Application of optogenetics and in vivo imaging approaches for elucidating the neurobiology of addiction

Abstract

The neurobiology of addiction has been an intense topic of investigation for more than 50 years. Over this time, technological innovation in methods for studying brain function rapidly progressed, leading to increasingly sophisticated experimental approaches. To understand how specific brain regions, cell types, and circuits are affected by drugs of abuse and drive behaviors characteristic of addiction, it is necessary both to observe and manipulate neural activity in addiction-related behavioral paradigms. In pursuit of this goal, there have been several key technological advancements in in vivo imaging and neural circuit modulation in recent years, which have shed light on the cellular and circuit mechanisms of addiction. Here we discuss some of these key technologies, including circuit modulation with optogenetics, in vivo imaging with miniaturized single-photon microscopy (miniscope) and fiber photometry, and how the application of these technologies has garnered novel insights into the neurobiology of addiction.

Introduction

Despite the commonly held belief that addiction is a failure of moral agency [1, 2], a large body of human and animal studies have established neurobiological factors as powerful drivers of drug-seeking behavior and addiction [3–5]. Evidence supports that genetic heritability contributes to addiction risk, with roughly half of addiction vulnerability being heritable [6–8].

Early pioneering studies from Olds and Milner [9, 10] identified key brain regions involved in reward and reinforcement, which made apparent that discrete brain regions and projections underlie behaviors central to addiction. These early experiments were a major impetus for intense scientific interest in identifying the neural correlates of drug addiction. At the time, there were few techniques for modulating brain function, mainly consisting of targeted lesions, electrical stimulation, and drug microinjection [11–13]. These techniques were restricted by a lack of cell-type and projection specificity and poor temporal resolution. In the last two decades, the rapid development of novel approaches for precisely observing and manipulating neuronal activity has allowed for the study of neural circuits that regulate reward and motivated behaviors. In this review article, our goal is not to provide a comprehensive discussion of drug addiction research, but rather to highlight how modern techniques in neuroscience have been applied to study cellular and circuit mechanisms of drug addiction (Fig. 1). We will first discuss representative studies that utilized optogenetics for mapping and manipulating reward circuitry and examine their impacts on drug-induced synaptic plasticity and behavior. We aim to update and complement earlier excellent reviews on these topics [5, 14–17]. Next, we will review the application of microendoscopic (miniscope) imaging and fiber photometry of genetically encoded Ca²⁺ indicators (GECIs) and fluorescent biosensors in the context of drug addiction. We refer the reader to recent in-depth reviews that discuss the different in vivo fluorescent imaging techniques with an emphasis on their strengths and limitations [18, 19]. In addition, recent reviews have covered the development, optimization, properties, and applications of fluorescent biosensors for various neurotransmitters and modulators [20, 21]. Miniscope Ca²⁺ imaging of activity changes in various brain cell types in animal models of neurodegenerative diseases has been reviewed [22]; the present review will focus on the application of miniscope and fiber photometry techniques to study circuit mechanisms of reward and addiction.

Fig. 1: In vivo optical tools for studying addiction neural circuitry.

Optogenetics, miniscope, and fiber photometry have distinct but complementary applications for manipulating and observing neural activity.

1. Optogenetics

One of the most remarkable accomplishments in the field of neuroscience has been the development of optogenetics, which, for the first time, allowed for the targeted modulation of defined cell types and projections in the brain, both in vitro and in vivo [23–25]. In optogenetics, light-sensitive ion channels or ion pumps are selectively expressed in cell populations of interest, followed by implantation of an optical fiber. Light of particular wavelengths delivered to the expressed construct through the optical fiber elicits ion flux that results in neuronal excitation or inhibition [26]. Over the last 20 years, many microbial opsins have been discovered and progressively engineered to optimize and customize a library of unique optogenetic constructs. These opsins have differing conductance magnitudes, ionic selectivity, kinetics, excitation spectra, and subcellular targeting profiles, thereby offering a diverse array of tools for investigating neural circuit function [26, 27]. The application of optogenetics has advanced our understanding of neural mechanisms of drug addiction on two major fronts: (1) anatomic circuit mapping with investigation of drug-induced cellular adaptations and (2) in vivo modulation of neural activity in behaving animals. In this section, we highlight several representative studies where these applications of optogenetics were leveraged to further elucidate addiction neurobiology.

1.1. Optogenetic circuit mapping

In optogenetic circuit mapping, opsins are expressed in a specific cell type, followed by photostimulation of axon terminals in the projection area, which induces postsynaptic currents in the receiving neurons [28]. Optogenetic circuit mapping has been instrumental in defining and characterizing the neurocircuitry of addiction [4, 29]. Central to this circuitry is the mesocorticolimbic pathway, consisting of dopaminergic projections from the ventral tegmental area (VTA) to the nucleus accumbens (NAc) and prefrontal cortex (PFC). Optogenetic circuit mapping has been used to delineate the cell-type and projection-specific organization of the mesocorticolimbic circuit, which was otherwise not possible with conventional electrophysiological approaches. Optogenetic circuit mapping challenged the dogma that the direct and indirect pathways of the NAc mirrored that of the dorsal striatum [30]. In the dorsal striatum, medium spiny neurons (MSNs) of the direct pathway almost exclusively express dopamine D1 receptors, whereas MSNs of the indirect pathway express dopamine D2 receptors [31]. Largely due to a lack of specific tools for assessing circuit connectivity, it was assumed that MSNs of the NAc were similarly segregated with regard to dopamine receptor type and projection target. Kupchik et al. [30] expressed channelrhodopsin 2 (ChR2) in NAc D1- or D2-expressing MSNs of the NAc core and demonstrated that about half of the neurons in the ventral pallidum (VP)—canonically part of the indirect pathway—receive synaptic input from D1-MSNs. Further, they demonstrated that VP neurons that project directly to the thalamus receive input from both D1- and D2-MSNs. These results indicate that the NAc direct and indirect pathways are not simply defined by dopamine receptor expression nor by the effect on thalamic activation or inhibition. Similarly, optogenetic circuit mapping has provided a more nuanced understanding of the nature of projections from different subnuclei of the NAc to different cell types within the VTA [32, 33].

Although past research has focused on VTA dopamine neurons, recent studies have also revealed critical roles for VTA glutamate, GABA, and neurotransmitter co-releasing populations in the regulation of reward and aversion [34]. For example, optogenetics was used to demonstrate that VTA glutamate neurons project to numerous cell types within the NAc including parvalbumin GABAergic interneurons, MSNs, and cholinergic interneurons, and that they can regulate reinforcement-related behavior independent of dopamine [35–37]. Further, VTA glutamate/GABA co-releasing neurons project to the VP and lateral habenula, where they have opposing effects on neuronal activity due to differences in the relative strength of glutamatergic and GABAergic input [36].

In addition to mapping anatomic connections, optogenetics provides a means for investigating the cellular and synaptic adaptations that occur in the reward circuity after exposure to drugs of abuse. Animals with opsins expressed in defined circuits can be exposed to a drug, such as in a drug self-administration paradigm, followed by ex vivo electrophysiological recording of photostimulation-induced synaptic currents to identify the pathway-specific synaptic adaptations. This approach was used to study input-specific glutamatergic plasticity in the NAc, which plays a fundamental role in cocaine-seeking behavior [38, 39]. Although it was known that calcium-permeable AMPAR (CP-AMPAR) expression increases during the incubation of cocaine craving and contributes to drug seeking [40], it remained unknown whether specific inputs were differentially regulated during withdrawal. By expressing ChR2 in the basolateral amygdala (BLA), infralimbic (IL) cortex, or prelimbic (PrL) cortex, it was demonstrated that IL and BLA projections to the NAc underwent insertion of CP-AMPARs during the incubation period, whereas PrL to NAc projections underwent insertion of non-CP-AMPARs [41, 42]. Strikingly, cocaine self-administration led to the generation of NMDAR-expressing silent synapses in BLA-NAc, IL-NAc, and PrL-NAc synapses, which were subsequently unsilenced by AMPAR insertion during the incubation period. Optogenetics was also utilized to induce long-term depression (LTD) of these projections in vivo, which reversed the incubation-induced plasticity via removal of synaptic AMPARs and, depending on the pathway targeted, either promoted or reduced the incubation of cocaine craving. These representative studies demonstrate how optogenetics can serve as a powerful tool for both identifying the anatomical organization of neural circuits involved in addiction and for interrogating synaptic adaptations that occur with exposure to drugs of abuse or in unique behavioral contexts. Further, as illustrated by the studies that utilized optogenetics to induce LTD in vivo, optogenetic circuit modulation can be leveraged to investigate the role of discrete pathways and synaptic adaptations in regulating addiction-related behaviors.

1.2. Optogenetic circuit modulation in addiction

In vivo optogenetic manipulation can directly test how key cell types and pathways regulate reward processing and drug-seeking behavior. In addition, optogenetics can be used in creative ways to manipulate synaptic and circuit physiology that extends beyond simple activation or inhibition. In this section, we highlight representative studies that illustrate the application of optogenetics to investigate circuit mechanisms of drug seeking, incubation of drug craving, and triggered relapse.

A foundational observation in the study of addiction was that midbrain dopamine neurons encode reward prediction error, whereby their firing rate increases or decreases depending on the presence of a reward and the animal’s expectation for that reward [43, 44]. However, it was challenging to resolve whether these neurons drive reinforcement behavior without a means of selectively manipulating dopamine neurons. By specifically expressing ChR2 in VTA dopamine neurons, Tsai et al. [45] showed that in vivo optogenetic induction of phasic but not tonic action potential firing produced conditioned place preference (CPP) in mice. Further, optogenetic self-stimulation of VTA dopamine neurons was reinforcing [46, 47] and sufficient to induce addiction-like behavior, including cue-induced relapse and a persistence of self-stimulation despite paired foot-shock punishment [48]. Subsequent studies further demonstrated a central role of VTA dopamine neuron activation in drug-induced synaptic plasticity. Cocaine or morphine exposure induced the expression of CP-AMPARs in VTA dopamine neurons and in NAc MSNs, and optogenetic stimulation of VTA dopamine neurons alone mimicked this effect [48–50].

The major projection targets of VTA dopamine neurons are D1- and D2-MSNs in the NAc [51]. However, it was difficult to determine the discrete roles of these neurons prior to the advent of optogenetics, as they are interspersed throughout the NAc [52]. Implementation of optogenetic manipulation in D1- or D2-Cre transgenic mice has demonstrated that stimulation of NAc D1- and D2-MSNs had opposing effects on cocaine CPP; D1-MSN stimulation enhanced, whereas D2-MSN stimulation reduced cocaine CPP [53]. Similarly, optogenetic stimulation of dorsal striatal D1- and D2-MSNs induced reinforcement and aversion, respectively [54]. However, this dichotomy between D1- and D2-MSNs was challenged by findings that both MSN populations can promote reward or aversion depending on the pattern of optogenetic stimulation; brief stimulation promoted reinforcement and cocaine CPP, whereas prolonged stimulation promoted aversion [55]. Thus, optogenetics has revealed specific cell types and activity patterns that regulate reinforcement and drug reward in critical nodes of addiction circuitry.

Caveats should be considered when interpreting optogenetics experiments—particularly that optogenetic activation can induce a robust, artificial activation of neurons and projections that does not accurately mimic natural signaling patterns [56, 57]. Optogenetic inhibition provides an alternative approach for interrogating circuit function that can help determine the necessity of a pathway for a particular behavior. In an early optogenetics study, inhibition of neurons or axon terminals in the PrL or NAc core reduced both cocaine-primed and cue-induced reinstatement of drug seeking [58]. Another group showed that optogenetic inhibition of lateral orbitofrontal cortex (OFC) to BLA projections disrupted cocaine-seeking induced by conditioned light and tone stimuli [59]. Both studies demonstrated that these projections are critical to reinstated cocaine seeking.

Lastly, an important extension of optogenetics has been to induce targeted manipulation of synaptic plasticity, particularly through the optogenetic depotentiation of specific pathways that underwent potentiation in addiction-related behavioral paradigms. An early study has shown that cocaine exposure potentiates excitatory transmission in the IL to NAc D1-MSN pathway, and that in vivo optogenetic depotentiation of this pathway reverses cocaine-induced locomotor sensitization [60]. Further, as referenced earlier, in vivo optogenetic LTD of NAc inputs from the BLA, IL, or PrL during withdrawal from cocaine self-administration reversed the “unsilencing” of these synapses that occurred during incubation of cocaine craving [41, 42]. Interestingly, depotentiation of inputs from the BLA and PrL inhibited incubation of cocaine craving, whereas depotentiation of input from the IL potentiated craving, thus highlighting how input-specific synaptic plasticity can have differential behavioral effects. Similarly, in vivo optogenetic reversal of cocaine-induced plasticity at the ventral hippocampus and medial prefrontal cortex (mPFC) inputs to NAc D1-MSNs impaired response discrimination and reduced response vigor, respectively [61].

Recent studies have also demonstrated that in vivo optogenetic induction of long-term potentiation (LTP) can be used to induce targeted synaptic plasticity. Although NAc MSNs have been well studied in the context of addiction, the region also contains fast-spiking interneurons (FSIs) that can orchestrate the output of MSN ensembles through GABAergic inhibitory gating that suppresses motivated action [62]. In vivo LTP of excitatory BLA synapses to NAc FSIs increased inhibition of MSNs and expedited the acquisition of cocaine self-administration, indicating that this proportionally small interneuron population significantly modulates the output of NAc principal neurons [63]. These studies highlight the breadth of insights that can be garnered by in vivo optogenetic manipulation of both circuit activity and synaptic plasticity in behavioral models of addiction.

2. Biosensors for in vivo optical imaging

Observing neural activity in specific cell types and circuits in freely behaving animals is a powerful means for identifying and characterizing the neural correlates of addiction-related behavior. Our understanding of how these behaviors are encoded has been facilitated by recent developments in in vivo optical imaging techniques and indicators of neuronal activity or neurotransmitter release. We first review basic principles of these tools and technologies for in vivo imaging, then discuss how their application has led to advances in the study of addiction.

In the past decade, GECIs [64] and indicators for membrane voltage changes [65] have been developed to monitor neuronal activity in real-time (reviewed by ref. [66]. GCaMP6 has been the most widely used GECI, although additional GECIs continue to be developed with different fluorescence spectra, Ca²⁺ sensitivities, kinetics, and signal-to-noise ratios that can be tailored to experimental needs [66]. GECIs can be expressed in specific neuronal populations and their projections. For example, a double-floxed inverse open reading frame (DIO) or flip-excision (Flex) version of GCaMP can be expressed in select neuronal cell types using Cre-driver mice or rats, and fluorescent imaging can be performed at the level of the cell body or axon terminals. Promotor-driven GECI expression provides another method for cell-type-specific labeling. For example, the CaMKIIα promoter can drive GCaMP6 expression in forebrain excitatory neurons [67], whereas the mDlx promoter has been used to drive GCaMP6 expression in forebrain GABAergic interneurons [68]. Pathway-specific expression of indicators can also be achieved by injecting the retrogradely transported viral vector canine adenovirus-2 (CAV-2) [69] or recombinant adeno-associated virus (rAAV) vector rAAV2-retro [70] expressing Cre recombinase into a projection target together with injection of a Cre-dependent GECI into an upstream brain region. In vivo biosensor imaging can also be combined with optogenetics or electrophysiology in multi-modal experimental designs.

Numerous fluorescent biosensors have recently been developed for detecting the release of various neurotransmitters and neuromodulators, including dopamine [71, 72], norepinephrine [73], acetylcholine [74], and endocannabinoids [75], with extensive ongoing research to engineer novel biosensors. Recent comprehensive reviews have highlighted the development and application of these biosensors [21, 76]. Concurrently, advancements in methods for in vivo optical imaging, such as two-photon microscopy, single-photon miniaturized microscopy (miniscopes), and fiber photometry have enabled GECI and neurotransmitter biosensor imaging in behaving animals. Readers are directed to excellent reviews on details and comparisons of these techniques, including their advantages and limitations [18, 77]. In the following sections, we will highlight their applications within the drug addiction field.

3. Miniaturized single-photon microscopy

Miniscopes use gradient-index (GRIN) lenses to image hundreds of cells at single-cell resolution in behaving animals [78]. Similar to conventional fluorescence microscopy, a miniscope is equipped with an excitation light source (typically a 470 nm light-emitting diode), a dichroic mirror, objective lens, and a sensor to detect emitted light. Two key differences include the GRIN relay lens and the small footprint complementary metal oxide semiconductor sensor for detecting emitted light. Miniscopes provide several key advantages compared to other technologies for observing neural activity. Although electrophysiological microelectrode arrays are similarly capable of recording from tens to hundreds of neurons simultaneously, miniscope imaging of GECIs allows for the observation of activity from specific cell types and pathways. Two-photon microscopy can also accomplish this and allows for high-resolution imaging but is limited by its restriction to regions at or near the brain surface. In contrast, the use of microendoscopic lenses of varying lengths and diameters enables miniscope imaging in most brain regions. Although miniscopes detect greater background signal than two-photon microscopy, a comparison of these technologies showed highly correlated signal amplitudes as well as signal-to-noise ratios [79]. Miniscopes may also be less susceptible to brain motion due to faster acquisition frame-rates [80]. These and other advantages of miniscopes have allowed for novel insights into the neural mechanisms of drug addiction. For further technical details of miniscope design and function, readers are directed to other reviews [18, 77, 81].

Miniscope imaging of GECIs in brain regions implicated in addiction has emerged as a powerful tool to study the activity of specific cell types and pathways associated with drug reward and seeking. Here we discuss representative studies that utilized miniscopes to achieve this goal. We focus on the following applications: (1) study of changes in neuronal activity during distinct behaviors; (2) identification and investigation of functional neuronal ensembles; and (3) application of machine learning to decode behavioral state information represented in neural activity. To study changes in neuronal activity during drug-seeking behaviors, the frequency and pattern of Ca²⁺ transients can be quantified and linked to behavioral states. For example, miniscope imaging was used to investigate the activity of different cell types in the VP, a key output nucleus of the NAc, in a mouse model of cocaine self-administration [82]. GABAergic and glutamatergic VP neurons expressing GCaMP6f were imaged during extinction or cue-induced cocaine seeking; in aggregate, GABA neurons had increased activity during cued cocaine seeking, whereas glutamate neurons had increased activity during extinction learning. However, the responses of different GABA and glutamate neurons were heterogenous, as subpopulations of these neurons had distinct activity patterns that deviated from the aggregate signal. This study highlights how miniscopes can reveal both the aggregate and individual responses of specific cell types in distinct drug-seeking behaviors.

The implementation of miniscopes in freely moving animals also allows for correlation of neuronal activity with spatial positioning of the animal. The glutamatergic projection from the IL to the NAc has been thought to be an “anti-relapse” circuit that inhibits drug seeking [38, 83]. Miniscopes were used to observe the activity of NAc shell-projecting IL neurons in a rat model of cocaine self-administration [84]. The IL-NAc shell-projecting neurons were targeted by expressing CAV-2-Cre in the NAc shell and Cre-dependent GCaMP6f in the IL. IL-NAc shell neurons displayed reduced activity during active lever presses. The extent of inhibition in IL activity decreased between withdrawal day 1 and day 15, and correlated with increased drug seeking on day 15. Also, a large proportion of IL-NAc shell neurons were found to be spatially selective and to code for the animal’s location in the self-administration chamber without obvious bias to the locations of the active or inactive levers. Such spatial selectivity was significantly reduced during the incubation period, which could contribute to increased cocaine motivation.

Miniscopes have proven to be particularly beneficial for the study of neuronal ensembles in animal models of addiction. A neuronal ensemble can be defined as a group of neurons that is recruited during a particular behavioral task or neural computation, an idea first proposed by Semon [85] and further described by Donald Hebb [86, 87]. Techniques for studying neuronal ensembles historically have relied on post hoc immunostaining for cellular markers of neuronal activity, including cytochrome oxidase [88] and immediate early genes (IEGs) such as c-Fos and Arc [89]. These markers are expressed within minutes of increased neuronal activity and enable the functional mapping of activated neurons throughout the brain [90]. However, post hoc immunostaining for IEGs is limited by its low temporal resolution and requirement for sacrificing animals, which precludes the investigation of neural activity dynamics during behavior.

Although traditional approaches for studying neuronal ensembles have provided key insights into the distributed neuronal populations involved in addiction [91], miniscopes have allowed for the identification and observation of ensembles in behaving animals. In the dorsal striatum, miniscope imaging of GCaMP6s-expressing D1- or D2-MSNs was carried out in freely moving mice [92]. D1- and D2-MSN ensembles were identified and found to be spatially clustered within the dorsal striatum. Both D1- and D2-MSNs displayed increased activity after motion initiation and decreased activity after motion termination. Although cocaine increased locomotor activity, it did not result in increased activity of these neuron clusters. It did, however, alter the organization and connectivity of MSN clusters relative to normal locomotion [92]. Miniscope imaging of neuronal ensembles was also carried out to investigate how dorsal hippocampal CA1 pyramidal neurons encode contextual associations for nicotine place conditioning [93]. Although it is well-established that hippocampal “place cells” in the CA1 encode for spatial location in an environment and are essential for spatial learning and memory [94], the involvement of these neurons in drug-induced contextual conditioning remains incompletely understood. Miniscope imaging of CA1 pyramidal neurons in the dorsal hippocampus showed that nicotine place conditioning activates a unique CA1 neuronal ensemble, and that this ensemble was reactivated in the nicotine-paired context in the absence of nicotine. Further, activation of these CA1 neurons was necessary for the expression of nicotine place preference. This study demonstrates that CA1 ensembles integrate nicotine-contextual information for subsequent recall and are required for the reward-context association.

Although statistical analyses can correlate neural activity with behavior, machine learning approaches can further extend these conclusions to approach causality. In the aforementioned study on D1- and D2-MSN clusters in the dorsal striatum, a supervised machine learning algorithm (decision tree) was implemented to predict ambulation and non-ambulation states based on cluster activity [92]. The investigators compared algorithm predictions based on neural clusters, randomly selected groups of cells, and whole population activity, to show that the neural clusters indeed were responsible for encoding ambulation start and stop. Predictions can also be made for future behaviors. In an investigation of the involvement of the mPFC to dorsal periaqueductal gray projection (dPAG) in compulsive alcohol drinking, Ca²⁺ dynamics of these neurons during the initial alcohol experience on day 1 were shown to correlate with emergent drinking phenotypes 2 weeks later [95]. mPFC-dPAG neurons were most inhibited during alcohol drinking on day 1 in mice that later displayed a compulsive drinking phenotype; these mice continued to consume alcohol in the presence of a bitter deterrent 2 weeks later. A supervised machine learning algorithm (support vector machine classifier) was successfully trained on this data to predict the compulsive drinking phenotype, providing support for the idea that differences in neural activity can predispose to addiction-like behavior even with exposure to identical environmental conditions. As implemented in the above studies [92, 95], supervised machine learning methods train an algorithm on a set of data where the predicted variable is verified during training. However, unsupervised methods rely on no such a priori knowledge. Unsupervised machine learning methods have also been implemented on miniscope data to identify neural representations of behavioral states [96, 97].

Advancements in lens and miniscope technology, as well as computational analysis methods, have greatly expanded the scope of in vivo fluorescent imaging in recent years. Commercial products and open-source designs have been developed with various specifications customized to different applications, including wired and wireless models (for a comparison, see ref. [98]). Several newer advancements promise to aid the future investigation of addiction neural circuitry; cranial windows allow observation of dorsal cortical surfaces, GRIN lenses allow for visualization of subcortical and deeper brain regions [99], and multiple cortical layers can be visualized using an angled prism probe [100]. The compact NINscope (named after the institute of origin) is an open-source design with further reduced weight and footprint, which can be used to observe two separate brain regions in one mouse [101]. Two miniscopes have been used simultaneously in two different animals to measure the neural correlates of social interaction [102]. Miniaturized two-photon microscopy that enables imaging in freely-moving or head-fixed animals is also currently under development [103]; however, this method still requires expensive table-top equipment with limited portability. Finally, the Inscopix nVoke miniscope (Palo Alto, CA, USA) offers integrated optogenetic manipulation capability to confirm causal links in neural circuits [104]. Combined with rapidly advancing technologies, creative experimental design and innovative analysis methods will continue to allow miniscopes to resolve the cellular and network-level underpinnings of addiction.

4. Fiber photometry

Fiber photometry is used to observe and quantify fluorescence signals from cells expressing fluorescent indicators such as GECIs or fluorescent neurotransmitter biosensors, thereby providing a read-out of neural activity or neurotransmitter release. Unlike miniscopes, fiber photometry records an aggregate signal from groups of neurons near an implanted optical fiber. Dichroic mirrors and collimators direct excitation light of one or more wavelengths to an implanted optical fiber, and the emitted fluorescent signals are detected by photoreceivers. In order to discern signals from fluorophores with overlapping emission spectra, modulating the voltages driving excitation light with phase-offset frequencies is followed by signal demodulation with lock-in amplification [18, 81]. Isosbestic excitation (405 nm) is frequently used to control for artifacts arising from fiber bending and tissue autofluorescence. Alternatively, an activity-independent fluorophore such as green fluorescent protein (e.g., for jRGECO1a) or mCherry (e.g., for GCaMP) can be used as a control. Fiber photometry offers reduced technical and analytical complexity relative to miniscope imaging. It also decreases tissue damage when imaging deep brain regions, which can be advantageous for experiments that do not require single-cell resolution. Regardless, the two methods are highly complementary to each other [97]. Studies using fiber photometry or miniscopes that have particular relevance to addiction are compiled in Table 1.

Table 1.

Psychoactive drugs alter GCaMP activity and dopamine release in reward circuits of the brain.

Alcohol	↓ GCaMP6m, mPFC-dPAG; during initial alcohol exposure	[95]
	GCaMP6f, dLight; differential activity patterns in subregional VTA-NAc projections	[117]
Cocaine	↓ GCaMP6m, VTA (DA) ↓ GCaMP6m, DRN (5-HT)	[105]
	↑ GCaMP6f, NAc (D1-MSN); prior to entry into cocaine-paired chamber ↓ GCaMP6f, NAc (D2-MSN); after entry into cocaine-paired chamber	[106]
	↑ GCaMP6f, VP (GABA); cued cocaine seeking ↑ GCaMP6f, VP (Glu); extinction learning	[82, 106]
	↑ GCaMP6f, IL-NAc; withdrawal	[84]
	↑ dLight, NAc	[116]
	GCaMP6s, dorsal striatum; altered organization and connectivity of MSN clusters	[92]
Heroin	↑ GCaMP6m, VTA (DA)	[105, 116]
	↑ GCaMP6m, DRN (5-HT)	[105]
	↑ dLight, NAc	[116]
MDMA	↓ GCaMP6m, VTA (DA) ↓ GCaMP6m, DRN (5-HT)	[105]
Methylphenidate	↑ GRABDA, dorsal striatum	[71]
Morphine	↑ dLight, NAc; intermittent dosing ↓ dLight, NAc; continuous dosing	[115]
Nicotine	↑ GCaMP6m, VTA (DA)	[105]
	GCaMP6f, CA1; neural ensemble activated during CPP reactivated in the nicotine-paired context in the absence of nicotine	[93]

↑ Increase, ↓ decrease, DA dopamine, Glu glutamate, 5-HT 5-hydroxytryptamine.

Traditionally, in vivo electrophysiology has been used to study how drugs of abuse affect neuronal activity. However, it remains challenging to determine the anatomical, morphological, or genetic identity of the recorded cells. Using fiber photometry, neuronal activity can be observed in genetically defined cell populations and specific projections. A straightforward yet illuminating study used fiber photometry to investigate how drugs with diverse mechanisms of action affect the activity of VTA dopamine and dorsal raphe nucleus serotonin neurons, which are key substrates that contribute to drug reinforcement [105]. Cocaine and 3,4-methylenedioxy methamphetamine both caused long-lasting suppression of dopamine and serotonin neuron activity, with both drugs similarly inhibiting these neurons via D2 or 5-HT_1A autoreceptors. In contrast, heroin increased the activity of both cell populations, whereas nicotine activated only VTA dopamine neurons. This study highlights the utility of fiber photometry for investigating the mechanisms and time course of drug effects on specific cell populations of interest.

Similarly, a study utilized fiber photometry to investigate how D1- and D2-MSNs in the NAc core are affected by cocaine and associated environmental cues during cocaine CPP [106]. Baseline activity in D2-MSNs was greater than in D1-MSNs and cocaine administration inhibited D2- but enhanced D1-MSN activity. In test sessions after cocaine place preference conditioning, D1-MSN activity peaked just prior to mouse entry to the cocaine-paired chamber, whereas D2-MSN activity was suppressed immediately after entering the chamber. Notably, chemogenetic inhibition of D1-MSNs eliminated the increase in D1-MSN activity prior to entry into the cocaine-paired chamber and abolished the acquisition and expression of cocaine CPP. Thus, observation of neuronal activity linked to specific behaviors or environments can help elucidate the neural encoding of drug-paired contexts.

A powerful experimental approach is to identify neuronal activity patterns with fiber photometry, then modulate the activity of these neurons to determine their specific contribution to behavior. Particularly suitable for this is the combination of fiber photometry with optogenetics, which can re-create or inhibit specific firing patterns with temporal precision. Mice learned to lever press for optogenetic self-stimulation of VTA dopamine neurons and were later exposed to painful footshocks in association with light delivery; this addition of footshocks differentiated mice into two groups: “renouncers” that decreased lever pressing with footshocks and “perseverers” that continued to compulsively self-stimulate despite this punishment [107]. Previously, it was found that perseverers had significantly increased numbers of c-Fos-expressing cells in the OFC [48]. To better determine whether OFC activation contributes to the perseverant phenotype, fiber photometry was employed to record GCaMP6m fluorescence from OFC axon terminals projecting to the dorsal striatum. During punished sessions, GCaMP6m fluorescence increased around lever-press events in perseverers but decreased in renouncers. Critically, optogenetic inhibition of OFC neurons in persevering mice decreased punishment resistance, indicating that increased OFC-dorsal striatum activity underlies this compulsive behavior.

On the other hand, fiber photometry has been used to identify the consequences of behaviors induced by optogenetic stimulation. It was recently demonstrated that optogenetic stimulation of VTA dopamine neurons is itself sufficient to induce Pavlovian conditioning in the absence of natural rewards [108]. When light and tone cues were paired with phasic stimulation of VTA dopamine neurons, despite the absence of an external rewarding stimulus, the cues subsequently evoked conditioned locomotor responses and time-locked Ca²⁺ increases in dopamine neurons. These effects did not occur when cues and optogenetic stimulation were separated in time. These studies illustrate how multi-modal experimental approaches with fiber photometry and optogenetics can link changes in neuronal activity with behavioral outputs. A distinct advantage of fiber photometry is that it allows for the simultaneous detection of multiple fluorophores expressed in different cell types in the same brain region. One approach is to inject green and red fluorophores that are activated by Cre (Cre-On) or inactivated by Cre (Cre-Off) into Cre-driver mice in brain regions with heterogeneous cell populations [109]. Using this approach, GCaMP6f and jRGECO1a were expressed in direct- and indirect-pathway spiny projection neurons (SPNs) of the dorsal striatum in both D1-Cre (direct-pathway-specific) and A2A-Cre (indirect-pathway-specific) mice [110]. Strikingly, Ca²⁺ transients in direct- and indirect-pathway SPNs were highly synchronous within each hemisphere but were asynchronous across hemispheres. The magnitude of activation in the direct and indirect pathways coordinately determined the direction and vigor of movements. These results extend on an earlier study using only one fluorophore (GCaMP3), which determined that both direct and indirect pathways are activated during movement initiation but could not record activity from both pathways simultaneously [111]. Simultaneous imaging of direct- and indirect-pathway SPNs further illuminated how the relative magnitude of activity of these neurons predicts movement dynamics [110].

Microdialysis and voltammetry have traditionally been used to detect neurotransmitters and chemical substances in the brain in awake animals. In vivo microdialysis involves perfusing fluid into the brain using a probe with a semi-permeable membrane and then collecting dialysates for subsequent analysis [112]. Microdialysis, however, lacks temporal resolution and is labor-intensive. Fast-scan cyclic voltammetry (FSCV) is an electrochemical technique for detecting neurotransmitter release using a carbon fiber electrode [113]. However, since monoamines oxidize at similar potentials and can be released in the same brain region, FSCV lacks specificity for differentiating between dopamine, serotonin, and norepinephrine [114]. The development of fluorescent biosensors for neurotransmitter detection has overcome many of these limitations and provides exciting opportunities for visualizing neurotransmitter dynamics in specific cell types in behaving animals.

Perhaps the most well-studied fluorescent biosensors are the genetically encoded dopamine sensors, GPCR-activation-based-dopamine (GRAB_DA) [71] and dLight [72]. These sensors detect in vivo dopamine release with subsecond resolution, submicromolar affinity, and high molecular specificity. Fiber photometry is a powerful tool for in vivo imaging of biosensors, as they can be targeted to specific cell types, and single-cell resolution is typically not required [71, 72]. In proof-of-concept studies, these sensors were used to detect dopamine release in response to natural and drug rewards [71, 72]. Sexual behavior and water delivery in water-restricted mice induced an increase in GRAB_DA fluorescence, whereas systemic administration of methylphenidate, a psychostimulant and norepinephrine-dopamine reuptake inhibitor, caused a prolonged increase in dopamine release in the dorsal striatum [71]. Similarly, sucrose reward consumption induced an increase in dLight and jRGECO1a fluorescence in NAc neurons. On the other hand, footshocks suppressed dLight fluorescence while enhancing jRGECO1a fluorescence, indicating a dissociation between dopamine release and local circuit activity [72]. In addition, cocaine, but neither the selective serotonin reuptake inhibitor citalopram nor the norepinephrine reuptake inhibitor reboxetine, increased the intensity of dLight fluorescence in the NAc [115]. These studies validate GRAB_DA and dLight as powerful tools for detecting dopamine release that occurs in response to natural and drug rewards in freely behaving animals.

Several studies have taken advantage of these tools to study how opioids and alcohol affect dopamine dynamics in rodents. Heroin administration increased GCaMP6m fluorescence in VTA dopamine neurons and dLight fluorescence in the NAc [116]. Consistent with this, heroin increased the number of c-Fos-expressing dopamine neurons in the VTA and occluded the optogenetic self-inhibition of VTA GABA neurons. These results support the idea that heroin increases dopamine release through disinhibition of VTA dopamine neurons. Dopamine biosensors were also used to study how different patterns of drug exposure can produce differing effects on behavior [115]. Morphine was given in intermittent doses that resulted in short periods of withdrawal; compared to a continuous infusion, intermittent dosing led to increased amplitude of spontaneous dLight transients in the NAc and locomotor sensitization, despite similar levels of serum morphine after a week under both protocols. In addition, intermittent dosing exacerbated morphine-evoked transcriptional adaptations in the NAc and dorsal striatum, perhaps as a consequence of altered dopamine dynamics. dLight was also used to investigate the spatiotemporal dynamics of dopamine release in various phases of alcohol self-administration [117]. Patterns of correlated VTA Ca²⁺ and NAc dLight fluorescence emerged over the course of self-administration training, which vanished during extinction and reappeared during relapse. Further analysis suggested that differential activity patterns in subregional VTA-NAc projections were involved in context-induced reinstatement of drug-seeking and reacquisition of alcohol consumption, which are related but distinct models of relapse. Extinction training did not return the mesolimbic circuits to a naive state, which may help explain why extinction training is not typically an effective treatment for alcohol dependence. Continued innovation in biosensors and new technological developments in fiber photometry, such as wireless fiber photometry [118] and multi-site recording using high density fiber arrays [119], will provide further flexibility for experimental design and expand our understanding of how cell activity in relevant neural circuits contributes to behavioral dysfunction in addiction models.

Conclusion

Enormous strides have been made in developing novel tools for manipulating and observing neural activity in behaving animals. Application of these tools has led to a better understanding of the mechanisms underlying reward processing and drug addiction. Optogenetics enables dissection of functional neural circuits involved in addiction and characterization of drug-induced cellular and synaptic adaptations. Meanwhile, miniscopes and fiber photometry provide real-time monitoring of cell-type- and pathway-specific neuronal activity in vivo. Further, optogenetics can be combined with miniscope or fiber photometry, allowing for simultaneous manipulation and observation of neural activity [104]. Looking forward, the continued development of wireless miniscope [120–122] and fiber photometry [118] technology will likely facilitate and improve behavioral studies, as animals would no longer be tethered, minimizing behavioral interference. This will be particularly beneficial in the setting of self-administration experiments where additional cables can interfere with intravenous infusion tubing. Also mentioned earlier, the compact NINscope enables observation of two separate brain regions in one mouse [101], while multi-fiber arrays for fiber photometry can record up to 48 regions at once [119]. To our knowledge, wireless technologies, NINscope, and multi-fiber arrays have yet to be utilized in addiction research. There remains immense untapped potential for implementing these exciting technologies to advance research into the neurobiology of drug addiction.