Injectable hydrogel microbeads for fluorescence-based in vivo continuous glucose monitoringHideaki Shibataa,b,c,1, Yun Jung Heoa,b,1, Teru Okitsua,d, Yukiko Matsunagaa,b, Tetsuro Kawanishia,c, and Shoji Takeuchia,b,2

aLife Bio Electromechanical Autonomous Nano Systems (BEANS) Center, BEANS Project, University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo 153-8505,Japan; bInstitute of Industrial Science, University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo 153-8505, Japan; cTERUMO Co. R and D Headquarters,1500 Inokuchi, Nakai-machi, Ashigarakami-gun, Kanagawa 259-0151, Japan; and dTransplant Unit, Kyoto University Hospital, 54 Shogoin-kawaharacho,Sakyo-ku, Kyoto 606-8507, Japan

Edited by Mark E. Davis, California Institute of Technology, Pasadena, CA, and approved August 17, 2010 (received for review May 20, 2010)

Fluorescent microbeads hold great promise for in vivo continuousglucose monitoring with wireless transdermal transmission andlong-lasting activity. The full potential of fluorescent microbeadshas yet to be realized due to insufficient intensity for transdermaltransmission and material toxicity. This paper illustrates the highly-sensitive, biostable, long-lasting, and injectable fluorescent micro-beads for in vivo continuous glucose monitoring. We synthesizeda fluorescent monomer composed of glucose-recognition sites, afluorogenic site, spacers, and polymerization sites. The spacersaredesigned tobe longandhydrophilic for increasingopportunitiesto bindglucosemolecules; consequently, the fluorescentmonomersenable high-intensive responsiveness to glucose. We then fabri-cated injectable-sized fluorescent polyacrylamide hydrogel beadswith highuniformity andhigh throughput.We found that our fluor-escent beads provide sufficient intensity to transdermally monitorglucose concentrations in vivo. The fluorescence intensity success-fully traced the blood glucose concentration fluctuation, indicatingour method has potential uses in highly-sensitive and minimallyinvasive continuous blood glucose monitoring.

fluorescent hydrogel ∣ microfluidics ∣ diboronic acid ∣ continuous glucosemonitoring ∣ diabetes mellitus

Diabetes is a global pandemic affecting over 200 million peo-ple (1, 2). Maintaining normal blood glucose concentrations

is crucial for preventing diabetic complications in the heart,kidney, retina, and neural system (3–6). The fingertip prick meth-od for collecting a blood sample is used at present to accuratelyanalyze blood glucose concentrations. However, the methodprovides intermittent information concerning glucose concentra-tions, which is not suitable to predict the trend of blood glucosechange. In contrast, continuous glucose monitoring (CGM)allows diabetic patients to effortlessly recognize changes in bloodglucose concentrations and signals a warning in the case of hypo-and hyperglycemia patients; even when diabetic patients aresleeping (7, 8).

Fully-implantable glucose sensors, embedded in the body, areideal for CGM. Previously, microdialysis and enzyme-tippedcatheters have been developed as fully-implantable glucose sen-sors for CGM. Although these sensors are capable of providingthe sequential information of blood glucose concentrations todiabetic patients, they need to have external links for continu-ously collecting samples or transmitting signals; thereby thesemethods cause discomfort and the risk of infection. Recently,an optical method using fluorescent beads (9–12) was proposedfor CGM. This method provides wireless transmission throughthe skin, and long-lasting activity in vivo compared to enzyme-based methods (13–16) that require an electrochemical reaction.However, fully-implantable glucose sensors based on the fluores-cent principle have not yet been developed mainly due to theinsufficient fluorescent intensity required for transdermal detec-tion and the toxicity of the material (17).

Here, we developed the highly-sensitive, biostable, long-lasting, and injectable fluorescent microbeads for in vivo CGM

(Fig. 1) to solve the aforementioned problems. We designed andsynthesized a unique fluorescent monomer based on diboronicacid that enables reversible responsiveness to glucose withoutany reagents and enzymes. The fluorescent monomer has long,hydrophilic spacers and polymerization sites to bind flexible sup-ports. As a result, the fluorescent monomer shows high mobilityoriginating from the increase in opportunities to contact glucosemolecules. Therefore, the fluorescent monomer has sufficientintensity for in vivo transdermal monitoring; even when it isimmobilized in a solid support. Then, we immobilized the fluor-escent monomer in microbeads. Due to the virtue of their smallsize, the fluorescent microbeads are injectable, minimally inva-sive, and rapidly respond to glucose change. By employing micro-fluidic devices, we succeeded in fabricating the microbeads ofthe order of 1 × 102μm with high throughput and high unifor-mity. Then, we experimentally verified in vitro and in vivo glucose

Fig. 1. A schematic illustration of the injectable fluorescent microbeadsfor in vivo CGM. The fluorescence intensity of the microbeads increases asglucose concentration increases. This reversible, powerless, and highly-sensi-tive reaction allows the fluorescent microbeads to continuously monitorglucose concentrations in vivo with long-lasting activity and transdermaltransmission. The fluorescent microbeads are simply injected under thedermis using a needle, minimizing pain and damage to tissues.

Author contributions: H.S., Y.J.H., T.O., Y.M., T.K., and S.T. designed research; H.S., Y.J.H.,T.O., and Y.M. performed research; T.K. contributed new reagents/analytic tools; H.S.,Y.J.H., Y.M., and S.T. analyzed data; and H.S., Y.J.H., T.O., Y.M., T.K., and S.T. wrotethe paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1H.S. and Y.J.H. contributed equally to this work.2To whom correspondence should be addressed. E-mail: takeuchi@iis.u-tokyo.ac.jp.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1006911107/-/DCSupplemental.

17894–17898 ∣ PNAS ∣ October 19, 2010 ∣ vol. 107 ∣ no. 42 www.pnas.org/cgi/doi/10.1073/pnas.1006911107

Dow

nloa

ded

by g

uest

on

May

10,

202

0

responsiveness of the fluorescent microbeads to show the poten-tial of applying to in vivo CGM.

Results and DiscussionDesign and Synthesis of Glucose-Responsive Monomer. Our glucose-responsive fluorescent monomer (GF-monomer) was modifiedfrom previously developed glucose-responsive fluorescent dye(GF-dye) (18). The glucose sensing principle of GF-dye is brieflyexplained as follows: GF-dye comprises of diboronic acid moietyand an anthracene moiety that act as the specific glucose-recogni-tion site and the fluorogenic site, respectively. In the absence ofglucosemolecules, the fluorescence of the anthracene is quenchedby a photo-induced electron transfer (PET) that occurs from theunshared electron pair of the nitrogen atom to the anthracene.When glucose molecules bind to the diboronic acid, a strongreaction between the nitrogen atom and boron atom inhibitsPET. As a result, the fluorescence of anthracene becomes higher(Fig. 2A) compared to glucose-free condition (18–21). In addition,the diboronic acid moiety has a high selectivity of glucose due tothe more than 10 times stronger reaction between the glucose-recognition site and the glucose compared to other sugars (22).

To apply fully-implantable glucose sensors, GF-dye needs to beimmobilized to a solid support. However, the fluorescence inten-sity of GF-dye decreased 14 times after immobilization due to thedecrease in the mobility of GF-dye (18); the decrease in mobilityinduced the decrease in opportunities to bind glucose molecules.Here, we synthesized GF-monomer from GF-dye to have better

mobility even when GF-monomer is immobilized. There aretwo design challenges: a long spacer and a polymerization sitethat binds to a sparse and flexible immobilization material. Weemployed polyethylene glycol (PEG) and acrylamide (AAm)group as the spacers and the polymerization site, respectively.Advantages of using PEG was that PEG is not only long (23)(Mw ¼ 3;400) but also hydrophilic (24). Thus, the PEG spacerallows better contact between theGF-monomer andwater-solubleglucose molecules. Meanwhile, AAm group facilitates the poly-merization of polyacrylamide (PAAm) gel that is a malleable,flexible hydrogel, with excellent water-binding capabilities (hydro-philicity) and biocompatibility (25). The sparse matrix and hydro-philic characteristics of PAAmgel contribute to the highly sensitiveglucose responsiveness of GF-monomer even when immobilizedin hydrogel scaffolds. Moreover, the biocompatibility of PAAmgel makes it possible to carry out in vivo application.

As described in Materials and Methods, GF-monomer wassynthesized in a way that met the aforementioned requirements.(Fig. 2B). First, PEG with acrylamide group and carboxyl group(AAm-PEG-COOH) was prepared by coupling amino-PEG-car-boxyl acid (NH2-PEG-COOH) and acrylyl chloride. Then, thedouble amino terminated the fluorescent dye (GF-dye) and theprepared AAm-PEG-COOH were coupled in the presence ofdehydration-condensation agent to obtain GF-monomer. Glu-cose-responsive fluorescent hydrogel (GF-hydrogel) (Fig. 2C)was obtained by polymerization of GF-monomer with AAmmonomers and cross linkers (see SI Text and Fig. S1).

Fig. 2. Glucose sensing principle and synthetic scheme of the fluorescent monomer (GF-monomer) and hydrogel (GF-hydrogel). (A) Fluorescence intensitychanges depending on the existence of a glucose molecule. (B) GF-monomer is obtained by coupling the diboronic dye (GF-dye) and long, hydrophilic spacerwith polymerization group (AAm-PEG-COOH). (C) GF-monomer is immobilized in polyacrylamide hydrogels having a sparse and flexible matrix.

Shibata et al. PNAS ∣ October 19, 2010 ∣ vol. 107 ∣ no. 42 ∣ 17895

ENGINEE

RING

Dow

nloa

ded

by g

uest

on

May

10,

202

0

Fabrication of the Fluorescent Microbeads. We fabricated glucose-responsive fluorescent microbeads (GF-beads), using a axisym-metric flow-focusing microfluidics device (AFFD) (26–28) thatcan produce monodisperse droplets with high uniformity(Fig. 3A). After gelation of the obtained droplets, we washedGF-beads to remove remaining unreacted GF-monomer. GF-beads were stored in buffer solution (Fig. 3B) with Pluronic®surfactant to prevent the adhesion of GF-beads to the surfaceof vials, tips, and needles; this surface treatment results in easyhandling of GF-beads during implantation. The fluorescentimage of GF-beads (Fig. 3C) reveals that GF-monomer wassuccessfully immobilized in the PAAm hydrogel microbeads.The diameter of GF-beads was measured as 130 μm with narrowsize distribution (see SI Text and Fig. S2). The size of GF-beadswas smaller than the inner diameter of a general injection needle.The injectable size of GF-beads can minimize the damage totissues when the microbeads are implanted in vivo. Furthermore,the beads of 130μm in diameter, unlike smaller sized beads(<1 μm), can stay between tissue layers without permeating thecell-membrane.

In Vitro Glucose Responsiveness Test. We performed the in vitroglucose responsiveness test using the fabricated GF-beads (seeSI Text). From the fluorescent images for various glucose concen-trations (Fig. 4 A–C), the glucose response curves were obtainedfor the glucose concentration of 0–1;000 mg·dL−1 at the emissionwavelength of 488 nanometers (nm) (see Fig. S3). When the glu-cose concentration increased from 0 mg·dL−1 to 1;000 mg·dL−1,the fluorescence intensity also increased depending on the glu-cose concentration (see Table S1). When the glucose concentra-

tion decreased from 1;000 mg·dL−1 to 0 mg·dL−1, the fluore-scence intensity concurrently decreased depending on the glucoseconcentration (see Table S1). The fluorescence intensity curve forthe increase in glucose concentrations matched the fluorescenceintensity curve for the decrease in glucose concentrations(Fig. 4D). These results indicate that glucose association anddissociation occur reversibly in the glucose-recognition site ofGF-monomer. The fluorescence intensity of GF-beads provideda suitable curve for measuring blood glucose concentrations inthe physiological range (62.5–500 mg·dL−1). At a glucose concen-tration of 500 mg·dL−1, the relative fluorescence intensity ofGF-beads was approximately 3 times higher than that of GF-dyeimmobilized in a rigid membrane (18).

We also measured the influence of photobleaching in oursystem, as shown in Fig. S4. As a result, the fluorescence intensityof the microbeads was reduced to 80% after 25 min when wecontinuously applied the excitation light (405 nm, 5.7 mW·cm−2)to the fluorescent microbeads. To minimize photobleaching ofthe microbeads, we believe that a pulse excitation system can beapplied to long-term CGM. For example, if the fluorescenceintensity of the microbeads is measured every 5 min with 1 msexpose, the photobleaching of microbeads will be less than 1%after 3 months.

In Vivo Glucose Monitoring. We implanted GF-beads into the earskin of mice to test the hypothesis that the fluorescence intensityof GF-beads is transdermally detectable (see SI Text). We selectedthe ear skin for the implantation site due to its higher transmis-sion (see Fig. S5A) compared to the other tissues such as abdom-inal muscle and urinary bladder (see Fig. S5B). GF-beads wereimplanted under the ear skin of a mouse (see Fig. S7) usingan injection needle commonly used in a clinical setting. The im-planted GF-beads, as shown in Fig. 5 A and B, were visible eventhough skin layers thicker than 200 μm with biological noises,such as tissues, hair, and secretions. In addition, mice possessingthe implanted GF-beads in their ears moved as usual after im-plantation (see Movie S1), and remained alive without showingany abnormalities for over 30 days.

We also experimentally verified the correlation between thefluorescence intensity of the implanted GF-beads and the in vivoblood glucose concentrations by nine intravenous glucose chal-lenges in five mice. Using the process as described in SI Text,we intravenously injected glucose to temporally elevate up to370 mg·dL−1 within the hyperglycemic range, and insulin todecrease to 130 mg·dL−1 within the euglycemic range. Blood

Fig. 3. Fluorescent microbeads fabricated by a microfluidic system. (A) Sche-matic diagram of the AFFD. The pregel solution flows into the inner channelof the AFFD, while the silicone oil flows into the outer channel. The dropletsin silicone oil are collected from the solution with TEMED at 37 °C, thenthey turn into microbeads after gelation. (B) Bright field image of GF-beads.The fluorescent microbeads were obtained with a uniform diameter.(C) Fluorescent image of GF-beads. GF-beads showed high intensity at theemission light. The scale bar indicates 100 μm.

Fig. 4. Glucose responsiveness of GF-beads. (A–C) Fluorescent images ofGF-beads at the glucose concentration of 0 mg·dL−1, 250 mg·dL−1, and1;000 mg·dL−1, respectively. (D) The fluorescence intensity of GF-beads(n ¼ 19) changes according to the increase and decrease in glucose concen-trations (see Table S1). The overlapping curves are evidence of the reversiblereaction between GF-beads and glucose.

17896 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1006911107 Shibata et al.

Dow

nloa

ded

by g

uest

on

May

10,

202

0

glucose concentrations were measured with the glucose sensorsby using a blood sample from the snipped tail. Fig. 5 D and Eillustrate the fluorescent images of the implantation site in themouse ear (Fig. 5C) before and after the glucose challenge; pseu-docolored images of Fig. 5 D and E are shown in Fig. 5 F and G,respectively. Fig. 5H plots the measured blood glucose concen-trations and the fluorescence intensity of GF-beads by time.The fluorescence intensity constantly tracked the blood glucoseconcentration fluctuation of one or two up-and-down cycles,as shown in Fig. 5H and Fig. S6. The response of fluorescence

intensity lagged 11� 5 min behind the change in blood glucoseconcentrations. Since the fluorescence intensity reflects theglucose concentration in subdermal interstitial tissues, the timelag mainly originated from the time lag of the changes insubdermal interstitial glucose concentration behind the changesin blood glucose concentrations (29, 30).

ConclusionsOur material has highly sensitive glucose responsiveness originat-ing from the high mobility of GF-monomer. We combined micro-fluidics technology to immobilize our material in injectable-sizedmicrobeads. These beads provide minimally invasive implanta-tion and a fluorescent signal by transdermal transmission withoutany external links or electric power sources. The fluorescenceintensity of the microbeads continuously corresponds to bloodglucose concentrations in vivo, showing practical and efficientglucose monitoring. Due to the virtue of malleable PAAm gel,we can optimize the shapes of the glucose-responsive hydrogeldepending on the implantation methods and sites. Moreover,since hydrogels are generally used in sensors to recognize specificmolecules (31, 32) and controlled release carriers (33), the fully-implantable fluorescent microbeads provide a unique route to theintelligent, versatile CGM sensors.

Materials and MethodsSynthesis of Glucose-Responsive Monomer. Synthetic scheme of GF-monomeris described in Fig. 2B. The AAm-PEG-COOH (2.0 g) was dissolved in dichlor-omethane (12 mL) at 0 °C. Diisopropylethylamine (DIEA) (460 μL) and acrylylchloride (216 μL) were added to the solution and reacted for 60 min. Thereactant solution was evaporated and applied to silica gel column chroma-tography for purification. Appropriate fractions were dried under reducedpressure, then amino and carboxyl terminated PEG (NH2-PEG-COOH)(1.40 g) was obtained. GF-dye (108 mg) was dissolved in ethanol (3 mL).NH2-PEG-COOH (1,080 mg), Milli Q water (7 mL), DIEA (105 μL), and 4-(4,6-Dimethoxy-1,3,5-triazin-2-yl)-4-methylmorpholinium chloride) (130mg) wereadded to the solution under a nitrogen atmosphere. Next, the mixturewas allowed to react at 25 °C for 3 h. The reactant solution was evaporatedand dried under reduced pressure. Then, we obtained crude GF-monomer.The product was dissolved in a mixture of acetonitrile and Milli Q andthen purified using the RP-HPLC. Appropriate fractions were evaporatedand then lyophilized to obtain pure GF-monomer (235 mg). See SI Textfor additional details.

Fabrication of Fluorescent Microbeads Using a Microfluidic System. TheGF-beads were fabricated using an AFFD. We designed the AFFD with three-dimensional modelling software (Rhinoceros, AppliCraft) and used a stereo-lithography modelling machine (Perfactory, Envision Tec.) to fabricate theAFFD. The pregel solution contained 5 wt% GF-monomer, 10 wt% AAm,0.2 wt% Bis-AAm, and 0.09 wt% sodium persulfate (SPS) in 60 mMphosphatebuffer with 1.0 mM EDTA, pH 7.4. The pregel solution flows into the innerchannel of the AFFD, while silicone oil (Dow Corning Toray Co., Ltd.) flowsinto its outer channel. The inlets of pregel solution and silicone oil wereconnected to a syringe pump (KDS210, KdScientific) through ethylene tetra-fluoroethylene tubes of 0.5 mm diameter. We set the flow rates of the innerand outer fluids at 10 μL·min−1 and 150 μL·min−1, respectively. The pregelsolution was broken into droplets at the orifice by the flow of silicone oil.The collected droplets were gathered in phosphate buffer with N,N,N',N'-tet-ramethylethylenediamine at 37 °C. After 30 min, we washed the polymerizedGF-beads three times with hexane (Kanto Chemical Co., Inc.), ethanol, Milli Qwater, and phosphate buffer. In addition, Pluronic® F127 (Sigma-Aldrich)surfactant was dissolved in buffer solution (0.05 wt%) to prevent adhesionof the microbeads to the surface of vials, tips, and needles. See SI Text andFig. S2 for additional details.

ACKNOWLEDGMENTS. We appreciate the support by Ms. N. Yamamoto foranimal experiments. This work was supported by the New Energy andIndustrial Technology Development Organization (NEDO).

1. Zimmet P, Alberti KGMM, Jonathan S (2001) Global and societal implications of the

diabetes epidemic. Nature 414:782–788.

2. Guerra SD, et al. (2005) Functional and molecular defects of pancreatic islets in human

type 2 diabetes. Diabetes 54:727–735.

3. Koya D, King GL (1998) Protein kinase C activation and the development of diabetic

complications. Diabetes 47:859–866.

4. Brownlee M (2001) Biochemistry and molecular cell biology of diabetic complications.

Nature 414:813–820.

Fig. 5. In vivo CGM in a mouse using injected GF-beads. (A, B) GF-beadsunder the dermis of a mouse ear. (C) Enlarged view of the implantation sitein the mouse ear. (D, E) Fluorescent images for glucose concentrations withinthe euglycemic and hyperglycemic ranges, respectively. (F, G) Pseudocoloredimages of D and E. (H) The fluorescence intensity traces blood glucoseconcentrations. The time lag mainly comes from the time response betweeninterstitial fluid glucose concentration and blood glucose concentration.

Shibata et al. PNAS ∣ October 19, 2010 ∣ vol. 107 ∣ no. 42 ∣ 17897

ENGINEE

RING

Dow

nloa

ded

by g

uest

on

May

10,

202

0

5. Lee TH, et al. (1999) Derivation and prospective validation of a simple index forprediction of cardiac risk of major noncardiac surgery. Circulation 100:1043–1049.

6. Nestler EJ, et al. (2007) Neurobiology of depression. Neuron 34:13–25.7. Kondepati VR, Heise HM (2007) Recent progress in analytical instrumentation for

glycemic control in diabetic and critically ill patients. Anal Bioanal Chem 388:545–563.8. Klonoff DC (2005) Continuous glucose monitoring: roadmap for 21st century diabetes

therapy. Diabetes Care 28:1231–1239.9. Zenkl G, Mayr T, Klimant I (2008) Sugar-responsive fluorescent nanospheres.

Macromol Biosci 8:146–152.10. Zenkl G, Klimant I (2009) Fluorescent acrylamide nanoparticles for boronic acid based

sugar sensing—from probes to sensors. Microchim Acta 166:123–131.11. Rounds RM, Ibey BL, Beier HT, Pishko MV, Cote GL (2007) Microporated PEG spheres

for fluorescent analyte detection. J Fluoresc 17:57–63.12. Russell RJ, Pishko MV, Gefrides CC, McShane MJ, Cote GL (1999) A fluorescence-based

glucose biosensor using concanavalin A and dextran encapsulated in a poly(ethyleneglycol) hydrogel. Anal Chem 71:3126–3132.

13. Turner RFB, Harrison DJ, Rayotte RV, Baltes HP (1990) A biocompatible enzymeelectrode for continuous in vivo glucose monitoring in whole blood. Sensors andActuators B1:561–564.

14. Abel PU, von Woedtke T (2002) Biosensors for in vivo glucose measurement: can wecross the experimental stage. Biosens Bioelectron 17:1059–1070.

15. Kenausis G, Chen Q, Heller A (1997) Electrochemical glucose and lactate sensorsbased on “wired” thermostable soybean peroxidase operating continuously andstably at 37 °C. Anal Chem 69:1054–1060.

16. Wilson GS, Hu Y (2000) Enzyme-based biosensors for in vivo measurements. Chem Rev100:2693–2704.

17. Ballerstadt R, Evans C, McNichols R, Gowda A (2006) Concanavalin A for in vivo glucosesensing: a biotoxicity review. Biosens Bioelectron 22:275–284.

18. Kawanishi T, Romey MA, Zhu PC, Holody MZ, Shinkai S (2004) Study of boronic acidbased fluorescent glucose sensors. J Fluoresc 14:499–512.

19. James TD, Sandanayake KRAS, Shinkai S (1995) Chiral discrimination of monosacchar-ides using a fluorescent molecular sensor. Nature 374:345–347.

20. James TD, Sandanayake KRAS, Shinkai S (1994) Novel photoinduced electron-transfer sensor for saccharides based on the interaction of boronic acid and amine.J Chem Soc, Chem Commun 477–478.

21. James TD (2007) Saccharide-selective boronic acid based photoinduced electrontransfer (PET) fluorescent sensors. Top Curr Chem 277:107–152.

22. James TD, Sandanayake KRAS, Iguchi R, Shinkai S (1995) Novel saccharide-photoin-duced electron transfer sensors based on the interaction of boronic acid and amine.J Am Chem Soc 117:8982–8987.

23. Mosqueria VCF, et al. (2001) Biodistribution of long-circulating PEG-grafted nanocap-sules in mice: effects of PEG chain length and density. Pharm Res 18:1411–1419.

24. Discher DE, Eisenberg A (2002) Polymer vesicles. Science 297:967–973.25. Lu H, Li Z, Hu N (2003) Direct voltammetry and electrocatalytic properties of catalase

incorporated in polyacrylamide hydrogel films. Biophys Chem 104:623–632.26. Anna SL, Bontoux N, Stone HA (2003) Formation of dispersions using “flow focusing”

in microchannels. Appl Phys Lett 82:364–366.27. Banderas LM, et al. (2005) Flow focusing: a versatile technology to produce size-

controlled and specific-morphology microparticles. Small 1:2–6.28. Huang S-H, Tan W-H, Tseng F-G, Takeuchi S (2006) A monolithically three-dimensional

flow-focusing device for formation of single/double emulsions in closed/open micro-fluidic systems. J Micromech Microeng 16:2336–2344.

29. Bantle JP, Thomas W (1997) Glucose measurement in patients with diabetes mellituswith dermal interstitial fluid. J Lab Clin Med 130:436–441.

30. Aussedat B, et al. (2000) Interstitial glucose concentration and glycemia: implicationsfor continuous subcutaneous glucose monitoring. Am J Physiol-Endoc M 278:E716–E728.

31. Miyata T, Asami N, Uragami T (1999) A reversibly antigen-responsive hydrogel. Nature399:766–769.

32. Gajovic N, Binyamin G, Warsinke A, Scheller FW, Heller A (2000) Operation of aminiature redox hydrogel-based pyruvate sensor in undiluted deoxygenated calfserum. Anal Chem 72:2963–2968.

33. Lee KY, Peters MC, Anderson KW, Mooney DJ (2000) Controlled growth factor releasefrom synthetic extracellular matrices. Nature 408:998–1000.

17898 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1006911107 Shibata et al.

Dow

nloa

ded

by g

uest

on

May

10,

202

0