Impairing one sensory modality enhances another by reprogramming peptidergic circuits in Caenorhabditis elegans

Animals that lose one sensory modality often show augmented responses to other sensory inputs. The mechanisms underpinning this cross-modal plasticity are poorly understood. To probe these mechanisms, we perform a forward genetic screen for mutants with enhanced O2 perception in C. elegans. Multiple mutants exhibiting increased responsiveness to O2 concomitantly show defects in other sensory responses. One mutant, qui-1, defective in a conserved NACHT/WD40 protein, abolishes pheromone-evoked Ca2+ responses in the ADL chemosensory neurons. We find that ADL’s responsiveness to pre-synaptic input from O2- sensing neurons is heightened in qui-1 and other sensory defective mutants resulting in an enhanced neurosecretion. Expressing qui-1 selectively in ADL rescues both the qui-1 ADL neurosecretory phenotype and enhanced escape from 21% O2. Profiling of ADL neurons indicates its acquired O2-evoked neurosecretion is the result of a transcriptional reprogramming that up-regulates neuropeptide signalling. We show that the conserved neuropeptide receptor NPR-22 is necessary and sufficient in ADL to enhance its neurosecretion levels. Sensory loss can thus confer cross-modal plasticity by re-wiring peptidergic circuits.


25
Animals that lose a sensory modality can adapt to the loss by increasing their sensitivity to 26 other sensory inputs. This change can involve re-purposing neurons or brain areas that 27 normally mediate responses to the lost modality such that they process other sensory inputs. 28 For example, in blind people the absence of visual stimulation promotes rewiring of inputs into 29 primary visual cortex (V1), so that it becomes responsive to tactile stimuli, a characteristic 30 absent in sighted individuals (Büchel et al., 1998;Dietrich et al., 2013;Sadato et al., 1996;31 neural circuits are incompletely understood, but at some level are thought to reflect Here, we employ forward genetics to identify mechanisms that promote cross-modal plasticity 69 across the RMG hub-and-spoke circuit in response to genetic lesions. Since the circuit 70 integrates several sensory modalities, we use genetic backgrounds that reduce signalling 71 across the hub-and-spoke circuit and suppress C. elegans arousal in response to 21% O2. 72 Mutations that promote cross-modal plasticity are likely to restore its O2-responses. We 73 identify several sensory defective mutants that increase ADL's ability to transmit information 74 from its pre-synaptic O2 sensors, including URX and RMG. Specifically, the mutants show 75 increased O2-evoked secretion of neuropeptides from ADL in response to oxygen inputs. 76 Using RNAseq we profile ADL neurons in wild type and one enhancer mutant, qui-1, and 77 discover rewiring of peptidergic circuits. Reprogramming of ADL's peptidergic proprieties 78 increase expression of the neuropeptide receptor NPR-22, which is necessary and sufficient 79 to increase neurosecretion from ADL. In summary, our data suggest that loss of sensory 80 reception in the ADL pheromone sensors increases ADL's responsiveness to input from the 81 O2-circuit by reprogramming its peptidergic properties. Rewiring peptidergic signalling across 82 circuits may be an unappreciated mechanism by which loss of one sensory modality enhances 83 responsiveness to another.

85
A genetic screen for enhancers of C. elegans aggregation behavior 86 Wild isolates of C. elegans avoid and escape 21% O2 (de Bono and Bargmann, 1998). On a 87 bacterial lawn these animals move rapidly and continuously while seeking lower O2  To confirm our mutants also showed an elevated O2-escape behavior, we further selected 107 from our collection 6 mutants that displayed a larger O2-escape behavior than wild type ( Figure   108 1B -Step 2). To identify the genetic defects causing increased O2-escape behavior in these 109 mutants we used a Deep Sequence Mapping strategy (See Methods and Zuryn et al., 2010). 110 A list of de novo high impact mutations highlighted only one likely loss-of-function mutation, a 111 premature stop-codon (Q966Stop), within the qui-1 gene ( Figure S1A). Previous work 112 suggested qui-1 mutants lay eggs where bacteria are thickest (Neal et al., 2016). qui-1(db104) 113 mutants isolated in our screen displayed both aggregation and O2-escape behavior ( Figure   114 1C-1D). We next compared the O2-escape behavior of the db104 mutant with a strain carrying 115 a deletion allele, qui-1(ok3571) ( Figure S1A). These strains showed indistinguishable 116 responses ( Figure S1B), confirming qui-1 (db104) is a loss-of-function mutation that leads to 117 a strong O2-escape behavior. To confirm this further, we showed that a wild type qui-1 118 transgene completely rescued qui-1 (db104) O2-escape phenotype ( Figure 1D).

120
In the hub-and-spoke circuit, the URX O2 sensors are tonically activated by 21% O2 and in 121 turn tonically activate the RMG hub interneurons (Busch et al., 2012). Optogenetic 122 experiments show that increasing URX or RMG activity is sufficient to stimulate rapid 123 movement (Busch et al., 2012). To probe the qui-1 phenotype, we imaged O2-evoked Ca 2+ 124 responses in URX and RMG neurons in qui-1 mutants ( Figure 1B -Step 3). qui-1 Ca 2+ 125 responses did not differ from those of wild type controls ( Figure S1C and S1D), suggesting 126 that its augmented O2 escape behavior does not reflect a simple increase in URX or RMG 127 activity.

129
The NACHT/WD40 protein QUI-1 acts in the ASH and ADL spoke neurons to inhibit O2-   and pheromones (ADL). We used cell-specific rescue of qui-1 mutants to ask if QUI-1 acts in 160 ASH and/or ADL neurons to inhibit O2-evoked escape behavior. Expressing qui-1 selectively 161 in ASH neurons reduced the O2-escape response of qui-1 mutants compared to controls 162 ( Figure 2A). The rescue was not complete: transgenic animals retained a significant O2-163 response compared to wild type controls ( Figure 2A). Expressing qui-1 only in ADL also 164 significantly reduced the O2-evoked escape behavior of qui-1 mutants ( Figure 2B), but as with 165 targeted expression in ASH, rescue was incomplete and transgenic animals responded 166 significantly more to a 21% O2 stimulus than wild type controls ( Figure 2B). Expressing QUI-1 167 in both ASH and ADL neurons did not show an additive rescue effect ( Figure 2C), consistent 168 with ablation studies suggesting these neurons act redundantly to promote aggregation 169 behavior (de Bono et al., 2002). We conclude that QUI-1 acts in ASH, ADL, and potentially 170 other neurons to downregulate O2-escape behavior. but QUI-1's role in these responses is not understood. Since O2-singlling reprograms the hub-175 and-spoke circuit, including ADL neurons (Fenk and de Bono, 2017), we speculated that 176 disrupting qui-1 alters ADL properties in a way that enhances circuit output in response to O2 stimuli. To probe how loss of qui-1 alters ADL function, we first examined ADL responses to 178 pheromones. In wild type control animals ADL neurons responded to the C9 ascaroside 179 pheromone with a Ca 2+ response, as expected (Jang et al., 2012), however this response was 180 completely abolished in qui-1 mutants ( Figure 2D). This suggests that QUI-1 is required for 181 sensory transduction of pheromone stimuli.   (Lee and Ashrafi, 2008). Using this assay, we found a striking increase in ADL 207 neurosecretion in qui-1 mutants compared to wild type ( Figure 3A). Expressing wild type qui-208 1 exclusively in ADL fully rescued this enhanced neurosecretion phenotype ( Figure 3A).

209
Increased insulin secretion levels cannot be explained by increased expression from the srh-   Our experiments with qui-1;gcy-35 double mutants predict that manipulating ambient O2 levels 243 should shape ADL neurosecretion in qui-1 mutants. To investigate this hypothesis, we grew 244 wild type and qui-1 mutants at 7% and 21% O2 and assayed neurosecretion from ADL. In wild 245 type animals ADL neurosecretion was unaffected by O2 experience ( Figure 3E). By contrast, 246 neurosecretion from ADL was significantly modulated by O2 experience in qui-1 mutants 247 ( Figure 3E). Mutants kept at low O2 concentrations showed markedly less ADL neurosecretion 248 than animals kept at 21% O2 ( Figure 3E). Together, these data support the hypothesis that 249 disrupting qui-1 confers O2-evoked neurosecretion on ADL neurons. activity alters ADL neurosecretion, we assayed qui-1 and qui-1;npr-1 double mutants. We 257 observed higher levels of neurosecretion from ADL in qui-1;npr-1 double mutants ( Figure 3F).

258
Over-expressing NPR-1 215V in RMG partially rescued the ADL phenotype of qui-1;npr-1 259 double mutants ( Figure 3F). We conclude that NPR-1 signalling in RMG neurons can suppress 260 neurosecretion from ADL. Taken together, these and previous data suggest enhanced ADL 261 neurosecretion in qui-1 mutants is principally driven by increased ADL responsiveness to O2 262 input from the hub-and-spoke circuit.  Together, our data suggest that compromising sensory input increases ADL's responsiveness 294 to O2 input from the hub-and-spoke circuit. To test this, we measured ADL neurosecretion in unlikely to regulate neurosecretion directly. We next asked if enhanced neurosecretion from 298 ADL in sensory defective mutants depended on O2 input. We raised bbs-7 mutants, which 299 showed the strongest O2-escape behavior among the sensory defective mutants we had 300 studied, at 7% and 21% O2 and measured ADL neurosecretion. bbs-7 mutants grown at 7% 301 O2, when URX-RMG activity is low, lost their enhanced neurosecretion phenotype and 302 showed secretion levels indistinguishable from wild type animals reared at 7% O2 ( Figure 4E).

303
These data suggest that ADL neurons release more DCVs in bbs, wrt-6, and fig-1 mutants 304 than wild type animals in response to input from the O2 circuit.  neurosecretion. Consistent with this, when we selected all known genes associated with DCV 328 release (Hobert, 2006) that were also differentially regulated, almost all them were elevated 329 in qui-1 mutants ( Figure 5A). This is likely to sustain the higher rate of neurosecretion. The 330 absence of qui-1 also reprogramed chemosensory receptor levels: more than half of all the 331 chemoreceptors expressed in ADL were differentially regulated ( Figure S5E and S5F). We 332 conclude that loss of qui-1 largely reprograms ADL properties and enhances genes associated 333 with synaptic release.  Here, we use forward genetics as our entry point to sought mechanisms that increase C.  Figure S1C and S1D), this suggests that in qui-1 mutants ADL neurons are more 389 sensitive to incoming pre-synaptic activity.

390
Selectively expressing gcy-35 cDNA in the URX O2-sensor partially rescues the ADL 391 neurosecretion phenotype of qui-1;gcy-35 mutants ( Figure 3D), confirming that URX helps 392 drive increased ADL neurosecretion in qui-1 mutants. Disrupting the inhibitory neuropeptide receptor npr-1 further increases ADL neurosecretion in qui-1 mutants, and this phenotype is 394 partially rescued by expressing npr-1 cDNA specifically in RMG interneurons ( Figure 3F). This 395 adds further support to our model, and indicates that activity from the hub-and-spoke circuit 396 propagates from URX-RMG to ADL to stimulate neurosecretion. Consistent with this, in qui-    (Brenner, 1974). To isolate mutants that preferentially aggregated on thick      responses, a pseudo-ratio of GCaMP6s over mKate2 signal was computed to account for 576 changes in GCaMP6s intensity due to animal movement. O2-evoked Ca 2+ responses were 577 calculated using the average GCaMP6s/mKate2 signal for a 15 sec window centred around 578 the peak of the response (R) and a 30 sec window before stimulation with 21% O2 (R0) and 579 expressed as . (Nikon) wide-field microscope using a 10x air lens. Z-stack images were taken at sub-603 saturating exposure for both GFP and mCherry intensities and analysed using ImageJ.

604
Coelomocytes were delineated using the GFP signal and the mCherry signal measured in the 605 same area. Values were plotted as arbitrary units of intensity.
srh-220p validation: To validate the srh-220 promoter construct we imaged wild type and qui-608 1 mutants carrying a transgene expressing mKate under the control of the srh-220 promoter 609 (srh-220p:mKate). We imaged ADL cell body using a Ti2 (Nikon) wide-field microscope using 610 a 40X air lens. Z-stack were taken avoiding saturating mKate signal. The boundary of ADL 611 cell body was taken and intensities extracted using ImageJ. Data were plotted using Prism. Analysis (GSA). Beside the classic GO term annotation, we functionally annotated C. elegans 646 neural genes using annotations from previously published reviews (Hobert, 2006;Robertson 647 and Thomas, 2006) and used these in the same way as described for GO term analysis. These 648 annotations were also used to extract data for particular classes of genes such as "Synaptic  The hub and spoke circuit associated with the URX O2 sensors and O2 escape behavior. Updated version according to (Cook et al., 2019). (B) Schematic of the genetic screen. We selected mutants that preferentially accumulate on thicker bacteria, a behaviour that depends on oxygen responses (Step 1), screened these for increased O2-escape behavior (Step 2), and then identified strains with overtly normal O2-evoked Ca 2+ responses in the URX O2-sensors and RMG interneurons (Step 3). (C) Bar graphs quantifying aggregation and bordering behavior. N=4-6 assays. (D) A wild type copy of qui-1 rescues the O2-escape phenotype of qui-1(db104) mutants. Left: Line shows average speed, shading shows standard error of the mean (SEM) and grey bars show 30s time intervals used to average the animal's speed. Right: Bar graph shows the fold change in average speed at 21% O2 compared to 7% O2 N= 6-9 assays. (E) QUI-1 expression and localization in an mNeonGreen::qui-1 translational fusion knock-in strain. Fluorescent neurons include ADL, ASH (Head) and PVQ, PHB and PHA (Tail), and potentially M3, AWB and ASJ based on position and morphology. Also visible is yellow gut autofluorescence. Statistics:**=p value ≤0.01, ***=p value ≤0.001, ns=not significant, Mann-Whitney U test. Comparisons are with WT.  . qui-1 acts in ADL and ASH chemosensory neurons to inhibit O2-escape behavior, and it is required for pheromoneevoked Ca 2+ responses in ADL.