Characterization of DNA-Conjugated Compounds Using a

Dec 12, 2014 - Fatih Nadi Gür, and Yixin Zhang*. B CUBE Center for Molecular Bioengineering, Technische Universität Dresden, 01307 Dresden, Germany...
0 downloads 0 Views 2MB Size
Technical Note pubs.acs.org/ac

Characterization of DNA-Conjugated Compounds Using a Regenerable Chip Weilin Lin,† Francesco V. Reddavide,† Veselina Uzunova,‡ Fatih Nadi Gür, and Yixin Zhang* B CUBE Center for Molecular Bioengineering, Technische Universität Dresden, 01307 Dresden, Germany S Supporting Information *

ABSTRACT: DNA-encoded chemical library (DECL) technology has emerged as a new avenue in the field of drug discovery. Combined with high-throughput sequencing, DECL selection experiments can provide not only many lead compounds but also insights into the structure−affinity relationship. However, the counts of individual DNA codes reflect, but cannot be used to precisely rank, the binding affinities of the corresponding compounds to protein targets. Herein, we describe a chip-based approach to realize an automated high-throughput assay for the kinetic characterization of the interaction between DNA-conjugated small organic compounds and protein targets. Importantly, this method can be applied to both single-pharmacophore DECLs and self-assembled dual-pharmacophore DECLs.

M

cannot be changed. In addition, tuning ligand density for ideal measurements is also a limiting preliminary step. Therefore, one chip can be used for immobilizing only one small molecule ligand at a certain density. These approaches cause high assay cost for analyzing many potential binders and prevents the high-throughput automation of the characterization process. In the second case, using immobilized protein to measure the binding of different small-molecule compounds in mobile phase always results in low signal-to-noise ratio. Moreover, the low solubility of many small-molecule compounds in aqueous solution often makes performing hit validation with this chip configuration very cumbersome. Herein, we describe a chip-based approach to realize an automated high-throughput assay for measuring the affinities between DNA-conjugated small organic compounds and protein targets. Moreover, this method can be applied to analyze compounds from both the widely used single pharmacophore DECL and self-assembled dual pharmacophore DECL.8 We took advantage of DNA-tagged small molecule compounds, which can be noncovalently captured on a biosensor chip displaying the complementary sequence (Scheme 1). Following the affinity measurements with the target proteins at different concentrations, the chip can be rapidly regenerated by destabilizing the duplex and then reused in the next cycles of experiments. The experiments were performed on a QCM biosensor and an interferometer, while the technology is also suitable for other biosensors including

any recent developments have offered promising tools to discover small-molecule hit compounds against protein targets of interest, including high-throughput screening, fragment-based drug design, virtual screening, and biologyoriented and diversity-oriented organic synthesis.1−4 As one of the most encouraging approaches that have emerged in the past years, screening of DNA-encoded chemical libraries (DECLs) has opened a new avenue in drug discovery.5−8 In combination with high-throughput sequencing, DECL technology can provide not only lead compounds but also a comprehensive structure−affinity relationship, since the readouts of selection experiments contain information not only about the potent binders, but also moderately and weakly binding compounds.9 However, although the counts of individual DNA codes reflect the relative binding affinities of the corresponding compounds, it cannot be used to calculate the absolute affinity value.7,9,10 Therefore, conventional technologies, e.g., enzyme inhibition, isothermal titration calorimetry (ITC), and fluorescence polarization assays are typically used to characterize the selected lead compounds. However, these methods only permit measurement of the inhibition constant (Ki) or the dissociation constant (Kd), but not the association rate (kon) and dissociation rate (koff), which can give more insight into drug properties.11 Compared to the above-mentioned in-solution assays, chipbased biosensor technologies (e.g., surface plasmon resonance (SPR), quartz crystal microbalance (QCM), and interferometry) have led to automated instrumentations to characterize protein−ligand interactions, including Kd, kon, and koff. This technology has two main configurations regarding whether the small molecule or protein is immobilized on the surface. In the first case, automation is restricted to alternating proteins in the mobile phase, whereas the covalently immobilized ligand © 2014 American Chemical Society

Received: October 23, 2014 Accepted: December 12, 2014 Published: December 12, 2014 864

DOI: 10.1021/ac503960z Anal. Chem. 2015, 87, 864−868

Technical Note

Analytical Chemistry Scheme 1. Principle of Affinity Measurement on a Regenerable Biosensor Chipa

a

Carboxylic acid groups on the chip were coupled to the amino groups of 5′-amino DNA (1). (Middle) A small molecule ligand (A) was conjugated to the amino group of 5′-amino DNA (1′), which is complementary to the sequence of 1. After annealing the conjugate A-1′ to 1 on the chip, the target protein was injected at different concentrations to assess the binding affinity. A-1′ can be washed off from the chip under dehybridization/ regeneration conditions. (Left cycle) Another DNA−ligand conjugate (B-1′) can then be immobilized on the regenerated chip, and its binding affinity to the same or a different protein can be evaluated. (Right cycle) This method can also be applied for a double-stranded DNA displaying two ligands (C-1′).

Scheme 2. Synthesis of DNA-Conjugated CsA Derivativesa

SPR and protein dynamics measurement by electrically actuated DNA levers.12 We used the immunosuppressive drug cyclosporin A (CsA) and biotin (and their derivatives) as DNA-conjugated ligands, while different cyclophilins (Cyps) and streptavidin were the target proteins, respectively. Binding to its receptor protein Cyp is essential for CsA’s gain-of-function inhibition of phosphatase calcineurin in activated T cells.13,14 CsA has various degrees of affinity with different Cyps, and at least six Cyps bind to CsA with low nanomolar affinity.15 CypA has been found promoting HIV16 and HCV (hepatitis C virus)17 infections and several CsA derivatives are under clinical trials against viral infections. Recently, the Cyp domain of retroviral restriction factor TRIM from rhesus macaque (RhTC) has raised great interest, because of its ability to target both HIV-1 and HIV-2.18,19 The chemical synthesis of DNA-conjugated CsA derivatives as well as their biophysical characterization can pave the way to generate large natural product-derived chemical libraries and to discover potent and subtype-specific inhibitors. As MeBmt-1 modification may change the affinity of CsA derivatives to CypA and cause different effects on various Cyps (see the Supporting Information),20,21 a MeBmt-1-modified CsA derivative with Fmoc-protected amino group and carboxylic acid (Fmoc-CsA-COOH) was synthesized22−24 and coupled to 5′-amino DNA 1′ 6 (Supporting Information). Two building blocks were coupled step-by-step via amide bond formation. Five different CsA derivatives were synthesized and characterized in this paper (see Scheme 2, as well as the Supporting Information). It is important to note that conventional methods to synthesize DECLs in solution resulted in very low yields, because of the large steric hindrance associated with CsA. A complementary sequence of 18 base pairs has been used to achieve efficient immobilization in hybridization buffer and fast dehybridization under chip regeneration conditions. To measure the binding kinetics of CypA to CsA-1′, CypA in the assay buffer was injected to the QCM chips. The binding and dissociation of CypA were recorded (see Figure 1a, as well as Figure S4d in the Supporting Information). The resulting Kd value of 6.67 ± 0.02 nM is in good agreement with the Ki value

a

The reactions were carried out on controlled pore glass (CPG), which had been used to synthesize amino-functionalized oligonucleotides. HATU and HOAt were used as activator and additive, respectively. Five DNA-conjugated CsA derivatives have been synthesized (Cs(1)-1′ to Cs(5)-1′). The structures of Fmoc-CsACOOH, Cs(n)-1′, building block A , and building block B are shown in the Supporting Information.

of 8.6 ± 1.1 nM of CsA, determined by CypA inhibition assay (see Figure S1 in the Supporting Information), as well as the Kd value of 8.8 ± 1 nM determined by ITC.25 Moreover, we measured the inhibition of CypA by single-stranded CsA-1′ and double-stranded CsA-1′/1 (see Figure S1 in the Supporting Information). Interestingly, the resulting Ki values of 8.0 ± 0.6 nM and 11.5 ± 1.8 nM have shown that DNA conjugation has only a minor effect on the protein−ligand interaction. Similiar to the conventional chip-based affinity measurement, the new method can be used to measure the binding affinity of 865

DOI: 10.1021/ac503960z Anal. Chem. 2015, 87, 864−868

Technical Note

Analytical Chemistry

Figure 1. Regenerable chip for kinetic measurement of DNA conjugated compounds. (a) Binding and dissociation curves (black) of CypA to CsA-1′ annealed with DNA 1 on the chip. The data were globally fitted (red) using a kinetic model where a 1:1 complex is formed between CypA and CsA and mass transport limitations were considered during the fitting. (b) The noncovalently captured CsA-1′ can be removed by washing the chip with 150 mM NaOH. The loading and regeneration procedure can be repeated over many cycles without losing the signal intensity. The curves show the processes at the 1st, 10th, and 20th cycles. (c) The curve of streptavidin binding to immobilized biotin-DNA conjugate and chip regeneration procedure. The treatment conditions are shown as arrows. (d) Curves of streptavidin binding according to the procedure shown in panel (c) in three repeats. (e−h) Dissociation constants of CypA (panel e), RhTC (panel f), CypB (panel g), and Cyp40 (panel h) to CsA derivatives. The experiments described in panels (a), (b), (c), and (d) were performed on QCM, while the experiments described in panels (e), (f), (g), and (h) were performed on a multichannel interferometer.

were used, and their affinities to GST-FKBP12 and trypsin were 0.57 ± 0.02 nM and 110 ± 19 μM, respectively, in good agreement with the reported values.28,29 Different from conventional methods, the DNA-encoded regenerable chip facilitates exchanging the immobilized ligand on the chip. Twenty cycles of loading CsA-1′ onto the QCM biosensor chip, followed by the regeneration procedure, did not cause any loss of signal intensity (Figure 1b). Moreover, the same chip used for Cyp/CsA measurements can be regenerated and used for biotin−DNA measurements. It is notoriously difficult to remove streptavidin from a biotinylated surface. The streptavidin bound to biotin on the chip cannot be eluted using SDS buffer, which is normally used to remove residue protein after an affinity measurement without destabilizing the DNA duplex (Figure 1c). Remarkably, the streptavidin/biotin-1′ complex can be rapidly washed off the chip, under the chip regeneration condition (150 mM NaOH), with no loss of signal intensity after three cycles (Figure 1d). The chip regeneration method is flexible and can be adapted to different drug discovery approaches. Self-assembled DECL technology resembles the fragment-based approach to exploring the chelate effect of a pair of ligands (dual-pharmacophore) to discover high-affinity binders.8,30 However, it is very difficult to characterize the bidentate interaction on a chip surface, because there is no routine chemical method to immobilize two ligands with controlled spatial arrangement, density, and stoichiometric ratio. In particular, the ligand density on the chip surface could have a dramatic effect on accessing bidentate interaction. The DNA-encoded regenerable chip allows us to control the surface density of a pair of self-assembled pharmacophores through mixing the double-stranded DNA (C-c1′, scheme) with unmodified DNA (c1′). Homotetrameric streptavidin and iminobiotin represent a classical model system for illustrating the bidentate effect of protein/ligand interaction.

a ligand toward different target proteins. The double helix structure is very stable through many rounds of binding, dissociation, and chip-cleaning (with SDS buffer) procedures. We used the above-mentioned CsA-1′/1 chip to measure its binding to RhTC, CypB, and Cyp40 (see Table 1, as well as Figure S4 in the Supporting Information). Table 1. Kinetics and Binding Affinity of CsA-1′/1 to Four Different Cyclophilins cyclophilin CypA RhTC CypB Cyp40

kon (×105 M−1 s−1) 13.1 7.03 5.97 3.45

± ± ± ±

0.1 0.29 0.24 0.30

koff (× 10−3 s−1) 8.75 58.1 1.46 36.4

± ± ± ±

0.07 1.7 0.05 2.3

Kd, kinetic (nM) 6.67 82.7 2.44 106

± ± ± ±

0.02 1 0.03 2.5

Importantly, this method is suitable not only for screening different CsA derivatives, but also on different biosensor instrumentation platforms such as an interferometer. As shown in Figures 1e and 1g, while most compounds have exhibited affinities with CypA and CypB, compound Cs(3)-1′ showed reduced affinity with both proteins. Similar to the parent compound, the affinities of these compounds with RhTC and Cyp40 are lower than those with CypA and CypB. Interestingly, Cs(1)-1′ is twice as potent in RhTC binding, when compared with the parental compound CsA-1′. Importantly, this work has a synthetic method to construct libraries based on CsA,26,27 to move DECL technology from simple combinatorial chemical libraries to more sophisticated natural product-like libraries. To demonstrate that this method can also be used to study other protein−ligand interactions, we have measured the bindings between FKBP12 binding ligand SLF and FKBP12, and protease inhibitor benzamidine and protease trypsin. Their conjugates, SLF-1′ and benzamidine-1′, 866

DOI: 10.1021/ac503960z Anal. Chem. 2015, 87, 864−868

Technical Note

Analytical Chemistry

Figure 2. Regenerable chip for kinetic measurement of DNA-conjugated dual-pharmacophores. (a) Different ligand patterns on the chip surface created by mixing nonmodified DNA duplexes with iminobiotin-modified DNA duplex at different ratios. The modified DNA duplex has either modifications on both chains (schematics 1 and 2) or a modification on only one chain (schematics 3 and 4). Tetrameric streptavidin is shown in black. (b−e) Experimental data showing the binding and dissociation curves for ligand loaded with different patterns. High-density bivalent ligand surface using DNA duplex with two iminobiotins on both chains (panel (b)), low-density bivalent ligand surface using DNA duplex with two iminobiotins on both chains (panel (c)), high-density monovalent ligand surface using DNA duplex with only one iminobiotin (panel (d)), and lowdensity monovalent ligand surface using DNA duplex with only one iminobiotin (panel (e)). Schematics 1, 2, 3, and 4 shown at the bottom of panel (a) correspond to panels (b), (c), (d), and (e), respectively.

chip surface can distinguish the bidentate binding of a pair of self-assembled ligands from the effect of high surface density. In summary, we describe a fast regenerable chip-based biosensor method for screening protein−drug interactions. The kinetic characterization is complementary to in-solution assays such as enzyme inhibition, ITC, and fluorescence polarization, providing additional important binding information such as the kon and koff values. This design can also be used to immobilize other DNA-conjugated compounds such as proteins32 to access their affinity with other proteins or antibodies. We tested CsA derivatives on two different types of biosensors. The chip regeneration procedure allows us to perform many experiments on a single chip and leads to almost completely automated high-throughput compound validation campaigns with minimal researcher intervention, through alternating both the proteins in the mobile phase as well as the ligands in the static phase. Furthermore, this method is very flexible, and thus can be adapted to other drug discovery methods such as fragmentbased drug discovery for hit validation. To our knowledge, it is the only high throughput on-chip biophysical analysis to test combinations of different motifs without synthesizing the individual bidentate compounds.

A DNA duplex with two iminobiotins has been studied in solution to confirm the suitability of self-assembled DECL technology as a fragment-based approach in drug discovery.31 To demonstrate the compatibility of the DNA-conjugate affinity measurement method with fragment-based drug discovery methods such as self-assembled DECL, we investigated the binding of streptavidin with DNA-iminobiotin conjugates (see Figure 2a). After injecting and annealing the double-stranded DNA I2c1′ onto the chip (Figure 2b), the binding and dissociation measurements of streptavidin showed a very strong binding affinity (