Biosensing Platform Based on Fluorescence Resonance Energy Transfer from Upconverting Nanocrystals to Graphene Oxide.pdf

BiosensorsDOI:10.1002/anie.201100769Biosensing Platform Based on Fluorescence Resonance EnergyTransfer from Upconverting Nanocrystals to Graphene Oxide*Cuiling Zhang,Yunxia Yuan,Shiming Zhang,Yuhui Wang,and Zhihong Liu*Fluorescence resonance energy transfer(FRET)is recog-nized as a sensitive and reliable analytical technique and hasbeen widely used in biological assays.1To obtain improvedFRETefficiency and analytical performance,it is of continu-ing interest to search for new energy donoracceptor pairs.Inthe past few years,the use of anti-Stokes fluorophoresincluding upconverting phosphors(UCP)and multiphoton-excited dyes as energy donors which can be excited in thenear-infrared(NIR)region has successfully circumvented theproblem of autofluorescence and scattering of light arisingfrom biological substances.2,3This has made it possible todirectly conduct FRET-based assays in biological samples.More recently,graphene,the newly emerging two-dimen-sional and zero-bandgap carbon nanomaterial,has attractedconsiderable attention in bioassays because of its uniqueelectronic,mechanical,and thermal properties.In the pio-neering work of Swathi et al.,it was proposed throughtheoretical calculations that graphene could act as a super-quencher of organic dyes,as a result of nonradiative transferof electronic excitation energy from dye excited states to thep system of graphene.4The rate of this long-range resonanceenergy transfer was suggested to have a d?4dependence ondistance d,in sharp contrast to traditional FRET,for whichthe rate has a d?6dependence.Inspired by this property,graphene and graphene oxide(GO)have been used as FRETacceptors with organic dyes and quantum dots as energydonors,511in which both graphene and GO exhibit highefficiency in quenching the donor emission and thus providegood sensitivity.Herein we reveal energy transfer from UCPto GO and thus construction of a new biosensing platformwhich could be used to detect glucose directly in serumsamples and extended to detection of other biologicallysignificant molecules.The previously reported FRET models based on grapheneor graphene oxide all rely on the pp stacking interactionbetween the carbon nanomaterial and nucleic acid chains,which bring the acceptor and donor(organic dyes or quantumdots)into close proximity.We hereby tried a different modelin which donor and acceptor are brought into FRETproximity through specific molecular recognition(Figure 1).We used GO as energy acceptor because the abundance ofcarboxy,hydroxy,and epoxy groups on the surface of GOsheets12makes the material more water-soluble and alsoenables covalent conjugation with other molecules.Conca-navalin A(conA)and chitosan(CS)were covalently attachedto UCP and GO,respectively.The known tight binding ofConA with CS may bring UCP and GO into appropriateproximity and hence induce energy transfer.Thereafter,theFRET process is anticipated to be inhibited(in part)becauseof competition between glucose and CS for ConA,whichcould be the foundation of glucose sensing.To realize such design,we first synthesized water-solubleNaYF4:Yb,Er UCP nanocrystals modified with polyacrylicacid(PAA).Details of UCP synthesis are given in theSupporting Information.The fluorescence intensity of UCPremains unchanged under continuous 980 nm illumination forup to several hours,which suggests good photostability.Highly dispersible PAA-functionalized UCP particles with anaverage particle size of about 50 nm were obtained,whichconsisted of a dominant hexagonal phase and a small amountof cubic phase(Figure S1,Supporting Information).TheUCPConA conjugate was prepared by an 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide(EDC)coupling proto-col,and successful conjugation was confirmed by UV/Visspectroscopy(Figure S2,Supporting Information).Grapheneoxide nanosheets were synthesized according to the reportedmethod,6and the obtained dispersion was dialyzed throughsemipermeable membranes to remove impurities.The prod-ucts were characterized by the XRD pattern(Figure S3,Supporting Information),which exhibits the characteristicdiffraction peak of GO at 2q=10.588 8.Chitosan moleculeswere attached to the surface of GO by EDC-mediatedcoupling,and FTIR spectra were recorded to characterize thechemical structure of GO and GOCS conjugates(Figure S4,Figure 1.Schematic illustration of the UCPGO biosensing platformand the mechanism of glucose determination.*C.L.Zhang,Y.X.Yuan,S.M.Zhang,Y.H.Wang,Prof.Z.H.LiuKey Laboratory of Analytical Chemistry for Biology and Medicine(Ministry of Education),College of Chemistry and MolecularSciences,Wuhan University,Wuhan 430072(P.R.China)Fax:[phone]E-mail:*The authors thank Prof.S.L.Chen for helping with the character-ization of GO.We also acknowledge the financial support from theNational Natural Science Foundation of China(Grant No.21075094),and the Science Fund for Creative Research Groups(Grant Nos.20621502 and 20921062)Supporting information for this article is available on the WWWunder http:/dx.doi.org/10.1002/anie.201100769.6851Angew.Chem.Int.Ed.2011,50,68516854?2011 Wiley-VCH Verlag GmbH&Co.KGaA,WeinheimSupporting Information).The characteristic absorption peaksof GO were observed,including peaks for O?H at 3413,C=Oat 1731,C=C at 1622,and C?O at 1092 cm?1(Figure S4a,Supporting Information).For the GOCS complex,charac-teristic peaks corresponding to the C=O stretching vibration(1631 cm?1),C?N stretching vibration(1384 cm?1),and theasymmetric and symmetric stretching vibrations of methylgroups(2938 and 2873 cm?1)were observed(Figure S4b,Supporting Information).Notably,both the UCPConA andGOCS complexes showed good dispersibility in aqueoussolution,which ensured the stability and reliability ofspectroscopic measurements.The formation of GOCSConAUCP complex by inter-action between CS and ConA was first verified with AFMmeasurements.The AFM height image shows that thethickness of the as-synthesized GO sheets was about 1 nm(Figure 2a).When UCPConA was mixed with GOCS,theheights of the complex were approximately 110 and 80 nm,which indicated that UCP was brought close to the GOsurface(Figure 2b).Energy transfer from UCP to GO wasthen investigated by measuring the fluorescence quenching ofUCP.On mixing GOCS and UCPConA solutions(in 0.01mTris-HCl buffer,pH 7.4),the fluorescence intensity of UCPgradually decreased with increasing concentration of GOCS(Figure 2c).The extent of fluorescence quenching linearlycorresponded to the concentration of GOCS(Figure 2c,inset).An overall degree of quenching of 81%was obtainedwith a GO concentration of 0.22 mgmL?1.Such a quenchingefficiency is somewhat lower than those reported with organicdyes as energy donor.This can be explained by the nature ofUCP materials,in which only the emitters(rare earth ions)ator near the surface of particles can be quenched.For assaysbased on such a quenchingrecovery model,higher fluores-cence quenching rates are generally preferable in terms ofdetermination sensitivity.Nonetheless,the extremely highluminescence efficiency of UCP nanocrystals still makes themcompetitive energy donors.As a kind of two-dimensionalmaterial with relatively large surface,adsorption of othersubstances(e.g.,the UCPConA complex)on the surface ofGO could be a concern that might cause false-positive signals.To rule out the possibility of nonspecific binding,a controlexperiment was done in which the UCPConA solution wasmixed with 0.25 mgmL?1of bare GO solution.No obvioussignal changewas observed in this case(Figure S5,SupportingInformation),and this implies that no nonspecific adsorptionof UCPConA on the GO surface occurs,and that thefluorescence quenching can be exclusively ascribed to recog-nition of CS by ConA.Furthermore,we performed the above FRETexperimentsin human serum from which inherent glucose was removed byglucose oxidase(vide infra),with two purposes:1)to takeadvantage of the NIR excitation of upconverting phosphors,that is,the ability to avoid background interference in acomplex sample matrix like serum;2)although the applica-tion of graphene materials in bioassay has been welldemonstrated in the above-mentioned literature,the possi-bility of using them in such a complex biological samplematrix has not yet been explored.We found that,when using20-fold diluted serum as medium,a FRET process that wasnearly the same as that in aqueous buffer occurred(Fig-ure 2d),except for a slight difference in the slope of the linearcalibration(inset).Thus,the various biomolecules in theserum do not have a significant influence on either the FRETprocess or the optical measurements.Considering the robustness of the FRET model shown bythe above experiments,we subsequently determined glucosein serum medium.To preclude the influence of the inherentglucose in serum,it was decomposed with glucose oxidasefollowed by deactivating the enzyme.The resulting serumcaused no significant signal alteration of the FRET sensor,that is,no inherent glucose was left(Figure S6,SupportingInformation).Then external free glucose with varying con-centrations was introduced to the sensor,which resulted inpartial deconstruction of the GOCSConAUCP complexdue to the stronger combination between glucose and ConA.Consequently,fluorescence of UCP donor was restored in aglucose concentration dependent manner(Figure 3).The plotof fluorescence intensity versus glucose concentration showeda linear calibration in the range from 0.56 to 2.0 mm.The limitof detection was 0.025 mm,calculated according to the 3sb/mcriterion,where m is the slope for the range of linearity usedand sbthe standard deviation of a blankn=11).Such assaysensitivity is comparable to or even better than those ofspectroscopic methods for glucose determination performedin aqueous solutions.1317The assay also exhibited goodreproducibility,as shown by standard deviations from inde-pendent measurements(Figure 3b).As compared to theFigure 2.AFM images and height profiles of GO(a)and the GOCSConAUCP complex(b),and fluorescence titration of UCPConA(0.45 mgmL?1UCP)by GOCS with GO concentration ranging from 0to 0.22 mgmL?1in c)Tris-HCl buffer(0.01m,pH 7.4)and d)humanserum(20-fold diluted with Tris-HCl buffer)from which inherentglucose had been removed.Inset:linear relationship between fluores-cence intensity of UCP and concentration of GO.Excitation wave-length:980 nm.Communications6852www.angewandte.org?2011 Wiley-VCH Verlag GmbH&Co.KGaA,WeinheimAngew.Chem.Int.Ed.2011,50,68516854noncovalent pp stacking between the ring structures innucleobases and the hexagonal cells of graphene(or grapheneoxide),the combination of target molecules on the surface ofGO nanosheets through more robust and stable covalentbonds may contribute to analytical performance criteria suchas sensitivity and precision,although it requires onemore stepfor labeling.More importantly,the flexibility in covalentcoupling reaction also laid a foundation for easily extendingthe FRET platform to other targets.The specificity of the FRET sensor towards glucose wasexamined in control experiments in which other biomoleculesincluding carbohydrate,protein,and amino acid were testedwith procedures identical to the above assay.At a concen-tration equal to that of glucose(2.4 mm),the tested substancesdid not cause obvious restoration of donor fluorescence,expect for mannose(Figure 4),which caused concentration-dependent recovery of UCP fluorescence(Figure S7,Sup-porting Information),as it is also recognized by conA.18Nevertheless,the assay of glucose in human serum wouldnot be affected,since mannose does not exist in mammalianblood.To validate the expanded application of the UCPGOFRET platform,a homogeneous hybridization model wasadopted for ssDNA detection.UCP was first tagged tostreptavidin(SA)by an EDC protocol,and successfulconjugation was confirmed by UV/Vis spectroscopy(Fig-ure S8,Supporting Information).The UCPSA complex wasthen linked to biotinylated ssDNA(5-biotin-GAGTTAG-CACCCGCATAGTCAAGAT-3)via the biotin(B)SAbridge.On mixing the UCPSABssDNA complex withincreasing amounts of GO solution,the fluorescence of UCPwas gradually quenched(Figure 5a),as a result of energytransfer from UCP to GO induced by the pp stackinginteraction between ssDNA and GO.5In the presence of thecomplementarytargetssDNA(5-ATCTTGAC-TATGCGGGTGCTAACTC-3),the stronger interaction ofcomplementary chains disturbs the interaction betweenUCPSABssDNA and GO,resulting in restoration of thefluorescence of UCP(Figure 5b).In conclusion,we have constructed a novel sensor forglucose determination based on FRET from upconvertingphosphors to graphene oxide.When excited with NIR light,the FRET sensor showed favorable analytical performance ina complex biological sample matrix.Unlike commonly usedheterogeneous methods like ELISA,which need multipleseparating steps,the proposed UCPGO sensor is capable ofhomogeneously detecting glucose in serum samples withoutany background interference,so the UCP-GO FRET systemcould be a promising platform for biosensing.The covalentcombination-based design can readily be extended to sensingof other biomolecules.This work may enrich the FRETtechnique and promote application of graphene materials inbioassay.Experimental SectionSensing of glucose:In a typical measurement,0.27 mgmL?1GOCSsolution was added to 3 mgmL?1UCPConA in 20-fold diluted(with0.01m Tris-HCl,pH 7.4)pretreated human serum.Thereafter,differ-ent concentrations of glucose were added to the above solution andthe mixtures incubated for 1.5 h before measurement.Fluorescenceemissionofthedonorwasmeasuredunderexcitationat 980 nmwithaCW laser.Detection of ssDNA:UCP(6.4 mg)was attached to SA(1.1 mg)in 8 mL 2-(N-morpholino)ethanesulfonic acid(MES,0.01m,pH 6.06)containing 3.2 mg EDC and 9.6 mg N-hydroxysulfosuccinimide(Sulfo-NHS),and then biotinylated ssDNA(biotin in fourfoldmolar excess of SA)was added to the UCPSA conjugate to formUCPSABssDNA complex.In a typical hybridization procedure,Figure 3.a)Fluorescence emission spectra of GOCSConAUCPcomplex in the presence of different concentrations of glucose(0.562.0 mm)in diluted serum.b)Linear relationship between donor fluores-cence intensity and glucose concentrations.Excitation wavelength:980 nm.Figure 4.Variations in fluorescence intensity induced by differentsubstances,all with a concentration of 2.4 mm.Blank represents theGOCSConAUCP complex,the fluorescence intensity of which isdefined as F0,and F is the fluorescence intensity in the presence oftested substances.Experiments were conducted in 20-fold diluted(with Tris-HCl buffer)human serum from which inherent glucose hadbeen removed.Excitation wavelength:980 nm.Figure 5.a)Fluorescence emission spectra of the UCPSABssDNAcomplex in the presence of different concentrations of GO(0,0.02,0.05,0.08,0.10,0.15,0.2 mgmL?1).b)Restoration of UCP fluores-cence after incubation with target ssDNA(0,3.33,6.65,13.3,26.6,53.2,106.4 nm)in the presence of 0.2 mgmL?1GO.6853Angew.Chem.Int.Ed.2011,50,68516854?2011 Wiley-VCH Verlag GmbH&Co.KGaA,Weinheimwww.an