Little is known about the genes and biological mechanisms controlling food preference in animals. In this study, by profiling the transcriptomes of dead prey fish feeders and nonfeeders in mandarin fish, we identified differentially expressed genes potentially influencing the unique food preference of live prey, including those affecting retinal photosensitivity, circadian rhythm, appetite control, learning and memory. Real-time RT-PCR confirmed the differential expression in selected genes. We also found dramatic difference in SNP abundance in feeders vs nonfeeders. These differences together might account for the different food preferences.
Differentially expressed genes involved in retinal photosensitivity pathway
Animals make food choices based on a number of physiological, nutritional, environmental, and sociocultural factors . Sensitivity of sensory system is critical to food preference. It is thus advantageous for mandarin fish to catch prey fish at night through the perception of motion and shape with the help of its well-developed scotopic vision . Although Salmo spp. also shows the selection of food motion and shape to some extent, the offered food pellet can be captured immediately before it falls down to the bottom of the tank because they have high visual acuity, thus can feed swiftly by darting [15–17]. This contrasts with mandarin fish, which has low visual acuity and feed only by stalking. Mandarin fish might not be able to accomplish its relatively long process of prey recognition before the offered food pellet has fallen down to the bottom of the tank and can no longer be perceived by its sensory organs. The more strict selection of prey motion and shape also makes it more difficult to feed mandarin fish with artificial diets and dead prey fish. Hence much better developed visual ability could improve the mandarin fish to accept dead prey fish or artificial diets .
Adaptation to dark in most animals is associated with increased 11-cis-retinal generation and rhodopsin reconstitution. Factors that interfere with the rhodopsin cycle or its downstream signaling pathways will affect vision, especially scotopic vision . We observed differential expression of Crbp, Rgr, Rdh8 and Gc in brains of feeders vs nonfeeders. Crbp is involved in the initial processing of retinol from food . Light-dependent formation of 11-cis-retinal by the retinal pigment epithelium and regeneration of rhodopsin under photic conditions involve the RGR opsin located in the retinal pigment epithelium . RDH8 is a critical regulator of chromophore regeneration . Gc, as a family of enzymes that catalyze the conversion of GTP to cGMP, are central in phototransduction cascade . Our results suggested that the up-regulation of Crbp, Rgr and Rdh8, and down-regulation of Gc in feeders might lead to increased 11-cis-retinal and rhodopsin levels, deceased cGMP generation, leading to greater visual ability and light sensitivity. Thus the feeders could capture the dead prey fish before it falls to the bottom of the tank. Moreover, the mRNA expression levels of connexin 35 (Cx35), Cyp4v2 and chicken ovalbumin upstream promoter transcription factor 2 (Coup-tf2) genes in the brain of feeders were all higher than those of nonfeeders, potentially contributing to the food preference [22–25] (Figure 4A).
Differentially expressed genes involved in circadian rhythm pathway
Previous studies demonstrated that the molecular mechanisms of circadian rhythm generation in zebrafish appear to be generally consistent with the mammalian model . We identified homologs of the mammalian clock genes in mandarin fish. We found differential expression in several clock genes, including Per1, Per2, cryptochrome (Cry), Clock, Bmal1, CkIα, CkIγ, CkIδ, CkIIβ, Rev-erbα, Skp1 and Rbx2 in feeders vs nonfeeders (Figure 4B). These genes are known to be critical regulators of circadian rhythm [27–33]. Taken together, changes in expression levels of these clock genes in feeders might reset circadian phase and contribute to the unique food preference.
Differentially expressed genes involved in appetite control pathway
Previous studies provide a framework for understanding the regulation of food intake in mammals and fish. Peripheral signals such as leptin from adipocytes, insulin from endocrine pancreas, cholecystokinin and peptide YY from gastrointestinal tract are incorporated in the hypothalamus to generate orexigenic (such as NPY and ghrelin) or anorexigenic (such as α-melanocyte stimulating hormone (α-MSH) derived from POMC) signals . We observed lower expression of orexigenic genes (Npy, proenkephalin, Gh, uncoupling protein 2 (Ucp2), Creb, eukaryotic initiation factor (eIF) 4E binding protein (4Ebp1), tuberous sclerosis 1 (Tsc1), ghrelin and leptin receptor gene-related protein (Ob-rgrp)) and higher expression of anorexigenic genes (Pomc, Pyy, preprosomatostatin (Srif), insulin, leptin, cholecystokinin (Cck) and tachykinin 1) in feeders compared with nonfeeders (Figure 4C). These genes are well established regulators of energy homeostasis and play important roles in determination of food preferences [35–46]. The changes in gene expression suggested that feeders had decreased appetite.
Differentially expressed genes involved in learning and memory pathway
Food habits can develop with learning experience [47, 48]. Our previous study indicated that sensory modality and associative learning appear to be critical factors in food discrimination of mandarin fish . However, study on genes involved in learning and memory of nocturnal piscivorous fish has received little attention. A number of molecules participate in learning and memory processes. The TORC1-mediated CREB regulation is a critical molecular step underlying synaptic plasticity and long-term memory [50, 51]. As a suppressor of CREB, protein phosphatase 1 (PP1) determines the efficacy of learning and memory by limiting acquisition and favoring memory decline . Neural cell adhesion molecule (NCAM) plays an important role in axonal growth, learning, and memory through activating the phosphorylation of MAPKs and CREB . In our study, compared with nonfeeders, the mRNA levels of Creb and Ncam were dramatically decreased and Pp1 were slightly increased in feeders, protentially resulting in significant deficiency in the maintenance of long-term memory of its natural food preference, and therefore feeders were able to accept novel food (dead prey fish) (Figure 4D).
Along the same line, as members of immediate early gene and the Fos family of transcription factors, c-fos and Fra-2 mRNA expressions are up-regulated in response to a variety of neuronal activation protocols, including LTP [54–56]. In the current study, c-fos, Fra-2 and zif268 genes were expressed at significantly lower levels in feeders than in nonfeeders, suggesting disruption of LTP and memory formation in feeders . We hypothesize that the acquisition of novel food preference might be closely associated with effacement of the original memory of natural food preference (live prey), and the enhanced learning capacity for new food preference (dead prey). We also showed that in mandarin fish, a number of genes necessary for memory formation and retention (C/EBP, Bdnf, Syt I, Syt IV and nitric oxide synthase (Nos)) were differentially expressed in feeders vs nonfeeders. They might also be involved in the determination of the unique food preference in mandarin fish (Figure 4D) [58–61].