Caveolae and scaffold detection from single molecule localization microscopy data using deep learning

Caveolae are plasma membrane invaginations whose formation requires caveolin-1 (Cav1), the adaptor protein polymerase I, and the transcript release factor (PTRF or CAVIN1). Caveolae have an important role in cell functioning, signaling, and disease. In the absence of CAVIN1/PTRF, Cav1 forms non-caveolar membrane domains called scaffolds. In this work, we train machine learning models to automatically distinguish between caveolae and scaffolds from single molecule localization microscopy (SMLM) data. We apply machine learning algorithms to discriminate biological structures from SMLM data. Our work is the first that is leveraging machine learning approaches (including deep learning models) to automatically identifying biological structures from SMLM data. In particular, we develop and compare three binary classification methods to identify whether or not a given 3D cluster of Cav1 proteins is a caveolae. The first uses a random forest classifier applied to 28 hand-crafted/designed features, the second uses a convolutional neural net (CNN) applied to a projection of the point clouds onto three planes, and the third uses a PointNet model, a recent development that can directly take point clouds as its input. We validate our methods on a dataset of super-resolution microscopy images of PC3 prostate cancer cells labeled for Cav1. Specifically, we have images from two cell populations: 10 PC3 and 10 CAVIN1/PTRF-transfected PC3 cells (PC3-PTRF cells) that form caveolae. We obtained a balanced set of 1714 different cellular structures. Our results show that both the random forest on hand-designed features and the deep learning approach achieve high accuracy in distinguishing the intrinsic features of the caveolae and non-caveolae biological structures. More specifically, both random forest and deep CNN classifiers achieve classification accuracy reaching 94% on our test set, while the PointNet model only reached 83% accuracy. We also discuss the pros and cons of the different approaches.

[1]  Subhransu Maji,et al.  Multi-view Convolutional Neural Networks for 3D Shape Recognition , 2015, 2015 IEEE International Conference on Computer Vision (ICCV).

[2]  S. Hell,et al.  Fluorescence nanoscopy by ground-state depletion and single-molecule return , 2008, Nature Methods.

[3]  Samuel T. Hess,et al.  Nanoscale Imaging of Caveolin-1 Membrane Domains In Vivo , 2015, PloS one.

[4]  D. Baddeley.  Detecting nano-scale protein clustering , 2015, Nature Methods.

[5]  Leonidas J. Guibas,et al.  PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[6]  Astrid Magenau,et al.  Sub-resolution lipid domains exist in the plasma membrane and regulate protein diffusion and distribution , 2012, Nature Communications.

[7]  Benjamin Recht,et al.  DeepLoco: Fast 3D Localization Microscopy Using Neural Networks , 2018, bioRxiv.

[8]  Mark E. J. Newman,et al.  The Structure and Function of Complex Networks , 2003, SIAM Rev..

[9]  Ghassan Hamarneh,et al.  Identification of caveolin-1 domain signatures via machine learning and graphlet analysis of single-molecule super-resolution data , 2019, Bioinform..

[10]  I. Nabi,et al.  Galectin-3 Overrides PTRF/Cavin-1 Reduction of PC3 Prostate Cancer Cell Migration , 2015, PloS one.

[11]  Dylan M Owen,et al.  Superresolution Microscopy Reveals Nanometer-Scale Reorganization of Inhibitory Natural Killer Cell Receptors upon Activation of NKG2D , 2013, Science Signaling.

[12]  Ghassan Hamarneh,et al.  Super Resolution Network Analysis Defines the Molecular Architecture of Caveolae and Caveolin-1 Scaffolds , 2018, Scientific Reports.

[13]  Geoffrey E. Hinton,et al.  Visualizing Data using t-SNE , 2008 .

[14]  Loris Nanni,et al.  Handcrafted vs. non-handcrafted features for computer vision classification , 2017, Pattern Recognit..

[15]  Mark Bates,et al.  Evaluation of fluorophores for optimal performance in localization-based super-resolution imaging , 2011, Nature Methods.

[16]  Tomer Michaeli,et al.  Deep-STORM: super-resolution single-molecule microscopy by deep learning , 2018, 1801.09631.

[17]  Z. Gong,et al.  Super Resolution Microscopy Reveals that Caveolin-1 Is Required for Spatial Organization of CRFB1 and Subsequent Antiviral Signaling in Zebrafish , 2013, PloS one.

[18]  Heung-Il Suk,et al.  Deep Learning in Medical Image Analysis. , 2017, Annual review of biomedical engineering.

[19]  Eilon Sherman.  Resolving protein interactions and organization downstream the T cell antigen receptor using single-molecule localization microscopy: a review , 2016 .

[20]  Leo Breiman,et al.  Random Forests , 2001, Machine Learning.

[21]  J. Dennis,et al.  JCB_200811059 381..386 , 2009 .

[22]  Ismail M. Khater,et al.  Super-resolution modularity analysis shows polyhedral caveolin-1 oligomers combine to form scaffolds and caveolae , 2018, Scientific Reports.

[23]  Robert G. Parton,et al.  Caveolae as plasma membrane sensors, protectors and organizers , 2013, Nature Reviews Molecular Cell Biology.

[24]  J. Dennis,et al.  Plasma membrane domain organization regulates EGFR signaling in tumor cells , 2007, The Journal of cell biology.

[25]  Daniel Choquet,et al.  SR-Tesseler: a method to segment and quantify localization-based super-resolution microscopy data , 2015, Nature Methods.

[26]  Matthew B Stone,et al.  Super-Resolution Microscopy: Shedding Light on the Cellular Plasma Membrane. , 2017, Chemical reviews.

[27]  Geoffrey E. Hinton,et al.  Deep Learning , 2015, Nature.

[28]  D. Craik,et al.  PTRF/cavin-1 neutralizes non-caveolar caveolin-1 microdomains in prostate cancer , 2013, Oncogene.

[29]  David J. Williamson,et al.  Bayesian cluster identification in single-molecule localization microscopy data , 2015, Nature Methods.

[30]  Carlos Guestrin,et al.  "Why Should I Trust You?": Explaining the Predictions of Any Classifier , 2016, ArXiv.

[31]  A. Mochizuki,et al.  Measurement of caveolin-1 densities in the cell membrane for quantification of caveolar deformation after exposure to hypotonic membrane tension , 2017, Scientific Reports.

[32]  Bryan Heit,et al.  MIiSR: Molecular Interactions in Super-Resolution Imaging Enables the Analysis of Protein Interactions, Dynamics and Formation of Multi-protein Structures , 2015, PLoS Comput. Biol..

[33]  M. Kirkham,et al.  PTRF-Cavin, a Conserved Cytoplasmic Protein Required for Caveola Formation and Function , 2008, Cell.

[34]  Mark M Davis,et al.  TCR and Lat are expressed on separate protein islands on T cell membranes and concatenate during activation , 2010, Nature Immunology.

[35]  M. Parat,et al.  Caveola-forming proteins caveolin-1 and PTRF in prostate cancer , 2013, Nature Reviews Urology.

[36]  Astrid Magenau,et al.  PALM imaging and cluster analysis of protein heterogeneity at the cell surface , 2010, Journal of biophotonics.

[37]  Mark Bates,et al.  Three-Dimensional Super-Resolution Imaging by Stochastic Optical Reconstruction Microscopy , 2008, Science.

[38]  Katharina Gaus,et al.  Conformational states of the kinase Lck regulate clustering in early T cell signaling , 2012, Nature Immunology.

[39]  Christophe Zimmer,et al.  Deep learning massively accelerates super-resolution localization microscopy , 2018, Nature Biotechnology.

[40]  Juan Carlos Fernández,et al.  Multiobjective evolutionary algorithms to identify highly autocorrelated areas: the case of spatial distribution in financially compromised farms , 2014, Ann. Oper. Res..

[41]  Chuan Zhou,et al.  Deep Learning in Medical Image Analysis. , 2017, Advances in experimental medicine and biology.