Fluorogenic Assay for Inhibitors of HIV-1 Protease with Sub-picomolar Affinity

A fluorogenic substrate for HIV-1 protease was designed and used as the basis for a hypersensitive assay. The substrate exhibits a kcat of 7.4 s−1, KM of 15 μM, and an increase in fluorescence intensity of 104-fold upon cleavage, thus providing sensitivity that is unmatched in a continuous assay of HIV-1 protease. These properties enabled the enzyme concentration in an activity assay to be reduced to 25 pM, which is close to the Kd value of the protease dimer. By fitting inhibition data to Morrison’s equation, Ki values of amprenavir, darunavir, and tipranavir were determined to be 135, 10, and 82 pM, respectively. This assay, which is capable of measuring Ki values as low as 0.25 pM, is well-suited for characterizing the next generation of HIV-1 protease inhibitors.

example, ITC has a theoretical limit in the low nanomolar range for the direct measure of an equilibrium disassociation constant (K d ).
The key to assessing the next generation of HIV-1 protease inhibitors is reducing the enzyme requirement in activity-based inhibition assays. Towards that end, we report here on the design and characterization of a novel fluorogenic substrate for HIV-1 protease. Its attributes-high k cat and k cat /K M values and high signal-to-noise ratio-are unprecedented, and enable the rapid, facile determination of sub-picomolar values of K i .

Results
Substrate Design. The first major design criterion for an improved fluorogenic substrate was identifying a peptide that was bound by HIV protease with high affinity and cleaved rapidly in a catalytic manner. To meet this criterion, we employed a peptide substrate that had been selected by phage display. The sequence GSGIFLETSL was reported to have k cat /K M = 1.3 μ M −1 s −1 for cleavage between the phenylalanine and leucine residues 18 . These values were determined by a discontinuous HPLC method. Under the same conditions, an endogenous cleavage site in the HIV polyprotein has k cat /K M = 0.022 μ M −1 s −1 18 .
The second major design criterion was employing a sensitive method to detect substrate turnover in a continuous manner. We focused on the loss of Förster resonance energy transfer (FRET), which underlies many useful assays 19 , and considered three donor/acceptor moieties. Appending the fluorophore p-aminonitrobenzoic acid (Abz) and installing a nitro group in the para position of phenylalanine as a quenching chromophore is the basis for a popular HIV protease substrate 11 . This pair, however, lacks sensitivity, as Abz is a weak fluorophore and its use provides a relatively low signal-to-noise ratio. FRET between fluorescein and rhodamine is the basis for some of the most sensitive known assays for enzymatic activity 20 , but protonation of a fluorescein moiety at pH 5, which is optimal for catalysis by HIV-1 protease 12 , compromises the utility of this pair for our purpose, and others 21,22 . We chose 5-((2-aminoethyl)amino)naphthalene-1-sulfonic acid (EDANS) and 4-(4-dimethylaminophenylazo)benzoic acid (DABCYL) as a FRET pair 12,20 . We installed these moieties into GSIFLETSL by replacing the glycine residue at the N-terminus with a glutamate-EDANS conjugate and by replacing the leucine residue at the C-terminus with a lysine-sDABCYL conjugate 12 . Finally, we added an arginine residue to each terminus to enhance aqueous solubility at pH 5, thereby generating substrate 1 (Fig. 1). Assay Design. Our assay was designed to minimize the enzyme concentration while maintaining a high signal-to-noise ratio. An inherent complexity is that HIV-1 protease is an obligate dimer. Significant attention has been paid to the K d value of the HIV-1 protease dimer, and conflicting ideas abound regarding an appropriate enzyme concentration for activity assays 23 . Freire and coworkers employed a rigorous thermodynamic approach to determine a dimer K d value of 23 pM 24 . We used this value as a lower limit for the enzyme concentration in our assays. Additionally, initial velocities require measurements from < 10% substrate turnover, and a convenient upper limit for the enzyme concentration in our assays was determined to be 6.5 nM. Upon cleavage by HIV-1 protease, the fluorescence intensity (I) of substrate 1 increases by I f /I o = 104.

Michaelis-Menten Kinetics.
Michaelis-Menten kinetics were used to evaluate the performance of substrate 1 as a substrate for HIV-1 protease. The initial velocity was directly proportional to enzyme concentration at a fixed substrate concentration ( Fig. 2A), and increased with substrate concentration at a fixed enzyme concentration (Fig. 2B). The latter data fitted well to the Michaelis-Menten equation (Fig. 2C). The observed V max value of 1.58 nM·s −1 for substrate 1 at an enzyme concentration of 214 pM afforded a k cat value of (7.4 ± 0.2) s −1 ; the K M value was (14.7 ± 1.0) μ M ( Table 2).

Determination of K i Values with Morrison's Equation.
Substrate 1 was used as the basis for assays of the inhibition of HIV-1 protease by amprenavir, darunavir, and tipranavir (Fig. 3) exhibited time-dependent inhibition; amprenavir did not. Pre-equilibrium data were omitted from the initial-velocity data fitted by linear regression.

Discussion
An assay of high sensitivity is critical for assessing the efficacy of high-affinity inhibitors of enzymatic activity. The HIV-1 protease substrate described herein provides initial velocity data of unmatched quality and sensitivity. Substrate 1 has a 1.5-fold higher k cat value, 7-fold lower K M value, and higher signal-to-noise ratio than does the parent substrate developed by Matayoshi and coworkers ( Table 2) 12 .
The Abz-based substrate developed by Toth and Marshall has a similar K M value, though substrate 1 provides a 17-fold greater signal-to-noise ratio 11 . These improvements in kinetic and spectroscopic parameters have enabled us to reduce the concentration of enzyme in standard assays to values close to the K d value of dimeric HIV-1 protease.
To quantify the utility of a substrate, we define the sensitivity (S) of an assay as the increase in fluoresence intensity brought about by the action of an enzyme on a low concentration of substrate. We express sensitivity (S) as the product of the kinetic parameter k cat /K M and the spectroscopic parameter I f /I o : S = (k cat /K M )(I f /I o ). By this measure, the sensitivity of an assay that uses substrate 1 is ≥ 10-fold greater than any fluorescence-based assay for HIV-1 protease activity ( Table 2).
The values of K i for amprenavir, darunavir, and tipranavir derived from initial velocities for the cleavage of substrate 1 are consistent with literature values (Table 1). Notably, the K i value for darunavir determined herein is comparable to literature values and much closer to values reported by other activity-based methods than is the value determined by an indirect, competitive displacement method 10 , which is two orders of magnitude lower. The high signal-to-noise, low variation in initial velocity measurements and low standard error for fits of inhibition data by Morrison's equation makes substrate 1 a useful probe for assaying high-affinity inhibitors.
According to a computational assessment, Morrison's equation can be used to determine values of K i that are up to 100-fold lower than the concentration of enzyme in an assay 15 . Because enzyme concentrations as low as 25 pM provided valuable data herein, we believe that our assay can be used to determine K i values that are ≥ 250 fM. We anticipate the use of substrate 1 in the development of the next generation of HIV-1 protease inhibitors.     substitutions 25 . Linear pET32b was prepared by PCR using primers that were the reverse complements of the DNA encoding HIV-1 protease. Gene and plasmid fragments were combined with Gibson assembly 26 .
Protein Purification. BL-21 codon-plus RIL from Agilent Technologies (Santa Clara, CA) was transformed freshly with the pET32b-HIV protease. A single colony was used to inoculate 1 L of Luria-Bertani medium containing ampicillin (200 μ M) in a Fernbach flask shaken at 37 °C. Expression was induced by the addition of IPTG (to 2 mM) upon reaching saturation (OD 600 nm 2.8-3.4), and the culture was grown for an additional hour. HIV-1 protease was purified and folded as described previously 27 . Cells were pelleted, resuspended in 20 mM Tris-HCl buffer, pH 7.4, containing EDTA (1 mM) and lysed at 18 kPSI using a cell disruptor from Constant Systems (Kennesaw, GA). Inclusion bodies were isolated by centrifugation at 10,000 g for 10 min. The pelleted inclusion bodies were washed with resuspension buffer containing urea (1.0 M) and Triton X-100 (1% v/v), and again with resuspension buffer. Inclusion bodies were isolated by centrifugation and lyophilized. Inclusion bodies were dissolved by sonication in aqueous acetic acid (50% v/v) at a concentration of 5 mg/mL. The solution was clarified by centrifugation, and soluble protein was applied to a Superdex 75 gel-filtration column from GE Healthcare Bio-Sciences (Pittsburgh, PA) that had been pre-equilibrated with aqueous acetic acid (50% v/v). Unfolded HIV-1 protease that eluted as major peak near one column-volume was pooled and lyophilized. HIV-1 protease was folded at a concentration of 0.1 mg/  Table 1.
Scientific RepoRts | 5:11286 | DOi: 10.1038/srep11286 mL in 100 mM sodium acetate buffer, pH 5.5, containing ethylene glycol (5% v/v) and glycerol (10% v/v). The solution of folded HIV-1 protease was clarified by centrifugation and concentrated with an Amicon stirred-cell concentrator equipped with a 10 K MWCO membrane from EMD Millipore (Billerica, MA). The concentrated protease was applied again to a Superdex 75 gel-filtration column that had been pre-equilibrated with the folding buffer. A new major peak containing dimeric HIV-1 protease was pooled and concentrated. The folding buffer was exchanged for 1 mM sodium acetate buffer, pH 5.0, containing NaCl (2 mM) using a PD-10 desalting column. A solution (~1.5 mg/mL) of purified HIV-1 protease was flash-frozen in liquid nitrogen and stored at − 80 °C until use.
Enzymatic Activity Assays. Substrate 1 was dissolved at a concentration of 1.0 mM in DMF containing TFA (0.1% v/v). Fluorescence of the EDANS moiety was measured on a M1000 Pro plate reader from Tecan (Maennedorf, Switzerland) by excitation at 340 nm and observation of emission at 490 nm. A fluorophore calibration was performed to enable quantitation of assay data. The product exhibits a fluorescence of 70 RFU/nM at a gain setting of 216, and all assays were performed at this gain setting unless indicated otherwise. Assays were performed in a Corning black, flat bottom, non-binding surface, 96-well plate. Assays were conducted at room temperature in 200 μ L of 50 mM sodium acetate buffer, pH 5.0, containing NaCl (0.10 M), DMF (2% v/v), substrate 1 (1-40 μ M), and HIV-1 protease (25 pM-6.5 nM). Assays with 30 and 40 μ M of substrate 1 required 3% and 4% v/v DMF, respectively. Inhibition assays were conducted with picomolar-nanomolar inhibitor (depending on the enzyme concentration and K i value) and 10 μ M substrate 1. Inhibition assays were monitored for until ≤ 7% of the substrate was converted to product. Initial velocities were measured in quadruplicate.
Solution concentrations of HIV-1 protease (10.7 kDa) was determined by measuring the absorbance at 280 nm and estimating the extinction coefficient as 12,500 M −1 cm −1 with software from ExPASy 28 . The fraction of active enzyme was determined by active-site titration using darunavir as the titrant and found to be 76% with respect to the value based on the A 280 nm . Fluorescence was monitored over the linear range of the detector, which corresponds to 700 nM of product formation at a gain setting of 216. Data Analysis. The velocity (v) of all enzyme-catalyzed reactions was obtained by linear fit of initial-velocity data using Prism 6 software from Graphpad (La Jolla, CA). Pre-equilibrium values from the beginning of data sets were removed to provide fluorescence measurements that were linear as a function of time (Fig. 3).
Values of v in the absence of an inhibitor were fitted to the Michaelis-Menten equation (eq. 1) by non-linear regression using Prism 6 software.  In eq 2, v o refers to the reaction in the absence of inhibitor. Enzymatic activity measured in the absence of an inhibitor was used to determine the enzyme concentration for data obtained in the presence of an inhibitor. These enzyme concentrations, which agreed (± 10%) with values estimated by active-site titration, were used as constraints for the non-linear regression analysis.