Thrombosis Research 136 (2015) 328–334

Contents lists available at ScienceDirect

Thrombosis Research

j ourna l homepage: www.e lsev ie r .com/ locate / th romres

Regular Article

A microfluidic cell culture system for monitoring of sequential changesin endothelial cells after heat stress

Hidekatsu Tazawa 1, Kenjiro Sato 2, Atsuhiro Tsutiya 2, Manabu Tokeshi 3,4, Ritsuko Ohtani-Kaneko 2,5,⁎1 Institute of Microchemical Technology Co. Ltd., 207 East KSP, 3-2-1 Sakado, Takatsu, Kawasaki, Kanagawa 213-0012, Japan2 Department of Life Sciences, Toyo University, 1-1-1 Itakura, Oura, Gunma 374-0193, Japan3 Division of Biotechnology and Macromolecular Chemistry, Faculty of Engineering, Hokkaido University, Kita 13 Nishi 8, Kita-ku, Sapporo 060-8628, Japan4 FIRST Research Center for Innovative Nanobiodevices and Innovative Research Center for Preventive Medical Engineering, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan5 Bio-Nano Electronic Research Centre, Toyo University, 2100 Kujirai, Kawagoe, Saitama 350-8585, Japan

⁎ Corresponding author at: Department of Life SciencesOura, Gunma 374–0193, Japan. Tel.: +81 276 82 9213.

E-mail address: r-kaneko@toyo.jp (R. Ohtani-Kaneko)

http://dx.doi.org/10.1016/j.thromres.2015.05.0080049-3848/© 2015 Elsevier Ltd. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 5 February 2015Received in revised form 30 April 2015Accepted 11 May 2015Available online 17 May 2015

Keywords:Microfluidic cell culture systemMicrochipEndothelial cellsHeat-shock responsePAI-1

Endothelial damage induced by a highly elevated body temperature is crucial in some diseases including viralhemorrhagic fevers. Here, we report the heat-induced sequential changes of endothelial cells under shear stress,which were determined with a microfluidic culture system. Although live cell imaging showed only minorchanges in the appearance of heat-treated cells, Hsp70 mRNA expression analysis demonstrated that the endo-thelial cells in channels of the system responded well to heat treatment. F-actin staining also revealed clearchanges in the bundles of actin filaments after heat treatment. Well-organized bundles of actin filaments in con-trol cells disappeared in heat-treated cells cultured in the channel. Furthermore, the system enabled detection ofsequential changes in plasminogen activator inhibitor-1 (PAI-1) secretion from endothelial cells. PAI-1 concen-tration in the effluent solution was significantly elevated for the first 15 min after initiation of heat treatment,and then decreased subsequently. This study provides fundamental information on heat-induced endothelialchanges under shear stress and introduces a potent tool for analyzing endothelial secretions.

© 2015 Elsevier Ltd. All rights reserved.

Introduction

A highly elevated body temperature can damage endothelial cells,leading to diffusion of serum proteins and blood cells and secretionchanges of coagulation and aggregation inhibitors. Endothelial damageis thus implicated in causing symptoms such as increased vascular per-meability, internal bleeding, and coagulation disorders. In particular,these symptoms are induced by some viral hemorrhagic fevers such asEbola virus infection, Crimean-Congo hemorrhagic fever, and denguehemorrhagic fever [1] (for review, [2,3]). The pathogenic mechanismsunderlying the endothelial damage associated with these viral hemor-rhagic fevers are very diverse and complex, includingboth direct and in-direct dysfunction of endothelial cells [2]. The direct effects of elevatedtemperature and viral infection on endothelial cells are particularly im-portant, and have been reviewed recently [4,5]. Other studies using an-imal models have suggested that endothelial cells are early targets ofheat stress injury [6,7]. Although the importance of heat injuries on en-dothelial functions in these viral hemorrhagic fevers or other diseasesthat are accompanied by high fever has been long recognized, only afew in vivo and in vitro studies have focused specifically on the effectsof heat stress on endothelial cells [8–10]. In addition, all previous

, Toyo University, 1-1-1 Itakura,

.

in vitro reports were carried out using stationary culture methodswith standard culture flasks, wells, or transwell cell culture inserts.However, since vascular endothelial cells in vivo are constantly subject-ed to shear stress associated with blood flow [11], it is likely that theirresponses in conventional stationary cultures in vitrodonot properly re-flect their responses in vivo.

Therefore, in order to examine the effect of heat stress on endothelialcells cultured under shear stress, we constructed a microfluidic culturesystem for endothelial cells, andmonitored their morphological and se-cretion changes during heat stress. This system enabled detection of theheat-induced sequential changes in cytoskeletal actin filaments and se-cretion of plasminogen activator inhibitor 1 (PAI-1).

Materials and Methods

Cell Culture in Flasks

We used two endothelial cell lines: monkey RF/6A 135 cells (CellBank, RIKEN BioResource Center; Tsukuba, Japan) and human umbilicalvein endothelial cells (HUVECs; Life Technologies). We used RF/6A 135cells, which originate from monkey chorioretinal vessel endothelialcells, because overexpression of HSP70 is known to be easily inducedin endothelial cells of chorioretinal vessels by increased temperature[12]. In addition, HUVECswere used to ensure the evaluation ofmonkeyPAI-1 cells with the ‘Human PAI-1 ELISA kit’ employed in the present

329H. Tazawa et al. / Thrombosis Research 136 (2015) 328–334

study. These cells were cultured in 75-cm2 tissue culture flasks (TechnoPlastic Products AG; Switzerland) at 37.5 °C in a humidified 5% CO2 at-mosphere in RPMI 1640medium (MP Biomedicals, Japan) supplement-ed with 10% fetal bovine serum (FBS; Biological Industries; Kibbutz BeitHaemek, Israel), 50 IU/ml penicillin (Meiji Seika Pharma Co., Ltd.;Japan), and 25 μg/ml streptomycin (Meiji Seika Pharma Co., Ltd.;Japan) for RF/6A 135 cells, and in MCDB 107 endothelial basal medium(COSMO BIO Co., Ltd.) supplemented with 10% FBS, 50 μg/ml endotheli-al cell growth supplement (Takara Bio, Japan), 100 μg/ml heparin (hep-arin sodium, Wako Pure Chemical Industries, Ltd.; Japan), 50 IU/mlpenicillin, and 25 μg/ml streptomycin for HUVECs.

To analyze the expression of Hsp70mRNA and PAI-1 secretion fromcells cultured in flasks, RF/6A 135 cells were seeded into 12 cultureflasks, six for controls (three for Hsp70 mRNA analysis and three forPAI-1 analysis) and the other six for the heat-treated groups. After thecells were cultured at 37.5 °C overnight, they were incubated for 1 hat 42.5 °C (heat-treated group) or at 37.5 °C (control group). Thereafter,mRNA or the media were collected and analyzed as described below.

Cell Culture in the Channels of the Microfluidic Culture System

In this study, we used a microchip (Fig. 1A, Sumitomo BakeliteCo., Ltd.) with a parallel flow channel made of polystyrene (PS), aspreliminary studies determined that this was the optimal chip mate-rial for culture of RF/6A 135 endothelial cells and HUVECs. A channelfor fluidic cell culture was formed in the interior of the base PS plate.The base and cover PS plates, both of which were 1-mm thick, were

Fig. 1. A. Schematic illustration of the chip platform with two channels. B. Schematic diagramillustration of the overall experimental setup of the microfluidic culture system. D. Image of th

thermally laminated. Each microchip had 2 channels, which were6-cm long, 300-μm wide, and 100-μm deep (Fig. 1B). Before thecells were introduced into the channel, the microchip was sterilizedwith 70% ethyl alcohol and rinsed with sterilized distilled water.Then, each channel was pre-coated with collagen type IV solution(Nitta Gelatin Inc.; Japan) as follows. First, the channel was washedwith 0.1 M NaOH, 70% ethanol, and sterilized distilled water in suc-cession. Next, the inner surface of each channel was coated by fillingit with collagen type IV solution for 2 h. The channels were thenwashed with sterilized distilled water five times and the chip plat-form was connected with sterilized tubes, valves, and syringes(Fig. 1C).

After filling the channels with culture medium, the inlet wasclosed. The microchip was then placed on the inserted heatingplate (ITO temperature controller) set on the microscope stage(Nikon; Tokyo, Japan) (Fig. 1D). The temperature of the heatingplate was set to 37.5 °C. Subconfluent endothelial cells were dissoci-ated with papain (Worthington Biochemical Corporation; Lakewood, NJ,USA) and dispersed in the medium to prepare the cell suspension at aconcentration of 2×107 cells/ml.We then introduced 35 μl of the cell sus-pension into the channel via the inlet. When the cells passed through thechannel, the same cell suspensionwas also inserted into another channel.After seeding the cells, the inlet was closed and the microchip wasimmobilized for 2 h at 37.5 °C to allow the cells to attach to the bottomof the channel. After 2 h, the valve was turned to the medium port andperfusion was started at 0.2 μl/min at 37.5 °C. Then, after 1 h perfusion,the perfusion flow rate was gradually increased to 2 μl/min.

showing the partial cross-section of the channel inside the chip platform. C. Schematice chip platform placed on the inserted heating plate set on the microscope stage.

0

10

20

30

40

50

60

70

80

Culture in flasks Culture in channels

Exp

ress

ion

of H

sp70

mR

NA **

**

Control Heat-treated Control Heat-treated

Fig. 2. Expression of Hsp70mRNA (mean ± SD) in control cells and heat-treated (42.5 °Cfor 1 h) cells cultured in flasks (left columns) or in the channel of the microfluidic culturesystem (right columns). The expression ofHsp70mRNAwas normalized to the expressionof the housekeeping gene, Rpl32 mRNA and y-axis indicates the relative expressionof Hsp70 mRNA vs. Rpl32 mRNA. Significant differences are denoted with asterisks(* p b 0.05, Student’s t-test). ND = not detected.

330 H. Tazawa et al. / Thrombosis Research 136 (2015) 328–334

To monitor cellular changes, we raised the temperature of theheating plate from 37.5 °C to 42.5 °C 1 h after the start of medium per-fusion at 2 μl/min, and the temperature wasmaintained for 1 h ormore.Live-cell images of RF/6A 135 cells were captured before and duringheat treatment using phase-contrast microscopy (Axiovert 40C, Zeiss;Tokyo, Japan). The effluent solution was collected to measure the con-centration of PAI-1 secreted from RF/6A 135 cells and HUVECs beforeand during heat treatment. At the end of the heat treatment, eitherRNA was extracted from the RF/6A 135 cells in the channel, or thecells were fixed for immunocytochemistry. In the latter case, the cellsin the channel were fixed for 1 h with 4% paraformaldehyde in 0.1 Mphosphate buffer (pH 7.3).

Real-time Polymerase Chain Reaction (PCR) Analysis for Hsp70 mRNA

Total RNA was extracted from cultured cells with RNAiso Plus(Takara Bio, Japan) and then converted to cDNA using a reverse tran-scription kit (QIAGEN; Valencia, CA). The synthesized cDNA was usedas a template in subsequent PCR analysis. Quantitative real-time PCRanalysis was performed using a Thermal Cycler Dice Real Time SystemTP800 (Takara Bio) according to themanufacturer’s suggested protocol.Gene expression levels were normalized to ribosomal protein L32(Rpl32) mRNA expression, which was measured simultaneously.Primers used for gene amplification were as follows: Hsp70 FW primer:5′-CGACCTGAACAAGAGCATCA-3′, Hsp70 RV primer: 5′-AAGATCTGCGTCTGCTTGGT-3′; Rpl32 FW primer: 5′-GCCCAAGATCGTCAAAAAGA-3′,Rpl32 RV primer: 5′-GTTGCACATCAGCAGCACTT-3′, PAI-1 Fw primer: 5’-CGAAGTTGACCTCAGGAAGC-3’, PAI-1 Rev primer: 5’-TGACAGCTGGTGGATGAGGAG-3’, Bax FW primer: 5’-AACTGGACAGTAACATGGAGCTG-3’,Bax RV primer: 5’-CTGGCAAAGTAGAAAAGGGCGA-3’, Bcl-2 FW primer:5’-TTCTTTGAGTTCGGTGGGGTC-3’, Bcl-2 RV primer: 5’-AGCCAGGAGAAATCAAACAGAGG-3’. Amplifications were carried out in a 20-μl volume,with each reaction containing 1 μl of cDNA, 200 nM of each primer pair,and 1× SYBR Premix Extaq (Takara Bio). The reaction mixture was sub-jected to 40 cycles of amplification, followed by post-PCR fluorescencedissociation curve analysis to confirm the specificity of the primer sets.

F-actin Staining

After fixation, cytoskeletal F-actin staining was performed withCytoPainter F-actin Staining kit (Abcam; Cambridge, UK). First, cells cul-tured in the channels were incubated with 10 mM PBS containing 0.1%Triton X-100 (Sigma-Aldrich) at room temperature for 5min. After rins-ing, cells were treated with fluorescent phalloidin conjugate dilutedwith labeling buffer at room temperature for 30 min. The cells werethen washed with PBS and observed under a fluorescence microscope(LSM5 Pascal, Carl Zeiss; Oberkochen, Germany).

Enzyme-linked Immunosorbent Assay (ELISA) for PAI-1

The cell culture effluent solution from each channel (n=4 for RF/6A135 cells and n= 3 for HUVECs) was collected on ice 15min before theheat treatment at 42.5 °C and at 15-min intervals during the heat treat-ment (0–15, 15–30, 30–45, 45–60, and 60–75min after the start of heattreatment). Then, the effluent solutions were centrifuged for 10 min at3000 × g at 4 °C to remove debris. Supernatants were subjected to aPAI-1 assay using Human PAI-1 ELISA Kit (ASSAYPRO; St. Charles, MO,USA). The estimation of PAI-1 concentration was performed accordingto the manufacturer’s protocol. In brief, 20 μl of samples and 30 μl ofthe dilute mix included in the kit, or 50 μl of diluted human PAI-1 stan-dard included in the kit were added into each well of the ELISA plate.Then, the wells were covered with a sealing tape and incubated for2 h. After each well was completely washed five times with 200 μl ofwash buffer, 50 μl of biotinylated human PAI-1 antibody was added toeach well and incubated for 1 h. After washing, 50 μl of streptavidin-peroxidase conjugate was added and incubated for 30 min. Following

an additional wash, 50 μl of chromogen substrate was added and sampleswere incubated until a blue color developed. Then, 50 μl of stop solutionwas added into each well and the absorbance was immediately read atawavelength of 450 nm (microplate readerMTP-800, CORONA ELECTRICCo., Ltd., Japan). The mean value of the duplicate readings for each stan-dard and samplewas calculated. Sample concentrations of PAI-1were de-termined from the standard curve.

Statistical Analysis

For statistical analysis, p-valueswere determined using the Student’st-test or the Student’s t-test for paired samples. p-values of less than0.05 were considered significant.

Results

The heat-shock response of endothelial cells was first examined bycomparing Hsp70 mRNA expression levels between heat-treated andcontrol RF/6A 135 endothelial cells. As shown in Fig. 2, the expressionof Hsp70 mRNA was significantly increased in heat-treated cells cul-tured in flasks, compared to that of control cells (p b 0.01, Student’st-test; left columns in Fig. 2). The expression of Hsp70 mRNA was alsosignificantly increased in the heat-treated cells cultured in channels ofthe microfluidic culture system, compared to that of control cells(p b 0.01, Student’s t-test; right columns in Fig. 2). These results demon-strated that themicrofluidic culture system is capable of detecting cellu-lar heat-shock responses.

Live images of control and heat-treated RF/6A 135 cells in thechannel were captured 1 h before the heat stress and every hour for3 h during the heat-stress treatment. Heat treatment scarcely causedmorphological changes for 3 h, as shown in pictures 6–8 in the right col-umn of Fig. 3, compared to control cells (pictures 2–4 in Fig. 3) or cellsbefore the heat stress (pictures 1 and 5 in Fig. 3). However, cytoskeletalchanges were already evident at 1 h after initiation of heat stress, basedon positive F-actin staining in RF/6A 135 cells cultured in channels(Fig. 4). In controls, actin filaments were well organized into intracellu-lar bundles, though some actin filaments were abundant beneath theplasma membrane (Fig. 4A and C). On the contrary, the intracellularfilamentous organization of actin bundles almost disappeared in heat-treated cells, whereas strong actin stainingwas found beneath the plasmamembrane, especially at the junction between adjacent cells (arrows inFig. 4B andD). These observations clearly demonstratedheat-induced cel-lular cytoskeletal changes.

We then studied the expression of PAI-1mRNA in control and heat-treated cells (Fig. 5). The expression of PAI-1 mRNA did not increase

Before heat stress

1 h heat stress

2 h heat stress

3 h heat stress

Heat –treated cellsControl cells

5

7

6

8

2

3

4

1

Fig. 3.Representative images of cells cultured in the channel at 37.5 °C (control cells, left column) and at 42.5 °C (heat-treated cells, right column). Images of control cells were taken everyhour for 4 hours (1–4). Images of heat-treated cells were taken before the heat stress (5) and every hour after the start of the heat stress (6–8). Scale bars = 200 μm.

331H. Tazawa et al. / Thrombosis Research 136 (2015) 328–334

after exposure to higher temperature. Next, we examined sequential se-cretion changes of PAI-1 from heat-treated RF/6A 135 endothelial cells,since severe heat stroke is known to cause disseminated intravascularcoagulation (DIC), which is reportedly associated with significantly ele-vated plasma PAI-1 levels in vivo [13]. However, the concentration ofPAI-1 in the cell medium from conventional culture flasks 1 h afterheat stress tended to be decreased, compared to that from control flasks(Fig. 6, p = 0.077, control vs heat-stressed group). On the contrary,heat-stress rapidly induced sequential changes of PAI-1 concentrationin the effluent solution from RF/6A 135 cells cultured in the channel,measured at 15-min intervals (Fig. 7). The relative concentrations ofPAI-1 secreted from endothelial cells were significantly increased

Fig. 4. Representative images of phalloidin staining for F-actin in control cells (A and C) and heamarked in white boxes in A and B are magnified and shown in C and D, respectively. Scale bar

within the first 15 min and at 15–30 min after the start of heat stresscompared to the concentration before the heat stress (p b 0.01), andthen gradually decreased, followed by a significant decrease at 45–60and 60–75 min after the stress (p b 0.05 and p b 0.01, respectively).

The PAI-1 ELISA kit used in the present study is suitable for analysis ofhuman samples. To examine whether the kit is usable to evaluate PAI-1secreted frommonkey cells, we examined PAI-1 not only inmonkey sam-ples but also in human HUVEC samples under the same experimentalconditions (Fig. 8). Almost similar to the results obtained for RF/6A 135cells (Fig. 7), the concentration of PAI-1 in the effluent of HUVECs wassignificantly increased for the first 15 min after the start of heat stress(p b 0.01). Then, it was gradually decreased and became significantly

t-treated cells (B and D) in the channel one hour after the start of the heat stress. Regionss = 50 μm (A and B), 25 μm (C and D).

0

0.5

1

Control Heat-treated

Exp

ress

ion

of P

AI-

1 m

RN

A

Fig. 5. Relative expression of PAI-1 mRNA from control and heat-treated (1 h) endothelialcells (RF/6A 135 cells) cultured in channels of the microfluidic culture system.

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0-15 min 0-15 min 15-30 min 30-45 min 45-60 min 60-75 min

Rel

ativ

e C

once

ntra

tion

of P

AI-

1

**

**

*

*

Before Heat-stress

Fig. 7. Relative concentration of PAI-1 secreted from endothelial cells (RF/6A 135 cells)cultured in channels of the microfluidic culture system before and during the heat stress.Significant differences are denoted with asterisks, compared to the concentration beforethe heat stress [Values represent means ± SD. *p b 0.05 and **p b0.01, Student’s t-test(for paired samples)].

332 H. Tazawa et al. / Thrombosis Research 136 (2015) 328–334

reduced at 60–75 min compared to the concentration before heat stress(p = 0.081, before heat-stress vs after heat-stress at 45–60 min;p b 0.05, before heat-stress vs after heat-stress at 60–75min). These stud-ies clearly demonstrated that the heat stress induced a transient increasein PAI-1 secretion frombothmonkey andhumanendothelial cells, follow-ed by its suppression.

Discussion

Here, we showed that heat stress induced rapid changes in endothe-lial cells cultured under shear stress conditions by using a microfluidicculture system.We confirmed a significant increase inHsp70mRNA ex-pression 1 h after the initiation of heat stress. Furthermore, F-actinstaining demonstrated clear cytoskeletal changes in endothelial cellscultured in the channels, even within 1 h after initiation of the heattreatment. We were also able to detect rapid heat-induced changes inPAI-1 secretion from endothelial cells. This is the first study to demon-strate stress-induced sequential changes of PAI-1 secreted from cul-tured endothelial cells.

The wall shear stress under our present experimental condition wascalculated as τ = 6μQ/ωd2, where τ is the shear stress in dyn cm−2, μthe fluid viscosity in dyn s cm−2 or Q the volumetric flow rate in cm3

s−1 and ω and d are the channel width and depth in cm, respectively[14,15]. When the volumetric flow rate was 2 μL/min and the fluidic

Rel

ativ

e C

once

ntra

tion

of P

AI-

1

0

0.5

1

Control Heat-treated

Fig. 6. Relative concentration of PAI-I secreted from control and heat-treated endothelialcells (RF/6A 135 cells) cultured in flasks. The concentration of PAI-1 was analyzed in theculture medium of control cells (incubated for 1 h at 37.5 °C) and heat-treated cells (incu-bated for 1 h at 42.5 °C).

viscosity was 0.0145 dyn s cm−2 [15], the shear stress was 0.7 dyn/cm2, which was lower than venous levels of shear stress (1 dyne/cm2). Although we examined PAI-1 secretion under low shear stressin the present study, we are currently studying it under differentshear stresses.

It is well known that exposure to high temperature induces the ex-pression of Hsp70mRNA in various kinds of cells, including endothelialcells. For example, rapid increases inHsp70mRNAwere observed whenJornot et al. [16] exposed HUVECs to 43–45 °C, Fukao et al. [17] exposedHUVECs to 43 °C, and Nakabe et al. [18] exposed human arterial endo-thelial cells at 42 °C. We first confirmed that this indicator of a heat-shock response was well induced in endothelial cells cultured in themicrofluidic cell culture system, based on the initial increase in Hsp70mRNA expression (Fig. 2).

We then showed that heat stress at 42.5 °C for 1 h disrupted theintracellular bundle organization of actin filaments in endothelial cellscultured in the microfluidic cell culture system (Fig. 4). Such heat-induced disruption of actin organization is consistent with previous ob-servations obtained from endothelial cells cultured in flasks [19,20].

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0-15 min 0-15 min 15-30 min 30-45 min 45-60 min 60-75 min

Rel

ativ

e C

once

ntra

tion

of P

AI-

1

Before Heat-stress

**

*

Fig. 8. Relative concentration of PAI-1 secreted from endothelial cells (HUVECs) culturedin channels of microfluidic culture system before and during the heat stress. Significantdifferences are denoted with asterisks, compared to the concentration before the heatstress [Values represent means ± SD. *p b 0.05, Student’s t-test (for paired samples)].

333H. Tazawa et al. / Thrombosis Research 136 (2015) 328–334

Romanov et al. [20] found that short-lasting hyperthermia (1–4 h) trig-gered the disappearance of cytoplasmic actin filaments and their redis-tribution to areas of cell-to-cell contact. DeMeester et al. [19] found thatinflammatory and/or heat shock treatments induced a dramatic cyto-skeletal collapse in endothelial cells, suggesting that cytoskeletal rear-rangement is likely a critical event in the pathway to apoptosis. Inorder to reveal the relationship between cytoskeletal collapse and apo-ptosis, we additionally examined expression of Bax and Bcl-2mRNAs asindicators of pro- and anti-apoptotic gene expression, respectively.After a 1-h heat treatment, although the expression of Bax mRNA wasnot changed [Bax mRNA/Rpl32 mRNA = 0.0955 ± 0.00522 (mean ±SD) in control cells (n = 3) and 0.0958 ± 0.0156 in heat-treated cells(n = 3)], the Bcl-2 mRNA expression was significantly increased [Bcl-2mRNA/Rpl32 mRNA = 0.0964 ± 0.00615 in control cells (n = 3) and0.122 ± 0.00607 in heat-treated cells (n = 3; p b 0.001; Control vsHeat-treated)]. Thus, the heat treatment used in the present study in-duced not a pro-apoptotic but an anti-apoptotic response. Combinedwith our data on the cytoskeletal disruption and increase in Bcl-2mRNA, our results suggest that cytoskeletal rearrangement induced byheat treatment is not directly related to the pathway to apoptosis.

One of the major advantages of a microfluidic culture system is thecapability of monitoring the sequential secretion of certain chemicalsin the effluent solution. Our system enabled monitoring of stress-induced sequential changes in PAI-1 concentration in the effluent forthefirst time. Under physiological conditions in vivo, PAI-1 is consideredto be released into the circulation and the extracellular space by only afew types of cells, including liver cells, smooth muscle cells, adiposecells, and platelets [21]. Severe sepsis, as well as other acute or chronicinflammatory diseases, causes high PAI-1 plasma levels. Under thesepathological conditions, several tissues, including endothelial cells, areknown to secrete quite large amounts of PAI-1 [21]. PAI-1 releasefrom endothelial cells is up-regulated by inflammatory cytokines andsignal transduction pathways, including the nuclear factor-kappaB andmitogen-activated protein kinase pathways (for review, see [22]). Si-multaneously, endothelial cells alter their secretory function in re-sponse to the shear stress generated by flowing blood [23]. Therefore,in order to elucidate the precise regulatorymechanismof PAI-1 in endo-thelial cells, their PAI-1 secretion should be examined under conditionsof shear stress.

In the present study, since PAI-1 increased quickly (within 15 min)upon exposure to higher temperature, it is unlikely that the increasewas mediated by enhanced protein synthesis. To confirm this hypothe-sis, we measured PAI-1 mRNA levels and revealed that the mRNA didnot increase after exposure to higher temperature (Fig. 5). Thus, the in-crease in PAI-1 antigen level was most likely the result of enhanced se-cretion. Other mechanisms, however, may also be involved, such asshedding of either cell-surface PAI-1 or matrix-bound PAI-1, as wasshown by Iacoviello et al. [24]. There are many reports showing thatPAI-1 secreted from vascular endothelial cells plays important rolesunder a variety of physiological and pathological conditions such asthrombolytic resistance [25], endothelial cell death [26], and smoothmuscle cell migration [27]. Thus, it is possible that the sudden increasein PAI-1 secretion after exposure to high temperature plays importantroles under certain pathological conditions. In the present study, wemeasured PAI-1 antigen levels but not its activity, because we perfuseda simple culture medium that did not contain vitronectin, which bindsPAI-1 and stabilizes its activity [28,29]. Wewould like to evaluate activePAI-1 level in the mediumwith vitronectin by using the microchip cul-ture system in the near future.

The present study is the first to report sequential change of PAI-1 se-cretion from endothelial cells during heat and shear stress. Elevation ofbody temperature is a common feature of inflammatory reactions, andis also considered to enhance the body’s defense mechanism. However,outbreaks of viral hemorrhagic fevers are emerging as an increasingthreat to human health worldwide. Moreover, dengue virus-infectedpatients have a high plasma concentration of PAI-1, and survival was

shown to be significantly worse in patients with higher PAI-1 concen-trations [30]. Elevated PAI-1 concentration in children suffering fromdengue hemorrhagic fever was also reported [3,31,32]. In addition, arecent study on Ebola virus infection showed that PAI-1 levels wereelevated in pediatric patients, and were strongly elevated in patientswho died and in those with hemorrhagic manifestations [33]. Althoughthe mechanism underlying the increase of PAI-1in these patientsremained unclear, Shyu et al. [34] suggested that domain III of the enve-lope glycoprotein of dengue virus serotypes 2 (EIII) induces PAI-1 geneexpression via theMEK/ERK signaling pathways. Our present study pro-vides basic information and a new tool for further elucidation of thismechanism. In addition, our system is expected to become a very usefultool for investigating many other responses of endothelial cells undershear stress, since vascular dysfunction plays an important role in vari-ous major diseases such as atherosclerosis [35], cancer [36], and diabe-tes [37]. Therefore, our novel microfluidic culture systemwill be helpfulto study the responses of endothelial cells under sheer stress to variousbiological, chemical, or physical cues.

Conflicts of Interest Statement

The authors whose names are listed immediately below certifythat they have NO affiliations with or involvement in any organizationor entity with any financial interests (such as honoraria; educationalgrants; participation in speakers’ bureaus; membership, employment,consultancies, stock ownership, or other equity interest; and experttestimony or patent-licensing arrangements), or non-financial interest(such as personal or professional relationships, affiliations, knowledgeor beliefs) in the subject matter ormaterials discussed in this manuscript.

Acknowledgements

We thank Prof. T. Urano, Hamamatsu University School of Medicine,for his helpful advice and discussions. This study was partially support-ed by the Japan Science and Technology Agency (JST) A-STEP SeedsActualization Type program. Part of this study was also supported by aGrant for the Program for the Strategic Research Foundation at PrivateUniversities S1101017, organized by the Ministry of Education, Culture,Sports, Science, and Technology (MEXT), Japan.

References

[1] H. Bodur, E. Akinci, P. Ongürü, Y. Uyar, B. Baştürk, M.G. Gözel, et al., Evidence of vas-cular endothelial damage in Crimean-Congo hemorrhagic fever, Int J Infect Dis 14(8) (2010) e704–e707, http://dx.doi.org/10.1016/j.ijid.2010.02.2240.

[2] S. Paessler, D.H. Walker, Pathogenesis of the viral hemorrhagic fevers, Annu RevPathol 8 (2013) 411–440, http://dx.doi.org/10.1146/annurev-pathol-020712-164041.

[3] A. Chuansumrit, W. Chaiyaratana, Hemostatic derangement in dengue hemorrhagicfever, Thromb Res 133 (1) (2014) 10–16, http://dx.doi.org/10.1016/j.thromres.2013.09.028.

[4] N.A. Dalrymple, E.R. Mackow, Virus interactions with endothelial cell receptors: im-plications for viral pathogenesis, Curr Opin Virol 7 (2014) 134–140, http://dx.doi.org/10.1016/j.coviro.2014.06.006.

[5] S. Antoniak, N. Mackman, Multiple roles of the coagulation protease cascade duringvirus infection, Blood 123 (17) (2014) 2605–2613, http://dx.doi.org/10.1182/blood-2013-09-526277.

[6] A. Bouchama, M.M. Hammami, A. Haq, J. Jackson, S. Al-Sedairy, Evidence for endo-thelial cell activation/injury in heatstroke, Crit Care Med 24 (7) (1996) 1173–1178.

[7] G.T. Roberts, H. Ghebeh, M.A. Chishti, F. Al-Mohanna, R. El-Sayed, F. Al-Mohanna,et al., Microvascular injury, thrombosis, inflammation, and apoptosis in the patho-genesis of heatstroke: a study in baboon model, Arterioscler Thromb Vasc Biol 28(6) (2008) 1130–1136, http://dx.doi.org/10.1161/ATVBAHA.107.158709.

[8] M. Wagner, I. Hermanns, F. Bittinger, C.J. Kirkpatrick, Induction of stress proteins inhuman endothelial cells by heavy metal ions and heat shock, Am J Physiol 277 (5 Pt1) (1999) L1026–L1033.

[9] K.Y. Ng, C.W. Cho, T.K. Henthorn, R.L. Tanguay, Effect of heat preconditioning on theuptake and permeability of R123 in brain microvessel endothelial cells during mildheat treatment, J Pharm Sci 93 (4) (2004) 896–907.

[10] Z.T. Gu, H. Wang, L. Li, Y.S. Liu, X.B. Deng, S.F. Huo, et al., Heat stress induces apopto-sis through transcription-independent p53-mediated mitochondrial pathways inhuman umbilical vein endothelial cell, Sci Rep 4 (2014) 4469, http://dx.doi.org/10.1038/srep04469.

334 H. Tazawa et al. / Thrombosis Research 136 (2015) 328–334

[11] Y.S. Li, J.H. Haga, S. Chien, Molecular basis of the effects of shear stress on vascularendothelial cells, J Biomech 38 (10) (2005) 1949–1971.

[12] T. Desmettre, C.A. Maurage, S. Mordon, Heat shock protein hyperexpression onchorioretinal layers after transpupillary thermotherapy, Invest Ophthalmol Vis Sci42 (12) (2001) 2976–2980.

[13] K. Aoki, A. Yoshino, Y. Ueda, T. Urano, A. Takada, Severe heat stroke associated withhigh plasma levels of plasminogen activator inhibitor 1, Burns 24 (1) (1998) 74–77.

[14] C.R. Jacobs, C.E. Yellowley, B.R. Davis, Z. Zhou, J.M. Cimbala, H.J. Donahue, Differentialeffect of steady versus oscillating flow on bone cells, J Biomech 31 (11) (1998)969–976.

[15] X. Zhang, P. Jones, S.J. Haswell, Attachment and detachment of living cells on mod-ifiedmicrochannel surfaces in a microfluidic-based lab-on-a-chip system, Chem EngJ 135 (2008) S82–S88.

[16] L. Jornot, M.E. Mirault, A.F. Junod, Differential expression of hsp70 stress proteins inhuman endothelial cells exposed to heat shock and hydrogen peroxide, Am J RespirCell Mol Biol 5 (3) (1991) 265–275.

[17] H. Fukao, Y. Hagiya, S. Ueshima, K. Okada, T. Takaishi, S. Izaki, et al., Effect of heatshock on the expression of urokinase-type plasminogen activator receptor inhuman umbilical vein endothelial cells, Thromb Haemost 75 (2) (1996) 352–358.

[18] N. Nakabe, S. Kokura, M. Shimozawa, K. Katada, N. Sakamoto, T. Ishikawa, et al., Hy-perthermia attenuates TNF-alpha-induced up regulation of endothelial cell adhesionmolecules in human arterial endothelial cells, Int J Hyperthermia 23 (3) (2007)217–224.

[19] S.L. DeMeester, J.P. Cobb, R.S. Hotchkiss, D.F. Osborne, I.E. Karl, K.W. Tinsley, et al.,Stress-induced fractal rearrangement of the endothelial cell cytoskeleton causes ap-optosis, Surgery 124 (1998) 362–371.

[20] Iu.A. Romanov, N.V. Kabaeva, L.B. Buravkova, Morphological and functional study ofhuman endothelium response to hyperthermia in vitro, Aviakosm Ekolog Med 35(2001) 40–46.

[21] B.R. Binder, G. Christ, F. Gruber, N. Grubic, P. Hufnagl, M. Krebs, et al., Plasminogenactivator inhibitor 1: physiological and pathophysiological roles, News Physiol Sci17 (2002) 56–61.

[22] E.K. Kruithof, Regulation of plasminogen activator inhibitor type 1 gene expressionby inflammatory mediators and statins, Thromb Haemost 100 (6) (2008) 969–975.

[23] J. Ando, K. Yamamoto, Vascular mechanobiology: endothelial cell responses to fluidshear stress, Circ J 73 (11) (2009) 1983–1992.

[24] L. Iacoviello, V. Kolpakov, L. Salvatore, C. Amore, G. Pintucci, G. de Gaetano, et al.,Human endothelial cell damage by neutrophil-derived cathepsin G. Role ofcytoskeleton rearrangement and matrix-bound plasminogen activator inhibitor-1,Arterioscler Thromb Vasc Biol 15 (11) (1995) 2037–2046.

[25] S. Handt, W.G. Jerome, L. Tietze, R.R. Hantgan, Plasminogen activator inhibitor-1 se-cretion of endothelial cells increases fibrinolytic resistance of an in vitro fibrin clot:evidence for a key role of endothelial cells in thrombolytic resistance, Blood 87 (10)(1996) 4204–4213.

[26] R. Abderrahmani, A. François, V. Buard, G. Tarlet, K. Blirando, M. Hneino, et al., PAI-1-dependent endothelial cell death determines severity of radiation-induced intesti-nal injury, PLoS One 7 (4) (2012) e35740, http://dx.doi.org/10.1371/journal.pone.0035740.

[27] R.R. Proia, P.R. Nelson, M.J. Mulligan-Kehoe, R.J. Wagner, A.J. Kehas, R.J. Powell, Theeffect of endothelial cell overexpression of plasminogen activator inhibitor-1 onsmooth muscle cell migration, J Vasc Surg 36 (1) (2002) 164–171.

[28] D. Seiffert, G. Ciambrone, N.V. Wagner, B.R. Binder, D.J. Loskutoff, The somatomedinB domain of vitronectin. Structural requirements for the binding and stabilizationof active type 1 plasminogen activator inhibitor, J Biol Chem 269 (4) (1994)2659–2666.

[29] T.C. Wun, M.O. Palmier, N.R. Siegel, C.E. Smith, Affinity purification of activeplasminogen activator inhibitor-1 (PAI-1) using immobilized anhydrourokinase.Demonstration of the binding, stabilization, and activation of PAI-1 by vitronectin,J Biol Chem 264 (14) (1989) 7862–7868.

[30] A.T. Mairuhu, T.E. Setiati, P. Koraka, C.E. Hack, A. Leyte, S.M. Faradz, et al., IncreasedPAI-1 plasma levels and risk of death from dengue: no association with the 4G/5Gpromoter polymorphism, Thromb J 3 (2005) 17, http://dx.doi.org/10.1186/1477-9560-3-17.

[31] D. Sosothikul, P. Seksarn, S. Pongsewalak, U. Thisyakorn, J. Lusher, Activation of en-dothelial cells, coagulation and fibrinolysis in children with Dengue virus infection,Thromb Haemost 97 (4) (2007) 627–634, http://dx.doi.org/10.1160/TH06-02-0094.

[32] K. Djamiatun, S.M. Faradz, T.E. Setiati, M.G. Netea, A.J. van der Ven, W.M. Dolmans, JTrop Pediatr 57 (6) (2011) 424–432, http://dx.doi.org/10.1093/tropej/fmq122.

[33] A.K. McElroy, B.R. Erickson, T.D. Flietstra, P.E. Rollin, S.T. Nichol, J.S. Towner, et al.,Ebola hemorrhagic Fever: novel biomarker correlates of clinical outcome, J InfectDis 210 (4) (2014) 558–566, http://dx.doi.org/10.1093/infdis/jiu088.

[34] H.W. Shyu, Y.Y. Lin, L.C. Chen, Y.F. Wang, T.M. Yeh, S.J. Su, et al., The dengue virus en-velope protein induced PAI-1 gene expression via MEK/ERK pathways, ThrombHaemost 104 (6) (2010) 1219–1227, http://dx.doi.org/10.1160/TH10-05-0302.

[35] J. Davignon, P. Ganz, Role of endothelial dysfunction in atherosclerosis, Circulation109 (23 Suppl. 1) (2004) III27–III32.

[36] D.J. Hicklin, Ellis LM Role of the vascular endothelial growth factor pathway in tumorgrowth and angiogenesis, J Clin Oncol 23 (5) (2005) 1011–1027.

[37] F. Cosentino, T.F. Lüscher, Endothelial dysfunction in diabetes mellitus, J CardiovascPharmacol 32 (Suppl. 3) (1998) S54–S61.