A, The aversive olfactory learning ability of wild type animals and chemosensory mutants. The learning ability of mutants and wild type animals were compared with two tailed Student’s t-test and multiple comparisons were adjusted with Bonferroni correction (Data are presented as mean +/− SEM. *** P < 0.001, n ≥ 6 assays, error bars: SEM). B, Expression of osm-6 cDNA in AWB and AWC olfactory sensory neurons rescues the learning defect of osm-6 mutants, but the expression of osm-6 cDNA in the AWC or in the gustatory neurons ASE does not rescue. The learning ability of transgenic animals was compared with that of non-transgenic siblings with paired two tailed Student’s t-test (Data are presented as mean +/− SEM. *** P < 0.001, n ≥ 4 assays, error bars: SEM, N.S.: not significant). C, A candidate neuronal network downstream of sensory neurons AWB and AWC to regulate aversive olfactory learning. Please also see Supplemental Figure 2 and Supplemental Table 1 and 2.

A, B, In the microdroplet assay, laser ablation of RIA generated significant defects in the trained choice index (A) and learning (B) without affecting naïve choice index. The choice indexes or the learning index of ablated animals were compared with that of mock control animals with two tailed Student’s t-test (Data are presented as mean +/− SEM. *** P < 0.001, n ≥ 9 assays, error bars: SEM). C, D, In the two-choice assay, killing RIA generated significant defects in the trained choice index (C) and learning (D) without affecting naïve choice index. The choice indexes or the learning indexes of transgenic animals and non-transgenic siblings were compared with that of wild type animals with two tailed Student’s t-test. Multiple comparisons were adjusted with Bonferroni correction (Data are presented as mean +/− SEM. *** P < 0.001, ** P < 0.01, n ≥ 3 assays, error bars: SEM). E, F, In both the microdroplet assay (E) and the two-choice assay (F), osm-6 mutants exhibited significant defects in the trained choice index and expression of osm-6 cDNA in AWB and AWC neurons rescued the defects. The choice indexes of transgenic animals and non-transgenic siblings were compared with that of wild type animals with two tailed Student’s t-test. Multiple comparisons were adjusted with Bonferroni correction (Data are presented as mean +/− SEM. *** P < 0.001, ** P < 0.01, n ≥ 6 assays, error bars: SEM).

A–F, G-CaMP calcium response of AWC^ON (A–C) and AWB (D–F) neurons in trained animals when stimuli alternated between OP50-condtioined medium and PA14-conditioned medium in the order of OP50-PA14-OP50 (A, D), or when stimuli alternated between buffer and OP50-condtioined medium (B, E), or between buffer and PA14-condtioined medium (C, F). The red traces indicate the average percentage changes in G-CaMP intensity and the grey areas around the traces indicate SEM. In A–F, the calcium signals within 3 second window immediately before the switch of the stimuli and the calcium signals within 3 second window that begins 1.5 second after the switch were quantified. Values were compared using a paired two-tailed Student’s t-test (*** P < 0.001, ** P < 0.01, n ≥ 5 animals). Compared to the calcium response in naïve animals (Figure 5A–F), training had no effect under each condition. G, Effects of neuronal ablations on the turning rates exhibited by trained animals when stimuli alternated between the smell of OP50 and PA14. The turning rates of ablated animals were compared with that of matched mock animals with two tailed Student’s t-test (Data are presented as mean +/− SEM. *** P < 0.001, ** P < 0.01, * P < 0.05, n ≥ 6 assays, error bars: SEM). H, In trained animals, RIA interneurons and SMD motor neurons regulate the turning rate towards the smell of PA14 to generate the trained preference. The AWB-AWC sensorimotor circuit is highlighted in blue and the ADF modulatory circuit is highlighted in pink. Please also see Supplemental Figure 4.

A, Training protocol for aversive olfactory learning on adult animals. B, Diagram of the olfactory microdroplet assay. One representative image and data output of 12 animals in one assay are shown. Results are vertically arranged on raster panels and each dot represents one Ω bend. Each row displays the result of one cycle of olfactory assay on one animal and each animal was tested for 12 cycles continuously in every assay. C, The aversive olfactory learning on the smell of PA14 in the microdroplet assays. Turning rates towards OP50 were compared with turning rates towards PA14 in naïve and trained animals with two tailed Student’s t-test. The naïve choice index was compared with the trained choice index with two tailed Student’s t-test. (Data are presented as mean +/− SEM. *** P < 0.001, n > 50 assays, error bars: SEM, N.S.: not significant). D, Wild type animals generate frequent turns when exposed to the clean air that passes mineral oil and carries no odorant (error bars: SEM). E, The aversive olfactory learning ability of wild type animals and mutants defective in neurotransmitter biosynthesis in the microdroplet assay. The learning ability of mutants was compared with that of wild type animals and the P values were calculated by two tailed Student’s t-test. Multiple comparisons were adjusted with Bonferroni correction (Data are presented as mean +/− SEM. * P < 0.05, *** P < 0.001, n ≥ 6 assays, error bars: SEM). F, The mod-1(ok103) mutant is defective in learning in the microdroplet assay. The learning ability of mod-1 was compared with that of wild type animals and the P value was calculated by two tailed Student’s t-test (Data are presented as mean +/− SEM. * P < 0.05, n ≥ 8 assays, error bars: SEM). Please also see Supplemental Figure 1.

A, Ablation of AWB sensory neurons abolished the naïve olfactory preference and learning. B, Ablation of RIA interneurons abolished the learned olfactory preference and learning without altering the naïve preference. C – E, Effects of ablating different neuronal types on the naïve olfactory preference (C), the learned olfactory preference (D) and learning (E). For A – E, choice indexes and learning indexes of ablated animals were compared with that of matched mock animals with two tailed Student’s t-test. (Data are presented as mean +/− SEM. *** P < 0.001, ** P < 0.01, * P < 0.05, n ≥ 6 assays, P values between 0.05 and 0.1 were indicated above the bar graphs. error bars: SEM). F, Two different neural circuits for the naïve and learned preferences. The AWB-AWC sensorimotor circuit is highlighted in blue and the ADF modulatory circuit is highlighted in pink.

A–F, G-CaMP calcium response of AWC^ON (A–C) and AWB (D–F) neurons in naïve animals when stimuli alternated between OP50-condtioined medium and PA14-conditioned medium in the order of OP50-PA14-OP50 (A, D), or when stimuli alternated between buffer and OP50-condtioined medium (B, E), or between buffer and PA14-condtioined medium (C, F). The blue traces indicate the average percentage changes in G-CaMP intensity for multiple recordings and the grey areas around the traces indicate SEM. In A–F, the calcium signals within 3 second window immediately before the switch of the stimuli and the calcium signals within 3 second window that begins 1.5 second after the switch were quantified. Values were compared using a paired two-tailed Student’s t-test (*** P < 0.001, ** P < 0.01, * P < 0.05, n ≥ 5 animals). G, Effects of neuronal ablations on turning rate exhibited by naïve animals when stimuli alternated between the smells of OP50 and PA14. The turning rates of ablated animals were compared with that of matched mock animals with two tailed Student’s t-test (Data are presented as mean +/− SEM. *** P < 0.001, ** P < 0.01, * P < 0.05, n ≥ 6 assays, error bars: SEM). H, In naïve animals, differential neuronal responses of AWC sensory neurons to the smell of OP50 and PA14 propagate through downstream circuit to generate preference. The AWB-AWC sensorimotor circuit is highlighted in blue and the ADF modulatory circuit is highlighted in pink. Please also see Supplemental Figure 4.