Emerging Trends in in Vivo Neurochemical Monitoring by Microdialysis

Abstract

Mapping chemical dynamics in the brain of live subjects is a challenging but highly rewarding goal because it allows neurotransmitter fluctuations to be related to behavior, drug effects, and disease states. A popular method for such measurements is microdialysis sampling coupled to analytical measurements. This method has become well-established for monitoring low molecular weight neurotransmitters, metabolites, and drugs, especially in pharmacological and pharmacokinetic studies. Recent technological developments which improve the temporal and spatial resolution of the methods will enable it to be used for studying behavior and small brain nuclei. Better assays allow monitoring more neurotransmitters simultaneously. Extension to analysis of aggregating proteins like amyloid β are proving extremely useful for uncovering the roles of these molecules and how they contribute to neurodegenerative diseases.

Introduction

Understanding the brain is one of the grand challenges facing science. Basic studies of brain function also have tremendous health and economic consequences as mental and neurological diseases such as Alzheimer’s, Parkinson’s, and depression are frequently long term, debilitating, and costly to treat resulting in significant loss of human potential and economic drain. In this review we consider progress made in monitoring neurochemicals in live animals as a method to study the brain. Neurons communicate by releasing neurotransmitters which diffuse across the synaptic cleft to interact with receptors on post-synaptic neurons. Besides this classical point-to-point communication, neurons may communicate by “volume transmission” wherein they release neurotransmitter or neuromodulator that diffuses beyond the synapse to affect neighboring neurons [1–2]. Measuring signaling chemicals and metabolites in the extracellular space can provide important insight into this chemical communication. By making such measurements in awake subjects, it is possible to correlate the chemical dynamics with behavior, disease progression, and drug effects in intact circuitry.

Tremendous challenges confront the scientist attempting in vivo neurochemical measurements. Neurotransmitters may change concentration rapidly requiring high temporal resolution. In some cases, one would like to follow changes that develop slowly over time, such as in the progression of drug dependency or a neurodegenerative disease, requiring great stability in measurements. Brain extracellular space is a soup of neurotransmitters, metabolites, energy molecules, and proteins so the chemical complexity of samples is high. Spatial compartmentalization of concentrations creates a challenge of spatial resolution as well. Measurement of dynamics within individual synaptic clefts represents the most extreme challenge. At a more macroscale level, the measurement within specific brain nuclei and sub-nuclei are important as such structures can be activated independent of neighboring regions and control specific behaviors or functions. Chemical release within sub-regions of brain nuclei < 1 mm³ can have identifiable functions [3–4]. A practical challenge is that for most experiments involving animal subjects, it is desirable to allow movement to observe behavior during the measurement. Therefore analytical measurements must be compatible with “samples on legs”.

Several techniques have been developed that allow in vivo neurotransmitter measurement including positron emission tomography (for reviews see [5–8]), cell based sensors [9], fluorescent tracers [10], implantable electrochemical sensors [11–15], and microdialysis sampling [16–17]. Although exciting advances have been made in all these areas recently, we will restrict our review to microdialysis sampling. In this approach, samples are collected from the brain extracellular space using a microdialysis sampling probe and resulting fractions analyzed for neurochemicals of interest. Traditionally this method has lower temporal resolution than the best sensors and lower spatial resolution because of the probe sizes. On the other hand, decoupling of sampling and detection allows the method to be extremely versatile. Thus, any analytical method can be coupled with the sampling probe to achieve the necessary sensitivity and selectivity for measurement. Methods may also allow multiple components to be determined in one fraction to study interactions among neurotransmitter systems. This method has been used for decades and is well established for pharmacological and pharmacokinetic studies. While many important neurochemical studies have used microdialysis in the past few years, in this review we will focus on recent technical advances and new applications for this approach to in vivo neurochemical monitoring. Although is microdialysis sampling is increasingly used for clinical studies [18–20], in this review we focus on advances related to fundamental neuroscience.

Improvements in Sampling: Large Molecules

In microdialysis, a semi-permeable membrane is used as the probe. Substances in the extracellular space can cross the membrane according to their concentration gradient. Perfusion of the interior space both maintains the concentration gradient and drives the collected sample out of the brain for collection and analysis. A particular advantage of microdialysis is the ability to detect larger molecules (peptides and proteins) in the extracellular space that might be difficult to detect at conventional sensors. Many peptides, including both neuropeptides and other signaling molecules like cytokines, are present at low concentrations making their collection and assay difficult. An improvement in sampling is to add antibodies or other affinity agents to the perfusion flow to enhance the concentration gradient and recovery of the probes (see Figure 1) [21–25]. This approach has been used to enhance recovery several fold for neuropeptides and cytokines.

Figure 1.

Illustration of using antibodies to enhance recovery in microdialysis. In normal microdialysis, analytes (represented by shapes) cross the membrane according to their concentration gradient. In “antibody-enhanced” microdialysis, antibody added to the perfusion fluid (“y”-shapes) results in low concentration of free analyte in the dialysate. This maintains a high concentration gradient of free analytes resulting in higher recovered total concentration. (Adapted from [23].)

Measuring even larger proteins requires membranes that have high molecular weight cut-offs. Although it has been known for some time that large proteins can be sampled using microdialysis membranes with high molecular weight cut-off; this approach has been taken in an exciting direction by sampling proteins like amyloid-β (Aβ) [26–30] and tau [31]. After the initial validation of this method (see Figure 2), it has been used extensively to show that these proteins, which are known to form toxic plaques associated with Alzheimer’s disease, are released by neuronal activity in a way that explains regional vulnerability to plaque formation [26,28]. This important discovery has led to several insights about the pathology of Alzheimer’s and the role of these proteins in normal function. Similar studies of sampling of α-synuclein [32], the aggregating protein associated with Parkinson’s disease, have been reported suggesting that microdialysis will be a generally useful tool for studying this family of diseases and these proteins.

Figure 2.

In vitro validation of method to measure free amyloid β (Aβ) in the extracellular space of a living brain by microdialysis. (A) Diagram of exchangeable Aβ. Triangles represent potential Aβ binding molecules, e.g., apoE, clusterin, and α2M. Only highlighted Aβ molecules are of the appropriate size to pass through a 35 kDA MWCO membrane on the microdialysis probe. (B) Interpolated zero flow method to quantify the pool of measurable Aβ_1-x and Aβ40 within samples of human CSF (n = 4). At 2.2 µl/min, the percentage recovery of eAβ was 9.74 ± 1.53% (mean ± SEM). C, In vitro percentage recoveries for each Aβ species using the interpolated zero flow method. Each recovery point contains error bars and are overlapping for each species (n = 4). In vitro recovery of each Aβ species by microdialysis is the same. (D) The concentration of eAβ and total soluble Aβ are highly correlated within a sample of human CSF (Pearson’s r = 0.9487; p < 0.0001; n = 9). The mean concentrations of soluble Aβ 1-x and e Aβ 1-x were 30.47 ± 4.23 ng/ml (mean_SEM) and 196.2 ± 28.65pg/ml, respectively. (E,F) Human CSF immunoprecipitated (IP’d) for Aβ has undetectable levels of total soluble Aβ and eAβ. Human CSF spiked with an amount of exogenous Aβ40 peptide expected to double Aβ concentration resulted in a 2.1-fold increase in total soluble Aβ and a 1.9-fold increase in eAβ. (Figure and legend adapted from [27].)

Temporal Resolution

To a first approximation, temporal resolution of microdialysis is limited by the mass sensitivity of the analytical method used for assay. That is, enough sample must be collected per fraction to be able to detect the compounds of interest. As shown several years ago, highly sensitive methods like capillary electrophoresis with laser-induced fluorescence detection allow assay of fractions collected every few seconds to dramatically improve the temporal resolution over the minutes required when using HPLC for analysis [33–35]. The high speed of CE separations also made it compatible with the large number of samples generated by collecting fractions at seconds intervals. Work with such methods revealed that further improvements in temporal resolution were limited by flow and diffusion broadening of concentration zones as they were transported from the probe to the analytical system. While this broadening could be limited by using short connecting tubing, this approach was impractical for experiments with freely moving animals. This problem was solved by segmenting the outflow of a dialysis probe into droplets separated by an immiscible fluid (fluorinated oil) [36–37] (see Figure 3A). Segmentation was performed using microfluidic tee that could be mounted on the head of the animal to create a nanoliter volume fraction collector. Segmented samples do not mix by flow or diffusion and therefore retain the temporal information down to a few seconds regardless of the length of connection tubing needed. The limit on temporal resolution with such a system is broadening within the probe itself.

Figure 3.

Illustration of segmented flow sampling and coupling to mass spectrometry. (A) Dialysate is pumped into a cross fluidic structure mounted on the subject’s head. In the cross, fluorinated oil and internal standard solution are pumped into separate arms. Resulting segmented flow emerges from the fourth arm and is pumped to the electrospray source of the mass spectrometer (MS). Inset shows a photo where aqueous solution of green dye was sampled and segmented within the cross by perfluorodecalin. HV stands for high voltage. (B) Recording of response to in vivo microinjection of neostigmine obtained using system in A with probe inserted in the striatum of a rat. Figure shows a sample trace for simultaneous detection of acetylcholine (ACh), internal standard, d4-acetylcholine (d4-ACh), and (C) neostigmine (Neo). Each compound is recorded at a specific m/z by the mass spectrometer. Arrow indicates beginning of each microinjection. Insert is a magnified view of the 1^st microinjection for neostigmine detection showing the detection of individual droplets and rapid response time possible. (Adapted from [42]).

Microchip electrophoresis (amino acids) [38–39], enzyme assay (glutamate) [40], electrochemistry [41], and direct infusion mass spectrometry (acetylcholine) [42] (see Figure 3B) have been coupled to segmented flow systems to create high throughput assays of samples collected at a few seconds intervals. The mass spectrometry work is particularly interesting because it generates a “sensor” with the selectivity, multi-channel detection, and convenience of mass spectrometry detection. So far the segmented flow MS method has been used for acetylcholine. A challenge is to determine if conditions can be developed to allow direct MS analysis of other neurotransmitters by this approach. While most of the droplet work has been done with on-line detection, it is also feasible to collect and store droplets for later analysis [43].

Spatial Resolution

Microdialysis probes are usually 200 to 400 um diameter with a 1–4 mm sampling length. This size limits the potential to sampling from smaller brain regions. Smaller probes can be made (e.g. 0.5 mm long by 200 um diameter) and successfully used if the analytical method used is sensitive enough (Figure 4). A more radical approach is use of low-flow push-pull perfusion [44]. In this method, sampling occurs at the tip of two closely spaced capillaries. One fluid withdraws sample (“pull”) while make-up fluid is pumped from the other capillary (“push”) to maintain fluid balance in the sampling region. Sampling occurs just at the tip so spatial resolution is enhanced (Figure 4). Flow rates of 50 nL/min or less are used to minimize the potentially damaging effects of having direct contact of fluid with tissue. A limitation of these probes is poor temporal resolution. Consider that at 50 nL/min, it can take 20 min just to collect 1 µL for assay. Direct coupling to capillary electrophoresis can alleviate this concern [45]. Our group coupled the method with segmented flow to show that both high temporal and spatial resolution can be obtained [40]. In vitro results suggested sub-second sampling was possible creating the exciting prospect of combining the advantages of sampling methods (wide choice of analytical methods) with sensors (high spatial and temporal resolution).

Figure 4.

Comparison of the size of different sampling probes. Top probe is a conventional microdialysis sampling probe from a commercial vendor. White material is the sampling membrane. “Mini” is a small microdialysis probe that can be fabricated manually. The sampling area is the clear area (about 0.5 mm long by 0.2 mm diameter) at the tip. “Push-pull” is two fused silica capillaries, coated in polyimide, fused together with epoxy. Sampling occurs just at the tip. In practice, a rigid sheath if often added to support the fused silica to near the tip. “Microfab push-pull” a Si-based probe made by lithography and bulk etching techniques. The probe is 85 um wide by 70 thick along the shank and tip. Sampling occurs from a 20 um diameter orifice at the tip (not visible in the photo).

Although this approach is still in early developmental stage, initial applications have revealed encouraging results that highlight the spatial resolution [46–48]. For example, the method was coupled with LC-MS to show sharp gradients in basal concentration of several neurotransmitters even within small brain regions such as the ventral tegmental area [49]. This was the first direct detection of concentration gradients maintained across such small brain nuclei. The probes were estimated to sample from 4 nL voxels. The method was also used in the eye demonstrating potential beyond the CNS measurements. [46]

So far all push-pull probes have been assembled by hand from capillary tubes. We have recently developed an approach to microfabricating sampling probes in Si based on “buried channel technology” [50] (Figure 4). These probes have been successfully used in vivo. Microfabricated probes are much smaller than other sampling probes creating even better spatial resolution. Further, they offer the potential of mass fabrication, incorporating electrodes, and adding sample preparation steps into the probe.

Assay Methods

As discussed above, improvements in sampling were by necessity coupled to improved assay methods. Assays commonly used for microdialysis samples have been adapted to droplet methods and to the smaller probes as discussed above. Sensitive immunoassays were developed for measuring the larger proteins such as Aβ and tau. Many other clever assays have been developed for different neurotransmitters and coupled to sampling probes (see e.g. [51]). Besides these advances, a trend has been use of LC-MS for analyzing dialysate. For years HPLC with electrochemical or fluorescence detection was the method of choice for analyzing dialysate fractions; however, this method could only allow a sub-set of neurotransmitters to be analyzed in one fraction. Thus, easily oxidized neurotransmitters like dopamine and serotonin could be detected by LC-EC and amino acids were typically detected using fluorescence assays. Several groups have recognized the power of LC-MS and developed assays that allow different groups of neurotransmitters to be measured in one assay [52–57]. Perhaps the most comprehensive method is one that allows all the amino acid neurotransmitters, dopamine, serotonin, norepinephrine, adenosine, acetylcholine, carnosine, and many metabolites to measured in 8 min [58]. This approach relies on benzoyl chloride to derivatize the amine and phenol containing molecules making them easily retained on reversed phase LC and improving their ionization efficiency. Acetylcholine is not derivatized but can still be detected in the same assay. Use of MS greatly relaxes the separation requirements compared to EC or fluorescence detection since the method adds selectivity to detection. This method was shown to be useful for 1 min fractions.

For several years many groups have been developing capillary LC coupled to MS detecting neuropeptides in dialysate [59–67]. Much work over the past few years have been primarily demonstration and validation of detection of different neuropeptides. While very challenging because of the low concentration and instability of neuropeptide samples, this development is starting to bear fruit. For example, for the first time peptides were measured during a spontaneous behavior (feeding) [68]. This study revealed that enkephalins but not dynorphins increased in the dorsal striatum during feeding behavior. Microinjection of opioid agonist into the same brain region evoked feeding in sated rats showing that the opioid peptides were a signal to start eating. These results show the potential of the LC-MS methods to begin unraveling the role of neuropeptides in behavior.

Conclusions

In vivo sampling methods have maintained a stable popularity for studying neuropharmacology, pharmacokinetics and in measurement of relatively slow changes. Expansion to important larger molecules and improvements in temporal and spatial resolution are yielding renewed interest in these methods.

Highlights.

• Enhanced recovery sampling allows trace level substances to be monitored in brain

• Amyloid-β can be monitored to reveal its dynamics in plaque formation

• Temporal resolution of 2 s for neurotransmitter monitoring can be achieved

• Spatial resolution of 4 nL has been achieved with miniaturized sampling probes

• LC-MS allows comprehensive monitoring of small molecule neurotransmitters