Circulating miRNA repertoire as a biomarker of metabolic and reproductive states in rainbow trout

Abstract

Background: Circulating miRNAs (c-miRNAs) are found in most, if not all, biological fluids and are becoming well-established non-invasive biomarkers of many human pathologies. However, their features in non-pathological contexts and whether their expression profiles reflect normal life history events have received little attention, especially in non-mammalian species. The aim of the present study was to investigate the potential of c-miRNAs to serve as biomarkers of reproductive and metabolic states in fish.

Results: The blood plasma was sampled throughout the reproductive cycle of female rainbow trout subjected to two different feeding regimes that triggered contrasting metabolic states. In addition, ovarian fluid was sampled at ovulation, and all samples were subjected to small RNA-seq analysis, leading to the establishment of a comprehensive miRNA repertoire (i.e., miRNAome) and enabling subsequent comparative analyses to a panel of RNA-seq libraries from a wide variety of tissues and organs. We showed that biological fluid miRNAomes are complex and encompass a high proportion of the overall rainbow trout miRNAome. While sharing a high proportion of common miRNAs, the blood plasma and ovarian fluid miRNAomes exhibited strong fluid-specific signatures. We further revealed that the blood plasma miRNAome significantly changed depending on metabolic and reproductive states. We subsequently identified three evolutionarily conserved muscle-specific miRNAs or myomiRs (miR-1-1/2-3p, miR-133a-1/2-3p, and miR-206-3p) that accumulated in the blood plasma in response to high feeding rates, making these myomiRs strong candidate biomarkers of active myogenesis. We also identified miR-202-5p as a candidate biomarker for reproductive success that could be used to predict ovulation and/or egg quality.

Conclusions: Together, these promising results reveal the high potential of c-miRNAs, including evolutionarily conserved myomiRs, as physiologically relevant biomarker candidates and pave the way for the use of c-miRNAs for non-invasive phenotyping in various fish species.

Keywords: Biological fluid; Biomarker; Fish; Non-invasive phenotyping; The authors Emilie Cardona and Cervin Guyomar contributed equally to this work.; mir202; mir375; myomiR.

Fig. 1

Experimental design. Fish rearing reproductive stages, fluid sampling, and biological samples used for RNA-seq and QPCR analysis are described. Additional details on fish rearing including feeding levels and monitoring of reproduction can be found in [27]

Fig. 2

Circulating miRNA repertoires. A Venn diagram of miRNAs detected in the blood plasma, ovarian fluid, and in 21 other sample types (brain, eggs, gills, gonad, head-kidney, heart, intestine, leucocytes, liver, muscle, myoblasts, myotubes, ovary, pituitary, skin, spermatogonia, spleen, stomach, testis, trunk-kidney, whole embryos). A miRNA was considered expressed in a sample type when its averaged normalized abundance exceeded 10 RPM (reads per million reads). B Distribution of miRNA normalized counts across sample types listed above in RPM. Corresponding data are available in Additional file 4

Fig. 3

Putative origin and expression patterns of c-miRNAs. A For each fluid c-miRNA, the organ in which its expression was the highest was considered as the putative organ of origin. The number of c-miRNAs exhibiting highest expression in a specific organ are shown on the chart. The heart is displayed even though no c-miRNA had a maximum expression in this organ. B Heatmap of miRNA expression across different organs, blood plasma, and ovarian fluid. The heatmap was built using a row-scaled matrix of the normalized read counts (RPM) supplied by Prost!. Normalized RPM counts were averaged for all samples of a given sample type. Corresponding data are available in Additional file 1

Fig. 4

Detection of blood plasma and ovarian fluid miRNAs in various organs and cell types. A PCA analysis of normalized RPM counts supplied by Prost! using all samples (N=52). PCA was centered but not scaled. Corresponding data are available in additional file 4. B The number of circulating miRNAs detected in all possible combinations of analyzed libraries is displayed. Libraries in which miRNAs were detected are indicated by a black dot. miRNAs were considered present in an organ or a fluid if their normalized abundance exceeded 10 RPM, on average, for each type of sample. The analysis was limited to the two biological fluids (blood plasma and ovarian fluid) and a subset of 13 different female organs displayed in Fig. 3

Fig. 5

Normalized expression values of the most abundant c-miRNAs across organs and biological fluids. The analysis was performed using the two biological fluids (blood plasma and ovarian fluid) and a subset of 13 different female organs displayed in Fig. 3. Normalized RPM counts were averaged for all samples of a given sample type. miR-451-5p, let-7a-5p, miR-21-5p, miR-16b-5p, miR-30d-5p, and miR-26a-5p were among the 10 most abundant miRNAs for both blood plasma and ovarian fluid. miR-92a-3p, miR-150-5p, and miR-128-3p were among the most abundant miRNAs in the blood plasma. miR-202-5p, miR-22a-1-3p, and miR-146a-5p were among the most abundant miRNAs in ovarian fluid. Corresponding data are available in Additional file 1

Fig. 6

Principal Component Analyses of biological fluid c-miRNAs. PCAs were centered but not scaled and computed from normalized miRNAs counts (RPM) supplied by Prost!. Ellipses represent 95% confidence intervals and were drawn for conditions represented by at least three individuals. Points are samples named under the following pattern: “Fluid_Reproductive stage_feeding_replicate.” The value for “Fluid” is either the blood plasma (P) or the ovarian fluid (OF). The following reproductive stages were analyzed: pre-vitellogenesis (PV), early vitellogenesis (EV), and ovulation (OV). The feeding level was either ad libitum (al) or restricted (r). Corresponding data are available in Additional file 4. A PCA of all 10 blood plasma and 4 ovarian fluid samples. B PCA using only the blood plasma samples (n=10) collected at the three different reproductive stages

Fig. 7

Expression profiling of c-miRNAs differentially expressed during the reproductive cycle. A Expression heatmap for c-miRNAs differentially expressed across reproductive stages. Expression values were log-transformed read counts of the 107 c-miRNAs differentially expressed during the reproductive cycle. The heatmap was scaled by row. Expression values were standardized and are expressed in standard deviation (sd) units. Expression dynamic clusters are indicated by colors in the rightmost column. The following reproductive stages were analyzed: pre-vitellogenesis (PV), early vitellogenesis (EV), and ovulation (OV). Corresponding data are available in Additional file 4. B Expression dynamics by cluster of the 107 differentially expressed c-miRNAs. All differentially expressed miRNAs were assigned to one of four clusters based on their expression dynamic during the reproductive period

Fig. 8

Blood plasma abundance of candidate biomarker c-miRNAs. Quantitative PCR analysis of selected miRNAs during the reproductive cycle and in response to feeding rate. The following reproductive stages were analyzed: pre-vitellogenesis (PV), early vitellogenesis (EV), late vitellogenesis (LV), and ovulation (OV). The feeding rate was either ad libitum (al) or restricted (r). Significant differences (ANOVA) in expression levels (arbitrary units, a.u.) are indicated for reproductive stage (RS), feeding rate (FR), and feeding rate—stage interactions (RS X FR). *p<0.05, **p<0.05, n.s. not significant (p>0.05). Replicates (N= 5 or 6) correspond to different individual fish. Corresponding data are available in Additional file 6

Fig. 9

Candidate biomarker c-miRNA abundance in various organs. Quantitative PCR analysis of selected miRNAs in different organs. Replicates (N= 3, except for heart N=1) correspond to different individual fish. Means and standard error mean are displayed. Corresponding data are available in Additional file 6