HIV RGB: Automated Single-Cell Analysis of HIV-1 Rev-Dependent RNA Nuclear Export and Translation Using Image Processing in KNIME

Abstract

Single-cell imaging has emerged as a powerful means to study viral replication dynamics and identify sites of virus−host interactions. Multivariate aspects of viral replication cycles yield challenges inherent to handling large, complex imaging datasets. Herein, we describe the design and implementation of an automated, imaging-based strategy, “Human Immunodeficiency Virus Red-Green-Blue” (HIV RGB), for deriving comprehensive single-cell measurements of HIV-1 unspliced (US) RNA nuclear export, translation, and bulk changes to viral RNA and protein (HIV-1 Rev and Gag) subcellular distribution over time. Differentially tagged fluorescent viral RNA and protein species are recorded using multicolor long-term (>24 h) time-lapse video microscopy, followed by image processing using a new open-source computational imaging workflow dubbed “Nuclear Ring Segmentation Analysis and Tracking” (NR-SAT) based on ImageJ plugins that have been integrated into the Konstanz Information Miner (KNIME) analytics platform. We describe a typical HIV RGB experimental setup, detail the image acquisition and NR-SAT workflow accompanied by a step-by-step tutorial, and demonstrate a use case wherein we test the effects of perturbing subcellular localization of the Rev protein, which is essential for viral US RNA nuclear export, on the kinetics of HIV-1 late-stage gene regulation. Collectively, HIV RGB represents a powerful platform for single-cell studies of HIV-1 post-transcriptional RNA regulation. Moreover, we discuss how similar NR-SAT-based design principles and open-source tools might be readily adapted to study a broad range of dynamic viral or cellular processes.

Keywords: CRM1; Exportin-1; Gag; ImageJ; KNIME; Konstanz Information Miner; Nuclear Ring Segmentation Analysis and Tracking (NR-SAT); RNA nuclear export; RNA regulation; RNA trafficking; Rev; Rev response element; XPO1; chromosomal region maintenance-1; human immunodeficiency virus type 1; image analysis; live cell imaging; open-source; retrovirus; translation; unspliced RNA.

Figure 1

HIV RGB: multicolor single-cell tracking of HIV-1’s post-transcriptional stages. (A) Scheme of HIV RGB strategy, wherein nascent vRNA transcripts are visualized in the nucleus using MS2-YFP tagging (Stage 1) prior to Rev-mCherry synthesis (Stage 2) triggering US vRNA nuclear export (Stage 3) and translation of Gag-CFP (Stage 4). Gag and US vRNA then accumulate in the cytoplasm (Stage 5) prior to Gag molecules ultimately multimerizing to form virus particles, packing two dimerized US VRNA genomes (ψ) (Stage 6). Note that stages 2–4 are tightly coupled and proceed almost concurrently. (B) Images from live cell imaging of a single HeLa.MS2-YFP cell, demonstrating the progression of stages defined for (A). Here, T = 0 was defined as the acquisition time point just prior to the first detection of Rev-mCherry. Nuclear puncta representative of transcription sites are indicated by black arrowheads in panel v. Black and white images on the right (panels viii. and xii.) demonstrate that vRNA MS2-YFP signals co-localize with Gag-CFP puncta (black arrows), indicating vRNA genome packaging. “N” designates the nuclear compartment and “C” the cytoplasm. Abbreviation “roi” stands for region of interest. Figure 1B corresponds to Video S1. (C) Manual tracking of 20 transfected cells, chosen at random, co-expressing Gag-CFP US VRNA with either Rev-mCherry (left) or an mCherry (No Rev) control (right). Each colored line represents a single cell as it progresses to a distinct post-transcriptional stage. Figure 1C corresponds to Video S2.

Figure 2

General image acquisition and analysis workflow. (A) Adherent cells were plated in 8-well glass-bottomed dishes and allowed to settle for 24 h. (B) Cells were transfected to express visible US vRNAs and Rev-mCherry. (C) Cells were imaged using multipoint acquisition (3 points per well) 4–6 h post-transfection using a 20× objective, capturing > 100 transfected cells per condition. Images were acquired for each channel over a prescribed time course. (D) Images were post-processed for NR-SAT and analysis using KNIME, with data output to graphing software.

Figure 3

HIV RGB node and component overview. The HIV RGB workflow is organized into several sections and components. The “Initialize Workflow” section loads the image stacks to be processed. These data are then passed to the “Measurement Channel” and “Nuclear Segmentation” components, respectively. The “Measurement Channel” component processes all measurement channels (default: background subtraction), while the “Nuclear Segmentation” component only processes the nuclear channel. The nuclear channel is first segmented, tracked, and then passed to the “ROI Formation” channel where the cytoplasmic rings are generated. Finally, the data derived from the “Measurement Channel” is rejoined to the now-generated nuclear masks and cytoplasmic rings, and measurements are made followed by data output in a CSV formatted file.

Figure 4

NR-SAT single-cell segmentation and tracking scheme. (Phase 1) The input nuclear channel is illumination-corrected using background subtraction, followed by local contrast enhancement. The resulting images then undergo thresholding (Mean method) and are dilated and then labeled. An object filter is applied to remove threshold artifacts prior to converting the images into binary images and filling holes. Finally, the data is passed to the Wählby Cell Clump Splitter, which separates nuclei that are in proximity to one another. (Phase 2) These data are then passed to the nuclei tracking nodes where the ImageJ plugin TrackMate is implemented to track each cell, accounting for splitting and merging events. (Phase 3) After the nuclei are tracked, the nuclear mask is duplicated and dilated (number of dilations is user-defined and applied to all images in the same manner, 10× in this example) to generate two masks, where the newly dilated mask is larger than the original source mask. These two masks (the newly larger dilated mask and the smaller original mask) are then segmented via the Voronoi segmentation node, generating cytoplasmic rings that are approximately a fixed pixel-width. (Phase 4) Finally these cytoplasmic rings, along with the nuclear masks, are applied to the measurement channel(s) and written to a CSV output file.

Figure 5

Rev-mCherry functional mutants and trafficking variants used to test the HIV RGB strategy. (A) Depiction of Rev-mCherry fusion protein and key functional domains. The arginine-rich domain (ARD) (indicated in purple) encodes residues important for Rev’s nuclear import and binding to the RRE. The ARD is flanked by sequences involved in Rev-Rev multimerization (indicated in gray) to form a functional export complex. Rev’s nuclear export signal (NES) (indicated in blue) binds to XPO1. The Rev M5 mutant bears a R42D/R43L to substitution in the ARD that abrogates Rev-RRE binding. SLT40 encodes I59D and I60D substitutions reducing Rev’s capacity to self-associate. The Rev-M10 mutant bears an L78D/E79L substitution in the NES that abolishes Rev’s ability to interact with XPO1. (B) Functional assessment of Rev-mCherry variants used in this study. To gauge infectious virion production, HIV-1 virions were generated using a HEK293T viral infectivity assay wherein we generated Rev-inactivated HIV-1 YFP reporter virus in the presence or absence of trans-delivered Rev-mCherry or the indicated Rev-mCherry variant protein. Supernatants were harvested at 48 h post-transfection and used to infect target cells, with infectivity measured based on the YFP reporter. Lysates were also harvested from producer cells at 48 h for immunoblot-based detection of HIV-1 Gag, Rev-mCherry, and HSP90 (loading control). (C) Images of individual HeLa cells acquired at 24 h post-transfection and expressing US vRNA in the presence or absence of Rev-mCherry or the indicated Rev trafficking variants. Transcripts are visible in the nucleus for all conditions (see frames ii., v., viii., xi., xiv., with bright nuclear puncta indicated in some images with arrowheads). Only the Rev-mCherry variants deemed functional based on (B) were competent to activate US vRNA nuclear export (panels v., xvii., xx. and xxiii.), Gag-CFP synthesis (panels vi., xviii., xxi. and xxiv.), and virus particle assembly (panels vi., xviii., xxi. and xxiv, see thin arrows). “N” designates the nuclear compartment and “C” the cytoplasm. Images represent compressed z-stacks so that some of the images (notably for Rev-M10-NES, Rev-2xNES, and Rev-2xARD) include signals from the ventral plasma membrane.

Figure 6

NR-SAT output. Color-coded traces for unbiased HIV RGB analysis of the effects of individual Rev-mCherry variants on Rev localization (top left), US vRNA nuclear export (top right), and US vRNA translation to generate Gag-CFP (bottom left). Cells were imaged for 24 h, with images processed using the NR-SAT workflow, tracking the relative subcellular location of each Rev-mCherry variant or the mCherry control (black line). Lines indicate mean fluorescence intensity (MFI) with standard error of the mean (SEM) for each time point, shown as a similarly colored delimited background for each condition.