(A,B) Device used to immobilize C. elegans for calcium imaging. (A) Layout of the microfluidic worm channel (red) and the overlying O₂ flow chamber (blue). The surrounding grey shaded box represents PDMS cast. Image at top shows an animal in the channel. (B) Schematic cross section of the device. Scale bar represents 100 μm. (C) Fluorescence imaging of the O₂-sensitive dye Ru(phen)₃Cl₂ dissolved in the worm channel. Each trace is a fluorescence intensity profile recorded while switching between the indicated O₂ concentrations. (D–I) Measurements of neural activity by calcium imaging of URX and BAG. (D,E,G,H) Black traces show the average percent change of G-CaMP fluorescence (ΔF/F₀) and dark shading shows standard error of the mean (SEM). Concentrations were 21% and 10% O₂ (D,E) or 21% and 4% O₂ (G,H). Light grey shading indicates the intervals at 21% O₂. (D) Average URX response to O₂ upshift. (E) Average URX response to O₂ downshift (a significant decrease is observed following a 21-10% downshift, as compared to a 21-21% control shift (Figure S3) (P=0.0195 by t-test)). (F) Individual URX responses to O₂ upshift, n=42. Each row represents one neuron. (G) Average BAG response to O₂ upshift. (H) Average BAG response to O₂ downshift. (I) Individual BAG responses to O₂ downshift, n=67. Each row represents one neuron. Control traces for URX and BAG calcium imaging are shown in Figure S3.

(A–D) Locomotion speed of indicated mutant strains. Traces show average speed and dark shading indicates standard error of the mean (SEM). Oxygen concentrations were switched between 21% and 10%; light grey shading marks intervals at 21%. (A) gcy-35(ok769) mutants. (B) gcy-36(db66) mutants. (C) gcy-31(ok296) mutants. (D) gcy-33(ok232) mutants. (E–H) Average speed changes (differences of the means of 10 second intervals before and after the switch) of animals with indicated genotypes. Error bars indicate SEM. Asterisks indicate significance by one-way ANOVA with Dunnett's post test, using the mutant in each panel as control group (* P=0.01–0.05, ** P=0.001–0.01, *** P<0.001). The numbers of animal tracks analyzed are indicated. Wild type data are from Figure 1D. (E) Transgenic rescue of gcy-35(ok769) by a cDNA under its own promoter (expressed in URX, AQR, PQR, BDU, SDQ, AVM, PVM, ALN, and PLN) or the flp-8 promoter (expressed in URX, AUA, and PVM). (F) Transgenic rescue of gcy-36(db66) by a cDNA under its own promoter (expressed in URX, AQR and PQR) or the flp-8 promoter. (G) Transgenic rescue of gcy-31(ok296) with genomic clones. (H) Transgenic rescue of gcy-33(ok232) by a cDNA under a BAG-specific gcy-31 promoter.

(A–C) Distributions of fed and starved animals in gradients of 0% to 21% O₂. About 80–100 animals per assay were allowed to distribute through a device with a linear O₂ gradient, and their positions were scored after 25 minutes. Error bars indicate standard error of the mean (SEM). (A) Wild type. (B) gcy-33(ok232) mutants. (C) gcy-31(ok296) mutants. (D) The hyperoxia avoidance index describes the preferential accumulation of animals in the middle three bins (7–14% O₂) compared to the left three bins (14–21% O₂), and is calculated as [(# at 7–14%)−(# at 14–21%)]/(# in 7–21%). Error bars indicate SEM. Asterisks indicate significance by t-test, P < 0.05. Crosses indicate significance by one-way ANOVA with Dunnett's post test using N2 starved as control group (P < 0.05). The numbers of experiments performed are indicated.

(A–D) Locomotion speed of animals with the indicated genotypes. Traces show average speed and shading indicates SEM. Concentrations were 21% and 10% O₂; shading marks intervals at 21% O₂. (A) gcy-35(ok769); gcy-31(ok296) double mutants. (B) gcy-35(ok769); gcy-31(ok296); BAG::gcy-35 BAG::gcy-36 sGC swap animals. (C) gcy-35(ok769); gcy-33(ok232); gcy-31(ok296) triple mutants. (D) gcy-35(ok769); gcy-33(ok232); gcy-31(ok296); BAG::gcy-35 BAG::gcy-36 sGC swap animals. (E) Average speed changes (difference of the means of 10 second intervals before and after the switch) of animals with the indicated genotypes. Error bars indicate SEM. Asterisks indicate significance by one-way ANOVA with Dunnett's post test using the corresponding mutants in each panel as control groups (***P<0.001). The numbers of animal tracks analyzed are indicated. Wild type data are from Figure 1D. (F–H) Locomotion speed of control animals (F) or animals that express the light gated ion channel ChR2 in BAG under the BAG specific flp-17 promoter (G). Traces show average speed and shading indicates SEM. Blue shaded bars show when animals were exposed to blue light for the indicated time intervals. (H) Average speed changes (difference of the means of 10 second intervals before and during the 50 second light pulse). Error bars indicate SEM. Asterisks indicate significance by t-test (P= 0.0012). The numbers of animal tracks analyzed are indicated. (I) Model of sensory function in URX and BAG neurons, in wild-type and sGC swap (gcy-35(ok769); gcy-33(ok232); gcy-31(ok296); BAG::gcy-35 BAG::gcy-36) animals. Cell bodies, sensory processes, and axons are diagrammed. Left, in wild-type animals URX is depolarized by O₂ upshift and BAG by O₂ downshift. Right, in sGC-swap animals, BAG is depolarized by O₂ upshift, and URX is silenced by the gcy-35 mutation.

(A) A subset of sGC-expressing sensory neurons of C. elegans. Additional sGC-expressing neurons (some of ALN, BDU, SDQ, PLN) and non-sGC expressing sensory neurons (ASH, ADL, ADF) can also affect O₂ responses (Chang et al., 2006; Rogers et al., 2006). (B–I) Locomotion speed and reversal rates of C. elegans off food. Traces show averages and dark shading indicates standard error of the mean (SEM). Oxygen concentrations were switched between 21% and 10%; light shading marks intervals at 21%. (B,D,F,H) Locomotion speed, calculated in 5 second bins. (C,E,G,I) Reversals per animal per minute, calculated in 15 second bins. (B–E) Wild type animals, well-fed (B,C) or 1 hour starved (D,E). For well-fed animals, average basal speed was 0.19±0.03 (SD) mm/s, range 0.14–0.28 mm/s; average reversal rate was 1.5±0.29 (SD) reversals/min, range 0.9–2.1 reversals/min (n=13 assays). For starved animals, average basal speed was 0.17±0.02 mm/s, range 0.13–0.20 mm/s; average reversal rate was 1.1±0.21 reversals/min, range 0.7–1.5 reversals/min (n=21 assays). (F–I) animals in which either URX (F,G) or BAG (H,I) neurons were killed by laser surgery. (J,K) Average speed and reversal changes to O₂ upshift and downshift; (J) Difference of the mean speeds in 10 seconds before and after the switch; (K) Difference in the number of reversals in one minute before and after the switch. Error bars indicate SEM. Asterisks indicate significance by one-way ANOVA with Dunnett's post test using mock-ablated animals as a control group (* P=0.01–0.05, *** P<0.001). The numbers of animal tracks analyzed are indicated.

Measurements of neural activity by calcium imaging. Traces show the average percent change of G-CaMP fluorescence (ΔF/F₀). (A) URX responses. Asterisks indicate significant fluorescence changes comparing mean ΔF/F₀ from 0–10 seconds vs 12–22 seconds (** P=0.001–0.01, *** P<0.001, paired t-test). (B) URX average ΔF/F₀ responses (12–22 s), same color code as in (A), arranged to show effects of target O₂ level and O₂ shift magnitude. Error bars indicate SEM. The numbers of recordings analyzed are indicated. (C) BAG responses. Asterisks indicate significant fluorescence changes comparing mean ΔF/F₀ from 0–10 seconds vs 12–22 seconds (** P=0.001–0.01, *** P<0.001, paired t-test). (D) BAG average ΔF/F₀ responses (12–22 s), same color code as in (C), arranged to show effects of target O₂ level and O₂ shift magnitude. Error bars indicate standard error of the mean (SEM). The numbers of recordings analyzed are indicated. Statistics are in Table S1.

(A–E) Measurements of neural activity by calcium imaging of URX (A–C) and BAG (D,E). Black traces show the average percent change of G-CaMP fluorescence (ΔF/F₀) and dark shading shows standard error of the mean (SEM). Concentrations were 21% and 10% O₂ (A–C) or 21% and 4% O₂ (D,E). (A) URX, gcy-35(ok769) mutants. (B) URX, gcy- 36(db66) mutants. (C) URX, gcy-33(ok232) mutants. (D) BAG, gcy-31(ok296) mutants. (E) BAG, gcy-33(ok232) mutants. Light grey shading indicates the intervals at 21% O₂. (F,G) Average ΔF/F₀ from 12–22 seconds for indicated genotypes, (F) URX; (G) BAG. The data for wild type animals correspond to Figure 2D,H. Error bars show SEM. Asterisks indicate significance by one-way ANOVA with Dunnett's post-hoc test using wild type in each panel as control groups (***P<0.001, ns not significant). The numbers of recordings analyzed are indicated.

(A–E) Calcium imaging of BAG neurons with altered sGC expression. Black traces show the average percent change of G- CaMP fluorescence (ΔF/F₀) and dark shading shows standard error of the mean (SEM). Concentrations were 21% and 4% O₂. Light grey shading indicates the intervals at 21% O₂. (A,B) Calcium responses of gcy-35(ok769); gcy-31(ok296) (A) and gcy-35(ok769); gcy-31(ok296); BAG::gcy-35 BAG::gcy-36 sGC swap animals (B) to O₂ upshift. (C) Calcium responses of gcy-35(ok769); gcy-31(ok296); BAG::gcy-35 BAG::gcy-36 sGC swap animals to O₂ downshift. (D,E) Calcium responses of gcy-35(ok769); gcy-33(ok232); gcy-31(ok296) (D) and gcy-35(ok769); gcy-33(ok232); gcy-31(ok296); BAG::gcy-35 BAG::gcy-36 sGC swap animals (E) to O₂ upshift. (F) Average ΔF/F₀ (12–22 s) of animals with indicated genotypes. Error bars indicate SEM. Asterisks indicate significance by t-test (***P<0.0001). The numbers of recordings analyzed are indicated.