Automated subcellular location determination and high-throughput microscopy

A generative model of microtubule distributions, and indirect estimation of its parameters from fluorescence microscopy images

The microtubule network plays critical roles in many cellular processes, and quantitative models of how its organization varies across cell types and conditions are required for understanding those roles and as input to cell simulations. High-throughput image acquisition technologies are potentially valuable for this purpose, but do not provide sufficient resolution for current analysis methods that rely on tracing of individual microtubules. We describe a parametric conditional model of microtubule distribution that can generate a microtubule network in intact cells using a persistent random walk approach. The model parameters are physically meaningful as they directly describe the spatial distribution of microtubules and include the number of microtubules as well as the mean of the length distribution. We also present an indirect method for estimating the parameters of the model from three-dimensional fluorescence microscope images of cells that relies on comparing acquired images with simulated images generated from the model. Our results show that our method can reasonably recover parameters for a given query image, and we present the distributions of parameters estimated by our method for a collection of HeLa cell images. (c) 2010 International Society for Advancement of Cytometry.

A Graphical Model to Determine the Subcellular Protein Location in Artificial Tissues

Location proteomics is concerned with the systematic analysis of the subcellular location of proteins. In order to perform comprehensive analysis of all protein location patterns, automated methods are needed. With the goal of extending automated subcellular location pattern analysis methods to high resolution images of tissues, 3D confocal microscope images of polarized CaCo2 cells immunostained for various proteins were collected. A three-color staining protocol was developed that permits parallel imaging of proteins of interest as well as DNA and the actin cytoskeleton. The collection is composed of 11 to 21 images for each of the 9 proteins that depict major subcellular patterns. A classifier was trained to recognize the subcellular location pattern of segmented cells with an accuracy of 89.2%. Using the Prior Updating method allowed improvement of this accuracy to 99.6%. This study demonstrates the benefit of using a graphical model approach for improving the pattern classification in tissue images.

High content image cytometry in the context of subnuclear organization

The organization of proteins in space and time is essential to their function. To accurately quantify subcellular protein characteristics in a population of cells with regard for the stochasticity of events in a natural context, there is a fast-growing need for image-based cytometry. Simultaneously, the massive amount of data that is generated by image-cytometric analyses, calls for tools that enable pattern recognition and automated classification. In this article, we present a general approach for multivariate phenotypic profiling of individual cell nuclei and quantification of subnuclear spots using automated fluorescence mosaic microscopy, optimized image processing tools, and supervised classification. We demonstrate the efficiency of our analysis by determination of differential DNA damage repair patterns in response to genotoxic stress and radiation, and we show the potential of data mining in pinpointing specific phenotypes after transient transfection. The presented approach allowed for systematic analysis of subnuclear features in large image data sets and accurate classification of phenotypes at the level of the single cell. Consequently, this type of nuclear fingerprinting shows potential for high-throughput applications, such as functional protein assays or drug compound screening.

Phenotypic profiling of the human genome by time-lapse microscopy reveals cell division genes

Despite our rapidly growing knowledge about the human genome, we do not know all of the genes required for some of the most basic functions of life. To start to fill this gap we developed a high-throughput phenotypic screening platform combining potent gene silencing by RNA interference, time-lapse microscopy and computational image processing. We carried out a genome-wide phenotypic profiling of each of the approximately 21,000 human protein-coding genes by two-day live imaging of fluorescently labelled chromosomes. Phenotypes were scored quantitatively by computational image processing, which allowed us to identify hundreds of human genes involved in diverse biological functions including cell division, migration and survival. As part of the Mitocheck consortium, this study provides an in-depth analysis of cell division phenotypes and makes the entire high-content data set available as a resource to the community.

A new method for predicting the subcellular localization of eukaryotic proteins with both single and multiple sites: Euk-mPLoc 2.0

Information of subcellular locations of proteins is important for in-depth studies of cell biology. It is very useful for proteomics, system biology and drug development as well. However, most existing methods for predicting protein subcellular location can only cover 5 to 12 location sites. Also, they are limited to deal with single-location proteins and hence failed to work for multiplex proteins, which can simultaneously exist at, or move between, two or more location sites. Actually, multiplex proteins of this kind usually posses some important biological functions worthy of our special notice. A new predictor called "Euk-mPLoc 2.0" is developed by hybridizing the gene ontology information, functional domain information, and sequential evolutionary information through three different modes of pseudo amino acid composition. It can be used to identify eukaryotic proteins among the following 22 locations: (1) acrosome, (2) cell wall, (3) centriole, (4) chloroplast, (5) cyanelle, (6) cytoplasm, (7) cytoskeleton, (8) endoplasmic reticulum, (9) endosome, (10) extracell, (11) Golgi apparatus, (12) hydrogenosome, (13) lysosome, (14) melanosome, (15) microsome (16) mitochondria, (17) nucleus, (18) peroxisome, (19) plasma membrane, (20) plastid, (21) spindle pole body, and (22) vacuole. Compared with the existing methods for predicting eukaryotic protein subcellular localization, the new predictor is much more powerful and flexible, particularly in dealing with proteins with multiple locations and proteins without available accession numbers. For a newly-constructed stringent benchmark dataset which contains both single- and multiple-location proteins and in which none of proteins has pairwise sequence identity to any other in a same location, the overall jackknife success rate achieved by Euk-mPLoc 2.0 is more than 24% higher than those by any of the existing predictors. As a user-friendly web-server, Euk-mPLoc 2.0 is freely accessible at http://www.csbio.sjtu.edu.cn/bioinf/euk-multi-2/. For a query protein sequence of 400 amino acids, it will take about 15 seconds for the web-server to yield the predicted result; the longer the sequence is, the more time it may usually need. It is anticipated that the novel approach and the powerful predictor as presented in this paper will have a significant impact to Molecular Cell Biology, System Biology, Proteomics, Bioinformatics, and Drug Development.

Automated microscopy for high-content RNAi screening

Fluorescence microscopy is one of the most powerful tools to investigate complex cellular processes such as cell division, cell motility, or intracellular trafficking. The availability of RNA interference (RNAi) technology and automated microscopy has opened the possibility to perform cellular imaging in functional genomics and other large-scale applications. Although imaging often dramatically increases the content of a screening assay, it poses new challenges to achieve accurate quantitative annotation and therefore needs to be carefully adjusted to the specific needs of individual screening applications. In this review, we discuss principles of assay design, large-scale RNAi, microscope automation, and computational data analysis. We highlight strategies for imaging-based RNAi screening adapted to different library and assay designs.

Multi label learning for prediction of human protein subcellular localizations

Predicting protein subcellular locations has attracted much attention in the past decade. However, one of the most challenging problems is that many proteins were found simultaneously existing in, or moving between, two or more different cell components in a eukaryotic cell. Seldom previous predictors were able to deal with such multiplex proteins although they have extremely important implications in future drug discovery in terms of their specific subcellular targeting. Approximately 20% of the human proteome consists of such multiplex proteins with multiple sample labels. In order to efficiently handle such multiplex human proteins, we have developed a novel multi-label (ML) learning and prediction framework called ML-PLoc, which decomposes the multi-label prediction problem into multiple independent binary classification problems. ML-PLoc is constructed based on support vector machine (SVM) and sequential evolution information. Experimental results show that ML-PLoc can achieve an overall accuracy 64.6% and recall ratio 67.2% on a benchmark dataset consisting of 14 human subcellular locations, and is very powerful for dealing with multiplex proteins. The current approach represents a new strategy to deal with the multi-label biological problems. ML-PLoc software is freely available for academic use at: http://www.csbio.sjtu.edu.cn/bioinf/ML-PLoc .

Local binary patterns variants as texture descriptors for medical image analysis

OBJECTIVE

This paper focuses on the use of image-based machine learning techniques in medical image analysis. In particular, we present some variants of local binary patterns (LBP), which are widely considered the state of the art among texture descriptors. After we provide a detailed review of the literature about existing LBP variants and discuss the most salient approaches, along with their pros and cons, we report new experiments using several LBP-based descriptors and propose a set of novel texture descriptors for the representation of biomedical images. The standard LBP operator is defined as a gray-scale invariant texture measure, derived from a general definition of texture in a local neighborhood. Our variants are obtained by considering different shapes for the neighborhood calculation and different encodings for the evaluation of the local gray-scale difference. These sets of features are then used for training a machine-learning classifier (a stand-alone support vector machine).

METHODS AND MATERIALS

Extensive experiments are conducted using the following three datasets:

RESULTS AND CONCLUSION

Our results show that the novel variant named elongated quinary patterns (EQP) is a very performing method among those proposed in this work for extracting information from a texture in all the tested datasets. EQP is based on an elliptic neighborhood and a 5 levels scale for encoding the local gray-scale difference. Particularly interesting are the results on the widely studied 2D-HeLa dataset, where, to the best of our knowledge, the proposed descriptor obtains the highest performance among all the several texture descriptors tested in the literature.

Automated evaluation of autoantibodies on human epithelial-2 cells as an approach to standardize cell-based immunofluorescence tests

Introduction

Analysis of autoantibodies (AAB) by indirect immunofluorescence (IIF) is a basic tool for the serological diagnosis of systemic rheumatic disorders. Automation of autoantibody IIF reading including pattern recognition may improve intra- and inter-laboratory variability and meet the demand for cost-effective assessment of large numbers of samples. Comparing automated and visual interpretation, the usefulness for routine laboratory diagnostics was investigated.

Methods

Autoantibody detection by IIF on human epithelial-2 (HEp-2) cells was conducted in a total of 1222 consecutive sera of patients with suspected systemic rheumatic diseases from a university routine laboratory (n = 924) and a private referral laboratory (n = 298). IIF results from routine diagnostics were compared with a novel automated interpretation system.

Results

Both diagnostic procedures showed a very good agreement in detecting AAB (kappa = 0.828) and differentiating respective immunofluorescence patterns. Only 98 (8.0%) of 1222 sera demonstrated discrepant results in the differentiation of positive from negative samples. The contingency coefficients of chi-square statistics were 0.646 for the university laboratory cohort with an agreement of 93.0% and 0.695 for the private laboratory cohort with an agreement of 90.6%, P < 0.0001, respectively. Comparing immunofluorescence patterns, 111 (15.3%) sera yielded differing results.

Conclusions

Automated assessment of AAB by IIF on HEp-2 cells using an automated interpretation system is a reliable and robust method for positive/negative differentiation. Employing novel mathematical algorithms, automated interpretation provides reproducible detection of specific immunofluorescence patterns on HEp-2 cells. Automated interpretation can reduce drawbacks of IIF for AAB detection in routine diagnostics providing more reliable data for clinicians.

Visualization of image data from cells to organisms

Advances in imaging techniques and high-throughput technologies are providing scientists with unprecedented possibilities to visualize internal structures of cells, organs and organisms and to collect systematic image data characterizing genes and proteins on a large scale. To make the best use of these increasingly complex and large image data resources, the scientific community must be provided with methods to query, analyze and crosslink these resources to give an intuitive visual representation of the data. This review gives an overview of existing methods and tools for this purpose and highlights some of their limitations and challenges.

Inference of RhoGAP/GTPase regulation using single-cell morphological data from a combinatorial RNAi screen

Biological networks are highly complex systems, consisting largely of enzymes that act as molecular switches to activate/inhibit downstream targets via post-translational modification. Computational techniques have been developed to perform signaling network inference using some high-throughput data sources, such as those generated from transcriptional and proteomic studies, but comparable methods have not been developed to use high-content morphological data, which are emerging principally from large-scale RNAi screens, to these ends. Here, we describe a systematic computational framework based on a classification model for identifying genetic interactions using high-dimensional single-cell morphological data from genetic screens, apply it to RhoGAP/GTPase regulation in Drosophila, and evaluate its efficacy. Augmented by knowledge of the basic structure of RhoGAP/GTPase signaling, namely, that GAPs act directly upstream of GTPases, we apply our framework for identifying genetic interactions to predict signaling relationships between these proteins. We find that our method makes mediocre predictions using only RhoGAP single-knockdown morphological data, yet achieves vastly improved accuracy by including original data from a double-knockdown RhoGAP genetic screen, which likely reflects the redundant network structure of RhoGAP/GTPase signaling. We consider other possible methods for inference and show that our primary model outperforms the alternatives. This work demonstrates the fundamental fact that high-throughput morphological data can be used in a systematic, successful fashion to identify genetic interactions and, using additional elementary knowledge of network structure, to infer signaling relations.

Determining the distribution of probes between different subcellular locations through automated unmixing of subcellular patterns

Many proteins or other biological macromolecules are localized to more than one subcellular structure. The fraction of a protein in different cellular compartments is often measured by colocalization with organelle-specific fluorescent markers, requiring availability of fluorescent probes for each compartment and acquisition of images for each in conjunction with the macromolecule of interest. Alternatively, tailored algorithms allow finding particular regions in images and quantifying the amount of fluorescence they contain. Unfortunately, this approach requires extensive hand-tuning of algorithms and is often cell type-dependent. Here we describe a machine-learning approach for estimating the amount of fluorescent signal in different subcellular compartments without hand tuning, requiring only the acquisition of separate training images of markers for each compartment. In testing on images of cells stained with mixtures of probes for different organelles, we achieved a 93% correlation between estimated and expected amounts of probes in each compartment. We also demonstrated that the method can be used to quantify drug-dependent protein translocations. The method enables automated and unbiased determination of the distributions of protein across cellular compartments, and will significantly improve imaging-based high-throughput assays and facilitate proteome-scale localization efforts.

Quantification and cluster analysis of actin cytoskeletal structures in plant cells: role of actin bundling in stomatal movement during diurnal cycles in Arabidopsis guard cells

Manual evaluation of cellular structures is a popular approach in cell biological studies. However, such approaches are laborious and are prone to error, especially when large quantities of image data need to be analyzed. Here, we introduce an image analysis framework that overcomes these limitations by semi-automatic quantification and clustering of cytoskeletal structures. In our framework, cytoskeletal orientation, bundling and density are quantified by measurement of newly-developed, robust metric parameters from microscopic images. Thereafter, the microscopic images are classified without supervision by clustering based on the metric patterns. Clustering allows us to collectively investigate the large number of cytoskeletal structure images without laborious inspection. Application of this framework to images of GFP-actin binding domain 2 (GFP-ABD2)-labeled actin cytoskeletons in Arabidopsis guard cells determined that microfilaments (MFs) are radially oriented and transiently bundled in the process of diurnal stomatal opening. The framework also revealed that the expression of mouse talin GFP-ABD (GFP-mTn) continuously induced MF bundling and suppressed the diurnal patterns of stomatal opening, suggesting that changes in the level of MF bundling are crucial for promoting stomatal opening. These results clearly demonstrate the utility of our image analysis framework.

Light microscopic analysis of mitochondrial heterogeneity in cell populations and within single cells

Heterogeneity in the shapes of individual multicellular organisms is a daily experience. Likewise, even a quick glance through the ocular of a light microscope reveals the morphological heterogeneities in genetically identical cultured cells, whereas heterogeneities on the level of the organelles are much less obvious. This short review focuses on intracellular heterogeneities at the example of the mitochondria and their analysis by fluorescence microscopy. The overall mitochondrial shape as well as mitochondrial dynamics can be studied by classical (fluorescence) light microscopy. However, with an organelle diameter generally close to the resolution limit of light, the heterogeneities within mitochondria cannot be resolved with conventional light microscopy. Therefore, we briefly discuss here the potential of subdiffraction light microscopy (nanoscopy) to study inner-mitochondrial heterogeneities.

Integrated analysis of receptor activation and downstream signaling with EXTassays

The ability to measure multiple cellular signaling events is essential to better understand the underlying complex biological processes that occur in living cells. Microarray-based technologies are now commonly used to study changes in transcription. This information, however, is not sufficient to understand the regulatory mechanisms that lead to gene expression changes. Here we present an approach to monitor signaling events upstream of gene expression. We coupled different reporter gene assays to unique expressed oligonucleotide tags (EXTs) that serve as identifiers and quantitative reporters. Multiple EXT reporters can be isolated as a pool and analyzed by hybridization to microarrays. To test the feasibility of our approach, we integrated complementation assays based on a protease from tobacco etch virus (TEV protease) and transcription factor activity profiling. Thereby, we simultaneously monitored Neuregulin-dependent mouse ErbB receptor tyrosine kinase dimerization, effector recruitment and downstream signaling.

A single fixation protocol for proteome-wide immunofluorescence localization studies

Immunofluorescence microscopy is a valuable tool for analyzing protein expression and localization at a subcellular level thus providing information regarding protein function, interaction partners and its role in cellular processes. When performing sample fixation, parameters such as difference in accessibility of proteins present in various cellular compartments as well as the chemical composition of the protein to be studied, needs to be taken into account. However, in systematic and proteome-wide efforts, a need exists for standard fixation protocol(s) that works well for the majority of all proteins independent of subcellular localization. Here, we report on a study with the goal to find a standardized protocol based on the analysis of 18 human proteins localized in 11 different organelles and subcellular structures. Six fixation protocols were tested based on either dehydration by alcohols (methanol, ethanol or iso-propanol) or cross-linking by paraformaldehyde followed by detergent permeabilization (Triton X-100 or saponin) in three human cell lines. Our results show that cross-linking is essential for proteome-wide localization studies and that cross-linking using paraformaldehyde followed by Triton X-100 permeabilization successfully can be used as a single fixation protocol for systematic studies.

High-content screening of primary neurons: ready for prime time

High-content screening (HCS), historically limited to drug-development companies, is now a powerful and affordable technology for academic researchers. Through automated routines, this technology acquires large datasets of fluorescence images depicting the functional states of thousands to millions of cells. Information on shapes, textures, intensities, and localizations is then used to create unique representations, or 'phenotypic signatures,' of each cell. These signatures quantify physiologic or diseased states, for example, dendritic arborization, drug response, or cell coping strategies. Live-cell imaging in HCS adds the ability to correlate cellular events at different points in time, thereby allowing sensitivities and observations not possible with fixed endpoint analysis. HCS with live-cell imaging therefore provides an unprecedented capability to detect spatiotemporal changes in cells and is particularly suited for time-dependent, stochastic processes such as neurodegenerative disorders.

Histopathological image analysis: a review

Over the past decade, dramatic increases in computational power and improvement in image analysis algorithms have allowed the development of powerful computer-assisted analytical approaches to radiological data. With the recent advent of whole slide digital scanners, tissue histopathology slides can now be digitized and stored in digital image form. Consequently, digitized tissue histopathology has now become amenable to the application of computerized image analysis and machine learning techniques. Analogous to the role of computer-assisted diagnosis (CAD) algorithms in medical imaging to complement the opinion of a radiologist, CAD algorithms have begun to be developed for disease detection, diagnosis, and prognosis prediction to complement the opinion of the pathologist. In this paper, we review the recent state of the art CAD technology for digitized histopathology. This paper also briefly describes the development and application of novel image analysis technology for a few specific histopathology related problems being pursued in the United States and Europe.

Histopathological image analysis: a review

Over the past decade, dramatic increases in computational power and improvement in image analysis algorithms have allowed the development of powerful computer-assisted analytical approaches to radiological data. With the recent advent of whole slide digital scanners, tissue histopathology slides can now be digitized and stored in digital image form. Consequently, digitized tissue histopathology has now become amenable to the application of computerized image analysis and machine learning techniques. Analogous to the role of computer-assisted diagnosis (CAD) algorithms in medical imaging to complement the opinion of a radiologist, CAD algorithms have begun to be developed for disease detection, diagnosis, and prognosis prediction to complement the opinion of the pathologist. In this paper, we review the recent state of the art CAD technology for digitized histopathology. This paper also briefly describes the development and application of novel image analysis technology for a few specific histopathology related problems being pursued in the United States and Europe.

Human cell chips: adapting DNA microarray spotting technology to cell-based imaging assays

Here we describe human spotted cell chips, a technology for determining cellular state across arrays of cells subjected to chemical or genetic perturbation. Cells are grown and treated under standard tissue culture conditions before being fixed and printed onto replicate glass slides, effectively decoupling the experimental conditions from the assay technique. Each slide is then probed using immunofluorescence or other optical reporter and assayed by automated microscopy. We show potential applications of the cell chip by assaying HeLa and A549 samples for changes in target protein abundance (of the dsRNA-activated protein kinase PKR), subcellular localization (nuclear translocation of NFκB) and activation state (phosphorylation of STAT1 and of the p38 and JNK stress kinases) in response to treatment by several chemical effectors (anisomycin, TNFα, and interferon), and we demonstrate scalability by printing a chip with ∼4,700 discrete samples of HeLa cells. Coupling this technology to high-throughput methods for culturing and treating cell lines could enable researchers to examine the impact of exogenous effectors on the same population of experimentally treated cells across multiple reporter targets potentially representing a variety of molecular systems, thus producing a highly multiplexed dataset with minimized experimental variance and at reduced reagent cost compared to alternative techniques. The ability to prepare and store chips also allows researchers to follow up on observations gleaned from initial screens with maximal repeatability.

Construction of functional interaction networks through consensus localization predictions of the human proteome

Characterizing the subcellular localization of a protein provides a key clue for understanding protein function. However, different protein localization prediction programs often deliver conflicting results regarding the localization of the same protein. As the number of available localization prediction programs continues to grow, there is a need for a consensus prediction approach. To address this need, we developed a consensus localization prediction method called ConLoc based on a large-scale, systematic integration of 13 available programs that make predictions for five major subcellular localizations (cytosol, extracellular, mitochondria, nucleus, and plasma membrane). The ability of ConLoc to accurately predict protein localization was substantially better than existing programs. Using ConLoc prediction, we built a localization-guided functional interaction network of the human proteome and mapped known disease associations within this network. We found a high degree of shared disease associations among functionally interacting proteins that are localized to the same cellular compartment. Thus, the use of consensus localization prediction, such as ConLoc, is a new approach for the identification of novel disease associated genes.

Robust signal detection in 3D fluorescence microscopy

Robust detection and localization of biomolecules inside cells is of great importance to better understand the functions related to them. Fluorescence microscopy and specific staining methods make biomolecules appear as point-like signals on image data, often acquired in 3D. Visual detection of such point-like signals can be time consuming and problematic if the 3D images are large, containing many, sometimes overlapping, signals. This sets a demand for robust automated methods for accurate detection of signals in 3D fluorescence microscopy. We propose a new 3D point-source signal detection method that is based on Fourier series. The method consists of two parts, a detector, which is a cosine filter to enhance the point-like signals, and a verifier, which is a sine filter to validate the result from the detector. Compared to conventional methods, our method shows better robustness to noise and good ability to resolve signals that are spatially close. Tests on image data show that the method has equivalent accuracy in signal detection in comparison to visual detection by experts. The proposed method can be used as an efficient point-like signal detection tool for various types of biological 3D image data.

Automated quantification of DNA demethylation effects in cells via 3D mapping of nuclear signatures and population homogeneity assessment

Today's advanced microscopic imaging applies to the preclinical stages of drug discovery that employ high-throughput and high-content three-dimensional (3D) analysis of cells to more efficiently screen candidate compounds. Drug efficacy can be assessed by measuring response homogeneity to treatment within a cell population. In this study, topologically quantified nuclear patterns of methylated cytosine and global nuclear DNA are utilized as signatures of cellular response to the treatment of cultured cells with the demethylating anti-cancer agents: 5-azacytidine (5-AZA) and octreotide (OCT). Mouse pituitary folliculostellate TtT-GF cells treated with 5-AZA and OCT for 48 hours, and untreated populations, were studied by immunofluorescence with a specific antibody against 5-methylcytosine (MeC), and 4,6-diamidino-2-phenylindole (DAPI) for delineation of methylated sites and global DNA in nuclei (n = 163). Cell images were processed utilizing an automated 3D analysis software that we developed by combining seeded watershed segmentation to extract nuclear shells with measurements of Kullback-Leibler's (K-L) divergence to analyze cell population homogeneity in the relative nuclear distribution patterns of MeC versus DAPI stained sites. Each cell was assigned to one of the four classes: similar, likely similar, unlikely similar, and dissimilar. Evaluation of the different cell groups revealed a significantly higher number of cells with similar or likely similar MeC/DAPI patterns among untreated cells (approximately 100%), 5-AZA-treated cells (90%), and a lower degree of same type of cells (64%) in the OCT-treated population. The latter group contained (28%) of unlikely similar or dissimilar (7%) cells. Our approach was successful in the assessment of cellular behavior relevant to the biological impact of the applied drugs, i.e., the reorganization of MeC/DAPI distribution by demethylation. In a comparison with other metrics, K-L divergence has proven to be a more valuable and robust tool for categorization of individual cells within a population, with potential applications in epigenetic drug screening.

Intelligent acquisition and learning of fluorescence microscope data models

We propose a mathematical framework and algorithms both to build accurate models of fluorescence microscope time series, as well as to design intelligent acquisition systems based on these models. Model building allows the information contained in the 2-D and 3-D time series to be presented in a more useful and concise form than the raw image data. This is particularly relevant as the trend in biology tends more and more towards high-throughput applications, and the resulting increase in the amount of acquired image data makes visual inspection impractical. The intelligent acquisition system uses an active learning approach to choose the acquisition regions that let us build our model most efficiently, resulting in a shorter acquisition time, as well as a reduction of the amount of photobleaching and phototoxicity incurred during acquisition. We validate our methodology by modeling object motion within a cell. For intelligent acquisition, we propose a set of algorithms to evaluate the information contained in a given acquisition region, as well as the costs associated with acquiring this region in terms of the resulting photobleaching and phototoxicity and the amount of time taken for acquisition. We use these algorithms to determine an acquisition strategy: where and when to acquire, as well as when to stop acquiring. Results, both on synthetic as well as real data, demonstrate accurate model building and large efficiency gains during acquisition.

A detailed analysis of 3D subcellular signal localization

Detection and localization of fluorescent signals in relation to other subcellular structures is an important task in various biological studies. Many methods for analysis of fluorescence microscopy image data are limited to 2D. As cells are in fact 3D structures, there is a growing need for robust methods for analysis of 3D data. This article presents an approach for detecting point-like fluorescent signals and analyzing their subnuclear position. Cell nuclei are delineated using marker-controlled (seeded) 3D watershed segmentation. User-defined object and background seeds are given as input, and gradient information defines merging and splitting criteria. Point-like signals are detected using a modified stable wave detector and localized in relation to the nuclear membrane using distance shells. The method was applied to a set of biological data studying the localization of Smad2-Smad4 protein complexes in relation to the nuclear membrane. Smad complexes appear as early as 1 min after stimulation while the highest signal concentration is observed 45 min after stimulation, followed by a concentration decrease. The robust 3D signal detection and concentration measures obtained using the proposed method agree with previous observations while also revealing new information regarding the complex formation.

Assays for mitotic chromosome condensation in live yeast and mammalian cells

The dynamic reorganization of chromatin into rigid and compact mitotic chromosomes is of fundamental importance for faithful chromosome segregation. Owing to the difficulty of investigating this process under physiological conditions, the exact morphological transitions and the molecular machinery driving chromosome condensation remain poorly defined. Here, we review how imaging-based methods can be used to quantitate chromosome condensation in vivo, focusing on yeast and animal tissue culture cells as widely used model systems. We discuss approaches how to address structural dynamics of condensing chromosomes and chromosome segments, as well as to probe for mechanical properties of mitotic chromosomes. Application of such methods to systematic perturbation studies will provide a means to reveal the molecular networks underlying the regulation of mitotic chromosome condensation.

Detection of protein-protein interactions through vesicle targeting

The detection of protein-protein interactions through two-hybrid assays has revolutionized our understanding of biology. The remarkable impact of two-hybrid assay platforms derives from their speed, simplicity, and broad applicability. Yet for many organisms, the need to express test proteins in Saccharomyces cerevisiae or Escherichia coli presents a substantial barrier because variations in codon specificity or bias may result in aberrant protein expression. In particular, nonstandard genetic codes are characteristic of several eukaryotic pathogens, for which there are currently no genetically based systems for detection of protein-protein interactions. We have developed a protein-protein interaction assay that is carried out in native host cells by using GFP as the only foreign protein moiety, thus circumventing these problems. We show that interaction can be detected between two protein pairs in both the model yeast S. cerevisiae and the fungal pathogen Candida albicans. We use computational analysis of microscopic images to provide a quantitative and automated assessment of confidence.

Challenges of automated screening and differentiation of non-organ specific autoantibodies on HEp-2 cells

Analysis of autoantibodies (AAB) by indirect immunofluorescence (IIF) remains the hallmark of diagnosing autoimmune diseases despite the introduction of multiplex techniques. Non-organ specific AAB are screened in routine diagnostics by IIF on HEp-2 cells. However, IIF results vary due to objective (e.g., cell fixation) and subjective factors (e.g., expert knowledge). Therefore, inter- and intralaboratory variance is relatively high. Standardisation of AAB testing by IIF remains a critical issue in and between routine laboratories and may be improved by automated interpretation systems. An overview of existing interpretation techniques will be given taking into account own data of the first fully automated reading system AKLIDES. The novel system provides fully automated reading of IIF images and software algorithms for the mathematical description of IIF AAB patterns. It can be used for screening and preclassification of non-organ specific AAB in routine diagnostics regarding systemic autoimmune and autoimmune liver diseases. Furthermore, this system paves the way for economic data processing of cell-based IIF assays and can contribute to the reduction of interlaboratory variance of AAB testing. More sophisticated pattern recognition algorithms and novel calibration systems will improve standardised quantifications of IIF image interpretation.

Automated Analysis of Human Protein Atlas Immunofluorescence Images

The Human Protein Atlas is a rich source of location proteomics data. In this work, we present an automated approach for processing and classifying major subcellular patterns in the Atlas images. We demonstrate that two different classification frameworks (support vector machine and random forest) are effective at determining subcellular locations; we can analyze over 3500 Atlas images with a high degree of accuracy, up to 87.5% for all of the samples and 98.5% when only considering samples in whose classification assignments we are most confident. Moreover, the features obtained in both of these frameworks are observed to be highly consistent and generalizable. Additionally, we observe that the features relating the proteins to cell markers are especially important in automated learning approaches.

An incremental approach to automated protein localisation

Background

The subcellular localisation of proteins in intact living cells is an important means for gaining information about protein functions. Even dynamic processes can be captured, which can barely be predicted based on amino acid sequences. Besides increasing our knowledge about intracellular processes, this information facilitates the development of innovative therapies and new diagnostic methods. In order to perform such a localisation, the proteins under analysis are usually fused with a fluorescent protein. So, they can be observed by means of a fluorescence microscope and analysed. In recent years, several automated methods have been proposed for performing such analyses. Here, two different types of approaches can be distinguished: techniques which enable the recognition of a fixed set of protein locations and methods that identify new ones. To our knowledge, a combination of both approaches – i.e. a technique, which enables supervised learning using a known set of protein locations and is able to identify and incorporate new protein locations afterwards – has not been presented yet. Furthermore, associated problems, e.g. the recognition of cells to be analysed, have usually been neglected.

Results

We introduce a novel approach to automated protein localisation in living cells. In contrast to well-known techniques, the protein localisation technique presented in this article aims at combining the two types of approaches described above: After an automatic identification of unknown protein locations, a potential user is enabled to incorporate them into the pre-trained system. An incremental neural network allows the classification of a fixed set of protein location as well as the detection, clustering and incorporation of additional patterns that occur during an experiment. Here, the proposed technique achieves promising results with respect to both tasks. In addition, the protein localisation procedure has been adapted to an existing cell recognition approach. Therefore, it is especially well-suited for high-throughput investigations where user interactions have to be avoided.

Conclusion

We have shown that several aspects required for developing an automatic protein localisation technique – namely the recognition of cells, the classification of protein distribution patterns into a set of learnt protein locations, and the detection and learning of new locations – can be combined successfully. So, the proposed method constitutes a crucial step to render image-based protein localisation techniques amenable to large-scale experiments.

Identification of rab20 as a potential regulator of connexin 43 trafficking

Connexin oligomerization and trafficking are regulated processes. To identify proteins that control connexin 43 (Cx43), a screen was designed using HeLa cells expressing a Cx43 construct with di-lysine endoplasmic reticulum (ER)-retention/retrieval motif, Cx43-HKKSL. At moderate levels of expression, Cx43-HKKSL is retained in the ER as monomers; however, Cx43-HKKSL stably overexpressed by HeLa cells localizes to the perinuclear region and oligomerizes. HeLa/Cx43-HKKSL overexpressors were transiently transfected with pooled clones from a human kidney cDNA library and used immunofluorescence microscopy to identify cDNAs that enabled overexpressed Cx43-HKKSL to convert from a perinuclear to ER localization pattern. Using this approach, a small molecular weight GTPase, rab20, was identified as a candidate protein with the ability to regulate Cx43 trafficking. Enhanced green fluorescent protein (EGFP)-tagged rab20 showed a predominantly perinuclear and ER localization pattern and caused wild-type Cx43 to be retained inside the cell. By contrast, mutant EGFP-rab20T19N, which lacks the ability to bind GTP, had no effect on Cx43. These results suggest Cx43 is transported through an intracellular compartment regulated by rab20 along the secretory pathway.

Bioimage informatics: a new area of engineering biology

In recent years, the deluge of complicated molecular and cellular microscopic images creates compelling challenges for the image computing community. There has been an increasing focus on developing novel image processing, data mining, database and visualization techniques to extract, compare, search and manage the biological knowledge in these data-intensive problems. This emerging new area of bioinformatics can be called ‘bioimage informatics’. This article reviews the advances of this field from several aspects, including applications, key techniques, available tools and resources. Application examples such as high-throughput/high-content phenotyping and atlas building for model organisms demonstrate the importance of bioimage informatics. The essential techniques to the success of these applications, such as bioimage feature identification, segmentation and tracking, registration, annotation, mining, image data management and visualization, are further summarized, along with a brief overview of the available bioimage databases, analysis tools and other resources.

Contact:pengh@janelia.hhmi.org

Supplementary information:Supplementary data are available at Bioinformatics online.

Automated recognition system to classify subcellular protein localizations in images of different cell lines acquired by different imaging systems

Systemic analysis of subcellular protein localization (location proteomics) provides clues for understanding gene functions and physiological condition of the cells. However, recognition of cell images of subcellular structures highly depends on experience and becomes the rate-limiting step when classifying subcellular protein localization. Several research groups have extracted specific numerical features for the recognition of subcellular protein localization, but these recognition systems are restricted to images of single particular cell line acquired by one specific imaging system and not applied to recognize a range of cell image sources. In this study, we establish a single system for automated subcellular structure recognition to identify cell images from various sources. Two different sources of cell images, 317 Vero (http://gfp-cdna.embl.de) and 875 CHO cell images of subcellular structures, were used to train and test the system. When the system was trained by a single source of images, the recognition rate is high and specific to the trained source. The system trained by the CHO cell images gave high average recognition accuracy for CHO cells of 96%, but this was reduced to 46% with Vero images. When we trained the system using a mixture of CHO and Vero cell images, an average accuracy of recognition reached 86.6% for both CHO and Vero cell images. The system can reject images with low confidence and identify the cell images correctly recognized to avoid manual reconfirmation. In summary, we have established a single system that can recognize subcellular protein localizations from two different sources for location-proteomic studies. studies.

High throughput microscopy: from raw images to discoveries

Technological advances in automated microscopy now allow rapid acquisition of many images without human intervention, images that can be used for large-scale screens. The main challenge in such screens is the conversion of the raw images into interpretable information and hence discoveries. This post-acquisition component of image-based screens requires computational steps to identify cells, choose the cells of interest, assess their phenotype, and identify statistically significant 'hits'. Designing such an analysis pipeline requires careful consideration of the necessary hardware and software components, image analysis, statistical analysis and data presentation tools. Given the increasing availability of such hardware and software, these types of experiments have come within the reach of individual labs, heralding many interesting new ways of acquiring biological knowledge.

Automated Proteome-Wide Determination of Subcellular Location Using High Throughput Microscopy

A major source of information for identifying subcellular location on a proteome-wide basis will be imaging of tagged proteins in living cells using fluorescence microscopy. We have previously developed automated systems to interpret images from such experiments and demonstrated that they can perform as well or better than visual inspection. Recent work demonstrates that these methods can be applied to large collections of images from sources as diverse as yeast expressing GFP-tagged proteins and human tissues imaged by immunocytochemistry. A distinct but related task is learning what location patterns exist. We have demonstrated clustering of mouse proteins into subcellular location families that share a statistically indistinguishable pattern. To communicate each pattern, we have developed approaches to learning generative models of subcellular patterns. Integration of high-throughput microscopy and automated model building with cell modeling systems will permit accurate, well-structured information on subcellular location to be incorporated into systems biology efforts.

AUTOMATED COMPARISON OF PROTEIN SUBCELLULAR LOCATION PATTERNS BETWEEN IMAGES OF NORMAL AND CANCEROUS TISSUES

Early cancer diagnosis and evaluation of cancer progression during treatment are two important factors for clinical therapy. In this study we propose a novel approach which automatically compares the subcellular location of proteins between normal and cancerous tissues in order to identify proteins whose distribution is modified by oncogenesis. This study analyzes 258 proteins in 14 different cancer tissues and their corresponding normal tissues using images provided by the tissue microarray collection of the Human Protein Atlas. Using texture features automatically extracted from the tissue images, 14 machine classifiers were trained to recognize the patterns of eight major organelles in each tissue. For each tissue-protein combination, the results of the classifier for normal and cancerous tissues were compared. Eleven proteins were identified as showing differences in location; these proteins may have potential as biomarkers.

A framework for the automated analysis of subcellular patterns in human protein atlas images

The systematic study of subcellular location patterns is required to fully characterize the human proteome, as subcellular location provides critical context necessary for understanding a protein's function. The analysis of tens of thousands of expressed proteins for the many cell types and cellular conditions under which they may be found creates a need for automated subcellular pattern analysis. We therefore describe the application of automated methods, previously developed and validated by our laboratory on fluorescence micrographs of cultured cell lines, to analyze subcellular patterns in tissue images from the Human Protein Atlas. The Atlas currently contains images of over 3000 protein patterns in various human tissues obtained using immunohistochemistry. We chose a 16 protein subset from the Atlas that reflects the major classes of subcellular location. We then separated DNA and protein staining in the images, extracted various features from each image, and trained a support vector machine classifier to recognize the protein patterns. Our results show that our system can distinguish the patterns with 83% accuracy in 45 different tissues, and when only the most confident classifications are considered, this rises to 97%. These results are encouraging given that the tissues contain many different cell types organized in different manners, and that the Atlas images are of moderate resolution. The approach described is an important starting point for automatically assigning subcellular locations on a proteome-wide basis for collections of tissue images such as the Atlas.

Subcellular localization of intracellular human proteins by construction of tagged fusion proteins and transient expression in COS-7 Cells

Identifying the subcellular compartment of a protein is an important step toward assigning protein function. Starting with a clone containing the open reading frame (ORF) of interest, it is possible to attach a variety of short amino acid tags or fluorescent proteins and detect the location of the protein, after transfection into a cell line, using fluorescent microscopy. By collecting data from various expression clone constructs, using a range of cell lines and double labeling with cellular compartment markers, a picture of the localization of a gene can be built up. This chapter describes how to obtain the ORF clone for your gene of interest, clone it into your choice of mammalian expression vector or vectors, transiently transfect for visualization, and where to get started when interpreting the results.

Visualizing and clustering high throughput sub-cellular localization imaging

Background

The expansion of automatic imaging technologies has created a need to be able to efficiently compare and review large sets of image data. To enable comparisons of image data between samples we need to define the normal variation within distinct images of the same sample. Even with tightly controlled experimental conditions, protein expression can vary widely between cells, and because of the difficulty in viewing and comparing large image sets this might not be observed. Here we introduce a novel methodology, iCluster, for visualizing, clustering and comparing large sub-cellular localization image sets. For each member of an image set, iCluster generates statistics that have been found to be useful in distinguishing sub-cellular localization. The statistics are mapped into two or three dimensions such as to preserve distances between the statistics vectors. The complete image set is then visualized in two or three dimensions using the coordinates so determined. The result is images that are statistically similar are spatially close in the visualization allowing for easy comparison of images that are similar and distinguishment of dissimilar images into distinct clusters.

Results

The methodology was tested on a set of 502 previously published images containing 10 known sub-cellular localizations. The clustering of images of like type was evaluated both by examining the classes of nearest neighbors to each image and by visual inspection. In three dimensions, 3-neighbor classification accuracy was 83.2%. Visually, each class clustered well with the majority of classes localizing to distinct regions of the space. In two dimensions, 3-neighbor classification accuracy was 68.9%, though visually clustering into classes could be readily discerned. Computational expense was found to be relatively low, and sets of up to 1400 images visualized and interacted with in real time.

Conclusion

The feasibility of automated spatial layout to allow comparison and discrimination of high throughput sub-cellular imaging has been demonstrated. There are many potential applications such as image database curation, semi-automated interactive classification, outlier detection and reference image comparison. By allowing the observation of the full range of imaging data available using modern microscopes these methods will provide an invaluable tool for cell biologists.

Toward a confocal subcellular atlas of the human proteome

Information on protein localization on the subcellular level is important to map and characterize the proteome and to better understand cellular functions of proteins. Here we report on a pilot study of 466 proteins in three human cell lines aimed to allow large scale confocal microscopy analysis using protein-specific antibodies. Approximately 3000 high resolution images were generated, and more than 80% of the analyzed proteins could be classified in one or multiple subcellular compartment(s). The localizations of the proteins showed, in many cases, good agreement with the Gene Ontology localization prediction model. This is the first large scale antibody-based study to localize proteins into subcellular compartments using antibodies and confocal microscopy. The results suggest that this approach might be a valuable tool in conjunction with predictive models for protein localization.

Maturation of the mammalian secretome

A recent use of quantitative proteomics to determine the constituents of the endoplasmic reticulum and Golgi complex is discussed in the light of other available methodologies for cataloging the proteins associated with the mammalian secretory pathway.