Journal of Neuroscience Methods 225 (2014) 1–12 Contents lists available at ScienceDirect Journal of Neuroscience Methods jo urnal homepage: www.elsevier.com/locate/jneumeth Extraction of protein profiles from primary neurons using active ଝ contour models and wavelets a,∗ b a a Danny Misiak , Stefan Posch , Marcell Lederer , Claudia Reinke , a b Stefan Hüttelmaier , Birgit Möller a Institute of Molecular Medicine, Martin Luther University Halle-Wittenberg, Heinrich-Damerow-Str. 1, 06120 Halle, Germany b Institute of Computer Science, Martin Luther University Halle-Wittenberg, Von-Seckendorff-Platz 1, 06099 Halle, Germany h i g h l i g h t s • Extraction of neuron areas using active contours with a new distance based energy. • Location of structural neuron parts by a wavelet based approach. • Automatic extraction of spatial and quantitative data about distributions of proteins. • Extraction of various morphological parameters, in a fully automated manner. • Extraction of a distinctive profile for the Zipcode binding protein (ZBP1/IGF2BP1). a r t i c l e i n f o a b s t r a c t Article history: The function of complex networks in the nervous system relies on the proper formation of neuronal Received 26 October 2013 contacts and their remodeling. To decipher the molecular mechanisms underlying these processes, it is Received in revised form essential to establish unbiased automated tools allowing the correlation of neurite morphology and the 18 December 2013 subcellular distribution of molecules by quantitative means. Accepted 19 December 2013 We developed NeuronAnalyzer2D, a plugin for ImageJ, which allows the extraction of neuronal cell morphologies from two dimensional high resolution images, and in particular their correlation with Keywords: protein profiles determined by indirect immunostaining of primary neurons. The prominent feature of Active contours our approach is the ability to extract subcellular distributions of distinct biomolecules along neurites. Fluorescence microscopy To extract the complete areas of neurons, required for this analysis, we employ active contours with High-content analysis Neuron morphology a new distance based energy. For locating the structural parts of neurons and various morphological Protein distribution parameters we adopt a wavelet based approach. The presented approach is able to extract distinctive Segmentation profiles of several proteins and reports detailed morphology measurements on neurites. Wavelets We compare the detected neurons from NeuronAnalyzer2D with those obtained by NeuriteTracer and Vaa3D-Neuron, two popular tools for automatic neurite tracing. The distinctive profiles extracted for several proteins, for example, of the mRNA binding protein ZBP1, and a comparative evaluation of the neuron segmentation results proves the high quality of the quantitative data and proves its practical utility for biomedical analyses. © 2014 The Authors. Published by Elsevier B.V. All rights reserved. 1. Introduction Neurons are essential components of higher organisms. They form simple up to extraordinary complex networks via their axonal ଝ and dendritic cell extensions. During early development, neurons This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike License, which permits non- extend elongated protrusions of cytoplasm, called neurites, which commercial use, distribution, and reproduction in any medium, provided the finally differentiate into functional axons or dendrites. The mor- original author and source are credited. ∗ phology of single neurites essentially facilitates the function of Corresponding author at: Institute of Molecular Medicine, Sect. for Molecular neuronal networks, and defects in neurite architecture frequently Cell Biology, Martin Luther University Halle-Wittenberg, Heinrich-Damerow-Str. 1, correlate with severe brain disorders and neurological defects. 06120 Halle, Germany. Phone: +49 345 5522862; fax: +49 345 5522894. E-mail address: [email protected] (D. Misiak). Thus, the analysis of neurite morphology and underlying molecular 0165-0270/$ – see front matter © 2014 The Authors. Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jneumeth.2013.12.009 2 D. Misiak et al. / Journal of Neuroscience Methods 225 (2014) 1–12 regulatory mechanisms is important to understand the basis of NeuriteAnalyzer2D is able to extract distinctive profiles of both neuronal function and neurological diseases. several proteins, demonstrating its ability to extract fluorescence During development or in response to environmental con- intensity distributions of labeled molecules along the localized neu- straints, neurons like most cells modulate their gene expression rites. In particular, a distinctive profile for the Zipcode binding program resulting in the synthesis and subsequent subcellular sort- protein (ZBP1/IGF2BP1) was extracted for the first time. In addition ing of mRNAs or the final product, the proteins. These underlying to basic measurements of neuronal morphology, like centerlines, regulatory mechanisms modulate intrinsic cell functions, control end points or branching patterns, our approach reports detailed cell migration, and direct the morphology of cellular protrusions morphology measurements on neurites. For example, average including neurites. To understand the function of these regulatory neurite width, number of filopodia-like protrusions (Mattila and processes, it is essential to reveal the localization of mRNAs and pro- Lappalainen, 2008) and size of growth cone areas are extracted. teins in relation to cell morphology by quantitative means. To this The presented method to extract profiles of labeled molecules is end, fully automated quantitative evaluation of molecular profiles validated by comparing the profiles of F-actin and ˛-tubulin to their remains a major obstacle. published distributions in neuronal cells. To validate our fully auto- A fully automated quantitative evaluation of molecular profiles mated extraction of neuron areas we compare the detected neurons requires an unbiased assessment of images acquired by high- from our approach and two other automatic tracing approaches to resolution fluorescence microscopy subsequent to the specific neurons manually analyzed by a biomedical expert. labeling of molecules by fluorescent dyes. Although these proce- dures are well established, the quantitative and high-throughput 2. Related work evaluation of acquired images remains rudimentary and currently the ability to extract and analyze the subcellular distributions of The segmentation of elongated tube-like structures, for exam- distinct biomolecules in a high-throughput manner does not exist. ple, blood vessels (Kirbas and Quek, 2004; Läthén et al., 2010), The extraction of distinctive molecular profiles based on flu- plant roots (Erz et al., 2005) or neuronal cells (Capowski, 1983; orescence intensities requires an exact segmentation of neurons. Ramm et al., 2003), has been a challenge in computer vision for a The vast majority of published neurite detection approaches are long period of time. Efforts towards computer-aided segmentation limited to tracing the centerlines of neurites and evaluating the of neurons and the analysis of neuronal morphology reach back global cell morphology, without a complete segmentation of cell 45 years (Glaser and Van Der Loos, 1965). Numerous findings in areas. Moreover, various approaches require manual or semiau- the fields of neurite tracing, quantitative methods of analysis and tomatic segmentation of neurites and, thus, still require manual morphology extraction have been investigated. intervention. However, to the authors’ knowledge, no tools for spatial and The goal of our work is to spatially and quantitatively assess quantitative assessment of distinct biomolecules along neurites are distinct biomolecules along neurites. To achieve this, molecu- currently available to correlate biomolecular and morphological lar profiles are extracted upon fluorescence labeling of relevant changes in neurons in a fully automated manner. molecules, for example, proteins. Additionally, morphological and Schmitz et al. present an automatic image analysis program biomolecular changes in neurons induced by stress or influence of called SynD (Synapse Detector) to calculate the synaptic recruit- molecular stimuli are to be characterized in a fully automated man- ment of proteins of interest (Schmitz et al., 2011). The calculation ner by correlating the neuronal cell morphology with the extracted of the recruitment is only given as the ratio between synaptic and protein profiles. This requires an automatic extraction of complete somatic intensity. Furthermore, the program reports dendritic and neuron and neurite areas, as well as an automatic extraction of synaptic characteristics, like dendrite length as well as number of protein profiles along the identified neurites. To detect the neuron branches and synapses. To measure the dendritic branching the areas, we apply energy based active contour models (Kass et al., Sholl analysis is applied (Sholl, 1953). This analysis is widely used 1988), where we propose a new distance based energy to detect but offers a limited sensitivity to detect differences between groups low contrast object boundaries. To identify the neurites within the of neurons and disregards orientation as well as topology of the detected neuronal cells the individual structural parts of the neu- dendritic tree (Uylings and van Pelt, 2002).