Quantifying drug tissue biodistribution by integrating high content screening with deep-learning analysis.

Quantitatively determining in vivo achievable drug concentrations in targeted organs of animal models and subsequent target engagement confirmation is a challenge to drug discovery and translation due to lack of bioassay technologies that can discriminate drug binding with different mechanisms. We have developed a multiplexed and high-throughput method to quantify drug distribution in tissues by integrating high content screening (HCS) with U-Net based deep learning (DL) image analysis models. This technology combination allowed direct visualization and quantification of biologics drug binding in targeted tissues with cellular resolution, thus enabling biologists to objectively determine drug binding kinetics.

Challenges and Opportunities in Enabling High-Throughput, Miniaturized High Content Screening.

Within the Drug Discovery industry, there is a growing recognition of the value of high content screening (HCS), particularly as researchers aim to screen compounds and identify hits using more physiologically relevant in vitro cell-based assays. Image-based high content screening, with its combined ability to yield multiparametric data, provide subcellular resolution, and enable cell population analysis, is well suited to this challenge. While HCS has been in routine use for over a decade, a number of hurdles have historically prohibited very large, miniaturized high-throughput screening efforts with this platform. Suitable hardware and consumables for conducting 1536-well HCS have only recently become available, and developing a reliable informatics framework to accommodate the scale of high-throughput HCS data remains a considerable challenge. Additionally, innovative approaches are needed to interpret the large volumes of content-rich information generated. Despite these hurdles, there has been a growing interest in screening large compound inventories using this platform. Here, we outline the infrastructure developed and applied at Bristol-Myers Squibb for 1536-well high content screening and discuss key lessons learned.

HCS road: an enterprise system for integrated HCS data management and analysis.

The effective analysis and interpretation of high-content screening (HCS) data requires joining results to information on experimental treatments and controls, normalizing data, and selecting hits or fitting concentration-response curves. HCS data have unique requirements that are not supported by traditional high-throughput screening databases, including the ability to designate separate positive and negative controls for different measurements in multiplexed assays; the ability to capture information on the cell lines, fluorescent reagents, and treatments in each assay; the ability to store and use individual-cell and image data; and the ability to support HCS readers and software from multiple vendors along with third-party image analysis tools. To address these requirements, the authors developed an enterprise system for the storage and processing of HCS images and results. This system, HCS Road, supports target identification, lead discovery, lead evaluation, and lead profiling activities. A dedicated client supports experimental design, data review, and core analyses and displays images together with results for assay development, hit assessment, and troubleshooting. Data can be exported to third-party applications for further analysis and exploration. HCS Road provides a single source for high-content results across the organization, regardless of the group or instrument that produced them.

Best practices in compound management for preserving compound integrity and accurately providing samples for assays.

Preserving the integrity of the compound collection and providing high-quality materials for drug discovery in an efficient and cost-effective manner are 2 major challenges faced by compound management (CM) at Bristol-Myers Squibb (BMS). The demands on CM include delivering hundreds of thousands of compounds a year to a variety of operations. These operations range from single-compound requests to hit identification support and just-in-time assay plate provision for lead optimization. Support needs for these processes consist of the ability to rapidly provide compounds as solids or solutions in a variety of formats, establishing proper long- and short-term storage conditions and creating appropriate methods for handling concentrated, potent compounds for delivery to sensitive biological assays. A series of experiments evaluating the effects of processing compounds with volatile solvents, storage conditions that can induce freeze/thaw cycles, and the delivery of compounds were performed. This article presents the results of these experiments and how they affect compound integrity and the accuracy of compound management processes.

Albumin binding, relaxivity, and water exchange kinetics of the diastereoisomers of MS-325, a gadolinium(III)-based magnetic resonance angiography contrast agent.

The amphiphilic gadolinium complex MS-325 ((trisodium-{(2-(R)-[(4,4-diphenylcyclohexyl) phosphonooxymethyl] diethylenetriaminepentaacetato) (aquo)gadolinium(III)}) is a contrast agent for magnetic resonance angiography (MRA). MS-325 consists of two slowly interconverting diastereoisomers, A and B (65:35 ratio), which can be isolated at pH > 8.5 (TyeklAr, Z.; Dunham, S. U.; Midelfort, K.; Scott, D. M.; Sajiki, H.; Ong, K.; Lauffer, R. B.; Caravan, P.; McMurry, T. J. Inorg. Chem. 2007, 46, 6621-6631). MS-325 binds to human serum albumin (HSA) in plasma resulting in an extended plasma half-life, retention of the agent within the blood compartment, and an increased relaxation rate of water protons in plasma. Under physiological conditions (37 degrees C, pH 7.4, phosphate buffered saline (PBS), 4.5% HSA, 0.05 mM complex), there is no statistical difference in HSA affinity or relaxivity between the two isomers (A 88.6 +/- 0.6% bound, r1 = 42.0 +/- 1.0 mM(-1) s(-1) at 20 MHz; B 90.2 +/- 0.6% bound, r1 = 38.3 +/- 1.0 mM(-1) s(-1) at 20 MHz; errors represent 1 standard deviation). At lower temperatures, isomer A has a higher relaxivity than isomer B. The water exchange rates in the absence of HSA at 298 K, kA298 = 5.9 +/- 2.8 x 10(6) s(-1), kB298 = 3.2 +/- 1.8 x 10(6) s(-1), and heats of activation, DeltaHA = 56 +/- 8 kJ/mol, DeltaHB = 59 +/- 11 kJ/mol, were determined by variable-temperature 17O NMR at 7.05 T. Proton nuclear magnetic relaxation dispersion (NMRD) profiles were recorded over the frequency range of 0.01-50 MHz at 5, 15, 25, and 35 degrees C in a 4.5% HSA in PBS solution for each isomer (0.1 mM). Differences in the relaxivity in HSA between the two isomers could be attributed to the differing water exchange rates.

When are two waters worse than one? Doubling the hydration number of a Gd-DTPA derivative decreases relaxivity.

The synthesis of a novel ligand, based on N-methyl-diethylenetriaminetetraacetate and containing a diphenylcyclohexyl serum albumin binding group (L1) is described and the coordination chemistry and biophysical properties of its Gd(III) complex Gd-L1 are reported. The Gd(III) complex of the diethylenetriaminepentaacetate analogue of the ligand described here (L2) is the MRI contrast agent MS-325. The effect of converting an acetate to a methyl group on metal-ligand stability, hydration number, water-exchange rate, relaxivity, and binding to the protein human serum albumin (HSA) is explored. The complex Gd-L1 has two coordinated water molecules in solution, that is, [Gd(L1)(H2O)2]2- as shown by D-band proton ENDOR spectroscopy and implied by 1H and 17O NMR relaxation rate measurements. The Gd-H(water) distance of the coordinated waters was found to be identical to that found for Gd-L2, 3.08 A. Loss of the acetate group destabilizes the Gd(III) complex by 1.7 log units (log K(ML) = 20.34) relative to the complex with L2. The affinity of Gd-L1 for HSA is essentially the same as that of Gd-L2. The water-exchange rate of the two coordinated waters on Gd-L1 (k(ex) = 4.4x10(5) s(-1)) is slowed by an order of magnitude relative to Gd-L2. As a result of this slow water-exchange rate, the observed proton relaxivity of Gd-L1 is much lower in a solution of HSA under physiological conditions (r1(obs) = 22.0 mM(-1) s(-1) for 0.1 mM Gd-L1 in 0.67 mM HSA, HEPES buffer, pH 7.4, 35 degrees C at 20 MHz) than that of Gd-L2 (r1(obs) = 41.5 mM(-1) s(-1)) measured under the same conditions. Despite having two exchangeable water molecules, slow water exchange limits the potential efficacy of Gd-L1 as an MRI contrast agent.

The interaction of MS-325 with human serum albumin and its effect on proton relaxation rates.

MS-325 is a novel blood pool contrast agent for magnetic resonance imaging currently undergoing clinical trials to assess blockage in arteries. MS-325 functions by binding to human serum albumin (HSA) in plasma. Binding to HSA serves to prolong plasma half-life, retain the agent in the blood pool, and increase the relaxation rate of water protons in plasma. Ultrafiltration studies with a 5 kDa molecular weight cutoff filter show that MS-325 binds to HSA with stepwise stoichiometric affinity constants (mM(-1)) of K(a1) = 11.0 +/- 2.7, K(a2) = 0.84 +/- 0.16, K(a3) = 0.26 +/- 0.14, and K(a4) = 0.43 +/- 0.24. Under the conditions 0.1 mM MS-325, 4.5% HSA, pH 7.4 (phosphate-buffered saline), and 37 degrees C, 88 +/- 2% of MS-325 is bound to albumin. Fluorescent probe displacement studies show that MS-325 can displace dansyl sarcosine and dansyl-L-asparagine from HSA with inhibition constants (K(i)) of 85 +/- 3 microM and 1500 +/- 850 microM, respectively; however, MS-325 is unable to displace warfarin. These results suggest that MS-325 binds primarily to site II on HSA. The relaxivity of MS-325 when bound to HSA is shown to be site dependent. The Eu(III) analogue of MS-325 is shown to contain one inner-sphere water molecule in the presence and in the absence of HSA. The synthesis of an MS-325 analogue, 5, containing no inner-sphere water molecules is described. Compound 5 is used to estimate the contribution to relaxivity from the outer-sphere water molecules surrounding MS-325. The high relaxivity of MS-325 bound to HSA is primarily because of a 60-100-fold increase in the rotational correlation time of the molecule upon binding (tau(R) = 10.1 +/- 2.6 ns bound vs 115 ps free). Analysis of the nuclear magnetic relaxation dispersion (T(1) and T(2)) profiles also suggests a decrease in the electronic relaxation rate (1/T(1e) at 20 MHz = 2.0 x 10(8) s(-1) bound vs 1.1 x 10(9) s(-1) free) and an increase in the inner-sphere water residency time (tau(m) = 170 +/- 40 ns bound vs 69 +/- 20 ns free).