A Real-Time Biosensor for ERK Activity Reveals Signaling Dynamics during C. elegans Cell Fate Specification

Abstract

Kinase translocation reporters (KTRs) are genetically encoded fluorescent activity sensors that convert kinase activity into a nucleocytoplasmic shuttling equilibrium for visualizing single-cell signaling dynamics. Here, we adapt the first-generation KTR for extracellular signal-regulated kinase (ERK) to allow easy implementation in vivo. This sensor, "ERK-nKTR," allows quantitative and qualitative assessment of ERK activity by analysis of individual nuclei and faithfully reports ERK activity during development and neural function in diverse cell contexts in Caenorhabditis elegans. Analysis of ERK activity over time in the vulval precursor cells, a well-characterized paradigm of epidermal growth factor receptor (EGFR)-Ras-ERK signaling, has identified dynamic features not evident from analysis of developmental endpoints alone, including pulsatile frequency-modulated signaling associated with proximity to the EGF source. The toolkit described here will facilitate studies of ERK signaling in other C. elegans contexts, and the design features will enable implementation of this technology in other multicellular organisms.

Keywords: C. elegans; ERK; biosensor; extracellular signal-regulated kinase; kinase translocation reporter; vulval development.

Figure 1. The nuclear ratio ERK-nKTR design

(A) The KTR method (Regot et al., 2014). When kinase activity is low, the KTR sensor protein is not phosphorylated and is nuclear-enriched. When kinase activity is high, the KTR sensor is phosphorylated and excluded from the nucleus. (The KTR is always largely excluded from the nucleolus.) (B) The ERK-KTR sensor is composed of peptides from Elk1 (Elk1^312–356 and FQFP), which promote ERK docking to substrates, followed by a module containing NLS and NES sequences, S/T-P phospho-acceptor residues for phosphorylation by ERK (Phospho-Sites), and the green fluorescent protein mClover. An annotated protein sequence of ERK-nKTR-mClover is shown in Table S2. (C) Bicistronic transgene design of ERK-nKTR: the coding region for a viral 2A peptide (T2A) is inserted between the coding regions for ERK-KTR-mClover and mCherry-H2B. The T2A peptide triggers “ribosomal pausing” efficiently in C. elegans (Ahier and Jarriault, 2014), such that ERK-nKTR-mClover and mCherry-H2B are encoded by a single transcript but produced as two separate, functional proteins in an equimolar ratio. Tissue-specific expression of the bicistronic transcript is achieved using appropriate promoter (prom) and 3′ UTR sequences. (C) After imaging of live C. elegans, the ratio of mCherry to mClover intensity is calculated per pixel, producing a new image (Red/Green), a pictorial representation of the the Red/Green ratio within nuclei (arrows), as quantified in Figs. 2 and 3.

Figure 2. The ERK-nKTR reports MPK-1/ERK activity in a phosphorylation-dependent manner

(A) The six VPCs, P3.p-P8.p, receive an EGF signal from the Anchor Cell (AC). Genetic and transcriptional reporter analyses indicate that the EGF signal results in high MPK-1/ERK activity in P6.p (red), the VPC closest to the AC. In the L2 stage, there is low activation of the EGFR-Ras-ERK pathway in P5.p and P7.p (Yoo et al., 2004); VPCs further from the AC, such as P4.p (yellow) have no MPK-1 activity. This drawing and all images of VPCs are oriented with anterior at the left and dorsal at the top. In half of wild-type hermaphrodites, P3.p often fuses with the hypodermal syncytium in the L2 stage instead of remaining as a VPC, so it is omitted from our analysis. (B-D) Here, and in all following images of ERK-nKTR expression (Figs. 3–5), the ERK-nKTR-mClover reporter (ERK-nKTR) is shown in green and mCherry-H2B (H2B) is in magenta. ERK-nKTR, H2B, and the calculated Red/Green ratio (Red/Green) are also shown separately in grayscale. Scale bars are 10 μm. (B) The transgene arTi85 expresses ERK-KTR(NLS3), which we call “ERK-nKTR”, in VPCs. ERK-nKTR protein is excluded from the nucleus of P6.p (arrows). (C) Mutation of ERK-nKTR phospho-sites to alanines in the ERK-nKTR(AAA) transgene arTi101 prevents key phosphorylations that occur upon ERK activation; thus, this mutant form is not responsive to ERK activity and accumulates in the nucleus of P6.p. (D) Mutation of the phospho-sites to glutamic acids, mimicking constitutive phosphorylation, in the ERK-nKTR(EEE) transgene arTi100 also makes the mutant form independent of ERK activity, and is excluded from nuclei in all the VPCs. (E) Each boxplot graph indicates the median, upper quartile, lower quartile, maximum, minimum values, and outliers (values more than 1.5X the interquartile range from either the upper or lower quartile). (F-H) Red/Green ratios for VPC nuclei expressing (F) ERK-nKTR (n=23), (G) ERK-nKTR(AAA) (n=19), and (H) ERK-nKTR(EEE) (n=19). To assess whether the localization of each reporter is patterned in VPCs, a one-way ANOVA was performed to compare the Red/Green ratios in all VPCs. For ERK-nKTR, there is a highly significant effect on Red/Green ratios when different VPCs are compared (at the p<0.01 level, F(4,109)=32.1, p-value=1.16×10⁻¹⁷). Post hoc pairwise comparisons find significant differences between Red/Green ratios for the following: P6.p vs. P4.p, P5.p, P7.p, or P8.p (all p-values<0.01), P5.p vs. P4.p or P8.p (p-values<0.01), and P7.p vs. P8.p (p-value<0.05) (see statistical tests in Supplemental Materials). In contrast to ERK-nKTR, neither phospho-site mutant is patterned; there are not significant differences in Red/Green ratios between VPCs expressing ERK-nKTR(AAA) (at the p<0.01 level, F(4,90)=2.36, p-value=0.060) or ERK-nKTR(EEE) (at the p<0.01 level, F(4,89)=0.661, p-value=0.621).

Figure 3. The ERK-nKTR is sensitive to levels of MPK-1/ERK activity

Images of ERK-nKTR transgene arTi85 in genetic backgrounds with altered MPK-1/ERK activity show ERK-nKTR-mClover (ERK-nKTR), mCherry-H2B (H2B), and the calculated Red/Green ratio (Red/Green). (A) Loss of MPK-1/ERK activity in mpk-1(0) causes nuclear accumulation of ERK-nKTR in P6.p, similar to the pattern seen with the non-phosphorylatable ERK-nKTR(AAA) (see Fig. 2D). (B) Ectopic activation of MPK-1 in all VPCs, achieved by expression of a constitutively active and stable form of LIN-45/Raf (see STAR Methods), results in ectopic exclusion of ERK-nKTR from the nuclei of P4.p, P5.p, P7.p, and P8.p, similar to the result seen with the phospho-mimetic ERK-KTR(EEE) (see Fig. 2E). (C–D) Red/Green ratios for VPC nuclei expressing ERK-nKTR in (C) mpk-1(0) (n=19), or in (D) the presence of hyperactive LIN-45/Raf (n=16). To assess whether the localization of ERK-nKTR is patterned in VPCs of each genotype, one-way ANOVAs were performed to compare the Red/Green ratios in all VPCs. ERK-nKTR is not patterned in mpk-1(0); there are not significant differences in Red/Green ratios between VPCs (at the p<0.01 level, F(4,90)=1.66, p-value=0.165). However, in the presence of hyperactive LIN-45/Raf, there is a significant effect on Red/Green ratios when different VPCs are compared (at the p<0.01 level, F(4,79)=4.52, p-value=0.002). In this case, post hoc pairwise comparisons indicate that the Red/Green ratio of P6.p is significantly different from those of P8.p (p-value<0.01) and P4.p (p-value<0.05) (see statistical tests in Supplemental Materials).

Figure 4. The ERK-nKTR reports MPK-1/ERK activity in multiple cell contexts

(A) ERK-nKTR is excluded from nuclei of migrating Sex Myoblast (SM) cells in wild type, but not in mpk-1(0). SM-specific ERK-nKTR transgene arTi133, showing ERK-nKTR-mClover (ERK-nKTR) and mCherry-H2B (H2B) in wild type (+) and mpk-1(0). (B) ERK-nKTR is excluded from the nuclei of AWC and ASE sensory neurons in wild type, but not in mpk-1(0). AWC and ASE-specific ERK-nKTR transgene arTi137, showing ERK-nKTR-mClover (ERK-nKTR) and mCherry-H2B (H2B) in wild type (+) and mpk-1(0) AWC (white arrows) and ASE (yellow arrows) neurons. (C) ERK-nKTR is localized in a pattern correlated with MPK-1 activity in the hermaphrodite germline. Germline-expressed ERK-nKTR transgene arSi4, showing ERK-nKTR-mClover (ERK-nKTR) and mCherry-H2B (H2B). The most distal germline nuclei (oriented to left in this micrograph) are in mitosis. More proximally, nuclei enter meiosis and most of the length of germline remains in pachytene of Meiosis I. Proceeding proximally, the germline “loops”, and in this region nuclei enter diplotene, which is followed by diakinesis and oocyte maturation. (D-F) Images from insets indicated by boxes in (C). (D) ERK-nKTR accumulates in nuclei within the distal mitotic region, consistent with low MPK-1 activity detected in fixed samples (Lee et al., 2007). (E,F) ERK-nKTR levels are low in some nuclei in pachytene (E), but becomes highly excluded from nuclei in the proximal pachytene (white arrows) and loop region (yellow arrows) (F). In the most proximal region, ERK-nKTR is excluded from the nuclei undergoing diakinesis and oocyte maturation (C, labeled −1, −2, −3, −4). Nuclear exclusion of ERK-nKTR in proximal pachytene, diplotene, and diakinesis regions is consistent with regions where high levels of phosphorylated substrates are detected in fixed samples (Arur et al., 2011; Drake et al., 2014). Scale bars in all images are 10 μm.

Figure 5. Live imaging of ERK-nKTR during VPC development

Unless otherwise indicated, movies were made of strains containing the transgene arTi85, which expresses ERK-nKTR in VPCs. (A) To examine ERK activity dynamics in wild-type late L2 larvae, time-lapse movies were captured by imaging at 2 minute intervals over a 1 hour time period (see STAR Methods). The Red/Green ratio of individual VPC nuclei over time is shown in two ways: all individual traces overlaid for each VPC (top), and a heatmap (bottom) where each row represents a cell, and Red/Green ratio is represented by color (n=11 larvae). (B) Time-lapse movies were captured as described in (A) to examine arTi101 larvae that express non-phosphorylatable ERK-nKTR(AAA) in an otherwise wild-type background (top, n=5 larvae), and mpk-1(0) larvae that express ERK-nKTR (bottom, n=5 larvae). Shown are all individual traces overlaid for each VPC. (C) Time-lapse Red/Green data for all individual nuclei were analyzed for peaks (STAR Methods). Shown are five representative traces for P6.p cells, where Red/Green of individual nuclei is shown over time, and the position of activity peaks are indicated by black circles. (D) Peak frequency (peaks/hour) is shown for each VPC (top). A two-sample k-test was performed to make pairwise comparisons of peak number (per hour) between VPCs. The peak frequency found in P6.p differed significantly from peak frequency in all other VPCs (p-value<1×10⁻⁶). The relative peak amplitude (difference between Red/Green ratio at peak and nearest trough) is shown for each VPC (bottom). No significant differences in relative peak amplitude were found in pairwise comparisons of different VPCs using a two-sample k-test. (E) Time-lapse movies were captured by imaging at 10 minute intervals over 4 overlapping 3.5-hour time periods after egg lay to examine ERK activity in wild-type VPC development: 20–23.5 hours (Early L2), 23–26.5 hours (Mid L2), 26–29.5 hours (Late L2), and 30–33.5 hours (Early L3) (see also STAR Methods, Fig. S3A). Heat maps where each row represents a cell, and Red/Green ratio of nuclei is represented by color. To correct for variation between animals, the average Red/Green value for the P4.p nucleus of a given larva was used as an internal baseline for normalization of other VPCs in that individual, since P4.p experiences minimal EGF signal (see Discussion and STAR Methods). Occasional missing values in heatmaps (white) result when the nucleus is out of focus for the time point.