Exposure to pathogenic bacteria inhibits the ability of worms to re-enter the bacteria lawn.

(A) Above: representative sequence of phase contrast images of P. aeruginosa PA14 lawn evacuation at t=1, 9 and 15 hrs after deposition of C.elegans onto lawn. Evacuated worms are highlighted in red using imaging processing. Below: fractional occupancy of C.elegans on bacteria lawns plotted against time spent on pathogenic PA14 lawns (red) or non-pathogenic E.coli OP50 lawns (blue). Mean occupancy and SEM over n = 25 independent experiments with 10 worms each. (B) C.elegans fractional occupancy (black) and the rate constant of C.elegans exit (exits events/hour/animal, red) plotted against time spent on pathogenic PA14 lawns. The exit rate constant increased with time spent on the lawn. Mean and SEM over n = 5 samples of 10 worms each. (C) Representative phase contrast images of a worm attempting to re-enter PA14 lawn after 5 hours (left column) and 15 hours (right column) of PA14 exposure. Time labels denote time after first contact of worm with bacteria lawn. (D) Latency in re-entry (delay in the re-entry of the animal onto the bacteria lawn upon first contact with the lawn, red) and fractional occupancy (black) plotted against duration of exposure for worms on PA14 lawns. Mean and SEM over n = 5 samples of 10 worms each. (E) Representative sequence of phase contrast images of a worm (highlighted in red) attempting to enter a fresh PA14 bacterial lawn pre PA14 exposure (left) and post 15 hours of PA14 exposure (right). Time labels denote time after first contact of worm with bacteria lawn. (F) Representative plots of distance of center of mass of each animal from the edge of the PA4 lawn versus time after first contact with lawn, pre (black, n = 11) and post 15 hours of PA14 exposure (red, n = 9). (G) Latency of entry of worms pre (n=35) and 15 hours post (n=29) PA14 exposure. Post exposed worms increased latency of entry tenfold compared to pre-exposed control. (H) Model for lawn evacuation. Pre-exposed worms (left) have high re-entry rates and low exit rates. Exposure to pathogenic bacteria causes a transition in behavior (right), where worms increase exit rate and decrease re-entry rate, leading to net lawn evacuation.

An optogenetic screen of the C.elegans nervous system to identify neurons controlling exit and re-entry into pathogenic bacteria lawns.

(A) Experimental protocol used to identify neurons that modulate long term changes in PA14 lawn evacuation dynamics. In each experiment, animals of one transgenic line expressing archaerhodopsin were placed onto a lawn of PA14 and allowed to settle for 1 hour (blue segment, in schematic). Worms were illuminated with green light (525nm) to inhibit activity for 2 hours in neurons expressing archaerhodopsin (green segment, in schematic). Worm evacuation was monitored over an additional 15 hours (red time segment, schematic). (B) Differential retention metric for 29 transgenic lines. The differential retention metric for each transgenic line was determined by measuring the difference in the area under lawn evacuation curves with and without neural inhibition (see methods). Differential retention metrics were compared against the standard deviation in retention metrics from uninhibited control worms (gray) to determine statistical significance (see methods). Six of the twenty nine lines (dop-2, flp-4, mpz-prom1, sams-5, ttx-3, and npr-4) showed statistically significant differential retention metrics. Mean and SEM over n = 5 samples of 10 worms each. (C) Rate constant of exit (exits events/hour/animal) for statistically significant transgenic lines (B). Four lines showed statistically significant changes in rate constant (dop-2, sams-5, ttx-3, and npr-4) between inhibition (red, +ATR) and no inhibition (blue, -ATR) control. Mean and SEM over n = 5 samples of 10 worms each. (D) Left: fluorescence and phase image of Pdop-2::ARCH-mCherry (C). Scale bar 15um. Right: fractional lawn occupancy of Pdop-2::ARCH-mCherry as a function of time (right) with (red, +ATR) and without (blue, -ATR) neural inhibition. (E) Trajectories of worms exiting the lawn either with (left) or without (right) inhibition of neurons expressing dop-2 (D). Trajectories are color coded by time and are over n = 2 samples of 10 worms each. (F) Latency in re-entry (time of re-entry of the animal onto the bacteria lawn after first contact) of statistically significant transgenic lines (B). Four lines show statistically significant differences in latency in re-entry(dop-2, sams-5, ttx-3, and npr-4) between inhibition (red, +ATR) and no inhibition (blue, -ATR) control. (G) Representative fluorescence and phase image (left) of Pnpr-4::ARCH-mCherry (F). Scale bar 15um. Fractional lawn occupancy of this line as a function of time (right) with (red, +ATR) and without (blue, -ATR) neural inhibition. (H) Representative plots of the distance of individual Pnpr-4::ARCH-mCherry worms from the edge of the lawn with (red, n = 5) and without (black, n = 5) neural inhibition. Inset: representative trajectories (n = 5) color coded by time of Pnpr-4::ARCH-mCherry worms attempting to enter bacteria lawn with (left) and without (right) neural inhibition.

Compressed sensing analysis identified a set of neurons controlling worm re-entry and exit.

(A) Example of a neural network where behavior is controlled by a small number of key nodes (highlighted in red) that make up a small fraction of the total number of nodes (gray). (B) Single neuron perturbations can be framed as solving a N by N matrix equation to find the neural contributions to phenotypes. Each diagonal entry (white) corresponds to a single perturbation. Interrogating the entire nervous system to determine the contribution (weight, w) of each neuron to the behavior requires as many measurements as the number of neurons in the nervous system (N). (C) Using compressed sensing to determine neurons controlling entry and exit from the optogenetic screen by determining their weights (w) from an underdetermined set of equations. These equations can be represented as a 29×87 measurement matrix M (left). Rows are promoters and columns are neuron types. Matrix has an entry 1 (white) if the promoter drives expression in that neuron type, else 0 (black). Phenotype vectors Pexit and Pentry were obtained by taking the difference between the re-entry timescale or rate constant of exit with and without neural inhibition (see methods). From measurements of the phenotype of 29 lines(Pexit and Pentry), the weights of each of the 87 neurons (w) can be determined using compressed sensing (see methods) (D) Median neuron weight contributions to the exit phenotype from 10,000 lasso regression solutions from bootstrapping (see methods). Neurons with significant weights contributing to the exit phenotype (AVA, CEP, HSN,RIA,RID, SIA) are highlighted. (E) Median neuron weight contributions to the re-entry phenotype from 10,000 lasso regression solutions from bootstrapping (see methods). Neurons with significant weights contributing to the re-entry phenotype (AIY, SIA, AVK, MI) are highlighted.

Identified neurons responsible for re-entry show reduction in calcium activity following PA14 exposure.

(A) Schematic of tracking and image stabilization microscope allowing for simultaneous measurement of GCaMP activity and worm position with 1µm precision (orange: mKO emission, light green: GCaMP emission, blue: GCaMP excitation, dark green: mKO excitation). An mkOrange labeled marker neuron is imaged through Camera 1. This imaging data is transmitted to FPGA 1 for processing. Processed position information is used to control the x,y,z (stage) and rotational optics to stabilize the worm within the field of view. GCaMP imaging data is acquired via Camera 2. A tunable lens being controlled by FPGA 2 scans through the worm to allow for acquisition of GCaMP images at multiple focal planes. A DLP mirror array, controlled by a PC, is used to target light on specific neurons through structured illumination. (B) Fluorescent and phase image of Pttx-3::GCaMP line used to image AIY neural activity. Scale bar 10µm. (C) Histogram of AIY neural activity pre (blue) and post (red) 24 hours of PA14 exposure normalized to pre-exposure baseline (see methods). Data taken from n = 4 worms over 36 min. Inset: histogram of neural activity from a single worm pre (blue) and post (red) 24 hour PA14 exposure. (D) AIY neural activity over 36min for a single worm pre (blue) and post (red) 24 hours of PA14 exposure normalized to pre-exposure baseline (see methods). Inset: zoom in of AIY neural activity between 28 to 33.5minutes to illustrate neural activity transition in naive worms. (E) Fluorescent and phase image of Pnpr-4::GCaMP line used for AVK imaging. Scale bar 10µm. (F) Histogram of AVK neural activity pre (blue) and post (red) 24 hours of PA14 exposure normalized to pre-exposure baseline (see methods). Data taken from n = 7 worms over 36 min. Inset: histogram of neural activity from a single worm pre (blue) and post (red) 24 hour PA14 exposure. (G) AVK neural activity for a single worm pre (blue) and post (red) 24 hours of PA14 exposure as a function of time normalized to pre-exposure baseline (see methods). (H) Fluorescent and phase image of Pnpr-4::GCaMP used to image SIA neural activity. Scale bar 10µm. (I) Histogram of AVK neural activity pre (blue) and post (red) 24 hours of PA14 exposure normalized to pre-exposure baseline (see methods). Data was taken from n = 8 worms over 36 min. Inset: histogram of neural activity from a single worm(blue) and post (red) 24 hour PA14 exposure. (J) SIA neural activity for a single worm pre (blue) and post (red) 24 hours of PA14 exposure as a function of time normalized to pre-exposure baseline (see methods).

Entry onto pathogenic bacteria can be controlled through modulation of key neurons.

(A) Fluorescent and phase image of Pttx-3::ARCH line used for neuron specific AIY inhibition. Scale bar 10um. (B) Latency in entry (delay in the entry of the animal onto the bacteria lawn upon first contact) of naive worms onto PA14 (n > 18) and OP50 (n > 17) lawns with (red) and without (blue) AIY neural inhibition. Neural inhibition resulted in significant increases in latency in entry onto PA14, but not OP50. (C) Fluorescent and phase image of Pflp-1::NpHR line used for neuron specific AVK inhibition. Scale bar 10µm. (D) Latency in entry of worms onto PA14 lawns with (red) and without (blue) AVK neural inhibition (n > 27). (E) Latency in entry of naive worms onto PA14 (n > 11) and OP50 (n > 16) with (red) and without (blue) inhibition of npr-4 expressing neurons. Neural inhibition resulted in significant increases in latency in entry onto both PA14 and OP50. (F) Fluorescent images of Pnpr-4::ARCH with (+ATR) and without (-ATR) targeted illumination of SIA/SIB neuron cluster. (G) Representative sequence of phase contrast images of a naive worm attempting to enter PA14 bacteria lawn with (+ATR) and without (-ATR) SIA neural inhibition. Time labels denote time after first contact of worm with bacteria lawn. (H) Latency in entry of naive worms onto PA14 with (red) and without (blue) SIA neural inhibition. Neural inhibition resulted in significant increase in latency in entry for PA14. Mean and SEM over n = 5 samples of 10 worms each. (I) Fluorescent and phase image of Pttx-3::CHR2 line used for neuron specific AIY activation. Scale bar corresponds to a length of 10µm. (J) Re-entry rate constant of evacuated worms following 24 hours of exposure to PA14 lawns with (green) and without (blue) AIY neural activation. Mean and SEM of the rate constant over n = 5 samples of 10 worms each (see methods). (K) Fluorescent and phase image of Pnpr-4::CHR2 line used for neural activation. Scale bar corresponds to a length of 10µm. (L) Re-entry rate constant of evacuated worms following 24 hours of exposure to PA14 lawns with (green) and without (blue) npr-4 neural activation. Mean and SEM of the rate constant over n = 5 samples (see methods).

Lawn evacuation following neural inhibition.

Evacuation dynamics of all 29 transgenic lines following 2 hours of neural inhibition.

Compressed sensing solutions with sparsity parameters for lawn exit and entry

(A) Median neuron weights from 10,000 lasso regression solutions over three orders of magnitude of sparsity parameters for lawn exit rate. (B) Median neuron weights from 10,000 lasso regression solutions over three orders of magnitude of sparsity parameters for lawn entry timescale.

Validation of compressed sensing solutions via to Arch expression efficiency through measurement matrix corruption

Solutions obtained following perturbing the measurement matrix to mimic differences in expression of Archaerhodopsin (see methods).

Validation of robustness of compressed sensing solutions to choice of measurements through promoter removal

To validate that solutions to lawn entry were robust to choice of measurements, 1 (A) to 5 (E) promoters were removed at random from the measurement matrix (see methods) and solutions were evaluated. Our results showed that our solutions were robust to such removals. (F) Aggregated solutions across 1-5 to removals.

Recovery and false positive rate for measurement matrix of choice

(A) False positive and negative rate for recovery of measurement matrix for a randomly simulated set of key neurons (see methods). False positive and recovery rates for each key neuron identified as being important in lawn entry behavior: AVK (B), AIY (C), MI (D) and SIA (E).

Activation of re-entry neurons fails to impact lawn occupancy

Activation of AIY (A) and Npr-4 expressing neurons (B) fails to significantly increase lawn occupancy in evacuated colonies. (C) Worms that re-enter lawn after Npr-4 neural activation show very low residency time on PA14 lawns.

Candidate neuropeptide identification from re-entry neurons

(A) Scatterplot of neuropeptides plotted on an axis of maximum relative expression in one of the two candidate neurons (AIY and SIA) versus the specificity of their expression (as measured by calculated participation ratio, see methods). pdf-2 is highlighted as a neuropeptide of interest as it showed both high expression and high specificity. (B) Plot of worm latency of re-entry onto lawns as a function of time spent on PA14 lawn for loss of function mutant of the neuropeptide pdf-2 (red) and wild type worms (blue). Mutant worms showed systematic increases in latency of re-entry compared to wild type control. (C) Representative center of mass trajectories of worms at the edge of the PA14 lawn taken from 12 to 14hours of PA14 exposure for wild type worms (left) and PDF-2 mutant worms (right) color coded by time

Archaerhodopsin lines that cons0tute the measurement matrix

Channelrhodopsin and Halo lines

GCaMP Lines