A collective modulatory basis for multisensory integration in C. elegans

In the natural environment, animals often encounter multiple sensory cues that are simultaneously present. The nervous system integrates the relevant sensory information to generate behavioral responses that have adaptive values. However, the signal transduction pathways and the molecules that regulate integrated behavioral response to multiple sensory cues are not well defined. Here, we characterize a collective modulatory basis for a behavioral decision in C. elegans when the animal is presented with an attractive food source together with a repulsive odorant. We show that distributed neuronal components in the worm nervous system and several neuromodulators orchestrate the decision-making process, suggesting that various states and contexts may modulate the multisensory integration. Among these modulators, we identify a new function of a conserved TGF-β pathway that regulates the integrated decision by inhibiting the signaling from a set of central neurons. Interestingly, we find that a common set of modulators, including the TGF-β pathway, regulate the integrated response to the pairing of different foods and repellents. Together, our results provide insights into the modulatory signals regulating multisensory integration and reveal potential mechanistic basis for the complex pathology underlying defects in multisensory processing shared by common neurological diseases. Author Summary The present study characterizes the modulation of a behavioral decision in C. elegans when the worm is presented with a food lawn that is paired with a repulsive smell. We show that multiple sensory neurons and interneurons play roles in making the decision. We also identify several modulatory molecules that are essential for the integrated decision when the animal faces a choice between the cues of opposing valence. We further show that many of these factors, which often represent different states and contexts, are common for behavioral decisions that integrate sensory information from different types of foods and repellents. Overall, our results reveal a collective molecular and cellular basis for integration of simultaneously present attractive and repulsive cues to fine-tune decision-making.

a repulsive odorant. We show that distributed neuronal components in the worm nervous

23
system and several neuromodulators orchestrate the decision-making process, suggesting that 24 various states and contexts may modulate the multisensory integration. Among these 25 modulators, we identify a new function of a conserved TGF-β pathway that regulates the 26 integrated decision by inhibiting the signaling from a set of central neurons. Interestingly, we find 27 that a common set of modulators, including the TGF-β pathway, regulate the integrated 28 response to the pairing of different foods and repellents. Together, our results provide insights 29 into the modulatory signals regulating multisensory integration and reveal potential mechanistic 30 basis for the complex pathology underlying defects in multisensory processing shared by 31 common neurological diseases.

33
Author Summary 34 The present study characterizes the modulation of a behavioral decision in C. elegans when the 35 worm is presented with a food lawn that is paired with a repulsive smell. We show that multiple 36 sensory neurons and interneurons play roles in making the decision. We also identify several 37 modulatory molecules that are essential for the integrated decision when the animal faces a 38 choice between the cues of opposing valence. We further show that many of these factors,

39
which often represent different states and contexts, are common for behavioral decisions that

Introduction 44
An environment is often represented by numerous sensory cues. For example, a tainted food 45 source can produce both attractive and repulsive odorants. In order to better survive, an animal 46 often needs to detect and process simultaneously present sensory cues to make a behavioral decision [1][2][3][4][5][6][7][8]. Because integrating multiple sensory cues generates a more accurate evaluation neuropeptides, mediate many of these neurological effects on decision-making [3,4,[12][13][14].

57
Intriguingly, patients of several neurological diseases, including autism spectrum disorder,

61
Together, these studies reveal multisensory integration as a common neuronal and behavioral 62 process modulated by multiple contexts across the animal kingdom and highlight the importance 63 of understanding the underlying mechanisms in normal as well as disease states. . Interestingly, all of these mutants exhibited wild-type behavioral decision when 169 they were exposed to 2-nonanone on an OP50 food lawn ( Figure 3A-D and S6 Figure). These 170 results show that serotonin, dopamine, tyramine or octopamine are not required for 2-  Figure 5D and 5E). Together, these results indicate that the activity and the synaptic output of 265 the AIY interneurons promote the decision to leave the food lawn paired with 2-nonanone, while 266 AIB is dispensable for the decision-making. Next, we examined transgenic animals that lawn more than wild type ( Figure 5F and 5G). However, these transgenic animals are normal in 272 2-nonanone avoidance in the absence of food or spontaneous food leaving. They also do not 273 reach the edge of the lawn more rapidly than wild type (S1-3 Tables). Together, these results

274
show that different downstream neurons modulate the decision to leave the repellent tainted food lawn in opposite ways by promoting or inhibiting the decision-making process. These 276 neurons may act as the convergent sites to process multiple sensory signals in order to 277 generate specific behavioral outputs.

279
Multisensory integration is regulated by a common set of modulators lawn and then in about 10-15 minutes started to repel the animals off the food lawn ( Figure 6A 288 and S10 Movie). Interestingly, 1-octanol failed to stimulate food leaving under our experimental 289 conditions ( Figure 6A). We also paired a lawn of Comamonas sp with 100% 2-nonanone.

290
Comamonas is an attractive food source for C. elegans [41]. We found that pairing a

291
Comamonas bacterial lawn with 100% 2-nonanone repelled C. elegans off the lawn similarly as 292 the OP50 lawn paired with 2-nonanone ( Figure 7A). Interestingly, we found that several 293 modulators, particularly TGF-β/DAF-7, the TGF-β receptor DAF-1, and the sensory neurons 294 ASK, that regulated the integrated response to an OP50 lawn paired with 100% 2-nonanone 295 also similarly regulated the integrated response to OP50 lawn paired with 100% benzaldehyde  320 elegans is exposed to an attractive food lawn concurrently with a repulsive odorant. We confirm 321 the requirement of the AWB sensory neuron that is known to sense repellents, including 2-  Table),

371
suggesting a specific function of nlp-7 and daf-7 in regulating the decision-making process. inhibiting the activity of RIM and RIC, or blocking the synaptic outputs of these neurons, or 399 disrupting the biosynthesis of the common neurotransmitter of these neurons, tyramine, does 400 not significantly change the decision-making process. However, disrupting the production of 401 tyramine, but not octopamine, in these neurons suppresses the slow-decision phenotype in the daf-1 or daf-7 mutant animals (Figure 3 and 4). Together, these results indicate that the DAF-403 7/DAF-1 pathway promotes the decision to leave by inhibiting the tyramine signaling from these 404 interneurons. This regulatory mechanism of DAF-7 is reminiscent of that in feeding, where DAF-405 7 promotes the pumping rate by inhibiting the output from the RIM and/or RIC neurons.

406
However, different from the function of DAF-7 in regulating feeding that is dependent on

(B)
The time course for worms leaving OP50 lawn that is paired with 2-nonanone of different 466 concentrations over 60 minutes, n = 2 assays for 10% and n = 3 assays each for 30%, 50% and 467 100%.

(D)
The time taken for worms to reach the edge of the OP50 food lawn when the lawn is paired 472 with either 10% or 100% 2-nonanone, n = 2 assays each.

473
(E) I -IV, Sample images of wild-type animals leaving an OP50 lawn that is paired with 100% 2-474 nonanone at different time points of the 60-minute assay.

(E-K)
Mutations in the genes encoding the neuropeptide processing enzymes, kpc-1 (E, n = 4 499 and 5 assays for wild type and kpc-1 mutant, respectively), or egl-3 (F, n = 4 and 6 assays for 500 wild type and egl-3 mutant, respectively), or a TGF-β -encoding gene daf-7 (G, n = 6 and 5 501 assays for wild type and daf-7 mutant, respectively), or a neuropeptide-encoding gene nlp-7 (H, leave the OP50 food lawn paired with 2-nonanone, and expressing the genomic DNA of nlp-7 (I, 504 n = 6, 7 and 4 assays for wild type, transgenic animals and non-transgenic siblings, 505 respectively) or daf-7 (J, n = 4 assays each for wild type, transgenic animals and non-transgenic 506 siblings) rescues the delayed food leaving phenotype of the respective mutant animals.

K)
Expressing the wild-type daf-7 cDNA in the sensory neurons ADE rescues the delayed 508 decision in the daf-7(e1372) mutant animals, n = 4 assays each for wild type, transgenic 509 animals and non-transgenic siblings, respectively.

510
(L) Expressing the wild-type daf-7 cDNA in the sensory neurons OLQ also rescues the delayed 511 decision in the daf-7(e1372) mutant animals, n = 3 assays for wild type, 3 assays for transgenic 512 animals and 2 assays for non-transgenic siblings, respectively.

513
Each bar graph reports the average percentage of worms outside the lawn 15 minutes after the 514 start of the assay, mutants are compared with wild-type animals and transgenic animals are 515 compared with non-transgenic siblings using Student's t-test, * p ≤ 0.05, ** p ≤ 0.01, ***p ≤ 516 0.001, n.s., not significant.

(A-D)
Mutating daf-1 that encodes the type I TGF-β receptor delays the decision to leave the 521 OP50 lawn paired with 100% 2-nonanone (A, n = 8 assays each), and expressing the genomic 522 DNA of daf-1 (B, n = 6 assays each) or a wild-type daf-1 cDNA in the RIM and RIC neurons (C, 523 n = 6, 6 and 5 assays for wild type, transgenic animals and non-transgenic siblings, 524 respectively) in the daf-1(m40) mutant animals rescues the delayed decision, but expressing 525 wild-type daf-1 in the sensory neurons (D, n = 3, 5 and 2 assays for wild type, transgenic 526 animals and non-transgenic siblings, respectively) does not rescue. Mutants are compared with 527 wild type and transgenic animals are compared with non-transgenic siblings with Student t-test. gated chloride channel (E, n = 2 and 4 assays for wild type and transgenic animals, 530 respectively) or blocking the synaptic release from these neurons by selectively expressing the 531 tetanus toxin (F, n = 2 assays each) does not alter the decision to leave the OP50 food lawn 532 paired with 100% 2-nonanone. Transgenic animals are compared with wild type.

542
Double mutants were compared with the respective single mutants using student's t test.

543
Each bar graph reports the average percentage of worms outside the lawn 15 minutes after the 544 start of the assay. Mean ± SEM, ** p ≤ 0.01, *** p ≤ 0.001, n.s.; not significant.  transgenic animals were compared with non-transgenic siblings or wild-type animals tested in 635 parallel on the same day.

637
The bar graphs in the figures report the percentage of worms outside the lawn at the time point 638 when the significant difference between the tested genotypes was first observed. When there 639 was no significant difference, the bar graphs report the percentage of worms outside the lawn 640 15 minutes after the start of the assay.

657
transgene was injected at 30-50 ng/µl with the co-injection marker as previously described

661
Lawn-leaving assay was performed and analyzed similarly as the assay for multisensory 662 integration, except that no repulsive chemical was present. Briefly, animals were placed on a 1 663 cm-diameter round-shaped bacterial lawn of OP50 and left for 10 minutes to acclimatize before 664 examining food leaving over a period of one hour by counting the number of worms that were 665 present on the food lawn every 5 minutes for a total of 60 minutes.