Neuroendocrine modulation sustains the C. elegans forward motor state.

Strains and constructs

C. elegans were cultured on standard Nematode Growth Medium (NGM) plates seeded with OP50, and maintained at 22°C. hp701 was isolated by EMS mutagenesis of a strain carrying an integrated Pceh-10-GFP array ( hpIs202 ), identified through a combination of SNP mapping and whole genome sequencing ( Davis and Hammarlund, 2006 ; Doitsidou et al., 2010 ), followed by rescuing by a unc-39- containing fosmid and a unc-39 minimal genomic clone. hp701 and other mutants obtained from the Caenorhabditis Genetics Center (CGC) were backcrossed at least 4x against N2 prior to usage. See Supplemental Methods for a complete list of mutants, transgenic lines, and constructs.

Electron microscopy L1

(N2) or young adult animals (N2 and hp701 ) were packed into 3 mm diameter aluminum carriers and frozen using a Leica EM HP100 high-pressure freezer. Subsequent fixation and freeze substitution was performed as described ( Weimer, 2006 ). Samples were infiltrated and embedded in the Spurr resin. 70 nm transverse serial sections were prepared and imaged using the FEI Techai 20 TEM at 200 kV (x29,000). For L1, the entire length of RID axon along the dorsal cord was reconstructed. For adults, 20–30 µm of the dorsal nerve cord was reconstructed. See Supplemental Information on EM reconstruction and analyses. Embryonic lineage analysis Strains ( hpIs202 and unc-39; hpIs202 ) were cultured and recorded at 25°C. Embryos were mounted at 1–4 cell stages, and recorded every 35 s with 25 z-levels. After ~350 z-scans (200 min) with DIC optics, the GFP fluorescence optics was automatically activated and exposed for every 20th z-scan till 320 min of development, and then every 30th z-scan till the three-fold stage. Recording was performed at 630x on a 4D microscope controlled by TimeToLive ( Caenotec -Prof. Ralf Schnabel r.schnabel@tu-bs.de ). RID and neighboring lineage cells were traced using SIMI°Biocell (SIMI Reality Motion Systems GmbH, http://www.simi.com/de/home.html ) until the late 1.5-fold stage (340 and 360 min) as described ( Schnabel et al., 1997 ).

Preparation of dissociated larval cells for cell sorting Synchronized hpIs202

(Pceh-10-GFP) and unc-39(hp701); hpIs202 larval stage 1 (L1) animals were grown on 150 mm 8P plates seeded with NA22 for approximately ~12 hr until they reached larval stage 2 (L2). Cells were extracted from L2s. See Supplemental Information for detailed methods. Laser ablation Synchronized hpIs202 (Pceh-10-GFP) L1 animals were immobilized and RID ablated using the MicroPoint laser as described ( Fang-Yen et al., 2012 ). Mock-ablated animals were L1s of the same genotype processed similarly, except that they were not exposed to the laser. L4 RID- and mock-ablated animals were confirmed for the status of RID prior to locomotion analyses.

Locomotion assay

Automated tracking and locomotion analyses were performed as described ( Gao et al., 2015 ). Detailed description of experimental conditions is provided in the Supplemental Information.

Calcium imaging

The third stage (L3) hpIs587 (Pflp-14-GCaMP6::cherry) animals were placed on a drop of M9 buffer on top of a 2.5% agar pad, covered with a cover slip. Four ~1 mm × 1 mm 0.001 inch-thick polycarbonate pieces (Catalog # 9513K12, McMaster-Carr) between the agar pad and coverslip served as spacers to allow movement. Each recording lasted for 3 min. Images were captured using a 20x objective on a Zeiss Axioskop 2 Plus equipped with an ASI MS-40000 motorized stage, a dual-view beam splitter (Photometrics) and a CCD camera (Hamamatsu Orca-R2). The 4x-binned images were obtained at 10 frames per second. See Supplemental Information for detailed methods.

Optogenetic stimulation

We restricted the expression of a codon-optimized version of Chrimson to the RID neuron, and stimulated wild-type, flp-14(gk1055), unc-39(hp701) and npr-4(tm1782)npr-11 ( ok594 ) animals carrying the same transgene using a Diode red laser while animal locomotion was tracked under dark-field imaging by infrared light. The stimulation and tracking were performed using the Colbert (COntrol Locomotion and BEhavior in Real Time) system ( Leifer et al., 2011 ). The locomotion was recorded for each animals, and their trajectories were then extracted and analyzed as described analyzed as previously described ( Luo et al., 2014 ). See Supplemental Information for detailed methods.

Electron microscopy analysis

Images were stitched and aligned using TrakEM2 ( Cardona et al., 2012 ). In the L1 animal, all neurons were identified based on cell body position ( Sulston et al., 1983 ), synapse pattern and neurite trajectory ( White et al., 1986 ). In the adult EM reconstruction, all neurons were identified by characteristic synapse patterns and trajectories ( White et al., 1986 ). Volumetric reconstruction of neurons was performed using TrakEM2 and skeleton tracing using CATMAID ( Saalfeld et al., 2009 ) followed by rendering in Blender ( http://www.blender.org ).

Fluorescence microscopy and confocal imaging

For imaging dense core vesicle ( Pceh-10-IDA-1::cherry) and neuropeptide fluorescent markers ( Pceh-10-INS-22::GFP) , expression patterns of UNC-39 ( Punc-39-GFP, Punc-39-UNC-39::GFP) and RID phenotypes in unc-39 mutants using RID fluorescent markers ( Pceh-10::GFP ) and RID cell fate markers ( Pkal-1-GFP, Pser-2-GFP, Pmod-1-GFP) , images were captured using a 63x objective on a Zeiss Axioplan two connected to a Hamamatsu ORCA-ER digital camera and processed using Improvision Open Lab software. Images were processed using minimal deconvolution levels to remove background fluorescence. Confocal images of transgenic strains carrying either Pflp-14-GFP , Pins17-GFP , unc-3fosmid::SL2::GFP , and unc-39fosmid::GFP were acquired on a Nikon Eclipse 90i confocal microscope. Confocal image processing was conducted using Adobe Photoshop. The primer pair to generate a 4.3 kb Pflp-14 fragment are tactgtcgaccgacaaacacccaaatatcc (forward, SalI) and aactggatcctccttcggattgtgtggag (reverse, BamHI).

Larval cell extraction and cell sorting

Briefly, synchronized animals were pelleted and thoroughly washed with M9 Buffer to remove bacterial contamination. Prior to extraction, an aliquot of sample was flash-frozen in liquid nitrogen to be used as the All Cells reference sample for subsequent transcriptome analysis. To the remaining sample, freshly thawed SDS-DTT solution (0.25% SDS, 200 mM DTT, 20 mM HEPES, 3% sucrose pH = 7.5–8.0), which softens the cuticle, was added for no longer than 4 min (or until majority of animals started twitching, but did not become completely rod-like or rigid). Post-incubation with SDS-DTT, the sample was neutralized and washed 5x times with 1x Egg Buffer. To break apart animals, Pronase solution (15 mg/mL) was added, and the sample was pipetted ~40x for 30 min. After several washes, cells were resuspended in 1x Egg buffer. Propidium iodide (1 μg/mL) was added prior to cell sorting to identify damaged cells. GFP+ cells from L2 worms were sorted onto Trizol-LS (Invitrogen) using a BD FACSAria with a 70 micron nozzle (BD Biosciences) operated at the University of Toronto Flow Cytometry Facility. Profiles of GFP+ strains were compared to an N2 standard to identify and exclude autofluorescent cells. For each independent replicate, 20,000–50,000 events, most likely representing cells, were FACS-isolated from each strain. Three and four biological replicates with corresponding All Cells reference samples were collected for the hpIs202 control and the unc-39 (hp701); hpIs202 strains, respectively.

RNA-sequencing preparation and analysis RNA extraction

For the all cells reference samples, flash-frozen pellets were pulverized using a mortar and pestle and dissolved in Trizol-LS (Invitrogen). RNA was extracted from these pulverized worm pellets (All and from sorted cells collected directly in Trizol-LS (GFP+ cells sample). DNA contamination was removed using the Zymo DNA-free RNA Kit (Zymo Research, Irving, CA) according to manufacturer’s instructions. Library preparation and RNA-sequencing RNA sample concentration and quality were determined using an Agilent Bioanalyzer at The Centre for Applied Genomics (SickKids Hospital, Toronto, ON, Canada).

RNA Integrity Numbers

(RIN) scores of seven and above were used for subsequent RNA-sequencing analysis. RNA concentration was also verified using the Quibit RNA HS Kit (Thermo Fisher Scientific). 5 ng of RNA was used as starting material for cDNA library preparation. cDNA was synthesized from RNA using the SMARTer Ultra Low Input RNA Kit for Sequencing (Clontech) and cDNA libraries prepared using the Low Input DNA Library Prep Kit (Clontech) according to manufacturer’s instructions. RNA-sequencing was performed on an Illumina Hiseq 2000 according to standard protocols, generating 100 base paired-end reads.

Bioinformatics analysis

Sequencing reads were mapped to the C. elegans genome (WS235) using RNA STAR under default settings ( Dobin et al., 2013 ). Using these conditions on RNA STAR, 75–80% of reads aligned to a unique transcript. Gene expression quantification and differential expression were analyzed using HTSeq ( Anders et al., 2015 ) and DESeq ( Anders and Huber, 2010 ), respectively, under default settings. Using DESeq, we identified differentially expressed transcripts between GFP+ cells and All Cells reference samples in the hpIs202 and unc-39(hp701);hpIs202 datasets. For hpIs202 and unc-39(hp701);hpIs202 datasets, transcripts were considered significantly enriched in GFP+ cells over sample matched All Cells by applying the following criteria or filters: (1) Differentially expressed transcripts with False Discovery Rate (FDR) adjusted p values1). Final datasets resulted from analyses of >3 experimental replica from All cells and GFP+ samples for wild-type and unc-39 mutant strains.

Validation of the datasets

Our final datasets were validated using >3 replicates of the wild-type samples. We assessed its quality by two criteria: first, the identification of transcripts known to be expressed in RID, ALA, AIY, or CAN ( Supplementary file 1 ; Positive Controls). We detected significant enrichment for transcripts reported to be expressed by either all GFP+ cells ( Svendsen and McGhee, 1995 ), or a subset of GFP+ cells, e.g. in AIY ( ttx-3 , hen-1 , glc-3 , pdfr-1 ) ( Flavell et al., 2013 ; Ishihara et al., 2002 ; Wenick and Hobert, 2004 ), and in RID ( lim-4 , snf-11, zig-5, ser-2 and kal-1 ) ( Bülow et al., 2002 ; Mullen et al., 2006 ; Tsalik and Hobert, 2003 ); Second, the absence of transcripts from non-neuronal tissues, such as muscles ( myo-2 , myo-3) or neuronal subtypes not included in GFP+ Cells - the glutamatergic ( eat-4 ) and GABAergic ( unc-25 ) neurons ( Supplementary file 1 ; Negative Controls). Lastly, this study validated the presence of FLP-14 and INS-17 in RID ( Table 1 ) by reporter analyses ( Figure 3 ) and the functional relevance of FLP-14 ( Figure 6 ).

Locomotion behavior analysis Behavior acquisition and tracking

When transferred to a new, thinly seeded plate, C. elegans typically spend most of the time moving forward, with brief interruptions of backward movement. As previously described with some modifications ( Gao et al., 2015 ), 35 mm Nematode Growth Media (NGM) plates with limited food (lightly seeded OP50 bacteria) were used for automated tracking and behavioral analyses. Using this method, we quantified the percentage of time animals spent moving forward, backward, and pausing. We also quantified initiation frequency, duration, and velocity of larval stage 4 (L4) animals. For RID-ablated animals, controls were mock-ablated animals carrying the same GFP reporter. For unc-39, flp-14 , ins-17 , and npr-4 npr-11 genetic mutants, controls were N2 animals. Prior to recording, animals were placed in the center of a 50 mm lightly seeded NGM plate and allowed to habituate for 5 min prior to recording. Behavior was recorded for 3 min under a 40x objective using a 20Zeiss Axioskop 2 Plus equipped with an ASI MS-40000 motorized stage and a CCD camera (Hamamatsu Orca-R2). Tracking and analysis were performed using Micromanager and ImageJ software plugins developed in-house (courtesy of Dr. Taizo Kawano). Image sequences were sampled at 100-msec exposure (10 frames per second). The directionality of movement (forward vs. backward) was determined by first identifying the anterior-posterior axis or the ‘head' and 'tail’ points, which were manually defined at the first two frames and verified throughout the recording. To calculate directionality of movement, the displacement of the midline point in relation to the head and tail for each worm was determined based on its position in the field-of-view and the stage coordinates. Image sequences where animals touched the edge of the recording field or crossed over on themselves were not processed.

Quantification and data analyses

Analyses of the output data were carried out using an R-based code developed in-house (courtesy of Dr. Michelle Po). The following parameters were quantified by the program: (1) Initiation (defined as the frequency of directional change for each animal); (2) Duration (defined as the time spent moving in the same direction for >3 frames or 300 msec, calculated for each bout of forward or reversal initiation); (3) Velocity (defined by the speed, displacement of animal divided by the # of frames, and directionality of the animal). Frequency of initiations, durations, and velocities were calculated for forward and reversal locomotion separately.

Calcium imaging and data analysis Regions of interest

(ROIs) containing the RID neuron was defined using a MATLAB script developed in-house. GCaMP and RFP fluorescence intensities from RID were then measured. To analyze overall RID activity, the ratio of GCaMP to RFP was calculated in order to control for possible motion artifacts detected. The velocity of each time point was measured using an Image J plug-in developed in-house ( Gao et al., 2015 ; Kawano et al., 2011 ). The rate for Acceleration during each transition from backward to forward locomotion was calculated by subtracting the lowest velocity point during backward locomotion from the highest velocity point during forward locomotion, and normalizing it to the number of frames in-between these two points (Velocity peak – Velocity trough / # of Frames). The rate of Deceleration during transitions from forward to backward locomotion was calculated similarly (Velocity trough – Velocity peak / # of Frames). Rise and decay in calcium transients during transitions from backward to forward locomotion and vice versa, respectively, were calculated using the linear slope. Cross-correlation analyses were performed between rate of acceleration and calcium rise, as well as deceleration and calcium decay.

Optogenetic stimulation and data analyses

To restrict expression of the Chrimson protein to RID, we drove the expression of a Chrimson::GFP:ZF construct by Pceh-10 , in a subset of neurons (RID, ALA, AIY, and CAN), and the expression of ZIF-1::SL2::GFP by Pgpa-14 , Pttx-3 , and Parr-1 , which overlap with Pceh-10 for ALA, AIY, and CAN, respectively. ZIF-1 targets chrimson::GFP::ZF for degradation. GFP and RFP were used to confirm the specificity of expression pattern of these constructions in targeted neurons. We first generated an integrated transgene hpIs626 from an Ex array made from co-injected Pceh-10-chrimson::GFP::ZF , Pgpa-14-ZIF::SL2::RFP , Pttx-3-ZIF::SL2::RFP and the lin-15 co-injection marker. This transgene exhibits restricted chrimson::GFP expression in RID and CAN. Parr-1-ZIF-1::SL2-RPF was then injected in hpIs626; ; ~10% of the double transgenic animals that carries the Ex array ( hpIs626;hpEx3808 ) exhibited RID-specific chrimson::GFP expression. These array were then crossed into unc-39 , flp-14, and npr-4 npr-11 backgrounds. For optogenetics experiments, we first picked late larvae to young adult transgenic animals by selecting for those with RID-specific GFP expression onto the OP-50 retinal plates under the dissecting fluorescent microscope. After 24 hr incubation, each animal was individually recorded using the COLBERT stimulation system (see below). After recording, each animal was mounted on slides to be examined for GFP signals; data collected from animals that exhibited residual GFP signals in CAN were discarded. To make OP50-retinal plates, we seeded each 60 mm NGM plate with a mixture of 250 μL OP50 in LB with 1 μL of 100 mM retinal in ethanol. Animals were then individually washed in the NGM buffer before allowed to navigate on the surface of a 100 mm NGM plate without food. We used the Colbert system to stimulate Chrimson with a Diode red laser (MRL-III-635, 635 nm wavelength, 200 mW maximum power, CNI Laser) and carry out dark-field imaging using infrared light and a dark field condenser (Nikon) using a 10X Plano Apo objective (NA = 0.45). A motorized stage and custom real-time computer vision software kept the animal in the center of the field of view. The locomotion of each animal was recorded for 30 min or until it reached the edge of the plate. The laser was set at a 3 min on/3 min off cycle during the recording. Data was collected at 30 f/s and analyzed using a customized particle-tracking and shape analysis algorithms. Each trajectory was segmented into periods of forward movement (runs) separated by sharp orientations (turns). Turns were automatically flagged when the heading change of the center of the mass trajectory was >60° over 1 s. To quantify the change of phase velocities before and after the laser was switched on ( Figure 6A,B ), we quantified velocity when the laser was on during a run with 10 s of forward movement before and after the switch for wild-type, flp-14 and npr-4 npr-11 animals, and 4 s before and 4 s after the switch for unc-39 animals. To quantify the forward run length, all runs during the laser on phase and laser off phase were included.

Statistical analysis

For locomotion (# of initiations, durations, velocities), statistical significance was determined using Mann-Whitney or Kruskal-Wallis tests for comparing two and more than two variables, respectively, and subjected to post-hoc analysis. For calcium imaging (correlations between acceleration/deceleration and rise/decay of calcium transients), Pearson tests were used to calculate correlation coefficients and statistical significance. For optogenetic stimulation and locomotion analyses, Wilcoxon matched-pairs signed rank test was used to compare the velocity difference during periods of Lights On and Off for the same strain. The Kruskal-Wallis Test with post-hoc test (Dunn’s Multiple Comparison) was used to assess the significance of difference between groups. p