Functional profiling technologies using arrayed collections of genome-scale siRNA and cDNA arrayed libraries enable the comprehensive global analysis of gene function. However, the current repertoire of high-throughput detection methodologies has limited the scope of cellular phenotypes that can be studied. In this report, we describe the systematic identification of mammalian growth-regulatory factors achieved through the integration of automated microscopy, pattern recognition analysis, and cell-based functional genomics. The effects of 7364 human and mouse proteins, encoded by individually arrayed cDNAs, upon proliferation and viability in U2OS osteosarcoma cells were evaluated in a live-cell, kinetic assay using quantitative image analysis. Overexpression of more than 86 cDNAs (1.15%) conferred dramatic increases in the proliferation, as determined cell enumeration. These included several known growth regulators, as well as previously uncharacterized ones (LRRK1, Ankrd25). In addition, novel functional roles for two genes (5033414D02Rik, 2810429O05Rik), now termed Gatp1 and Gatp2, respectively, were identified. Further analysis demonstrated that these encoded proteins promoted cellular proliferation and transformation in primary cells. Conversely, cells depleted for Gatp1 underwent apoptosis upon serum reduction, suggesting that Gatp1 is essential for cell survival under growth-factor-restricted conditions. Taken together, our findings offer new insight into the regulation of cellular growth and proliferation, and demonstrate the value and feasibility of assessing cellular phenotypes through genome-level computational image analysis

This report describes statistical validation methods implemented on assay data for inhibition of subcellular redistribution of nuclear factor kappaB (NF kappaB) in HeLa cells. We quantified cellular inhibition of cytoplasmic-nuclear translocation of NF kappaB in response to a range of concentrations of interleukin-1 (IL-1) receptor antagonist in the presence of IL-1alpha using eight replicate rows in each four 96-well plates scanned five times on each of 2 days. Translocation was measured as the fractional localized intensity of the nucleus (FLIN), an implementation of our more general fractional localized intensity of the compartments (FLIC), which analyzes whole compartments in the context of the entire cell. The NF kappaB antagonist assay (inhibition of IL-1- induced NF kappaB translocation) data were collected on a Q3DM (San Diego, CA) EIDAQtrade mark 100 high throughput microscopy system. [In 2003, Q3DM was purchased by Beckman Coulter Inc. (Fullerton, CA), which released the IC 100 successor to the EIDAQ 100.] The generalized FLIC method is described along with two-point (minimum-maximum) and multiple point titration statistical methods. As a ratio of compartment intensities that tend to change proportionally, FLIN was resistant to photobleaching errors. Two-point minimum-maximum statistical analyses yielded the following: a Z' of 0.174 with the data as n = 320 independent well samples; Z' by row data in a range of 0.393-0.933, with a mean of 0.766; by-plate Z' data of 0.310, 0.443, 0.545, and 0.794; and by-plate means of columns Z' data of 0.879, 0.927, 0.945, and 0.963. The mean 50% inhibitory concentration (IC50) for IL-1 receptor antagonist over all experiments was 213 ng/ml. The combined IC50 coefficients of variation (CVs) were 0.74%, 0.85%, 2.09%, and 2.52% for the four plates. Repeatability IC50 CVs were as follows: day to day 3.0%, row to row 8.0%, plate to plate 2.8%, and day to day 0.6%. The number of cells required for statistically resolvable differences in dose concentrations, plotted in a family of FLIN sigma/deltamicro (SD/range) curves and tabulated, demonstrated cell-by-cell assay precision with our combined sigma/deltamicro = 0.32 that required approximately 10-fold fewer cells than in a previously reported NF kappaB assay with sigma/deltamicro = 1.52. To better understand the relationship between cell-by-cell measurements and IC50 precision, 500 Monte Carlo simulations with varying cell-measurement SDs were used to explore three-, five-, seven-, and 11-point model titrations. The reductions in deltaIC50 90% confidence intervals from 11- to three-point titrations were 10-fold with the previously reported sigma/deltamicro = 1.52 and twofold with our sigma/deltamicro = 0.32. With these normalized parameters, this report provides a common statistical foundation, independent of the assay details, for evaluating the performance of imaging data on any instrument.

Lipolysis in adipocytes is associated with phosphorylation of hormone sensitive lipase (HSL) and translocation of HSL to lipid droplets. In this study, adipocytes were cultured in a high-throughput format (96-well dishes), exposed to lipolytic agents, and then fixed and labeled for nuclei, lipid droplets, and HSL (or HSL phosphorylated on serine 660 [pHSLser660]). The cells were imaged via automated digital fluorescence microscopy, and high-content analysis (HCA) methods were used to quantify HSL phosphorylation and the degree to which HSL (or pHSLser660) colocalizes with the lipid droplets. HSL:lipid droplet colocalization was quantified through use of Pearson's correlation, Mander's M1 Colocalization, and the Tanimoto coefficient. For murine 3T3L1 adipocytes, isoproterenol, Lys-γ3-melanocyte stimulating hormone, and forskolin elicited the appearance and colocalization of pHSLser660, whereas atrial natriuretic peptide (ANP) did not. For human subcutaneous adipocytes, isoproterenol, forskolin, and ANP activated HSL phosphorylation/colocalization, but Lys-γ3-melanocyte stimulating hormone had little or no effect. Since ANP activates guanosine 3',5'-cyclic monophosphate (cGMP)-dependent protein kinase, HSL serine 660 is likely a substrate for cGMP-dependent protein kinase in human adipocytes. For both adipocyte model systems, adipocytes with the greatest lipid content displayed the greatest lipolytic responses. The results for pHSLser660 were consistent with release of glycerol by the cells, a well-established assay of lipolysis, and the HCA methods yielded Z' values >0.50. The results illustrate several key differences between human and murine adipocytes and demonstrate advantages of utilizing HCA techniques to study lipolysis in cultured adipocytes.

Efficient image cytometry of a conventional microscope slide means rapid acquisition and analysis of 20 gigapixels of image data (at 0.3-microm sampling). The voluminous data motivate increased acquisition speed to enable many biomedical applications. Continuous-motion time-delay-and-integrate (TDI) scanning has the potential to speed image acquisition while retaining sensitivity, but the challenge of implementing high-resolution autofocus operating simultaneously with acquisition has limited its adoption. We develop a dynamic autofocus system for this need using: 1. a "volume camera," consisting of nine fiber optic imaging conduits to charge-coupled device (CCD) sensors, that acquires images in parallel from different focal planes, 2. an array of mixed analog-digital processing circuits that measure the high spatial frequencies of the multiple image streams to create focus indices, and 3. a software system that reads and analyzes the focus data streams and calculates best focus for closed feedback loop control. Our system updates autofocus at 56 Hz (or once every 21 microm of stage travel) to collect sharply focused images sampled at 0.3x0.3 microm(2)/pixel at a stage speed of 2.3 mms. The system, tested by focusing in phase contrast and imaging long fluorescence strips, achieves high-performance closed-loop image-content-based autofocus in continuous scanning for the first time.

No abstract