BJD
British Journal of Dermatology
T R AN SLA T IO NA L RE SE AR CH
In vivo single human sweat gland activity monitoring using
coherent anti-Stokes Raman scattering and two-photon
excited autofluorescence microscopy
X. Chen,1 P. Gasecka,1 F. Formanek,2 J.-B. Galey2 and H. Rigneault1
1
Aix-Marseille Universite, CNRS, Centrale Marseille, Institut Fresnel, UMR 7249, Domaine Universitaire de Saint Jer^ome, F-13397 Marseille CEDEX 20,
France
2
L’Oreal Recherche Avancee, 1 Avenue Eugene Schueller, 93600 Aulnay-sous-Bois, France
Linked Comment: Osseiran and Evans. Br J Dermatol 2016; 174:714–715.
Summary
Correspondence
Herve Rigneault.
E-mail: [email protected]
Accepted for publication
9 November 2015
Funding sources
Centre National de la Recherche Scientifique
(CNRS), Aix-Marseille University A*Midex (no.
ANR-11-IDEX-0001-02) and ANR grants from
the France Bio Imaging (ANR-10-INSB-04-01)
and France Life Imaging (ANR-11-INSB-0006)
infrastructure networks. X.C. was funded by the
Chinese Science Council.
Conflicts of interest
None declared.
DOI 10.1111/bjd.14292
Background Eccrine sweat secretion is of central importance for control of body
temperature. Although the incidence of sweat gland dysfunction might appear of
minor importance, it can be a real concern for people with either hypohidrosis
or hyperhidrosis. However, sweat gland function remains relatively poorly
explored.
Objectives To investigate the function of single human sweat glands.
Methods We describe a new approach for noninvasive imaging of single sweat
gland activity in human palms in vivo up to a depth of 100 lm, based on nonlinear two-photon excited autofluorescence (TPEF) and coherent anti-Stokes
Raman scattering (CARS).
Results These techniques appear to be useful compared with approaches already
described for imaging single sweat gland activity, as they allow better threedimensional spatial resolution of sweat pore inner morphology and real-time
monitoring of individual sweat events. By filling the sweat pore with oil and tuning the CARS contrast at 2845 cm 1, we imaged the ejection of sweat droplets
from a single sweat gland when oil is pushed out by sweat flow. On average,
sweat events lasted for about 30 s every 3 min under the conditions studied. On
the other hand, about 20% of sweat glands were found inactive. TPEF and CARS
were also used to study, at the single pore level, the antiperspirant action of aluminium chlorohydrate (ACH) and to reveal, for the first time in vivo, the formation of a plug at the pore entrance, in agreement with reported ACH
antiperspirant mechanisms.
Conclusions Although data were acquired on human palms, these techniques show
great promise for a better understanding of sweat secretion physiology and
should be helpful to improve the efficacy of antiperspirant formulations.
What’s already known about this topic?
•
Coherent anti-Stokes Raman scattering (CARS) can be used to perform label-free
chemical imaging.
What does this study add?
•
© 2015 British Association of Dermatologists
CARS and two-photon excited autofluorescence can image single sweat gland activity in vivo.
British Journal of Dermatology (2016) 174, pp803–812
803
804 Imaging single human sweat gland activity, X. Chen et al.
The eccrine sweat gland is among the major cutaneous appendages. It is distributed widely across the skin surface in
humans, with densities varying from < 50 glands cm 2 in
some leg areas to > 500 glands cm 2 in the palms of the
hands.1 The major function of eccrine sweat glands is to regulate the body’s core temperature by water evaporative heat
dissipation under thermal stress conditions.2 In addition to
thermal stimulation, sweating is related to a variety of triggers
in everyday modern life, such as spicy food consumption,
mental stress and anxiety.3–5
Primary sweat is almost isotonic to plasma when entering
the duct, but is then rendered hypotonic during its passage to
the skin surface by reabsorption of sodium chloride and water
along the duct epithelium. Sweat collected at the skin surface
is therefore composed mainly of water containing sodium
chloride and low levels of solutes such as potassium, lactate,
urea, bicarbonate, glycoproteins and peptides at pH 6–7.2 Elevated sweat sodium chloride concentration is well known to
be one of the diagnostic criteria for cystic fibrosis.
The palms of the hands are a site of permanent sweating,
which can be stimulated by physical or mental stress. This
hydration increases the skin friction coefficient considerably
and therefore improves the adherence to objects and contributes to the sense of touch. Sweat is also involved in skin
homeostasis as a constituent of the hydrolipidic film that covers the skin surface. It also exerts both a moisturizing action
and antimicrobial activity related to its content of antimicrobial peptides such as dermcidin.6
Although many eccrine sweat gland dysfunctions are
known,7 sweat gland research remains relatively poor. In this
context, and with the scope of improving our understanding
of eccrine sweat delivery, this paper focuses on recently developed nonlinear optical imaging techniques to access previously
inaccessible information on single sweat gland activity (SSGA).
We focus here on the inner morphology of the sweat pore
with high three-dimensional spatial resolution, as well as realtime monitoring of individual sweat events. These nonlinear
imaging techniques provide imaging up to a depth of
100 lm, giving access to the intraepidermal part of the sweat
duct, called the acrosyringium, which has a luminal diameter
of 20–60 lm. Sweat duct orifices, known as ductal pores,
have a funnel-like appearance in specific areas such as the
palm of the hand, where they are distributed along skin
ridges, making them most accessible for microscopic investigation.
Various techniques have been explored over the years to
monitor sweat activity. Older techniques include starch–iodine
paper,8 plastic imprint and collection of sweat accumulated
under metallic oil.9 Recent techniques include infrared thermographic imaging,10 direct video monitoring,11 optical
coherence tomography (OCT)12,13 and colorimetric mapping
using hydrochromic polymers.14
Except for OCT, most of these techniques are not suitable
to measure instantaneous sweating. Indeed, in the case of
mental and physical stress, sweat glands were shown to discharge sweat in a pulsatile mode.15 SSGA has also been moniBritish Journal of Dermatology (2016) 174, pp803–812
tored continuously at the skin surface using a dedicated
chamber and electrical conductivity measurement.16 Nevertheless, the method measured an electrolyte secretion rate that
differs from the real sweat rate. Furthermore it provided no
information about the inner sweat pore morphology.
A different approach is taken in the present work, aiming at
exploring, for the first time, in vivo human SSGA using nonlinear imaging. Vibrational imaging techniques such as coherent anti-Stokes Raman scattering (CARS)17 and stimulated
Raman scattering18,19 have provided an affordable, chemically
specific, label-free contrast mechanism suitable for fast imaging of the skin.20–22 Quite noticeably, these vibrational contrast mechanisms are very efficient at imaging the aliphatic C–
H bonds at 2845 cm 1 found in oils, and in vivo imaging of
sebaceous glands has been recently reported using CARS
microscopy.23
In this work we use two techniques: firstly, two-photon
excited autofluorescence (TPEF) microscopy to map the sweat
pore’s inner morphology, and secondly, CARS microscopy to
monitor SSGA when exogenous oil, inserted in the sweat pore,
is flushed out during pulsatile sweat discharges. The activity
and mechanism of a commercially available antiperspirant product (aluminium chlorohydrate salt) are also explored.
Materials and methods
Two-photon excited autofluorescence and coherent antiStokes Raman scattering nonlinear imaging
Nonlinear imaging (TPEF, CARS) was performed using a custom-built set-up incorporating a picosecond stimulated Raman
optical source.24 This source is composed of two optical parametric oscillators (OPO1 and OPO2, Emerald; APE GmbH,
Berlin, Germany) synchronously pumped by a mode-lock frequency-doubled neodymium-doped yttrium vanadate laser
(PicoTrain; High Q Laser GmbH, Rankwell, Austria) operating
at 532 nm. The two beams from OPO1 (pump) and OPO2
(Stokes) (pulse duration 5 ps, repetition rate 76 MHz) are
overlapped in time and space and sent into a custom-made
scanning microscope. For C–H bond (olive oil) imaging
OPO1 operates at a wavelength of 735 nm, whereas OPO2 is
set at 929 nm to generate a CARS signal at 608 nm
(2845 cm 1). The two beams are recombined using a 750nm short-pass filter. The CARS and TPEF signals are detected
at the same time in the backward direction using two photomultiplier tubes (H10682; Hamamatsu Photonics, Hamamatsu,
Japan) working in a photon-counting regime.
Two-photon excited autofluorescence results mostly from
skin interaction with the pump beam (735 nm) and originates
mostly from nicotinamide adenine dinucleotide phosphate,
flavin adenine dinucleotide, tryptophan, keratin and melanin25
present in the epidermis. Excitation and epifluorescence collection are provided by a numerical aperture 115 objective lens
(Nikon 409; Nikon, Tokyo, Japan). Incident powers at the
sample plane were 30 mW for both pump and Stokes beams.
The objective can be controlled in the z-direction with a
© 2015 British Association of Dermatologists
Imaging single human sweat gland activity, X. Chen et al. 805
piezoelectric scanner with a z-step of 1 lm (max range
400 lm). The signals (TPEF and CARS) are first filtered from
the laser beams by a 700-nm short-pass filter, then the TPEF
signal is separated from the anti-Stokes by a short-pass
dichroic beam splitter at 553 nm. The TPEF signal is collected
in the 400- to 550-nm spectral window using a band-pass filter, whereas the CARS signal is collected within a narrow
spectral window of 15 nm with a band-pass filter centred at
605 nm. Typical CARS/TPEF images acquired throughout this
work were x–y images, 250 9 250 pixels (field of view
100 9 100 lm, 03 s per image) and x–z images
(100 9 100 lm, 10 s per image). Figure S1 (see Supporting
Information) shows a diagram of the set-up.
fixed (Fig. 1d). The home-designed sample holder is shown
in detail in Figure 1e. The excitation laser (wavelength
735 nm, pulse duration 5 ps, repetition rate 76 MHz, power
at the sample plane ~30 mW) comes from a water objective
(Nikon 409; numerical aperture 115) and the generated
TPEF (or CARS – see below) signal is collected by the same
objective in epifluorescence detection. Details on the nonlinear
microscope24 can be found in the Materials and methods section and Figure S1 (see Supporting Information).
Figure 2 illustrates the TPEF imaging process of the morphology of a single sweat pore. A stack of x–y images is taken
from the skin surface down to a depth of 150 lm (acrosyringium), with a z-step of 1 lm. The dark areas, corresponding to the sweat pore and the following spiral section of the
eccrine duct, appear contrasted over the bright epithelial cells
surrounding the duct (Fig. 2b). The contrast fades with
increasing depth (z > 100 lm) because light scattering and
absorption in the skin alter the sharpness of the beam focus.
Only a weak fluorescence coming from the epidermis can be
recorded at a depth of 150 lm. Although confocal reflectance
microscopy can also be used for skin imaging,26 TPEF provides a better signal-to-noise ratio, especially when scattering
is strong.
Retrieving the three-dimensional morphology of a single
sweat pore requires further image processing. We have chosen
a semiautomatic image segmentation method named a ‘region
growing method’ that is well adapted to the delineation of the
dark areas inside the duct. The ‘region growing method’ looks
for groups of pixels of similar intensity starting from a
selected pixel or group of pixels (called the ‘seed’). The intensity of the neighbouring pixels is then examined to decide
whether they have to be removed or added to the growing
region. The object is finally represented by all accepted pixels
during the growing procedure.27 The ‘seed’ is at first manually selected within the dark region corresponding to the duct,
Results
In vivo single human sweat pore imaging using twophoton excited autofluorescence
(a)
Sweat pore
Epidermis
Figure 1 shows the experimental scheme we followed to
image in vivo human sweat pore and sweat gland activity. The
morphology of the full eccrine sweat gland is depicted in Figure 1a. Throughout this work the microscopy imaging depth
is limited to the epidermis, and the reported investigations
concentrate on the acrosyringium. The left hand of a volunteer
individual is cleaned with water and a region on the palm is
delineated (Fig. 1b) where a glass coverslip (diameter
25 mm, thickness 150 lm) is positioned directly on the skin.
Next, a custom-made metallic holder featuring a 20-mm clear
aperture is firmly attached to the skin using doubled-sided
adhesive so that the glass coverslip fits into its aperture
(Fig. 1c). The glass coverslip flattens the skin surface and eases
the image acquisition. The hand together with its holder is
then securely fixed on the microscope stage with magnets for
in vivo imaging, and the volunteer’s left elbow is maintained
Acrosyringium
(b)
(d)
Dermis
Straight
eccrine duct
(c)
Magnet
(e)
Coverslip
Hand holder
(magnet)
NA 1·15
Microscope
sample stage
40 x
Excitation laser
(735 nm, 5 ps, 76 MHz)
Eccrine
sweat gland
TPEF
CARS
Fig 1. In vivo single sweat pore nonlinear imaging. (a) Diagram of eccrine sweat gland morphology; (b) selected location on the palm; (c) sample
holder that can be attached to the microscope sample stage with magnets; (d) hand position on the microscope for in vivo two-photon excited
autofluorescence (TPEF)/coherent anti-Stokes Raman scattering (CARS) imaging and (e) detailed design of the sample holder.
© 2015 British Association of Dermatologists
British Journal of Dermatology (2016) 174, pp803–812
806 Imaging single human sweat gland activity, X. Chen et al.
(a) Sample
(b) TPEF stack
(c) 3D view (threshold)
Skin surface
Image size (X-Y):
150 µm x 150 µm
Z = 0 µm
m
Z = 10 µ
85 µm
70
µm
m
Z = 55 µ
m
Z = 70 µ
100 µm
z-step = 1 µm
m
Z = 26 µ
m
Z = 85 µ
Z = 100
µm
Z = 130
µm
Z = 150
µm
Fig 2. Single sweat pore two-photon excited autofluorescence (TPEF) imaging. (a) Selected palm area; (b) TPEF z-stack images and (c) threedimensional morphology of the acrosyringium segment (sweat pore and the following spiral eccrine duct – greyscale is used for rendering
purposes).
for each image in the z-stack (Fig. 2b). Details of the image
processing are given in Figure S2 and the accompanying text
(see Supporting Information).
Figure 2c shows the result of the image segmentation process where the three-dimensional sweat pore geometry is
revealed. The funnel-like appearance of the sweat pore and the
spiral shape of the sweat duct located in the epidermis can be
clearly distinguished up to a depth of 100 lm.
Single sweat gland activity monitored using coherent
anti-Stokes Raman scattering
In vivo SSGA is studied on the volunteers’ palms in a geometry
similar to that shown in Figure 1. We implemented a method
where the discharged sweat droplets are visualized when they
develop into added olive oil that fills the sweat pore cavity.
Although the use of an oil to observe sweat droplets is not
new, and was first devised by J€
urgensen in 1924,28 here we
use CARS microscopy to image olive oil directly in a single
sweat pore in vivo. This strategy is superior to direct CARS
sweat water imaging, as water immediately fills the spacing
British Journal of Dermatology (2016) 174, pp803–812
between the glass coverslip and the skin surface. CARS uses a
pump and a Stokes beam with a frequency difference matching the vibrational frequency of the targeted molecular
bond,29 and allows fast imaging of C–H bonds present in
lipids to be recorded.30 To achieve the strongest contrast in
olive oil we target the aliphatic CH2 vibrational resonance at
2845 cm 1. The excitation pump beam is set at 735 nm and
the Stokes beam at 929 nm to generate a CARS signal at
608 nm. At the same time, the skin-native TPEF is collected to
reveal the sweat pore morphology with increasing depth. To
allow better observation of sweat droplets in oil in vivo, the
hand holder was modified by adding a 05-mm spacer
between the skin surface and the coverslip to create a ‘pool’
of oil. Details of the sample holder are shown in Figure S3
(see Supporting Information).
Figure 3 shows examples of SSGA from a 28-year-old Asian
woman. A single sweat pore volume filled with olive oil is
imaged with TPEF and CARS microscopy at various depths.
The TPEF image reveals the sweat pore morphology (Fig. 3a),
whereas the CARS image shows how the oil penetrates into
the sweat pore cavity (Fig. 3b). During sweat discharge the
© 2015 British Association of Dermatologists
Imaging single human sweat gland activity, X. Chen et al. 807
TPEF
(a)
Z = 0 µm
Z = 12 µm
Z = 22 µm
20 µm
Z = 32 µm
Z = 0 µm
Z = 12 µm
Z = 22 µm
20 µm
Z = 32 µm
Z = 0 µm
Z = 12 µm
Z = 22 µm
20 µm
Z = 32 µm
CARS
(b)
CARS
(c)
Fig 3. Two-photon excited autofluorescence (TPEF) and coherent anti-Stokes Raman scattering (CARS) images of single sweat gland activity. (a)
TPEF images of the sweat pore with increasing depth, (b) CARS images (same location) of the sweat pore filled with olive oil and (c) a sweat
discharge as seen with CARS (the ejected aqueous sweat is visualized in black).
oil is flushed out from the sweat pore, and the aqueous sweat
volume, free of lipids, is visualized in black (no CARS signal)
(Fig. 3c). In contrast to CARS, TPEF is necessary to reveal the
in-depth sweat pore morphology, as oil penetrates only inside
the opening of the spiral eccrine duct (up to 30 lm; Fig. 3b).
Furthermore, the CARS signal from oil is not constant in time
because of the sweating process.
Figure 4a and Movie S1 (see Supporting Information) show
time-lapse CARS imaging at the skin surface (opening of sweat
pore) of an SSGA from a 28-year-old Asian woman. A single
pore is observed for 10 min (as shown following the arrow
in Fig. 4a). Starting from time t = 0 the oil slowly penetrates
into the sweat pore. At 43 s a first sweat droplet is secreted
from the pore with a sudden expansion of its area (within
2 s), then it follows a slower but gradual increase of the sweat
area during 20–30 s, which corresponds to the sweat release
at the skin surface. At 1 min 18 s the sweat area has reached
its maximum. Then the sweat area decreases as the oil penetrates back into the pore cavity until 7 min 12 s. At 7 min
18 s a second sweat event happens, until it reaches its maximum sweat area 30 s later at 7 min 48 s. Figure 4b presents
the SSGA chronogram, associated with Figure 4a, where we
have distinguished the ‘sweat time’ sections corresponding to
© 2015 British Association of Dermatologists
an expansion of the sweat area as seen by CARS microscopy
and the ‘relax time’ sections in between.
Figure 4c presents the results of a statistical study to quantify the ‘relax’ and ‘sweat’ times and sweat ‘period’. Thirtyone sweat events coming from 13 different sweat pores were
recorded from two volunteers (both 28-year-old women, one
Asian and one white). The mean values were calculated, and
the boxes show that 50% of the data are distributed around
the mean. Most of the ‘relax time’ happens within a 2- to 4min duration with an average of 3 min. The ‘sweat time’ histogram is more dispersed in the range 0–50 s, yet the average
sweat time is 30 s. A mean 30-s ‘sweat time’ is in good
agreement with recent OCT measurements of sweat droplet
formation at the skin surface.13 Meanwhile, the ‘sweat period’,
defined from the beginning of one sweat event to the beginning of another sweat event that follows, is 198 s on average,
indicating that for a single sweat pore, sweat events happen
on average every 33 min.
Such recordings also make the quantification of the early
sweat event possible, corresponding to the rapid (few seconds) expansion of the sweat area immediately before its start.
Figure 5 shows the x–z plane time-lapse CARS image of the
sweat pore cavity opening during the first minute of an SSGA.
British Journal of Dermatology (2016) 174, pp803–812
808 Imaging single human sweat gland activity, X. Chen et al.
(a)
20 µm
00’00·0’’
00’21·4’’
07’18·7’’
07’19·3’’
(b)
07’18·1’’
07’21·5’’
Relax time
43 s
0
00’42·2’’
00’43·2’’
07’12·6’’
07’48·7’’
00’43·8’’
02’51·6’’
08’09·0’’
02’05·0’’
08’55·8’’
00’44·9’’
00’50·4’’
01’47·2’’
09’33·8’’
01’18·0’’
10’12·4’’
Sweat time
34 s
5 min 58 s
2 min
4 min
2 min 3 s
42 s
6 min
8 min
10 min
(c)
Fig 4. Single sweat gland activity (SSGA) as seen by time-lapse coherent anti-Stokes Raman scattering imaging. (a) SSGA as seen at the skin
surface; the ejected aqueous sweat appears in black; (b) chronogram of the same and (c) statistical histogram of ‘relax time’, ‘sweat time’ and
‘sweat period’ (defined as the sum of the relax and sweat times).
The images are taken directly from the inverted microscope
(Fig. 1d), with the skin surface located at the bottom (z = 0)
of each time-lapse image. At t = 0 the cavity is full of oil,
whereas at 32 s the pulsatile sweat discharge has flushed out
all of the oil from the sweat pore. Indeed, Figure 5 shows
how the sweat expansion is mapped, although the sweat droplet remains in the sweat pore. From the experimental data
(Fig. 5), we find that a sweat volume of 154 pL is ejected in
18 s, giving a sweat flow of 09 pL s 1.
This result suggests that what is measured here is the very
early step of the sweating process when sweat flow is still
increasing. Indeed, the reported total sweat volume ejected
from an SSGA is in the order of few nanolitres per minute,31,32 which gives a sweat flow of about tens of picolitres
per second, a value one order of magnitude higher than our
British Journal of Dermatology (2016) 174, pp803–812
measurements. Nevertheless, previously reported techniques31,32 were unable to follow an SSGA at its early stage,
and rather the dynamics were integrated over many sweat
pore events. We believe that our measurements highlight the
early ‘transient sweat flow’ that has not yet reached its steady
value. However, the total sweat volume cannot be measured;
this is because when the water channel opens into the oil, our
field of view and three-dimensional imaging speed are not
high enough to follow the total water volume that extends in
three dimensions into the skin/coverslip spacing.
Aluminium chlorohydrate antiperspirant action
The most widely used topical antiperspirant agents comprise
aluminium salts such as aluminium chlorohydrate (ACH).
© 2015 British Association of Dermatologists
Imaging single human sweat gland activity, X. Chen et al. 809
00’00’’
20 µm
Z stack
00’14’’
Layer reconstruction
Fig 5. Left: coherent anti-Stokes Raman
scattering time-lapse images of the sweat pore
cavity during the first minute of single sweat
gland activity. The oil is ejected from the top
to the bottom (the skin surface is located at
the bottom following the experimental
geometry of Fig. 1d, e). Right: the ejected oil
area is selected on each z-stack by applying an
intensity threshold filter, and the threedimensional total ejected volume of oil is
calculated by summing up all of the selected
voxels (voxel size 2 9 2 9 1 lm)
(represented in yellow).
00’32’’
3D volume
00’46’’
Z
Introduced for the first time in 1916,33 their efficiencies have
since been extensively studied.34–36 ACH is generally believed
to dissolve in the sweat on the skin surface and, to some
extents, to diffuse down to the sweat duct. The dissolved salts
forms a gel in situ, in the presence of proteins, creating a
‘plug’ in the duct near the sweat pore, thus reducing the
sweat flux reaching the skin surface. Additional biological
effects of aluminium salts upon sweat production have also
been described, such as ionic exchange blockage at the distal
acrosyringium and structural changes of eccrine glands.35
In a recent article, Yanagishita et al., through histology, clarified the localization of aluminium in palmar skin after topical
ACH treatment. They confirmed the precipitation of an amorphous gel that includes keratin and polysaccharides in the
sweat duct in the stratum corneum.37 However, visualization
of these plugs in vivo with a noninvasive technique has never
been published. Moreover, a precise and comprehensive
description of what happens exactly inside sweat pores when
sweat gets in contact with ACH is still lacking. Such knowledge could be very helpful either to improve ACH antiperspirant efficacy or to replace it with aluminium-free alternatives.
Other issues concern the biological impact of pore plugging
on sweat secretion and reabsorption. In this context, we
describe here how we used our SSGA CARS imaging method
to study, in vivo, the effect of topical ACH formulation
applications.
A commercial antiperspirant roll-on product containing
15% ACH was topically applied for 2 days every 12 h onto
the palm of a 28-year-old Asian woman. A group of treated
sweat pores along the ridge lines was then observed at 24, 36
© 2015 British Association of Dermatologists
Z
X
X
and 48 h during the ACH application and compared with
nontreated sweat pores located on the same palm, left as controls. In total 116 sweat pores were observed to examine their
activities (whether or not sweat droplets develop into oil that
fills the sweat pore cavity), each of them for a duration of at
least 5 min (a time longer than the sweat period). Figure 6
shows the result of our study in terms of the percentage of
inactive sweat glands (defined over the total number of sweat
pores tested).
In the control areas, 19% of sweat glands were inactive
(seven of 36 tested). This value is in good agreement with
previously reported results.1 In the areas where the ACH product was applied, at 24 h (i.e. after two applications at 0 and
12 h), 10% of the tested sweat glands were inactive (three of
31 tested). The inactive gland percentage increased to 29%
(six of 21 tested) at 36 h (after three applications at 0, 12
and 24 h) and reached 89% (25 of 28 tested) at 48 h (after
four applications at 0, 12, 24 and 36 h). This implies that the
number of inactive sweat glands increases by 468 times following ACH treatment. In another words, the number of
active sweat glands, and therefore the sweat volume produced,
decreases by 864%, assuming that the remaining active sweat
pores experience no change in their pulsing frequency and
pulse strength under ACH treatment. These results demonstrate
the antiperspirant efficiency of the tested ACH product at the
single sweat pore level.
To observe the ‘plug’ resulting from the interaction
between ACH and sweat or skin-surface proteins, a focus
was made on sweat pore morphology and oil distribution
in the acrosyringium segment under ACH application. FigBritish Journal of Dermatology (2016) 174, pp803–812
810 Imaging single human sweat gland activity, X. Chen et al.
Nonactive sweat glands / sweat pores tested
100%
89%
80%
60%
40%
20%
29%
No treatment
(control)
10%
0%
Observaon me
ure 7 shows an example of TPEF and CARS imaging of the
sweat pore cavity at 48 h. As previously mentioned, TPEF
gives access to the pore morphology (Fig. 7a), whereas
CARS reveals the presence of the oil that penetrates into the
pore cavity (Fig. 7b). At a depth of 20 lm a ‘plug’ structure can be seen in both the TPEF and CARS images, which
develops down to the imaging depth limit at 35 lm (the
penetration depth of the oil). The ‘plug’ structure and location can be better observed in the x–z (Fig. 7c) and y–z
Fig 6. Percentage of inactive sweat glands
under aluminium chlorohydrate treatment
(one application every 12 h) on human palm
skin as observed at 24, 36 and 48 h.
(Fig. 7d) planes. Note that in these images the skin surface
is located at z = 0 (following the experimental geometry of
Fig. 1c).
As aluminium chlorohydrate has no fluorescence properties,
TPEF contrast in plugs is likely to be linked to the presence of
proteins embedded in the gel originating from either sweat or
corneocytes detached from the skin surface. Eccrine sweat is
indeed known to contain small amounts of proteins and glycoproteins, at a concentration around 50 lg mL 1.2 Although
(a)
(b)
(c)
(d)
Fig 7. Aluminium chlorohydrate (ACH)/protein ‘plug’ as seen by two-photon excited autofluorescence (TPEF) and coherent anti-Stokes Raman
scattering (CARS) imaging in a human palm sweat pore at 48 h following ACH applications (one application every 12 h). (a) TPEF images and
(b) CARS images with increasing depth. (c, d) Combined TPEF and CARS images in the (c) x–z and (d) y–z planes. The ‘plug’ is highlighted with
a red dashed circle in (c) and (d). In (c) and (d) the skin surface is located at bottom following the experimental geometry of Figure 1d, e.
British Journal of Dermatology (2016) 174, pp803–812
© 2015 British Association of Dermatologists
Imaging single human sweat gland activity, X. Chen et al. 811
this concentration may appear very low compared with the
aluminium concentration, it is conceivable that, once the aluminium protein aggregation is initiated, the sweat flow could
feed the growing plug with protein, thereby leading to progressive obstruction of the sweat pore. We also tried to detect
plugs by tuning CARS contrast to a vibrational region that
could be specific to aluminium chlorohydrate. However,
Raman spectra of such inorganic structures show bands only
at very low wave numbers (< 500 cm 1), which are not
reachable with our set-up. As highlighted with the red dashed
circle in Figure 7c,d, the ‘plug’ appears as a disorganized
structure located at the end of the sweat pore and closing the
spiral duct entrance. The oil is seen not to penetrate deeper
than this ‘plug’ because of the sweat duct obstruction. Concomitantly, the ‘plug’ is believed to block sweat droplet ejection and to prevent it from reaching the skin surface. To our
knowledge, such plug formation has never been reported
in vivo.
Discussion
In this study, SSGA, a relatively unexplored field in dermatology, is imaged in vivo in human palm skin using the nonlinear
microscopy techniques TPEF and CARS. We show that the
combined TPEF and CARS nonlinear images can reveal morphological and functional activity in vivo at the level of a single
human sweat gland. Filling the pore with oil, we imaged pulsatile sweat discharges and described single sweat gland
dynamics in terms of ‘sweat’ and ‘relax’ events. As the sweat
dynamics are observed directly in the sweat pore cavity, this
work highlights an early transient sweat flow regime occurring during the first tens of seconds of a sweat event. Combined TPEF and CARS imaging appears to be superior to
previously described SSGA imaging approaches as it allows a
better three-dimensional spatial resolution of the sweat pore
inner morphology and real-time monitoring of individual
sweat events.
Our study additionally explored the action of ACH, a
known antiperspirant ingredient, on the sweat gland dynamics at the single pore level and confirmed its efficiency. We
have shown that the combined TPEF and CARS images can
reveal the appearance of a ‘plug’, located at the junction
between the sweat pore and the spiral eccrine duct, resulting from the interaction between ACH and skin or sweat
proteins. The visualization of these plugs in vivo with a
noninvasive technique has never been reported before, and
clarifies the antiperspirant mechanism of ACH. Altogether,
these results show that nonlinear microscopy offers the
powerful ability to study in vivo the activity of the human
eccrine system with potential novel applications in dermatology.
Acknowledgments
The authors acknowledge Patrick Ferrand for the implementation of the scanning NLO microscope software.
© 2015 British Association of Dermatologists
References
1 Taylor NA, Machado-Moreira CA. Regional variations in transepidermal water loss, eccrine sweat gland density, sweat secretion
rates and electrolyte composition in resting and exercising
humans. Extrem Physiol Med 2013; 2:4.
2 Sato K, Kang WH, Saga K, Sato KT. Biology of sweat glands and
their disorders. I. Normal sweat gland function. J Am Acad Dermatol
1989; 20:537–63.
3 Amir M, Arish A, Weinstein Y et al. Impairment in quality of life
among patients seeking surgery for hyperhidrosis (excessive
sweating): preliminary results. Isr J Psychiatry Relat Sci 2000;
37:25–31.
4 Davidson JR, Foa EB, Connor KM, Churchill LE. Hyperhidrosis in
social anxiety disorder. Prog Neuropsychopharmacol Biol Psychiatry 2002;
26:1327–31.
5 Lore RK. Palmar sweating and transitory anxiety in children. Child
Dev 1966; 37:116–24.
6 Murota H, Matsui S, Ono E et al. Sweat, the driving force behind
normal skin: an emerging perspective on functional biology and
regulatory mechanisms. J Dermatol Sci 2015; 77:3–10.
7 Sato K, Kang WH, Saga K, Sato KT. Biology of sweat glands and
their disorders. II. Disorders of sweat gland function. J Am Acad Dermatol 1989; 20:713–26.
8 Sato K, Dobson RL. Regional and individual variations in function
of human eccrine sweat gland. J Invest Dermatol 1970; 54:443–9.
9 Ohhashi T, Sakaguchi M, Tsuda T. Human perspiration measurement. Physiol Meas 1998; 19:449–61.
10 Kim Y-S, Yoon Y-S. The new diagnostic method for hyperhydrosis
using digital infrared thermographic imaging. Clin Neurol Neurosurg
1997; 99:S156.
11 Gomez DS, Mariani U, Leirner AA et al. A new method for scoring
active sweat glands. Clinics (Sao Paulo) 2005; 60:505–6.
12 Jonathan E. In vivo sweat film layer thickness measured with Fourier-domain optical coherence tomography (FD-OCT). Optics Lasers
Eng 2008; 46:424–7.
13 Ohmi M, Tanigawa M, Wada Y, Haruna M. Dynamic analysis for
mental sweating of a group of eccrine sweat glands on a human
fingertip by optical coherence tomography. Skin Res Technol 2012;
18:378–83.
14 Lee J, Pyo M, Lee SH et al. Hydrochromic conjugated polymers for
human sweat pore mapping. Nat Commun 2014; 5:3736.
15 Ogawa T, Sugenoya J. Pulsatile sweating and sympathetic sudomotor activity. Jpn J Physiol 1993; 43:275–89.
16 Shamsuddin AKM, Togawa T. Continuous monitoring of singlesweat-gland activity. Physiol Meas 2000; 21:535–40.
17 Zumbusch A, Langbein W, Borri P. Nonlinear vibrational microscopy applied to lipid biology. Prog Lipid Res 2013; 52:615–32.
18 Freudiger CW, Min W, Saar BG et al. Label-free biomedical imaging with high sensitivity by stimulated Raman scattering microscopy. Science 2008; 322:1857–61.
19 Nandakumar P, Kovalev A, Volkmer A. Vibrational imaging based
on stimulated Raman scattering microscopy. N J Phys 2009;
11:033026.
20 Saar BG, Contreras-Rojas LR, Xie XS, Guy RH. Imaging drug delivery to skin with stimulated Raman scattering microscopy. Mol
Pharm 2011; 8:969–75.
21 Belsey NA, Garrett NL, Contreras-Rojas LR et al. Evaluation of
drug delivery to intact and porated skin by coherent Raman scattering and fluorescence microscopies. J Control Release 2014;
174:37–42.
22 Chen X, Gregoire S, Formanek F et al. Quantitative 3D molecular
cutaneous absorption in human skin using label free nonlinear
microscopy. J Control Release 2015; 200:78–86.
British Journal of Dermatology (2016) 174, pp803–812
812 Imaging single human sweat gland activity, X. Chen et al.
23 Jung Y, Tam J, Jalian HR et al. Longitudinal, 3D in vivo imaging of
sebaceous glands by coherent anti-Stokes Raman scattering microscopy: normal function and response to cryotherapy. J Invest Dermatol 2015; 135:39–44.
24 Brustlein S, Ferrand P, Walther N et al. Optical parametric oscillator-based light source for coherent Raman scattering microscopy:
practical overview. J Biomed Opt 2011; 16:021106.
25 Zipfel WR, Williams RM, Christie R et al. Live tissue intrinsic emission microscopy using multiphoton-excited native fluorescence
and second harmonic generation. Proc Natl Acad Sci USA 2003;
100:7075–80.
26 Wang H, Lee AM, Frehlick Z et al. Perfectly registered multiphoton
and reflectance confocal video rate imaging of in vivo human skin. J
Biophotonics 2013; 6:305–9.
27 Bankman IN, ed. Handbook of Medical Image Processing and Analysis. Amsterdam: Academic Press, Elsevier, 2008.
28 J€
urgensen E. Mikrobeobachtungen der Schweisssekretion der Haut
des Menschen unter Kontrastf€arbung. Arch F Klin Med 1924;
144:193–201.
29 Zumbusch A, Holtom GR, Xie XS. Vibrational microscopy using
coherent anti-Stokes Raman scattering. Phys Rev Lett 1999;
82:4014–17.
30 Evans CL, Potma EO, Puoris’haag M et al. Chemical imaging of tissue in vivo with video-rate coherent anti-Stokes Raman scattering
microscopy. Proc Natl Acad Sci USA 2005; 102:16807–12.
31 Lee JB, Kim TW, Shin YO et al. Effect of the heat-exposure on
peripheral sudomotor activity including the density of active sweat
glands and single sweat gland output. Korean J Physiol Pharmacol
2010; 14:273–8.
32 Provitera V, Nolano M, Caporaso G et al. Evaluation of sudomotor
function in diabetes using the dynamic sweat test. Neurology 2010;
74:50–6.
British Journal of Dermatology (2016) 174, pp803–812
33 Stillians AW. The control of localized hyperhidrosis. JAMA 1916;
67:2015.
34 Reller HH, Lueders WL. Pharmacologic and toxicologic effects of
topically applied agents on the eccrine sweat glands. Adv Mod Toxicol
1977; 4:1–54.
35 Holze E, Kligman AM. Mechanism of antiperspirant action of aluminum salts. J Soc Cosmet Chem 1979; 30:279–95.
36 Quatrale RP, Coble DW, Stoner KL et al. The mechanism of
antiperspirant by aluminum salts II: histological observations of
human eccrine sweat glands inhibited by aluminum chlorohydrate.
J Soc Cosmet Chem 1981; 32:107–36.
37 Yanagishita T, Tamada Y, Ohshima Y et al. Histological localization
of aluminum in topical aluminum chloride treatment for palmar
hyperhidrosis. J Dermatol Sci 2012; 67:69–71.
Supporting Information
Additional Supporting Information may be found in the online
version of this article at the publisher’s website:
Movie S1. Sweat events visualized with time-lapse coherent
anti-Stokes Raman scattering imaging from a 28-year-old
Asian woman.
Figure S1. Two-photon excited autofluorescence and coherent anti-Stokes Raman scattering nonlinear microscope.
Figure S2. Image segmentation of a sweat pore and the spiral eccrine duct segment using the ‘region growing method’
algorithm.
Figure S3. Modified sample holder for single sweat gland
activity imaging.
© 2015 British Association of Dermatologists