Annals of The New York Academy of Sciences - 2017 - Volz - Pitfalls in Using Fluorescence Tagging of Nanomaterials

Ann. N.Y. Acad. Sci.
ISSN 0077-8923
A N N A L S O F T H E N E W Y O R K A C A D E M Y O F SC I E N C E S
Issue: Tight Junctions and their Proteins
ORIGINAL ARTICLE
Pitfalls in using fluorescence tagging of nanomaterials:

tecto-dendrimers in skin tissue as investigated by
Cluster-FLIM
Pierre Volz,1 Priscila Schilrreff,2 Robert Brodwolf,1 Christopher Wolff,3 Johannes Stellmacher,1
Jens Balke,1 Maria J. Morilla,2 Christian Zoschke,3 Monika Schäfer-Korting,3
and Ulrike Alexiev1
1
Institute of Experimental Physics, Freie Universität Berlin, Berlin, Germany. 2 Nanomedicine Research Program
(Departamento de Ciencia y Tecnologia), Universidad Nacional de Quilmes, Buenos Aires, Argentina. 3 Institute for Pharmacy
(Pharmacology and Toxicology), Freie Universität Berlin, Berlin, Germany
Addresses for correspondence: Professor Dr. Ulrike Alexiev, Institute of Experimental Physics, Freie Universität Berlin,
Arnimallee 14, 14195 Berlin, Germany. [email protected]; Professor Dr. Monika Schäfer-Korting, Institute for
Pharmacy (Pharmacology and Toxicology), Freie Universität Berlin, Königin-Luise Str. 2+4, 14195 Berlin, Germany.
[email protected]
Targeted topical application promises high drug concentrations in the skin and low systemic adverse effects. To
locate drugs and drug-delivery systems like nanocarriers, fluorescent dyes are commonly used as drug surrogates
or nanocarrier labels in micrographs of tissue sections. Here, we investigate how labeling degree, concentration of
fluorophore, and nanocarrier may affect the interpretation of these micrographs. False-negative penetration results
due to inter- and intramolecular quenching effects are likely. Using tecto-dendrimers as an example, we present a
detailed analysis of pitfalls in the (semi-)quantitative evaluation of skin nanocarrier penetration. Fluorescence lifetime
imaging microscopy (FLIM) allows distinguishing the target fluorescence of dye-tagged nanocarriers from skin
autofluorescence, providing a highly sensitive tool for clear-cut localization of the nanocarriers. Cluster-FLIM images
reveal that FITC-labeled tecto-dendrimers penetrate the stratum corneum of human skin ex vivo and reconstructed
human skin but do not cross the tight junction barrier.
Keywords: biomacromolecules; drug delivery system; FLIM; time-resolved fluorescence spectroscopy; tight junctions;
reconstructed human skin
Introduction or are used as labels to track the nanocarriers

themselves. Intensity readings with (confocal) flu-
Nanocarriers have been developed for numerous
orescence microscopy are widely used to obtain
applications in enhanced or targeted drug deliv-
a first insight into the penetration profiles and
ery. In particular, the notoriously low skin pene-
the nanocarrier-modulated uptake of loaded drugs
tration of drugs can be significantly enhanced by
or dyes.4–6 On the basis of these results, novel
nanocarriers,1 which might allow topical treatment
concepts in nanocarrier-modulated drug delivery
and reduction of systemic adverse effects.2,3 How-
will be improved or terminated. Moreover, flu-
ever, stratum corneum (SC) lipids and corneocytes,
orescence lifetime imaging microscopy (FLIM) is
filled with keratin, as well as tight junction proteins
increasingly becoming an important tool in nano-
in viable layers of the epidermis, limit the access to
medical research.7–12 Given that the fluorescence
the drug target site. If nanocarriers overcome these
decay is determined by both the fluorophore and
barriers, their effects on viable cells and their excre-
its microenvironment,13–15 FLIM measurements
tion will be of toxicological concern.
allow a detailed and highly specific insight into
In early phases of nanocarrier development, flu-
nanocarrier–tissue interactions and provide the
orescent dyes serve as surrogates for loaded drugs
doi: 10.1111/nyas.13473
202 Ann. N.Y. Acad. Sci. 1405 (2017) 202–214
C 2017 New York Academy of Sciences.
17496632, 2017, 1, Downloaded from https://nyaspubs.onlinelibrary.wiley.com/doi/10.1111/nyas.13473 by Brunel University, Wiley Online Library on [06/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Volz et al. Tecto-dendrimer penetration in skin
potential to distinguish nanocarrier fluorescence were from Corning (Corning, NY). The anti-
from tissue autofluorescence using novel multivari- body against claudin-1 (1:200, 1C5-D9, cat-no.
ate analysis tools.7 H00009076-M01) was purchased from Novus Bio-
Tecto-dendrimers were developed to overcome logicals (Cambridge, UK), and the Alexa 594–
the limited tolerability of cationic polyamidoamine labeled secondary antibody was from Dianova
dendritic nanocarriers and to increase the number (Hamburg, Germany). Fluorescein isothiocyanate
of cavities for drug transport.16 Tecto-dendrimers (FITC) was obtained from Thermo Fisher Scientific
induce necrosis specifically in melanoma cells, (Waltham, MA). All other chemicals were of the
which might be related to the increased uptake highest purity available.
into SK-Mel-28 cells compared with HaCaT cells.17
Loading methotrexate and zoledronic acid into Tecto-dendrimers
tecto-dendrimers increased the cytotoxicity of Synthesis and labeling. G5G2.5 tecto-denrimers
these anticancer drugs to melanoma cells and (megamer, MG) were synthesized and purified as
keratinocytes.18 described in Schilrreff et al.17 FITC (5 mg/mL) was
The investigation of novel drug delivery systems dissolved in acetone and slowly added to G5G2.5
like tecto-dendrimers requires a human test plat- solutions in phosphate-buffered saline (PBS) (pH
form with in vivo–like properties. High safety stan- 7.4) at various (MG/FITC) molar ratios with
dards in clinical trials cause tremendous efforts in increasing excess of FITC. Subsequently, this solu-
toxicity testing, and, despite comprehensive testing tion was incubated at 4 °C for 24 h in the dark
in the animal, many drug candidates fail because to allow the amino groups of the G5 core to react
of lacking efficacy or unforeseen adverse effects.19 with the FITC. The FITC-labeled tecto-dendrimers
Human skin ex vivo is as limited as animal tests are (MG–FITC) were then separated from free FITC by
of scientific and ethical concern. Three-dimensional size-exclusion chromatography on a Sephadex-G25
skin models (e.g., reconstructed human epidermis fine column (GE Healthcare, Chicago, IL). Elution
and reconstructed human skin (RHS)) from pri- of MG–FITC was facilitated by centrifugation (1000
mary human cells proved useful for testing skin × g) of the loaded column for 1-min runs to obtain
absorption and local effects of small molecules with the individual fractions.
validated test guidelines.20–22 Aiming not only to
assess skin absorption but also to understand the Determination of MG–FITC concentration. MG
molecular mechanisms of skin permeation, RHS and MG–FITC were analyzed by polyacrylamide
features auspicious advantages. RHS is a metaboli- gel electrophoresis (SDS-PAGE; 15% acrylamide,
cally active system,23 considers fibroblast effects on 0.375 M Tris–HCl (pH 8.9), and 0.1% (w/v)
keratinocyte phenotypes,24 and can be transferred SDS). Sample MGs of unknown concentrations and
into more complex test systems, including disease reference MGs17 with known concentrations were
models.25,26 solubilized in PBS (pH 7.4) and 17% glycerol. Elec-
Using FITC-labeled tecto-dendrimers as an trophoresis was performed using a running buffer
example, we investigated penetration of the nano- of 25 mM Tris–HCl (pH 8.8) and 1% (w/v) SDS
carrier into human skin ex vivo as well as into RHS. at a constant voltage of 200 V. After electrophore-
Moreover, we analyzed the pitfalls from nanocar- sis, gels were stained with Coomassie Brilliant
rier labeling and the consequences for the assess- Blue R-250 (AppliChem, Darmstadt, Germany) in
ment for cutaneous uptake of nanocarriers. This 20% methanol and 10% acetic acid. Images of the
technical guide will improve nonclinical studies on destained gels were taken and band intensities quan-
nanocarrier–tissue interactions and might help to tified using ImageJ (National Institute of Health).
unravel the effect of nanocarriers on tight junction The reference bands of the MGs with known par-
proteins. ticle concentration were used to generate a calibra-
tion curve (band intensity versus concentration) by
Materials and methods
a linear regression.17 Using that calibration curve,
Materials the concentration (in mol/L) of the MG–FITC
Collagen I was obtained from Biochrom (Berlin, sample was determined from the respective band
Germany), cell culture inserts (0.4 ␮m pore size) intensities.
Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

C 2017 New York Academy of Sciences. 203
Tecto-dendrimer penetration in skin Volz et al.
Determination of FITC concentration and label- side length of 450 ␮m. FITC and MG–FITC fluores-
ing stoichiometry. Molar concentrations of cence was excited at 488 nm and Alexa Fluor 594
attached FITC molecules were determined by UV– fluorescence at 561 nm using an acousto-optical
visible (UV–Vis) spectroscopy using the Lambert– tunable filter (SELECT UV-VIS, NKT Photonics)
Beer law. The absorbance of FITC at 488 nm (PBS, at 19.5 MHz. Fluorescence emission was spectrally
pH 7.4) and the corresponding extinction coeffi- selected by a long-pass filter with ␭em >515 nm
cient of 65,000 M–1 cm–1 was used. Reference mea- (515LP BrightLine HC, Semrock, Rochester, NY) for
surements were performed at alkaline pH with a FITC and MG–FITC, and with ␭em >575 nm (575LP
fully deprotonated fluorescein and an extinction ET, Chroma, Rockingham, VT) for Alexa Fluor
coefficient of 78,000 ± 7000 M–1 cm–1 (50 mM 594. Emitted photons were collected into 1024 time
potassium phosphate buffer at pH 9),27 and the channels with a channel width of 19.97 picoseconds.
corresponding maximum absorbance of the pH- The instrument response function of the system was
sensitive absorption band was determined. The less than 100 ps full width at half-maximum. FLIM
molar labeling ratio of MG–FITC was then deter- data were analyzed using self-written routines in
mined as the molar ratio FITC/MG. C2+ . Fluorescence decay traces of the individual
pixels were partitioned into classes (i.e., clusters)
Methods for characterization of MG–FITC. using a multivariate pattern-recognition method.
Size and ␨ -potential. Size and ␨ -potential of MG False-color images were generated by assigning
and MG–FITC samples in PBS were determined by a distinct color to all pixels containing a fluo-
dynamic light scattering and phase analysis light rescence decay curve that belonged to a certain
scattering, respectively, using a Nanosizer (ZEN cluster (e.g., autofluorescence versus MG–FITC flu-
3600, Malvern Instruments, Malvern, UK). MG– orescence). Solution FLIM measurements of FITC
FITC shows comparable size to unlabeled MG.17 and MG–FITC were performed using Cellview
UV–Vis spectroscopy. Absorption spectra of dishes (Greiner Bio-One, Kremsmünster, Austria).
FITC and MG–FITC were measured either in a The recorded solution fluorescence decay traces
3 × 3 mm quartz cuvette with a UV2450 (Shimadzu, were fitted using a multiexponential model func-
Kyoto, Japan) or using a NanoDrop One (Thermo tion:
Fisher Scientific) UV–Vis spectrophotometer. The n
absorbance was recorded between 300 and 600 nm I (t) = ␣i e −t/␶i , (1)
with a spectral resolution of 0.5 nm. i
Fluorescence spectroscopy. Fluorescence emission with n the total number of decay components, ␣i the
spectra of FITC and MG–FITC were recorded with amplitude, and ␶ i the fluorescence lifetime of the
a Fluoromax-3 (Horiba Jobin Yvon, Kyoto, Japan) ith component.29 The analysis of the time-resolved
using a 3 × 3 mm quartz cuvette. The temperature quenching data30,31 is based on the mean fluores-
was set to 20 °C, and all samples were excited at cence lifetime ␶ m calculated by
488 nm. The fluorescence emission was recorded

n
␣i ␶i
between 500 and 700 nm with a spectral resolution ␶m = n ␶i . (2)
of 1 nm. i i ␣i ␶i
Time-resolved fluorescence and fluorescence life-
time imaging microscopy. Time-resolved fluores- Skin penetration
cence measurements were conducted in a home- Skin preparation. Human abdominal skin with
made FLIM setup.28 The setup consists of an no visible damages, stretch marks, or scars was
inverted Microscope (IX71, Olympus, Shinjuku, obtained from plastic surgeries with permission and
Japan), a tunable ps-supercontinuum laser (SuperK informed consent from male donors (age: 40–60,
Extreme EUV3, NKT Photonics, Blokken, Den- ethical approval EA4/091/10). After surgery, skin
mark), a confocal scanning unit (DCS120, B&H, was handled according to standardized procedures
Berlin, Germany), a hybrid PMT detector (HPM- to avoid surface contamination with subcutaneous
100-40, B&H) and TCPSC electronics (SPC150, lipids.20 Cryoconserved and thawed human skin was
B&H). The FLIM images were recorded using a 40× dermatomized to 500 ␮m and punched for a surface
Objective (Olympus) resulting in a total field of view area of 2 cm².
204 Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

RHS was grown as described previously.25 Briefly, 594. After washing, the secondary antibody was
we used 0.6 × 106 normal human dermal fibrob- incubated for 1 h at room temperature. The slices
lasts and 3.0 × 106 normal human keratinocytes were finally washed and subjected to FLIM.
per construct, and constructs were cultivated for
3 weeks. Primary normal human keratinocytes and Results
fibroblasts (passage 3, pooled from three donors) Fluorescence intensity–based evaluation
were from therapeutically indicated circumcisions of skin penetration
(ethical approval EA1/081/13) after parents had FITC-labeled tecto-dendrimers (MG–FITC) were
signed the written informed consent. Cell cul- topically applied to human skin ex vivo (Fig. 1A
ture was performed according to standard oper- and B) and RHS (Fig. 1E and F). An aqueous FITC
ating procedures and referred to good cell culture solution served as reference (Fig. 1C and D and G
practice. and H, respectively). The fluorescence intensity of
For skin penetration studies, 30 ␮L/cm2 of FITC alone exceeded the fluorescence intensity of
test formulation (50 ␮mol/L FITC-labeled tecto- MG–FITC in the skin samples. These results indi-
dendrimers in PBS (pH 7.4) or an equimolar FITC cate some penetration into the upper epidermis but
solution in PBS) was applied onto the tissue sur- increased FITC uptake, in both human skin ex vivo
face. To determine the concentration of fluorescein and RHS. Without further characterization of the
and MG–FITC for application, we used a previ- fluorescence properties of MG–FITC, this would
ously published procedure.32 In brief, cryosections lead to the conclusion that only a marginal amount
of tissues were exposed to a series of concentra- of MG–FITC is entering the upper skin compared
tions of MG–FITC, and the fluorescence intensities with FITC alone.
were measured. We obtained a linear correlation
(R² = 0.9784) between concentration and fluores- How labeling stoichiometry of MG–FITC
cence intensity and proved that 5 ␮mol/L MG–FITC affects fluorescence properties
was 10-fold above our detection limit in epifluores- A detailed characterization of MG–FITC as a func-
cence microscopy. Assuming a maximal penetration tion of labeling stoichiometry and tecto-dendrimer
of about 10% of MG–FITC after topical application, concentration was performed using UV–Vis and
we decided to use 50 ␮mol/L MG–FITC for further steady-state and time-resolved fluorescence spec-
experiments. Aqueous FITC solution (50 ␮mol/L) troscopy. To achieve a high FITC labeling yield,
served as control. Human skin ex vivo and RHS FITC was added in varying molar excess to the
were incubated at 37 °C, 5% CO2 for 6 h with the tecto-dendrimers, resulting in labeling stoichiome-
test solutions. Subsequent to removal of excess for- tries ranging from 0.3 to 40 molecules of FITC
mulation with PBS, the skin samples and RHS were per tecto-dendrimer. For the low-labeling sam-
snap frozen and sectioned into 7-␮m slices (Leica ples, a labeling degree of five FITC molecules per
CM 1510S, Wetzlar, Germany). tecto-dendrimer and for the high-labeling samples
20–40 FITC molecules per tecto-dendrimer was
Skin immunostaining. Immunofluorescence determined. Unexpectedly, a higher level of label-
staining for claudin-1 detection was performed ing did not result in higher fluorescence inten-
according to standard protocols. Briefly, tissue sec- sity when compared to equal amounts of FITC
tions were fixed using ice-cold acetone and washed (Fig. 2A–D), indicating quenching of the fluores-
with PBS containing 0.1% bovine serum albumin cence. The quenching factor was determined by
(BSA) and 0.1% Tween 20 (PBS-BSA-Tween). comparison of FITC and MG–FITC fluorescence
Subsequently, skin sections were incubated with (Fig. 2C and D). With regard to visibility of an
goat serum (5% in PBS–BSA–Tween) to block individual tecto-dendrimer and taking the calcu-
unspecific antibody binding. Next, the primary lated quenching factor of 5 (low labeling, Fig. 2C)
anti-claudin-1 antibody was applied to the tissue into account, this would result on average in one
sections and incubated overnight at 4 °C. To emitting fluorophore per tecto-dendrimer for the
avoid any spectral overlap with the absorbance low-labeled samples. However, for the highly labeled
and emission of FITC and MG–FITC, we selected sample, only every second or third tecto-dendrimer
a secondary antibody labeled with Alexa Fluor would fluoresce, based on a quenching factor of 90
Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

MG--FITC FITC
(A) (B) (C) (D)
(E) (F) (G) (H)
Figure 1. Wide-field microscopy of (A, B, E, F) MG–FITC and (C, F, G, H) FITC topically applied on human skin and RHS. (A,
C, E, G) Bright-field images; (B, D, F, H) corresponding epifluorescence images. Scale bar: 200 ␮m. Conditions as described in
Materials and methods.
(high labeling, Fig. 2D) and a labeling degree of 40 (fluorescence lifetime). Since this is an additional
FITC/tecto-dendrimer. Thus, a higher labeling, for means of depopulating the excited state, it thus
the case at hand, leads to less visibility of the car- affects the rate of fluorescence decay (Fig. 2G and
rier and consequently to a worse localization in the H, red curves). For static quenching, the fluo-
micrograph. At the same time, the physicochemi- rophore and the quencher form a nonfluorescent
cal properties of the labeled tecto-dendrimer will be complex, with the number of these complexes tend-
changed owing to high labeling with a pH-sensitive ing to increase with increasing concentration of
(i.e., ionizable) dye. Depending on the environment quencher molecules. For MG–FITC, the concentra-
and pH, the net charge of FITC varies. Consequently, tion of bound fluorescein increases, and thus the
the environment and the labeling degree may affect probability to form dark complexes (ground-state
␨ -potential, electrostatic interactions, and confor- quenching) at higher labeling stoichiometries would
mation or aggregation of the dendrimer and thus increase.36 The nonfluorescent complexes formed
influence skin penetration. The ␨ -potential of unla- in static quenching reduce the observed steady-state
beled tecto-dendrimers was slightly negative at fluorescence yield, but have no effect on the flu-
–4 mV,17 while after FITC labeling this value became orescence decay rates for the remaining noncom-
more negative. Furthermore, pH–titration curves of plexed fluorophores.31 This would yield different
MG–FITC were measured, and the pKa of bound quenching ratios between steady-state and time-
FITC was determined. The pKa increases from 6.5 resolved fluorescence. Higher ground-state quench-
for free FITC in 10 mM salt to pKa 7 for MG– ing corresponds to a higher quenching factor for
FITC with low labeling and to pKa 8 for MG–FITC steady-state fluorescence. From the comparison of
with high labeling stoichiometry. A shift to higher the intensity ratio (Fig. 2E and F) obtained from the
pKa values indicates a more negative electrostatic steady-state emission curves (Fig. 2C and D) with
potential,33,34 , consistent with a higher number of the fluorescence lifetime ratios (Fig. 2G and H), we
negatively charged FITC molecules bound to the conclude that, in the case of the highly labeled sam-
tecto-dendrimer. ple, a strong ground-state quenching of the attached
Factors that could reduce the measured flu- fluorescein occurs. This only takes place when the
orescence intensity are reabsorption (inner filter molecules are in close proximity (several Å) to each
effect35 ), intra- and intermolecular quenching, and other.36 Naturally, this probability increases when
static or dynamic quenching.31 To address the latter, the labeling stoichiometry is higher and more FITC
we determined the fluorescence lifetime of MG– molecules are attached within one tecto-dendrimer,
FITC in relation to the lifetime of FITC. In the as explained above. Figure 3A shows the depen-
case of dynamic quenching, the quencher molecule dence of the quenching factor on the labeling
collides with the fluorophore in the excited state ratio. The data indicate a relative sharp labeling
206 Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

Figure 2. Labeling effects on MG–FITC fluorescence dependent on a high and low labeling ratio. (A, B) Absorbance measurements
of FITC (black) and MG–FITC (red) for a low (A) and high (B) labeling ratio (PBS pH 7.4). (C, D) Emission spectra and quenching
factor of the FITC (black) and MG–FITC (red) samples shown in A and B. (E, F) Comparison of static and dynamic quenching for
high (E) and low (F) labeling ratios, as revealed by steady-state (intensity (I)) and time-resolved fluorescence measurements (mean
fluorescence lifetime (␶ m )). The values obtained for FITC (I0 , ␶ m,0 ) are compared to those obtained from MG–FITC (I,␶ m ). (G, H)
Fluorescence decay traces of FITC (black) and MG–FITC (red) at high (G) and low (H) labeling ratios in PBS (pH 7.4). The mean
fluorescence lifetimes were determined to be 3.95 ns for FITC and 1.44 ns (high labeling) and 1.75 ns (low labeling) for MG–FITC.
ratio–dependent increase in the quenching factor from dilutions of a highly concentrated, highly
above a labeling ratio of 7 FITC/tecto-dendrimer. labeled sample (20 FITC/tecto-dendrimer)
To test for the dependence on tecto-dendrimer and used different concentrations of low-labeled
concentration (i.e., the existence of intermolecular samples (<5 FITC/tecto-dendrimer). The results
quenching), we determined the quenching factor are presented in Figure 3B, where the dilution of
Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

Figure 3. Quenching factors of MG–FITC. (A) Dependence on labeling ratio (FITC/MG). (B) Relationship of MG–FITC concen-
tration and quenching factor. Quenching factors were determined from steady-state fluorescence intensities.
the tecto-dendrimer concentration is expressed via cence intensity for the low-labeled sample, as indi-
the concentration of the attached FITC molecules. cated by quenching factors in the range of about
For the low-labeled sample, no changes of the 0.5–1.6 when compared with MG–FITC in aqueous
quenching factor as a function of tecto-dendrimer solution at pH 7.4.
concentration were found. The quenching factor,
however, increases for the highly labeled sample How background autofluorescence affects
with increasing tecto-dendrimer concentration. evaluation of skin penetration
This leads to the conclusion that, for the highly Intensity-based readings of fluorescence microscopy
labeled samples, not only intramolecular but also might be flawed owing to fluorescence properties of
intermolecular quenching occurs (i.e., the higher the tagged sample and the discrimination of sam-
the tecto-dendrimer concentration, the fewer fluo- ple fluorescence from background autofluorescence.
rescence counts are detected). This behavior renders Figure 5F shows the autofluorescent background
a quantitative analysis impossible. When extrapo- of a skin sample. Background-intensity subtraction
lating to a highly diluted sample, the intersection most likely induces false-positive readings when the
with the y-axis in Figure 3B yields a quenching factor threshold intensity is assumed to be low, making a
of about 18, suggesting that, under highly diluted correct assessment of the penetration behavior diffi-
conditions, one fluorescein per tecto-dendrimer cult. In Figure 5H–J, different degrees of background
also fluoresces for the highly labeled samples. thresholding are presented. Background threshold-
Moreover, the complex environment in biologi- ing was based on the control sample in Figure 5F.
cal tissue might affect the fluorescence intensities of To simulate different degrees of background, the
the label. In skin, the pH profile as well as polarity threshold was gradually increased to more conserva-
and viscosity matter, depending on the location of tive values in the analysis (Fig. 5H–J). Figure 5G and
the fluorescently labeled nanocarrier within the lipid H show the intensity-based localizations of MG–
layers of the SC or within the more hydrophilic envi- FITC in the SC and in lower viable epidermal layers.
ronment in the viable epidermis. Thus, we tested In contrast, a more conservative background thresh-
the effect of these parameters on the FITC fluo- olding (Fig. 5I and J) limits the localization to the
rescence lifetime. Figure 4 shows the fluorescence upper SC.
lifetime curves and quenching analysis for a low- The use of FLIM solves this problem (Fig. 5K–O).
labeled MG–FITC sample as a function of different Since the fluorescence lifetime depends on the elec-
pH values (Fig. 4A and D), in solutions with dif- tronic structure of the fluorescing molecule as well
ferent polarity (Fig. 4B and E), and in solutions as the environment, FLIM allows us to distinguish
with different viscosity (Fig. 4C and F). In strong background fluorescence from target molecule flu-
contrast to the effect of low and high labeling on orescence via the unique fluorescence lifetime
the quenching factor (Fig. 2E–H), the change in of the target molecule.10 This procedure, how-
the environment only marginally affects the fluores- ever, becomes more complicated when a complex
208 Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

Figure 4. Environmental effects on MG–FITC fluorescence quenching. (A–C): I0 and ␶ m,0 were obtained from MG–FITC in PBS
at pH 7.4 and compared to MG–FITC fluorescence intensities and mean lifetimes (I,␶ m ) in different buffer conditions. (D–F)
Environmental effects on the fluorescence decay traces of MG–FITC. The fluorescence decay trace of MG–FITC in PBS (pH 7.4) is
shown as reference. (A, D) Comparison of pH 5.5 and 7.4. (B, E) Polarity effect shown for MG–FITC in a 50% PBS/ethanol (EtOH)
mixture and in an aqueous solution of small unilamellar vesicles (SUV) containing 0.9 mM DMPC. (C, F) Viscosity effect shown
for MG–FITC in PBS/glycerol mixtures with 10% and 50% glycerol.
fluorescence lifetime decay is involved.37 To meet ex vivo and RHS (Fig. 6A and D, red and green).
this challenge, we recently introduced a cluster- Immunofluorescence staining of the tight junction
based multivariate FLIM analysis tool7 that allows protein claudin-1 in skin38 was clearly visualized
us to discriminate background from target fluo- by FLIM. We detected claudin-1 at the keratinocyte
rescence despite complex fluorescence decays and membranes up to the stratum granulosum (Fig. 6B
multiple target fluorescence subspecies. Such sub- and E, yellow), and immunostaining apparently did
species may be induced (e.g., by concentration- not influence the detection of MG–FITC, as seen by
dependent fluorescence quenching) as observed for the comparable results for excised human skin in
the highly labeled MG–FITC. The complex mul- Figures 5 and 6.
tiexponential decay curves of the autofluorescence Although we can localize even small amounts of
background (cyan) and MG–FITC (red) in the skin MG–FITC with high precision and resolution by
as extracted by Cluster-FLIM are shown in Figure 5P. Cluster-FLIM, we did not detect MG–FITC beyond
Figure 5K–O show the Cluster-FLIM images for the tight junction protein barrier in the stratum
both the background autofluorescence (cyan) and granulosum (dashed white lines). While two differ-
the target fluorescence of MG–FITC (red). MG– ent fluorescence decays of MG–FITC in human skin
FITC can be clearly discriminated from autofluo- ex vivo indicate heterogeneous interactions of MG–
rescence, and this is independent of the amount of FITC with the SC (Fig. 6C, green versus red decay
autofluorescence present in the image. Figure 5Q curves), we detected only a single fluorescence life-
and R show a higher magnification of the images in time signature of MG–FITC within RHS (Fig. 6F).
Figure 5D and L.
Discussion
Cluster-FLIM–based evaluation of skin
penetration Nanomedicine describes the successful trans-
As determined by Cluster-FLIM, MG-FITC pene- lation from fundamental nanocarrier research
trates only into the SC but not into viable epider- to novel drug-delivery systems. Among more
mal layers. This is the case for both human skin than 50 nanocarrier-based drugs with marketing
Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

Figure 5. FLIM study of MG–FITC penetration into excised human skin and the effect of background thresholding. (A–E)
Bright-field image of untreated (A) and overlay of the bright-field images of MG–FITC-treated (B–E) human skin with the MG–
FITC localizations from FLIM analysis as shown in L and M. MG–FITC intensities are false-color coded in red on the basis of
fluorescence lifetime cluster analysis. (G–J) Confocal intensities of MG–FITC-treated skin with different background thresholds
((G: 100 cts/px; H: 300 cts/px; I: 1200 cts/px); and J: 2500 cts/px). (F) Confocal intensity of untreated skin. (K–O) False-color–coded
FLIM micrographs: autofluorescence is colored in cyan, and MG–FITC fluorescence is colored in red. (P) Normalized fluorescence
decay traces (red: MG–FITC; cyan: autofluorescence) as extracted with Cluster-FLIM from A–E and K–O. (Q, R) Magnification of
D and L, respectively. Scale bars: 200 ␮m.
authorization, most are based on two approaches surements exist,40 novel nanocarriers must be char-
only: liposomal or polyethylene glycol coating. acterized in detail for their fluorescence properties
Despite advantages like monodisperse nature, tun- after labeling. In most cases, the respective fluo-
able end groups for target recognition, and cleav- rescence label is added in excess to ensure a high
able spacers for triggered drug release, dendrimeric labeling yield. Depending on the nanocarrier, this
structures are rarely translated from bench to might lead to complications when aiming for an
bedside.39 assessment of the penetration profile and a (semi-)
To improve the nonclinical evaluation of novel quantitative evaluation, as the high labeling degree
concepts in drug delivery, we demonstrated the does not automatically result in higher photon emis-
impact of advanced techniques like Cluster-FLIM sion and thus higher visibility of the nanocarrier.
on the readout of nanocarrier effects. First studies Additionally, intermolecular quenching effects have
on nanocarriers frequently aim to track them and to be taken into account, as well as the depen-
their cargo to the target site. Fluorescence label- dence on environmental factors, such as pH, polar-
ing appears to be as easy and quick, as fluorescence ity, or viscosity. In this study, we determined the
microscopy provides results on nanocarrier uptake. best labeling stoichiometry to ensure the visibility
Although guides for standardized fluorescence mea- and linear concentration dependence (Figs. 2 and 3,
210 Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

Figure 6. Comparison of MG–FITC penetration and localization of the tight junction protein claudin-1 in human epidermis. (A,
D) Overlay image of MG–FITC-treated human skin and RHS. White dashed line indicates the border/barrier of the SC and viable
epidermis. (B) False-color–coded FLIM micrograph of the sample shown in A. Claudin-1 is stained in yellow and MG–FITC in green
and red. (E) FLIM micrographs of the sample shown in D with claudin-1-staining in yellow and MG–FITC in red. (C) Normalized
fluorescence decay traces of MG–FITC as extracted with Cluster-FLIM from B. (F) Comparison of MG–FITC fluorescence decays
(red traces) in human skin and RHS (extracted from E). Scale bars: 50 ␮m; SC, stratum corneum; SG, stratum granulosum; SSp,
stratum spinosum; D, dermis.
low-labeled MG–FITC) for quantitative evaluation samples, induces additional variation. Tissue blocks
in skin. In contrast to the high impact of the are cut to slices of a few micrometers, but cutting
labeling stoichiometry on fluorescence quenching blocks into slices of exactly the same thickness
(Fig. 3), changes in the environment, such as the requires a very well-maintained cryotome and years
pH, only marginally affect the fluorescence quench- of experience in cutting. Variation in slice thickness
ing factor when using the low-labeled MG–FITC causes variation in autofluorescence within the
sample (Fig. 4A and D). In summary, only a careful same study. Taken together, setting the threshold
analysis of the fluorescence properties of the fluo- of fluorescence exposure time to exclude autofluo-
rescently tagged nanocarrier will allow for a quan- rescence remains arbitrary and prone to mistakes.
titative evaluation of the penetration behavior into Depending on the bleaching of the fluorescent
the skin. Otherwise, the various quenching effects dye, its intensity will be reduced and false-negative
may mask the real localization and concentration of results might occur. If the background autofluo-
the nanocarrier in skin tissue. rescence is underestimated, as shown in Figure 5G
Furthermore, cutaneous autofluorescence limits and H, false-positive results will occur. Again using
the sensitivity of intensity-based fluorescence. Since the FITC–tecto-dendrimers as an example, we
cutaneous autofluorescence depends on the skin highlight the importance of correct background
color and the body site of the skin,41 the limit of subtraction that is not entirely possible using
detection varies from skin sample to skin sample. intensity-based fluorescence microscopy. FLIM
Moreover, snap freezing or paraffin embedding, provides a tool to meet these challenges and to allow
standard approaches in the analysis of tissue distinguishing nanocarrier fluorescence from tissue
Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

autofluorescence, as shown in Figure 5L–O. can be explained by differences in SC lipid composi-

Although FLIM is a highly specialized microscopic tion and/or packing in RHS. Nevertheless, RHS cor-
technique and not readily accessible, the number rectly predicted the penetration of tecto-dendrimers
of publications using FLIM as an analytical tool is into the SC without penetration into the viable epi-
increasing, indicating the impact of this technique dermis. To the best of our knowledge, this is the first
in skin research.7,9,42–44 FLIM in combination with study investigating the cutaneous uptake of tecto-
confocal laser scanning microscopy (CLSM) adds dendrimers into RHS. While the predictive power
to the advantages of CLSM over conventional of reconstructed human epidermis for the absorp-
fluorescence, imaging the fluorescence lifetime tion of small molecules has been validated,20 further
information. Single-photon counting–based FLIM, studies with skin disease models will require RHS
however, requires longer exposure times than with epidermal–dermal interactions. In this study,
steady-state CLSM, owing to the reduced excitation we proved the utility of RHS for the investigation of
energies necessary for single-photon counting tecto-dendrimer penetration.
and the need of a certain number of photons for In conclusion, the Cluster-FLIM–based readout
fluorescence lifetime analyses. New and advanced improved the accuracy of nonclinical research dur-
analysis tools, as shown here with our Cluster-FLIM ing the early development of novel nanocarriers and
method, enable accurate fluorescence background might provide molecular insights into nanocarrier–
subtraction, even at low photon counts. tissue interactions. Testing nanocarriers with RHS
Employing Cluster-FLIM, tecto-dendrimers were overcomes the shortage of human skin ex vivo and
exclusively found within the SC of human skin ex promotes the replacement, reduction, and refine-
vivo and RHS and not within viable layers. This ment (3R principle) of animal tests.
result is consistent with the limited permeation of
macromolecules through tight junction proteins.45 Acknowledgments
The absence of tecto-dendrimers in viable epider- This work was financially supported by the Ger-
mal layers limits their drug-like use as described for man Research Foundation (Collaborative Research
melanoma cells.17 On the other hand, it reduces the Center 1112, Project B03 to U.A. and C02 to
toxicological concern of a topical nanocarrier appli- M.S.K.). The authors also would like to acknowledge
cation, since they access neither viable layers nor the Helmholtz Association through the Helmholtz
the systemic circulation. However, diseases might Virtual Institute ‘‘Multifunctional Biomaterials for
change the skin barrier function, which could enable Medicine” (Kantstr. 55, 14513 Teltow, Germany).
tecto-dendrimer penetration into viable layers and Priscila Schilrreff thanks Freie Universität for a Flex-
require further studies. Among other conditions, ible Funds travel grant. Priscila Schilrreff and Maria
nonmelanoma skin cancer alters the expression of Jose Morilla are members of the Research Career
tight junction proteins.46 Finally, a controlled inter- Program from the National Council for Scientific
action between tight junction proteins and nanocar- and Technological Research (CONICET).
riers would allow targeted delivery.
To obtain insight into the molecular interactions Competing interests
of tecto-dendrimers with cutaneous lipids and pro-
The authors declare no competing interests.
teins, FLIM analysis provides additional informa-
tion on the microenvironment of the nanocarrier. References
Concerning MG–FITC, changes in the pH value
or polarity alter the fluorescence lifetime, whereas 1. Vogt, A., C. Wischke, A.T. Neffe, et al. 2016. Nanocarriers
for drug delivery into and through the skin—do existing
a slightly increased viscosity does not affect the technologies match clinical challenges? J. Control. Release
fluorescence lifetime (fluorescence decay traces in 242: 3–15.
Fig. 4D–F). This might aid further nanocarrier stud- 2. Korting, H.C. & M. Schäfer-Korting. 2010. Carriers in the
ies on cellular or direct protein interactions to inves- topical treatment of skin disease. Handb. Exp. Pharmacol.
tigate the above-mentioned opening mechanism for 2010: 435–468.
3. Yamamoto, K., A. Klossek, R. Flesch, et al. 2016.
tight junctions. Core–multishell nanocarriers: transport and release of
The additional fluorescence lifetime signature in dexamethasone probed by soft X-ray spectromicroscopy. J.
human skin ex vivo (Fig. 6C) compared with RHS Control. Release 242: 64–70.
212 Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

4. Alnasif, N., C. Zoschke, E. Fleige, et al. 2014. Penetration 20. Schäfer-Korting, M., U. Bock, W. Diembeck, et al. 2008. The
of normal, damaged and diseased skin—an in vitro study use of reconstructed human epidermis for skin absorption
on dendritic core–multishell nanotransporters. J. Control. testing: results of the validation study. Altern. Lab. Anim. 36:
Release 185: 45–50. 161–187.
5. Campbell, C.S.J., L.R. Contreras-Rojas, M.B. Delgado- 21. OECD 428. 2004. Test No. 428: skin absorption: in vitro
Charro, et al. 2012. Objective assessment of nanoparticle method. In OECD Guidelines for the Testing of Chemicals,
disposition in mammalian skin after topical exposure. J. Section 4: Health Effects. OECD Publishing. Accessed August
Control. Release 162: 201–207. 17, 2017. https://doi.org/10.1787/9789264071087-en.
6. Prow, T.W., N.A. Monteiro-Riviere, A.O. Inman, et al. 2012. 22. OECD 439. 2015. Test No. 439: in vitro skin irritation. In
Quantum dot penetration into viable human skin. Nanotox- OECD Guidelines for the Testing of Chemicals, Section 4:
icology 6: 173–185. Health Effects. OECD Publishing. Accessed August 17, 2017.
7. Alexiev, U., P. Volz, A. Boreham, et al. 2017. Time-resolved https://doi.org/10.1787/9789264242845-en.
fluorescence microscopy (FLIM) as an analytical tool in skin 23. Bätz, F.M., W. Klipper, H.C. Korting, et al. 2013. Esterase
nanomedicine. Eur. J. Pharm. Biopharm. 116: 111–124. activity in excised and reconstructed human skin—
8. Leite-Silva, V.R., M. Le Lamer, W.Y. Sanchez, et al. 2013. biotransformation of prednicarbate and the model dye
The effect of formulation on the penetration of coated and fluorescein diacetate. Eur. J. Pharm. Biopharm. 84: 374–
uncoated zinc oxide nanoparticles into the viable epidermis 385.
of human skin in vivo. Eur. J. Pharm. Biopharm. 84: 297– 24. Berroth, A., J. Kuhnl, N. Kurschat, et al. 2013. Role of fibrob-
308. lasts in the pathogenesis of atopic dermatitis. J. Allergy Clin.
9. Zhu, Y., C.-S. Choe, S. Ahlberg, et al. 2015. Penetration of Immunol. 131: 1547–1554.
silver nanoparticles into porcine skin ex vivo using fluores- 25. Zoschke, C., M. Ulrich, M. Sochorová, et al. 2016. The barrier
cence lifetime imaging microscopy, Raman microscopy, and function of organotypic non-melanoma skin cancer models.
surface-enhanced Raman scattering microscopy. J. Biomed. J. Control. Release 233: 10–18.
Opt. 20: 051006. 26. Hönzke, S., C. Gerecke, A. Elpelt, et al. 2016. Tailored den-
10. Boreham, A., T.-Y. Kim, V. Spahn, et al. 2011. Exploiting flu- dritic core–multishell nanocarriers for efficient dermal drug
orescence lifetime plasticity in FLIM: target molecule local- delivery: a systematic top-down approach from synthesis to
ization in cells and tissues. ACS Med. Chem. Lett. 2: 724–728. preclinical testing. J. Control. Release 242: 50–63.
11. Huck, V., C. Gorzelanny, K. Thomas, et al. 2016. From mor- 27. Molecular Probes, Inc. 2015. Certificate of analysis. 2017.
phology to biochemical state—intravital multiphoton fluo- 28. Boreham, A., R. Brodwolf, K. Walker, et al. 2016. Time-
rescence lifetime imaging of inflamed human skin. Sci. Rep. resolved fluorescence spectroscopy and fluorescence life-
6: https://doi.org/10.1038/srep22789. time imaging microscopy for characterization of dendritic
12. Roberts, M.S., Y. Dancik, T.W. Prow, et al. 2011. Non- polymer nanoparticles and applications in nanomedicine.
invasive imaging of skin physiology and percutaneous pen- Molecules 22: E17.
etration using fluorescence spectral and lifetime imaging
29. Kim, T.Y., K. Winkler & U. Alexiev. 2007. Picosecond multi-
with multiphoton and confocal microscopy. Eur. J. Pharm.
dimensional fluorescence spectroscopy: a tool to measure
Biopharm. 77: 469–488.
real-time protein dynamics during function. Photochem.
13. Boreham, A., M. Pfaff, E. Fleige, et al. 2014. Nanodynamics of
Photobiol. 83: 378–384.
dendritic core–multishell nanocarriers. Langmuir 30: 1686–
30. Volz, P., N. Krause, J. Balke, et al. 2016. Light and pH-
1695.
induced changes in structure and accessibility of trans-
14. Boreham, A., J. Pikkemaat, P. Volz, et al. 2016. Detecting
membrane helix B and its immediate environment in
and quantifying biomolecular interactions of a dendritic
channelrhodopsin-2. J. Biol. Chem. 291: 17382–17393.
polyglycerol sulfate nanoparticle using fluorescence lifetime
31. Alexiev, U. & D.L. Farrens. 2014. Fluorescence spectroscopy
measurements. Molecules 21: E22.
of rhodopsins: insights and approaches. Biochim. Biophys.
15. Marcu, L. 2012. Fluorescence lifetime techniques in medical
Acta 1837: 694–709.
applications. Ann. Biomed. Eng. 40: 304–331.
32. Lombardi Borgia, S., M. Regehly, R. Sivaramakrishnan,
16. Tomalia, D.A., S. Uppuluri, D.R. Swanson, et al. 2000.
et al. 2005. Lipid nanoparticles for skin penetration
Dendrimers as reactive modules for the synthesis of new
enhancement-correlation to drug localization within the
structure-controlled, higher-complexity megamers. Pure
particle matrix as determined by fluorescence and parelec-
Appl. Chem. 72: 2343.
tric spectroscopy. J. Control. Release 110: 151–163.
17. Schilrreff, P., C. Mundina-Weilenmann, E.L. Romero, et al.
2012. Selective cytotoxicity of PAMAM G5 core–PAMAM 33. Alexiev, U., P. Scherrer, T. Marti, et al. 1995. Time-resolved
G2.5 shell tecto-dendrimers on melanoma cells. Int. J. surface charge change on the cytoplasmic side of bacteri-
Nanomed. 7: 4121–4133. orhodopsin. FEBS Lett. 373: 81–84.
18. Schilrreff, P., G. Cervini, E.L. Romero, et al. 2014. Enhanced 34. Möller, M. & U. Alexiev. 2009. Surface charge changes upon
antimelanoma activity of methotrexate and zoledronic acid formation of the signaling state in visual rhodopsin. Pho-
within polymeric sandwiches. Colloids Surf. B Biointerfaces tochem. Photobiol. 85: 501–508.
122: 19–29. 35. Fonin, A.V., A.I. Sulatskaya, I.M. Kuznetsova, et al. 2014. Flu-
19. Arrowsmith, J. 2011. Trial watch: phase II failures: 2008– orescence of dyes in solutions with high absorbance. Inner
2010. Nat. Rev. Drug Discov. 10: 328–329. filter effect correction. PLoS One 9: e103878.
Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

36. Runnels, L.W. & S.F. Scarlata. 1995. Theory and application 42. Ostrowski, A., D. Nordmeyer, A. Boreham, et al. 2014. Skin
of fluorescence homotransfer to melittin oligomerization. barrier disruptions in tape stripped and allergic dermatitis
Biophys. J. 69: 1569–1583. models have no effect on dermal penetration and systemic
37. Boreham, A., R. Brodwolf, M. Pfaff, et al. 2014. Temper- distribution of AHAPS-functionalized silica nanoparticles.
ature and environment dependent dynamic properties of a Nanomedicine 10: 1571–1581.
dendritic polyglycerol sulfate. Polym. Adv. Technol. 25: 1329– 43. Witting, M., A. Boreham, R. Brodwolf, et al. 2015. Interac-
1336. tions of hyaluronic acid with the skin and implications for
38. Kirschner, N. & J.M. Brandner. 2012. Barriers and more: the dermal delivery of biomacromolecules. Mol. Pharm. 12:
functions of tight junction proteins in the skin. Ann. N.Y. 1391–1401.
Acad. Sci. 1257: 158–166. 44. Liu, D.C., A.P. Raphael, D. Sundh, et al. 2012. The
39. Caster, J.M., A.N. Patel, T. Zhang, et al. 2017. Investiga- human stratum corneum prevents small gold nanoparti-
tional nanomedicines in 2016: a review of nanotherapeutics cle penetration and their potential toxic metabolic conse-
currently undergoing clinical trials. Wiley Interdiscip. Rev. quences. J. Nanomater. 2012. https://doi.org/10.1155/2012/
Nanomed. Nanobiotechnol. 9: e1416. 721706.
40. Resch-Genger, U., K. Hoffmann & A. Hoffmann. 2008. 45. Kirschner, N., R. Rosenthal, M. Furuse, et al. 2013. Con-
Standardization of fluorescence measurements: criteria for tribution of tight junction proteins to ion, macromolecule,
the choice of suitable standards and approaches to fit-for- and water barrier in keratinocytes. J. Invest. Dermatol. 133:
purpose calibration tools. Ann. N.Y. Acad. Sci. 1130: 35–43. 1161–1169.
41. Czekalla, C., K.H. Schönborn, N. Döge, et al. 2017. Body 46. Rachow, S., M. Zorn-Kruppa, U. Ohnemus, et al. 2013.
regions have an impact on the collagen/elastin index of the Occludin is involved in adhesion, apoptosis, differentiation
skin measured by non-invasive in vivo vertical two photon and Ca2+ -homeostasis of human keratinocytes: implica-
microscopy. Exp. Dermatol. 26: 822–824. tions for tumorigenesis. PLoS One 8: e55116.
214 Ann. N.Y. Acad. Sci. 1405 (2017) 202–214


Annals of The New York Academy of Sciences - 2017 - Volz - Pitfalls in Using Fluorescence Tagging of Nanomaterials

Uploaded by

Copyright:

Available Formats

Annals of The New York Academy of Sciences - 2017 - Volz - Pitfalls in Using Fluorescence Tagging of Nanomaterials

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Annals of The New York Academy of Sciences - 2017 - Volz - Pitfalls in Using Fluorescence Tagging of Nanomaterials

Uploaded by

Copyright:

Available Formats

Ann. N.Y. Acad. Sci.

Pitfalls in using fluorescence tagging of nanomaterials:

Introduction or are used as labels to track the nanocarriers

Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

204 Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

(E) (F) (G) (H)

206 Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

208 Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

210 Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

autofluorescence, as shown in Figure 5L–O. can be explained by differences in SC lipid composi-

212 Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

214 Ann. N.Y. Acad. Sci. 1405 (2017) 202–214

You might also like