Access to this full-text is provided by Springer Nature.
Content available from Scientific Reports
This content is subject to copyright. Terms and conditions apply.
1
Scientific REPORTS | (2018) 8:14101 | DOI:10.1038/s41598-018-32314-4
www.nature.com/scientificreports
Separation of blood microsamples
by exploiting sedimentation at the
microscale
D. Forchelet1, S. Béguin2, T. Sajic3, N. Bararpour4, Z. Pataky5, M. Frias6,7, S. Grabherr4,
M. Augsburger4, Y. Liu3,8, M. Charnley9, J. Déglon4, R. Aebersold
3,10, A. Thomas
4,11 &
P. Renaud1
Microsample analysis is highly benecial in blood-based testing where cutting-edge bioanalytical
technologies enable the analysis of volumes down to a few tens of microliters. Despite the availability of
analytical methods, the diculty in obtaining high-quality and standardized microsamples at the point
of collection remains a major limitation of the process. Here, we detail and model a blood separation
principle which exploits discrete viscosity dierences caused by blood particle sedimentation in a
laminar ow. Based on this phenomenon, we developed a portable capillary-driven microuidic device
that separates blood microsamples collected from nger-pricks and delivers 2 µL of metered serum for
bench-top analysis. Flow cytometric analysis demonstrated the high purity of generated microsamples.
Proteomic and metabolomic analyses of the microsamples of 283 proteins and 1351 metabolite features
was consistent with samples generated via a conventional centrifugation method. These results were
conrmed by a clinical study scrutinising 8 blood markers in obese patients.
Blood sample separation, consisting in the extraction and isolation of the liquid surrounding blood cells, is the
most common preparation operation performed before clinical biochemistry analysis. With recent advancement
of bioanalytical technologies in term of sensitivity and selectivity, the question of the minimal blood volume
required for biochemical analyses has become more central, indicating the need for microsampling solutions1.
Microsampling with typical volumes of 10–100 µL allows less invasive, more frequent and more convenient blood
sampling. For preclinical trials on small animals and populations at risk (such as newborns or polymedicated
patients), microsample analysis drastically increases subject welfare. Most common microsystems for blood
separation traditionally rely on sedimentation2,3, microltration4,5 or cell deviation6,7 in a microuidic chip8
(see Supplementary Note 1). However, typically, such systems require either sample pre-dilution, have com-
plex designs or suer from low extraction yields9. Additionally, o-chip sample retrieval is typically not per-
formed as those systems are destined to be integrated in single device as micro total analysis systems (i.e. labs
on chip). In this paper we report a device developed for sample preparation at the point of collection (SP-POC)
that addresses the need for a blood microsample separation device that generates standardized and stabilized
1Microsystems Laboratory (LMIS4), School of Engineering (STI), École Polytechnique Fédérale de Lausanne (EPFL),
Lausanne, CH, 1015, Switzerland. 2ARC Training Centre in Biodevices, Faculty of Science, Engineering and Technology,
Swinburne University of Technology, Hawthorn, VIC, 3122, Australia. 3Department of Biology, Institute of Molecular
Systems Biology, ETH Zurich, Zurich, CH, 8093, Switzerland. 4Unit of Toxicology, CURML, Lausanne University
Hospital, Geneva University Hospitals, rue Michel-Servet 1, Geneva, CH, 1211, Switzerland. 5Service of Therapeutic
Education for Chronic Diseases, WHO Collaborating Centre, Geneva University Hospitals, University of Geneva, rue
Gabrielle-Perret-Gentil 4, Geneva, CH, 1205, Switzerland. 6Division of Laboratory Medicine, Department of Genetics
and Laboratory Medicine, Geneva University Hospitals, rue Gabrielle-Perret-Gentil 4, Geneva, CH, 1205, Switzerland.
7Division of Endocrinology, Diabetes, Hypertension and Nutrition, Department of Internal Medicine Specialities,
Faculty of Medicine, University of Geneva, rue Gabrielle-Perret-Gentil 4, Geneva, CH, 1205, Switzerland. 8Department
of Pharmacology, Cancer Biology Institute, Yale University School of Medicine, West Haven, CT, 06516, USA. 9Centre
for Micro-Photonics, Faculty of Science, Engineering and Technology, Swinburne University of Technology, Hawthorn,
VIC, 3122, Australia. 10Faculty of Science, University of Zurich, Zurich, CH, 8006, Switzerland. 11Faculty of Biology and
Medicine, University of Lausanne, Vulliette 04, Lausanne, CH, 1000, Switzerland. D. Forchelet and S. Béguin contributed
equally.A. Thomas and P. Renaud jointly supervised this work. Correspondence and requests for materials should be
addressed to D.F. (email: david.forchelet@ep.ch)
Received: 25 June 2018
Accepted: 3 September 2018
Published: xx xx xxxx
OPEN
Content courtesy of Springer Nature, terms of use apply. Rights reserved
www.nature.com/scientificreports/
2
Scientific REPORTS | (2018) 8:14101 | DOI:10.1038/s41598-018-32314-4
microsamples10. e device allows separation of 25 µL undiluted blood samples in a complete passive manner
requiring no external equipment: both, separation and pumping are passive. Whole blood is separated into liquid
and residual cells using a separation principle based on sedimentation-induced viscosity dierences and simulta-
neous capillary-driven laminar ow. Volume-metered ejection of the separated samples allows o-chip retrieval
of 2 µL processed samples for subsequent analysis as illustrated in Fig.1a. e microuidic process (modeled in
Supplementary Note 2) leads to a low cellular contamination as characterized using ow cytometry. Mass spec-
trometry proteomics and metabolomics proling, as well as standard clinical chemistry analysis, were performed
on the chip-separated samples to demonstrate their quality and their suitability for gold standard bench-top
analysis methods.
Figure 1. Separation microdevice and uidic behaviour (a) Illustration of the study’s purpose: generation
of analytically relevant cell-free blood microsample from ngerprick; (le) sampling of capillary blood aer
ngerprick and device loading; (center) sequence of microscopic captures depicting the formation of a cell-free
plug at the air-liquid interface; (right) retrieval allowing o-chip gold standard analyses; (b) Device structure
containing two areas performing the main functions: separation and ejection; (c) Illustration of the separation
principle showing the expected velocity u distribution due to the distribution of cellular volume fraction φ; (d)
Typical extracted volume curve in time for 0.08 and 0.25 μl/min whole blood feeding rate; (e) Comparison of
anticoagulated and fresh sample yields showing a strong increase in fresh blood extraction yield (N = 23 and
N = 13 for anti-coagulated and fresh blood extraction yield respectively); (f) Ejection mechanism: air injection
allows the ejection of a 2 μL liquid sample from the metering area. e volume denition is performed by the
capillary valves present in the channel and at the outlet; x, y and z represent the channel longitudinal, transversal
and vertical directions respectively.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
www.nature.com/scientificreports/
3
Scientific REPORTS | (2018) 8:14101 | DOI:10.1038/s41598-018-32314-4
Results
A passive device for blood separation and volume metering. e device consists of a microuidic
system containing a separation area and an ejection area, as illustrated in Fig.1b. It is constituted of two poly-
dimethylsiloxane (PDMS) parts assembled together (see Supplementary Fig.1), one of which is modied with an
additive to allow adequate capillary pressure, and eectively triggers spontaneous ow in the device. Each area
executes a specic function: separation of blood samples and ejection of a volume-metered output cell-free sam-
ple, respectively. While owing in the separation area cells present in the sample, under the action of gravitational
forces, sediment towards the bottom of the channel where neither trenches nor complex structures are present11.
e formed sediment is at a high volume fraction (φ) of cells and is of higher viscosity than the cell-free super-
natant12 (see Supplementary Note2 and Supplementary Fig.2). e simultaneous pumping of the liquid and the
generation of the dierent sediment phases creates a velocity dierence between the lower viscosity supernatant
and the higher viscosity sediment13 as illustrated in Fig.1c. e extraction of plasma from anticoagulated blood
(resp. serum in case of untreated whole blood) starts aer a separation delay, necessary to establish through sedi-
mentation a viscosity contrast sucient to mediate separation as shown in Fig.1d. e duration of the separation
delay is dependent on cell sedimentation speed and distance: the process is thus inuenced by blood parameters
(e.g. hematocrit, cell size) and design parameters (channel height) but is independent of whole blood feeding
owrate (experimental Pearson’s r = 0.43, N = 20). Upon reaching sucient height, the clear supernatant starts
owing faster than the sediment and a plug of clear liquid appears at the air-liquid interface: a spatial separation of
a plasma (resp. serum) plug occurs along the channel length. is separation along the channel length allows easy
manipulation in standard planar technology microuidic systems. is novel separation mechanism is modeled
in Supplementary Note 2 and allows passive separation of whole blood microsamples in a simple device using
spontaneous separation phenomena. e separation phenomenon happens in the bulk of the sample and thus
allows great exibility with system design and material choices.
e separation delay, independent of ow rate, was determined to be 400 s ± 148 s (N = 20) and 430 s ± 88 s
(N = 13) for anti-coagulated samples and fresh samples respectively (Supplementary Fig.3). During the subse-
quent separated liquid extraction phase, the cell-free plug grows continuously as whole blood enters the system:
the sediment still progresses in the chip during extraction, however the air-liquid interface progresses at a faster
rate. To characterize the separation phenomenon, experiments at constant ow rates were devised with an exter-
nal uidic control and anticoagulated blood samples. As shown in Fig.1d, the extraction rate is constant during
the extraction phase and depends on the imposed feeding owrate. e extraction rate is linearly correlated with
the feeding owrate (adjusted-R2 = 0.96, N = 23) and the yield (see Materials and Methods) achieved is 17%
(N = 23; Fig.1e). e quantity of plasma (respectively serum) separated directly depends on the volume of blood
loaded and relates to the maximum respective yields reported. e geometry of the separation and metering
channels have an impact on the total separation time, maximum volume that ca be processed, and total volume
separated: wider and longer channel designs yield more separated sample volume, and longer channels impose
more total separation time. To characterize the fresh blood separation mechanism, the devices were run using the
capillary pressure as the driving mechanism. ese conditions recreate eld operation conditions, where capillary
pumping implies non-constant owrates and clotting drastically changes the sample viscosity. For fresh samples,
coagulation leads to a yield increase to 67% (N = 13; Fig.1e), due to an additional ltration through a formed clot
(see Supplementary Note 3).
e integrated sample ejection mechanism illustrated in Fig.1f allows the o-chip retrieval of the separated
sample in a form that is immediately compatible with a variety of downstream gold standard analyses such as ow
cytometry, immunoassays or mass spectrometry. e sample volume is restricted to 2 µL by two capillary valves
placed on each side of a volume metering area: an in-channel capillary valve and the open outlet. During channel
lling, the in-channel capillary valve acts as a delay valve: the liquid front stops an instant upon lling the capillary
valve area before spontaneously starting to ll the metering area. Upon complete lling of the metering area, ow
in the device stops spontaneously, when reaching the outlet, thereby simplifying sample ejection and increasing
the robustness of separation timing. e sample ejection is activated by collapsing a cavity in the chip through
external mechanical pressure: the air contained in the cavity (>2 µL) is injected into the channel and drives the liq-
uid movement. At the in-channel capillary valve, an air-liquid interface is created, and surface tension prevents air
from traveling towards the inlet. e air drives the liquid content of the metering to the outlet. us, the volume
ejected precisely corresponds to the volume contained between the in-channel valve and the outlet. e ejection
process is completed when the separated sample collected in the metering area is transferred outside the system
in the form of a 2 µL drop of separated blood. Excess air is released outside the chip aer formation of the drop.
Sample purity. To assess the purity of the chip-separated samples, cell contamination was determined using
ow cytometry. Particle size-based distribution was analyzed using forward scattering (FSC) data as shown in
Supplementary Fig.4. Data was compared between device-separated samples and centrifuged samples from cap-
illary whole blood. Typical normalized histograms of FSC values for each sample type is shown in Fig.2a (le).
ese results show that the chip-separated samples are marginally contaminated by particles smaller than RBC,
represented by the strongly populated range between 75 K and 175 K in the whole blood histograms. As illustrated
in Fig.2a (right), whole blood samples concentration was 3.6·106 particles/µL (coecient of variation, CV = 17%).
For conventional plasma and serum samples, particle concentration was 1.1·103 particles/µL (CV = 48%) and
8.7·102 particles/µL (CV = 32%), respectively. Chip-separated samples contain a signicantly lower particle con-
centration with 4.4 102 particles/µL (CV = 17%) compared to a conventional separation method. Plasma and serum
do not yield signicantly dierent particle counts. Purity (see Materials and Methods) achieved for chip-separated
samples was 99.987%. In comparison, lower purities were obtained for the reference samples (99.968% for plasma,
and 99.976% for serum samples). us, this experiment shows that the microdevice separates blood cellular com-
ponents with a higher repeatability and a higher purity upon comparison with reference samples.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
www.nature.com/scientificreports/
4
Scientific REPORTS | (2018) 8:14101 | DOI:10.1038/s41598-018-32314-4
Figure 2. Analytical comparison of three blood separation methods (a) Blood samples particle content was
analyzed by ow cytometry; (le) normalized event counts vs FSC showing the narrow FSC distribution of
events in separated samples vs whole blood; (right) event count per microliter of original samples (N = 5)
showing a signicantly lower number of particles in chip-separated samples compared to centrifuged plasma
or serum samples; (b) Proteomic data measured in the capillary blood of healthy volunteer; (le) boxplot
presenting coecient of variations calculated for 43 FDA approved blood biomarkers, in generated cell-free
blood samples (N = 5) by plasma, serum or chip-based method (right) non-supervised hierarchical clustering
of samples (N = 5) from three blood separation methods based on 283 quantied proteins; (c) Metabolomic
data measured in the capillary blood of healthy volunteer. Non-supervised hierarchical clustering of samples
(N = 3) from three blood separation methods based on 1351 metabolic features. Asterisks indicate statistically
signicant dierences among the tested groups and corresponds to the p-value adjusted for the multiple
comparisons: *P = <0.05, **P = <0.01, ***P = <0.001. N corresponds to the number of analytical repetition.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
www.nature.com/scientificreports/
5
Scientific REPORTS | (2018) 8:14101 | DOI:10.1038/s41598-018-32314-4
Biochemical sample content. Liquid blood samples generated with reference methods and chip separa-
tion method were compared for proteome content by using a bottom-up proteomic workow based on SWATH
mass spectrometry (SWATH-MS)14. For the generation of plasma, serum and chip-separated liquid, whole blood
originated from nger-prick capillary sampling on a single donor (young male adult) was used. e generation of
liquid samples for each of three blood preparation methods was repeated ve times. In total, 15 biological samples
were measured, and data metrics were compared between chip-separated liquid, plasma and serum samples. e
proteomic analysis identied 312 unique proteins (below 1% of protein FDR) of which 284 proteins were consist-
ently quantied in at least 30% of measured data15. Notably, among 284 quantied proteins, 43 have already been
approved as plasma biomarkers for various diseases by the Food and Drug Administration (FDA)16. To estimate
the protein variability between the three dierent blood separation methods, the coecients of variation (CVs)
of 43 FDA approved proteins were computed. CVs were calculated based on raw protein intensities between
ve replicates of each method. e protein Beta-2-Microglobulin (B2M) showed generally high variability in all
three-sample type. Besides B2M, the coecient of variation of the FDA biomarkers for conventional plasma and
serum data ranged between 3.5 and 30% and did not exceed 22% for the chip-separated liquid (Fig.2b le). By
using one-way ANOVA test, protein variability on chip-separated samples was shown to be signicantly lower
than for conventionally prepared plasma and serum samples. e dierences were statistically signicant in both
comparisons against plasma (adjusted p-value = 0.003) and also against serum (adjusted p-value < 0.001), with
median CV of chip-separated liquid of 7.5% compared to 10.9% and 12.2% for plasma and serum, respectively
(Fig.2b le).
en, hierarchical cluster analysis was used to group the samples according to the similarity of their quantita-
tive protein proles. e patterns generated from the 15 blood samples revealed clear division of three dierent
types of blood liquid separations as shown in Fig.2b (right). Five replicates per sample type were always adjacent
to each other, showing an excellent reproducibility for each of the three blood separations. However, two main
clusters on the heat-map separated plasma samples from the two closest sub-clusters of serum and chip separated
devices. e data indicate the high proteome similarity of chip-separated liquid and conventionally generated
blood serum based on the quantitative protein proles. Remarkably, the subset of proteins that mainly dier in
abundance levels in the plasma compared to other serum and chip-separated liquid, were related to blood clotting
cascade, such as platelet factors or brinogen chains (Supplementary Fig.5).
e proteomic characterization of the generated samples was conrmed by untargeted metabolomic analyses
of the same samples. Aer matching against known metabolites in the human metabolome database (HMDB,
http://www.hmdb.ca) using the mass to charge ratio (m/z), the putative metabolite list to interrogate was reduced
to 1351 metabolic features, starting from 2 µL of chip-separated samples. Based on quantitative metabolome data,
two main clusters on the heat-map distinguished plasma samples from the two closest sub-clusters of serum and
chip separated devices (Fig.2c) with remarkable reproducibility between three repeated samples of respective
matrices. Although signicant dierences of abundance were observed for some metabolites between the three
biological matrices, this result conrmed that the chip-separated samples are comparable to serum at the meta-
bolic level. is result also showed that the chip-separated samples would provide similar results than serum or
plasma in clinical and biological studies. is is in agreement with previous studies aiming at comparing both
plasma and serum matrices for metabolomic investigations17,18.
Taken together, proteomic and metabolomic data conrm the high reproducibility of the microuidic device
in generating analytically relevant cell-free liquid from capillary whole blood.
Diagnostic applications. e capacity of the microuidic chip to integrate in a laboratory test cycle was
tested on samples collected from 11 obese subjects. Anti-coagulated venous blood was separated in the microu-
idic device and diluted prior to being used for standard biochemical analysis. Samples were analyzed on a clinical
automated analyser platform that performed 8 typical blood markers for lipidic status, renal and liver functions,
and inammation (see Table1). In parallel, conventional plasma samples, obtained aer centrifugation, were
analysed for the same clinical panel in a central laboratory (CCL) and obtained values were used as references. As
expected, considering the nature of the subject population, values outside the normal range were found for lipid
prole (e.g. triglycerides and HDL) and liver function (e.g. gGT). For each of the eight clinical markers, strong
positive linear correlation was observed between chip-separated samples (N = 11) and reference CCL plasma
Name System Normal range
CCL vs chip-separated
Pearson r p-value
Cholesterol total Lipid prole <5 mmol/L 0.879 3.71E-4
High Density Lipoprotein (HDL) Lipid prole >1 mmol/L 0.918 6.70E-5
Triglycerides Lipid prole <1.7 mmol/L 0.996 8.1E-11
Creatinine Renal function 60–110 µmol/L 0.756 7.13E-3
Urea Renal function 2.5–7.9 mmol/L 0.969 8.82E-7
Urates Renal function 160–430 µmol/L 0.960 2.80E-6
γ-Glutamyltransferase (gGT) Liver Function 10–55 U/L 0.970 8.09E-7
C-Reactive protein(CRP) Inammatory state <10 mg/L 0.977 1.15–6
Table 1. Panel analytes. Analytes in the performed panel and associated systems. Normal range practiced in the
Geneva University Hospital for each analyte. Pearson correlations per analyte.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
www.nature.com/scientificreports/
6
Scientific REPORTS | (2018) 8:14101 | DOI:10.1038/s41598-018-32314-4
values (Pearson’s r range 0.756 to 0.996) (Table1). Furthermore, we found strong positive correlations with sta-
tistical signicance for the 8-panel markers measured between plasma CCL and chip-separated samples for each
subject individually (Pearson’s r range 0.991 to 0.999) (see Supplementary Table1). ese results are conrmed by
visual inspection of the heatmaps presenting the two tested methods (CCL and microuidic chip) and show the
capacity of the microdevice to integrate in a standard laboratory test cycle (Fig.3).
The use of conventional bench-top analytical processes requires substantial dilution factor for over-
coming instrument dead volume, thus limiting the panel size upon microsample analysis. e quality of the
chip-separated microsamples would allow a larger panel and improved resolution with dedicated low-dead vol-
ume analytical protocols and tools.
Discussion
e SP-POC microuidic device presented in this work enables the on-site operation-free separation of capillary
whole blood microsamples. e quality and reproducibility of the processed samples as well as the o-chip sample
retrieval allowed performing robust state-of-the-art bench-top analysis.
e passive separation and the simplicity of the device design allow a strong potential for immediate mass
production with high exibility both in terms of manufacturing methods and materials.
For subsequent bench-top analysis, the SP-POC device could be associated with a sample storage and trans-
port medium to preserve the sample quality from eld sampling to high-throughput analysis in a centralized
laboratory. In addition, specic lab-on-chip application could also be explored by integrating the microuidic
device described herein into micro total analysis system.
By exploiting the unique microuidic capabilities of our device for blood separation, this technology could
thus bring new insights into all biomedical areas where blood collection and testing is needed and contribute
to the active monitoring of diseases and wellness for personalized medicine, potentially impacting millions of
end-users.
Methods
Fabrication of microuidic devices. e microuidic device fabrication is based on a standard so lithog-
raphy polydimethylsiloxane (PDMS) process illustrated in Supplementary Fig.1. e device consists of two parts:
the top part contains the imprint of the structure while the bottom is at and modied with surfactant to obtain
an adequate capillary pressure in the device. e top part mold consists of a silicon wafer onto which a 200 μm
SU-8 (GM 1070, Gerstelltec, Switzerland) layer is spincoated and photostructured. A 2 mm PDMS (Sylgard 184,
Dow Corning, USA) layer is poured on the mold before degassing. e PDMS was prepared from a 1:10 mix-
ture of curing agent and base. e structured PDMS layer is partly cured for 30 min at 80 °C. e bottom part
mold consists of a bare silicon wafer exempt of structure. A 1 mm layer of PDMS with the addition of 0.58% of
surfactant (Silwet 618, Momentive, USA) is poured on the mold before degassing. e hydrophilic layer is partly
cured for 30 min at 80 °C. Both the top and bottom layers are cured simultaneously. Each structured imprint in
the top layer is cut to chip size and a 2 mm inlet is punched. e structured layer is put in contact with the hydro-
philic part and the chip outlet is created by cutting the channel at its utmost end. e devices are stored at room
temperature for at least 24 hours before use. is delay allows for complete polymerization of the materials and
the establishment of some adhesion force resulting from cross-linking between the layers.
Capillary blood collection. All samples used in this work, except those used for clinical chemistry
assay, are nger prick capillary blood microsamples. Finger pricks were performed on disinfected skin with
contact-activated lancets (Microtainer, BD, USA). e rst drop was wiped o to avoid interstitial liquid contam-
ination. e whole blood microsamples were retrieved directly from the skin of the subject using a pipette. e
research design and protocol were approved by the Swinburne’s Human Research Ethics Committee (SUHREC),
“SHR Project 2016/024 – Extracting and using small volumes of human blood for biomedical device testing”. All
methods were performed in accordance with the relevant guidelines and regulation
Figure 3. Diagnostic blood parameters. Heatmaps representing abundance of 8 clinical blood markers in 11
obese subjects. Comparison between central clinical laboratory (CCL) plasma values and the chip-separated
analytical values. Colors represent values relative to the normal range (see Table1). Red colors indicate high
concentrations, while blue colors represent low concentrations.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
www.nature.com/scientificreports/
7
Scientific REPORTS | (2018) 8:14101 | DOI:10.1038/s41598-018-32314-4
Chip-based separation of anticoagulated samples. A 200 μl sample of untreated skin-puncture capillary
whole blood was mixed with 1.6% of 2% Coomassie blue water solution for staining and observation convenience and
with 0.18 mg/100 μL of dried K2EDTA anticoagulant. e dilution due to the staining is minimal and the anticoagula-
tion doesn’t induce any dilution. e samples are thoroughly mixed before use through tube inversion. A 20 μL sample
of prepared blood was loaded in a portion of tubing with the pumping system (Nemesys, Cetoni, Germany). Tubing
and syringes were previously lled with water and an inert oil plug (Peruoromethyldecaline, PFD) to prevent any
compliance eect or mixing. e tubing was inserted in the inlet of the device placed in a closed container. e con-
tainer inner atmosphere was saturated with humidity by using an open water container, thus mitigating evaporation.
e pumping system is used during these experiments to impose a constant feeding ow rate. Time lapse imaging with
a digital camera (Panasonic, Lumix, Japan) was used to characterize front positions as well as plasma or total ow rate.
e chamber’s bottom was lined with millimeter graph paper for dimension calibration purposes. Cellular suspension
front and liquid-air front positions were measured, using soware (ImageJ), from the center of the inlet through the
center line of the channel to the desired front. Extracted volume is determined by computing the volume contained
between the cellular front and the liquid-air interface. e inltration distance is the channel length primed with liquid
and it is measured from the inlet center to the liquid-air interface. e feeding rate of whole blood was measured on
the time lapse imaging data by measuring the progression of the liquid-air interface, hence the changes in inltration
distance. e plasma extraction rate is computed by extracting the slope of a linear regression of non-zero extracted
volume data points; the feeding ow rate was extracted from the linear regression of total fed volume. e yield is com-
puted through ratio between extraction rate and feeding owrate during the extraction phase. e anticoagulated blood
separation delay is computed by using the time axis intercept of the linear regression.
Chip-based separation of fresh untreated samples. As for anticoagulated samples, a 200 μl sample of
untreated capillary whole blood was mixed with 1.6% of 2% Coomassie blue in water for staining and observation
convenience. e samples are thoroughly mixed before use through up/down pipetting operations. A 25 μL of pre-
pared blood was loaded with a pipette in the chip inlet port. is quantity is in excess, as the chip total volume is
7 μL. e sample totally lls the inlet and creates a drop on the chips top part. Liquid ows in the system through sole
capillary action. e chips were, prior to loading, placed in a closed container with high humidity to mitigate evapo-
ration during the experiments. Time lapse imaging with a digital camera (Panasonic, Lumix, Japan) was used to char-
acterize front positions. Yield and separation delay are computed on still images as no linear regression can be applied
if capillary lling is used. e cell-free liquid extraction yield is computed by measuring, in the time-lapse imaging,
the total volume loaded during the extraction phase in the chip main channel and the total generated cell-free liquid.
is value represents a time average yield during the complete extraction phase. e fresh blood separation delay is
determined by identifying the rst time point with a resolvable clear plug in the time lapse imaging data.
Reference separated sample generation. Separated samples were generated as references for analysis:
plasma and serum samples. Plasma samples were generated from fresh whole blood collected in EDTA-coated
tubes and centrifuged for 10 min at 2000 RCF. Plasma aliquots were retrieved through supernatant pipetting. To
generate serum samples, fresh whole blood was allowed to clot for 15 min at ambient condition before being cen-
trifuged for 10 min at 2000 RCF. Serum aliquots were retrieved through supernatant pipetting.
Flow cytometry analysis. Flow cytometry (BD FACSAriaTMIII, BD Biosciences, USA) was used to deter-
mine cell counts and size distribution in raw blood and separated samples. Four sample of dierent nature were
characterized in this comparison: whole blood, plasma, serum and chip-separated liquid from untreated whole
blood. e samples originated from the same subject and are all retrieved through skin puncture. For whole blood
samples, 2 μL of fresh whole blood were sampled from the surface of the subject punctured skin. e samples were
immediately diluted in 300 μL of Phosphate Buered Saline (PBS). 30 μL of this sample was further diluted 10x in
PBS. For plasma samples, 150 μL of fresh whole blood were anticoagulated before being centrifuged for 10 min at
2000RCF. A 2 μL aliquot of the resulting plasma was then diluted into 300 μL of PBS. For serum samples, 250 μL
of fresh whole blood were allowed to clot for 15 min at ambient condition before being centrifuged for 10 min
at 2000 RCF. A 2 μL aliquot of the resulting serum was then diluted into 300 μL of PBS. For the chip-separated
samples, each chip was loaded with a fresh 25 μL sample of untreated fresh whole blood. e 2 μL ejected from the
chip were recovered with a pipette before being diluted into 300 μL of PBS.
Proteomic analysis. ree dierent blood preparation protocols were characterized for proteomic content.
Equal amount of 2 µL of plasma, serum or chip-separated liquid was used for overnight trypsin digestion of single
samples. In total we digested 15 samples, ve replicates per each blood preparation protocol. Next day, peptide
digests were cleaned on MACROSpin Plate-VydacSilicaC18 (Nest Group Inc., USA), solubilized in 50 μL of 0.1%
aqueous formic acid (FA) with 2% acetonitrile (ACN) and were used for nal MS analysis.
SWATH assay library. We used publicly available SWATH assay library19 previously generated on TripleTOF
5600 mass spectrometer equipped with a NanoSpray III source and heated interface (AB Sciex, Canada) from
depleted plasma digest fractionated by strong anion exchanger (SAX) and full non-fractionated plasma samples19.
SWATH-MS Measurement and Data analysis. 15 blood samples were measured on TripleTOF 5600
mass spectrometer operated in SWATH mode as described earlier14. Reverse phase peptide separation was per-
formed with linear nanoLC gradient as described before. An accumulation time of 100 ms was used for 64 frag-
ment ion spectra of 12.5 m/z each and for the precursor scans (SWATHs) acquired at the beginning of each
cycle, resulting in a total cycle time of 3.3 s. e SWATHs were overlapping by 1 m/z and thus cover a range of
400–1200 m/z. Raw SWATH data les were converted into the mzXML format using ProteoWizard (version
3.0.3316)20 and data analysis was performed using the OpenSWATH tool15 integrated in the iPortal workow21.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
www.nature.com/scientificreports/
8
Scientific REPORTS | (2018) 8:14101 | DOI:10.1038/s41598-018-32314-4
e recorded feature intensities aer OpenSWATH identication were ltered through SWATH2stats22 to
reduce the size of the output data and remove low quality features. is resulted in a list of 312 proteins achieving
protein FDR below 1% in all the samples measured.
en these ltered fragment intensities were analysed by the R/Bioconductor package MSstats (version MSstats.
daily 2.3.5) and converted to relative protein abundances that were used for further statistical data analysis23.
Metabolic analyses. Similar to proteomic analysis, three dierent blood preparation protocols were character-
ized for proteomic content. Equal amount (2 µL) of serum, plasma or chip-separated samples were extracted using
50 µL of mix solvent EtOH:MetOH:H2O in the ratio of 2:2:1. Overall, 9 samples were extracted, three replications for
each subtype of blood samples. All samples were then vortexed mixed for 30 s, incubated for 10 min at 4 °C and cen-
trifuged for 10 min at 14000 rpm and 4 °C. e supernatants were removed and evaporated to dryness using speed
vacuum concentrator (SpeedVac SPD1010, ermo Fisher Scientic, USA) and stored at −80 °C until analysis.
Untargeted metabolomic analysis was performed on UPLC system (Dionex, ermo Fisher Scientic, USA)
hyphenated with HRMS, hybrid quadrupole-Orbitrap mass spectrometer (Q Exactive, ermo Fisher Scientic,
USA). HRMS was interfaced with an electrospray ionization (ESI) source. Data acquisition was done in both
negative (NEG) and positive (POS) polarities using similar setting: sheath gas ow rate 40, auxiliary gas ow rate
10, capillary temperature 320 °C, S-lens RF 50 and auxiliary gas heater temperature 300 °C. e sweep voltage was
optimized for each ionization mode to have the proper spray current (POS: 3.3 kV and NEG: 2.9 kV). Moreover,
lock mass was considered with respect to the acquisition mode which permit real-time recalibration by correct-
ing m/z shis due to instrumental dri. Reverse phase chromatography was done using C18 Kinetex, 2.6 μm,
50 mm × 2.1 mm I.D. column (Phenomenex, USA). Mobile phase was composed of A = 0.1% Formic acid in H2O
and B = 0.1% Formic acid in MeOH. We used identical mobile phase for both positive and negative ionization
modes. Elution was carried out in gradient condition with the mobile phase composition changed from 98% A
(0–6 min) to 100% B (6–9 min). To keep system reproducibility, we considered 3 min for system re-equilibrating.
e overall run time was roughly 13 min. Flow rate was set at 0.3 mL/min and the injection volume was 3 μL.
Raw data was converted to the mzXML format using ProteoWizard MsConvert version 3.0.7331. e mzXML
les were then processed using open-source freely available soware XCMS online for peak detection, chroma-
togram alignment and isotope annotation24. is process provided alignment of dri (retention time and accu-
rate mass) in data and ensured that a chromatographic signal (i.e., metabolite feature (m/z x RT X intensity)) is
identied with the same parameters in each sample25. To reduce data complexity, a home-made script was used to
remove background by ltering data matrix according to a predened m/z list of metabolites. is list, including
around ve thousand of present and/or detected metabolites in biouids, is built from the human metabolome
database (HMDB, http://www.hmdb.ca) and updated form our own library.
Statistical data analysis for proteomics and metabolomics. Using R package “pheatmap” on the
log-transformed, normalized relative protein and metabolite intensities, hierarchical data clustering analysis was
performed to generate two-dimensional centered heat map. Manhattan distance as distance measure was used for
clustering of scaled data. Specically for proteomic analysis, the coecient of variation between ve repetitions of
each protocol were calculated on the normalized raw protein intensities for the 43 detected FDA biomarkers. Box
plots were generated in Rstudio (version 3.0.2) by using the ggplot2 package.
Clinical chemistry. is study, approved by the local ethics committee, was based on 11 informed subjects
who declared their consent. Fasted subjects were taken two successive samples of whole blood through a single
venipuncture. e rst samples were processed in the Central clinical laboratory (CCL) of the Geneva University
Hospitals, where usual plasma separation and clinical automated analyser (Cobas 8000, Roche Diagnostics,
Switzerland) testing was performed. e second group of samples was taken in k2EDTA tubes (Vacutainer, BD,
USA). From these tubes, plasma was generated within 7 hours of sampling both through the microuidic separa-
tion device and through standard centrifugation protocol. During the interval, they were continuously mixed on
roller mixers. Prior to experiments, the samples were mixed by inverting the tube at least 3 times and vortexing.
For microuidic separation purposes, a 10 μL blood microsample was loaded in a portion of tubing with the
pumping system (Nemesys, Cetoni, Germany). Tubing and syringes were priory lled with water and a PFD oil
plug to prevent any compliance eect or dilution through mixing. e tubing was inserted in the inlet of the 5
devices operated simultaneously and the devices were placed in a closed Petri dish with an open water container.
e chip pumping operation protocol was the following: (i) load inlet until front is visible in the chip at 100 μl/min
(ii) loading of 2 µL at 10 μl/min (iii) separation at 0.1 μl/min (0.5 mm/min linear speed) until cellular front neared
the valve. e second operation was performed to ensure that blood entering the channel would not have under-
gone sedimentation in the inlet. Using the device with anticoagulated blood yields approximately 1 μL per device.
e liquid recovered from the outlet was transferred and merged in a container and total volume (of approximately
5 μL) was determined through weighing. e chip-separated samples were diluted in 100 μL (approximately 1:20
dilution) of 0.9% NaCl solution before use. e nal results were corrected to account for the precise dilution ratio
obtained. is dilution allowed obtaining sucient sample volume for the operation of analysis. Analysis was
performed through spectrophotometric techniques on an automated analyzer (AU480, Beckman-Coulter, USA)
References
1. Chapman, . et al. Overcoming the barriers to the uptae of nonclinical microsampling in regulatory safety studies. Drug Discov.
Today 19, 528–532 (2014).
2. Wang, M. M. et al. Microuidic sorting of mammalian cells by optical force switching. Nat. Biotechnol. 23, 83–87 (2005).
3. Dimov, I. . et al . Stand-alone self-powered integrated microuidic blood analysis system (SIMBAS). Lab Chip 11, 845–850 (2011).
Content courtesy of Springer Nature, terms of use apply. Rights reserved
www.nature.com/scientificreports/
9
Scientific REPORTS | (2018) 8:14101 | DOI:10.1038/s41598-018-32314-4
4. Crowley, T. A. & Pizziconi, V. Isolation of plasma from whole blood using planar microlters for lab-on-a-chip applications. Lab
Chip 5, 922–929 (2005).
5. Homsy, A. et al. Development and validation of a low cost blood ltration element separating plasma from undiluted whole blood.
Biomicrouidics 6, 12804 (2012).
6. Tripathi, S. et al. Microdevice for plasma separation from whole human blood using bio-physical and geometrical eects. Sci. Rep.
6, 26749 (2016).
7. Di Carlo, D. Inertial microuidics. Lab Chip 9, 3038–3046 (2009).
8. ersaudy-erhoas, M. & Sollier, E. Micro-scale blood plasma separation: from acoustophoresis to egg-beaters. Lab Chip 13, 3323–46
(2013).
9. Mielczare, W. S., Obaje, E. A., Bachmann, T. T. & ersaudy-erhoas, M. Microuidic blood plasma separation for medical
diagnostics: is it worth it? Lab Chip 16, 3441–3448 (2016).
10. Carraro, P. & Plebani, M. Errors in a stat laboratory: Types and frequencies 10 years later. Clin. Chem. 53, 1338–1342 (2007).
11. Sartory, W. . Prediction of Concentration Proles during Erythrocyte sedimentation by hindered settling model. Biorheology 11,
253–264 (1974).
12. Quemada, D. heology of concentrated disperse systems and minimum energy dissipation principle - I. Viscosity-concentration
relationship. R heol. Acta 16, 82–94 (1977).
13. Murata, T. Eects of sedimentation of small red blood cell aggregates on blood ow in narrow horizontal tubes. Biorheology 33,
267–283 (1996).
14. Gillet, L. C. et al. Targeted Data Extraction of the MS/MS Spectra Generated by Data-independent Acquisition: A New Concept for
Consistent and Accurate Proteome Analysis. Mol. Cell. Proteomics 11(O111), 016717 (2012).
15. öst, H. L. et al. OpenSWATH enables automated, targeted analysis of data-independent acquisition MS data. Nat. Biotechnol. 32,
219–23 (2014).
16. Anderson, N. L. e clinical plasma proteome: A survey of clinical assays for proteins in plasma and serum. Clin. Chem. 56, 177–185
(2010).
17. Yu, Z. et al. Dierences between human plasma and serum metabolite proles. PLoS One 6, 1–6 (2011).
18. Liu, L. et al. Dierences in metabolite prole between blood plasma and serum. Anal. Biochem. 406, 105–112 (2010).
19. Liu, Y. et al. Quantitative variability of 342 plasma proteins in a human twin population. Mol. Syst. Biol. 11, 786–786 (2015).
20. Chambers, M. C. et al. A cross-platform toolit for mass spectrometry and proteomics. Nat. Biotechnol. 30, 918–920 (2012).
21. unszt, P. et al. iPortal: the swiss grid proteomics portal: equirements and new features based on experience and usability
considerations. Concurr. Comput. Pract. Exp. 27, 433–445 (2015).
22. Blattmann, P., Heusel, M. & Aebersold, . SWATH2stats: An /bioconductor pacage to process and convert quantitative SWATH-
MS proteomics data for downstream analysis tools. PLoS One 11, 1–7 (2016).
23. Choi, M. et al. MSstats: an pacage for statistical analysis of quantitative mass spectrometry-based proteomic experiments.
Bioinformatics 30, 2524–2526 (2014).
24. Tautenhahn, ., Patti, G. J., inehart, D. & Siuzda, G. XCMS Online: A Web-Based Platform to Process Untargeted Metabolomic
Data. Anal. Chem. 84, 5035–5039 (2012).
25. Wolfender, J. L., Marti, G., omas, A. & Bertrand, S. Current approaches and challenges for the metabolite proling of complex
natural extracts. J. Chromatogr. A 1382, 136–164 (2015).
Acknowledgements
is work was scientically evaluated by the SNSF, nanced by the Swiss Confederation and funded by Nano-
Tera.ch. e authors would like to thank the Centre for Micro fabrication of the Swiss Institute of Technology
in Lausanne (CMI EPFL) for their support. is work was performed in part at the MCN and Biointerface
(Swinburne) Node of the Australian National Fabrication Facility (ANFF). A company established under the
National Collaborative Research Infrastructure Strategy to provide nano- and micro-fabrication facilities
for Australia’s researchers. T.S. was supported by the Swiss National Science Foundation 26 (grant # SNSF;
31003A_130530 to R.A). e contributions and expertise of Professor Richard James is gratefully acknowledged.
Author Contributions
S.B., D.F. developed microuidic chip devices. J.D., S.B., D.F., T.S. and A.T. designed the experiments and wrote
the manuscript. T.S. and Y.L. performed proteomic data analysis. N.B. and A.T. performed metabolomic data
analysis. S.B., D.F. and T.S. prepared the gures. S.B. and M.C. performed FACS analysis. Z.P. and M.F. recruited
subjects and obtained informed consent. S.G., M.A., R.A. and P.R. edited the manuscript and gures. All authors
reviewed this work.
Additional Information
Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-018-32314-4.
Competing Interests: e authors declare no competing interests.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre-
ative Commons license, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons license and your intended use is not per-
mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the
copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
© e Author(s) 2018
Content courtesy of Springer Nature, terms of use apply. Rights reserved
1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com
Available via license: CC BY
Content may be subject to copyright.
