Extraocular photoreception, the ability to detect and respond to light outside of the eye, has not been previously described in deep-sea invertebrates. Here, we investigate photosensitivity in the bioluminescent light organs (photophores) of deep-sea shrimp, an autogenic system in which the organism possesses the substrates and enzymes to produce light. Through the integration of transcriptomics, in situ hybridization and immunohistochemistry we find evidence for the expression of opsins and phototransduction genes known to play a role in light detection in most animals. Subsequent shipboard light exposure experiments showed ultrastructural changes in the photophore similar to those seen in crustacean eyes, providing further evidence that photophores are light sensitive. In many deep-sea species, it has long been documented that photophores emit light to aid in counterillumination – a dynamic form of camouflage that requires adjusting the organ’s light intensity to “hide” their silhouettes from predators below. However, it remains a mystery how animals fine-tune their photophore luminescence to match the intensity of downwelling light. Photophore photosensitivity allows us to reconsider the organ’s role in counterillumination - not only in light emission but also light detection and regulation.

Live specimens (n = 5) were preserved in RNAlater and stored at −80 °C. Eye and photophore tissue replicates were carefully dissected and homogenized in TRIzol reagent (ThermoFisher Scientific). Total RNA was discretely extracted from tissues and rDNase (Macherey-Nagel) treated following recommendations in DeLeo et al. 2018 (doi: 10.1093/biomethods/bpy012). Libraries were prepared from isolated mRNA using the NEBNext Ultra II Directional RNA Library Prep Kit for Illumina. Barcoded libraries were size selected (Pippin Prep, Sage Science) and sequenced across an Illumina HiSeq4000 lane. Raw sequencing data was quality assessed using FastQC to inform quality and adaptor trimming with Trimmomatic v0.36. Trimmed reads were processed with Rcorrector and BBnorm to error-correct reads and normalize read coverage. Trinity v2.6.5 was used to assemble tissue-specific (eye and photophore) reference assemblies de novo (minimum contig length of 200 bp, k-mer size of 23). Contamination was removed from each assembly using Kraken v1.0 with default parameters and NCBI’s (Refseq) bacteria, archaea and viral databases. Contaminate free assemblies were then passed through BBduk and dedupe (BBTools suite, available at: http://sourceforge.net/projects/bbmap) to remove duplicate transcripts and rRNA.