Ah

MNC

a

ARRAA

KBMNPE

1

fsmeeitcmf2fttea

0d

Biosensors and Bioelectronics 25 (2009) 68–75

Contents lists available at ScienceDirect

Biosensors and Bioelectronics

journa l homepage: www.e lsev ier .com/ locate /b ios

novel biofuel cell harvesting energy from activateduman macrophages

iho Sakai, Andreas Vonderheit, Xun Wei, Claudia Küttel, Andreas Stemmer ∗

anotechnology Group, Department of Mechanical and Process Engineering, ETH-Zurich,H-8092 Zurich, Switzerland

r t i c l e i n f o

rticle history:eceived 9 April 2009eceived in revised form 28 May 2009ccepted 2 June 2009vailable online 10 June 2009

eywords:iofuel cell

a b s t r a c t

Macrophage phagocytosis activates NADPH oxidase, an electron transport system in the plasma mem-brane. This study examined the feasibility of utilizing electrons transferred through the plasma membranevia NADPH oxidase to run a biofuel cell. THP-1 human monocytic cells were chemically stimulated todifferentiate into macrophages. Further they were activated to induce a phagocytic response. Duringdifferentiation, cells became adherent to a plain gold electrode which was used as anode in a two-compartment fuel cell system. The current production in the fuel cell always corresponded to the NADPHoxidase activity, which was evaluated by the amount of superoxide anion produced upon stimulation in

acrophageADPH oxidasehagocytosislectron transfer

combination with the expression levels of the different NADPH oxidase subunits in cells. Moreover, ourresults of different inhibitory tests let us conclude that (i) the current observed in the fuel cell originatesfrom NADPH oxidase in activated macrophages and (ii) there are multiple electron transport pathwaysfrom the cells to the electrode. One pathway involves superoxide anions produced upon stimulation,additional not yet identified electron transport occurs independently of superoxide anions.This type ofnovel biofuel cell driven by living human cells may eventually develop into a battery replacement forsmall medical devices.

. Introduction

Conventional batteries in implantable medical devices oftenorm the biggest and heaviest part and even may contain toxic sub-tances. Many attempts have been undertaken to lighten and toiniaturize the batteries, to improve their biocompatibility, and to

ventually eliminate batteries at all. The latter could be realized, forxample, in enzymatic fuel cells, where typically glucose, presentn the blood abundantly, is used as a substrate for catalytic reac-ion (Mano et al., 2002; Katz and Willner, 2003). Wiring the redoxenter of the enzymes to the electrodes not only has eliminated theembrane from the system but also facilitated the charge transport

rom enzyme to electrode resulting in higher current yield (Heller,004). When implanted in the human body, however, this type ofuel cell suffers from a short lifetime due to biofouling from adsorp-ion of biomolecules or degradation of materials including enzymes

hrough the inflammatory response of the body, which includesncapsulation and phagocytosis (Wisniewski et al., 2000; Wilsonnd Gifford, 2005; Kim et al., 2006).

∗ Corresponding author. Tel.: +41 44 632 4572; fax: +41 44 632 1278.E-mail address: astemmer@ethz.ch (A. Stemmer).

956-5663/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.bios.2009.06.005

© 2009 Elsevier B.V. All rights reserved.

During the inflammatory response leukocytes are recruited tosites of pathogens and activated to synthesize radicals, primar-ily superoxide anion through nicotinamide adenine dinucleotidephosphate-oxidase (NADPH oxidase) (Brozna et al., 1988). NADPHoxidase is a transmembrane enzyme transporting electrons acrossthe plasma membrane (Ly and Lawen, 2003). NADPH oxidase canbe activated by diverse intercellular signalling species.

The correlation between electrical energy and biological func-tion of membrane proteins has been known since the first half of thelast century, when the existence of a membrane potential was con-firmed (Kamada, 1934; Hodgkin and Huxley, 1952). Revealing thecoupling of electron and hydrogen transport across the membraneand the production of biological energy (Mitchell, 1961) provokedthe discovery of different transmembrane electron transport mech-anisms that involve membrane proteins, including NADPH oxidase.

Utilizing electrons transferred to extracellular space by NADPHoxidase in activated leukocytes would mark a step forward towarda novel type of a biofuel cell operated with human cells directly.Moreover, such type of biofuel cell would take advantage of the

cells’ ability to keep synthesizing fresh enzymes.

In this paper, we introduce a biofuel cell run by electricalenergy harvested by activated monocyte-derived macrophages. Wedemonstrate the correlation between current production in thefuel cell and NADPH oxidase activity and prove that the natural

nd Bioelectronics 25 (2009) 68–75 69

hh

2

2

sG(

Wptm2HimPwpo

2

h2bwl

ttmmapw(

2powbcS

2

Hw1a(2t3cPer

Fig. 1. Schematic setup of fuel cell. Cells differentiated in the media with chemicalstimuli adhered to a gold electrode (inset: light microscopy image of differentiatedcells in a culture well). The electrode was transferred to the fuel cell setup to serve

M. Sakai et al. / Biosensors a

ost defence mechanism allows extraction of electrical energy fromuman living cells.

. Materials and methods

.1. Cell culture and cell differentiation/activation

Human THP-1 monocytes were purchased from DSMZ (Braun-chweig, Germany) and maintained in RPMI1640 Media withlutamax (Invitrogen) supplemented with 10% fetal calf serum

FCS; Invitrogen) in a humidified atmosphere of 5% CO2 at 37 ◦C.4 × 105 THP-1 cells/ml were seeded either in culture wells (for

estern blot analysis and measurement of super oxide anionroduction) or on gold electrodes in custom-made incubation con-ainers (for current measurement in a fuel cell) in FCS containing

edia. To differentiate the cells 300 ng/ml LPS (Sigma–Aldrich),0 ng/ml Recombinant Human TNF-� and 20 ng/ml Recombinantuman IFN-� (both Invitrogen) were added, followed by a 2-day

ncubation. Subsequently cells were activated by 50 nM phorbolyristate acetate (PMA, Sigma–Aldrich). For Western blot analysis,

MA was added directly to incubation media. Otherwise the mediaere replaced with Hanks balanced salt solution (HBSS) withouthenol red (Invitrogen) before cell activation to avoid disturbancef spectrophotometrical or electrical signals.

.2. Western blot analysis

After four hours of incubation with PMA, cells were collected andomogenized in a 2× SDS loading buffer (160 mM Tris–HCl (pH 6.8),0% glycerol, 4% SDS, 10% �-mercaptoethanol, 0.04% bromophenollue). The cells for negative control were seeded simultaneouslyith the cells to be stimulated. When the treated cells were col-

ected, the un-stimulated cells were also collected for lysis.Samples were separated by 12% SDS-PAGE and subsequently

ransferred to a PVDF membrane. The blots were blocked by incuba-ion for 1 h in PBS-T (1× PBS, 0.2% Tween 20) containing 5% nonfat

ilk then incubated at 4 ◦C overnight with the appropriate pri-ary antibody in PBS-T at the following dilutions: anti-p22phox and

nti-p47phox (both Santa Cruz Biotechnology, Inc.) at 1:1000, anti-67phox (BD Biosciences) at 1:500. The same blots were probedith anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH)

HyTest Ltd., 1:20,000) for loading control.The blots were washed with TBS-T (1× TBS, 0.2% Tween

0), then incubated for 1.5 h with the appropriate horseradisheroxidase-conjugated and alkaline phosphatase-conjugated sec-ndary antibody (Promega) at a 1:10,000 dilution. After washingith TBS-T, the blots were visualized either with ECL Western

lotting detection reagents (Amersham Bioscience) or CDP-Starhemiluminescence substrate (Roche Diagnostics) and exposed touper RX x-ray film (Fujifilm).

.3. Measurement of superoxide anion production and inhibitors

After cell differentiation the media were replaced withBSS and, for inhibitory tests, the cells were pre-incubatedith inhibitors for half an hour. The inhibitors used were

00 U/ml superoxide dismutase (SOD), 100 �M N�-nitro-l-rginine methylester (L-NAME), 5 �M diphenylene iodoniumDPI) and 10 nM Staurosporine (all Sigma–Aldrich). Then 100 �M-(4-iodophenyl)-3-(4-nitrophenyl)-5-(2,4-dis-ulfophenyl)-2Hetrazolium, monosodium salt (WST-1) (Dojindo Laboratories),

00 ng/ml LPS, 20 ng/ml TNF-� and 20 ng/ml IFN-� (equivalentoncentrations as for cell differentiation), and subsequently 50 nMMA were added. The concentration of inhibitors represented theffective amount necessary for maximum inhibition of WST-1eduction by superoxide anion produced by the cells.

as anode. The anodic compartment was filled with HBSS containing differentiationstimuli, the cathodic compartment with potassium ferricyanide in HBSS. The twocompartments were separated by a cation exchange membrane. During currentmeasurements PMA as an activator was added to anodic compartment.

After four hours of incubation, the supernatants were collectedand the superoxide anion production was determined by measuringthe amount of reduced WST-1 with a spectrophotometer (Bio-RadLaboratories, Inc.) at 450 nm.

2.4. Fuel cell construction and operation

The two-compartment fuel cell (Fig. 1) was constructed withplexiglass elements that were bolted together. The anodic and thecathodic compartment had a volume of 6 ml each. The compart-ments were separated by a cation exchange membrane (BDH 55165,VWR International Ltd.) that was soaked in HBSS prior to use.

Electrodes were fabricated by depositing a 100 nm thick goldlayer (with a 10 nm chromium adhesion layer) on glass slides by e-beam evaporation. The surface area of the electrodes for cell seedingwas 4 cm2.

Cells were seeded on the electrode as mentioned above in Sec-tion 2.1. After 2 days of incubation the cells were differentiated andadhered to the electrode. This electrode was transferred into theanodic compartment just before the current measurement.

The fluid in the anode compartment was HBSS with LPS, TNF-�and IFN-� in equivalent concentrations as for cell differentiation,while the cathode compartment was filled with HBSS containing0.1 M potassium ferricyanide. The electric circuit was closed withcopper cables that had been attached to the electrodes with sil-ver paint, with the junctions being sealed with silicone paste. Toprevent evaporation of fluid from the fuel cell, the setup was cov-ered with an aluminum foil housing with a wet paper towel placedinside. The measurements were carried out at room temperature.The current was calculated from the voltage drop measured across aresistor with a digital multimeter (2000, Keithley Instruments Inc.).

The individual electrode potentials were determined using anAg/AgCl and a calomel reference electrode for the anode and thecathode, respectively, which have been connected via salt bridges.

For current measurements the activator PMA was added afterthe current stabilized and the inhibitors were added after a clearincrease in current.

3. Results and discussion

To investigate the feasibility of our setup to be used as afunctional fuel cell, we monitored the electrical current after thechemically stimulated cells were placed into the anode compart-

70 M. Sakai et al. / Biosensors and Bioelectronics 25 (2009) 68–75

Fig. 2. Current generation in the fuel cell. (A) Current flow without cells on electrode, (B–H) with cells adhering to electrode. (A) The measurement was started with HBSSa int ofo l TNFa (F) L-No e activ

mtowp

oph

lready containing LPS, TNF-� and IFN-� in the anodic compartment. At the time ponly activated with 50 nM PMA; (C) only differentiated with 300 ng/ml LPS, 20 ng/mctivation, inhibitors were added to the anode fluid. Inhibitors used were (E) SOD,n electrodes was measured ((E–H), grey lines). Solid arrows indicate addition of th

ent. To validate whether a current production can be attributedo active NADPH oxidase in macrophages, the expression of NADPHxidase subunits was examined and the NADPH oxidase activityas evaluated via a chemical assay that measures superoxide anion

roduction.

All experiments were repeated several times using THP-1 cellsf different passage numbers. Due to the state of cells, variations inerformance were unavoidable. Nevertheless, the data presentedere are representative for the experiments conducted.

the solid arrow, the activator PMA was added to the anodic compartment. (B) Cells-� and 20 ng/ml IFN-�; (D) cells differentiated and activated sequentially. After cellAME, (G) DPI, (H) Staurosporine. For each inhibitory test the current without cellsator PMA, dashed arrows indicate addition of inhibitors.

3.1. Background current from the system

To determine the background current induced by the fluid andthe chemicals in the fuel cell, current generation in the absence

of cells on the electrode was examined employing the same pro-cedure as the current measurement with cells on the electrode.Briefly, the anodic compartment was filled with HBSS containingdifferentiation stimuli, while the cathodic compartment was filledwith HBSS containing potassium ferricyanide. After the current sta-

nd Bioelectronics 25 (2009) 68–75 71

b(gactcwwe

ictisc

3

Ntec

3N

gAvml2taooub

cPtbtbo

etotnitabl

3p

tc

Fig. 3. Expression of NADPH oxidase subunits at differentiation/activation stages ofTHP-1. From top to bottom; p22phox , p47phox , p67phox . Immunodetection of GAPDHserves as loading control. Bar graphs represent the band intensity of immunoblots,

M. Sakai et al. / Biosensors a

ilized, the activator PMA was added to the anodic compartmentFig. 2A, grey lines in Fig. 2E–H, solid arrows). To determine back-round current in the inhibitory tests, inhibitors were added to thenodic compartment (grey lines in Fig. 2E–H, dashed arrows). Theurrent dropped in the beginning due to the equilibration betweenhe two compartments (Fig. 2A, grey lines in Fig. 2E–H). This firsturrent drop was steeper compared to that in the measurementith cells on the electrode (Fig. 2B–D and grey lines in Fig. 2E–H),hich was possibly due to the lower impedance of the cell-free

lectrode.The addition of the activator as well as the addition of the

nhibitors to the anodic compartment induced no obvious currenthange. This indicates that none of the chemicals used in the sys-em, i.e., the chemicals for differentiation and the activation, thenhibitors, potassium ferricyanide and the buffer, generate a sub-tantial electrical current. Thus the effect of a possible backgroundurrent can be neglected in the experiments reported below.

.2. Current generation and superoxide anion release

To investigate the correlation between current generation andADPH oxidase activity, first the superoxide anion release from

he cells and the protein expression in the cells during differ-ntiation/activation were examined. Secondly these results wereompared with the current generation in the fuel cell.

.2.1. Superoxide anion production and expression level ofADPH oxidase subunits

Monocytes had to first differentiate into macrophages and thenet activated. To this end, THP-1 cells were stimulated in two steps.s differentiation stimuli TNF-�, IFN-� and LPS were used. To acti-ate the macrophage we used PMA. NADPH oxidase consists of twoembrane-bound subunits, gp91phox and p22phox, and four cytoso-

ic subunits, p67phox, p47phox, p40phox and RacGTP (Babior et al.,002; Bey et al., 2004; Cathcart, 2004). Upon macrophage activa-ion, the cytosolic subunits translocate to the plasma membranend bind to the membrane-bound subunits to form the NADPHxidase complex (Clark et al., 1990). We analyzed the expressionf p22phox, a membrane-bound subunit, and two cytosolic sub-nits, p47phox and p67phox on the differentiation/activation stagesy Western blot.

All subunits analyzed were expressed in low levels in restingells (Fig. 3). When the cells were only exposed to the activatorMA, the expression of both cytosolic subunits increased, whereashis activator did not have any effect on the p22phox synthesis. Inoth cases the percentage of WST-1 reduction was small, reflectinghe low superoxide anion production from the cells (Fig. 4). This cane explained as a result of the small amount of assembled NADPHxidase due to low expression of subunits (Figs. 3 and 4).

The cells exposed to the differentiation stimuli and the cellsxposed to the differentiation stimuli plus the activator in sequen-ial order both expressed an enhanced level of all three NADPHxidase subunits investigated (Fig. 3). Although the addition ofhe activator to the cells treated with differentiation stimuli didot change the amount of NADPH oxidase subunits expressed, it

ncreased the WST-1 reduction abundantly (Fig. 4). This suggestshat the NADPH subunits expressed in the differentiated cells weressembled only after adding the activator. The detail of the assem-ly mechanism of the NADPH oxidase subunits will be discussed

ater in Section 3.3.4.

.2.2. Current generation correlates to superoxide anionroduction by NADPH oxidase

Cells not treated with any chemical stimuli did not adhere tohe electrode. Hence we did not measure any current in the fuelell system. When the cells were activated by PMA without dif-

normalized to GAPDH. From left: Untreated cells as negative control; Activationwith 50 nM PMA; Differentiation with 300 ng/ml LPS, 20 ng/ml TNF-�, and 20 ng/mlIFN-�; Differentiation + Activation with sequential stimuli of same concentrationsas before.

ferentiation treatment, they adhered to the electrode, however, nocurrent was generated (Fig. 2B). When the cells were exposed tothe differentiators, the cells adhered onto the electrode as welland no current increase was observed (Fig. 2C) unless the activa-tor PMA was added into the anodic compartment (Fig. 2D). Theseresults were in agreement with the superoxide anion produced bythe cells each in differentiation/activation stages in Fig. 4. Togetherwith the correlation of superoxide anion production and NADPHoxidase activity discussed above, it is obvious that the NADPH oxi-dase activity and the current generation in the fuel cell are stronglylinked.

3.3. Sources of current generation

To investigate the source of the current generation, the effect ofdifferent inhibitors on the current generation in the biofuel cell andthe superoxide anion detection were examined, as schematicallyillustrated in Fig. 5. The results of these inhibition studies are shownin Figs. 2 and 4.

72 M. Sakai et al. / Biosensors and Bioe

Fig. 4. WST-1 reduction upon differentiation/activation of THP-1 cells and in pres-ence of inhibitors. WST-1 reduction refers to the spectrophotometric absorbancenormalized to the value of fully stimulated (differentiated and activated) cells in eachexperiment. From left: Untreated cells as negative control; Activation with 50 nMPMA; Differentiation with 300 ng/ml LPS, 20 ng/ml TNF-�, and 20 ng/ml IFN-�; Dif-ferentiation + Activation with sequential stimuli of same concentration as before.Ft5

3

tc(a

dase is linked to the production of nitric oxide through inducible

FtaS

or inhibition experiments, cells were preincubated with inhibitors after differen-iation but before activation. Inhibitors used were 100 U/ml SOD, 100 �M L-NAME,�M DPI, and 10 nM Staurosporine.

.3.1. Superoxide anion and its derivativesThe electron transferred by NADPH oxidase is usually cap-

ured by oxygen to form superoxide anion (Babior, 1999). SODatalyses the reaction of superoxide anion to hydrogen peroxide2O2

•− + 2H+ → O2 + H2O2) and decreases the amount of superoxidenion rapidly. Therefore SOD is commonly used to confirm super-

ig. 5. Schematic illustration of NADPH oxidase assembly and inhibition routes. Upon dhe cytosolic subunits (p67phox , p47phox , p40phox and RacGTP) are phosphorylated, assemblssembled NADPH oxidase oxidizes NADPH in the cytosol to NADP+ and electrons are trataurosporine is also indicated.

lectronics 25 (2009) 68–75

oxide anion detected. At physiologic pH the rate constant of thereaction mentioned above with SOD is ≈2 × 109 M−1 s−1 (Michelsonet al., 1977), whereas the rate constant of the reaction betweensuperoxide anion and WST-1 is ≈3–4 × 104 M−1 s−1 (Ukeda, 2004).Thus the addition of SOD shortens the lifetime of superoxide anionand prevents almost all WST-1 reduction by the superoxide anion.

Our results confirmed that WST-1 reduction was stronglylowered when SOD was added (10% compared to 100% in Dif-ferentiation + Activation, Fig. 4). Also the current in the fuel celldecreased on addition of SOD (≈50% compared to maximum cur-rent, Fig. 2E), indicating a participation of the superoxide anionin the observed current production. However, in the experimentswith cells treated with differentiation and activation stimuli, thecurrent only decreased to 50% of the maximum (Fig. 2E), whereasthe amount of WST-1 reduction decreased to 10% of the maxi-mum (Fig. 4). Since hydrogen peroxide, the product of the catalyticreaction, can oxidize or reduce a variety of inorganic ions in aque-ous solution, hydrogen peroxide could contribute to the remainingcurrent. To examine the contribution of hydrogen peroxide to thecurrent flow, catalase was added to the anodic compartment afterSOD, to catalyze the reaction of hydrogen peroxide to oxygen. Theaddition of catalase did not cause any significant change in the cur-rent (data not shown), thus we conclude that hydrogen peroxidedoes not contribute to the current production under our experi-mental conditions.

3.3.2. Nitric oxide and its derivativesProduction of superoxide anion via the activated NADPH oxi-

nitric oxide synthase (Rosen et al., 1995). Nitric oxide can bind withsuperoxide anion to form another reactive species, peroxynitrite(Huie and Padmaja, 1993). These species are reactive and could alsoinfluence the current.

ifferentiation NADPH oxidase subunits are expressed abundantly. Upon activatione, and translocate to the membrane-bound subunits (gp91phox , p22phox). Completelynsported to the extracellular space. The action of inhibitors SOD, L-NAME, DPI, and

nd Bio

o1oaWpu

3

tpTbot

tameaattatot

3

mtta(ctdcuNsa

iiiNtacn1oTwW(grpr

M. Sakai et al. / Biosensors a

L-NAME competes against arginine, that interacts with nitricxide synthase, and inhibits the nitric oxide synthesis (Rees et al.,990). Thus L-NAME allows one to examine the effect of nitric oxider peroxynitrite on the current generation in the fuel cell. Since theddition of L-NAME did change neither the current (Fig. 2F) nor the

ST-1 reduction (Fig. 4), current production and superoxide anionroduction were not influenced by nitric oxide or peroxynitritender our experimental conditions.

.3.3. Electron transfer from NADPH oxidase subunitDPI binds to flavoproteins and suppresses reactions in which

hese proteins are involved. In phagocytotic cells DPI suppresses theroduction of superoxide anion and nitric oxide (Stuehr et al., 1991).he suppression of superoxide anion production is caused by DPIlocking the reduction of FAD in the plasma membrane componentf NADPH oxidase (Ellis et al., 1989), thereby inhibiting electronransport to the extracellular space.

The addition of DPI suppressed both the current generation andhe superoxide anion production by activated macrophages (Fig. 2Gnd Fig. 4), suggesting that superoxide anion and/or nitric oxideight take part in the current production in the fuel cell. How-

ver, according to our experimental results with L-NAME describedbove, nitric oxide does not have any influence on the superoxidenion production from the cells or on the current measurement inhe fuel cell, which proved that neither nitric oxide nor peroxyni-rite, the derivative of superoxide anion and nitric oxide, served asn electron mediator between the cells and the electrode. Hence,he current decrease upon inhibition by DPI supports an active rolef an electron transfer of NADPH oxidase in the current produc-ion.

.3.4. Assembly of NADPH oxidaseAssembly of the NADPH oxidase complex in activated

acrophages is necessary for current generation, as evidenced byhe difference in the NADPH oxidase activity in each differentia-ion/activation stages as shown in Figs. 3 and 4. PMA is a well-knownctivator of protein kinase C (PKC) known to phosphorylate p47phox

Nauseef et al., 1991; Bey et al., 2004). Without phosphorylation theytosolic subunits do not build up a complex that is translocatedogether to the plasma membrane to assemble the NADPH oxi-ase complex. When cells were differentiated but not activated, nourrent production was observed although the NADPH oxidase sub-nits were sufficiently expressed (Fig. 3). This may be because theADPH oxidase complex had not yet been assembled completely

ince the phosphorylation of p47phox is possibly first triggered uponctivation by PMA.

Staurosporine is one of the most potent and widely usednhibitors of protein kinases, and at low concentrations selectivelynhibits the activity of PKC (Yamaguchi et al., 1996). Since PKCs involved in the phosphorylation of the cytosolic subunits ofADPH oxidase, inhibition of PKC activity blocks the transloca-

ion of these subunits to the plasma membrane components. Thessembled NADPH oxidase complex is not permanently stable sinceytosolic subunits dissociate from the membrane-bound compo-ents through hydrolyzation of GTP-bound Rac (Bastian and Hibbs,994). The addition of staurosporine gradually stops the assemblyf NADPH oxidase and only dissociation of the complex continues.hus the amount of electron transfer through the NADPH oxidaseill drop with the decrease of fully assembled NADPH oxidase. TheST-1 reduction was lowered upon inhibition by staurosporine

Fig. 4) although it was not abolished completely and the currentenerated from activated macrophages decayed slowly (Fig. 2H),eflecting the kinetics of staurosporine inhibition. These findingsrove again that the assembled NADPH oxidase is essential for cur-ent production in the fuel cell.

electronics 25 (2009) 68–75 73

Consequently, we have proved that the activated NADPH oxidaseis the origin of the electron transferred to the electrode.

3.4. Electron paths between cells and the electrode

As shown in the previous section, NADPH oxidase is a sourceof the current generation in a fuel cell and superoxide anion par-ticipates in the current flow. Here we discuss how electrons aretransferred from activated macrophages to the electrode in ourbiofuel cell system. Between the cell membrane and the electrodethere are several conceivable electron pathways; (i) direct electrontransfer, (ii) indirect electron transfer through (a) superoxide anion,(b) derivatives of superoxide anion, (c) other molecules mediatingelectron transfer.

To evaluate the contribution of superoxide anion and itsderivatives to the current production, the amount of currentmeasured in our biofuel cell (1.5–2 �A/106 cells seeded) wascompared to the expected amount of current based on the super-oxide anion produced from activated macrophages. The amountof superoxide anion produced can be estimated from the spec-trophotometrical measurement of WST-1 reduction. Accordingto our control measurement of WST-1 reduction by superoxideanion in the hypoxanthine/xanthine oxidase system, only onethird of the superoxide anion produced was used for WST-1reduction (supplementary material). This is plausible since therate constant of the reaction of WST-1 with superoxide anion(≈3–4 × 104 M−1 s−1) is expected to be lower than that of spon-taneous conversion of superoxide anion to hydrogen peroxide(≈1 × 105 M−1 s−1 (Ukeda, 2004). Calculating with a representa-tive spectrophotometric absorption value of the reduced WST-1(1–1.5) and accounting for the control measurement, the super-oxide anion production of THP-1 cells in our experiments wasestimated 1.4–2 nmol/min/106 cells. This value is higher than thesuperoxide anion production of THP-1 cells reported in litera-ture (0.05–0.5 nmol/min/106 cells) (Wang et al., 2002; Almeidaet al., 2005; Lee et al., 2007). The superoxide anion detectionmethods using cytochrome c, chemiluminescence, or WST-1 areindirect measurements and due to the non-specificity of indi-cators and the reoxidation of indicators the reliability of thesedetection methods is an issue (Tarpey and Fridovich, 2001). More-over, due to the complexity of signal pathways within cells andthe difference in cell activation stimuli employed in each studythe activation degree of the cells responding to various stimulihas not been well-examined. Therefore, the direct comparisonof superoxide anion release between different reports is diffi-cult.

In our experiments, if all electrons flowing from cells to superox-ide anions were caught and transferred to the electrode, a currentof 1.6–2.4 �A/106 cells seeded would be produced in the biofuelcell, which matches roughly the current measured with activatedmacrophages in our setup (1.5–2 �A/106 cells seeded).

Whole-cell patch-clamp experiments showed that the amountof current flow from NADPH oxidase corresponds to the amount ofsuperoxide anion release from granulocytes (10–20 pA per cell cur-rent measured, 10 nmol/min/106 cells superoxide anion produced)(Schrenzel et al., 1998; DeCoursey et al., 2003). In patch-clampexperiments NADPH, oxidase substrate, or GTP-�S, nonhydrolyz-able analogue of GTP, was added to stabilize the measurements.Without adding these chemicals, less electron flow from cells wasdetected. Furthermore the yield of electron transfer from cells inpatch-clamp experiments should be in any case better than that

between cells and the anode in our biofuel cell due to lower systemimpedance. Thus, for our experimental condition, the contributionof superoxide anion to the current generation is expected to bemuch lower. However, the current measured in the biofuel cell islarger than the expected, indicating the existence of an additional

74 M. Sakai et al. / Biosensors and Bioe

Fig. 6. Current generation and the potential of the electrodes in the fuel cell. Cellswere activated with PMA. After a clear increase in current the NADPH oxidaseitcc

rs

(s(aiioa

3s

t

nhibitor DPI was added to the anodic compartment. The anode potential andhe cathode potential were measured using an Ag/AgCl reference electrode and aalomel reference electrode, respectively. The potential value of the cathode wasonverted to Ag/AgCl reference scale.

oute of electron flow from the cell to the electrode independent ofuperoxide anions.

Both SOD and DPI reduced the superoxide anion productionFig. 4), however, current suppression upon inhibition by DPI wastronger than that by SOD (Fig. 2E and G). SOD is cell-impermeableOldenborg et al., 2000) and scavenges the electron flow medi-ted by superoxide anion, whereas DPI is cell-permeable andnhibits the electron flow through the plasma membrane by bind-ng to a NADPH oxidase subunit. This also suggests the existencef an additional route of electron flow different from superoxidenion.

.5. Potential change in the anode and in the cathode duringtimulation

The current flowing through the fuel cell is driven by the poten-ial difference between the anode and the cathode. To investigate

lectronics 25 (2009) 68–75

the influence of each compartment on the current production,the potential of the anode and the cathode were determinedindependently with reference electrodes during the current mea-surement.

As shown in Fig. 6, the potential of the anode corresponded tothe change in current after stimulation by PMA and inhibition byDPI, while the cathode potential changed insignificantly (the vari-ation is less than 10% of the anode). This observation indicates thatthe current production is not substantially driven by the cathodicchemical reaction, but by the potential change at the anode throughthe activation of macrophages.

3.6. Limiting factors for current generation

Our novel biofuel cell utilizing activated macrophages as elec-tron donors produced a current of 1.5–2 �A per 106 cells seeded.This value could be improved by modifying materials and experi-mental environment.

In our current setup the buffer in the anodic compartment hadcontact with ambient air and thus contained a certain amount ofdissolved oxygen (≈8 mg/l) (Tölgyessy et al., 1988). Generally, theanolyte should be deoxygenated, since oxygen is a very strong elec-tron acceptor that reacts against the anode. However, mammaliancells need aerobic condition to survive. Therefore, the maximiza-tion of the electron flow from cells to the anode has to be achievedin the presence of oxygen.

By employing a cation exchange membrane to form a two-compartment fuel cell, a variety of oxidizers can be selected aselectron acceptors, since their competition with the anodic com-partment can be neglected. In general, however, the existence of amembrane in a fuel cell limits proton transport between the com-partments, which hinders the electron flow from the anode to thecathode. Additionally this low proton transport causes accumula-tion of acidity in the anode over time, inducing an unfavourableenvironment for cell physiology and suppressing NADPH oxidaseactivity through lowering intracellular pH (Morgan et al., 2005).Nevertheless, until higher current levels are reached, the voltageloss resulting from the membrane and the acidification of theanodic compartment can be neglected.

Finally, all experiments were carried out at ambient condition.Since cells are best maintained at 37 ◦C in 5% CO2, the experimentalenvironment could be further improved to prolong cell activity. Weassume that the decline of the current (Fig. 3D) was caused by theloss of cell activity or viability after a few hours.

4. Conclusion

We have demonstrated first results confirming the feasibilityof harvesting electricity from human cells, namely macrophages.To this end we took advantage of host defence mechanisms,where monocytes undergo differentiation into macrophages andadhere to the anode in a fuel cell setup. Activation and inhibi-tion of NADPH oxidase elucidated its role in transferring electronsthrough the plasma membrane of macrophages. With plain goldelectrodes the extracted current reached 1.5–2 �A per 106 cellsseeded in the mediator-less two-compartment fuel cell system.The use of a mediator facilitates the electron transport fromelectron donor to the electron acceptor. Improved current yieldin microbial biofuel cells and enzymatic fuel cells was success-fully demonstrated by employing mediators in solution (Kim

et al., 2000; Park and Zeikus, 2000) or immobilized on elec-trode surfaces (Kano and Ikeda, 2000; Katz and Willner, 2003;Calabrese Barton et al., 2004). Similarly, modifying the elec-trode material and surface is likely to lead to higher current aswell.

nd Bio

A

Aab

A

t

R

A

BB

BB

B

C

C

C

DEHHKH

plasty 17 (3), 335–346.

M. Sakai et al. / Biosensors a

cknowledgements

We thank the members of the lab of Prof. Ari Helenius and Dr.xel Niemann, ETH Zurich, for their courtesy and assistance, andlso offer special thanks to Dr. Laurence Vindevogel for her contri-ution when starting this research.

ppendix A. Supplementary data

Supplementary data associated with this article can be found, inhe online version, at doi:10.1016/j.bios.2009.06.005.

eferences

lmeida, A.C., Rehder, J., Severino, S.D., Martins-Filho, J., Newburger, P.E., Condino-Neto, A., 2005. Journal of Interferon & Cytokine Research 25 (9), 540–546.

abior, B.M., 1999. Blood 93 (5), 1464–1476.abior, B.M., Lambeth, J.D., Nauseef, W., 2002. Archives of Biochemistry and Bio-

physics 397 (2), 342–344.astian, N.R., Hibbs, J.B., 1994. Current Opinion in Immunology 6 (1), 131–139.ey, E.A., Xu, B., Bhattacharjee, A., Oldfield, C.M., Zhao, X., Li, Q., Subbulakshmi, V.,

Feldman, G.M., Wientjes, F.B., Cathcart, M.K., 2004. The Journal of Immunology173 (9), 5730–5738.

rozna, J.P., Hauff, N.F., Phillips, W.A., Johnston, R.B., 1988. Journal of Immunology141 (5), 1642–1647.

alabrese Barton, S., Gallaway, J., Atanassov, P., 2004. Chemical Reviews 104 (10),4867–4886.

athcart, M.K., 2004. Arteriosclerosis, Thrombosis, and Vascular Biology 24 (1),23–28.

lark, R.A., Volpp, B.D., Leidal, K.G., Nauseef, W.M., 1990. Journal of Clinical Investi-gation 85 (3), 714–721.

eCoursey, T.E., Morgan, D., Cherny, V.V., 2003. Nature 422 (6931), 531–534.llis, J.A., Cross, A.R., Jones, O.T., 1989. Biochemical Journal 262 (2), 575–579.eller, A., 2004. Physical Chemistry Chemical Physics 6 (2), 209–216.uie, R.E., Padmaja, S., 1993. Free Radical Research Communications 18 (4), 195–199.ano, K., Ikeda, T., 2000. Analytical Sciences 16 (10), 1013–1021.odgkin, A.L., Huxley, A.F., 1952. Journal of Physiology - London 117 (4), 500–544.

electronics 25 (2009) 68–75 75

Kamada, T., 1934. Journal of Experimental Biology 11 (1), 94–102.Katz, E., Willner, I., 2003. Journal of the American Chemical Society 125 (22),

6803–6813.Kim, J., Jia, H., Wang, P., 2006. Biotechnology Advances 24 (3), 296–308.Kim, N., Choi, Y., Jung, S., Kim, S., 2000. Biotechnology and Bioengineering 70 (1),

109–114.Lee, J.S., Nauseef, W.M., Moeenrezakhanlou, A., Sly, L.M., Noubir, S., Leidal, K.G., Schlo-

mann, J.M., Krystal, G., Reiner, N.E., 2007. Journal of Leukocyte Biology 81 (6),1548–1561.

Ly, J.D., Lawen, A., 2003. Redox Report 8 (1), 3–21.Mano, N., Mao, F., Heller, A., 2002. Journal of the American Chemical Society 124

(44), 12962–12963.Michelson, A.M., Mccord, J.M., Fridovich, I. (Eds.), 1977. Superoxide and Superoxide

Dismutases. Academic Press.Mitchell, P., 1961. Nature 191 (478), 144.Morgan, D., Cherny, V.V., Murphy, R., Katz, B.Z., DeCoursey, T.E., 2005. Journal of

Physiology-London 569 (2), 419–431.Nauseef, W.M., Volpp, B.D., McCormick, S., Leidal, K.G., Clark, R.A., 1991. The Journal

of Biological Chemistry 266 (9), 5911–5917.Oldenborg, P.A., Sundqvist, I.M., Sehlin, J., 2000. Diabetic Medicine 17 (7), 532–537.Park, D.H., Zeikus, J.G., 2000. Applied and Environmental Microbiology 66 (4),

1292–1297.Rees, D.D., Palmer, R.M.J., Schulz, R., Hodson, H.F., Moncada, S., 1990. British Journal

of Pharmacology 101 (3), 746–752.Rosen, G.M., Pou, S., Ramos, C.L., Cohen, M.S., Britigan, B.E., 1995. The FASEB Journal

9 (2), 200–209.Schrenzel, J., Serrander, L., Banfi, B., Nusse, O., Fouyouzi, R., Lew, D.P., Demaurex, N.,

Krause, K.-H., 1998. Nature 392 (6677), 734–737.Stuehr, D.J., Fasehun, O.A., Kwon, N.S., Gross, S.S., Gonzalez, J.A., Levi, R., Nathan, C.F.,

1991. The FASEB Journal 5 (1), 98–103.Tarpey, M.M., Fridovich, I., 2001. Circulation Research 89 (3), 224–236.Tölgyessy, P., Büchlerová, E., Brtko, J., 1988. Journal of Radioanalytical and Nuclear

Chemistry 126 (2), 139–143.Ukeda, H., 2004. Dojin News 112, 1.Wang, M.L., Hauschka, P.V., Tuan, R.S., Steinbeck, M.J., 2002. The Journal of Arthro-

Wilson, G.S., Gifford, R., 2005. Biosensors & Bioelectronics 20 (12), 2388–2403.Wisniewski, N., Moussy, F., Reichert, W.M., 2000. Fresenius Journal of Analytical

Chemistry 366 (6–7), 611–621.Yamaguchi, M., Saeki, S., Yamane, H., Okamura, N., Ishibashi, S., 1996. Biochemical

and Biophysical Research Communications 220 (3), 891–895.