Detection and Monitoring of Neurotransmitters - A Spectroscopic Analysis
Detection and Monitoring of Neurotransmitters - A Spectroscopic Analysis
Detection and Monitoring of Neurotransmitters - A Spectroscopic Analysis
Author Manuscript
Neuromodulation. Author manuscript; available in PMC 2014 May 01.
Published in final edited form as:
NIH-PA Author Manuscript
Abstract
Objectives—We demonstrate that confocal Raman mapping spectroscopy provides rapid,
NIH-PA Author Manuscript
detailed and accurate neurotransmitter analysis, enabling millisecond time resolution monitoring
of biochemical dynamics. As a prototypical demonstration of the power of the method, we present
real-time in vitro serotonin, adenosine, and dopamine detection, and dopamine diffusion in an
inhomogeneous organic gel, which was used as a substitute for neurologic tissue.
Materials and Methods—Dopamine, adenosine and serotonin were used to prepare
neurotransmitter solutions in DI water. The solutions were applied to the surfaces of glass slides,
where they inter-diffused. Raman mapping was achieved by detecting non-overlapping spectral
signatures characteristic of the neurotransmitters with an alpha 300 WITec confocal Raman
system, using 532 nm Nd:YAG laser excitation. Every local Raman spectrum was recorded in
milliseconds and complete Raman mapping in a few seconds.
Results—Without damage, dyeing, or preferential sample preparation, confocal Raman mapping
provided positive detection of each neurotransmitter, allowing association of the high-resolution
spectra with specific micro-scale image regions. Such information is particularly important for
complex, heterogeneous samples, where changes in composition can influence neurotransmission
processes. We also report an estimated dopamine diffusion coefficient two orders of magnitude
smaller than that calculated by the flow-injection method.
Conclusions—Accurate nondestructive characterization for real-time detection of
NIH-PA Author Manuscript
Keywords
Basic science; Raman spectroscopy; neurotransmitters; brain
*
Corresponding author: Felicia S. Manciu, Department of Physics, University of Texas at El Paso, 500 West University Ave., El Paso,
TX 79968, USA, Phone: (915) 747-8472; Fax: (915) 747-5447; fsmanciu@utep.edu.
Conflict of interest statement: The authors report no conflict of interest.
Manciu et al. Page 2
I. INTRODUCTION
The purpose of this work is to demonstrate the prospects of confocal Raman spectroscopy
NIH-PA Author Manuscript
Currently, the therapeutic success of deep brain stimulation for tremor associated with
Parkinson’s disease (PD) and essential tremor (ET) has led to the early application of deep
brain stimulation (DBS) for an increasing spectrum of conditions, ranging from movement
disorders to neuropsychiatric conditions (11). Preclinical studies using fast scan cyclic
voltammetry and carbon-fiber microelectrodes (CFM) have shown neurotransmitter release
in various efferent targets during DBS (12-14). For example, it has been demonstrated that
subthalamic nucleus (STN) DBS evokes dopamine release in the caudate in the intact rat and
pig and, most significantly, in the parkinsonian rat 6-hydroxydopamine (6-OHDA) model
(12-14). Another neurochemical mechanism that may be of particular importance to DBS is
NIH-PA Author Manuscript
adenosine release. Proposed as a chemical mediator of thalamic DBS for the treatment of
essential tremor (15), caudate adenosine release can be measured during electrical
stimulation of the nigrostriatal dopaminergic tract (16). It has also been demonstrated that
STN DBS elicits this release (14).
Whereas the use of CFM chemical microsensors offers the advantage of smaller real-time
measurement probes than those previously used in microdialysis (e.g., 5 to 10 μm versus
200 to 400 μm diameter for microdialysis probes), their chemical stability is affected by the
biological environment; they, therefore have the disadvantage of being long-term
degradable. There remains a high demand for technological development, since a complete
understanding of the DBS mechanism remains far from being achieved, in large part
because of the technical difficulties in combining measurement modalities for global
assessment of neural activity and chemical-specific sensing. As demonstrated in this in vitro
research work, Raman spectroscopy, a non-destructive method of detection, provides the
opportunity to combine precise, real time measurement of multiple neurotransmitters to
imaging data.
If used in future clinical studies for better elucidation of the mechanism of action of DBS,
NIH-PA Author Manuscript
Raman can also have the advantage of increasing the stability of the sensing mechanism by
employing non-degradable optical fiber as sensing probes; fibers that can be produced with
selected micro-scale dimensions. Furthermore, since light can be collected in real-time via
optical fibers, from living tissue, the benefit of high sensitivity to functional changes and,
consequently, of revealing the dynamics of cells in the nervous system, either via the
absorption of light or emission of light via elastic or inelastic scattering, can be achieved
(5-10). Whereas fluorescence/luminescence optical detection approaches have the advantage
of strong signals, they have limitations as to the chemical moieties that can be detected
simultaneously in a clinical environment. On the other hand, Raman spectroscopy can not
only provide the most detailed and accurate analysis of the chemical composition of the
sample under study, with no evidence of disruption of catecholamine detection due to the
presence of amine-containing metabolites and proteins (5-10), but also, as will be
demonstrated in this work, enables monitoring the chemical dynamics at millisecond time
resolution, which is another important requirement for in vivo applications.
Thus, in the past few years, various Raman spectroscopic techniques such as surface-
enhanced Raman (SERS) (5-7), surface enhanced spatially offset Raman spectroscopy
(SESORS) (8), localized surface plasmon resonance (LSPR) (9), and ultraviolet resonance
NIH-PA Author Manuscript
Raman (UVRRS) (10) have been proposed and developed for identification of biological
molecules. These Raman techniques have been used for a wide-range of in vitro and in vivo
diagnostic applications, such as non-invasively monitoring blood analytes and in working
with coronary artery and Alzheimer’s diseases, breast cancer and brain tumors, and for
minimally invasive but real-time diagnostics of superficial tissue (skin tissue) and deep
tissue (mammalian tissue) (5-10). Further progress in Raman applications for rapid analyte
detection is based on development of various correlation and classification algorithms such
as the Savitzky–Golay second-derivative (SGSD) method (17), multivariate calibration
(MVC) models (18), and multiobjective evolutionary algorithm (MOEA) (19).
Raman spectroscopy also has the advantage of label-free recognition of biomolecules (such
as neourotransmitters, in this case), recording a unique vibrational spectrum for each
different molecular species (every species has its own unique molecular bond
configuration). In particular, this advantage is important in dynamic processes such as
diffusion, since diffusion parameters can be measured by direct observation of the species of
interest, avoiding the possible inaccuracy introduced by the use of chemical labels with
molecules larger or smaller than that of the neurotransmitter analyzed, with possible
resultant changes in biological activity. As a prototypical confirmation of the power of the
NIH-PA Author Manuscript
method, in this article we also present an in vitro analysis of the real-time diffusion of
dopamine in an inhomogeneous organic gel medium.
The Raman measurements in this study were acquired at ambient conditions with an alpha
300 WITec confocal Raman system (WITec Inc., Ulm, Germany), using the 532 nm
excitation of a Nd:YAG laser and a 20X, NA=0.4, objective lens. Briefly, the experimental
set-up, which is described in detail elsewhere in the literature (20), consists of a laser source
that is coupled into a confocal microscope via a single mode optical fiber of 50 μm
diameter; the fiber also has the effective role of a pinhole source for confocal microscopy.
The reflected laser line and (elastically) Rayleigh-scattered light are eliminated by an edge
filter, which allows only the (inelastically) Raman-scattered light to be focused and collected
with a multimode optical fiber that is coupled to the spectrometer as an entrance slit. During
measurements, the power output of the Nd:YAG (532 nm) laser was kept low at ~10 mW.
For data acquisition, the WiTec Control software was employed; it also controls the
piezoelectric stage for sample scanning.
Locally (i. e., at every image pixel), the Raman spectrum was recorded in milliseconds and
the overall Raman mapping record of the inter-diffusion of the three neurotransmitters in a
few minutes. Each consecutive 41.5 × 46.7 μm2 image in this Raman mapping of dopamine
NIH-PA Author Manuscript
diffusion was acquired with an integration time of 6.4 s per image and 4 ms per local
spectrum. To obtain the Raman mapping images, the Raman signal was detected by a 1024
× 127 pixel peltier cooled CCD camera with a spectral resolution of 4 wavenumbers; at each
pixel a complete Raman spectrum was recorded.
Some vibrations are common in these neurotransmitters, as revealed by Fig. 1 (e), where the
standard Raman spectra of dopamine, adenosine, and serotonin are plotted using the same
NIH-PA Author Manuscript
pseudo-colors as mentioned above. However, in this case, the Raman mapping was achieved
by detecting and integrating non-overlapping characteristic spectral signatures of the
compounds, as follows: for dopamine the vibration attributed to C-O stretching at 1289
cm-1, for serotonin, the indole ring stretching vibration at 1540 cm-1, and for adenosine
either of the adenine ring vibrations at 320 cm-1 or 1336 cm-1 (10,21,22). One reason behind
considering these frequencies, although other non-overlapping vibrations exist in these
spectra at higher energies (the 2800 – 3500 cm-1 spectral region), is the potential
interference, in the latter energy range, of common vibrations arising from molecular
structures in living cells and other organic constituents of normal tissue, such as the strong
CH-stretching band, the valence vibrations of CH2 and CH3 moieties in proteins and lipids,
and the OH bands, mainly assigned to water. Our choice of frequencies thus establishes a
sound basis for the current study, and, at the same time, anticipates its applicability to future
in vivo investigations. The choice is also based on consideration of the more intense
characteristic non-overlapping vibrations of the compounds of interest, thus increasing
detection sensitivity and accuracy.
As reported in the literature, environmental factors (e.g., pH, solvents, noble metal surface
characteristics, etc.), affect neurotransmitters Raman vibrational lines, mainly through
NIH-PA Author Manuscript
Comparison of the frequency positions of these bands in Fig. 2 (a) and Fig. 2 (b) shows only
very slight shifts, validating the suitability of our frequency selection for Raman mapping.
More importantly, assessment of their relative intensities demonstrates that the Raman
technique could also be used as a quantitative method for estimating the relative
concentrations of neurotransmitters. For example, the adenosine 320 and 1336 cm-1
vibrations, which are highlighted with blue color in Fig. 2(a), have higher intensities in the
Raman integrated spectrum of image (d), corroborating the strong blue color observed in
NIH-PA Author Manuscript
this image (see Fig. 1 (d)). Furthermore, besides the dopamine vibration at 1289 cm-1, which
was considered for Raman mapping, there is another unmarked dopamine Raman line
around 400 cm-1 which has a higher intensity in the integrated spectrum of image (c) than in
that of image (a). This observation is again in good agreement with the intensity of the red
color seen in the Raman mapping presented in Fig 1 (c) as compared to that in Fig. 1 (a).
However, an accurate quantitative analysis by taking into account the ratio of Raman peak
cross-sections is obstructed by the above mentioned observed shifting and by the influence
of other closely located vibrational modes, an influence that results in a broadening of these
bands due to the possible convolution of various vibrations.
There are three important and also interrelated issues to be considered for future in vivo
investigations: a fast acquisition time, an acceptable S/N ratio, and the penetration depth of
NIH-PA Author Manuscript
the laser wavelength employed. Concerning the S/N ratio and the penetration depth, a 785
nm near infrared (NIR) laser excitation of a diode laser has been used most commonly for in
vivo experiments due to its deeper tissue penetration (of the order of millimeters) and its
good S/N ratio, in spite of fluorescence signals expected from the cellular molecules of
living tissue. An even longer-wavelength excitation, such as the 1064 nm line of the
frequency doubled Nd:YAG laser, will further reduce the tissue fluorescence by
approximately two orders of magnitude, therefore improving the S/N ratio. However, a
drawback in using this excitation is the increased signal absorption by water at longer
wavelengths.
On the other hand, because the intensity of the Raman signal is proportional to the fourth
power of the frequency of the excitation, the signal obtained from the green light excitation
(which was employed in our experiments) is much stronger than in the other cases, allowing
us to decrease the acquisition time by more than an order of magnitude (for a similar S/N
ratio). This short acquisition time was instrumental to obtaining information in real time
about the diffusion of dopamine in an organic medium and without fluorescence
interference, although just for a penetration depth of a few tens of micrometers, as is usual
for the 532 nm excitation.
NIH-PA Author Manuscript
The Knox gelatin was used for the following reasons: it has higher heterogeneity than the
standard agarose gel and has mechanical properties closer to those of brain tissue (23); thus,
in a very simplified way, the diffusion of dopamine in a gelatin medium may provide a more
appropriate model of the natural process occurring in brain tissue.
The sequential confocal Raman mapping images of Fig. 3 reveal the non-uniform “random
walk” character typical of diffusion on the microscopic scale, in this case for dopamine. The
estimation of the diffusion coefficient of locally deposited neurotransmitter was performed
by fitting two consecutive cross-sectional average concentration-time profiles, which were
obtained from two consecutive images. The results are presented in Fig. 4, where the thick
solid line is the Gaussian fitting curve. For fitting we applied a simplex algorithm to a
solution of the diffusion equation:
where C(x) is the concentration distribution, D is the diffusion coefficient, t is the time
elapsed from the beginning of the diffusion, and A is related to the total amount of
dopamine. The value of 1.28×10-8 cm2/s for the coefficient of diffusion D obtained in this
way is two orders of magnitude smaller than the one calculated by the flow-injection method
(24). Whereas the flow-injection method addresses the diffusion of dopamine in liquid, here
we study a more involved problem, in which gelatin generates a porous gel-like structure in
water (somewhat similar to the brain). Therefore, the diffusion depends strongly on the
dimensions of the channels available to dopamine as well as on the gelatin structure, which
might create traps (temporary or permanent) for dopamine. A temporary trap, where the
dopamine is initially stored and then slowly released back into the gel network, as well as
physical movement constraints due to the network structure, strongly reduce the long range
diffusion, as suggested by the results of the fitting from Fig. 4.
IV. CONCLUSIONS
Since real-time and accurate detection of neurotransmitters in inhomogeneous environments
by nondestructive characterization and without the requirement of sample labeling are key
NIH-PA Author Manuscript
issues in bioscience, the work reported here demonstrates the capabilities of Raman
spectroscopy in future neuroscience applications.
Although extensive research has been done in DBS, and preclinical studies (12-16) have
demonstrated that the function of neurotransmitters such as dopamine, adenosine, and
serotonin is affected by Parkinson’s, etc. diseases, the mechanism of DBS is far from being
completely understood. As a contribution to this active field, it is possible that optical
techniques such as those presented in this study, with their capacity for obtaining real time
maps of molecular species concentrations, can create a basis for future visualization and
measurement of neurotransmitter release in living systems and thereby provide significant
insights into the action of DBS.
In the work reported here, in vitro inter-diffusion of dopamine, adenosine and serotonin have
been directly visualized and analyzed using confocal Raman mapping, where integration of
characteristic non-overlapping signatures of these neurotransmitters such as the 320, 1289,
1336, and 1540 cm-1 vibrations were considered for detection. Not only is chemical
differentiation of these compounds observed at almost stationary time frames (4 ms
integration time per spectrum), but the acquired results could form a strong foundation for
NIH-PA Author Manuscript
We also report here a value of 1.28×10-8 cm2/s for the estimated diffusion coefficient of
locally deposited dopamine, a value that is two orders of magnitude smaller than the one
calculated by the flow-injection method. A lower value for this coefficient is expected in a
solid non-uniform porous medium such as gelatin, since the process will depend on the
various dimensions of the available channels for dopamine diffusion. Furthermore, the
diffusion process can be quite complex in this gel network structure, which creates
temporary or permanents traps for dopamine to be stored and then slowly released back into
the medium; phenomena that will strongly reduce the long range diffusion. From this
perspective, although in a very simplified way, the diffusion of neurotransmitters in gelatin
mimics quite well the natural process occurring in brain tissue.
NIH-PA Author Manuscript
In conclusion, Raman spectroscopy can provide critical insights into the problems of
accurate detection and fast monitoring of neurotransmitters’ diffusion and distribution in
inhomogeneous environments. This technique has tremendous clinical potential, and there is
substantial room for future advances.
Acknowledgments
Source(s) of financial support: This work has been supported by NIH K08 NS 52232 award and by a research
agreement between Mayo Clinic and the University of Texas at El Paso and The Grainger Foundation to KEB and
KHL.
Dr. Felicia S. Manciu provided the expertise in Raman Spectroscopy, designed the studies and the experimental
approach. Prof. Kevin E. Bennet and Dr. Kendall Lee provided the relevance, additions to the studies and obtained
the funding. Dr. Manciu developed the original manuscript with assistance from Prof. Bennet. Dr. Manciu and Dr.
William Durrer conducted the experiments, analyzed the data, and made the figures. All authors reviewed the data
and provided editorial improvements and approved the final manuscript.
References
NIH-PA Author Manuscript
13. Lee KH, Blaha CD, Garris PA, Mohseni P, Horne AE, Bennet KE, Agnesi F, Bledsoe JM, Lester
DB, Kimble C, Min H-K, Kim Y-B, Cho Z-H. Evolution of deep brain stimulation: human
electrometer and smart devices supporting the next generation of therapy. Neuromod. 2009;
NIH-PA Author Manuscript
12:85–103.
14. Shon YM, Chang SY, Tye SJ, Kimble CJ, Bennet KE, Blaha CD, Lee KH. Comonitoring of
adenosine and dopamine using the Wireless Instantaneous Neurotransmitter Concentration
System: proof of principle. J Neurosurg. 2010a; 112:539–548. [PubMed: 19731995]
15. Bekar L, Libionka W, Tian GF, Xu Q, Torres A, Wang X, Lovatt D, Williams E, Takano T,
Schnermann J, Bakos R, Nedergaard M. Adenosine is crucial for deep brain stimulation-mediated
attenuation of tremor. Nat Med. 2008; 14:75–80. [PubMed: 18157140]
16. Cechova S, Venton BJ. Transient adenosine efflux in the rat caudate-putamen. J Neurochem. 2008;
105:1253–1263. [PubMed: 18194431]
17. Zhang D, Ben-Amotz D. Enhanced Chemical Classification of Raman Images in the Presence of
Strong Fluorescence Interference. Appl Spectrosc. 2000; 54:1379–1383.
18. Dingari NC, Barman I, Singh G, Kang JW, Dasari R, Feld M. Investigation of the specificity of
Raman spectroscopy in non-invasive blood glucose measurements. Anal Bioanal Chem. 2011;
400:2871–2880. [PubMed: 21509482]
19. Jarvis RM, Rowe W, Yaffe N, O’Connor R, Knowles J, Blanch EW, Goodacre R. Multiobjective
evolutionary optimisation for surface-enhanced Raman scattering. Anal Bioanal Chem. 2010;
397:1893–1901. [PubMed: 20440481]
20. Hollricher O. Confocal Raman microscopy teams high-resolution capabilities with powerful
NIH-PA Author Manuscript
Figure 1.
(a) - (d) confocal Raman mapping images of dopamine (pseudo color: red), serotonin
NIH-PA Author Manuscript
(pseudo color: green), and adenosine (pseudo color: blue) recorded in different spots, and (e)
the standard Raman spectra of these neurotrasmitters, as labeled. The spectra are vertically
translated for clarity and recorded for the 150 to 3500 cm-1 spectral region.
Figure 2.
(a) integrated spectra of the previously acquired Raman mapping images and (b) standard
Raman spectra of the neurotrasmitters in the 150 to 1800 cm-1 spectral region of interest; the
latter spectra are presented for comparison purposes.
NIH-PA Author Manuscript
Figure 3.
Sequential confocal Raman mapping images of dopamine diffusion in gelatin.
NIH-PA Author Manuscript
Figure 4.
Dopamine diffusion curves. The solid line through each graph is the fitted theoretical curve.
NIH-PA Author Manuscript