Dissecting the catecholamines: how new approaches will facilitate the distinction between noradrenergic and dopaminergic systems.

Abstract

To overcome the limitations of fast scan cyclic voltammetry (FSCV) to discriminate between catecholamines, we discuss new approaches to monitor the dynamics of these neurochemicals with high spatial, genetic, and temporal specificity.

Catecholamine neurotransmitters are abundant and central for brain function. Dopamine, norepinephrine, and epinephrine each have important roles throughout development, and are targets for pharmacotherapeutics, as alteration of normal catecholaminergic function has been implicated in a wide range of neurobiological disorders (i.e. affective, neurodegenerative, etc.). Despite the importance of these neurochemicals, the ability to finely tune our real-time detection and manipulation of their signaling has proved elusive. However, new techniques are on the horizon for just this purpose. These approaches, such as selective optogenetics with fast scan cyclic voltammetry (FSCV), machine learning to distinguish signals from FSCV recordings, and biosensors coupled to selective fluorescent indicators, will progress our understanding of catecholaminergic function in physiological and pathological processes.

Limitations to FSCV

FSCV has become a common technique for measuring catecholaminergic function with high temporal precision. The millisecond resolution allows release and reuptake dynamics to be recorded following single, time-locked events such as reward/cue presentation. Because of their small probe size (50–150 μm × 5–7 μm), carbon fiber microelectrodes can be placed in neuroanatomically distinct regions, with either in vivo, or ex vivo electrochemistry. Therefore, FSCV uniquely provides the spatial and temporal resolution for the analysis of biogenic amines. However, there are some limitations to FSCV that need to be acknowledged, addressed, and improved upon. For example, the common usage of electrical stimulation to evoke release can be problematic because typically the stimulation parameters are set to supraphysiological frequencies compared to the firing frequencies of catecholamine neurons. Furthermore, these excessive stimulation parameters may be sufficient to also drive release of other chemical signals within the stimulated brain region, such as other neurotransmitters, growth factors, neuropeptides, etc. Additionally, although measurements of dopamine release in the striatum are relatively easy to achieve because of the dense dopaminergic inputs there, measuring catecholamines in other regions with less dense innervation becomes increasingly difficult. For instance, measurements of norepinephrine release in the prefrontal cortex has been notoriously complicated despite the known functions of norepinephrine there. Finally, the electroactive moiety that allows for the oxidation and reduction of both dopamine and norepinephrine is the catechol group integral to both transmitters, and traditional chemometric analysis cannot parse the relative contributions of each neurotransmitter to the overall signal. As the field has been dominated by striatal dopamine-centric hypotheses, assumptions are that any noradrenergic signal included in the voltammogram are negligible and have put the onus on noradrenergic researchers to show that their results are not due to dopamine via pharmacological manipulations. Recent data, however, have pointed to new hypotheses that further complicate our understanding of catecholaminergic neuronal function, which is that “norepinephrine” neurons can co-release dopamine with functional signaling¹. These studies complicate traditional FSCV analyses because the corelease of these catecholamines in the same neuroanatomical region and temporal specificity make the advantages of using FSCV now limitations. However, we believe that new approaches to FSCV experiments and additional methodology will begin to address some of these limitations and further dissect apart the roles of dopamine and norepinephrine in brain function.

Selective transgenic animals for optogenetic FSCV

Recently, the use of optogenetics has improved the neuroanatomical and cell-type specific control of select neuronal populations. Several transgenic mouse lines have been created to express opsin proteins selectively in catecholaminergic neurons. The use of these lines coupled with FSCV has enhanced our understanding of some of the neural mechanisms regulating catecholamine release. For example, by using a mouse line expressing Cre-recombinase under the regulation of the dopamine transporter (DAT) promoter crossed with an Ai32 line (expressing Cre-dependent channelrhodopsin2-eYFP in the ROSA locus) a DAT::CrexAi32 line was created to optically control dopamine neurons selectively without the need for viral surgeries. Using these mice to measure light-evoked dopamine activity, the authors were able to highlight the influence of acetylcholine signaling on facilitating dopamine release following multiple presentations of pulse trains³. We used a similar transgenic approach with a DBH::CrexAi32 mouse line to examine the regulation of norepinephrine release following stress exposure in a recent manuscript⁵. Again, using transgenic mouse lines to selectively express channelrhodopsin in neuronal populations provides enhanced specificity to probe catecholaminergic signaling. However, selective activation of noradrenergic neurons may still be limited, as it lacks the ability to address issues of co-release of both norepinephrine and dopamine from the same neurons or synaptic vesicles.

Machine Learning to separate FSCV signals

The implementation of machine learning algorithms trained to distinguish neurochemical signatures from the cyclic voltammogram may better resolve catecholamine signals than traditional chemometric analysis. Moran et al. successfully used a supervised-learning approach to distinguish serotonin from dopamine signals from FSCV traces². They went on to use their dual-transmitter model to analyze voltammograms from human striatum measured during a stock market game and found that serotonin signals encode elements of prediction error. The ability to apply FSCV approaches in a human disease state opens the door for future possibilities to investigate the neurochemical basis of other disorders as well. Additionally, we find that electrodes probed with equivalent concentrations of both dopamine and norepinephrine (0.3, 1.0, 3.0 μM) reveal a slight, but significant (F_{1, 4} = 126.9, p<0.001), shift in the oxidation potential at the peak current (Δ = 0.0371 ± 0.00345 V, n=5 electrodes, Figure 1 A, B). There was also a significant main effect of concentration on peak oxidation potential where larger concentrations resulted in anodic shifts (F_2,8 = 20.59, p<0.001, Figure 1B). Furthermore, with regards to the peak current measured by each concentration, there was a concentration by catecholamine interaction (F_2,8 = 70.80, p<0.0001), and while there was a linear relationship for peak current across concentrations for both catecholamines (dopamine r²=0.903, norepinephrine r²=0.9183) the slope of the lines were significantly different from each other (p<0.01). These findings suggest that the amount of endogenous catecholamine released has impact on the chemical reaction occuring on the electrode, and that there can be large differences with regard to peak oxidative current between electrodes examined, highlighting the need for careful calibration of electrodes, and consideration of how much catecholamine is being releaed in a given brain region. Furthermore, our data in combination with the approaches by Moran et al. provide an exciting possible expansion of FSCV analysis into designing similar machine learning algorithms to distinguish between catecholamines. These algorithms could then be applied to studies like in Kempadoo et al. (described above) in which pharmacological manipulations and ex vivo slice stimulation point to the functional release of dopamine from noradrenergic neurons.

Figure 1.

Shifts in the oxidation and reduction peaks between dopamine (DA) and norepinephrine (NE) may allow for supervised machine learning to distinguish the catecholamines. A) Cyclic voltammogram of equal concentrations of dopamine (black) and norepinephrine (blue) from the same carbon fiber electrode in a flow cell. Hash lines indicate potential at peaks. B) Quantification of cathodic shift in potential at peak of dopamine and norepinephrine samples. C) Differential sensitivity to catecholamines across concentrations.

*** p<0.001; **** p<0.0001.

Biosensors for fluorescent visualization of catecholamine activation

Finally, the development of fluorescent biosensors will also significantly advance our ability to monitor catecholamine signaling. Recently, a modified dopamine receptor that fluoresces when in a bound conformation (dLight1) was described for use with fiber photometry⁴. Excitingly, the authors included supplemental information describing similar sensors designed from modified β- and α-adrenergic receptors, κ- and μ-opioid receptors, a 5HT-2A receptor, and a melatonin type-2 receptor. As this method of modifying receptors to fluoresce in a bound state improves, we imagine a library of tools to quantify ligand availability for any receptor with a known structure. This approach also allows for projection specificity when combined with genetically-targeted viral delivery approaches. Furthermore, improvements in fiber photometry, microscopy, and endoscopy will only strengthen the usefulness of biosensors to visualize neurochemical signaling.

Conclusions

The use of FSCV to monitor electroactive neurochemicals in near real-time has been a major advance for studies of brain functions, but like all tools, there are limitations to its abilities. Here we have highlighted difficulties in using FSCV to divide the catecholamines. However, we believe that the development of new approaches to create or analyze FSCV data, along with new biosensors designed to selectively signal catecholamine availability will generate new answers to the role these neurotransmitters play in neural signaling.