Responsive Photonic Crystal Carbohydrate Hydrogel SensorMaterials for Selective and Sensitive Lectin Protein DetectionZhongyu Cai, Aniruddha Sasmal, Xinyu Liu,* and Sanford A. Asher*

Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania 15260, United States

*S Supporting Information

ABSTRACT: Lectin proteins, such as the highly toxic lectinprotein, ricin, and the immunochemically important lectin,jacalin, play significant roles in many biological functions. It ishighly desirable to develop a simple but efficient method toselectively detect lectin proteins. Here we report thedevelopment of carbohydrate containing responsive hydrogelsensing materials for the selective detection of lectin proteins.The copolymerization of a vinyl linked carbohydrate monomerwith acrylamide and acrylic acid forms a carbohydrate hydrogelthat shows specific “multivalent” binding to lectin proteins.The resulting carbohydrate hydrogels are attached to 2-Dphotonic crystals (PCs) that brightly diffract visible light. Thisdiffraction provides an optical readout that sensitivelymonitors the hydrogel volume. We utilize lactose, galactose, and mannose containing hydrogels to fabricate a series of 2-DPC sensors that show strong selective binding to the lectin proteins ricin, jacalin, and concanavalin A (Con A). This bindingcauses a carbohydrate hydrogel shrinkage which significantly shifts the diffraction wavelength. The resulting 2-D PC sensors canselectively detect the lectin proteins ricin, jacalin, and Con A. These unoptimized 2-D PC hydrogel sensors show a limit ofdetection (LoD) of 7.5 × 10−8 M for ricin, a LoD of 2.3 × 10−7 M for jacalin, and a LoD of 3.8 × 10−8 M for Con A, respectively.This sensor fabrication approach may enable numerous sensors for the selective detection of numerous lectin proteins.

KEYWORDS: photonic crystals, carbohydrate hydrogels, biosensors, lectin proteins detection, copolymerization

Proteins and carbohydrates are two of the central buildingblocks of life. Protein−carbohydrate interactions play

fundamental roles in many biological processes, such as cellrecognition and cell adhesion, and in the etiology of a broadrange of diseases.1−4 The binding between individual proteinsand carbohydrates is generally nonspecific and relatively weak,with association constants in the range of μM to mM.5

Fortunately, the specificity and affinity of protein−carbohydrateinteraction can be dramatically enhanced by the clustering ofmultiple carbohydrates, a phenomenon defined as the “clusterglycoside effect”.6,7 Multivalent protein−carbohydrate inter-actions are widely utilized in nature to achieve strong andspecific molecular and cellular interactions.8 For example, manyhighly poisonous toxins, including Shiga-like toxin and ricin,induce cellular toxicity through multivalent interactions.9,10

Numerous approaches that use multivalent binding havebeen developed to detect and to monitor protein−carbohydrateinteractions both in vitro and in vivo.4,11,12 Water-solublepolymer,13 dendrimers14−16 as well as solid carriers11,17 such asnanoparticles,11,18−22 carbon nanotubes,17 and graphene23 wereused as scaffolds to present multivalent carbohydrate ligands.These systems were utilized as biosensor devices andbioimaging agents for lectins,23−26 toxins,19−22,27 microbes,and tumor cells.28−32 These systems are quite versatile;33−36

they were applied to sense and quantify protein−carbohydrate

interactions by additionally utilizing fluorescence, NMR,circular dichroism, surface plasmon resonance spectroscopy,total internal reflection fluorescence spectroscopy, andmicroscopy.12,33 Unfortunately, the need for these sophisticatedinstrumentation approaches to detect protein−carbohydrateinteractions restricts their utility for use in limited resourceenvironments. The development of a simple and inexpensivemethod to monitor protein−carbohydrate interactions is highlydesirable.Recently, there has been significant interest in the use of

photonic crystals (PCs) for sensing due to their intrinsicsimplicity and their diffraction efficiency.37−47 3-D PC sensorswere developed to sense many species, including metal ions,glucose, sarin, and various saccharides.38−42,45 A majorlimitation of non-close-packed 3-D PC hydrogel sensors isthat the electrostatic self-assembly of highly ordered arraysrequires very low ionic strength conditions, which can beinconsistent with the sensor fabrication chemistry.44 Tosurmount this difficulty, we developed a 2-D PC sensingtechnology that avoids the need for low ionic strengthconditions.43,44 This 2-D PC sensing motif uses a monolayer

Received: June 24, 2017Accepted: September 22, 2017Published: September 22, 2017

Article

pubs.acs.org/acssensors

© 2017 American Chemical Society 1474 DOI: 10.1021/acssensors.7b00426ACS Sens. 2017, 2, 1474−1481

Cite This: ACS Sens. 2017, 2, 1474-1481

array that is prepared independently of the responsive hydrogel.We monitor the shift in diffraction of the 2-D array attached tothe surface of the responsive hydrogel. This approach is simpleand highly efficient. We used this approach to develop sensorsfor pH, surfactants, metals, proteins, and microbes.43,48−55

In this study, we synthesized carbohydrate containinghydrogels for developing sensors for lectin proteins, such asthe highly toxic lectin protein, ricin, and the immunochemicallyimportant lectin, jacalin.10,56 These sensor materials utilize our2-D PC sensing motif to quantify the concentrations of lectinproteins. As discussed in detail below we first self-assemble a 2-D PC at the air−water interface and transfer it onto a glassslide. We then polymerize a vinyl linked carbohydratemonomer solution containing acrylamide and acrylic acidonto the 2-D PC monolayer. This forms the sensing responsivehydrogel sensor material. These responsive hydrogels containcarbohydrates attached to a polymer backbone that multi-valently binds lectin proteins. This causes a hydrogel shrinkagethat blue-shifts the diffraction wavelength of the 2-D PC arrayattached to the hydrogel surface. The vivid optical diffractionreports on the lectin protein concentration.

■ EXPERIMENTAL SECTIONMaterials and Methods. Allyl α-D-galactose and allyl β-D-

galactose were purchased from Sigma-Aldrich and Fisher Scientific,respectively, and used as received. The seeds of the castor oil plant canproduce two similar proteins, ricin (RCA60) and an agglutinin(RCA120). The less toxic RCA120 were used in this study.Unconjugated Ricinus communis agglutinin I (RCA I, RCA120) andunconjugated jacalin were purchased from Vector Laboratories, Inc.,and were prepared at the required solution concentrations prior to use.Acrylamide (AAm), acrylic acid (AAc), N,N′-methylenebis-(acrylamide) (MBAAm), 2-hydroxy-1-(4-(2-hydroxyethoxy)-phenyl)-2-methyl-1-propanone (Irgacure 2959), dimethyl sulfoxide (DMSO),2-allyloxyethanol, styrene, HEPES, sodium azide, sodium chloride,calcium chloride dehydrate, bovine serum albumin (BSA), andmanganese chloride tetrahydrate were purchased from Sigma-Aldrich,and used as received. D-Lactose monohydrate and D-mannose werepurchased from Chem-Impex International Inc. Hg(CN)2, HgBr2, andBF3·Et2O were purchased from Acros Organics. Deuterium oxide waspurchased from Cambridge Isotope Laboratories, Inc. Concanavalin A(Con A) was donated by Sigma-Aldrich. 1-Propanol was purchasedfrom J. T. Baker Inc. Monodisperse, ∼650 ± 16-nm-diameterpolystyrene (PS) particles were synthesized by using an emulsifierfree emulsion polymerization as previously reported.57

Fabrication of 2-D PC Lectin Protein Sensors. 2-Allylethoxyl-β-D-lactose was synthesized from commercially available D-lactose in 3steps (synthesis protocols are shown in Supporting Information,Schemes S1−S3). Mannose monomer (2-allylethoxyl α-D-mannose)with the same 2-allylethoxy linker at its reduced ends was alsosynthesized. The synthesis protocol is shown in the SupportingInformation (synthesis protocols are shown in Supporting Informa-tion, Schemes S4 and S5). The 1H and 13C NMR spectra of these twocarbohydrate monomers and their analysis are shown in Figures S1−S5. The structure of commercial allyl α-D-galactose and allyl β-D-galactose was also confirmed with 1H NMR spectra (Figures S6 andS7).Figure 1 shows the procedures for the fabrication of our 2-D PC

PAAm-AAc-Carbohydrate hydrogels. The 2-D PS colloidal PC arraywas first assembled on a water surface by using our needle tip flowtechnique.48 The 2-D PC array on the water surface was thentransferred to a glass slide and dried in the air. A 50 μL solution ofAAm, AAc, MBAAm, and 2-allyloxyethyl lactose (or othercarbohydrate monomer) was layered onto the 2-D array on the glassslide (24 × 50 mm2). Another glass slide was placed on top to coverand flatten the polymerization solution.

The copolymerization was carried out by using 365 nm UV light(UVP, UVGL-55 hand-held UV Lamp, 6 Watt) at room temperature.After 20 min, we peeled the 2-D array hydrogel film from the glassslide and washed it at least 5 times in 10 mM phosphate buffered saline(PBS) solution (containing 0.15 M NaCl) to remove any unreactedmonomers and impurities. Scheme 1 illustrates the copolymerizationof PAAm-AAc-Carbohydrate hydrogels. A series of 2-D PC lactosehydrogels were fabricated by using monomer solutions containing 20,40, 80, and 120 mg/mL of 2-allylethoxyl lactose. These samples aredenoted as 2-D PC PAAm-AAc-Lactose-20, PAAm-AAc-Lactose-40,PAAm-AAc-Lactose-80, and PAAm-AAc-Lactose-120, respectively.Likewise, the galactose and mannose-containing hydrogels are similarlydenoted. The different reaction stoichiometries utilized for the 2-D PCcarbohydrate hydrogel preparations are listed in Table 1.

Protein Recognition. The response of these sensors to proteinswas characterized by using either a UV−vis reflection spectrometer, orby light diffracted from a green laser pointer (λ = 532 nm). For thePAAm-AAc-Lactose hydrogel, the hydrogel sensors were firstequilibrated in 10 mM PBS solution (containing 0.15 M NaCl and0.08 wt % sodium azide) at pH 7.8 for 24 h, during which the PBSsolution was frequently changed. Then small pieces of the sensor (8mm × 8 mm squares) were placed in 0 to 1.0 mg/mL ricin solutionscontaining 0.15 M NaCl and 0.08 wt % sodium azide. The 2-D PCPAAm-AAc-Lactose sensors were equilibrated overnight beforediffraction measurements.

Bragg diffraction from the 2-D PCs on the PAAm-AAc-Lactosehydrogel sensors was monitored by using an Ocean Optics USB2000-UV−vis Spectrometer, a LS-1 Tungsten Halogen Light Source, and anR-series Fiber Optic Reflection Probe. All diffraction measurementswere carried out with the 2-D PC-Carbohydrate hydrogels on a silverfront surface mirror (Thorlabs, VA). The diffraction measurementswere carried out in a Littrow configuration with the fiber at a ∼14°angle from the array normal.43 The response of other carbohydratehydrogels was measured in the same way.

For the detection of Con A with the PAAm-AAc-Mannosehydrogel, the hydrogel sensors were pre-equilibrated with a 0.1 MNaCl solution containing 1 mM Ca2+ and Mn2+. The Con A samplesolutions were prepared with a 0.1 M NaCl solution containing 1 mMCa2+ and Mn2+. For the detection of jacalin using the PAAm-AAc-Galactose hydrogel, the hydrogel sensors were equilibrated with 10mM HEPES buffered saline at pH 8.5 containing 0.1 mM Ca2+ and0.08 wt % sodium azide.

For the reversibility study, the 2-D PC PAAm-AAc-Lactosehydrogel sensors were immersed into a 2 mL 1.0 mg/mL ricinsolutions containing 10 mM PBS at pH 7.8 for 4 h before eachmeasurement. After the Debye diffraction ring diameter measure-

Figure 1. Illustration of the fabrication of 2-D PC-Carbohydratehydrogel sensors.

ACS Sensors Article

DOI: 10.1021/acssensors.7b00426ACS Sens. 2017, 2, 1474−1481

1475

ments, the samples were washed with a large amount of 10 mM PBS atpH 7.8 prior to the next round of measurements.Microscopy and NMR Spectroscopy Characterization. The

carbohydrate monomers were characterized with 1H NMR and 13CNMR (Bruker Avance III 600 and 400 MHz), respectively. The 2-DPS colloidal PC arrays and the 2-D PC PAAm-AAc-Carbohydratehydrogel sensors were sputter-coated with gold (Au). The surfacemorphology measurements were taken by using a scanning electronmicroscope (SEM, JEOL JSM6390LV).Optical Diffraction Characterization. Under the irradiation of a

laser pointer, the 2-D PC diffracts light at an angle that depends onboth the interparticle spacing and the laser wavelength.58−60 Therotationally disordered small 2-D PC array domains (20 × 20 μm2)diffract a Debye ring pattern. The first-order diffraction angle, α,depends upon the particle spacing: sin α = 2λlaser/(3

1/2d), where α isthe interior angle of the Debye diffraction ring, λlaser is the laserwavelength, and d is the particle spacing. The diffraction angle α isdetermined from the Debye ring diameter: α = tan−1(D/2h), where Dis the Debye ring diameter and h is the distance between the 2-D PCarray and screen. We monitor the 2-D PC array particle spacing by

measuring D and h and calculating d = λ +D hD

4 ( / 2)3

laser2 2

.49 In this

study, the h value was set to 36 mm. For each analyte concentration, 3identical samples were used and each sample was measured at 3different positions. The average and standard deviation of d wereobtained from these 9 values. The standard deviation of the particlespacing is calculated using the following equation:

∑=−

− =

sN

x x1

1( )

i

N

i1

2

The diffraction of our 2-D PC PAAm-AAc-Carbohydrate hydrogelsensors was also measured by using an Ocean Optics USB 2000-UV−vis spectrometer, a LS-1 tungsten halogen light source, and an R-series

fiber optic reflection probe in a Littrow configuration with the fiberprobe at an ∼14° angle from the array normal. The Debye diffractionring diameter measurement physically involves the same diffractionprocess as that measured by the UV−vis reflection probe fiber opticspectrometer. In a Littrow configuration, the 2-D Bragg diffractionrelationship is mλ = 31/2d sin θ, where m is the diffraction order, λ isthe diffracted wavelength (in vacuum), d is the 2-D particle spacing,and θ is the angle of the incident light relative to the 2-D PC arraynormal.60,61

■ RESULTS AND DISCUSSIONFigure 2a shows an SEM image of a 2-D PC fabricated on anair−water interface, which was transferred onto a glass slide,

and then dried in air. The resulting 2-D PC is highly orderedover dimensions of 20 μm × 20 μm. Cracks in the hexagonalordering presumably occur as the 2-D array shrinks as it drieson the slide. Figure 2b shows the white light diffractionspectrum of a 2-D PC PAAm-AAc-Lactose hydrogel whenilluminated from below by a collimated white light source

Scheme 1. Synthesis of PAAm-AAc-Carbohydrate Hydrogelsa

aR represents vinyl or allyl substituted carbohydrates.

Table 1. Stoichiometry of 2-D PC PAAm-AAc-CarbohydrateHydrogelsa

SamplesAAm(mg)

Carbohydratemonomer(mg)b

MBAAm(mg)

AAc(μL)

Irgacure2959(μL)c

2-D PC PAAm-AAc-carbohydrate-20

40 8 0.4 8 12

2-D PC PAAm-AAc-carbohydrate-40

40 16 0.4 8 12

2-D PC PAAm-AAc-carbohydrate-80

40 32 0.4 8 12

2-D PC PAAm-AAc-carbohydrate-120

40 48 0.4 8 12

2-D PC PAAm-AAc-carbohydrate-160

40 64 0.4 8 12

aTotal reaction solution volume is 400 μL. bCarbohydrate monomersused in this study include 2-allylethoxyl-β-D-lactose, allyl α-D-galactose,allyl β-D-galactose, and 2-allylethoxyl α-D-mannopyranoside. cInitiatoris Irgacure 2959 in DMSO (33%, w:v).

Figure 2. (a) SEM image of 2-D PCs of monodisperse PS particleswith a diameter of ∼650 nm. (b) Photograph of diffraction from 2-DPCs on the surface of a PAAm-AAc-Lactose hydrogel (∼80 μm thick)illuminated with a flashlight below at an angle of ∼50° to the normal.The white light forward diffraction gives rise to a spectrum of colors.(c) SEM image of 2-D PCs on a freshly prepared 2-D PC PAAm-AAc-Lactose hydrogel attached to a cover glass and dried in air prior to Ausputtering. (d) SEM image of 2-D PCs on a swollen 2-D PC PAAm-AAc-Lactose hydrogel that was attached to a cover glass and dried inair before Au sputtering. The PAAm-AAc-Lactose hydrogel adhered tothe cover slide, which prevented its area from shrinking. Thisprevented contraction of the expanded hydrogel 2-D array upondrying.

ACS Sensors Article

DOI: 10.1021/acssensors.7b00426ACS Sens. 2017, 2, 1474−1481

1476

incident at 50° from the normal. The diffraction colors varyfrom the deep red to blue. This diffraction spectrum is similarto that of the 2-D PC on the glass slide shown in Figure S8a.Figure 2c shows an SEM image of an almost close-packed 2-

D PC on the surface of an ∼80-μm-thick PAAm-AAc-Lactosehydrogel sensor that was dried in air. The 2-D PC remains wellordered and almost close-packed (Figure 2c). This hydrogelsensor was removed from the glass slide, swollen in a 0.1 MNaCl solution, and then dried onto a glass slide. During dryingthe swollen hydrogel surface adheres to the glass slide whichprevents the hydrogel area from shrinking. As a result, thehydrogel can only shrink along its thickness. Thus, the 2-Darray spacing is larger in Figure 2d compared to Figure 2c.The analyte lectin protein induced volume response of the

carbohydrate hydrogel sensor suspended in water can bedetermined, by measuring its white light diffraction shifts. Wecan monitor the 2-D PC diffraction wavelength maximum byusing a reflection optical fiber probe in the Littrowconfiguration, as we previously demonstrated.43 Alternatively,we can irradiate the 2-D PC carbohydrate hydrogel sensoralong its normal with a green laser pointer. The light is forwarddiffracted into a Debye ring as shown in Figure S8b. The Debyediffraction ring diameter is directly proportional to the 2-Darray nearest neighbor spacing.49,51,55 Both diffraction methodscan be used to monitor changes in the hydrogel volume thatgive rise to changes in the 2-D array spacing.51,53,55 Generally,Debye ring diffraction measurements are more convenient.Figure 3a shows the dependence of the diffraction spectrum

of the 2-D PC PAAm-AAc-Lactose-80 hydrogel sensor ondifferent concentrations of ricin. These diffraction spectra weremeasured with an Ocean Optics reflection fiber probe in aLittrow configuration. The diffraction wavelength maximum ofthe 2-D PC PAAm-AAc-Lactose-80 sensor shifts from 622 to583 nm as the ricin concentration increases from 0 to 1.0 mg/mL. The inset photographs show the colors of the forward-diffracted light taken with a camera along the normal and with awhite light source below at an angle of ∼76° to the 2D arraynormal. The diffraction color changes from red to green as thehydrogel sensor is exposed to increasing ricin concentrationsfrom 0 to 1.0 mg/mL.Figure 3b shows the ricin concentration dependence of the

particle spacing of the 2-D PC PAAm-AAc-Lactose hydrogelsensors measured by using the Debye diffraction rings. TheseDebye ring particle spacing measurements are consistent withthose measured by using the reflection fiber probe (Figure 3a).The 2-D particle spacing decreases as the ricin concentrationincreases because multivalent ricin binding to the hydrogellactose increases the hydrogel cross-link density. This shrinksthe hydrogel causing a 2-D array particle spacing decrease. Thisarray spacing decrease blue shifts the 2-D PC Bragg diffraction.Figure 3b also shows that larger particle spacing decreasesoccur as the hydrogel lactose concentration increases. A 16 nmparticle spacing decrease occurs upon exposure of the 2-D PCPAAm-AAc-Lactose-20 hydrogel sensor to 1.0 mg/mL ricin,while a larger 51 nm particle spacing decrease occurs uponexposure of the 2-D PC PAAm-AAc-Lactose-40 hydrogelsensor to 1.0 mg/mL ricin. The 2-D PC PAAm-AAc-Lactose-120 hydrogel sensor shows the largest particle spacing decrease(∼124 nm) for a ricin concentration of 1.0 mg/mL.The increasing diffraction blue shifts due to hydrogel lactose

concentration increases occur because an increased lactoseconcentration increases the binding between lactose and ricin,which gives rise to more cross-links and, thus, a larger hydrogel

shrinkage. The increased binding results from both an increasedlactose concentration and an increased affinity caused bymultivalent binding. Thus, a higher lactose content leads to alarger responsivity of the 2-D PC PAAm-AAc-Lactose sensor.We determined a limit of detection (LoD) of the 2-D PCPAAm-AAc-Lactose-120 hydrogel sensor of 9 μg/mL (7.5 ×10−8 M) ricin, which is far below the adult human lethal oraldose of 20−30 mg/kg.62 This LoD value is comparable to amouse bioassay method (∼7.5 μg/mL) but our ricin sensorresponds faster (100 min) compared to the mouse bioassaymethod (24 h).63,64 We also determined that the dynamic rangeof the 2-D PC PAAm-AAc-Lactose-120 hydrogel sensor isbetween 2.5 × 10−7 M and 24.3 × 10−7 M, while its linear rangeis from 7.5 × 10−8 M to 27.5 × 10−8 M (see SI and Figure S9for calculation details).Figure 4 shows the lectin protein and the BSA concentration

dependence of the particle spacing of the 2-D PC PAAm-AAc-Lactose-40 hydrogel sensor measured by using the Debyediffraction ring diameter. The 2-D array particle spacing

Figure 3. (a) Normalized and smoothed diffraction spectra of 2-D PCPAAm-AAc-Lactose hydrogel sensors for different ricin concentra-tions. These measurements were taken in a Littrow configuration withan angle of 14° between the probe and the normal to the 2-D array.The Littrow configuration occurs when the diffracted light is detectedin back diffraction. In the Littrow configuration, the 2-D Braggdiffraction relationship is mλ = 31/2d sin θ, where m is the diffractionorder, λ is the diffracted wavelength (in vacuum), d is the 2-D particlespacing, and θ is the angle of the light relative to the normal to the 2-Darray.60 (b) Ricin concentration dependence of 2-D PC PAAm-AAc-Lactose hydrogel particle spacing changes.

ACS Sensors Article

DOI: 10.1021/acssensors.7b00426ACS Sens. 2017, 2, 1474−1481

1477

decreases with increasing ricin concentrations. This response isdue to selective ricin binding to lactose as evident from thecontrol experiments, where little change occurs upon additionof the proteins BSA, or the other lectin proteins jacalin andCon A. Proteins that do not bind to lactose do not give rise toany diffraction shifts. Thus, this 2-D PC PAAm-AAc-Lactosesensor selectively and sensitively detects ricin.65,66

We further fabricated 2-D PC PAAm-AAc-Galactose and 2-DPC PAAm-AAc-Mannose hydrogels using similar syntheticmethods (Scheme 1). Figure 5 shows the ricin concentration

dependence of the particle spacing of 2-D PC PAAm-AAc-β-Galactose-40, PAAm-AAc-α-Galactose-40, and PAAm-AAc-Lactose-40 hydrogel sensors. At 1.0 mg/mL ricin concen-trations, the α-galactose, β-galactose, and lactose hydrogelsensors show 2, 15, and 51 nm particle spacing decreases,respectively. As expected, the PAAm-AAc-α-Galactose-40hydrogel sensor shows the smallest response because ricinhas a negligible binding affinity to α-galactose. In contrast, ricin(RCA60) is known to specifically bind to β-galactose with asingle binding site that has an association constant of ∼6900M−1. Ricin (RCA60) has larger affinity for lactose since ricin has

two binding sites for lactose with association constants of Ka1 =35 000 M−1 and Ka2 = 2800 M−1.67

The RCA120 used in this study has two binding sites for β-galactose and four binding sites for lactose, respectively, since itconsists of two RCA60, in which the two RCA60 (A-B) chainsare covalently connected by a disulfide bond.67,68 The multiplehydrogel lactose monomers enhance multivalent bindingbetween lactose and the RCA120 ricin protein, which enablessensitive detection of lectin ricin.In this work, we also examined the selectivity of our mannose

and α-galactose hydrogel sensors toward other lectin proteins,such as Con A that binds mannose and jacalin that binds α-galactose.52,69,70 Figure 6a shows the Con A concentration

dependence of the particle spacing change for 2-D PC PAAm-AAc-Mannose and 2-D PC PAAm-AAc-Lactose hydrogels. The2-D PC PAAm-AAc-Mannose hydrogel shows a much largerresponse to 2.0 mg/mL Con A (∼124 nm particle spacingdecrease) due to the strong Con A binding to mannose. Only a∼4 nm particle spacing decrease occurs for the 2-D PC PAAm-AAc-Lactose hydrogel. We fabricated more sensitive Con Asensors by simply increasing the mannose concentration to 160mg/mL (see Figure S10 in SI). The calculated Con A detectionlimit of this unoptimized 2-D PC PAAm-AAc-Mannose-160hydrogel sensor is 3.8 × 10−8 M. The dynamic range of the 2-DPC PAAm-AAc-Mannose-160 hydrogel sensor is between 1.3 ×10−7 M and 29.1 × 10−7 M, while its linear range is from 3.8 ×

Figure 4. Dependence of 2-D PC PAAm-AAc-Lactose-40 hydrogelparticle spacing on BSA and the lectin proteins ricin, Con A, andjacalin. Only ricin binding blue shifts the diffraction.

Figure 5. Ricin concentration dependence of 2-D PC PAAm-AAc-β-Galactose-40, PAAm-AAc-α-Galactose-40, and PAAm-AAc-Lactose-40hydrogel sensor particle spacing changes.

Figure 6. (a) 2-D PC PAAm-AAc-Mannose-40 and 2-D PC PAAm-AAc-Lactose-40 hydrogel sensors for the selective detection of Con A,and (b) 2-D PC PAAm-AAc-Lactose-40 and 2-D PC PAAm-AAc-α-Galactose-40 sensors for the selective detection of jacalin.

ACS Sensors Article

DOI: 10.1021/acssensors.7b00426ACS Sens. 2017, 2, 1474−1481

1478

10−8 M to 42.3 × 10−8 M (see SI and Figure S10 for calculationdetails). Our unoptimized 2-D PC PAAm-AAc-Mannose-160hydrogel sensor shows higher sensitivity and a larger linearrange than the reported colorimetric method for Con Adetection, which was reported to have a LoD of 10−7 M and alinear range from 8 × 10−8 M to 26 × 10−8 M (Table S1 inSI).71

We synthesized the α-galactose carbohydrate hydrogel forsensing jacalin. Figure 6b shows the highly selective response ofthe 2-D PC PAAm-AAc-α-Galactose sensor to the lectin jacalin.An ∼65 nm particle spacing decrease occurs when the PAAm-AAc-α-Galactose hydrogel sensor is exposed to 2 mL of a 1.0mg/mL solution of jacalin. In contrast, a negligible ∼2 nmparticle spacing decrease is observed for the 2-D PC PAAm-AAc-Lactose hydrogel sensor to jacalin. This is because ofjacalin’s much larger association constant to α-galactose ((2.2 ±0.8) × 107 M−1) compared to lactose and mannose.70,72 Byincreasing the α-galactose concentration to 160 mg/mL (seeFigure S11 in SI), we fabricated a more sensitive jacalin sensor.We determined a detection limit of 2.3 × 10−7 M for thisunoptimized 2-D PC PAAm-AAc-α-Galactose-160 hydrogelsensor toward jacalin. The dynamic range of the 2-D PCPAAm-AAc-α-Galactose-160 hydrogel sensor is between 7.6 ×10−7 M and 55.8 × 10−7 M, while its linear range is from 2.3 ×10−7 M to 3.8 × 10−7 M. To the best of our knowledge, this isthe first sensor for jacalin detection (see SI and Figure S11 forcalculation details).We probed the reversibility of our 2-D PC carbohydrate

hydrogel sensors. The response of our PAAm-AAc-Lactosehydrogel sensor over 5 cycles of exposure to 0 and 1.0 mg/mLricin solutions is highly reversible (Figure 7). The particlespacings are relatively constant before and after ricin addition.

The kinetics of 2-D PC PAAm-AAc-Lactose hydrogel sensorresponse to 1.0 mg/mL ricin is shown in Figure 8. The PAAm-AAc-Lactose hydrogel sensor particle spacing decrease saturates∼100 min after exposure to 2 mL of a 1.0 mg/mL ricinsolution. This response follows a single exponential decay e−kt

(k ∼ 0.033 min−1). This relatively slow response is presumablylimited by the slow diffusion rate of ricin (RCA120, Mw = 120000 Da) into the hydrogel which limits the rate of itsmultivalent cross-link formation.

■ CONCLUSIONSWe report the development of carbohydrate hydrogel sensormotifs for the sensitive and selective detection of lectinproteins. A 2-D PC array was attached to a hydrogel containinglectin protein specific carbohydrate to fabricate the sensormotif. Selective multivalent lectin protein binding to thehydrogel carbohydrate forms cross-links, which shrink thehydrogel volume. This blue shifts the diffraction of the 2-D PCsensor. The 2-D PC PAAm-AAc-Lactose sensor selectivelydetects ricin and shows a ricin detection limit of 9 μg/mL (LoD= 7.5 × 10−8 M). The α-galactose hydrogel sensor is selectivefor jacalin (LoD = 2.3 × 10−7 M), while the mannose hydrogelsensor is highly selective for Con A (LoD = 3.8 × 10−8 M).These sensor motifs show very bright diffraction readouts,which may enable visual biological and chemical agentdetection in applications such as food safety, healthcare, andchemical threat monitoring.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acssen-sors.7b00426.

Synthesis protocols and 1H NMR and 13C NMR ofcarbohydrate monomers, photographs of 2-D photoniccrystals and Debye diffraction ring, and details on thecalculations of limit of detection, dynamic range andlinear range of the photonic crystal hydrogel sensors(PDF)

■ AUTHOR INFORMATIONCorresponding Authors*E-mail: xinyuliu@pitt.edu.*E-mail: asher@pitt.edu.ORCIDZhongyu Cai: 0000-0001-9251-8154NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSWe thank Dr. Zhenmin Hong for helpful discussion, Mr.Kyeongwoo Jang for 1H NMR measurements, and Sigma-

Figure 7. Reversibility of the 2-D PC PAAm-AAc-Lactose-40 hydrogelfor ricin sensing.

Figure 8. Kinetics of 2-D PC PAAm-AAc-Lactose-40 hydrogel to ricinat a concentration of 1.0 mg/mL. The hydrogel is 80 μm thick.

ACS Sensors Article

DOI: 10.1021/acssensors.7b00426ACS Sens. 2017, 2, 1474−1481

1479

Aldrich for their generosity in providing expensive chemicals.The authors gratefully acknowledge HDTRA (Grant No. 1-15-1-0038 to S. A.) and University of Pittsburgh (to X. L.) forfinancial support.

■ REFERENCES(1) Lis, H.; Sharon, N. Lectins: Carbohydrate-Specific Proteins ThatMediate Cellular Recognition. Chem. Rev. 1998, 98, 637−674.(2) Karlsson, K.-A. Pathogen-Host Protein-Carbohydrate Interac-tions as the Basis of Important Infections. In The MolecularImmunology of Complex Carbohydrates 2, Wu, A., Ed.; SpringerUS, 2001; Vol. 491, pp 431−443.(3) Dwek, R. A. Glycobiology: Toward Understanding the Functionof Sugars. Chem. Rev. 1996, 96, 683−720.(4) Wang, S.-K.; Cheng, C.-M. Glycan-based diagnostic devices:current progress, challenges and perspectives. Chem. Commun. 2015,51, 16750−16762.(5) Collins, B. E.; Paulson, J. C. Cell surface biology mediated by lowaffinity multivalent protein−glycan interactions. Curr. Opin. Chem.Biol. 2004, 8, 617−625.(6) Lundquist, J. J.; Toone, E. J. The Cluster Glycoside Effect. Chem.Rev. 2002, 102, 555−578.(7) Lee, Y. C.; Lee, R. T. Carbohydrate-Protein Interactions: Basis ofGlycobiology. Acc. Chem. Res. 1995, 28, 321−327.(8) Mammen, M.; Choi, S.-K.; Whitesides, G. M. PolyvalentInteractions in Biological Systems: Implications for Design and Useof Multivalent Ligands and Inhibitors. Angew. Chem., Int. Ed. 1998, 37,2754−2794.(9) Kitov, P. I.; Sadowska, J. M.; Mulvey, G.; Armstrong, G. D.; Ling,H.; Pannu, N. S.; Read, R. J.; Bundle, D. R. Shiga-like toxins areneutralized by tailored multivalent carbohydrate ligands. Nature 2000,403, 669−672.(10) Bigalke, H.; Rummel, A. Medical aspects of toxin weapons.Toxicology 2005, 214, 210−220.(11) Adak, A. K.; Lin, H.-J.; Lin, C.-C. Multivalent glycosylatednanoparticles for studying carbohydrate-protein interactions. Org.Biomol. Chem. 2014, 12, 5563−5573.(12) Fasting, C.; Schalley, C. A.; Weber, M.; Seitz, O.; Hecht, S.;Koksch, B.; Dernedde, J.; Graf, C.; Knapp, E.-W.; Haag, R.Multivalency as a Chemical Organization and Action Principle.Angew. Chem., Int. Ed. 2012, 51, 10472−10498.(13) Eissa, A. M.; Cameron, N. R., Glycopolymer Conjugates. In Bio-synthetic Polymer Conjugates, Schlaad, H., Ed.; Springer: Berlin,Heidelberg, 2013; pp 71−114.(14) Chabre, Y. M.; Roy, R. Multivalent glycoconjugate synthesesand applications using aromatic scaffolds. Chem. Soc. Rev. 2013, 42,4657−4708.(15) Mrazek, H.; Weignerova, L.; Bojarova, P.; Novak, P.; Vanek, O.;Bezouska, K. Carbohydrate synthesis and biosynthesis technologies forcracking of the glycan code: Recent advances. Biotechnol. Adv. 2013,31, 17−37.(16) García, I.; Marradi, M.; Penades, S. Glyconanoparticles:multifunctional nanomaterials for biomedical applications. Nano-medicine 2010, 5, 777−792.(17) Gorityala, B. K.; Ma, J.; Wang, X.; Chen, P.; Liu, X.-W.Carbohydrate functionalized carbon nanotubes and their applications.Chem. Soc. Rev. 2010, 39, 2925−2934.(18) Kikkeri, R.; Lepenies, B.; Adibekian, A.; Laurino, P.; Seeberger,P. H. In Vitro Imaging and in Vivo Liver Targeting with CarbohydrateCapped Quantum Dots. J. Am. Chem. Soc. 2009, 131, 2110−2112.(19) Maljaars, C. E. P.; de Souza, A. C.; Halkes, K. M.; Upton, P. J.;Reeman, S. M.; Andre, S.; Gabius, H.-J.; McDonnell, M. B.; Kamerling,J. P. The application of neoglycopeptides in the development ofsensitive surface plasmon resonance-based biosensors. Biosens.Bioelectron. 2008, 24, 60−65.(20) Kulkarni, A. A.; Weiss, A. A.; Iyer, S. S. Detection ofCarbohydrate Binding Proteins Using Magnetic Relaxation Switches.Anal. Chem. 2010, 82, 7430−7435.

(21) Uzawa, H.; Ohga, K.; Shinozaki, Y.; Ohsawa, I.; Nagatsuka, T.;Seto, Y.; Nishida, Y. A novel sugar-probe biosensor for the deadlyplant proteinous toxin, ricin. Biosens. Bioelectron. 2008, 24, 923−927.(22) Nagatsuka, T.; Uzawa, H.; Sato, K.; Kondo, S.; Izumi, M.;Yokoyama, K.; Ohsawa, I.; Seto, Y.; Neri, P.; Mori, H.; Nishida, Y.;Saito, M.; Tamiya, E. Localized surface plasmon resonance detectionof biological toxins using cell surface oligosaccharides on glyco chips.ACS Appl. Mater. Interfaces 2013, 5, 4173−4180.(23) Ragoussi, M.-E.; Casado, S.; Ribeiro-Viana, R.; Torre, G. d. l.;Rojo, J.; Torres, T. Selective carbohydrate-lectin interactions incovalent graphene- and SWCNT-based molecular recognition systems.Chem. Sci. 2013, 4, 4035−4041.(24) Gorityala, B. K.; Lu, Z.; Leow, M. L.; Ma, J.; Liu, X.-W. Designof a “Turn-Off/Turn-On” Biosensor: Understanding Carbohydrate-Lectin Interactions for Use in Noncovalent Drug Delivery. J. Am.Chem. Soc. 2012, 134, 15229−15232.(25) Sakurai, K.; Hatai, Y.; Okada, A. Gold nanoparticle-basedmultivalent carbohydrate probes: selective photoaffinity labeling ofcarbohydrate-binding proteins. Chem. Sci. 2016, 7, 702−706.(26) Vedala, H.; Chen, Y.; Cecioni, S.; Imberty, A.; Vidal, S. b.; Star,A. Nanoelectronic Detection of Lectin-Carbohydrate InteractionsUsing Carbon Nanotubes. Nano Lett. 2011, 11, 170−175.(27) Schofield, C. L.; Field, R. A.; Russell, D. A. Glyconanoparticlesfor the Colorimetric Detection of Cholera Toxin. Anal. Chem. 2007,79, 1356−1361.(28) Kale, R. R.; Mukundan, H.; Price, D. N.; Harris, J. F.; Lewallen,D. M.; Swanson, B. I.; Schmidt, J. G.; Iyer, S. S. Detection of IntactInfluenza Viruses using Biotinylated Biantennary S-Sialosides. J. Am.Chem. Soc. 2008, 130, 8169−8171.(29) Dinh, H.; Zhang, X.; Sweeney, J.; Yang, Y.; He, Y.; Dhawane, A.;Iyer, S. S. Glycan based Detection and Drug Susceptibility of InfluenzaVirus. Anal. Chem. 2014, 86, 8238−8244.(30) Lin, C.-C.; Yeh, Y.-C.; Yang, C.-Y.; Chen, C.-L.; Chen, G.-F.;Chen, C.-C.; Wu, Y.-C. Selective Binding of Mannose-EncapsulatedGold Nanoparticles to Type 1 Pili in Escherichia coli. J. Am. Chem. Soc.2002, 124, 3508−3509.(31) Xue, C.; Velayudham, S.; Johnson, S.; Saha, R.; Smith, A.;Brewer, W.; Murthy, P.; Bagley, S. T.; Liu, H. Highly Water-Soluble,Fluorescent, Conjugated Fluorene-Based Glycopolymers with Poly-(ethylene glycol)-Tethered Spacers for Sensitive Detection ofEscherichia coli. Chem. - Eur. J. 2009, 15, 2289−2295.(32) El-Boubbou, K.; Zhu, D. C.; Vasileiou, C.; Borhan, B.; Prosperi,D.; Li, W.; Huang, X. Magnetic Glyco-Nanoparticles: A Tool ToDetect, Differentiate, and Unlock the Glyco-Codes of Cancer viaMagnetic Resonance Imaging. J. Am. Chem. Soc. 2010, 132, 4490−4499.(33) Krishnamurthy, V. M.; Estroff, L. A.; Whitesides, G. M.Multivalency in Ligand Design. In Fragment-based Approaches in DrugDiscovery; Wiley-VCH Verlag GmbH & Co. KGaA, 2006; pp 11−53.(34) Hao, N.; Neranon, K.; Ramstrom, O.; Yan, M. Glyconanoma-terials for biosensing applications. Biosens. Bioelectron. 2016, 76, 113−130.(35) Bojarova, P.; Kren, V. Sugared biomaterial binding lectins:achievements and perspectives. Biomater. Sci. 2016, 4, 1142−1160.(36) Yang, H.; Li, H.; Jiang, X. Detection of foodborne pathogensusing bioconjugated nanomaterials. Microfluid. Nanofluid. 2008, 5,571−583.(37) Holtz, J. H.; Asher, S. A. Polymerized colloidal crystal hydrogelfilms as intelligent chemical sensing materials. Nature 1997, 389, 829−832.(38) Asher, S. A.; Alexeev, V. L.; Goponenko, A. V.; Sharma, A. C.;Lednev, I. K.; Wilcox, C. S.; Finegold, D. N. Photonic CrystalCarbohydrate Sensors: Low Ionic Strength Sugar Sensing. J. Am.Chem. Soc. 2003, 125, 3322−3329.(39) Arunbabu, D.; Sannigrahi, A.; Jana, T. Photonic crystal hydrogelmaterial for the sensing of toxic mercury ions (Hg2+) in water. SoftMatter 2011, 7, 2592−2599.

ACS Sensors Article

DOI: 10.1021/acssensors.7b00426ACS Sens. 2017, 2, 1474−1481

1480

(40) Zhang, C.; Losego, M. D.; Braun, P. V. Hydrogel-Based GlucoseSensors: Effects of Phenylboronic Acid Chemical Structure onResponse. Chem. Mater. 2013, 25, 3239−3250.(41) Huang, Y.; Li, F.; Qin, M.; Jiang, L.; Song, Y. A Multi-stopbandPhotonic-Crystal Microchip for High-Performance Metal-Ion Recog-nition Based on Fluorescent Detection. Angew. Chem., Int. Ed. 2013,52, 7296−7299.(42) Qin, M.; Huang, Y.; Li, Y.; Su, M.; Chen, B.; Sun, H.; Yong, P.;Ye, C.; Li, F.; Song, Y. A Rainbow Structural-Color Chip forMultisaccharide Recognition. Angew. Chem., Int. Ed. 2016, 55, 6911−6914.(43) Zhang, J.-T.; Wang, L.; Luo, J.; Tikhonov, A.; Kornienko, N.;Asher, S. A. 2-D Array Photonic Crystal Sensing Motif. J. Am. Chem.Soc. 2011, 133, 9152−9155.(44) Cai, Z.; Smith, N. L.; Zhang, J.-T.; Asher, S. A. Two-Dimensional Photonic Crystal Chemical and Biomolecular Sensors.Anal. Chem. 2015, 87, 5013−5025.(45) Yan, C.; Qi, F.; Li, S.; Xu, J.; Liu, C.; Meng, Z.; Qiu, L.; Xue, M.;Lu, W.; Yan, Z. Functionalized photonic crystal for the sensing of Sarinagents. Talanta 2016, 159, 412−417.(46) Fenzl, C.; Hirsch, T.; Wolfbeis, O. S. Photonic Crystals forChemical Sensing and Biosensing. Angew. Chem., Int. Ed. 2014, 53,3318−3335.(47) Ge, J.; Yin, Y. Responsive Photonic Crystals. Angew. Chem., Int.Ed. 2011, 50, 1492−1522.(48) Zhang, J.-T.; Wang, L.; Lamont, D. N.; Velankar, S. S.; Asher, S.A. Fabrication of Large-Area Two-Dimensional Colloidal Crystals.Angew. Chem., Int. Ed. 2012, 51, 6117−6120.(49) Cai, Z.; Zhang, J.-T.; Xue, F.; Hong, Z.; Punihaole, D.; Asher, S.A. 2D Photonic Crystal Protein Hydrogel Coulometer for SensingSerum Albumin Ligand Binding. Anal. Chem. 2014, 86, 4840−4847.(50) Zhang, J.-T.; Smith, N.; Asher, S. A. Two-Dimensional PhotonicCrystal Surfactant Detection. Anal. Chem. 2012, 84, 6416−6420.(51) Zhang, J.-T.; Chao, X.; Liu, X.; Asher, S. A. Two-dimensionalarray Debye ring diffraction protein recognition sensing. Chem.Commun. 2013, 49, 6337−6339.(52) Zhang, J.-T.; Cai, Z.; Kwak, D. H.; Liu, X.; Asher, S. A. Two-Dimensional Photonic Crystal Sensors for Visual Detection of LectinConcanavalin A. Anal. Chem. 2014, 86, 9036−9041.(53) Cai, Z.; Luck, L. A.; Punihaole, D.; Madura, J. D.; Asher, S. A.Photonic crystal protein hydrogel sensor materials enabled byconformationally induced volume phase transition. Chem. Sci. 2016,7, 4557−4562.(54) Cai, Z.; Kwak, D. H.; Punihaole, D.; Hong, Z.; Velankar, S. S.;Liu, X.; Asher, S. A. A Photonic Crystal Protein Hydrogel Sensor forCandida albicans. Angew. Chem. 2015, 127, 13228−13232.(55) Cai, Z.; Kwak, D. H.; Punihaole, D.; Hong, Z.; Velankar, S. S.;Liu, X.; Asher, S. A. A Photonic Crystal Protein Hydrogel Sensor forCandida albicans. Angew. Chem., Int. Ed. 2015, 54, 13036−13040.(56) Kabir, S. Jacalin: a jackfruit (Artocarpus heterophyllus) seed-derived lectin of versatile applications in immunobiological research1.J. Immunol. Methods 1998, 212, 193−211.(57) Reese, C. E.; Asher, S. A. Emulsifier-Free Emulsion Polymer-ization Produces Highly Charged, Monodisperse Particles for NearInfrared Photonic Crystals. J. Colloid Interface Sci. 2002, 248, 41−46.(58) Lumsdon, S. O.; Kaler, E. W.; Velev, O. D. Two-DimensionalCrystallization of Microspheres by a Coplanar AC Electric Field.Langmuir 2004, 20, 2108−2116.(59) Pan, F.; Zhang, J.; Cai, C.; Wang, T. Rapid Fabrication of Large-Area Colloidal Crystal Monolayers by a Vortical Surface Method.Langmuir 2006, 22, 7101−7104.(60) Krieger, I. M.; O’Neill, F. M. Diffraction of light by arrays ofcolloidal spheres. J. Am. Chem. Soc. 1968, 90, 3114−3120.(61) Tikhonov, A.; Kornienko, N.; Zhang, J. T.; Wang, L. L.; Asher,S. A. Reflectivity Enhanced Two-Dimensional Dielectric Particle ArrayMonolayer Diffraction. J. Nanophotonics 2012, 6, 063509.(62) European Food Safety. Ricin (from Ricinus communis) asundesirable substances in animal feed - Scientific Opinion of the Panelon Contaminants in the Food Chain. EFSA J. 2008, 6, 726.

(63) Zhan, J.; Zhou, P. A simplified method to evaluate the acutetoxicity of ricin and ricinus agglutinin. Toxicology 2003, 186, 119−123.(64) Bozza, W. P.; Tolleson, W. H.; Rivera Rosado, L. A.; Zhang, B.Ricin detection: Tracking active toxin. Biotechnol. Adv. 2015, 33, 117−123.(65) Hester, G.; Kaku, H.; Goldstein, I. J.; Wright, C. S. Structure ofmannose-specific snowdrop (Galanthus nivalis) lectin is representativeof a new plant lectin family. Nat. Struct. Mol. Biol. 1995, 2, 472−479.(66) Drickamer, K. Multiplicity of lectin-carbohydrate interactions.Nat. Struct. Mol. Biol. 1995, 2, 437−439.(67) Zentz, C.; Frenoy, J.-P.; Bourrillon, R. Binding of galactose andlactose to ricin. Equilibrium studies. Biochim. Biophys. Acta, ProteinStruct. 1978, 536, 18−26.(68) Sweeney, E. C.; Tonevitsky, A. G.; Temiakov, D. E.; Agapov, I.I.; Saward, S.; Palmer, R. A. Preliminary crystallographic character-ization of ricin agglutinin. Proteins: Struct., Funct., Genet. 1997, 28,586−589.(69) Baenziger, J. U.; Fiete, D. Structural determinants ofconcanavalin A specificity for oligosaccharides. J. Biol. Chem. 1979,254, 2400−2407.(70) Bourne, Y.; Astoul, C. H.; Zamboni, V.; Peumans, W. J.; Menu-Bouaouiche, L.; Van Damme, E. J. M.; Barre, A.; RougE, P. Structuralbasis for the unusual carbohydrate-binding specificity of jacalintowards galactose and mannose. Biochem. J. 2002, 364, 173−180.(71) Schofield, C. L.; Haines, A. H.; Field, R. A.; Russell, D. A. Silverand Gold Glyconanoparticles for Colorimetric Bioassays. Langmuir2006, 22, 6707−6711.(72) Smith, E. A.; Thomas, W. D.; Kiessling, L. L.; Corn, R. M.Surface Plasmon Resonance Imaging Studies of Protein-CarbohydrateInteractions. J. Am. Chem. Soc. 2003, 125, 6140−6148.

ACS Sensors Article

DOI: 10.1021/acssensors.7b00426ACS Sens. 2017, 2, 1474−1481

1481