Temperature, Oxygen, and Salt-Sensing Neurons in C. elegans Are Carbon Dioxide Sensors that Control Avoidance Behavior

Summary Homeostatic control of body fluid CO2 is essential in animals but is poorly understood. C. elegans relies on diffusion for gas exchange and avoids environments with elevated CO2. We show that C. elegans temperature, O2, and salt-sensing neurons are also CO2 sensors mediating CO2 avoidance. AFD thermosensors respond to increasing CO2 by a fall and then rise in Ca2+ and show a Ca2+ spike when CO2 decreases. BAG O2 sensors and ASE salt sensors are both activated by CO2 and remain tonically active while high CO2 persists. CO2-evoked Ca2+ responses in AFD and BAG neurons require cGMP-gated ion channels. Atypical soluble guanylate cyclases mediating O2 responses also contribute to BAG CO2 responses. AFD and BAG neurons together stimulate turning when CO2 rises and inhibit turning when CO2 falls. Our results show that C. elegans senses CO2 using functionally diverse sensory neurons acting homeostatically to minimize exposure to elevated CO2.


Supplementary Figure 3. Expression patterns of the six C. elegans -carbonic anhydrase family members
A. Upstream promoter regions from cah-1, 2, 3, 4, 5 and 6 fused to gfp reveal differential expression of cah genes in head and tail neurons, the intestine, the hypodermis and the excretory cell. Shown are confocal gfp z projections and transmitted white light images of L2 animals and of young adult heads for cah-1, 2 and 3. cah-1::gfp labels ciliated head sensory neurons, head ganglia, the ventral nerve cord and tail neurons. cah-2::gfp labels a set of 4 quadrant neurons in the head, BAGL/R (see (B), below), other head neurons, the mid-body canal neurons CANL and CANR and tail neurons. cah-3::gfp intensely labels head neurons (note these are likely interneurons as gfp-filled dendrites appear absent), the ventral nerve cord and the intestine. cah-4::gfp strongly labels the hypodermis (excluding the seam cells) and the excretory cell. cah-5::gfp labels head neurons and the intestine in L2 animals but neuronal expression in the head was absent by the young adult stage (not shown).
cah-6::gfp strongly labels two bilateral pairs of head neurons and a pair of tail neurons in L2 animals and in adults (not shown). Scale bar, 50 m.

across % CO 2 increases and decreases
A -F. Statistical analysis of changes in speed across 0 -5% CO 2 increases and 5 -0% CO 2 decreases of wild type and neuron-ablated animals off food. Average speed traces, left, time-averaged speeds before and after gas switches, middle, and average changes in speed across gas switches, right. (A, D) Average speed during 0 -5% and 5 -0% gas switches. Error bars omitted for clarity. (B, E) Time-averaged speeds before (red bars) and after (dark grey bars) an increase (B) or decrease (E) in % CO 2 .
Intervals for comparison coincide with peaks in wild type behaviour. Error bars indicate SEM. (C, F) Average change in speed across an increase (C) or a decrease (F) in % CO 2 . Error bars indicate SEMs calculated from SEM values in (B, E) using error propagation formulae. Significance markers indicate comparisons with wild type, unless otherwise indicated.

Transgenic strain construction
Expression constructs were made using the Invitrogen Multisite Gateway 3- For expression studies of the six C. elegans -carbonic anhydrase family members cah-1 to cah-6, promoter fragments were cloned into Gateway expression vectors together with gfp and the unc-54 3' UTR. Promoter fragments from the cah-2,

Fabrication of Microfluidic Devices
Microfluidic devices were fabricated from polydimethylsiloxane (PDMS) using standard soft lithography technology (Xia and Whitesides, 1998;Qin et al., 2010). Devices were designed in AutoCAD (Autodesk) and printed on a film photomask (Photo Data Ltd, UK) at 128000 dots per inch (dpi

Details of temperature controlled Ca 2+ imaging
For imaging AFD at different temperatures (Supplementary Figure 1E, F)

Determination of the tax-2(p694) deletion limits
The tax-2(p694) mutant strain PR694 was obtained from the Caenorhabditis Genetics Centre. The forwards primer p694F4 GTTGATCGGTTGACAATCAGTAG and reverse primer p694R4 GCTCGAAGTAGCCCAAACATTTC amplify a 2.2 Kb fragment across the p694 lesion. Sequencing this PCR product with 4 internal primers revealed p694 to be a 390 bp deletion and a 36 bp insertion, compared with the previously reported ~ 1.6 Kb deletion (Coburn and Bargmann, 1996). p694 removes 264 bp of sequences upstream of tax-2, exon 1 (81 bp), 45 bp of the first intron and introduces a 36 bp A/T rich insert. This definition is confirmed by primers p694F5 ATGATGACTGCTTGGCAACGGAC and p694R4 that flank the deletion and by p694F7 TCATTTCTTTGCGTCTCCTTG that lies within the deletion. A p694F5/R4 PCR yields a 697 bp product from wild type and a 342 bp product from p694 template. A p694F7/R4 PCR yields a 480 bp product from wild type and no product from p694 template.