Sage Journals: Discover world-class research

Abstract

Antiretroviral therapy is able to effectively control but not eradicate HIV infection, which can persist, leading to the need for lifelong therapy. The existence of latently HIV-infected cells is a major barrier to the eradication of chronic HIV infection. Histone deacetylase inhibitors (HDACis), small molecules licensed for oncology indications, have shown the ability to produce HIV transcripts in vitro and in vivo. The pharmacologic parameters that drive optimal HIV latency reversal in vivo are unknown and could be influenced by such factors as the HDACi binding kinetics, concentration of compound, and duration of exposure. This study evaluates how these parameters affect HIV latency reversal for a series of novel HDACis that differ in their enzymatic on and off rates. Varying cellular exposure, using automated washout methods of HDACi in a Jurkat cell model of HIV latency, led to the investigation of the relationship between pharmacokinetic (PK) properties, target engagement (TE), and pharmacodynamic (PD) responses. Using an automated robotic platform enabled miniaturization of a suspension cell-based washout assay that required multiple manipulations over the 48 h duration of the assay. Quantification of histone acetylation (TE) revealed that HDACis showed early peaks and differences in the durability of response between different investigated HDACis. By expanding the sample times, the shift between TE and PD, as measured by green fluorescent protein, could be fully characterized. The comprehensive data set generated by automating the assays described here was used to establish a PK/PD model for HDACi-induced HIV latency reversal.

Keywords

HDAC HIV latency Automation PK/PD Modeling

Introduction

Since the start of the HIV pandemic, intensive research has resulted in the development of a number of antiretroviral drugs that target different stages of the HIV life cycle. Current therapy uses combined classes of antiretroviral drugs (combined antiviral therapy [cART]) to suppress plasma viremia to below the level of quantification by standard clinical viral load assays. However, in the overwhelming majority of patients, viral recrudescence occurs within 4 to 8 wk of interruption of cART,¹ resulting in a requirement for lifelong and uninterrupted cART to maintain viral suppression. Although cART is effective at preventing viral replication in the periphery, recent reports have indicated that it might be ineffective at preventing replication in some tissues.^2,3

HIV infection can also persist in resting memory T cells that harbor fully replication competent proviral DNA but are transcriptionally silent. As a consequence, these latently infected T cells do not express viral antigens and remain invisible to immune surveillance.⁴ Upon antigen recognition, latently infected cells are reactivated and proliferate, resurrecting HIV gene transcription and resulting in the production of nascent HIV virions that contribute to viral recrudescence. The half-life of these latently infected resting memory CD4+ T cells has been calculated to be 44 mo,^5,6 leading to estimates that it would take more than 70 years of cART therapy to fully eradicate the viral reservoir. As cART therapy does not affect integrated proviral DNA, the persistence of this reservoir represents a major hurdle in the eradication of the virus in chronically infected patients.

Strategies to accelerate the elimination of the latent HIV reservoir have centered on a “shock and kill” approach (also termed “kick and kill”). In this strategy, HIV-infected resting memory CD4+ T cells are first “shocked” with agents that induce HIV gene transcription and production of viral antigens.^7–9 An important consideration for shock agents is that they do not induce T-cell activation and cytokine production, as this would lead to detrimental and potentially life-threatening consequences. Once decorated with viral antigens, HIV-infected memory T cells can be distinguished from uninfected cells, rendering them susceptible to strategies that specifically clear (or “kill”) HIV-infected cells.

Jurkat cells containing integrated, but transcriptionally silent, HIV proviruses have been intensively studied as a model system for HIV latency.^10–20 Although these model systems do not recapitulate all aspects of HIV latency, HIV gene transcription is activated by most shock agents that have exhibited activity in resting memory T cells from HIV-infected patients.²¹ Studies outlined here specifically used the Jurkat 2D10 cell line.²²

HIV transcriptional elongation is influenced by the chromatin structure in latently infected T cells. HDACs (HDAC1, 2, and 3) have been isolated from the HIV promoter and are thought to play a key role in the restriction of HIV gene transcription in quiescent cells.^{14,16,23–25} The addition of HDAC inhibitors to these cells resulted in measurable changes in cellular histone acetylation (a proximal biomarker of target engagement), disruption of the nuc1 structure, and concomitant activation of HIV gene transcription.^26–29

Histone deacetylase inhibitors (HDACi) can induce latent HIV transcription based on key proof-of-concept clinical studies of the licensed oncology HDACis vorinostat,^30,31 panobinostat,³² and romidepsin,³³ which have demonstrated statistically significant increases in HIV mRNA in resting CD4+ T cells,^30–33 HIV gag p24 protein expression,³⁴ and plasma viremia^32,33 after drug administration. As these drugs were developed for cancer indications, the key pharmacokinetic (PK)/pharmacodynamic (PD) relationships that drive HIV latency reversal remain unknown.

As the PK profiles and biophysical properties of these molecules differ, we sought to determine the PK parameters that govern HIV latency reversal, as these data would be a key consideration in the design of novel HDACis for HIV latency reversal. To this end, a series of washout experiments were designed to allow variation of HDACi concentration and exposure time in a model system of HIV latency. Initially, a series of plate-based manual washout experiments were performed to determine the impact of concentration and time on HDACi-induced HIV latency reversal. However, such experiments are laborious and low throughput and did not allow convenient, simultaneous capture of multiple parameters. To generate a data set suitable to establish a PK/PD model for HDACi-induced HIV latency reversal, we sought to automate the washout experiment. Although automation is commonplace in PK/PD studies of antibiotics,^35,36 no system exists for monitoring HIV latency reversal in a mammalian suspension cell line. The automation of this assay allowed collection of both proximal target engagement (histone acetylation) and HIV latency reversal (enhanced green fluorescent protein [eGFP]expression) over a full-dose response of several novel HDACis at different exposures. This allowed the generation of a PK/PD model to inform on the design of a series of in vivo experiments to establish the tolerability of HDACis at these exposures.

Methods

Cell Culture and Compounds

Jurkat 2D10 cells were used as a model system for HIV latency with eGFP integrated into the cell under control of a transcriptionally silent HIV promoter, which can be induced.²² Cells were cultured in RPMI 1640 (11835-030, Gibco, Waltham, MA) supplemented with 5% heat-inactivated fetal bovine serum (10082-139, Gibco) and 100 U/mL penicillin 100 µg/mL streptomycin (15140-148, Gibco). Vorinostat, panobinostat, and novel aryl ketone³⁷– and ethyl ketone³⁸–based HDAC inhibitors in 10 mM DMSO stocks were synthesized in house. Romidepsin powder was commercially sourced (Selleck, Houston, TX) and reconstituted to a concentration of 10 mM in DMSO.

Manual Washout

Jurkat cells (5 × 10⁶ cells in 5 mL of media) were seeded in a T25 flask (Cat. No. 156340, Thermo Scientific USA, Waltham, MA). For each compound, flasks were prepared representing a single point in a 10-point dose titration with a DMSO control flask. Cells were incubated at 37 °C with 5% CO₂ with 80% relative humidity. Samples (1 mL) were taken at washout time points, pelleted by centrifugation at 300 × g for 5 min, media aspirated, and cells resuspended in 1 mL of fresh media without drug and plated in a 24-well plate (Falcon). Twenty-four hours after treatment initiation, regardless of exposure time, cells were stained with violet fixable viability dye (Invitrogen, Carlsbad, CA). Viability and percentage positive eGFP cells were determined using an LSRII flow cytometer (BD Biosciences, San Jose, CA) with a 405 nm laser and 440/40 nm band-pass filter and 488 nm laser and 505 nm long-pass dichroic mirror with a 525/50 nm bandpass filter, and data were batch analyzed in FACSDiva Software (BD Biosciences). Gates were placed on viable cells with a secondary gate then placed around eGFP-expressing viable cells.

The effective concentration that elicited 50% of maximal eGFP-positive cells (EC₅₀ values) was calculated based on the concentration or area under the curve (AUC) using GraphPad Prism 8.1.1 software (La Jolla, CA) with a variable slope (four parameters) least-squares fit for analysis.

Automated Washout

The required distribution of time points for appropriate interrogation of PK/PD characteristics was the driving force for workflow automation. The process described in this section was programmed, operationally tested, and biologically validated prior to experimental implementation. All instruments were controlled using the in-house–developed automation software Telios,³⁹ which would be analogous to Thermo’s Momentum or HRB’s Cellario.

Automation Robotics/Hardware/Driver

The instrumentation used on the physical robotic system included the following: two robotic arms TX90L (Staubli, Pfaffikon, Switzerland), an ATS acoustic dispenser (EDC Biosystems, Fremont, CA), Multidrop Combi (Thermo Scientific USA), Biomek FX (Beckman Coulter Inc., Brea, CA), PHERAstar FS (BMG Labtech, Ortenberg, Germany), Acumen Ex3 (SPT Labtech, Hertfordshire, UK), two GNF Washer Dispensers (GNF, La Jolla, CA), two Vspin plate centrifuges (Agilent, Santa Clara, CA), a Blue Washer (BlueCatBio, Concord, MA), 4 °C automated incubator (Cytomat 2C, Thermo Scientific USA), and cell culture automated incubator (Thermo Scientific USA). Each instrument was controlled by a Telios driver, which used an individual equipment automation property interface. All scheduling was programmed through Telios director software, which allowed for real-time compound and plate tracking through raw data acquisition.

Automated Compound Transfer, Cell Preparation, and eGFP Detection

The automated assay required the preproduction of assay compound plates which contained a 16-pt 1.5-fold serial dilution of each HDACi to be tested. Concentration ranges were selected to encompass previously determined EC₅₀ values, and dilutions were prepared in DMSO from 10 mM frozen stocks using a bravo liquid handler (Agilent). After dilution preparation, 7 µL of compound was immediately transferred on the bravo deck to a barcoded source plate (788876-2B, Greiner, Kremsmünster, Austria), and titration information was associated with the plate barcode. All remaining steps were performed on the fully integrated robotic platform described above.

An ATS acoustic dispenser was used to transfer 100 nL of compound from the source plate in triplicate to the assay plate, a 384-well poly-D-lysine–coated assay plate (789071-T-PDL, Greiner). A compound transfer file, based on the plate barcode, that contained well dispense information was automatically associated with each plate. Twelve replicate assay plates were made, with each plate representing a single time point in the assay as indicated by the arrows in Figure 1 . Jurkat 2D10 cells (100,000 cells/mL) were prepared and maintained under constant agitation with magnetic stirring to prevent cell settling during processing. Using a multidrop Combi, 50 µL of cell suspension (5000 cells per well) was added to each well of the assay plate. Treated cells were incubated at 37 °C, 5% CO₂, with 80% humidity in an automated cell culture incubator for time course–appropriate durations. Cells were washed by pelleting the cells by centrifugation for 1 min at 300 × g (VSpin, Agilent), followed by removal of the cell supernatant and addition of 50 µL of fresh media (GNF washer dispenser). This cycle was repeated two additional times. Cell loss was validated by quantification of cellular ATP levels during the dedicated optimization steps (not shown). Efficient compound removal was validated by quantification of a fluorescein dye added as a surrogate for test compounds following each wash step with a 250-fold dilution achieved after final wash (data not shown). Programming maintained an “uninterrupted flow” (i.e., no other plates can access the instrumentation used) when a plate entered this iterative series of centrifugation and washing described. After the final wash, cells were resuspended in a final volume of 40 µL media. As indicated in Figure 1 , plates were either returned to the cell culture incubator until a post–washout time point or incubated at room temperature for 10 min to allow cells to settle prior to imaging. eGFP expression was measured using an Acumen EX3 instrument with a 488 nm laser and 530 nm emission filter. Immediately after eGFP measurements, plates were processed for histone acetylation enzyme-linked immunosorbent assay (ELISA) as described below.

Figure 1.

Automated washout design. For each exposure time (shaded area), 12 replicate plates were made. Each plate corresponds to a time point (hours) along the assay duration (arrows). At each time point (except 0), histone acetylation and HIV latency reversal (green fluorescent protein) data were collected for each plate. The shaded area indicates the duration of drug exposure. The light area indicates the washout time points. On each plate, six compounds were tested in 1.5-fold, 16-point titrations in triplicate.

Raw data files were generated in Telios with barcode and time stamp information to track data source and assay time. Data generated for treated cells were normalized to DMSO controls, and analysis was performed to compare data with the manual washout in a similar manner. Time-course graphs for individual concentrations were generated in GraphPad Prism.

Automated Histone Acetylation ELISA

Histone acetylation was measured by sandwich ELISA histone H4 capture and H4K5/8/12/16 detection.⁴⁰ The ELISA workflow was decoupled into two steps: coating and assay. The coating step was performed once per week to support several instances of the actual assay portion, because precoated, blocked plates were found to work successfully for many weeks in storage. ELISA plates (Cat. No. 3576, Corning, Corning, NY) were coated with 45 µL of 1:2000 anti-H4 antibody (D214-3, MBL, Tokyo, Japan) in a coating buffer (C3041, Sigma, St. Louis, MO) and incubated at 4 °C overnight. Plates were washed 1X on the BlueWasher with phosphate-buffered saline (PBS) with 0.05% Tween-20 (Sigma). Using a Combi, 50 µL of blocking buffer (3% bovine serum albumin; A9205, Sigma) in PBS was added to each well of the ELISA plate. Blocked plates were sealed on a Plateloc plate sealer (Agilent) and stored at 4 °C in blocking buffer until ELISA was performed (no longer than 1 wk). Cells in the assay plate were lysed with 40 µL of 1% Triton X-100 (Sigma) in PBS. Blocking buffer was removed from the ELISA plate using the Blue Washer. Using a Biomek FX, 25 µL of lysed cell sample was transferred from the assay plate to the uniquely barcoded ELISA plate. A transfer file tracking the cell lysate transfer was automatically generated to track the source of the samples for ELISA evaluation. To each ELISA well, 12.5 µL of 1:500 alkaline phosphatase conjugated anti-acetyl histone H4 (Cat. No. 05-1355, Millipore, Burlington, MA) in blocking buffer was dispensed using the GNF dispenser. The ELISA plate was incubated at 4 °C for 16 h. ELISA plates were washed twice on the Blue Washer using 80 µL PBS with 0.05% Tween-20, ending with an evacuated plate. To measure luminescence, 20 µL Tropix CDP-Star Sapphire II substrate (Cat. No. T2214, Applied Biosystems, Foster City, CA) was added to the washed ELISA plate. Following a 1 h incubation at room temperature, luminescence was measured on a PHERAstar FS.

Raw data files, named with the ELISA plate barcode, were automatically generated and stored by Telios software. ELISA results were normalized to DMSO-only treated plate controls. Time-course graphs for individual concentrations were generated in GraphPad Prism.

Information Tracking and Compiling

All plates used for this assay were uniquely barcoded, and barcodes were used by Telios software to automatically store and associate transfer and data files with plate barcodes. The informational framework streamlined the postprocessing integration by preparing and consolidating data into formats easily integrated and interpreted by modeling software scripts described below. The ATS Telios driver generated tracking files with every plate-to-plate transfer, mapping the source transfer wells to the destination wells. The Biomek FX Telios driver generated a plate-to-plate transfer file for cell lysates to ELISA plates for automated tracking of sample transfer. Ultimately, these tracking files along with data files generated from the Acumen (eGFP) and the PHERAstar (ELISA) were compiled and organized so that a set analysis program could intake this formulated utility output. Following the robotic run completion, the postprocessing script was run for parsing and compiling all raw data with associated assay and compound information as tracked by barcodes stored within the robotic network. Ancillary data such as compound exposure time for each plate needed to be manually added to the script. A csv file formatted for the further processing for relationship analysis was the final product of the robotic run and processing. Data formatting and compilation operation used compound repository information to associate compound information at the plate-well level to link compound information to data generated from both assays, ensuring reliable data and well-information tracking.

PK/PD Relationship Analysis

PK/PD analysis was conducted in both R 3.5.3⁴¹ and GraphPad Prism 8.1.1 using the automated washout experimental data. The AUC was derived using the treatment duration and concentration variables, in which AUC was calculated by multiplying the treatment duration by the compound concentration in each well. The EC₅₀ of each compound from 24 h exposure was used to derive the potency-normalized AUC for each time point. The area under the effect curve of HA (AUECHA) was derived using the trapezoidal rule, measuring the DMSO-normalized fold change in histone acetylation as a function of time for each dose. Each point shown represents the area under the of DMSO-normalized HA effect over time for each dose level per exposure. Likewise, the area under the effect curve of eGFP (AUECGFP) was derived using eGFP time-concentration effect profiles. Exploratory analysis of (1) potency-normalized AUC and AUECHA and (2) potency-normalized AUC and AUECGFP was conducted for all compounds.

Results

In Vitro Activity of HDACis on Purified Enzymes

There are 11 HDAC isoforms expressed in mammalian cells, which are categorized into four classes: class I (HDAC 1, 2, 3, and 8), class IIa (HDAC 4, 5, 7, and 9), class IIb (HDAC 6 and 10), and class IV (HDAC 11). In vitro characterization of novel aryl ketone³⁷ and ethyl ketone³⁸ HDACi compounds and 3 licensed HDACis (vorinostat, panobinostat, and romidepsin) against HDAC isoforms 1, 2, 3, 6, and 8 is shown in Supplementary Table S1 . The licensed HDACis displayed a range of potencies in the enzymatic assay in the nanomolar range (ranging from 0.5 nM to 442 nM). Importantly, vorinostat, panobinostat, and romidepsin demonstrated potent activity against HDAC isoforms 1 and 3, which are required to reverse HIV latency in vitro,^42,43 although inhibition of HDAC 3 alone is not sufficient for latency-reversing activity (LRA).³⁷ As described previously,^44,45 measurements of enzyme residency and potency were used to estimate the enzymatic half-life of each HDACi on recombinant HDAC1. Vorinostat dissociated rapidly from the enzyme (t½ = 0.075–0.5 h), whereas the t½ for panobinostat (0.5–3 h) and romidepsin (>28 h) dissociated more slowly. Collectively, these data demonstrate that the licensed HDACis broadly inhibited HDAC class I and II enzymes but with different potencies and displayed differing enzyme residency kinetics. The novel HDACis, which were designed specifically for analysis of HIV latency reversal, were more selective for class I HDAC isoforms and longer calculated enzyme residency times (T_1/2). Aryl ketones displayed longer enzyme T_1/2 as compared with compounds from the ethyl ketone series ( Suppl. Table S1 ).

Washout Experiments with Vorinostat, Panobinostat, and Romidepsin

The model system employed in these studies used the 2D10 Jurkat cell line, which harbors a transcriptionally silent, minimal HIV1 provirus that includes an eGFP reporter gene.²² Treatment of this cell line with LRAs activates HIV transcription that can be quantified by measurement of eGFP expression.

Initially, a series of manual washout experiments were performed, in which cells were incubated with a range of HDACi concentrations for either 4 or 24 h. After 4 or 24 h, cells were removed, washed, and then incubated in HDACi-free media. After an additional 24 h, HIV latency reversal was measured via the quantitation of eGFP-positive cells by flow cytometry. Cytotoxicity in 2D10 cells was noted with 24 h exposure at the highest concentrations tested. After a 24 h exposure of HDACi, the E_max (percentage of eGFP-positive cells at the highest drug concentration with no cytotoxicity) of vorinostat, panobinostat, and romidepsin were similar, with 90%, 97%, and 86% eGFP-positive cells, respectively ( Fig. 2A–C ), suggesting that chromatin restriction is a major barrier to HIV transcription in this cell line, which can be overcome by HDACis. However, after only a 4 h exposure, there was a marked reduction in the percentage of eGFP-positive cells for vorinostat (33%) and panobinostat (70%) but not romidepsin (86%) as compared with 24 h exposures with the same compounds ( Fig. 2A–C ). It is possible that the higher LRA activity observed for panobinostat and romidepsin compared with vorinostat following shorter (4 h) incubations is a function of their slower dissociation from HDAC enzymes following washout. This suggests that the dissociation half-life of HDACis influences the ability of each to elicit maximal HIV latency reversal after short exposures.

Figure 2.

Manual washout. Histone deacetylase inhibitors vorinostat (fast binding kinetics), panobinostat (moderate binding kinetics), and romidepsin (slow off-rate kinetics) were tested with different exposure durations (4 h or 24 h). Enhanced green fluorescent protein (eGFP) responses measured by flow cytometry at 24 h after initial compound treatment. Dose response (A, B, C) and exposure (area under the curve [AUC], exposure duration multiplied by concentration) response (D, E, F) for 4 h and 24 h exposures are shown.

These experiments also allowed the calculation of EC₅₀ values (concentration that elicits 50% of the maximal response) for each compound following either a 4 h or 24 h incubation ( Fig. 2A–C ). As perhaps could be expected, the EC₅₀ values calculated for the HDACis revealed that the rank order of potency in the HIV latency model system was consistent with the potency of the compounds against class I HDACs in the enzymatic assays (i.e., romidepsin > panobinostat > vorinostat; Suppl. Table S1 ). In addition, EC₅₀ values differed depending on the length of time cells were incubated with the HDACis ( Fig. 2A–C ). After a 4 h exposure (EC₅₀ 4hrs), the calculated EC₅₀ values were higher than those obtained following a 24 hour exposure (EC₅₀ 24hrs); vorinostat (EC₅₀ 4hr 2461 nM vs. EC₅₀ 24hr 610 nM), panobinostat (EC₅₀ 4hr 92 nM vs. EC₅₀ 24hr 10 nM), and romidepsin (EC₅₀ 4hr 15.6 nM vs. EC₅₀ 24hr 3.1 nM). However, the exposure-adjusted EC₅₀ (EAUC₅₀), which is the concentration that elicits 50% of the maximal response with each exposure duration, was similar at 4 hours (EAUC₅₀ 4hrs) and 24 h (EAUC₅₀ 24hrs), indicating that HDACi-induced HIV latency reversal is driven by AUC, as summarized in Table 1 .

Table 1.

Summary of Assay Data from Manual and Automated Washout Experiments.^a

Name	Chemical Group	Washout Method	EC₅₀ (nM)		AUEC₅₀ (nM × h)		E_max (nM)
Name	Chemical Group	Washout Method	4 h	24 h	4 h	24 h	24 h
Vorinostat	Hydroxamic acid	Manual	2461	610	9800	14,600	5000
		Automated	1165 (666 to 2537)	875 (792 to 968)	4662 (2663 to 10,148)	21,008 (18,975 to 23,278)	4400
Panobinostat	Hydroxamic acid	Manual	92	10	368	243	61.7
		Automated	86.9 (69.8 to 145.3)	13.8 (11.7 to 16.2)	347.5 (279.1 to 581.1)	330.3 (279.9 to 388.7)	88.9
MRK1	Aryl ketone	Manual	Not tested	Not tested	Not tested	Not tested	Not tested
		Automated	16.9 (15.0 to 20.1)	3.6 (2.8 to 5.5)	77.6 (60.3 to 80.4)	86.5 (66.3 to 132.2)	17.8
MRK2	Aryl ketone	Manual	Not tested	Not tested	Not tested	Not tested	Not tested
		Automated	803.1 (624.5 to ?)^b	98.0 (89.9 to 107.0)	3212 (2498 to ?)^b	2353 (2157 to 2567)	444.4
MRK3	Ethyl ketone	Manual	Not tested	Not tested	Not tested	Not tested	Not tested
		Automated	113.9 (103.1 to 126.0)	18.3 (13.6 to 21.9)	455.5 (412.2 to 504.2)	439.5 (326.6 to 525.6)	98.8
MRK4	Ethyl ketone	Manual	Not tested	Not tested	Not tested	Not tested	Not tested
		Automated	12.9 (11.8 to 14.1)	3.46 (3.09 to 3.82)	51.6 (47.4 to 56.2)	83.7 (75.9 to 91.6)	44.4

Numbers in parentheses indicate 95% confidence intervals.

No defined E_max.

We next sought to define a PK/PD model of HDACi-induced HIV latency reversal. The goal of this model is to predict in vivo exposures that would result in maximal HIV latency reversal for novel HDACis. Toward this aim, we used vorinostat, panobinostat, and four novel HDACis (MRK-1–4; see the Methods section) that were generated specifically for the analysis of HIV latency reversal^37,38 and are selective for HDAC1, 2, and 3 (class I) isoforms. These HDACis belong to 2 different structural classes, defined by their zinc-binding pharmacophore, which dictates the affinity and t½ for the different HDAC isoforms ( Suppl. Table S1 ). Ethyl-ketone compounds, MRK3 and MRK4, exhibit short HDAC enzymatic binding half-lives, similar to vorinostat and panobinostat (hydroxamic acid pharmacophore),³⁸ whereas aryl-ketone compounds, MRK1 and MRK2, have higher affinity for HDACis and have longer half-lives that are more similar to romidepsin ( Suppl. Table S1 ).³⁷

To generate an exposure-response model for HDACi-induced HIV latency reversal required a larger data set, including a full dose response over different exposures between 4 and 24 h of each compound. To make the model more robust and help establish the relationship between the timing and magnitude of optimal target engagement for subsequent HIV latency reversal, we additionally sought to quantify histone acetylation as the proximal biomarker for HDACi target engagement. Automation allowed for higher throughput with more time points than could otherwise be managed manually.

Novel Automation and Data Processing for Latency Washout Time Course

Automation of the latency reversal assay allowed miniaturization to a 384-well format from a flask/24-well format and multiplexing with histone acetylation, which significantly increased the throughput of the assay as well as a more complete view of the target engagement and PD relationship. Optimization of the centrifuge, aspirate, and dispense cycle to minimize suspension cell loss but maximize HDACi removal was required to efficiently automate the assay. GFP and histone acetylation were normalized to untreated control wells and expressed as a fold-change over untreated cells. Raw data and relevant assay information were compiled and processed in as single csv file directly incorporated into modeling and analysis tools described in the Methods section.

Automation and programming of the washout enabled around-the-clock processing with data generation and compilation occurring upon completion of individual plates or bulk processing at the end of runs. This allowed for assessment of data during the run for quality control checks as well as bulk processing of all data to generate a single comprehensive tabular file, compiling all assay data over multiple runs and multiple days. Miniaturization and automation of cell plating, compound washout, eGFP reading, and HA ELISA processing allowed for the generation of technical replicates of larger ranges of dose response (single 10 point for manual versus triplicate 16 point automation) at more exposure durations (addition of 8 and 12 h exposures in automation experiments). These data enabled interrogation of time points (12 time points throughout 48 h as described in Fig. 1 ) that were not investigated in manual experiments (measured eGFP at 24 h post dose for all exposure durations) due to either logistical or capacity restraints.

Data obtained from the automated assay were similar to that obtained in our manual experiments, as EC₅₀ and EAUC₅₀ values obtained from the manual experiments were within the 95% confidence intervals of technical replicates from the automated assay for vorinostat and panobinostat ( Table 1 ; Suppl. Fig. S1 ). Automation of the washout process and the ability to track larger volumes of data automatically enabled a more thorough investigation of time points over the 48 h of the assay. Assessment of time points between 8 and 18 h and performance of washouts between 8 and 12 h postdose would be problematic in a manual assay, as they would require manual time in the lab throughout a 24 h time period outside of normal working hours. However, given the autonomous nature of the assay system employed in these experiments, these additional time points could easily be added to the experiment. The robotic platform was able to generate a process for these time points and required only in-lab system setup time, which could easily be managed within regular working hours. The data generated over time also characterized the change in expression of GPF and measured histone acetylation at time points after the removal of HDACi.

Hysteresis of Histone Acetylation and eGFP Expression

Inhibition of HDACs causes hyperacetylation of histones by histone acetyltransferases, resulting in chromatin decondensation and gene expression. In oncology clinical trials, histone acetylation is often used as a proximal biomarker of HDACi target engagement. Histone acetylation and HIV reporter gene expression (eGFP expression, a surrogate of HIV latency reversal) were measured throughout the assay and normalized to untreated control cells (arrows in Fig. 1 ). The ability to measure more frequent time points throughout the time course allowed a comprehensive characterization of the kinetics of histone acetylation and eGFP expression throughout the time course at different exposures and HDACi concentrations.

As the calculated EC₅₀ for the HDACis tested, robust levels of histone acetylation were noted after incubation of cells for 4 h ( Fig. 3 ), which declined following the removal of compound. However, there was a marked difference in histone acetylation between different HDACis after compound removal. After washout of vorinostat, histone acetylation levels rapidly returned to baseline levels (by the next time point) but remained elevated for 12 h for panobinostat and 48 h for MRK1 ( Fig. 3 ). The kinetics of histone deacetylation after compound removal were correlated with the HDACi enzymatic dissociation half-life ( Suppl. Table S1 ); namely, the more rapid the enzymatic dissociation, the more rapid return to baseline histone acetylation levels. The decline in histone acetylation levels observed with MRK3 and MRK4 ( Suppl. Fig. S2 ) was similar to that observed with panobinostat. However, the decline in histone acetylation levels observed with the aryl-ketone series (MRK1 and MRK2) was slower ( Fig. 3 ; Suppl. Fig. S2 , respectively). Despite the measured increase in histone acetylation for all HDACis after a 4 h incubation, it is notable that these levels were not sufficient to induce a significant amount of HIV reporter gene expression ( Fig. 3 ; Suppl. Fig. S2 ).

Figure 3.

Time course of normalized histone acetylation (HA) and enhanced green fluorescent protein (eGFP) responses from automated washout experiments following exposure to various histone deacetylase inhibitors at their EC₅₀ values in 2D10 Jurkat cells. Washout time point (indicated by arrows) following 4 h exposure with (A) vorinostat (EC₅₀ 877 nM), (B) panobinostat (EC₅₀ 11.7nM), and (C) MRK1 (EC₅₀ 3.51 nM) and following 24 h exposure with (D) vorinostat (EC₅₀ 877nM), (E) panobinostat (EC₅₀ 11.7 nM), and (F) MRK1 (EC₅₀ 3.51 nM).

As expected, a 24 h incubation at the EC₅₀ for each of the HDACis tested resulted in more prolonged elevation of histone acetylation. However, as with shorter incubations, removal of vorinostat at 24 h led to a rapid decline in histone acetylation levels but a slower decline observed for both panobinostat, MRK1 ( Fig. 3 ), MRK2, and MRK3 ( Suppl. Fig. S2 ). The maximal level of histone acetylation achieved with all HDACis was higher after a 24 h incubation than that after a 4 h incubation ( Fig. 3 ; Suppl. Fig. S2 ). In addition, following a 24 h incubation, HIV latency reversal (eGFP expression) was noted for each HDACi, which peaked at 30 h and declined by 48 h. This delay in HIV latency reversal was noted for all HDACis tested ( Fig. 3 ; Suppl. Fig. S2 ). It is interesting to note that HIV latency reversal was observed only following a 24 h incubation with HDACis, despite similar peak levels of histone acetylation achieved after more than 4 or 24 h incubations. This indicates that HIV latency reversal is not driven by achieving a specific threshold of HDACi target engagement but rather by prolonged elevation of histone acetylation.

At a 4 h exposure of higher HDACi concentrations (E_max), the time course of histone acetylation noted for each compound was similar to that of the EC₅₀ experiments, with notably higher peak levels of histone acetylation ( Fig. 4 ; Suppl. Fig. S3 ). However, unlike experiments at lower concentrations, the incubation of cells over short periods (4 h) at the E_max concentration resulted in a low level of HIV latency reversal as compared with 24 exposures ( Fig. 4 ). HIV latency reversal peaked at 18 to 24 h, which was after the peak of histone acetylation (4 h). For a 4 h exposure, eGFP expression returned to baseline levels by 48 h for all compounds tested. At 24 h exposure of E_max for each compound, higher levels of histone acetylation and HIV latency reversal were noted for all HDACis ( Fig. 4 ; Suppl. Fig. S3 ). Under these conditions, the time to peak of HIV latency reversal and histone acetylation was similar for all HDACis tested ( Fig. 4 ; Suppl. Fig. S3 ). Consistent with the mechanism of HDACis, the data revealed hysteresis between the peak of histone acetylation and the peak eGFP detection. The ability to generate all of the time points enabled a more thorough characterization of this hysteresis. Next, the relationship between exposure, target engagement (histone acetylation), and latency reversal (eGFP) was investigated.

Figure 4.

Time course of normalized histone acetylation (HA) and enhanced green fluorescent protein (eGFP) responses from automated washout experiments following exposure to various histone deacetylase inhibitors at their E_max values in 2D10 Jurkat cells. Washout time point (indicated by arrows) following 24 h exposure with (A) vorinostat (E_max 4.4 µM), (B) panobinostat (E_max 88.9 nM), and (C) MRK1 (E_max 17.6 nM) and following 24 h exposure with (D) vorinostat (E_max 4.4 µM), (E) panobinostat (E_max 88.9 nM), and (F) MRK1 (E_max 17.6 nM).

PK/PD Relationship

Using the robust data set for multiple chemical classes, relationships between exposure and the proximal (histone acetylation) and distal (eGFP) markers of latency reversal were investigated. Exposure metrics such as C_max and time above EC₅₀ were leveraged as part of the PK/PD exploration. Of the metrics analyzed, only potency-normalized AUC and area under effect curves for HA and eGFP showed significant relationships. Normalizing AUC, using EC₅₀ calculations, allowed the comparison of data from HDACis with different potencies. A relationship could be derived when comparing the AUECHA or AUECGFP to potency-normalized AUC. The relationship between AUECHA and AUECGFP and potency-normalized AUC at 4 h and 24 h exposures is shown in Figure 5 . The AUECHA ( Fig. 5A ) and AUECGFP ( Fig. 5C ) relationship for vorinostat deviated from the other HDACis following short (4 h) exposure durations but trended toward the other HDACis after 24 h exposures ( Fig. 5B, D ). This is likely because of the rapid loss of target engagement (HA) following vorinostat washout (Figs. 3 and 4), indicating that robust HIV latency reversal with vorinostat would require longer exposures in vivo.

Figure 5.

Pharmacokinetic/pharmacodynamic relationships established by comparing histone acetylation (HA) and green fluorescent protein (GFP) response over time to potency (EC₅₀)–normalized exposure. The area under the effect curve of HA (AUECHA) and area under the effect curve of GFP (AUECGFP) were derived using the trapezoidal rule measuring the DMSO normalized fold change as a function of time for each dose. Each point shown represents the area under the DMSO-normalized HA or eGFP effect over time for each dose level. The x-axis is a normalization of AUC, product of treatment concentration and exposure time, to the 24 h treatment calculated EC₅₀ for each compound. The y-axis uses the normalization of HA (A, 4 h; B, 24 h) or GFP (C, 4 h; D, 24 h) over assay duration. As these are both relationships of normalized values, units are relative.

Collectively, these data indicate that target engagement and HIV LRA activity can be modeled for different HDACis based on exposure duration. Automating the assay enabled a larger volume of data points to be collected to more precisely define the relationship between exposure, histone acetylation, and eGFP expression. This model can be used to model HIV latency reversal and establish exposure and target engagement thresholds for subsequent in vivo experiments of novel HDACis at exposures predicted to yield HIV latency reversal. It is noteworthy that these experiments also helped inform on optimal sample times from in vivo experiments to capture the peak of histone acetylation.

Discussion

Here, we sought to determine the PK parameters that defined the exposure/response relationship of HDACis to reverse HIV latency. To accomplish this goal, we performed a series of washout experiments using a Jurkat T-cell model of HIV latency. In many in vitro experiments, designed to assess the extent of HIV latency reversal, compounds are incubated with cells over a 24 h period often at the highest noncytotoxic concentration. As sustained high concentrations of LRAs are unlikely to be achieved over a 24 h period in vivo, such experiments can overestimate the magnitude of LRA activity or toxicity.^7–9 Exposure response relationships have been investigated in manual experiments with limited time points²⁷ and provide only a basic view of the exposure/response relationship.

In response to HDACis, histone acetylation occurs rapidly in the Jurkat T-cell model system of HIV latency and results in removal of the chromatin restriction of transcription. The delay in the peak of histone acetylation and reporter gene detection is likely due to the need to accumulate HIV mRNA transcripts, following removal of the chromatin blockade, that are subsequently translated into protein. Our data demonstrate that the timing of peak reporter gene expression is dictated by both the concentration and duration of HDACi. In this study, we used a Jurkat 2D10 cell line that contains a single integrated HIV provirus.²² Future studies using a population of cells harboring a number of unique HIV provirus would help determine the impact of different HIV integration sites on HDACi-induced latency reversal. Without automation, it would not have been practical to generate the data set to thoroughly characterize the hysteresis between histone acetylation and eGFP expression.

The model demonstrated that the area under the effect curve was the most relevant target engagement and PD parameter in determining exposure relationships. Our data suggest that compounds with rapid enzyme dissociation rates that are also rapidly cleared from the body are unlikely to achieve sustained HIV transcription in vivo, resulting in a lower level of HIV latency reversal. It is possible that synthesis of HIV TAT, following sustained HIV latency reversal, reaches a threshold that results in a positive feedback loop of HIV transcription. The robust data collected over multiple exposures were used to model PK thresholds for future in vivo studies to assess the tolerability of HDACis at exposures predicted to yield maximal HIV latency reversal. Furthermore, these experiments helped inform on optimal time points from in vivo studies to capture the peak of histone acetylation or HIV LRA activity.

Our studies allowed modeling of peak HIV LRA activity, following HDACi exposure, which would be key to the timing of the introduction of a “kill” agent. Furthermore, our studies demonstrated that the target pharmacologic parameters required for optimal HIV gene expression differs between HDACis with different enzyme residency times. Because HIV gene expression occurs as a function of HDACi concentration and time, the model best fitting the PK/PD relationship measures histone acetylation and latency reversal as a function of potency normalized AUC.

Automating the workflow and data processing enabled washout experiments to be performed to simultaneously generate high-density data sets that included quantifying histone acetylation (proximal biomarker of HDACi target engagement) and GFP (surrogate HIV latency reversal) for six HDACis that represent three different chemical classes. The Telios software platform was used to manage and track assay automation workflow, and data³⁹ and well results with compound mapping for both ELISA and GFP assays were able to be reformatted, consolidated, and harmonized for PK/PD studies by a custom script generated for this purpose. Flexible data analysis and processing allowed for data tracking in real time, and ultimately, this allowed for expedited analysis of a PK/PD relationship across many compounds with minimal manual interventions. Generation of automated tracking reduces the risks inherent with manual tracking, and processing of multiple plate transfers generated multiple large data sets over time. This comprehensive automated washout protocol can easily accommodate future workflows addressing PK/PD relationships in vitro, including but not limited to other HIV latency reversal compounds, regardless of cells in suspension or adherent. Furthermore, the process described here would be amenable to many automation platforms.

Supplemental Material

sj-pdf-1-jbx-10.1177_2472555220983810 – Supplemental material for In Vitro Pharmacokinetic/Pharmacodynamic Modeling of HIV Latency Reversal by Novel HDAC Inhibitors Using an Automated Platform

Supplemental material, sj-pdf-1-jbx-10.1177_2472555220983810 for In Vitro Pharmacokinetic/Pharmacodynamic Modeling of HIV Latency Reversal by Novel HDAC Inhibitors Using an Automated Platform by William Newhard, Munjal Patel, Jason Cassaday, Jeanine Ballard, Brian Squadroni, Guoxin Wu, Jian Liu, Wensheng Yu, Joe Kozlowski, Paul Zuck, Bonnie Howell, Daria Hazuda, Ryan Vargo and Richard Barnard in SLAS Discovery

Footnotes

Acknowledgements

We thank Steve Carroll, Tracy Diamond, and Julie Strizki for reviewing our manuscript and providing useful input for this work.

Supplemental material is available online with this article.

Declaration of Conflicting Interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: All authors are employed by Merck and Co., and their research and authorship of this article was completed within the scope of their employment with Merck and Co.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Jeanine Ballard

References

Bongiovanni

Casana

Tincati

; et al. Treatment Interruptions in HIV-Infected Subjects. J. Antimicrob. Chemother. 2006, 58, 502–505.

Lorenzo-Redondo

Fryer

H. R.

Bedford

; et al. Persistent HIV-1 Replication Maintains the Tissue Reservoir during Therapy. Nature 2016, 530, 51–56.

Martinez-Picado

Deeks

S. G.

Persistent HIV-1 Replication during Antiretroviral Therapy. Curr. Opin. HIV AIDS 2016, 11, 417–423.

Donahue

D. A.

Wainberg

M. A.

Cellular and Molecular Mechanisms Involved in the Establishment of HIV-1 latency. Retrovirology 2013, 10, 11.

Crooks

A. M.

Bateson

Cope

A. B.

; et al. Precise Quantitation of the Latent HIV-1 Reservoir: Implications for Eradication Strategies. J. Infect. Dis. 2015, 212, 1361–1365.

Siliciano

J. D.

Kajdas

Finzi

; et al. Long-term Follow-Up Studies Confirm the Stability of the Latent Reservoir for HIV-1 in Resting CD4+ T Cells. Nat. Med. 2003, 9, 727–728.

Grau-Expósito

Luque-Ballesteros

Navarro

; et al. Latency Reversal Agents Affect Differently the Latent Reservoir Present in Distinct CD4+ T Subpopulations. PLoS Pathog. 2019, 15, e1007991.

Khoury

Darcis

Lee

M. Y.

; et al. The Molecular Biology of HIV Latency. Adv. Exp. Med. Biol. 2018, 1075, 187–212.

Vemula

S. V.

Maxwell

J. W.

Nefedov

; et al. Identification of Proximal Biomarkers of PKC Agonism and Evaluation of Their Role in HIV reactivation. Antiviral Res. 2017, 139, 161–170.

10.

Folks

T. M.

Justement

Kinter

; et al. Cytokine-Induced Expression of HIV-1 in a Chronically Infected Promonocyte Cell Line. Science 1987, 238, 800–802.

11.

Chan

J. K.

Bhattacharyya

Lassen

K. G.

; et al. Calcium/Calcineurin Synergizes with Prostratin to Promote NF-κB Dependent Activation of Latent HIV. PLoS One 2013, 8, e77749.

12.

Coull

J. J.

Romerio

Sun

J. M.

; et al. The Human Factors YY1 and LSF Repress the Human Immunodeficiency Virus Type 1 Long Terminal Repeat via Recruitment of Histone Deacetylase 1. J. Virol. 2000, 74, 6790–6799.

13.

Friedman

Cho

W. K.

Chu

C. K.

; et al. Epigenetic Silencing of HIV-1 by the Histone H3 Lysine 27 Methyltransferase Enhancer of Zeste 2. J. Virol. 2011, 85, 9078–9089.

14.

Margolis

D. M.

Counterregulation of Chromatin Deacetylation and Histone Deacetylase Occupancy at the Integrated Promoter of Human Immunodeficiency Virus Type 1 (HIV-1) by the HIV-1 Repressor YY1 and HIV-1 Activator Tat. Mol. Cell Biol. 2002, 22, 2965–2973.

15.

Imai

Okamoto

Transcriptional Repression of Human Immunodeficiency Virus Type 1 by AP-4. J. Biol. Chem. 2006, 281, 12495–12505.

16.

Jiang

Espeseth

Hazuda

D. J.

; et al. c-Myc and Sp1 Contribute to Proviral Latency by Recruiting Histone Deacetylase 1 to the Human Immunodeficiency Virus Type 1 Promoter. J. Virol. 2007, 81, 10914–10923.

17.

Marban

Suzanne

Dequiedt

; et al. Recruitment of Chromatin-Modifying Enzymes by CTIP2 Promotes HIV-1 Transcriptional Silencing. EMBO J. 2007, 26, 412–423.

18.

Sherrill-Mix

Lewinski

M. K.

Famiglietti

; et al. HIV Latency and Integration Site Placement in Five Cell-Based Models. Retrovirology 2013, 10, 90.

19.

Shirakawa

Chavez

Hakre

; et al. Reactivation of Latent HIV by Histone Deacetylase Inhibitors. Trends Microbiol. 2013, 21, 277–285.

20.

Williams

S. A.

Chen

L. F.

Kwon

; et al. NF-kappaB p50 Promotes HIV Latency through HDAC Recruitment and Repression of Transcriptional Initiation. EMBO J. 2006, 25, 139–149.

21.

Spina

C. A.

Anderson

Archin

N. M.

; et al. An In-Depth Comparison of Latent HIV-1 Reactivation in Multiple Cell Model Systems and Resting CD4+ T Cells from Aviremic Patients. PLoS Pathog. 2013, 9, e1003834.

22.

Pearson

Kim

Y. K.

Hokello

; et al. Epigenetic Silencing of Human Immunodeficiency Virus (HIV) Transcription by Formation of Restrictive Chromatin Structures at the Viral Long Terminal Repeat Drives the Progressive Entry of HIV into Latency. J. Virol. 2008, 82, 12291–12303.

23.

Verdin

DNase I-Hypersensitive Sites Are Associated with Both Long Terminal Repeats and with the Intragenic Enhancer of Integrated Human Immunodeficiency Virus Type 1. J. Virol. 1991, 65, 6790–6769.

24.

Verdin

Paras

Jr. Van Lint

Chromatin Disruption in the Promoter of Human Immunodeficiency Virus Type 1 during Transcriptional Activation. EMBO J. 1993, 12, 3249–3259.

25.

Archin

N. M.

Eron

J. J.

Palmer

; et al. Valproic Acid without Intensified Antiviral Therapy Has Limited Impact on Persistent HIV Infection of Resting CD4+ T Cells. AIDS 2008, 22, 1131–1135.

26.

Edelstein

L. C.

Micheva-Viteva

Phelan

B. D.

; et al. Short Communication: Activation of Latent HIV Type 1 Gene Expression by Suberoylanilide Hydroxamic Acid (SAHA), an HDAC Inhibitor Approved for Use to Treat Cutaneous T Cell Lymphoma. AIDS Res. Hum. Retroviruses 2009, 25, 883–887.

27.

Maxwell

J. W.

Falcinelli

S. D.

Nefedov

; et al. Cellular Gene Modulation of HIV Infected CD4 T Cells in Response to Serial Treatment with the Histone Deacetylase (HDAC) Inhibitor Vorinostat. J. Virol. 2020, 94, e00351–20.

28.

Van Lint

Emiliani

Ott

; et al. Transcriptional Activation and Chromatin Remodeling of the HIV-1 Promoter in Response to Histone Acetylation. EMBO J. 1996, 15, 1112–1120.

29.

Archin

N. M.

Bateson

Tripathy

M. K.

; et al. HIV-1 Expression within Resting CD4+ T Cells after Multiple Doses of Vorinostat. J. Infect. Dis. 2014, 210, 728–735.

30.

Archin

N. M.

Liberty

A. L.

Kashuba

A. D.

; et al. Administration of Vorinostat Disrupts HIV-1 Latency in Patients on Antiretroviral Therapy. Nature 2012, 487, 482–485.

31.

Rasmussen

T. A.

Tolstrup

Brinkmann

C. R.

; et al. Panobinostat, a Histone Deacetylase Inhibitor, for Latent-Virus Reactivation in HIV-Infected Patients on Suppressive Antiretroviral Therapy: A Phase 1/2, Single Group, Clinical Trial. Lancet HIV 2014, 1, e13–e21.

32.

Søgaard

O. S.

Graversen

M. E.

Leth

; et al. The Depsipeptide Romidepsin Reverses HIV-1 Latency In Vivo. PLoS Pathog. 2015, 11, e1005142.

33.

Swanson

Talla

; et al. HDAC Inhibition Induces HIV-1 Protein and Enables Immune-Based Clearance following Latency Reversal. JCI Insight 2017, 2, e92901.

34.

Rasmussen

T. A.

Schmeltz Sogaard

Brinkmann

; et al. Comparison of HDAC Inhibitors in Clinical Development: Effect on HIV Production in Latently Infected Cells and T-Cell Activation. Hum. Vaccin. Immunother. 2013, 9, 993–1001.

35.

VanScoy

B. D.

Mendes

R. E.

Castanheira

; et al. Relationship between Ceftolozane-Tazobactam Exposure and Selection for Pseudomonas aeruginosa Resistance in a Hollow-Fiber Infection Model. Antimicrob. Agents Chemother. 2014, 58, 6024–6031.

36.

Wang

Wismer

M. K.

Racine

; et al. Development of an Integrated Semi-Automated System for In Vitro Pharmacodynamic Modelling. J. Antimicrob. Chemother. 2008, 62, 1070–1077.

37.

Liu

Kelly

; et al. Selective Class I HDAC Inhibitors Based on Aryl Ketone Zinc Binding Induce HIV-1 Protein for Clearance. ACS Med. Chem. Lett. 2020, 11, 1476–1483.

38.

Liu

; et al. Discovery of Ethyl Ketone-Based HDACs 1, 2, and 3 Selective Inhibitors for Reversing HIV Latency. Bioorg. Med. Chem. Lett. 2020, 13, 127–197.

39.

Cassaday

Finley

Squadroni

; et al. Development of a Platform to Enable Fully Automated Cross-Titration Experiments. SLAS Technol. 2017, 22, 195–205.

40.

Tsai

Baker

C. E.

; et al. In Vivo Analysis of the Effect of Panobinostat on Cell-Associated HIV RNA and DNA Levels and Latent HIV Infection. Retrovirology 2016, 13, 36.

41.

Team

R. C.

R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing: Vienna, Austria. https://www.R-project.org. 2019.

42.

Huber

Doyon

Plaks

; et al. Inhibitors of Histone Deacetylases: Correlation between Isoform Specificity and Reactivation of HIV Type 1 (HIV-1) from Latently Infected Cells. J. Biol. Chem. 2011, 286, 22211–22218.

43.

Keedy

K. S.

Archin

N. M.

Gates

A. T.

; et al. A Limited Group of Class I Histone Deacetylases Acts to Repress Human Immunodeficiency Virus Type 1 Expression. J. Virol. 2009, 83, 4749–4756.

44.

Zhang

Barbieri

C. M.

Garcia-Calvo

; et al. Moderate to High Throughput In Vitro Binding Kinetics for Drug Discovery. Front. Biosci. (Schol. Ed.) 2016, 8, 278–297.

45.

Tummino

P. J.

Copeland

R. A.

Residence Time of Receptor-Ligand Complexes and Its Effect on Biological Function. Biochemistry 2008, 47, 5481–5492.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.30 MB