Article Cite This: Bioconjugate Chem. XXXX, XXX, XXX−XXX
pubs.acs.org/bc
Mimicking Cell Surface Enhancement of Protease Activity on the Surface of a Quantum Dot Nanoparticle Tiffany Jeen and W. Russ Algar* Department of Chemistry, University of British Columbia, 2036 Main Mall, Vancouver, British Columbia V6T 1Z1, Canada
Downloaded via UNIV OF SOUTH AUSTRALIA on October 28, 2018 at 12:02:02 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
S Supporting Information *
ABSTRACT: Nature has evolved mechanisms to increase the specificity of enzymes and enhance their activity to support particular biological functions. Mimicking these mechanisms with artificial systems, such as nanoparticles, would greatly benefit bioanalysis. Here, we have taken inspiration from platelet cells to create a fluorescent nanoparticle probe with enhanced sensitivity toward its protease target. Platelets use protease-activated receptor 1 (PAR1) to enhance thrombin activity that initiates their aggregation in the early stages of blood clotting. We therefore coconjugated multiple copies each of a peptide substrate and a fragment of PAR1 to a semiconductor quantum dot (QD) to mimic this behavior. Thrombin activity toward conjugates with and without the PAR1 fragment were tracked via Förster resonance energy transfer (FRET). The co-conjugated PAR1 increased thrombin-catalyzed hydrolysis of the substrate by several-fold up to multiple orders of magnitude, albeit dependent on the surface chemistry of the QD. The enhancement effect arose from a combination of selective binding between the PAR1 fragment and thrombin, and from colocalization of the substrate and the PAR1 fragment at the QD interface. Substrate−receptor co-conjugation is thus a promising strategy for the rational design of nanoparticle bioconjugates with optimized sensitivity and specificity for biosensing and imaging.
■
INTRODUCTION Colloidal semiconductor quantum dots (QDs) offer an exceptional combination of optical properties for bioimaging and bioanalysis.1−5 In particular, QDs exhibit bright photoluminescence (PL) that is spectrally narrow, tunable across a wide range of wavelengths through control of size and composition, and resistant to photobleaching.1,2 The surface of QDs can also be functionalized with a variety of ligand or polymer coatings and conjugated with biomolecules such as proteins, peptides, and oligonucleotides.6,7 Successful pairing of these chemistries with the superior optical properties of QDs has led to the development of a wide array of QD-based imaging probes and biosensors, the latter of which often use Förster resonance energy transfer (FRET) to signal detection of their target analyte or biomarker.8 Given the importance of enzymes in biochemistry and disease pathology, it is no surprise that many QD-based biosensors have targeted enzymes and their catalytic activity, where proteases and kinases are common examples of targets.9−17 The aforementioned PL properties, amenability to FRETbased sensing, and well-developed surface functionalization methods also make QDs a good model material for fundamental studies of how inorganic nanoparticles interact with biological molecules and systems.18 These interactions may, for example, be most easily and directly followed using QDs and FRET, but the results may be extensible to other inorganic materials such as gold, lanthanide, and metal oxide © XXXX American Chemical Society
and alloy magnetic nanoparticles. In this context, several studies have sought to better understand how proteases turn over their substrates when the latter are conjugated to a QD.19−23 One of the most interesting findings is that there is often an acceleration of substrate turnover.19−22 The acceleration factors to date are between ∼2-fold and ∼80fold versus equal amounts of substrate without the nanoparticle. Although the causes of this acceleration are not yet fully understood, several contributions to the acceleration have been proposed: a high local concentration of substrate at the surface of a QD, weak affinity between the enzyme and QD, and larger collisional cross sections and more favorable steric factors (from the collision theory of reaction rates). The acceleration appears to be tempered by strong adsorption of protease to the QD and by steric hindrance from bulky ligands on the QD, both potentially leading to deceleration (i.e., inhibition) of turnover.19,20,22,23 Although these studies have shown that interfacial chemistry is able to modulate enzymatic activity toward QD-conjugated substrates, a broadly applicable and predictive model does not yet exist for how enzyme− nanoparticle interactions modulate substrate turnover. Biology has its own mechanisms for modulating enzyme activity. These mechanisms include allostery, post-translation Received: September 12, 2018 Revised: October 9, 2018
A
DOI: 10.1021/acs.bioconjchem.8b00647 Bioconjugate Chem. XXXX, XXX, XXX−XXX
Article
Bioconjugate Chemistry
Figure 1. (A) Model of the primary experimental system. Orange-emitting CdSe/CdS/ZnS QDs are coated with small molecule ligands (represented by the mesh) and conjugated with multiple copies of Sub1(A647), a dye-labeled peptide substrate for thrombin and trypsin, and multiple copies of hpPAR1, a peptide that features the thrombin-binding PAR1 amino acid sequence. Important parts of the amino acid sequences are labeled in the inset, including the hirudin-like sequence (Hir) of hpPAR1. Turnover of substrate by thrombin was measured through changes in FRET. A680 was substituted for A647 in some experiments. (B) TEM image of the QDs. (C) Normalized absorption and PL emission spectra for the QD and A647 and A680 dyes. (D) Structures of the ligands used to coat the QDs. The quantum yield (Φ) of the coated QDs and their approximate hydrodynamic diameter (DH) and zeta potential (ζ) are listed below each structure. (E) PL images of agarose gels showing (top) assembly of various equivalents of hpPAR1 to the QDs and (bottom) lack of assembly of the pPAR1 analog without the polyhistidine tag.
to the surface of QDs as model inorganic nanoparticles. A previous study showed that the adjacent display of PAR1-like substrate and thrombin-binding peptides on a bacterial cell surface increased turnover of the substrate by approximately an order of magnitude,41 suggesting that the enhancement could indeed be engineered into a non-native system, albeit still biotic rather than abiotic. With QDs and other inorganic nanoparticles, previous studies19−23,42−44 have been purely exploratory with respect to the acceleration or inhibition of enzymatic activity toward nanoparticle−substrate conjugates. For example, small libraries of interfacial chemistries and enzymes have been surveyed, but the a priori design of enzymatic acceleration with nanoparticle−substrate conjugates has not been achieved. In contrast, we envisioned coconjugation of the thrombin-binding portion of PAR1 to a QD-peptide substrate conjugate as a specific and rationally designed means of accelerating the turnover of substrate by thrombin. Indeed, we saw an acceleration of thrombin activity with PAR1 coconjugated to a QD-peptide substrate conjugate that, for certain surface chemistries, ranged from 3-fold to orders of magnitude faster than the corresponding QD-peptide substrate conjugates without PAR1. No effect was observed for a protease other than thrombin, nor for unconjugated PAR1 in bulk solution. Acceleration was retained across different batches of QDs functionalized with zwitterionic ligands, where these ligands were essential to the acceleration. Our results show that the capacity of a nanoparticle to display multiple copies of multiple biomolecules is a new opportunity to tune and optimize enzymatic activity. Such opportunities stand to
modifications, the presence or absence of cofactors and endogenous inhibitors, and activation of proenzymes. In the context of proteases, thrombin is an excellent example. This protease has an essential role in blood-clotting, activating platelets to form a clot to staunch bleeding.24,25 Thrombin activity is modulated through several mechanisms, including but not limited to conversion of pro-thrombin by Factor Xa, thrombin activation of other pro-thrombin activating factors, and inhibition through protein C, protein S, antithrombin, heparin cofactor II, thrombomodulin, and tissue factor pathway inhibitor, among others.26−29 As part of its role in clotting, thrombin contributes to the activation of platelets by binding to and cleaving protease-activated receptor 1 (PAR1).25,30 The N-terminal exodomain of this transmembrane protein contains a hirudin-like amino acid sequence (KYEPF) that binds to exosite I of thrombin through a combination of hydrophobic, polar, and electrostatic interactions.31−34 (Hirudin is a thrombin-binding anticoagulant peptide from leeches.) This binding interaction enhances the efficiency of cleavage of PAR1 through both allosteric effects and localization of thrombin to increase the frequency of productive (e.g., properly oriented) interactions with the substrate sequence in the exodomain of PAR1.35−37 It has also been proposed that localization of thrombin to the platelet surface through PAR1 also facilitates cleavage of other nearby receptors (e.g., PAR4) that lack an affinity sequence in their exodomain.34,38−40 Here, we investigated if the ability of PAR1 to enhance thrombin activity was transferable from the surface of platelets B
DOI: 10.1021/acs.bioconjchem.8b00647 Bioconjugate Chem. XXXX, XXX, XXX−XXX
Article
Bioconjugate Chemistry
loading capacity of the QDs (∼60 peptides, see SI for details). An additional control was a peptide with the same sequence as hpPAR1, except for the terminal H6, so it would not assemble to the surface of QDs, as confirmed by gel electrophoresis (see Figure 1E). This peptide is abbreviated as pPAR1. The generic abbreviation Y is used to refer to interchangeable hpPAR1, pPAR1, and Sub2, as X-QD-[Sub1(dye)]8-[Y]N. By virtue of their arginine residues, Sub1(dye) and Sub2 were also substrates for trypsin, which was used as a control protease versus thrombin. hpPAR1 and pPAR1 also had the potential to be substrates for trypsin because of the two lysine residues in their sequence; however, trypsin activity is known to be inhibited by anionic residues at the P2−P5 positions relative to the cleavage site.53 It is thus unlikely that the lysine residue at position 30 of hpPAR1 (and the corresponding residue of pPAR1) was a cleavage site (four preceding anionic residues), but some activity toward the lysine residue at position 20 was likely retained (only one preceding anionic residue). FRET. A647 and A680 were selected as the dye labels for these experiments because they were good FRET acceptors for the orange-emitting QDs yet their PL emission was spectrally well separated from the QD PL emission. Figure 1C shows the absorption and PL emission spectra for A647 and A680 overlaid with those for the QD. The spectral overlap and Förster distance for the QD-A647 pair were J = 1.1 × 10−9 cm6 mol−1 and R0 ranged from 5.0 nm (X = DHLA) to 6.2 nm (X = GSH), respectively. The corresponding values for the QDA680 pair were J = 7.9 × 10−10 cm6 mol−1 and R0 = 4.8−5.9 nm. The range in the values of the Förster distances arose from the different quantum yields between the various X-QD. The ensemble average center-to-center distances between the X-QD and A647 dye was estimated from the FRET efficiency and Förster theory (see SI for details). These distances were in the range ∼6.7−6.9 nm for X = DHLA, DHLA-SB, and DHLA-CB, and ∼7.7 nm for GSH. These distances are well below the maximum possible center-tocenter distance, which is 14.3 nm as the sum of the radius of the QD, the contour length (i.e., full extension) of the peptide, and the length of the dye linker. The FRET-measured centerto-center distance puts the dye ∼2−3 nm from the surface of the QD, similar to what is depicted in the model in Figure 1A, which uses quasi-simulated peptide conformations (see SI for details).54−56 Progress Curves. Figure 2A shows representative examples of progress curves (average of three replicates) from assays of the thrombin-catalyzed hydrolysis of Sub1(A647) as part of XQD-[Sub1(A647)]8-[Y]N conjugates. This data is plotted as the normalized PL ratio versus time and is not converted into a plot of a stoichiometric quantity (e.g., moles of substrate) versus time. The conversion would require a yet unproven assumption of mechanism, and the stochiometric value would differ between an assumption of homogeneous versus heterogeneous hydrolysis of the QD−substrate peptide conjugates (see SI for a detailed discussion).19 This limitation precluded absolute determination of classical enzyme kinetic metrics such as the Michaelis constant (Km) and turnover number (kcat). Instead, the initial rate of change in the PL ratio and the weighted average rate coefficient were used for analysis. Both metrics were determined by fitting the progress curves in Figure 2A to an empirical exponential decay function, as described in the SI. As done previously,19 the metrics were made relative values by normalization to the corresponding
greatly benefit biosensors and imaging probes for enzyme targets.
■
RESULTS QD−Peptide Conjugates. Figure 1A shows a schematic of the QD−peptide conjugates that were primarily used to measure thrombin activity via FRET. The QDs were CdSe/ CdS/ZnS core/shell/shell materials and a TEM image of the as-synthesized nanocrystals is shown in Figure 1B. The QDs had an overall diameter of 9.8 ± 1.3 nm, a first exciton peak at 582 nm, and a PL emission maximum at ∼602 nm with a fullwidth-at-half-maximum of ∼31 nm. Absorption and PL emission spectra for the QD are shown in Figure 1C. For dispersion in aqueous buffer, the QDs were coated with one of four hydrophilic, small-molecule thiol or dithiol ligands. Figure 1D shows the structures of the ligands: glutathione (GSH), a pseudozwitterionic ligand; dihydrolipoic acid (DHLA), an anionic ligand; and DHLA-sulfobetaine (DHLA-SB) and DHLA-carboxybetaine (DHLA-CB), two zwitterionic ligands. The zeta potentials of QDs coated with these ligands were estimated by Ferguson analysis45−47 and are listed below the ligand structures in Figure 1D, along with the hydrodynamic diameters (measured by dynamic light scattering) and representative PL quantum yield values for the ligandcoated QDs in buffer (pH 9.2). The variation in quantum yield arises from the different ligands and corresponding variation in ligand exchange methodology. The X-QDs, where X denotes the ligand coating, were selfassembled with 8 equiv of Alexa Fluor 647- or Alexa Fluor 680 (A647 or A680)-labeled peptide substrates, which had an LVPRGS sequence that was recognized and cleaved by thrombin, and N equivalents of a peptide designed to mimic the exodomain of PAR1 through inclusion of the KYEPFWEDEE amino acid sequence.31,34,41,48−50 These peptides are abbreviated as Sub1(A647 or A680) and hpPAR1, respectively. Sub1(dye) is used as a general abbreviation if the context is interchangeable between A647 and A680. A third peptide, which was also a substrate for thrombin, is abbreviated as Sub2 and was used as a control versus hpPAR1. Full peptide sequences are given in Table 1. The peptides had either a CTable 1. Peptide Sequences Amino acid sequence (N-terminal to C-terminal)a Ac-H6SP6GSDGNESGLVPR↓GSGC-(A647 or A680) H6GGSGGSGGYNPNDKYEPFWEDEEKNESG GGSGGSGGYNPNDKYEPFWEDEEKNESG GGNGSGQNGAAYALVPR↓GSGP5GH6-Am Ac-H6GP5GSDGNEGNLAGSGC
MW (kDa) 3.0 3.8 3.0 3.1 2.6
Abbrev. Sub1(A647 or A680) hpPAR1 pPAR1 Sub2 Pep
a
Single-letter amino acid codes. Am = amidation. Ac = acetylation. The downward arrow (↓) represents the cleavage site. Dye is labeled on the cysteine side-chain of Sub1(dye).
terminal or N-terminal hexahistidine (H6) sequence that is well-known to bind to the ZnS shell of ligand-coated QDs.6,51,52 Figure 1E shows a PL image of an agarose gel with mobility shifts for the QDs that indicate binding of H6terminated peptide. The conjugates used in protease activity assays were X-QD-[Sub1(dye)]8-[hpPAR1]N and X-QD[Sub1(dye)]8-[Sub2]N. The total average number of assembled peptides per QD, N + 8, was always less than the maximum C
DOI: 10.1021/acs.bioconjchem.8b00647 Bioconjugate Chem. XXXX, XXX, XXX−XXX
Article
Bioconjugate Chemistry
Figure 2. (A) FRET-based progress curves for the thrombin-catalyzed hydrolysis of Sub1(A647) in X-QD-[Sub1(A647)]8-[Y]N conjugates, where X = DHLA, DHLA-SB, DHLA-CB, or GSH; Y is nothing, hpPAR1, pPAR1 (does not assemble to the QD), or Sub2; and N varies between 10 and 40. The progress curves are colored according to the identity of Y. For clarity, there is no distinction in color for different values of N; however, the blue arrow points in the direction of the trend with increasing N for hpPAR1. The thrombin (Thr) concentrations are indicated above each graph. (B) Kinetic metrics derived from the progress curves in panels A. Plots of the relative empirical rate coefficient (RERC; left vertical axis for each panel) and the relative initial rate (RIR; right vertical axis for each panel).
The effect of Y = Sub2 was either a small increase in the rate of hydrolysis (X = DHLA, GSH) or a small decrease (X = DHLA-SB, DHLA-CB). Figure 2B summarizes the progress curve results for all X-QD-[Sub1(A647)]8-[Y]N and all values of N. The enhancements in initial rate with thrombin and Y = hpPAR1 were between 7- and 13-fold and 3- and 15-fold for X = DHLA-SB and DHLA-CB, respectively. The enhancements in the average empirical rate coefficient were between 4- and 8fold and 2- and 7-fold, respectively. In contrast, modulation of both the initial rate and empirical rate coefficient of hydrolysis with Y = pPAR1 was between 0.5- and 1.4-fold across all X-QD and all values of N. Likewise, modulation of these values with Y = Sub2 was less than 2-fold. To evaluate the selectivity of the enhancement from hpPAR1, the experiments with X = DHLA-SB were repeated with trypsin as the protease instead of thrombin. As shown in Figure 3, the changes in the initial rates and average empirical rate coefficients of hydrolysis were between 0.8- and 1.7-fold for hpPAR1, pPAR1, and Sub2, collectively, with no special enhancement effect for hpPAR1. This result indicated selectivity and was consistent with the known less-efficient activation of PAR1 by trypsin in biological systems.31,34 Next, the effect of thrombin concentration on hydrolysis rates was evaluated with and without hpPAR1. Figure 4A,B shows progress curves and plots of the average empirical rate coefficient and initial rate for DHLA-SB-QD-[Sub1(A680)]8[hpPAR1]40 and DHLA-SB-QD-[Sub1(A680)]8 as a function of thrombin concentration. Without hpPAR1, turnover of substrate was reliably detectable at ∼50 nM thrombin, whereas with hpPAR1, turnover was reliably detectable at ∼3 nM thrombin (using a 10% decrease in normalized PL ratio over the assay time as the threshold for reliability). For comparison,
values for the control sample, X-QD-[Sub1(A647)]8 conjugates, to aid in comparison of the effect of different Y because PL ratios vary between different X-QD and different values of N, and because different X-QDs engender different rates of proteolysis. As such, if the absolute values of these metrics do not change versus the control sample with no Y, then a value of unity is obtained. Otherwise, the relative enhancement (>1) or inhibition (
