the mesocorticolimbic circuit. In DBS for PD and movement disorders, the improvements in motor and non-motor symptoms following treatment have primarily been attributed to modulation of the striatal dopaminergic system through the protection of dopaminergic cells and normalization of dopamine signalling. In models of dementia and epilepsy, DBS attenuates aberrant cellular and molecular changes in the hippocampus, a key structure involved in both memory function and seizure generation. Electrophysiological data were limited in models of dementia and memory, OCD, physical injury, and SUD and were not reported in anxiety disorder, bladder function, and eating disorders. In contrast, the majority of studies using non-pathological strains reported on DBS-induced electrophysiology effects, which were target- and parameter-specific.

Table 1 Behavioural, neurochemical, and electrophysiological changes associated with deep brain stimulation based on disease modeled and brain target.
Full size table

Behavioural outcomes

To investigate if DBS treatment would improve symptoms or behaviours, a large portion of the studies included behavioural tests performed either at the end or during treatment (Tables 1–2, Supplementary Table 5). While 73% of rat studies employed behavioural tests, with positive outcomes reported in 83.4%, 81.4% of mouse research investigated behavioural outcomes, reporting improvement in 88.6% of the studies. Although some behavioural tests are more frequently used in one species over the other, several tests are established and commonly performed in both rats and mice. The open-field test was widely used to assess locomotor activity and/or anxiety-like behaviour in several disease models. The Morris Water Maze was used to investigate spatial memory in models of dementia, the Forced Swimming Test was used for assessing immobility (as a measure of behaviour despair) in models of depression, and the Drug Self-Administration paradigm was used in substance use disorder models. Moreover, quantification of food intake was used in models of eating disorders for both species, as well as recording seizure frequency in models of epilepsy.

Table 2 Behavioural outcomes of DBS in rat and mouse models of neuropsychiatric disorders.
Full size table

Adverse events

Only a few studies reported adverse effects from DBS treatment (rat: n = 24, mouse: n = 1). The most commonly reported side effects were brain injury at the insertion site (i.e., hemorrhage, lesions, inflammation) [70, 112,113,114,115,116] and seizures (onset or worsening) [117,118,119,120,121,122] when very high frequencies or amplitudes were used [122]. Following DBS, two articles observed depressive-like behaviours in models of movement disorders [69, 109], three articles reported increased impulsivity [21, 22, 123], and two reported tremors [20, 124]. Few studies observed side effects of anxiety-like behaviour [125], mania [126], unexpected weight change [127], fragmented sleep-wake cycles [53], impaired memory [128], and kindling [55].

Discussion

Rodent models of DBS

The strength of rodent models for DBS research is rooted in dimensions of face (model’s ability to reflect clinical symptoms), construct (similar disease etiology between human condition and preclinical models), and predictive (response to treatments seen in clinical populations) validity. Most rodent models used in DBS studies demonstrate high face and construct validity, being valuable for analyzing the pathophysiology and behavioural effects of treatment, thus facilitating the development and optimization of therapies [129,130,131]. Transgenic models of neurological and psychiatric disorders, as well as healthy strains, have been extensively used to provide insights into the intrinsic mechanisms of DBS, treatment safety, and potential adverse effects of stimulation, significantly contributing to advancing the current knowledge and showing great potential for developing targeted therapies [129,130,131]. These models, however, do not capture all features of human disease but mimic specific relevant symptoms [129,130,131]. The unilateral 6-OHDA model is the most common model of PD and involves the degeneration of nigrostriatal dopaminergic neurons [132, 133] leading to major motor dysfunctions (e.g., akinesia, tremor) and basal ganglia neuroplasticity similar to what is observed in patients [134]. Although these features support the strong face validity of the model, the behavioural and cellular effects produced in this model depend on the site of drug injection [133], and the resulting changes in glutamate release and beta band frequency in response to DBS are different from those observed in humans [135]. Transgenic models of PD are a good alternative, as they present multiple and more complex symptoms, such as observed in patients [107,108,109, 136, 137]. Neurodegenerative diseases of aging, such as AD, have also been studied in rodent models either using aged animals [55], intracranial injection of chemicals [138,139,140,141,142] or transgenic models with mutations in AD-related genes [66, 111, 143,144,145,146]. These models exhibit clinically relevant features, including the development of β-amyloid plaques and Tau protein aggregates and cognitive decline, which are excellent models for DBS research.

In epilepsy research, chemoconvulsants (e.g., kainic acid, pilocarpine, pentylenetetrazole) are frequently used to generate rodents with spontaneously recurrent seizures [147]. These drugs induce extensive hippocampal sclerosis and mossy fibre sprouting in the dentate gyrus [148,149,150], and these models have strong face validity for temporal lobe epilepsy. Electrical kindling models can be generated by delivering chronic stimulation to the hippocampus or the amygdala, allowing for the screening and study of antiseizure treatments for chronic cases of epilepsy [151]. Genetic models are also very attractive to study epilepsy, providing strong construct validity as the mechanisms underlying epileptogenesis more accurately reflect the disease etiology in humans. However, the induction of seizures through chemoconvulsants is not representative of the clinical causes of epilepsy, kindled animals do not typically exhibit spontaneous recurrent seizures, thus limiting the translatability of findings, and transgenic animals may not exhibit the full range of complex alterations observed in patients.

Depression/anxiety are widely investigated models for DBS, having strong predictive and construct validity. Models of depression/anxiety can be generated through neurobiological and genetic manipulation (i.e., selective breeding of rats with a predisposition to depression or knockouts of serotonin-related genes) or via chronic exposure to stressors, which captures the delay in antidepressant efficacy seen in patients [152]. Additionally, these models lead to impairments in the hypothalamic-pituitary-adrenal axis, altered immune reaction, and changes in monoaminergic transmission similar to those detected in patients with depression [153]. Generally, DBS applied to rodent models of depression has demonstrated comparable effects to those observed in patients, such as increased monoamine concentrations in the prefrontal cortex (PFC) [154] and hippocampus [155] and normalization of local field potential activity at the ventral tegmental area (VTA) [156, 157]. Transgenic mice, but not rats, are also widely used in investigations of neurodevelopmental disorders. These include animals with mutations targeting genes known to be related to the disease in humans, such as MECP2 [58,59,60], Shank3 [57], and CDKL5 [61]. In contrast, rat studies focused on exposing naïve rats to chemicals to induce epigenetic changes [158] and subsequent ASD-like phenotypes [47,48,49] or injecting animals with patient-derived antinuclear antibodies to induce Tourette-like symptoms [50].

One major aspect of the experimental DBS landscape that requires dire attention is the need to include females in DBS studies. In this review, we found that only 1.9% of rat studies and 19.2% of mouse studies included females. The consideration of both sexes is important because DBS can have distinct effects on males and females due to differing biological attributes, such as genetics, physiology, brain anatomy, and hormones. For instance, several groups have reported behavioural and morphological differences among male and female 6-OHDA-lesioned rats [159, 160], which may subsequently affect DBS response. Similar observations have been made in models of addiction, whereby female rodents are more likely to consume and seek drugs in self-administration paradigms [161]. Thus, future experimental DBS studies should evolve to include both male and female models in order to address differing therapeutic effects influenced by sex.

DBS parameters and electrode characteristics

DBS parameters such as amplitude, frequency, pulse width, and duration of treatment can vary across brain targets, disease indication, and electrode design. In the clinic, the ranges within which most stimulation parameters fall are in amplitudes between 1–3.5 V, frequencies between 80–185 Hz, and pulse width between 60–210 µs [162]. By means of translation to human DBS, more groups have started to apply clinically relevant DBS paradigms in rodent models. The use of high- or low-frequency stimulation to specific brain areas plays a large role in the overall effects of DBS in both rodent models and humans and may result in different local and widespread plastic brain changes that can be captured in studies evaluating electrophysiological and neurochemical responses associated with DBS [7]. For instance, high-frequency stimulation (HFS; >90 Hz) of the STN in PD models improves motor symptoms, whereas low-frequency stimulation (<50 Hz) is ineffective or exacerbates symptoms [163]. In contrast, although the majority of addiction studies employed HFS, some applying <20 Hz stimulation to the nAcc [25], lateral habenula [164], ventral striatum [165], or STN [67] demonstrated beneficial behavioural effects. In the same manner, there is debate regarding the use of intermittent vs. continuous stimulation. DBS has traditionally been delivered in a continuous fashion in the clinical setting; nonetheless, recent preclinical studies have suggested that a cyclic DBS programming approach may produce beneficial effects [55, 120]. Furthermore, several groups have begun to develop closed-loop adaptive DBS systems and advancements in this field (using both preclinical models and humans) will help us better understand the neural basis and overall effectiveness of such a technology.

The size, shape, and area of microelectrode implantation can affect the volume of tissue activated by stimulation and overall brain tissue reactivity or potential neural damage [166]. The large majority of DBS studies in rodents used bipolar stainless steel electrodes with a diameter between 125–300 µm for rats and 50–125 µm for mice. In recent years, there has been an increase in options for electrode materials such as platinum-iridium, tungsten, and carbon fibres. Platinum is a relatively non-toxic and biologically magnetic inert material in brain tissue and is similar to that used in humans [167]. Tungsten electrodes present low impedance, high conductivity, uniform plastic deformation, and MRI compatibility [168]. The use of carbon fibre electrodes for DBS and neural recording has also gained traction because of its ability to produce fewer MRI artifacts, improve magnetic field homogeneity, and induce smaller temperature changes in MR environments when compared to other metal-based electrodes [167, 169]. However, there are technical challenges surrounding the assembly and implantation of these flexible fibre electrodes, requiring a high level of surgical expertise [170].

DBS brain targets

Regarding brain targets, there is currently no universal consensus on the best target for DBS, notwithstanding the disease indication. Thus, preclinical studies provide great value in our understanding and development of potential therapeutic targets. The most evident example is the DBS of the STN in PD, which is now an extensively used target in the clinic. The STN was found to exhibit unusually increased activity in animals with Parkinsonian symptoms, and experimental lesions of the STN in rats resulted in evident improvements in motor dysfunction [171]. These findings supported the hypothesis that pathological activity occurs in the STN in PD, and modulation of this area can improve Parkinsonian symptoms. Accordingly, studies using rodent models have demonstrated that STN-DBS significantly reduces tremors, rigidity, and bradykinesia associated with PD. However, consistent with clinical observations, some rodents treated with STN-DBS also exhibited enhanced depressive-like behaviours [68, 69, 172]. Thus, other key brain structures, such as the entopeduncular nucleus [35, 173], pedunculopontine tegmental nucleus [116, 174,175,176,177], substantia nigra [178, 179], and zona incerta [172] have been considered for DBS targeting in movement disorder rodent models in hopes of alleviating non-motor symptoms.

Epilepsy research has concentrated on stimulating various aspects of the thalamus to alleviate seizure activity. Thalamic nuclei are highly interconnected with substantial bidirectional projections via the mammillothalamic, thalamocortical, and spinothalamic tracts. Thus, thalamic DBS has been shown to significantly reduce seizure frequency and increase seizure latency, as detected using electroencephalography [180]. DBS targeting the thalamus has also been studied in the context of AD, whereby high-frequency stimulation was shown to be less effective at facilitating spatial memory than the entorhinal cortex or fornix DBS [141]. In line with this finding, Hamani et al. [128]. reported that DBS delivered to the ANT at high amplitude (i.e., 500 µA) disrupts the acquisition of contextual fear conditioning and, in turn, may explain the impaired performance on spatial alternating tasks observed in these rats. In contrast, a recent study by Chamaa et al. showed that both single and repeated sessions of ANT-DBS may induce a significant increase in neurogenesis in the ipsilateral dentate gyrus [181] and improve spatial reference memory on the Y-maze test [182].

In the field of psychiatric disorders, the nAcc is a key DBS target explored for the treatment of SUD, eating disorders, treatment-resistant depression, OCD, and psychosis. Rodents receiving nAcc stimulation show a decrease in consumption, preference, and/or motivation for consuming several substances such as ethanol [26, 106, 183, 184], cocaine [24, 25, 67, 91, 92, 164, 185, 186], and morphine [93, 112, 187,188,189,190]. While most studies showed success in modulating pathologic behaviour using nAcc-DBS, there was conflicting evidence regarding the efficacy of stimulating the nAcc core vs. shell. Several addiction studies showed reduced morphine-induced conditioned place preference and ethanol self-administration regardless of whether the DBS was targeting the nAcc core or shell. In contrast, Vassoler et al. [91] and Wilden et al. [183] demonstrated that DBS of the nAcc shell, but not the nAcc core, reduced cocaine-induced reinstatement. Similar to this observation, Oterdoom et al. [191] reported a reduction in binge-eating behaviour among rats treated with DBS targeting the nAcc shell but not in those targeted at the core aspect of the nucleus. Furthermore, Sesia et al. [21] reported reduced impulsivity when targeting the nAcc core and the opposite effect when targeting the nAcc shell of OCD models. The contradicting findings regarding the optimal DBS target may be explained by the variation in volumes of tissue activated due to differing stimulation parameters applied and differences in cellular characteristics between models.

Mechanisms of action

In rodent DBS research, the most frequent study goal is to offer a better understanding of the mechanisms of action of treatment. Several hypotheses on the mechanism of DBS have been proposed, such as direct inhibition/excitation of neural activity, whereby DBS modulates pathological oscillatory activity within brain networks [7]. However, the precise action of DBS in each brain target and in each neurological or psychiatric disorder remains elusive.

Parkinson’s disease

DBS is considered the standard of care for patients with PD, and the STN represents the most well-studied target for this indication. Several studies describe a neuroprotective effect of STN-DBS on dopaminergic neurons in the substantia nigra [30, 31, 192,193,194] via increased signalling strength in the BDNF-trkB pathway (i.e., brain-derived neurotrophic factor and its receptor tropomyosin receptor kinase B), and increased autophagy through the blocking of protein phosphatase 2A activation [195, 196]. However, these results failed to be replicated in transgenic models of α-Synuclein pathology [197]. Increased levels of BDNF and cerebral dopamine neurotrophic factor in the nigrostriatal system and primary motor cortex have also been associated with positive outcomes following STN-DBS [115, 195, 198,199,200]. Furthermore, STN-DBS is thought to reduce neuroinflammation [201,202,203] due to decreased fractalkine pathway signalling [201,202,203]. Conflicting results have been reported on extracellular levels of dopamine following STN-DBS [204,205,206,207,208], with some studies showing the increase in dopamine metabolites to be associated with a better outcome [204,205,206,207,208]. STN-DBS has also been reported to inhibit serotonergic neurons in the raphe nuclei [29, 209,210,211] and decrease serotonergic transmission in the striatum, hippocampus, and PFC [212, 213], leading to depression [109, 172, 212, 213]. This effect was not observed following entopeduncular nucleus DBS, suggesting that the therapeutic effect of DBS via decreased serotonergic transmission is characteristic of STN-DBS [214, 215]. Electrophysiological studies have associated decreased beta oscillations in cortical and basal ganglia networks [43, 208, 211, 216, 217] and increased alpha power in the medial PFC [218] with improved motor and non-motor symptoms following treatment. In models of motor symptoms not associated with dopamine degeneration (e.g., haloperidol-induced catalepsy, pharmacologically-induced tremor and tardive dyskinesia [219, 220]), STN-DBS improves motor function through the modulation of serotonin transmission hippocampal-striatal coherence, and cortical beta oscillations [29, 209,210,211].

Dementias and Alzheimer’s disease

DBS represents an emerging therapeutic approach for patients with dementia. Improved memory and enhanced hippocampal neurogenesis in the dentate gyrus have been reported following DBS of several brain targets, including the fornix [54, 66], nucleus basalis of Meynert (NBM) [221], medial septum [142, 222], entorhinal cortex (EC) [144], ventral striatum [223], and thalamus [181, 182], and this effect was associated with the restoration of theta oscillations in thalamic nuclei and striatum following treatment [224]. Fornix- and NBM-DBS have also been shown to reduce hippocampal neuronal loss and neuroinflammation [54, 66], whereby fornix-DBS increased hippocampal acetylcholine levels and activity [142, 225]. NBM-DBS promoted cholinergic fibre growth in the cingulate cortex without altering hippocampal acetylcholine levels [221]. Additionally, DBS of the intralaminar thalamic nuclei leads to increased dendritic spine density in cortical and hippocampal pyramidal neurons [64, 138, 221] and altered cortical glutamatergic neurotransmission [226]. The neuropathological hallmarks of AD have also been explored following EC- and fornix-DBS treatments. While both EC- and fornix-DBS reduced cortical and hippocampal amyloid deposition [54, 143, 144, 146], EC-DBS also decreased tau protein levels in the cortex and hippocampus [143, 144, 146] by enhancing enzymes involved in tau clearance [143, 144, 146].

Epilepsy

The ANT is the primary target for treating epilepsy with DBS, as it is approved by the Food and Drug Administration as an adjunctive treatment for reducing the frequency of partial-onset seizures in adult populations [227]. Positive changes in the hippocampus following ANT-DBS have been shown, with enhanced neurogenesis, increased neurotrophic factors, reduced neuroinflammation, reduced mossy fibre sprouting, and anti-apoptotic and neuroprotective effects [75,76,77,78,79,80, 228,229,230]. ANT-DBS anti-seizure effect was also associated with increased adenosine, reduced adenosine kinase levels and changes in gamma-aminobutyric acid (GABA) transmission in the hippocampus [230,231,232,233,234]. Differentially expressed genes involved in ion channel activity, glutamatergic synapse, and regulation of immune response have also been reported in the hippocampus following ANT-DBS [235]. Electrophysiological studies described ANT-DBS to enhance the seizure threshold by decreasing theta oscillations in the hippocampus and cortex and by suppressing delta oscillations during non-rapid eye movement sleep, which results in an increased amount of rapid eye movement sleep [82, 230, 234].

Addiction

It is widely accepted that the initial reinforcing effects of addictive drugs are mediated by enhanced synaptic concentrations of dopamine in forebrain subcortical structures, particularly the nAcc [236, 237], and the prolonged use of drugs leads to an imbalance between excitatory glutamate and inhibitory GABA [236, 238]. Preclinical studies on nAcc-DBS for the treatment of addiction have shown antidromic stimulation of the cortex via cortico-accumbal afferents [91, 165, 239], decreased glutamate and increased GABA levels in the VTA, ventral pallidum, and nAcc [190], upregulation of glutamatergic receptors in the nAcc [105] and amygdala [240], and downregulation of glutamatergic receptors in the VTA [164]. Interestingly, the selective blockade of dopamine 1 receptor (D1R) along with nAcc-DBS reverses cocaine-induced hyperlocomotion and plasticity changes in GABAergic medium spiny neurons expressing D1R via activation of metabotropic glutamatergic receptors [105], suggesting an intricate interaction between neurotransmitter systems is required for the therapeutic effects of DBS.

Eating Disorder

DBS treatment for eating disorders has been described in patients suffering from morbid obesity. The nAcc and the ventral and/or lateral aspects of the hypothalamus are the targets of choice in this patient population, and although DBS was considered to be a safe procedure with encouraging outcomes, it has been associated with side effects such as hardware infection and stimulation-induced mania [241, 242]. nAcc-DBS in rodent models of obesity is associated with decreased high fat/ high sucrose intake [104, 191], improved glucagon and glucose concentrations in plasma and increased neuronal activation in the lateral hypothalamus [243]. Lateral hypothalamus DBS led to reduced body weight and food intake [244, 245], increased metabolism in the mammillary body, hippocampus, and amygdala, and decreased metabolism in the thalamus, caudate, temporal cortex, and cerebellum [245]. This treatment was also associated with reduced markers of activity-induced synaptic plasticity and memory formation in the hippocampus and EC and vesicular GABA transporter [246].

Obsessive-compulsive disorder

Changes in the mesolimbic dopaminergic reward system play a significant role in the pathological habit formation observed in OCD [236, 237]. In clinical populations, the anterior limb of the internal capsule, the nAcc, and the inferior thalamic peduncle are common targets for neuromodulatory treatments of drug-resistant OCD [247,248,249]. In rodent models of OCD, DBS targeting the internal capsule led to increased neuronal activity in the targeted area and in the PFC, which was associated with therapeutic effects [62]. Similarly, DBS of the ventral striatum increased neuronal activity in the targeted area and in the PFC, orbitofrontal cortex (OFC), and amygdala [62, 123]. When targeting the core aspect of the nAcc specifically, DBS reduced OFC neuronal firing and enhanced spontaneous local field potential oscillatory activity in the slow frequency band [239, 250]. nAcc-DBS was also shown to increase dopaminergic and serotonergic in nAcc [21]. Thalamic DBS, however, has a distinct behavioural outcome when different subnuclei are targeted; while DBS of the mediodorsal nucleus of the thalamus increases impulsivity and decreases neuronal activity in cerebellar nuclei and PFC [22], DBS of the parafascicular nucleus of the thalamus alleviates prepulse inhibition [251], a measure of sensorimotor gating which is disrupted in OCD [252, 253]

Depression

Given its key role in emotional processing and depressotypic behaviour, the infralimbic (IL) nucleus (also referred to as the mouse ventromedial PFC, a structure considered to be homologous to the human subgenual cingulum) is the most commonly studied DBS target in models of depression. IL-DBS was shown to enhance serotonin release and induce long-term alterations of serotonin receptor expression [85, 254, 255], an effect that can be blocked by depleting serotonergic transmission [255], and is thought to be mediated by direct modulation of prefrontal projections to the dorsal raphe nucleus [254,255,256,257]. Enhanced levels of serotonin were also reported following DBS of the nAcc [258,259,260], lateral habenula, and medial forebrain bundle [261, 262]. In addition to serotonergic transmission, IL-DBS is also modulated by adenosinergic and glutamatergic neurotransmission as pre-treatment with selective antagonists attenuates the antidepressant effect of treatment [257, 263]. Furthermore, IL-DBS induces beneficial neuroplasticity, such as hippocampal neurogenesis [264, 265], and increased dendrite length [154, 266], with associated increases in BDNF [264, 265, 267,268,269,270].

Post-traumatic stress disorder

The amygdala and IL have been the primary DBS targets for treating PTSD, as hyperactivity of the basolateral amygdala (BLA) is associated with PTSD symptom severity [271]. BLA-DBS has been shown to decrease and normalize local neuronal activation [272, 273] and reverse fear condition-induced changes to synaptic plasticity in the cortical-amygdala connections, suggesting BLA-DBS may disrupt the long-term retention of fear memory [274]. However, it is important to note that the chronic electrical stimulation of the temporal lobes (especially the amygdala) is known to induce seizures, therefore being a well-established model of kindling [275, 276]. IL-DBS was shown to mitigate PTSD-like behaviours while reducing BLA neuronal activity by activating GABAergic interneurons through PFC-BLA projections [100], being an effective and safer target for DBS.

Neurodevelopmental disorders

ASD and Tourette’s syndrome are the most commonly studied neurodevelopmental disorders in preclinical DBS research. In ASD models, IL-DBS restores the abnormal serotonergic transmission and modulates the expression of glutamatergic and GABAergic receptors in the targeted area while improving social deficits, anxiety-like behaviour, and hyperactivity [48]. Fornix-DBS in ASD models shows an antidepressant effect [58] and reduces memory deficits [60, 61] via normalization of the expression of genes related to intellectual disabilities [58] and restoration of hippocampal synchrony, spontaneous excitatory postsynaptic currents, long-term potentiation and feedforward inhibition [59,60,61]. Thalamus-DBS has been shown to improve repetitive behaviours in ASD models by restoring functional connectivity in corticostriatal and corticolimbic circuits and decreasing dopaminergic receptor expression in the striatum [47]. In a Tourette’s syndrome model, however, thalamus-DBS reduced local theta band activity and local field potentials in the striatum [49]. In this model, entopeduncular DBS decreased dopamine concentration and dopamine transporter in the striatum, leading to decreased tic behaviour [50].

Schizophrenia and psychosis

Several targets have been investigated for improving positive symptoms in models of schizophrenia and psychosis. While targeting the hippocampus with DBS restores the dopaminergic activity of the VTA [277], targeting the VTA with DBS leads to increased GABA neurotransmission but not dopaminergic activity [278]. Medial septum-DBS has also been shown to restore dopaminergic transmission while improving serotonergic activity [94, 95] and reducing hippocampal gamma oscillations [20], which leads to attenuated psychotic schizophrenic symptoms. Similarly, thalamic-DBS modulates neural oscillations, alleviating aberrant thalamic oscillatory activity and modulating coherence in thalamocortical networks [279].

Traumatic Brain Injury

In the context of TBI, DBS has been used to address decreased levels of consciousness, deficits in cognitive function, and motor recovery. The hypothalamus is a key region involved with wakefulness and alertness, particularly through orexin neurons of the lateral hypothalamus that have widespread projections throughout the brain [280]. In a weight-drop model of TBI, lateral hypothalamus-DBS promoted consciousness recovery along with increases in adenosine A1 receptor, decreases in GABA receptor, increases in orexin receptor, and decreased low-frequency delta oscillations in the PFC [280]. In addition, brain regions with strong connectivity with the hippocampus, such as the medial septum, have also been targeted, as these regions play a critical role in cognition. DBS of the medial septum was linked to the restoration of hippocampal theta oscillations and improved cognitive functions following TBI [281]. Furthermore, the cerebellum has been targeted for its role in influencing motor function through extensive thalamocortical projections, and DBS of the cerebellum was shown to enhance motor functions alongside the elevation of perilesional neuronal activity and suppression of neuroinflammation and apoptotic markers [45].

Limitations of DBS studies with rodent models

A few limitations need to be taken into account when considering the direct translation of data from rodent models to the clinical setting. Although experimental models have provided much value to our understanding of the underlying mechanisms and effects of DBS, it remains important to consider that these models do not fully reflect human disease but rather mimic a series of key symptoms. For instance, DBS of the fornix and lateral hypothalamus has shown promising outcomes in rodent models of AD [54, 111] and obesity [244, 245], respectively. However, the application of these approaches has not been as successful within the clinical population [282, 283], which alludes to the caution that needs to be taken when translating such therapeutic regimens to humans. The high rate of positive behavioural responses observed in these studies should also be interpreted with caution, as there is a tendency in scientific publications to focus on publishing positive results rather than negative ones. Furthermore, the anatomical organization of rodent brains is similar to that of the human brain in many aspects; however, key differences in cortical processing and, therefore, cognitive abilities prevent the direct translation of findings, especially in the context of top-down inhibitory control of motivated behaviours. Also, anatomical differences between rodents and humans in specific clinically relevant brain targets may be impeditive to a translation to rodent models. Lastly, the make of DBS electrodes, the stimulation parameters and stimulation settings (i.e., acute, chronic) used in rodents are highly variable and are typically different from those used in humans [284].

Despite these limitations, rodent studies have contributed substantially to our current knowledge of the mechanisms underlying DBS treatment. Advancements in the field have led to the development of novel stimulation techniques and rodent DBS devices that are fully implantable and/or with wireless stimulators for long-term use in freely moving animals [285]. These applications are only growing as new technologies such as closed-loop circuits [286] and specialized stimulation electrodes [167,168,169] improve the precision of targeting and decrease adverse effects.

Ethical considerations

While DBS is an efficacious surgical treatment for many neurological and psychiatric disorders in adult and pediatric populations, the precise neurobiological mechanism of action of DBS treatment remains elusive. Thus, translational research is needed to advance our understanding of disease processes and treatment mechanisms to develop novel, less invasive and more efficacious therapies. Animal experimentation, however, must be performed under strict ethical guidelines to answer sound scientific questions that cannot be addressed in in-vitro or in-silico models [287,288,289]. The “Three Rs” principle (Reduction, Refinement, and Replacement) sets the ethical standard for in-vivo research. The reduction principle refers to reducing the number of animals used per experiment or study while keeping statistical power, the Refinement principle refers to the improvement of all methods used in research to minimize pain, suffering, distress or lasting harm and improve animal welfare, and the Replacement principle refers to the full or partial replacement of live animals with technologies or alternative approaches [287,288,289].

In clinical settings, distinct ethical considerations must be made when performing DBS trials. Questions on how to provide the best care for vulnerable patients at the late stages of life, especially those with progressive degenerative disorders, must be made to assess if the benefits of treatment are superior to the potential complications [290, 291]. In patients with psychiatric disorders, major concerns arise related to the patient’s capacity to provide informed consent, the possibility of altering behaviour via brain stimulation, as well as the risks and benefits compared to traditional and less invasive treatment methods [292, 293]. Also, DBS for treating psychiatric disorders shares many features with psychosurgeries and, therefore, raises several ethical and legal concerns that must be openly discussed [294]. For the pediatric population, it is crucial to assert the potential risks and benefits of DBS, the optimal time for offering the treatment and the long-term consequences of brain stimulation [295]. Also, it is necessary to discuss the role of the informed assent given by the child in addition to the informed consent signed by the parent/guardian [295]. A bioethical framework to advance ethical discussions regarding pediatric DBS has been proposed and involves the protection of the child’s best interest, the consideration of the developmental context, the creation of strategies for mitigating known and unknown risks, the critical appraisal of the adult literature, and fostering communication and collaboration among practitioners [296]. Finally, there are several concerns related to specific devices’ capacity for detecting neuronal activity and the safe storage of these data [297]. With the advancement of DBS devices, new policies must be developed to maximize benefits and minimize harm to patients.

Conclusion

Though there has been extensive research into the effects of DBS targeted to key brain structures, there remains a dire need for standardized, protocol-based approaches to find optimal stimulation targets and parameters for distinct pathologies. Studies on rodent models have not only shed light on possible mechanisms of action of DBS treatment but also improved our understanding of the underlying disease processes. Models of psychiatric and neurodevelopmental disorders have implicated dysfunctions in the monoaminergic system and in the functional connectivity of the mesocorticolimbic network in the pathophysiology of diseases, and, therefore, modulation of these aspects with DBS resulted in improved behaviour. Similarly, the cellular and molecular complexity of PD has been further explored in rodent models, which are capable of capturing both motor and non-motor symptoms of the disease. In these models, DBS improves symptoms via modulation of the striatal dopaminergic system and suggests that treatment response is associated with dopamine availability. Although there are several limitations in translating findings from preclinical to clinical settings, these studies have shown that direct modulation of neural activity (both cellular and molecular) is among the major mechanisms of action, albeit further research is necessary for a complete understanding of this neuromodulation therapy.