key: cord-0317033-1uahaoam
authors: Revilla-i-Domingo, Roger; Veedin Rajan, Vinoth Babu; Waldherr, Monika; Prohaczka, Günther; Musset, Hugo; Orel, Lukas; Gerrard, Elliot; Smolka, Moritz; Farlik, Matthias; Lucas, Robert J.; Raible, Florian; Tessmar-Raible, Kristin
title: Characterization of cephalic and non-cephalic sensory cell types provides insight into joint photo- and mechanoreceptor evolution
date: 2021-01-11
journal: bioRxiv
DOI: 10.1101/2021.01.10.426124
sha: 41425d379e3156dba39a15551f1f0c37667dcc80
doc_id: 317033
cord_uid: 1uahaoam

Rhabdomeric Opsins (r-Opsins) are light-sensors in cephalic eye photoreceptors, but also function in additional sensory organs. This has prompted questions on the evolutionary relationship of these cell types, and if ancient r-Opsins cells were non-photosensory. Our profiling of cephalic and non-cephalic r-opsin1-expressing cells of the marine bristleworm Platynereis dumerilii reveals shared and distinct features. Non-cephalic cells possess a full set of phototransduction components, but also a mechanosensory signature. We determine that Pdu-r-Opsin1 is a Gαq-coupled blue-light receptor. Profiling of cells from r-opsin1 mutants versus wild-types, and a comparison under different light conditions reveals that in the non-cephalic cells, light – mediated by r-Opsin1 – adjusts the expression level of a calcium transporter relevant for auditory mechanosensation in vertebrates. We establish a deep learning-based quantitative behavioral analysis for animal trunk movements, and identify a light-and r-Opsin-1-dependent fine-tuning of the worm’s undulatory movements in headless trunks, which are known to require mechanosensory feedback. Our results suggest an evolutionary concept in which r-Opsins act as ancient, light-dependent modulators of mechanosensation, and suggest that light-independent mechanosensory roles of r-Opsins likely evolved secondarily.

INTRODUCTION results are consistent with the idea that photo-and mechanosensory systems 148 have a common evolutionary origin in a multimodal sensory cell type. 149

Shared and distinct molecular signatures of eye photoreceptor and 151 trunk r-opsin1-expressing cells 152 In order to gain insights into the molecular signatures of EP and TRE cells, we 153 established a mechanical dissociation protocol compatible with FACS, and 154 benchmarked to minimize cell death. We next dissected heads and trunks of 155 the same pMos{rops::egfp} vbci2 individuals (Fig. 1A) , isolated EGFP-positive 156 cells from heads and trunks, and established transcriptomes for both sorted 157 and unsorted cells using Illumina HiSeq sequencing on cDNA amplified by the 158 Smart-Seq2 protocol [25] (Fig. 1B) . Gates for FACS (Fig. 1C,D) were 159 calibrated using dissociated cells from wildtype heads ( Fig.1-figure  160 supplement 1A) and trunks ( Fig.1-figure supplement 1B) , to exclude 161 isolation of autofluorescent cells. 162

To validate the sampling strategy, we investigated if this procedure 163 reproduced expected results for genes known to be enriched in both EP and 164 TRE cells. Both r-opsin1 and egfp were up to several thousand times more 165 abundant in libraries derived from EGFP-positive cells than in those of 166 unsorted cells (Fig. 1E,F) . In further support of successful enrichment, 167 signatures of EGFP-positive cells were consistently enriched in the gq gene 168 encoding the G alpha subunit Gq (Fig. 1G) . Gq was previously shown to be 169 strongly expressed in EP and TRE cells [17] . By contrast, the genes encoding 170 the ribosomal subunit Rps9 or the Polo-like Kinase Cdc5, previously 171 established as internal controls for gene expression quantification 172 experiments [26] , were not enriched in either EP or TRE cell populations (Fig.  173   1-figure supplement 1C,D) . 174

As these results indicated that the experimental procedure allowed for 175 significant enrichment and profiling of EP and TRE cells, we next used EdgeR 176 [27] to systematically calculate enrichment scores for each of the EGFP-177 positive populations, compared to the combined set of head and trunk 178 unsorted cells. From a total of 39575 genes, we determined a set of 278 179 genes (0.7%) to be significantly enriched in EP cells, and a set of 361 genes 180 (0.9 %) significantly enriched in TRE cells (FDR < 0.05) ( Fig. 2A) . 133 genes 181 (0.3 % of total) were shared between the EP and TRE cells (common EP-182 /TRE-enriched genes), including, expectedly, r-opsin1 and gq (Figure 2-183 figure supplement 2), and leaving 145 (0.4 % of total) EP-specific genes, 184 and 228 (0.6 % of total) TRE-specific genes ( Fig. 2A) As to potential differences between EPs and TREs, prior analyses had 204 pointed to the expression of circadian clock genes in the EPs and the 205 adjacent brain lobes [26] , and both classical and molecular studies suggested 206 the retina as a site of continuous neurogenic activity [31, 32] , contrasting with 207 the appearance of the TREs as sparse, differentiated neurons [17] . In line with 208 these expectations, we found the EP-, but not TRE-derived transcriptomes to 209 be enriched, respectively, in the circadian clock gene bmal, as well as a 210 homolog of the embryonic lethal, abnormal vision/elav gene, a marker 211 characteristic for committed neurons [33, 34] . 212 we found that putative homologs of 9 and 8 of the 12 components of the r-235

Opsin phototransduction pathway are enriched in the P. dumerilii EP and TRE 236 cells, respectively (Fig. 2B) . Statistical analysis with 10 4 random gene sets of 237 matching size (see Methods) revealed these results to be highly significant (p 238 < 10 -4 , for both EP and TRE). Of note, all 12 components of the r-Opsin 239 phototransduction pathway were found to be expressed in the TRE cells of P. Following the same strategy, to further explore potential additional functions of 244 the TRE cells, we next tested the molecular relationship between the worm's 245 TREs and the r-Opsin-expressing, mechanosensory JO neurons of 246

Drosophila. For this, we took advantage of 101 genes identified as JO neuron 247 specific in a microarray analysis [15] , and 80 P. dumerilii homologs of these. 248

Significant subsets of these were found in the common EP-/TRE-enriched 249 signature (9 genes; p < 10 -4 ), and the TRE-specific signature (7 genes; p = 250 1.3x10 -3 ) (Fig. 3A) . The common EP-/TRE-enriched genes essentially reflect 251 the P. dumerilii homologs of the aforementioned phototransduction pathway 252 (rh3/rh4, rh5/rh6, trp/trpl, norpa, gβ76c, pip5k59b, arr2 and klp68D; Fig. 3B Fig. 3B ), whereas the other 5 have not been tested for mechanosensory 265 functions. In order to compensate for this lack of functional information, we 266 also performed a comparison with mouse, where the largest number of genes 267 involved in mechanosensation is known. We systematically determined 268 putative P. dumerilii homologs of all mouse genes assigned to the gene 269 ontology (GO) category "sensory perception of mechanical stimulus" 270 (GO:0050954), and then assessed their overlap with EP or TRE expressed 271 genes (Fig. 3F,G) . Indeed, these homologs are significantly overrepresented 272 in the TRE-specific signature (p=0.029; Fig. 3G ). Similar analyses with GO 273 categories for other sensory perception modalities associated to the JO, such 274 as "sensory perception of temperature stimulus" (GO:0050951) and "sensory 275 perception of pain" (GO:0019233), showed no statistically significant results 276 ( Fig. 3H,I) . 277 A closer analysis of those mouse mechanosensory genes whose bristleworm 278 counterparts are expressed in TRE cells (Fig. 3-figure supplement 2) , 279  points at a gene signature shared between TRE cells and mouse inner ear  280   hair (IEH) cells: 18 out of the 19 TRE-specific gene homologs have reported  281 effects on hearing function in the mouse (yellow shading in Fig. 3-figure  282 supplement 2 that also express fish r-Opsin orthologs [17] . activates Gα q , similar to Drosophila r-Opsins. 329

The relative responsiveness of a photoreceptor cell to different wavelengths of 330 light is a fundamental determinant of its sensory capabilities. We therefore 331 next determined the spectral sensitivity of P. dumerilii r-Opsin1, using our 332 HEK293 cells second messenger assay to measure changes in calcium 333 concentration in response to near monochromatic stimuli spanning the visible 334 spectrum ( Fig.4-figure supplement 1C-F) . The EC 50 values (irradiance 335 required to elicit 50% response; see fitted with an opsin:retinaldehyde pigment template function [54] . The optimal 338 λ max for the template was determined by least squares as 471 nm (Fig. 4B,C) . We next identified the genes differentially expressed between the EP or TRE 363 cells of mutant vs non-mutant worms using the EdgeR algorithm. Genes with 364 an FDR < 0.05 were considered significantly differentially expressed. We then 365 focused on the P. dumerilii homologs of all mouse hearing genes that were 366 expressed in either EP or TRE cells of mutant or non-mutant worms. In the 367 EP cells, none of these candidate genes was significantly differentially 368 expressed between mutant and non-mutant worms. By contrast, one gene 369 (atp2b/c7424, P. dumerilii homolog of mouse atp2b2; In a first experiment to assess the possible requirement of r-opsin1 for 422 coordinated segmental movements, we assessed the crawling movement 423 exhibited by decapitated trunks when stimulated by a focal bright light 424 stimulus [17] . Transheterozygous r-opsin1 1/17 individuals clearly responded 425 to such stimuli, but exhibited a significantly reduced net distance when 426 compared to wild-type animals (p=0.02; Wilcoxon rank sum test; Fig.5 - figure  427 supplement 2). Whereas this result is consistent with the notion that r-Opsin1 428 is involved in the correct execution of motor movements, the experiment does 429 not discriminate between r-Opsin1 triggering the response and/or modulating 430 its motor execution. 431

We therefore decided to investigate a very regularly performed behavior that 432 does not require light as a stimulus. Annelids from the Platynereis genus 433 exhibit a stereotypical undulatory behavior that is thought to increase water 434 flow and oxygenation [62] . The presence of this behavior in Platynereis 435 dumerilii is seemingly independent of time [26] , and requires a tight 436 coordination between segments. Thus, we reasoned that if r-Opsin1 in the 437 segmentally arranged TRE cells plays a role in the modulation of motor 438 movements, this behavior presents a good test. We recorded the movement 439 of r-opsin1 mutant and wild-type trunks of de-capitated worms for five 440 consecutive days, using a previously established infrared video system [63] . Analyses on a total of 64 trunks revealed that wild-type (black graphs) 459 exhibited a light-dependent modulation of the undulatory movements, which 460 were higher during darkness ( Fig. 5F-H) . This modulation was abolished in r-461 ops1-/-worms, whose trunks exhibited equally high undulatory movements 462 during light and dark ( Fig.5F-H From an ecological perspective, a light-modulatory function could effectively 542 serve to adjust mechanosensory functions in species exposed to varying light 543 conditions, allowing them to tune mechanoreceptive responses to ambient 544 light. Whereas our functional results are restricted to the bristleworm model, 545

we reason that a modulatory function as proposed here might plausibly also 546 reflect the functionality of an ancestral "protosensory" cell [47] , that could 547 egfp/c13611 7x10 -3 4x10 -3 common EP/TRE Fig. 1F , ref. [17] gq/c6424 0.018 0.010 common EP/TRE Fig. 1G , ref. [17] ngbl/c10609 7x10 -3 7x10 -3 common EP/TRE Platynereis dumerilii were raised and bred in the Max Perutz Labs marine 804 facility according to established procedures [76] . Experimental animals were 805 immature adults fed last 4 to 6 days prior to the day of the experiment. 806

Remaining food was removed a day after feeding, and the seawater changed, 807 leaving the worms unperturbed for 3 to 5 days prior to sampling. All 808 pMos{rops::egfp} vbci2 transgenic worms [17] used for transcriptome profiling 809 were screened for strong EGFP fluorescence under a stereo microscope 810 system (Zeiss SteREO Lumar V12) at least 6 days before the experiment. To 811 partially immobilize the worms for the screening, worms were shortly 812 transferred to a dry petri dish. 813 FSC-W channels. To separate real EGFP fluorescence from 837 autofluorescence, we followed a previously established strategy [77] , 838 measuring fluorescence elicited by a 488nm laser using two distinct detectors 839 (see Fig. 1C,D) . One quantified fluorescence in the 515-545nm range ("FITC" 840 axis in Fig. 1C,D; Fig.1-figure supplement 1A,B) , while the other quantified 841 fluorescence in the 600-620nm range ("PE" axis in Fig. 1C,D; Fig.1-figure  842   supplement 1A,B) . Comparison between stained cell suspensions from 843 transgenic (Fig. 1C,D) and wild-type ( Fig.1-figure supplement 1A,B ) 844 specimens allowed for the definition of the gate containing EGFP+ events 845 (boxes in Fig. 1C,D) . 846 [27, 80] . For each experiment, we used the raw read counts to first 904 filter out all genes that did not have more than 1 count per million in at least 3 905 samples within the experiment, and to then calculate normalization factors for 906 each sample by comparing all samples of the same experiment. 907

Subsequently, we used the quantile-adjusted conditional maximum likelihood 908 (qCML) method to calculate the common and gene-wise dispersion, and the 909 exact test for the negative binomial distribution to test for differentially 910 expressed genes [27, 80] Two proteins (Gg and InaD) had no P. dumerilii tblastn hits that satisfied this 945 stringent threshold. Therefore, to assign P. dumerilii homologs to these 946 proteins, we lowered the stringency of the E value threshold to 1e-8. To 947 corroborate that c33855 is a bona fide homolog of Gg (E value 2e-9), we 948 confirmed that this P. dumerilii gene is the best tblastn hit of the M. musculus 949 Gg counterpart (Gng; E value against c33855: 2e-9). Similarly, to corroborate 950 that c7982 is a homolog of InaD (E value 2e-16), we confirmed that this gene 951 is the best tblastn hit of the M. musculus InaD counterpart (Mpdz; E value 952 against c7982: 2e-81). 953

Statistical assessment of subset specificity. To assess whether the 954 number of EP-and/or TRE-expressed/enriched genes overlapping with the P. 955 dumerilii homologs of a set of N D. melanogaster or M. musculus genes was 956 meaningful, we generated 10 4 sets of N randomly-picked D. melanogaster or 957 M. musculus genes, and performed the same analysis as for our real set of N 958 D. melanogaster or M. musculus genes. We then determined the frequency 959 with which such randomly generated sets resulted in an overlap that was 960 equal or higher than that found for our real set. 961

Transporting ATPases and related proteins. Candidate Plasma membrane 963

Calcium-Transporting Atpases and related proteins were identified from the 964 Platynereis dumerilii transcriptome with the tblastn algorithm, using selected 965 animal homologs as query (see Fig.5-figure supplement 1A) 

To validate the results of our differential expression analysis, we selected 2 972 common EP-/TRE-enriched genes (ngbl/c10609 and tmdc/c2433), and 3 973 TRE-specific genes (f8a/c6996, dmd/c7924 and trpA/c7677). The genes 974 selected cover a wide FDR range in our statistical analysis (Fig.2-figure  975 supplement 2). ngbl/c10609 and tmdc/c2433 are among the top enriched 976 genes in both EP and TRE samples (FDR < 0.01), whereas trpA/c7677 (FDR 977 = 0.038) is close to the significance threshold ( Fig.2-figure supplement 2) . 978

The low FDR values for ngbl/c10609 and tmdc/c2433 reflect the high level of 979 expression of these genes in the EP and TRE samples for all three biological 980 replicates, and the low level of expression in the unsorted samples ( Fig.2-981   figure supplement 1A,B) . From these data, we expected that ngbl/c10609 982 and tmdc/c2433 would be expressed at low levels (or not expressed at all) in 983 any cell type other than EP and TRE cells. We used the established single-or 984 two-color whole mount in situ hybridization (WMISH) [75] with r-opsin1 as 985 reference. Within the head, r-opsin1 is prominently expressed in the four adult 986 eyes [17] , which is reproduced in our controls (Fig.2-figure supplement  987 3C,D, detected in red). Of note, a dense pigment cup covers the internal 988 portion of each eye that contains the photosensitive outer segments of the 989 retinal photoreceptors [31] . This pigmented area can be seen as a dark area 990 in the eyes (Fig.2-figure supplement 3C.D) , which partially shields the r-991 opsin1 staining. However, due to the localization of the photoreceptor cell 992 deletion). Mutant worms were raised and crossed for several generations to 1223 generate both homozygous incross strains and respective wild-type relatives. 1224

Reconstructing the eyes of Urbilateria

Development of pigment-cup eyes in the polychaete 1259

Platynereis dumerilii and evolutionary conservation of larval eyes in 1260 Bilateria

The last common ancestor of most bilaterian animals 1263 possessed at least nine opsins

Phototransduction in Drosophila. Curr Opin

Opsins and clusters of sensory G-protein-coupled receptors in the 1269 sea urchin genome

1272 Unique system of photoreceptors in sea urchin tube feet

1275 Cephalochordate melanopsin: evolutionary linkage between 1276 invertebrate visual cells and vertebrate photosensitive retinal ganglion 1277 cells

Extraocular 1279 vision in a brittle star is mediated by chromatophore movement in 1280 response to ambient light

Early Cambrian lobopodian sclerites and associated fossils 1283 from Kazakhstan

Chance and necessity in eye evolution

Photoreceptor cells 1288 and eyes in Annelida

Segmentation, metamerism and the Cambrian explosion

Hedgehog signaling regulates segment formation in 1294 the annelid Platynereis

Death march of a segmented and 1297 trilobate bilaterian elucidates early animal evolution

Drosophila auditory organ genes and 1301 genetic hearing defects

Proprioceptive Opsin functions in Drosophila larval locomotion

Stable transgenesis in the marine annelid 1308

Platynereis dumerilii sheds new light on photoreceptor evolution

Douglas 1311 RH. Light sensitivity in a vertebrate mechanoreceptor?

Ancestry of photic and mechanic sensation?

TALENs mediate efficient and heritable mutation of endogenous 1318 genes in the marine annelid Platynereis dumerilii

Spectral tuning of phototaxis by a Go-opsin in the 1322 rhabdomeric eyes of Platynereis

Neural circuitry of a polycystin-mediated 1326 hydrodynamic startle response for predator avoidance

The polychaete Platynereis dumerilii 1329 (Annelida): a laboratory animal with spiralian cleavage, lifelong 1330 segment proliferation and a mixed benthic/pelagic life cycle

1333 Expression dynamics and protein localization of rhabdomeric opsins in 1334 Platynereis larvae

Full-length RNA-seq from single cells using Smart-seq2

Circadian and circalunar clock interactions in a marine 1341 annelid

edgeR: a Bioconductor 1343 package for differential expression analysis of digital gene expression 1344 data

Profiling by image 1347 registration reveals common origin of annelid mushroom bodies and 1348 vertebrate pallium

Neuronal connectome of a sensory-motor 1352 circuit for visual navigation

A Go-type opsin mediates the shadow reflex in the annelid 1356 Platynereis dumerilii

Das Auge von Platynereis dumerilii 1359 (Polychaeta): Sein Feinbau im ontogenetischen und adaptiven Wandel 1360 [The eye of Platynereis dumerilii (Polychaeta): Its fine structure

A versatile depigmentation, clearing, and labeling 1365 method for exploring nervous system diversity

Molecular architecture of annelid nerve cord 1369 supports common origin of nervous system centralization in Bilateria

Orthologs of key 1372 vertebrate neural genes are expressed during neurogenesis in the 1373 annelid Platynereis dumerilii

Isolation of mRNA from specific 1376 tissues of Drosophila by mRNA tagging

Deletion of 1379 PDZD7 disrupts the Usher syndrome type 2 protein complex in 1380 cochlear hair cells and causes hearing loss in mice

Myosin VIIa, harmonin and cadherin 23, three Usher I gene 1384 products that cooperate to shape the sensory hair cell bundle

Modes and regulation of endocytic membrane retrieval in mouse 1388 auditory hair cells

ATP8B1 is essential for maintaining normal 1392 hearing

A new compartment at stereocilia tips defined by spatial and 1396 temporal patterns of myosin IIIa expression

Sox2 and sox3 1399 play unique roles in development of hair cells and neurons in the 1400 zebrafish inner ear

The Notch 1403 ligands DLL1 and JAG2 act synergistically to regulate hair cell 1404 development in the mammalian inner ear

1407 CRYM mutations cause deafness through thyroid hormone binding 1408 properties in the fibrocytes of the cochlea

Absence of SERPINB6A 1411 causes sensorineural hearing loss with multiple histopathologies in the 1412 mouse inner ear

Nonmuscle myosin heavy-chain gene MYH14 is expressed in 1416 cochlea and mutated in patients affected by autosomal dominant 1417 hearing impairment (DFNA4)

Loss of Myh14 1420 increases susceptibility to noise-Induced hearing loss in CBA/CaJ 1421

A conserved developmental program for 1423 sensory organ formation in Drosophila melanogaster

Molecular evolution of the 1426 vertebrate mechanosensory cell and ear

A GFP-based genetic screen 1429 reveals mutations that disrupt the architecture of the zebrafish 1430 retinotectal projection

Gq alpha protein function 1433 in vivo: genetic dissection of its role in photoreceptor cell physiology

Human melanopsin forms a pigment maximally 1436 sensitive to blue light (λmax ≈ 479 nm) supporting activation of 1437

G(i/o) signalling cascades

Ciliary photoreceptors with a vertebrate-type opsin in an invertebrate 1442 brain

Reproducible and sustained 1444 regulation of Gαs signalling using a metazoan opsin as an optogenetic 1445 tool

In 1447 search of the visual pigment template

Mutations in a plasma membrane Ca2+-ATPase gene cause 1451 deafness in deafwaddler mice

Deficiency in 1454 plasma membrane calcium ATPase isoform 2 increases susceptibility 1455 to noise-induced hearing loss in mice

Role of ATP-1458 dependent calcium regulation in modulation of Drosophila synaptic 1459 thermotolerance

Identification of nonviable genes 1462 affecting touch sensitivity in Caenorhabditis elegans using neuronally 1463 enhanced feeding RNA interference

Differential 1466 impacts of the head on Platynereis dumerilii peripheral circadian 1467 rhythms

Proprioceptors, bristle receptors, efferent sensory 1469 impulses, neurofibrils and number of axons in the parapodial nerve of 1470 the polychaete Harmothoë

The sensory and motor innervation of Nereis

A morphometric 1476 comparison of dissimilar early development in sibling species of 1477 Platynereis (Annelida, Polychaeta)

Seasonal variation in UVA light drives hormonal and 1481 behavioral changes in a marine annelid via a ciliary opsin

The 1484 statistical geometry of transcriptome divergence in cell-type evolution 1485 and cancer

The origin and evolution of cell types

Evolution of neuronal 1490 types and families

Cubozoan jellyfish: an Evo/Devo model for 1493 eyes and other sensory systems

A short history of nearly every sense -The evolutionary 1496 history of vertebrate sensory cell types

Usher protein functions in hair cells and 1499 photoreceptors

Chromophore-independent roles of Opsin apoproteins in 1503 Drosophila mechanoreceptors

Photomechanical responses in Drosophila 1506 photoreceptors

1508 Functions of Opsins in Drosophila taste

The evolution of 1511 phototransduction from an ancestral cyclic nucleotide gated pathway

Cnidocyte discharge is regulated 1515 by light and opsin-mediated phototransduction

1518 Fluorescent two-color whole mount in situ hybridization in Platynereis 1519 dumerilii (Polychaeta, Annelida), an emerging marine molecular model 1520 for evolution and development

Großes 1524

Establishment of 1526 transgenesis in the demosponge Suberites domuncula

Full-length transcriptome assembly from RNA-Seq data without a 1530 reference genome

NextGenMap: fast 1533 and accurate read mapping in highly polymorphic genomes

A scaling normalization method for 1536 differential expression analysis of RNA-seq data

MUSCLE: multiple sequence alignment with high accuracy 1539 and high throughput

IQ-TREE: a fast 1542 and effective stochastic algorithm for estimating maximum-likelihood 1543 phylogenies

Cloning and sequence analysis of cDNA for the luminescent protein 1547 aequorin

SciPy 1.0: fundamental algorithms for scientific 1551 computing in Python

Data Structures for Statistical Computing in Python

Array programming with NumPy

VanDyk 1559 JK, et al. TAL Effector-Nucleotide Targeter (TALE-NT) 2.0: tools for 1560 TAL effector design and target prediction

1563 Efficient design and assembly of custom TALEN and other TAL 1564 effector-based constructs for DNA targeting

Simple methods for generating and detecting locus-specific 1568 mutations induced with TALENs in the zebrafish genome

bodies (and those of the support cells) outside the pigment cup, gene 993 expression can be assessed in this apparent circle around the pigment cup 994 (broken white contour in Fig.2-figure supplement 3D) . In this non-995 pigmented area of the eyes, the red staining for r-opsin was clearly discernible 996 ( Fig.2-figure supplement 3D) . Single-color ISH using a probe against 997 ngbl/c10609 showed expression of this gene in the EP as well (Fig.2-figure  998 supplement 3E,F, blue staining), confirmed by two-color-WMISH ( trpA/c7677 is detected in small spots in each parapodium ( Fig.2-figure  1014 supplement 4F,G, blue staining). Two-colour-WMISH shows that one of the 1015 spots in each parapodium overlaps with ropsin1 expression, limited to only 1016 part of the cell (Fig. 3E) . f8a/c6996 and dmd/c7924 were also expressed in a 1017 single cell in the ventral flap of the dorsal arm of each parapodium, in a 1018 position that is consistent with the TRE cell ( Fig.2-figure supplement 4H-K) , 1019 while expression in the eyes was undetected (f8a/c6996; Fig.2-figure  1020 supplement 3K,L) or extremely weak (dmd/c7924; Fig.2-figure supplement  1021 3M,N). Along with our set of control genes, these additional validations yield a 1022 total of 10 genes that confirm our enrichment analysis (Fig.2-figure  1023 supplement 2). The confirmation of f8a/c6996, dmd/c7924 and trpA/c7677, 1024 with relatively low level of expression and moderate enrichment FDR values 1025 ( Fig.2-figure supplement 2) particularly strengthens the validity of our 1026 51 analysis. It is worth noting that genes expressed at low levels are more likely 1027 affected by stochasticity effects during cDNA synthesis and amplification than 1028 their highly abundant counterparts. This provides a likely explanation why 1029 dmd/c7924 and trpA/c7677 are detected, respectively, in two and one of the 1030 three biological replicates in TRE cells (Fig.2-figure supplement 1D,E) . [53] ), human Rhodopsin 1 (G i/o assay; ref. [53] ) and human Opn4 (G q 1061 assay, ref.[51]). 1062

To determine the spectral sensitivity of Platynereis r-Opsin1, the 1064 aforementioned bioluminescence assay was further refined. Band-pass (420, 1065 442, 458, 480, 500, 520, 540, 568 and 600nm) and neutral density filters (0- spectrum and intensity of the light (see Fig. 5B ) was determined using the 1240 same spectroradiometer as described above. The temperature was monitored 1241 with a similar device as described above, and was kept within the same range 1242 as in the blue light conditions (18.5