Mottin et al. Appl. Spectrosc. 1993
590
FIBER-OPTIC TIME-RESOLVED FLUORESCENCE SENSOR FOR IN VITRO
SEROTONIN DETERMINATION.
STEPHANE MOTTIN*, CANH TRAN-MINH, PIERRE LAPORTE, RAYMOND CESPUGLIO,
MICHEL JOUVET
Laboratoire de Biotechnologie, Ecole Nationale Superieure des Mines de St-Etienne, F-42023 St-Etienne Cedex 2,
FRANCE (S.M., C.T.); Laboratoire T.S.I., URA CNRS 842, Univ. de St-Etienne, F-42023 St-Etienne Cedex 2, FRANCE
(P.L.); Département de Médecine Expérimentale, INSERM U 52, Univ. Cl. Bernard, F-69373 Lyon Cedex 08, FRANCE
(R.C., M.J.).
[email protected]
Mottin, S. identifier: http://orcid.org/0000-0002-7088-4353
Datacite DOI: https://doi.org/10.5281/zenodo.438934
Reference BibTeX
@Article{1993_mottin_4,
author = {Mottin, Stéphane and Tran-Minh, Canh and Laporte, Pierre and Cespuglio, Raymond and Jouvet, Michel},
title = {Fiber optic time-resolved fluorescence sensor for in vitro serotonin determination},
journal = {Appl. Spectro.},
year = {1993},
volume = {47},
number = {5},
pages = {590-597},
doi = {10.1366/0003702934067180},
keywords = {Time-resolved spectroscopy;fluorescence;serotonin;Sensor;in vitro;fluorescence lifetime;UV;optical fibre},
}
In a phosphate buffer pH 7 and with an excitation at 337 nm, a previously unreported
fluorescence of 5-HT (5-hydroxytryptamine) is presented. The 70 nm wide fluorescence
emission has a maximum at around 390 nm. Two emission states were established for 5-HT:
the first, well known, emits at around 340 nm with a decay-time near to 2.7 ns and the other
at around 390 nm with a lifeime near to 1 ns. After this preliminary study, a fiber-optic
chemical sensor (FOCS) is developed in order to measure 5-HT with a single-fiber
configuration and nitrogen laser excitation. Temporal and spectral effects including
fluorescence of fiber-optic are studied and compared for measurements both directly in
cuvettes and through a fiber-optic. Our FOCS is fast with less than thirty seconds required
for each measurement and reaches a detection limit of 5-HT in the range of 5 µM. Our
system would be a possible "tool" for in vivo determination of 5-HT.
Index Headings: Fluorescence; Time-resolved spectroscopy; Serotonin; Optical fluorometric
sensors.
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
591
Introduction
In the past, different approaches (anatomical, neurophysiological and pharmacological) have
helped to establish a causal relation between the brain's serotonin (5-hydroxytryptamine or 5-HT)
level and the alternation of the different states of vigilance. Later new techniques have offered the
possibility of studying the activity of serotoninergic neurons throughout the sleep-waking cyclei.
Methods measuring 5-HT or 5-HIAA (5-hydroxyindoleacetic acid) levels in vivo , which would
allow the investigation of its release and metabolism, are divided into two categories, namely
voltammetry and dialysis (push-pull perfusion)ii,iii,iv. In vivo voltammetry presents some
limitations regarding absolute identification of the monitored compoundsv. In this regard, the
dialysis technique has an advantage. However, the dialysis tubing has a fairly short 'functional '
life in vivo and presents others drawbacks such as the limitation attached to the relative thickness
of the probe vi.
On account of the above limitations and therefore having in mind choices always compatible with
in vivo applications, we have searched an alternative compatible with the development of a
miniature sensor characterized by the rapidity, specificity and sensitivity of the measurements. In
this paper, we show the feasibility of in vitro rapid determination of 5-HT (less than 30 seconds)
by a fiber-optic chemical sensor (FOCS). The single-fiber configuration is preferred rather than
the double-fiber or multiple-fiber configuration despite a higher blank noisevii. In addition, the
injection and collection of light complicate the set-up. For the possible extension to an in vivo
sensor, the interactions between sensor and media must be the simpliest in order to miniaturize,
sterilize and minimize problems of incompatibility of materials such as the resins used for
attaching the ends of the two fibers. Sensitivity in term of detection limit depends on instrumental
choices such as the magnitude of the parasitic fluorescence and the efficiency of the single-fiber
configuration optical coupler. Specificity is investigated and the signal given by 5-HIAA is also
characterized. Relations between results and our instrumental selections are discussed.
One of the main choices is the use of time-resolved fluorescence techniques. Such methods have
been widely used in fields as diverse as molecular biology, polymers science and solid-states
physicsviii. In most cases, only qualitative measurements are researched, nevertheless quantitative
methods appears specially if the steady-state fluorescence intensity is not sufficientix,x,xi. Two
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
592
approaches for quantitative applications have emerged. Firstly, introduction of fiber-optic and
construction of an extrinsic time-resolved fluorescence optode for remote sensingxii and a semiintrinsic optode for detection of chemical compounds which have no fluorescencexiii,xiv. Secondly,
the other approach was in the field of time-resolved fluorescence instrumentation so as to obtain
a rapid and simplified techniquexv,xvi.
Photophysical properties of 5-HT are mainly due to the chromophore 5-hydroxyindole. In order
to measure selectively 5-HT by a remote sensor using optical fibers, we looked at the existence of
specific photophysical properties for this group with regard to indole compounds (e.g. tryptophan)
and inside the group of 5-hydroxyindolic compounds specially between 5-HT and 5-HIAA. UV
absorbance between simple indole and indole which are substituted in the ring at the five position
in polar solvent like water show some differencesxvii : in contrast with the 1La band ( notation of
Plattxviii ), many differences occur for the 1Lb band such as a red shift and a decrease of molar
extinction.
Native fluorescence (called UV fluorescence) of 5-HT in neutral aqueous solutions has been
reported since 1955 : the excitation and emission of the 5-HT peak at respectively 295 and 340
nmxix. Another emission with a maximum near 550 nm (called green fluorescence) was found in
strongly acidic solutions without any change in the absorption spectrumxx. This unusual
fluorescence with a such large Stokes'shift is very specific to 5-hydroxy or 5-alkoxy indolxxi and
was widely used for the fluorimetric assay of 5-HT.
Finally, at physiological pH levels, native UV fluorescence of 5-HT does not seem to be enough
for our purpose because of the lack of specificity. Native green fluorescence of 5-hydroxyindole
compounds becomes important at a pH near to 2, which is, according to our goal, difficult to
exploit .
On account of these data, we have first studied photophysical properties of 5-HT when excited at
337 nm despite the fact that this wavelength lies at the extreme limit of the absorption spectrum.
The idea was to choose a wavelength avoiding absorption (and fluorescence) of indol compounds
such as tryptophan and still allowing an absorption of 5-hydroxyindolic compounds. The 337 nm
excitation leads to a broad band blue fluorescence emission ( maximum: 390 nm, width: 70 nm)
which to our best knowledge, has not been previously reported. An extrinsic FOCS with
subnanosecond resolution and rapid determination of lifetime (less than 30 seconds) has been
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
593
developed with this near UV excitation. Similar techniques have been previously described in the
literature but with visible excitationxxii. Additional technical difficulties occur when using UV
excitation such as the great attenuation in the fiber and fluorescence of fiber and dust. Some
interferences generated by fiber's fluorescence are time-filtered by using time-resolved
fluorescence technique. A single-fiber configuration is described with a view for use at 337.1 nm
as the excitation wavelength. This present method can be applied for in-line and in vivo analyses
when compounds measured have some photophysical properties such as absorbance in the near
UV and fluorescence with a Stoke shift of more than 40 nm.
Theory
If [C*] represents the concentration of excited molecules,
we have the following formula:
∗]
𝑛
(1)
𝑑[𝐶
= − ∑ 𝑘𝑖 [𝐶 ∗ ] + 𝛾 𝐼𝑎𝑏𝑠
𝑑𝑡
𝑖=1
where ki (i=1,n) are the rate constants for each mode of deactivation of the excited state and Iabs
the flux of photons absorbed.  ( (mole/l)/ photon ) is the light efficiency for the molecular
excitation.
In case of instantaneous excitation we have :
(2)
So we can write : [C*](t) = [C*]0 exp (-t / m ) with m the measured fluorescence lifetime:
(3)
The intensity of the fluorescence observed at a time t after excitation is expressed by :
(4)
F(t) = kF [C*]=[C*]0 exp (-t / m ) / 0 .
where kF = 1 / 0 ; kF is the rate constant for the emission of photons as a mode of deactivation
of the excited state . And 0 is called the intrinsic fluorescence lifetime.
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
594
With the Beer-Lambert law, we can write : [C*]0 = I0-10-L[C] where I0 is the flux of excitation
photons,  is the molar absorption coefficient, L the optical path and [C] the molar concentration.
If the absorbance is low ( L[C]<0.05 ) we can write :
F(t) ≈ 2.3 L[C] I0 exp (-t / m ) / 0 .
(5)
From this equation, the integrated fluorescence intensity ( H ) is
(6)
∞
𝐻([𝐶]) = ∫ 𝐹(𝑡) 𝑑𝑡 ≈
0
2.3 𝜀 𝐿 𝐼0 𝜏𝑚
[𝐶]
𝜏0
H is a linear function of the concentration for small absorbances.
With our laser system, the fluorescence lifetimes are evaluated by the method based on the
numerical convolution of the instrument response and a least-square optimized comparison with
the experimental signal XXII. The 512 points of the instrument response is convoluted with a single
exponential decay. The result is compared to the experimental fluorescence response between 1
to 10ns by means of a unweighted Chi-squared test. The Chi-square test is a measure of the
goodness of fit but it has absolute meaning only if the weighting factors is specified. In our
measurements with our laser fluorometer based upon sampling technique, the noise is neither
normally distributed nor similar to the poisson distribution, i.e. the formula of the weighting
factors is neither a constant nor the same used in photon-counting techniquesxxiii,xxiv. In addition
relative residues ri (in percent), figure on our results:
ri =
y i - y( xi )
yi
(7)
where yi are the experimental data and y(xi) the fitting function.
Temporal multimodal broadening is negligible in our application because of the small length of
optical fiberxxv. A small time shift of the fluorescence response is necessary because of the
difference between the transit times for the scattered light of the laser which is used to build the
instrumental response function and fluorescences collected at different wavelengths. This
difference comes from the refractive index change of fiber core with wavelength.
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
595
Materials and methods
Reagents. 5-HT hydrochlorid (97%) and 5-HIAA free acid (98-100%) are obtained from Sigma
Chemical Co. 5-HT hydrochlorid (99%) from Aldrich is also used for static fluorescence and
single photon counting experiments with no great change. Solutions were prepared by using
commercial phosphate buffer (Merck) for solutions at pH 7. Solutions were freshly prepared each
day.
Absorption Measurements.
The apparatus used is a Kontron model 860 connected to a microcomputer by RS232 to transfer
data.
Steady-state Fluorescence Measurements.
A LS-50 luminescence spectrometer from Perkin Elmer was used with 1 second as time
integration and respectively 1 and 5 nm bandwidth for excitation and emission monochromators.
Time-resolved Fluorescence Measurements.
We perform time-resolved fluorescence of 5-HT at different excitation wavelengths with a model
299T time domain fluorometer from Edinburgh Instrument and with its nitrogen flashlamp. The
excitation and emission bandwiths were set at respectively 10 nm and 20 nm.
A diagram of the experimental setup used with laser excitation is shown in Figure 1. The nitrogen
laser (Model LN120C, PRA Laser Inc.,Laser Photonics Company) delivers subnanosecond pulses
of 300 ps (FWHM) and 70 µJ. The 337.1 nm radiation is first filtered by a black filter type UG 1
(Schott,France) to eliminate visible light generated by some nitrogen plasma discharge between
the electrodes of the cavity. A small mirror is placed on the border of the light beam and focused
on a fast photodiode EGG model UV 1B which is used as trigger for the digitizer. The laser beam
is reflected by a mirror model 614 (Schott) named M2 and focused on the optical fiber by a small
lens with a focal length such that the input aperture is less than the numerical aperture of the
optical fiber used. The fiber used is a step index multimode model PCS 1000 from Quartz et Silice
(plastic-clad silica fiber-optic, core+cladding diameter of 1260µm with a 1000µm core, numerical
aperture equal to 0.4). The single-fiber configuration of the FOCS uses the same fiber to transmit
the excitation radiation to the sample and to guide the collected fluorescence to the detection
system.The fluorescence signal goes through the mirror M2 (model 614 from Schott). This mirror
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
596
has a reflectance more than 90% for wavelength lower than 370 nm and a transmittance more than
90% for wavelength upper than 415 nm .The fluorescence signal is focused on the entrance slit of
the monochromator ( Jobin Yvon) with the identical lens used for injection of the laser into the
fiber. The fixed slits provide a 4 nm resolution. The photomultiplier (PMT) is a R3810 from
Hamamatsu selected in gain to obtain such a response that single events are easily detected by a
50 Ω input fast electronics. The voltage divider is a E850-13 MOD from Hamamatsu. The overall
response time of the PMT and divider is 0.8 ns (FWHM). The main limitation of this photodetector
is the saturation which occurs at a typical flux of a few tens photons within the time window of
1ns. The 50 Ω output of the PMT is sent to a transient digitizer (Tektronix Model 7912 AD
mainframe, Model 7A19 amplifier unit, Model 7B90P time base unit), which is triggered by the
above mentioned fast photodiode. Each individual point of the sample fluorescence decay is
transferred by a GPIB bus to a microcomputer using a program written by ourselves. For the time
window used (10 ns), temporal memory resolution is around 20 picoseconds i.e. much shorter than
the digitizer time bandwidth.
For measurements performed directly with cuvettes, removing the fiber mirror M2 and lenses L1,
the laser beam is focused after mirror M1 with a cylindric lens. The cuvette is placed in front of
the slit of the emission monochromator.
The use of a low repetition frequency laser source (less than 100 Hz for commercial nitrogen
lasers) prevents from a realistic use of a single-photon counting technique. We have chosen a
transient digitizer since it is very flexible due to its ability to account as well for single or multiple
events thus allowing a fast acquisition. In addition it is less expensive than a single-photon
counting systemxxvi. In this paper, our aim is not to obtain the best accuracy on lifetimes but to
characterize relative concentration within less than thirty seconds. In practice it was sufficient to
collect and average signals for 20 seconds at 15 Hz for either instrumental or fluorescence
responses.
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
597
Results and discussion
Absorption of 5-HT.
In a buffer at pH 7, absorption spectra of 5-HT and 5-HIAA present no measurable differences.
Figure 2 (curve a, top) presents the absorption spectrum between 240 and 340 nm of 5-HT in the
Merck buffer pH7 used. Low molar extinction can however be measured since 5-HT is very
soluble in water. Special attention is given to molar extinction of 5-HT at 337 nm: a typical value
is of the order of 2.5 ± 0.5 M-1. cm-1 , i.e one thousand times less than the value of the maximum
of the 1Lb band. Therefore the nitrogen laser wavelength lies in the extreme limit of the
absorbance of 5-HT. This fact is probably the reason why 5-hydroxyindole compounds were never
studied at this wavelength .
Fluorescence of 5-HT in cuvette.
In Figure 2 (curve b, bottom), the emission spectra of the fluorescence of 5-HT in Merck buffer
pH7 (concentration is set at 66µM) is reported with different excitation wavelengths. When the
range of excitation wavelength is less than 320 nm, emission spectra of 5-HT correspond to the
well known UV spectrumXIX. Since 320nm excitation wavelength, a new band appears with a
maximum at approximately 390 nm and a bandwidth around 70 nm. Tryptamine (10mM) was
chosen to investigate if a comparable phenomenon could be measured for indole compounds not
hydroxylated on the five position of the ring. In such a case no fluorescence in visible domain was
detected for the same excitation. On the other hand and in the same conditions, 5-HIAA exhibits
the same spectral properties for its emission fluorescence (maxima and bandwidth) as compared
to 5-HT. However, emission intensity of 5-HT is greater than the one of 5-HIAA by a factor
typically of about ten.
We have investigated the time-resolved fluorescence of 5-HT at 415nm emission wavelength and at
different excitation wavelengths from 285 to 335 nm with the flash lamp fluorometer. The emission
wavelength was chosen at 415 nm in order to totally avoid raman band and to get best signal for the
390nm fluorescence. The 5-HT concentration was set at 500µM. No concentration quenching was
found for this concentration. With the Edinburgh Instrument's software, biexponential fits gave best
results. Table I shows the biexponential analysis for each emission fluorescence pulses where the
preexponential weighting factors (in percent) are i 's, the lifetimes are ti (in ns), the estimated error
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
598
on ti of the Edinburgh Instrument program are ∆t, the reduced Chi-square is 2 , the Durbin Watson
parameter is DW and the fractional contributions to the total emision for each component are fi xxvii:
(8)
.
𝛼𝑖 𝑡𝑖
𝑓𝑖 =
∑𝑖 𝛼𝑖 𝑡𝑖
The first component has a decay time near of 1 ns and the second near of 2.7 ns. With the 290 nm
excitation wavelength and 340 nm emission wavelength, Chen,R.F
xxviii
previously reported a
decay time of 5-HT egal to 2.7 ns which correspond to our second component. The reduced Chisquare 2 and the Durbin Watson parameters are correct except analysis at the excitation
wavelength 310 nm which exhibits a high 2 and a bad DW for the bi-exponential model. When
excitation wavelength increases, the fractional contribution of the first component increases also.
At the 315 nm excitation wavelength and 415 nm emission wavelength, the fractional
contributions are quite egal to. The 315 nm excitation wavelength could be named "toppling
point". Less than 295 nm the first component is low and least-square fitting does not reach with
accuracy the lifetime of 1ns. Since 320 nm, the second component becomes so weak that large
errors occur on its lifetime calculated by least-square fitting.
The static fluorescence emission spectra at different excitation wavelengths, the biexponential
model found to be the best and the evolution of the fractional contribution of each component
versus the excitation wavelength establish two emission states for 5-HT: the first emits at around
340nm with a lifetime near to 2.7 ns and the other at around 390 nm with a lifetime near to 1 ns.
In Figure 3 (curve a, top), with the nitrogen laser system, the signal dependence versus
concentration of 5-HT is clearly shown. Due to the saturation of the photomultiplier, the distortion
of the fluorescence response function appears on the curve associated with 300 µM 5-HT
concentration. Alternately at low intensities the detection is limited by parasitic fluorescence.
Therefore the analysis of lifetimes has to take into account these two opposing limitations. In
Figure 3 (curves b and c, bottom left and right), experimental results and convolution calculation
for 5-HT and 5-HIAA are presented. The concentrations of 5-HT and 5-HIAA are respectively 30
µM and 600 µM. These concentrations have been found to be optimal in our conditions in view
of the above mentioned opposing limitations. A noticeable difference between lifetimes of 5-HT
and 5-HIAA is measured. Relative lifetimes have been found to be identical at two different
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
599
wavelengths inside this blue band, namely at 430 nm and 450 nm, giving 1.2 ± 0.2 ns for 5-HT
and 2.1 ± 0.2 ns for 5-HIAA. As for the emission intensity of 5-HT, it is found to be greater than
that of 5-HIAA by a factor of twenty. The higher range concentration of 5-HIAA than 5-HT does
not play an inner filtering because of the low absorption extinction coefficient of these two
compounds at 337nm. For 5-HIAA, further experiments are necessary to measure a possible
concentration quenching.
Fiber-optic Time-resolved Fluorescence Sensor .
Spectral and Temporal effects of the Optical Fiber.
Two different quartz optical fibers with high numerical apertures (0.4) have been tested : "TECS
fiber" and "Quartz et Silice" fiber. With 337.1 nm as the excitation wavelength, the "TECS fiber"
exhibits an high purple fluorescence with a lifetime of 3.6 ns generated by the cladding, whilst the
other one (purple fluorescence) appears significant only at the injection of the laser into the fiber
and at the other end of the fiber. The "Quartz et Silice" fiber has its fluorescence maxima below
400 nm with a decay time of 4.5 ± 0.3 ns. In Figure 4, the time resolved fluorescence of "Quartz
et Silice" fiber model PCS 1000 is reported to illustrate the signal generated at both ends of the
fiber. A length of 6.2 meters for the fiber was chosen in order to separate the fluorescence
generated at the laser injection from that collected at the tip of the fiber. Moreover, by increasing
the length of the fiber, the attenuation of the laser becomes non-negligible. From fabricator's data
at 337 nm and for 6.2 meters of PCS 1000, transmittance of the fiber is approximatively 80 %.
For in vivo or in-line monitoring this length allows a disposition where the apparatus and the
measurement take place in different rooms. 415nm is chosen as the emission wavelength because
the signal to noise ratio is at an optimum. When we consider the three principal factors i.e. the UV
excitation, the fluorescence emission (which is at around 400 nm) and optical fiber use, the main
advantage of time-resolved fluorescence over steady state fluorescence is to decrease the parasitic
light generated by fiber by temporal separation.
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
600
The time taken by light for the distance of a round-trip through 6.2 m of the fiber (L) is around
65.5 ns (see Figure 4). The refractive index (n) of the core fiber is 1.467. The numerical aperture
(na) is 0.4. With the geometric model of guided wavesxxix, we write:
(9)
Δ𝑡𝑚𝑎𝑥 =
2𝑛𝐿
𝑐 √1 − 𝑛 𝑎2
(10)
Δ𝑡𝑚𝑖𝑛, =
2𝑛𝐿
𝑐
So tmin = 60.8 ns and tmax = 66.4 ns which is in agreement with the time measured. Choosing
a time window of 100ns lets us characterize the fluorescence of the fiber. In practice
photobiochemical signals take place in a narrow time window close to the exit fluorescence. With
a time window (1ns by division), the noise is reduced to the fluorescence generated at the end of
fiber in contact with the measurement media. Using different media (even in air), this fluorescence
measured at 415 nm emission wavelength, kept the same decay time. By time-resolved
fluorescence, the single-fiber configuration can be used to optimize a FOCS, especially if
detectability is the principal aim. In order to measure 5-HT, a major difference of the extrinsic
FOCS with respect to voltammetry and dialysis methods relies upon the difference of the type of
interaction between tissue and apparatus. For voltammetry, interactions between 5-hydroxyindole
compounds either of tissue or in the little pool on an electrodexxx and electrons are superficial. For
dialysis methods, interactions between 5-HT of an extracellular liquid and 5-HT in dialysat are
through a membrane and are also superficial. For the extrinsic FOCS, interactions between 5-HT
of tissue and extracellular liquid with light, are volumic.
Processing of the signal.
Figure 5 exhibits the integrated intensity of 5-HT and 5-HIAA fluorescences collected in the same
time window versus concentration of each compound. The difference of emission fluorescence
measured through our FOCS appears between these two compounds. For the same signal
(integrated intensity), the concentration of 5-HT is around twenty times more than that of 5-HIAA.
With 415 nm as the emission wavelength, and a bandwidth of 4 nm, 5 µM for 5-HT and 100 µM
for 5-HIAA are the range of the concentration limit determined using the integrated fluorescence
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
601
signal. It is well-known that the normal baseline of extracellular endogenous 5-HT and 5-HIAA
in the brain are respectively in the range of 1 nM to 100 nM and 100 nM to 1 µM
xxxi
. Further
improvements are therefore required in order to reach these values. Nevertheless the feasibility of
a FOCS for the determination of 5-HT using the present technique is established in these rather
stringent conditions arising from the use of UV light which is necessary for our analytical aim.
Several improvements of this first approach may be expected since the detection limit reached is
reduced by noise which clearly depends on our instrumental choices. The nitrogen laser excitation
wavelength is not the optimal one. The use of other optical fibers (even specially developed ones)
together with special end treatments could greatly improve the results. We avoided the use of a
twin fiber configuration to maintain a maximum miniaturization capability. This choice may be
changed if shown to be absolutely necessary. Improvements to the experimental setup using the
modifications listed above could result in an increase to the signal to noise ratio of a factor of
hundred.
Time resolved fluorescence methods allow in principle a more refined signal analysis than
integrated intensity. Temporal data leads to another parameter: fluorescence lifetime. In practice,
the fluorescence response is temporally shifted as compared to the instrumental response function
essentially because of the temporal chromatic shift. This obliges us to move fluorescence
responses forward in order to start with an identical time origin. The instrumental response
function was taken at 337 nm by measuring backscattered light. This was done by tuning the
monochromator at 337 nm and no particles were added to increase scattering. Complications
appear when measuring the instrumental response by reflection of the laser light at the tip of the
fiber i.e. band broadening as reported on Figure 6. The excitation instrumental response cannot
therefore be registered systematically before and after each measurement. To extract the lifetime
from our signal, we have used instrumental responses registered with the cuvette system which do
not suffer from the above mentioned drawback. This may decrease the accuracy of results of
lifetime because we did not take in account possible slight variations of the electronics. If accurate
lifetime must be measured with our system, the method of internal calibration with the use of
reference lifetimes of
POPOP
(p-bis[2-(5-phenyloxazolyl)]
benzene) or PPO (2,5-
diphenyloxazole) could be applied xxxii. Emission fluorescence pulses collected with a single fiber,
are presented in Figure 7. We have normalized some fluorescence pulses in order to show temporal
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
602
evolution from noise (Merck buffer sole) to fluorescence signal with 5-HT concentration. Using
415 nm as the emission wavelength and 4 nm as the bandwidth, experimental lifetimes for 5-HT
(27.7 µM) and 5-HIAA (200 µM) are respectively 2ns and 3ns. These apparent lifetimes are
therefore longer than those measured in the cuvette. Our principal aim when using optical fibers
is not to measure with accuracy the lifetime fluorescence of 5-HT. But rather extracting from the
signal an evaluation of the lifetime in addition to its integration, gives qualitative information and
may be a powerful tool for the identification of the signal measured in vivo .
The in vivo application has been started. Measurements of a signal (excitation wavelength: 337
nm; emission wavelengths: 390 nm to 530 nm) in the cortex and in deep tissue of unanesthetized
rats is possible. Chemical dependence of such an in vivo signal is now under determination. The
results will be published elsewhere. Our technique seems to be powerful and promising for
developing a FOCS adapted to analytical chemistry measurements in the brain.
Conclusion
The static fluorescence emission spectra at different excitation wavelengths, the analysis of the
lifetimes and fractional weighting factors versus excitation wavelengths with the flash lamp
system establish two emission states for 5-HT: the first and well-known, emits at around 340 nm
with a lifetime near to 2.7 ns and the other at around 390 nm with a lifetime near to 1 ns.
With the laser system, we show the feasability of a FOCS. The specificity of the measure is based
on three parameters: excitation and emission wavelengths and lifetimes. No signal has been
observed for indol compounds such as tryptamine. 5-HT and 5-HIAA gave a fluorescence signal
with the same spectral properties but their fluorescence lifetimes are different enough to allow a
separate analytical exploitation. In addition, the blue fluorescence of 5-HT is ten times more
intense than the one of 5-HIAA which could permit measurements of 5-HT in presence of 5-HIAA
with a ratio of up to one to one thousand. Using only the integrated fluorescence intensity, the
detection limit of 5-HT of our FOCS is in the range of 5µM. The noise limits our detectability and
depends mainly on instrumental choices such as the nitrogen laser excitation and the single-fiber
configuration. Moreover, the optical fiber model choice appears to be rather critical. Normal
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
603
baselines of extracellular endogenous 5-HT and 5-HIAA in brain are in the range of 1 nM to 100
nM and 100 nM to 1 µM respectively. Further improvements to our system are necessary so as to
increase the signal to noise ratio by one hundred times . The main drawback of our choices for the
determination of 5-HT is the lowness of the molar extinction of 5-hydroxyindole compounds at
the nitrogen laser excitation wavelength used. An other picosecond source which emits near of
320 nm would allow a better detection limit.
In addition to the characterization of two emission state of 5-HT (excitation and emission
wavelengths and lifetimes), this paper demonstrates the capability of our method as an extrinsic
FOCS for the in vitro determination of 5-HT within less than in thirty seconds. We describe our
investigation to solve the specific problems occurring by the use of a single-fiber configuration at
an excitation wavelength in near UV. In practice, the interest of using time-resolved fluorescence
method to develop a FOCS is clearly shown.
Acknowledgment
The authors wish to thank Professor Teoule for the use of the static fluorometer and Professor J.C.
Talbot for his interest in this work and for making the lamp flash fluorometer available.This work
was supported by INSERM U52, CNRS 1195 and DRET (Contract 87-215).
Registry No. 5-HT, 153-98-0; 5-HIAA, 54-16-0;Tryptamine, 343-94-2.
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
604
Figure 1
Fig. 1. Representationof the setup used for time-resolved fluorescence measurements with the
optical fiber: F, filter;M, mirror;L, lens; PMT, photomultiplier;PD, photodiode.
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
605
Figure 2
Fig. 2. Results in cuvette: (a, top) absorption spectrum of 5-HT in Merck buffer pH 7; (b, bottom)
emission spectra of 5-HT fluorescence in Merck buffer pH 7 at different excitation wavelengths
(concentration equal to 66 µM). Black solid arrows refer to the right ordinates.
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
606
Figure 3
Fig3. Laser time-resolved fluorescence of 5-HT in Merck buffer pH 7 at different concentrations
in cuvette. The emission wavelength is 430 nm with a 4-nm bandwidth.
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
607
Figure 4
Fig. 4. Time-resolved fluorescence of (a, top) 5-HT (30µM) in Merck buffer (pH 7) in cuvette.
The emission wavelength is 450 nm with a 4-nm bandwidth. The excitation wavelength is 3a7 nm
from a nitrogen laser. Curve a: instrument response function; Curve B: emission fluo- rescence
response; Curve 7 (dotted line): calculated response with a 1.2 ns lifetime. (b, bottom) 5-HIAA
(600 µM) with the same conditions used for part a, and calculated response with a 2.1ns lifetime.
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
608
Figure 5
Fig. 5. Emission intensity (log scale) at 415 nm through a 6.2-m-long PCS 1000 optical fiber with
the exit end (end B) dipped into the Merck buffer at pH 7. Peak A is due to retro-fluorescence of
optical fiber mainly at the laser beam injection.
Retro-fluorescence of the fiber tip generates peak B.
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
609
Figure 6
Figure 6. Integrated intensity of emission fluorescence pulse in the Merk buffer at pH7, guided
and collected by the same optical fiber. The emission wavelength is 415 nm with a bandwidth of
4 nm. The excitation wavelength is 337 nm.
• 5-HT fluorescence;
•
5-HIAA fluorescence.
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
Figure 7
Fig 7. Instrument response function (IRF) at a wavelength of 337 nm.
License CC-BY-NC-ND
610
Mottin et al. Appl. Spectrosc. 1993
611
Figure 8
Fia. 8. Scaled fluorescence responses of 5-HT in the Merck buffer at pH 7. The curve with the "C"
symbol is obtained with 5-HT in a cuvette at 430 nm. Other pulses come from fluorescence of 5HT at 415 nm collected with the single-fiber configuration.
License CC-BY-NC-ND
Mottin et al. Appl. Spectrosc. 1993
612
Table 1
exc(nm) 1(%) t1 (ns)
285
295
300
305
310
315
320
325
335
18.9
18.8
24.0
31.6
47.8
75.2
79.3
98.5
96.0
0.55
0.87
0.96
0.85
0.85
1.00
0.84
0.99
1.00
∆t
0.02
0.02
0.03
0.07
0.01
0.03
0.02
0.01
0.01
2(%)
81.1
81.2
76.0
68.4
52.2
24.8
20.7
1.5
4.0
t2
(ns)
2.8
2.6
2.8
2.6
2.8
3.2
1.8
13.3
6.5
∆t
0.01
0.05
0.01
0.04
0.02
0.10
0.09
6.8
1.4
2
DW
f1
f2
1.20
1.14
1.27
1.11
1.51
1.11
1.23
1.12
1.09
1.54
1.85
1.61
1.79
1.36
1.73
1.82
1.90
1.90
4.4
7.2
9.7
13.1
21.6
48.8
64.0
83.0
79.4
95.6
92.8
90.3
86.9
78.4
51.2
36.0
17.0
20.6
Table I. Bi-exponential analysis of 415 nm emission fluorescence of 5-HT in Merck buffer pH7 at
different excitation wavelengths (exc).
i are the pre-exponential weighting factors in percent. ti are the decay times in ns. ∆t are the
estimated error on ti of the Edinburgh Instrument program. 2 is the reduced Chi-square. DW is
the Durbin Watson parameter. fi are the fractional contributions to the total emission for each
component:
𝑓𝑖 =
License CC-BY-NC-ND
𝛼𝑖 𝑡𝑖
∑𝑖 𝛼𝑖 𝑡𝑖
Mottin et al. Appl. Spectrosc. 1993
613
References
i
M. Sallanon, M. Denoyer, and M. Jouvet, Etats de veille et de sommeil; P.Meyer, J.L.Elghozi and A.Q.Salva,
Eds.(Masson, Paris, 1989), pp.36-47.
ii
J. L. Ponchon, R. Cespuglio, F. Gonon, M. Jouvet, and J. F. Pujol, Anal.Chem. 51, 1483 (1979).
iii
T. Sharp, T. Zetterström, L. Christmanson, and U. Ungerstedt, Neuroscience Letters 72, 320 (1986).
iv
P. Kalen, E. Strecker, E. Rosengren, and A. Björklund, J.Neurochem. 5, 1422 (1988).
v
R. Cespuglio, N. Sarda, A. Gharib, H. Faradji, and N. Chastrette, Esp.Brain Res. 64, 589 (1986).
vi
Neuropharmacology of Serotonin, A. R. Green, Ed. (Oxford University Press,1985), p.218 and p.414.
vii
J. Louch and J. D. Ingle, Anal.Chem. 60, 2537 (1988).
viii
Application of Time-resolved Opical Spectroscopy; V. Brüchner and K. H. Feller, Eds.(Elsevier, Amsterdam,
1990), Chap. 3-5.
ix
G. M. Hiefje and G. R. Haugen, Anal.Chim.Acta. 123, 255 (1981).
x
F. V. Bright and K. S. Litwiler, Anal.Chem. 61, 1510 (1989).
xi
T. Matsui, K. Suzuki, M. Sakagami, and T. Kitamori, Appl Spectrosc. 45, 32 (1991).
xii
Y. Kawabata, T. Imasaka, and N. Ishibashi, Anal.Chim.Acta. 173, 367 (1985).
xiii
M. K Carroll, F. V. Bright, and G. M. Hiefje, Anal.Chem. 61, 1768 (1989).
xiv
M.E. Lippish, J. Pusterhofer, M. J. P. Leiner, and O. S. Wolfbeis, Anal.Chim.Acta. 205, 1 (1988).
xv
D. J. Desilets, J. T. Coburn, D. A. Lantrip, P. T. Kissinger and F. E. Lytle, Anal.Chem. 58, 1123 (1986).
xvi
S. A. Nowak, F. Basile, J. T. Kivi, and F. E. Lytle, Appl Spectrosc. 45, 1026 (1991).
xvii
N. Glasser, "Transitions radiatives et non radiatives dans l'indole et ses dérivés." Thesis.University Louis
Pasteur, Strasbourg, France, 15-17 (1979).
xviii
J. R. Platt, Journal of chemical Physics 17, 484 (1949).
xix
S. Udenfriend, D. F. Bogdanski, and H. Weissbach, Science 122, 32 (1955).
xx
S. Udenfriend, D. F. Bogdanski, and H. Weissbach, Science 122, 972 (1955).
xxi
R. F. Chen, Proc.Nat.Acad.Sci.USA 60, 598 (1968).
xxii
G. H.Vickers, R. M. Miller, and G. M. Hieftje, Anal.Chim.Acta. 192, 145 (1987).
xxiii
D. V. O'Connor and D. Phillips, Time-correlated single photon counting; (Academic Press, London, 1984), p.
172.
xxiv
A. Grinvald and I. Z. Steinberg, Anal. Biochem. 59,583 (1974).
xxv
M. K. Carrol and G. M. Hieftje, Appl. Spectosc. 45,1053 (1991).
xxvi
J. N. Demas, Molecular Lumiscence Spectroscopy, Part 2, S. G. Schulman, Ed. (Wiley, N.Y.,1988), p. 93.
xxvii
E. R. Carraway, J. N. Demas, and B. A. Degraff, Anal.Chem. 63, 332 (1991).
xxviii
R. F. Chen, G. G. Vurek, and N. Alexander, Science 156, 949 (1967).
xxix
A. W. Snyder and J. D. Love, Optical waveguide theory (Chapman Hall,London,1983), p.32.
xxx
R. Cespuglio, H. Faradji, Z. Hahn, and M. Jouvet, Measurement of Neurotransmetter Release In Vivo, C. A.
Marsden, Ed. (John Wiley and Sons, Chichester, 1984), pp. 173-191.
xxxi
N. T. Maidment, C. Routledge, K. F. Martin, M. Brazell, and C.A.Marsden, Monitoring Neurotransmitter
Release during behavior, Joseph,M.F.;Fillenz,M., Eds. (Ellis Harwood Health Sciences Series,Chichester,1986), pp.
73-93.
xxxi
F. Castelli, Rev.Sci.Instrum. 56, 538 (1985).
License CC-BY-NC-ND