Activation Kinetics of Zipper Molecular Beacons - The Journal of

Dec 4, 2014 - Proteases play key roles in the regulation of normal cellular function, and thus, their deregulation leads to many disease states. Molec...
0 downloads 0 Views 5MB Size
Article pubs.acs.org/JPCB

Activation Kinetics of Zipper Molecular Beacons Tracy W. Liu,†,∥ Juan Chen,† Laura Burgess,† Brian C. Wilson,† Gang Zheng,*,† Lixin Zhan,‡ Wing-Ki Liu,§ and Bae-Yeun Ha*,§ †

Department of Medical Biophysics, Ontario Cancer Institute, University of Toronto, Toronto, Ontario, Canada M5G 1L7 Grand River Regional Cancer Centre, 835 King Street West, Kitchener, Ontario, Canada N2G 1G3 § Department of Physics and Astronomy and Guelph-Waterloo Physics Institute, University of Waterloo, Waterloo, Ontario, Canada N2L 3G1 ‡

ABSTRACT: Proteases play key roles in the regulation of normal cellular function, and thus, their deregulation leads to many disease states. Molecular beacons are promising protease-imaging probes for the detection and characterization of disease as well as for the evaluation of treatment. Inspired by this, we examined the efficiency of zipper molecular beacons (ZMBs) as imaging probes. First, we showed experimentally that the symmetrical ZMB (zip5e5r), bearing 5-arginine and 5-glutamate arms, is as efficient as the asymmetrical zip5e8r in enhancing cell uptake but without the dark toxicity exhibited by the asymmetric zipper. Also, zip5e5r was shown to dissociate more efficiently at pH’s greater than 5. Using a simple two-state binding model, we attributed this to a larger number of charge-pair conformations for zip5e8r. We then measured the ability of soluble matrix metalloproteinases (MMPs) to cleave zip5e5r, and compared their cleavage efficiency with the original photodynamic molecular beacon (PMB). Finally, as a first step toward understanding our observations quantitatively, we simulated the native structures of the peptides GPLGLARK and EGPLGLARRK with charged termini NH3+ and COO− that approximate the PMB and ZMB (with one pair of arginine/ glutamate electrostatic zipper), respectively. We concluded that inclusion of the zipper changes the native structure of the MBs, altering the cleavage efficiency of different MMPs.



INTRODUCTION Proteases are catalytic enzymes involved in many essential reactions for physiological processes. Often during disease development, changes in protease activity occur through the deregulation, upregulation, and/or downregulation of normal proteolytic activity and/or initiation of new enzymes.1−3 This has stimulated the development of protease-targeted imaging strategies for the detection and characterization of disease as well as for the evaluation of treatments.1 Protease imaging provides specific parameters with regard to disease environments including premalignant molecular abnormalities and growth kinetics, angiogenesis growth factors, and disease biomarkers. This is largely due to the identification of a number of protease substrates and the ability to image the change in expression levels of proteases.1 Molecular beacons (MBs) for protease imaging are generating a great amount of promise with the continual discovery of disease biomarkers and the high expression levels of proteases associated with diseases. The MBs are attractive, as their activation is confined to tissues overexpressing the target protease, whereas they remain inactive in nonexpressing tissues. An advantage of protease activation is the high signal amplification, as a small amount of enzyme can continually cleave and activate a countless number of MBs. Protease activated probes also possess a reduced background as a result of the quenching © 2014 American Chemical Society

from the intact form. The MB template can be applied to various enzymes as long as its cleavable peptide substrate is known. MB specificity arises from how specifically the target protease cleaves the peptide linker.4 A natural extension of molecular beacons is to exploit their selective nature for smart therapeutics. To this end, our group has utilized the multifunctional properties of photosensitizers (PSs); they are not only useful as an image-guidance and diagnostic tool but also show photodynamic capabilities for therapeutics.5 Photodynamic therapy (PDT) causes cell damage through the production of cytotoxic singlet oxygen (1O2), other reactive oxygen species, or a photoinduced electron transfer generated by a light activated PS.6 PDT MBs (PMBs), which mimic the concept of classic imaging MBs, have been developed to control the ability of a PS to generate 1O2 only in target cells. PMBs consist of an enzyme-specific peptide linker with a PS and quencher conjugated at its opposite ends. The PS’s photoreactivity and fluorescence remain inactivated until the peptide is cleaved, allowing the PS and quencher to dissociate from one another. Once activated, PMBs restore its photoreactivity and fluorescence production.5 Received: August 27, 2014 Revised: December 4, 2014 Published: December 4, 2014 44

dx.doi.org/10.1021/jp5086813 | J. Phys. Chem. B 2015, 119, 44−53

The Journal of Physical Chemistry B

Article

and compare it with that against the original photodynamic molecular beacon (PMB). To shed some quantitative insight into this observation, we carry out molecular simulations of the native structures of the peptides GPLGLARK and EGPLGLARRK with charged termini NH3+ and COO− that represent approximately the PMB and the ZMB (with one pair of arginine/glutamate electrostatic zipper), respectively. Inclusion of the zipper changes the native structure of the MBs, which can in turn influence the stability of the native structure, altering the cleavage efficiency of different MMPs differently.

Furthermore, we recently introduced an asymmetrical polyarginine/polyglutamate electrostatic “zipper” to address some of the challenges plaguing PMBs.5,7 For example, we note that MBs should contain enzyme-specific linkers that have conformations favorable for the effective silencing of the MBs, i.e., those that allow the quencher and PS to be brought together for the silencing (i.e., for minimal background fluorescence). This requirement then sets a limit on types of proteases that can be imaged this way, since this is not satisfied by all protease cleavable linkers. Furthermore, extracellularly activated MBs may diffuse to nontarget cells/tissues before target cell uptake occurs, and MBs use nonspecific/passive delivery to target cells/tissues, which is suboptimal. Zipper molecular beacons were designed to enhance activation kinetics and target-cell uptake as dissociation of the zipper releases the cell penetrating polycation, while eliminating a specific peptide sequence/ structure required for conventional MBs and PMBs.8 The inclusion of the zipper allows any protease to be targeted, if the cleavage sequence is known. Unfortunately, this approach presented several challenges with activation kinetics and toxicity. With a growing interest in implementing intelligent probes into the clinic, understanding the complexity associated with the polycation/polyanion construct is desirable. In this paper, we examine activation and cleavage of zipper molecular beacons (ZMBs) with varying pH so as to mimic a disease environment, in which the pH can be as low as ∼4, i.e., a hypoxic environment. We first show experimentally that the symmetrical ZMB including 5-arginine and 5-glutamate arms (zip5e5r) (here the lower (upper)-case letters refer to D (L) amino acids) is a better imaging probe than the asymmetrical zip5e8r. First, the symmetric zip5e5r is as efficient as the asymmetrical zip5e8r in terms of cell uptake, but importantly, it does not show the dark toxicity exhibited by the asymmetric zipper. Also, the symmetrical zip5e5r dissociates more efficiently at pH’s greater than 5, as desired for effective imaging probes. Using a simple twostate binding model, we explain this observation: the reduced efficiency of the asymmetric zipper is attributed to a larger number of its charge-pair conformations, which tends to stabilize the zipped state entropically. We then measure the cleavage efficiency of soluble matrix metalloproteinases (MMPs) against zip5e5r,



EXPERIMENTS AND RESULTS

Our earlier asymmetrical ZMB, (e)5GPLGLAR(r)8K (abbreviated as zip5e8r), has several limitations. Compared to our PMB,5 the quenching of the ZMB was several-fold superior, but its activation kinetics in vitro were at best on the same order. Furthermore, zip5e8r demonstrated dark toxicity not seen with the PMB (Appendix B, Figure 9). This is not surprising, as an 8-amino acid polycation cell penetrating peptide (e.g., 8r) has been reported to have toxicity issues.9−11 A related point is that effective cell penetration appears to require at least six consecutive cationic amino acids,9 which we had also interrogated using fluorescein as the cargo.8 However, the efficiency of a cell penetrating peptide must also be influenced by the cargo it is delivering. Indeed, the cargo pyropheophorbide-α (Pyro) has cell penetrating capabilities itself and, thus, may enhance the cell delivery function of the polyarginine. This nontrivial effect was assessed by conjugating Pyro to a 5-amino acid polyarginine, the minimum length of the polycation required to form a zipper but previously shown to have limited cell penetrating capabilities; this combined system was tested against an 8-amino acid polyarginine and Pyro alone (Figure 1); detailed experimental methods can be found in Appendix C. Within 10 min, Pyro fluorescence is detectable within the cytoplasm of cells by both 5-polyarginine-Pyro and 8-polyarginine-Pyro conjugates; over time, regardless of the polyarginine length, the delivery of Pyro into the cell is enhanced when compared to Pyro alone. This suggests that, when Pyro is used as the cargo, a 5-polyarginine sequence is just as efficient as an 8-polyarginine sequence in enhancing cell uptake. By decreasing the number of

Figure 1. Representative confocal images of cell penetrating capability of Pyro alone and Pyro-5r and Pyro-8r incubated with MT-1 cells for 10, 30, and 60 min. Left columns represent fluorescence images, while right columns show corresponding brightfield images of KB cells. The white scale bar represents 50 μm. See Appendix C for details. 45

dx.doi.org/10.1021/jp5086813 | J. Phys. Chem. B 2015, 119, 44−53

The Journal of Physical Chemistry B

Article

positive charges in the zipper, as in a zipper consisting of five polyarginines and five polyglutamates (zip5e5r), it may show more efficient activation kinetics than the asymmetric zipper, as the electrostatic stability is not as strong, while maintaining the enhanced delivery function. The in vitro activation kinetics of zip5e8r were on a similar, if not slower, time scale as PMBs, a dilemma arising from the two-step activation requirements possibly due to the slow dissociation of the polycation/polyanion. The dissociation status of the zipper mechanism versus pH was then measured for the asymmetric and symmetric ZMB (Figure 2, see Appendix C for

pKa, the acid dissociation constant for glutamic acid, the relative dissociation between the two zip PMBs should be similar. This was mirrored in the fluorescent fold increase from dissociation of the zipper arms at different pH’s (Figure 2C). To understand the experimental observations of the zipper dissociation, we introduce a simple polymer model (see Figure 3). Details of this model are given in Appendix A, where we derive an expression for the relative dissociation, eq A.7. The relevant parameters are (1) the Bjerrum length SB, which is the length scale at which the electrostatic energy between two charges equals the thermal energy; (2) the Debye length 1/κ beyond which the electrostatic interaction is screened; (3) the nearest neighbor distance a between ion pairs on the ionic arms of the MB; (4) the difference between the number of cations and the anions on the polyionic arms δn = n+ − n−; (5) the pK values of the amino acids; and (6) the pH of the solution. The pH and pK values determine the probabilities of protonation for the amino acids (see eq A.2). In Figure 4, we have plotted our theoretical results for the relative dissociation, which can be compared with the experimental results of Figure 2A,B. The experiments were done in the presence of NaCl (137 mM), KCl (2.7 mM), Na2HPO4 (10 mM), and KH2PO4 (2 mM). Accordingly, we have chosen κ−1 ≈ 0.74 nm. The choice of a needs to be biologically inspired. When we chose a = 0.21 nm, the agreement between theoretical and experimental results for the case of zip5e5r was best. [Ionic sizes in water are influenced by hydration. The resulting hydrated radii are appreciably larger than the bare values. When two ions form a pair, the degree of their hydration is reduced so as to enhance their Coulomb attraction, thus altering a. In our coarse-grained (implicit-solvent) model, all this effect is subsumed into the single parameter a. Our choice of this parameter is in part to ensure the best agreement between theoretical and experimental data; also it falls in the accepted range (see, for instance, Israelachvili12).] At low pH, the negative charge on acidic glutamate monomers is neutralized more effectively, and their average charge is thus smaller in magnitude. This diminishes the electrostatic binding and promotes dissociation, as shown in the figure. Below we discuss both similarities and differences between the theoretical and experimental results. The general agreement between the theoretical and experimental results, especially in the large pH range ≳5, suggests that our theoretical model captures some of the features observed experimentally. In particular, it offers a physical picture of why the asymmetric case is more stable against dissociation. In contrast to the symmetric beacon, the asymmetric one can be in any of four different zipped states and is entropically more stable. As pH decreases, however, the relative dissociation reaches a plateau more easily in the theoretical results. Some discrepancy is expected, however, since our theoretical approach leaves out some molecular details. First, the spatial dependence of pH values is ignored, even though the concentration of H+ depends on the “surface” potential of polyions (e.g., the polycationic or polyanionic arm). Neutralization of anionic groups at low pH will be less effective near a cationic surface than in the bulk. If this were taken into account, as pH is lowered, the dissociation curve would change more gradually, as seen in Figure 2, than indicated in Figure 4. Also, the twostate model is less realistic at low pH, since in this case local opening of ion pairs becomes more appreciable and contributes to both the binding energy and the chain entropy. Considering the simplicity of our model, however, the agreement is encouraging.

Figure 2. Relative dissociation of (A) zip5e5r and (B) zip5e8r and corresponding (C) fluorescent fold increase of zip beacons at different pH’s. All experiments done in triplicate, where * represents p < 0.05.

experimental methods). Statistically significant differences were seen in the relative dissociation of zip5e5r versus zip5e8r at pH 7.3; approximately 55% of the zip5e5r polycation/polyanion was dissociated corresponding to a 40-fold increase in fluorescence, whereas only 35% of the zip5e8r polycation/polyanion construct was dissociated corresponding to a 20-fold increase in fluorescence. Statistically significant differences in the relative dissociation of the zipper arms between the two constructs were observed up to pH 5.5. As the pH decreases and approaches the 46

dx.doi.org/10.1021/jp5086813 | J. Phys. Chem. B 2015, 119, 44−53

The Journal of Physical Chemistry B

Article

Figure 3. Two-state polymer model of the molecular zippers. Monomers are strung together through covalent bonds represented by solid lines. They are identical except for their ionization status. A polycationic arm is attached to one end of a nonionic chain, consisting of s monomers, and a polyanionic arm to the other end. Red dashed lines describe ion pairing. (a) Closed and open states of the symmetric zip5e5r. In the closed state, all charges are paired; in the open state, all charges are “free”. The chain entropy is solely determined by the number of looped conformations. (b) Distinct charge-pairing conformations of the asymmetric zip5e8r. They also contribute to the chain entropy.

native structure of a protein/peptide that lies at the global minimum of free energy can be approximated by the global energy-minimum state. The zipper molecular beacon studied in Figure 6, zip5e5r, consists of five arginines and five glutamates. For the simulation efficiency, however, we only introduce one arginine and one glutamate in the zipper part. Using the simplified system, we examine how the incorporation of the zipper would affect the native structures of molecular beacons GPLGLARK and eGPLGLARrK, with lower case letters indicating D-amino acids. However, due to the limitation of the force field we employed, no D-amino acid could be simulated. Instead, we replaced the D-amino acids with their corresponding “standard” ones and studied EGPLGLARRK, with the expectation that the replacement would not alter the native structures significantly. In our simulations, the empirical force field ECEPP/315 was employed to describe the interatomic interactions of the system and the implicit solvent model OONS16 to account for the solvent effect. To improve the simulation efficiency, the basin hopping (BH) MC method17,18 was used in the search, and peptide dihedral angles ω were fixed at 180°. Since the interactions between the photosensitizer (PS) and the quencher are not well understood, we simulated two situations: with and without terminus Coulomb interactions. For the former, we set the N-terminus and C-terminus to NH3+ and COO− groups, respectively, and for the latter, to neutral NH2 and COOH groups. Figure 7 shows the native structures of (a) GPLGLARK and (b) EGPLGLARRK, when there is no Coulomb interaction between the two termini. All the structures obtained are α-helices, with the two termini well separated apart. In molecular beacons, such configurations will push away the PS and the quencher, which is in conflict with their characteristics of optical silence. Figure 8 shows the native structures of the two beacons [(a) GPLGLARK and (b) EGPLGLARRK] when the Coulomb interaction between the two termini is turned on. Clearly, panel b exhibits an obvious beta hairpin structure which is similar to the polymer model of Figure 3, while panel a shows no beta hairpin structure but a looplike structure. If additional E’s and R’s are included, it may become more like a hairpin. For both configurations, the sensitizer and the quencher are close enough to maintain optical silence. In all cases, the amino acid connection (GPLG*LARK) was exposed to the environment. While a more complete picture is lacking, it is tempting to interpret it as enhancing cleavage in the presence of cleavage enzymes, since the beacon is now better poised for enzyme binding and activity. To better relate the computational model to the theoretical one, one should sample a much wider conformational space

Figure 4. Relative dissociation of zip5e5r and zip5e8r as a function of pH. At low pH, the average charge on the acidic glutamate monomer is smaller in magnitude. This diminishes the electrostatic binding and promotes dissociation. The lines through the theoretical results are drawn to guide the eye.

The ability of different matrix metalloproetinases (MMPs) to cleave the MMP-targeted peptide linker in ZMBs and PMBs is compared in Figure 5. The activation specificity and cleavage efficiency shown by different MMPs were determined by individually incubating PMB or zip5e5r with each secreted MMP. Activation analysis was determined by high-performance liquid chromatography (HPLC). The details of methods and materials can be found in Appendix C. The MMP’s target linker was modeled after a secreted MMP substrate, and thus, we focused our studies on using secreted MMPs only. PMB and zip5e5r are specifically cleaved by MMP3, MMP7, MMP9, MMP10, and MMP12MMPs from the gelatinase and stromelysin family and matrilysin. Furthermore, all five MMPs cleave the peptide substrate linker specifically between the glycine and lysine in our ZMB/ PMB peptide sequence, GPLG*LARK (the cleavage site shown by the asterisk). Surprisingly, inclusion of the zipper did not affect the number of MMPs that were able to cleave the peptide linker, GPLGLARK. However, when the cleavage efficiency was compared for the two probes, different cleavage efficiencies were observed (Figure 6). A statistically significant difference was seen in the cleavage efficiency by MMP7, MMP9, and MMP12. The ZMB zip5e5r was cleaved more efficiently by MMP7 and MMP9, whereas PMB was cleaved more effectively by MMP12. In order to supplement the simple model employed in Appendix A, we performed Monte Carlo (MC) simulations in searching for the native states of the beacons. It is our view that they are implicated in the cleavage efficiency/specificity of MMPs. According to the thermodynamic hypothesis,13,14 the 47

dx.doi.org/10.1021/jp5086813 | J. Phys. Chem. B 2015, 119, 44−53

The Journal of Physical Chemistry B

Article

Figure 5. Cleavage specificity of the zipper beacons. HPLC traces of (A) PMB and (B) zip5e5r showing the MMPs that cleave the beacons after 24 h of incubation. Time measured in minutes. Colors of the traces are irrelevant, while the red and blue vertical lines indicate the intact and cleaved zip5e5r.

beyond what the native state represents and understand how likely non-native states will be populated. We will then acquire a better sense of how much detail should be included. Because of the highly demanding computational power for such considerations, we content ourselves with the complementary nature of the two approaches: if the theoretical model captures pH-dependent dissociation of the beacon as well as the interplay between chain entropy and energy in determining the dissociation, the numerical one offers a detailed picture of the native state, which will be eventually beneficial for understanding the relationship between the beacon structure and the cleavage activity of an enzyme. An emerging picture from our results in Figures 7 and 8 is that the attraction between the sensitizer and the quencher by themselves is not strong enough to hold stably the sensitizer− quencher pair. The change in structure of the cleavable linker by inclusion of merely one anionic and one cationic amino acid on each end indicates that the inclusion of the zipper will bring

about similar effects. To explain quantitatively the observed differences in cleavage efficiency between PMB and zip5e5r by different enzymes within the same family requires detailed knowledge of the interaction between the MMPs and the molecular beacons, which unfortunately is not available at present.



DISCUSSION While PMBs have demonstrated much promise, the inclusion of the zipper allows superior activation kinetics and enhanced cell uptake, since the zipper causes the beacon to be better quenched and dissociation of the zipper releases the cell penetrating polycation arm. Here, we have investigated the activation kinetics (here kinetics refers loosely to zipper stability) of ZMBs, showing that the symmetric zip5e5r can enhance the activation kinetics of PMBs, while demonstrating similar cell uptake kinetics but reduced toxicity compared with the asymmetric zip5e8r (Figure 10). Furthermore, the inclusion of the zipper eliminates the quenching dependence on the 48

dx.doi.org/10.1021/jp5086813 | J. Phys. Chem. B 2015, 119, 44−53

The Journal of Physical Chemistry B

Article

Figure 6. Cleavage efficiency of the zip5e5r versus the original beacon (PMB) for specific MMP enzymes. There is a significant increase (* represents p < 0.01) in the cleavage efficiency of zip5e5r by MMP7 and MMP9; however, there is a significant decrease in the cleavage efficiency of the zipper beacon by MMP12 when compared to the original beacon.

Figure 8. Native structures of (a) GPLGLARK and (b) EGPLGLARRK with charged termini NH3+ and COO−.

targeted peptide−linker sequence, allowing any cleavable sequence to be used and opening the possibility to target any protease. Originally, we reported that PMB was specifically activated by matrilysin (MMP7), but upon closer inspection, we found that PMB was cleaved by five different MMPs (Figure 5A), the same group that specifically cleaved the ZMB zip5e5r (Figure 5B). This is not surprising as MMPs are a family of structurally related, multifunctional zinc-dependent endopeptidases that degrade extracellular matrix proteins and process many biological molecules.19 MMP3, MMP7, MMP9, MMP10, and MMP12 have several similar substrates including collagen IV, gelatin, elastin, fibronectin, and proteoglycan.20 Each MMP has the capability to cleave a large number of substrates and it is this multifunctional nature of MMPs that gives rise to their cross-reactivity.21 Cross-reactivity of proteases is almost unavoidable, especially when targeting MMPs.21−23 Depending on the application of the beacon, the cleavage of PMB by more than one target MMP may prove to be advantageous as the activation is further amplified. MMPs are often overexpressed in many diseases where the entire family, itself, is upregulated. Thus, a small amount of each MMP can cleave multiple beacons, enhancing the sensitivity of ZMB to detect molecular changes in disease. MMPs often coexist in vivo, and a number of MMPs are found within a single

Figure 7. Native structures of (a) GPLGLARK and (b) EGPLGLARRK with neutral termini NH2 and COOH. Green, carbon; blue, nitrogen; red, oxygen; white, hydrogen. PLGLA in part a and GPLGLAR in part b are in the helix. In both panels, the N-termini are at the left side. 49

dx.doi.org/10.1021/jp5086813 | J. Phys. Chem. B 2015, 119, 44−53

The Journal of Physical Chemistry B

Article

As a first step toward understanding these observations, we simulated the native structures of the peptides GPLGLARK and EGPLGLARRK with charged termini NH3+ and COO− that approximate the PMB and a ZMB (with one pair of arginine/ glutamate electrostatic zipper), respectively. We concluded that inclusion of the zipper does change the native structure of the MBs, and it could influence the stability of the native structure. However, a quantitative understanding of the experimental results requires further investigation of the interaction of the MMPs with the MPs.

signaling cascade. In addition, activated MMPs often participate in the processing of other MMPs, for example, activated MMP3 cleaves the MMP7 proenzyme releasing catalytic MMP7.24 Meanwhile, activated MMP7 cleaves the MMP9 proenzyme, activating MMP9.24 A growing body of evidence shows that MMPs contribute to tumor progression through a variety of mechanisms and play a major role in cell proliferation, migration, differentiation, angiogenesis, apoptosis, and host defense.19,25,26 MMP expression is upregulated in almost all types of human cancers, often associated with a more aggressive phenotype and an overall poor prognosis.19,25,26 Therapies that alter the tumor environment (alter growth, angiogenesis, or metastatic potential) may cause changes in MMP proteolytic activity which could be detected by ZMB and provide an indicator of tumor responsiveness for a treatment.27 MMP upregulation is not solely confined to cancer; it is also reported to play a role in the formation of atherosclerotic lesions.20,28 The overexpression of MMPs is thought to be involved in the growth, destabilization, and eventual rupture of atherosclerotic lesions.20 MMP3, MMP7, MMP9, MMP10, and MMP12 expression and activity are increased in atherosclerotic plaque20 as well as in associated macrophage-rich regions.28 Therefore, ZMB could reveal the role of MMPs in plaque formation and provide a means to prevent excessive MMP activity, resulting in the stabilization of vulnerable atherosclerotic lesions.28 By understanding the specific characteristics of each of these individual smart probes, we can better interrogate systems for MMP activity. For example, the activity of MMP7 and MMP9 will be observed more efficiently using zip5e5r, whereas MMP12 activity is better assessed using PMB. Thus, within the same system, even with the cross-reactivity of MMPs, utilizing different beacons will allow us to evaluate individual MMP activity. Here lies a need for the development of in vivo tools to dictate appropriate doses and therapeutic windows for therapy as well as the development of a method to identify suitable patients for specific therapies, a role that ZMBs could fulfill with the increasing involvement of MMPs in disease development. The knowledge that individual presents with distinct disease characteristics further supports the transition toward personalized medicine which tailors treatments based upon a patient’s individual disease signature. This goal will be partially realized through ZMB, as it targets proteases essential in disease progression but also using ZMB as a template to create other beacons targeting a vast array of disease biomarkers.



APPENDIX A

Polymer Model of the Molecular Beacon

In this Appendix, we introduce a coarse-grained, polymer model of the molecular beacons. In this model, except for their charge properties, all monomers (i.e., amino acids) are assumed to be identical (e.g., in their size b), as illustrated in Figure 3. A polycationic or polyanionic “arm” consisting of n+ or n− ionizable monomers, respectively, is attached to one end of a linear chain formed by s non-ionic monomers. Chain opening or closure is governed by the balance between chain entropy and energy.29 If chain closure is favored by the electrostatic attraction between opposite charges, chain opening is driven by chain entropy. As for chain conformations, we introduce a two-state model, in which they are classified in two subgroups: (i) “closed” or “zipped” (with all opposite charges paired when n+ = n− or maximally paired otherwise) and (ii) “open” or “unzipped” (all open). For simplicity, we do not consider “breathing” or local opening of ion pairs. For electrostatic considerations, it proves useful to introduce SB = e2/4πε0εrkBT, known as the Bjerrum length, where ε0 is the permittivity of free space and εr the dielectric constant of the solvent.30 This is a length scale at which the electrostatic energy of two elementary charges e is equal to kBT and plays a fundamental role in biomolecular electrostatics. In water, at room temperature, εr ≈ 80 and SB ≈ 7 Å. Besides, there are two other relevant length scales: (i) a (= a few Å), the distance between (the nearest neighbor) ion pairs on the cationic and anionic arms, and (ii) the Debye screening length, denoted as κ−1, i.e., a length scale beyond which the electrostatic interaction is exponentially screened. If ni is the concentration of ions of the ith kind and charge qi, κ2 = 4πSB∑i qi2ni. The electrostatic energy of the beacon in the closed state is given as a pairwise sum over all interaction pairs (i.e., the nearest neighbor, the next-nearest neighbor, ...). For the symmetric case (n+ = n− = n), it becomes



SUMMARY In this work, the symmetrical zipper molecular beacon zip5e5r has proven to be a better imaging probe than the asymmetrical zip5e8r. First, the symmetric one is as efficient as the asymmetric one in enhancing cell uptake, but only the latter shows dark toxicity. Furthermore, the symmetrical zip5e5r dissociates more readily at pH’s greater than 5, a desirable feat for an effective imaging probe. Our coarse-grained (two-state) binding model suggests that the asymmetric zipper is entropically more stable against dissociation. Since the symmetrical zip5e5r shows preferable activation kinetics, we have measured the ability of the matrix metalloproteinases MMP3, MMP7, MMP9, MMP10, and MMP12 to cleave zip5e5r, and compared the cleavage efficiency of these MMPs with that of the original PMB. While zip5e5r is clearly more efficiently cleaved by MMP7 and MMP9, the opposite is true for MMP2 and MMP12, whereas the cleavage efficiencies by MMP3 and MMP10 for zip5e5r and PMB are comparable.

n−1 n−1

2

2 2 1/2

En e−κ(a +|i − j| b ) = −SB(1 − Pα)Pα ′ ∑ ∑ kBT a 2 + |i − j|2 b2 i=0 j=0

(A.1)

where Pα = 1/(1 + 10 pH − pKα)

(A.2)

is the probability of protonation for an amino acid of type α; e(1 − Pα) and ePα are the average charges on basic (e.g., Arg or R) and acidic (e.g., Glu or E) amino acids, respectively. For the asymmetric case of n+ > n− = n, however, some of the cations will remain unpaired, possibly interacting with the paired ones (see Figure 3b). This charge−dipolar interaction is, however, a weaker effect, when the average charge on Arg is comparable in magnitude to that on Glu or (1 − Pα) ≈ Pα′ (i.e., around neutral pH). In this limit a/b → 0, as is often the case, 50

dx.doi.org/10.1021/jp5086813 | J. Phys. Chem. B 2015, 119, 44−53

The Journal of Physical Chemistry B

Article

straight, not to perturb the electrostatic interaction; it is our view that ion pairing between the cationic and anionic arms enhances straightening of the arms, similarly to what we would expect from β structures a polypeptide chain forms. This is equivalent to assuming that the energy of the zipper part is dominant over its conformational entropy. The statistical weight for the closed state Wclosed can thus be written as

this contribution vanishes. In other words, the unpaired cations do not contribute to the electrostatic binding energy of the beacon. This allows us to omit this term for the asymmetric case. Even at low pH, this simplification will not introduce a significant error, since the nearest-neighbor interaction dominates the binding energy. On the other hand, a polymer chain stores large degrees of freedom, whether linear or circular. The resulting entropy depends on the strength of self-avoidance, which keeps two monomers from approaching each other any closer than some length ≈ b. It thus reduces the size of the conformational space and chain entropy. An appropriate model needs to be chosen accordingly. For a short or stiff chain, self-avoidance is less important. For the beacon we consider here, the number of monomers in the loop part is s = 8. If the chain were perfectly flexible, the significance of self-avoidance would be marginal. However, the bond between amide planes is not completely flexible and as a result a poly-peptide chain shows local stiffness over about two monomers. This makes self-avoidance less significant. While it is, in principle, straightforward to include self-avoidance, we seek a minimalist model that captures the general trend seen in the experiments. We employ the random walk (RW) chain model and test it against the data. Let W0 be the number of looped conformations. Note that W0 depends on how it is counted or how the conformational space is constructed. Accordingly, for the computation of W0, insignificant numerical prefactors are dropped. For the symmetric case zip5e5r, it is given by29 W 0sym =

z s+1 (s + 1)3/2

sym W closed = W 0syme−En / kBT

For the asymmetric case zip5e8r, can be given as a sum over distinct ion pairing states, as illustrated in Figure 3b. Here, we assume that these closed states represent those realized in the experiment. Unlike the symmetric case in Figure 3a, this is rather a strong assumption, since it is not clear how the dye and quencher at the chain ends would affect chain statistics. Currently, we do not have a molecular picture of closed states especially for the asymmetric case. While our model can be refined as more details become known, we consider the conformations in Figure 3a as valid ones. For the reason explained above, the electrostatic energy for this case will be the same as that for the asymmetric case. As a result, we find δn asym W closed =

Wopen + Wclosed

z s + δn e−En / kBT (s + 1 + i)3/2

(A.5)

where δn = n+ − n− = 3 for zip5e8r. For the symmetric case zip5e5r, δn = 0, and this result reduces to the one in eq A.4. In sym other words, Wasym closed includes Wclosed as a special case. On the other hand, the statistical weight for the open state can readily be obtained, if we ignore the aforementioned “straightening” of each arm:

(A.3)

Wopen

∑ i=0

where the coordination number is z = 6 for a square lattice, on which the RW lives. The term on the numerator is the expected total number of steps the RW can take on the lattice. The term in the denominator describes the reduction of this number, because of the global constraint that the two ends meet. In contrast to the loop part, the zipper part tends to remain Pdiss =

(A.4)

Wasym closed

Wopen = z 2n + δn + s − 1

(A.6)

Note that (2n + δn + s − 1) is the total number of links or bonds in the chain, which is the total number of monomers minus one. Finally, the relative dissociation is given by the ratio 1

=

(

δn

1 + ∑i = 0



z1 − 2n 3/2

(s + 1 + i)

B

α

α′

2

2 2 1/2

n − 1 n − 1 e−κ(a +|i − j| b ) ∑j=0 2 i=0 a + |i − j|2 b2

) exp⎢⎣S (1 − P )P ∑

⎤ ⎥ ⎦

(A.7)

Figure 9. Toxicity of zip5e8r. (A) Dark toxicity of zip5e8r with varying concentrations and incubation times. Error bars represent +1 standard deviation for three independent experiments. Confocal imaging of MT-1 cells 4 h post incubation with 5 μM concentration of (B) zip5e8r or (C) PMB showing (i) fluorescence and (ii) corresponding brightfield images. 51

dx.doi.org/10.1021/jp5086813 | J. Phys. Chem. B 2015, 119, 44−53

The Journal of Physical Chemistry B

Article

monitored by real-time fluorescence increase at 37 °C on a HORIBA FluoroMax-4 spectrofluorometer (excitation 650 nm, emission 675 ± 5 nm). Ten μL of the cleaved zip beacon solutions was then diluted to 200 μL in different pH buffers (7.3, 6.9, 6.2, 5.5, 4.5, 3.8, 3, 2.3) and allowed to equilibrate for 30 min. The fluorescence due to zipper dissociation resulting from different pH solutions was evaluated. The relative dissociation amount was calculated by normalizing the fluorescence to pH 2.3. MMP Cleavage Studies. Beacons (1 nmol) were first dissolved in 2.5 μL of DMSO and 0.5 μL of Tween 80. The solution was diluted to 200 μL in MMP cleavage buffer and then incubated with 0.5 nmol of MMP (MMP kit, Enzo Lifesciences) for 24 h at 37 °C. The cleavage solutions were then analyzed by reverse-phase analytical HPLC using a Waters 2695 controller with a 2996 photodiode array detector. The HPLC method for PMB was performed on a Zorbax 300SB-C8 column (4.6 mm × 250 mm) using the following method: solvent A, 0.1% TFA and water; solvent B, acetonitrile; gradient, from 80% of A and 20% of B to 100% of B over 35 min, and finally back to 80% of A and 20% B in 1 min and kept for 4 min; flow rate, 0.8 mL/min. The HPLC method for zip5e5r was performed on a a Zorbax 300SB-C3 column (4.6 mm × 150 mm) using the following method: solvent A, 0.1% TFA and water; solvent B, acetonitrile; gradient, from 80% of A and 20% of B to 100% of B over 12 min, and finally back to 80% of A and 20% B in 1 min and kept for 2 min; flow rate, 0.8 mL/min. Dark Toxicity Studies. MT-1 cells were grown to 60% confluence in a NuncLab-TekII-96-well black walled plate. RPMI 1640 medium with 10% fetal bovine serum containing a 0.5, 1, and/or 5 μM probe (diluted in 2.5 μL of DMSO and 0.5 μL of Tween80) was added, and the cells were incubated for 24 h at 37 °C under 5% CO2. Cell viability was then determined by means of the colorimetric MTT assay. Briefly, after 24 h of incubation with probes, the medium was removed and 3-(4,5-imethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (Invitrogen) solution in medium (0.5 mg/mL, 150 μL) was added to each well followed by incubation for 2 h under the same environment. A 150 μL portion of a 1:1 ratio of DMSO to 70% iso-propanol in 0.1 M HCl (10% by weight, 100 μL) was then added to each well. The plate was agitated on a Spectra Max Plus microplate reader (Molecular Devices Corporation) for 5 s before the absorbance at 570 nm at each well was taken. Cell viability was calculated by normalizing to MT-1 cell viability under normal growth conditions (no probe).

For a given pKα, Pdiss can be obtained as a function of pH. We note that pK values vary from reference to reference. For instance, the pK value of Glu falls in the range 3.2−4.7.31,32 Here, we use some representative values: pKα = 4.1 for Glu and pKα = 12.5 for Arg.



APPENDIX B

Supplemental Figures

Figure 9 shows the toxicity of zip5e8r, while Figure 10 demonstrates the dark toxicity of zip5e5r.

Figure 10. Dark toxicity of 5 μM zip5e5r incubated for 24 h normalized to control MT-1 cells. The error bars represent ±1 standard deviation for three independent experiments.



APPENDIX C

Materials and Methods

Beacon Synthesis. All amino acid derivatives and the Sieber amide resin were purchased from Novabiochem (San Diego, CA, USA). The black hole quencher BHQ3 carboxylic acid succinimdyl ester (BHQ3-NHS) was obtained from Biosearch Technologies (Novato CA, USA). The pyropheophorbide carboxylic acid succinimidyl ester (Pyro-NHS) was prepared according to a published procedure.33 Other chemicals were obtained from Sigma-Aldrich (Oakville, ON) and were used as received. Reverse-phase HPLC was preformed on a XBridgeTM-C8 column (2.5 μm, 4.6 × 150 mm2) using a Waters 2695 controller with a 2996 photodiode array detector and a Waters ZQTM mass detector (Waters Limited, Mississauga, ON). The beacons consist of the fluorophore pyropheophorbide-α (Pyro), black hole quencher 3, linked together by a peptide sequence of GPLGLARK (PMB), (e)5GPLGLA(r)5K (zip5e5r), (e)5GPLGLA(r)8K (zip5e8r), (r)5 (Pyro-5r), and (r)8 (Pyro-8r) synthesized as described previously.6,14 Confocal Microscopy. MT-1 cells were grown to 60% confluence days in NuncLab-TekII-CC2 8-chambered slides. RPMI 1640 medium with 10% fetal bovine serum containing 5 μM probe (diluted in 2.5 L of DMSO and 0.5 μL of Tween80) was added, and the cells were incubated for 10, 30, and 60 min at 37 °C. The cells were then washed three times with PBS, and the chamber slides were then imaged on an Olympus FluoView 1000 laser scanning confocal microscope equipped with a 488 nm Ar laser and a 633 nm He−Ne laser. Dissociation Studies. Zip beacons (10 nmol) were first dissolved in 2.5 μL of DMSO and 0.5 μL of Tween80. The solution was diluted to 1 mL of phosphate buffered saline (PBS) and then incubated with 10 μL of human proteinase K (2 μg) to completely cleave all peptide linker for 1 h. The activation based upon dissociation of the zipper arm was



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. Present Address ∥

(T.W.L.) The University of Texas MD Anderson Cancer Center, Houston, TX 77030. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the Canadian Institute of Health Research, the Ontario Institute for Cancer Research through funding provided by the Government of Ontario, the Canadian Cancer Society Research Institute, the Joey and Toby 52

dx.doi.org/10.1021/jp5086813 | J. Phys. Chem. B 2015, 119, 44−53

The Journal of Physical Chemistry B

Article

(19) Egeblad, M.; Werb, Z. New functions for the matrix metalloproteinases in cancer progression. Nat. Rev. Cancer 2002, 2, 161−174. (20) Rodriguez, J. A.; Orbe, J.; Martinez de Lizarrondo, S.; Calvayrac, O.; Rodriguez, C.; Martinez-Gonzalez, J.; Paramo, J. A. Metalloproteinases and atherothrombosis: MMP-10 mediates vascular remodeling promoted by inflammatory stimuli. Front. Biosci. 2008, 13, 2916−2921. (21) Butler, G. S.; Overall, C. M. Updated biological roles for matrix metalloproteinases and new “intracellular” substrates revealed by degradomics. Biochemistry 2009, 48, 10830−10845. (22) Giambernardi, T. A.; Grant, G. M.; Taylor, G. P.; Hay, R. J.; Maher, V. M.; Mccormick, J. J.; Klebe, R. J. Overview of matrix metalloproteinase expression in cultured human cells. Matrix Biol. 1998, 16, 483−496. (23) Roy, R.; Yang, J.; Moses, M. A. Matrix metalloproteinases as novel biomarkers and potential therapeutic targets in human cancer. J. Clin. Oncol. 2009, 27, 5287−5297. (24) Visse, R.; Nagase, H. Matrix metalloproteinases and tissue inhibitors of metalloproteinases: structure, function, and biochemistry. Circ. Res. 2003, 92, 827−839. (25) Cicek, M.; Oursler, M. J. Breast cancer bone metastasis and current small therapeutics. Cancer Metastasis Rev. 2006, 25, 635−644. (26) Köhrmann, A.; Kammerer, U.; Kapp, M.; Dietl, J.; Anacker, J. Expression of matrix metalloproteinases (MMPs) in primary human breast cancer and breast cancer cell lines: New findings and review of the literature. BMC Cancer 2009, 9, 188. (27) Scherer, R. L.; McIntyre, J. O.; Matrisian, L. M. Imaging matrix metalloproteinases in cancer. Cancer Metastasis Rev. 2008, 27, 679− 690. (28) Newby, A. C. Metalloproteinase expression in monocytes and macrophages and its relationship to atherosclerotic plaque instability. Arterioscler., Thromb., Vasc. Biol. 2008, 28, 2108−2114. (29) de Gennes, P.-G. Scaling Concepts in Polymer Physics; Cornell University Press: Ithaca, NY, 1979. (30) Nelson, P. Biological Physics; W. H. Freeman and Company: New York, 2008. (31) Alberts, B.; et al. Molecular Biology of the Cell, 5th ed.; Garland Science: New York, 2007. (32) Williamson, M. How Proteins Work; Garland Science: New York, 2011. (33) Zhang, M.; Zhang, Z.; Blessington, D.; Li, H.; Busch, T. M.; Madrak, V.; Miles, J.; Chance, B.; Glickson, J. D.; Zheng, G. Pyropheophorbide 2-deoxyglucosamide: a new photosensitizer targeting glucose transporters. Bioconjugate Chem. 2003, 14, 709.

Tanenbaum/Brazilian Ball Chair in Prostate Cancer Research, Department of Defense BCRP Predoctoral Award W81XWH10-1-0115, and the Natural Sciences and Engineering Research Council (NSERC) of Canada. This work was made possible by the facilities of the Shared Hierarchical Academic Research Computing Network (SHARCNET: www.sharcnet.ca).



REFERENCES

(1) Turk, B. Targeting proteases: successes, failures and future prospects. Nat. Rev. Drug Discovery 2006, 5, 785−799. (2) Law, B.; Tung, C. H. Proteolysis: a biological process adapted in drug delivery, therapy, and imaging. Bioconjugate Chem. 2009, 20, 1683−1695. (3) Drag, M.; Salvesen, G. S. Emerging principles in protease-based drug discovery. Nat. Rev. Drug Discovery 2010, 9, 690−701. (4) Weissleder, R.; Ntziachristos, V. Shedding light onto live molecular targets. Nat. Med. 2003, 9, 123−128. (5) Zheng, G.; Chen, J.; Stefflova, K.; Jarvi, M.; Li, H.; Wilson, B. C. Photodynamic molecular beacon as an activatable photosensitizer based on protease-controlled singlet oxygen quenching and activation. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 8989−8994. (6) Wilson, B. C.; Patterson, M. S. The physics, biophysics and technology of photodynamic therapy. Phys. Med. Biol. 2008, 53, R61− 109. (7) Zheng, X.; Morgan, J.; Pandey, S. K.; Chen, Y.; Tracy, E.; Baumann, H.; Missert, J. R.; Batt, C.; Jackson, J.; Bellnier, D. A.; Henderson, B. W.; Pandey, R. K. Conjugation of HPPH to carbohydrates changes its subcellular distribution and enhances photodynamic activity in vivo. J. Med. Chem. 2009, 52, 4306. (8) Chen, J.; Liu, T. W.; Lo, P. C.; Wilson, B. C.; Zheng, G. “Zipper” molecular beacons: a generalized strategy to optimize the performance of activatable protease probes. Bioconjugate Chem. 2009, 20, 1836− 1842. (9) Mitchell, D. J.; Kim, D. T.; Steinman, L.; Fathman, C. G.; Rothbard, J. B. Polyarginine enters cells more efficiently than other polycationic homopolymers. J. Pept. Res. 2000, 56, 318−325. (10) Jiang, T.; Olson, E. S.; Nguyen, Q. T.; Roy, M.; Jennings, P. A.; Tsien, R. Y. Tumor imaging by means of proetolytic activation of cellpenetrating peptides. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 17867− 17872. (11) Jones, S. W.; Christison, R.; Bundell, K.; Voyce, C. J.; Brockbank, S. M.; Newham, P.; Lindsay, M. A. Characterisation of cell-penetrating peptide-mediated peptide delivery. Br. J. Pharmacol. 2005, 145, 1093−1102. (12) Israelachvili, J. N. Intermolecular and Surface Forces, 3rd ed.; Academic Press: Boston, 2011. (13) Anfinsen, C. B. Principles that govern the folding of protein chains. Science 1973, 181, 223−230. (14) Privalov, P. L. Stability of proteins: small globular proteins. Adv. Protein Chem. 1979, 33, 167−241. (15) Nemethy, G.; Gibson, K. D.; Palmer, K. A.; Yoon, C. N.; Paterlini, G.; Zagari, A.; Rumsey, S.; Scheraga, H. A. Energy Parameters in Polypeptides. 10. Improved geometric parameters and nonbonded interactions for use in the ECEPP/3 algorithm, with application to proline-containing peptides. J. Phys. Chem. 1992, 96, 6472−6484. (16) Ooi, T.; Oobatake, M.; Nemethy, G.; Scheraga, H. A. Accessible surface areas as a measure of the thermodynamic parameters of hydration of peptides. Proc. Natl. Acad. Sci. U. S. A. 1987, 84, 3086− 3090. (17) Wales, D. J.; Doye, J. P. K. Global optimization by basinhopping and the lowest energy structures of Lennard-Jones clusters containing up to 110 atoms. J. Phys. Chem. A 1997, 101, 5111−5116. (18) Li, Z.; Scheraga, H. A. Monte Carlo-minimization approach to the multiple-minima problem in protein folding. Proc. Natl. Acad. Sci. U. S. A. 1987, 84, 6611−6615. 53

dx.doi.org/10.1021/jp5086813 | J. Phys. Chem. B 2015, 119, 44−53