Conserved function of lincRNAs in vertebrate embryonic development despite rapid sequence evolution

Abstract

Thousands of long intervening noncoding RNAs (lincRNAs) have been identified in mammals. To better understand the evolution and functions of these enigmatic RNAs, we used chromatin marks, poly(A)-site mapping and RNA-Seq data to identify more than 550 distinct lincRNAs in zebrafish. Although these shared many characteristics with mammalian lincRNAs, only 29 had detectable sequence similarity with putative mammalian orthologs, typically restricted to a single short region of high conservation. Other lincRNAs had conserved genomic locations without detectable sequence conservation. Antisense reagents targeting conserved regions of two zebrafish lincRNAs caused developmental defects. Reagents targeting splice sites caused the same defects and were rescued by adding either the mature lincRNA or its human or mouse ortholog. Our study provides a roadmap for identification and analysis of lincRNAs in model organisms and shows that lincRNAs play crucial biological roles during embryonic development with functionality conserved despite limited sequence conservation.

Figure 1. Identification of Zebrafish lincRNA Genes

(A) Positions of H3K4me3 peaks from 24-hpf embryos with respect to annotations of known protein-coding genes and genes of small ncRNAs (<200 nt) annotated in Ensembl or RefSeq. (B) Positions of poly(A) sites with respect to annotated protein-coding and small ncRNA genes. (C) Pipeline for identification of lincRNAs. See text and extended experimental procedures for description.

Figure 2. Expression of Zebrafish lincRNAs

(A) Whole-mount in situ hybridizations of selected lincRNAs. Control experiments using sense probes for selected lincRNAs were also performed (Figure S2). (B) Expression levels of lincRNA and protein-coding genes evaluated using RNA-Seq results from ten stages/tissues. Plots indicate the median, quartiles, and 10^th and 90^th percentiles. RPKM is reads per kilobase per million reads. (C) Correlations between levels of neighboring transcripts. For each gene, the Spearman correlation between its expression profile (across ten stages/tissues) and that of the closest protein-coding gene was determined, and the average is plotted for the lincRNA and coding genes. Error bars are 95% confidence intervals based on 1000 random shuffles of lincRNA positions.

Figure 3. lincRNA Conservation in Vertebrates

(A) Conservation levels of lincRNA and protein-coding introns and exons, computed using phastCons (Siepel et al., 2005) applied to the 8-way whole-genome alignment. For each lincRNA locus, a computational control was generated by random sampling of a length-matched region from intergenic space of the same chromosome. Within this control region, exons were assigned to the same relative positions as in the authentic lincRNA locus (control exons). (B) Annotation of human or mouse genomic regions aligned to 188 zebrafish lincRNA genes in the 8-way whole-genome alignment. Regions aligned with zebrafish lincRNAs were tested for overlap with (i) lincRNAs (Table S3), (ii) protein-coding sequence, (iii) 5′UTR or 3′UTR, (iv) introns, or (v) GenBank mRNAs (unannotated cDNAs), in this order, and assigned to the first category for which overlap was observed. (C) Conserved orientation of protein-coding genes with adjacent lincRNAs. Orthologous protein-coding genes adjacent to zebrafish and mammalian lincRNAs were identified, and the corresponding lincRNAs were considered to have conserved positions, regardless of their sequence conservation. Plotted is the number of those with orientations conserved with respect to their anchoring proteins. The 95% confidence intervals were estimated using 200 cohorts of computational controls, generated as in (A). (D) linc-tmem106a and its positionally conserved lincRNAs in the human and mouse genomes. The number of 3P tags mapping to the plus and minus strands are indicated (red and blue, respectively).

Figure 4. lincRNAs with Short Conserved Segments

(A) Genomic context and sequence conservation of the linc-oip5 (cyrano) lincRNA gene. Gray boxes include the deeply conserved region. The conservation plot is relative to the human locus, and is based on aligned regions of 37 genomes, which do not include any fish genomes, as those do not contain any regions that are aligned with this human locus in the whole-genome alignments. The top consensus logo highlights the RNA sequence of the most conserved segment, which we identified in 45 vertebrate genomes, including fish genomes. Shown are the 67 aligned positions present in zebrafish, with a score of 2 bits indicating residues perfectly conserved in all 45 genomes. The bottom consensus logo shows conservation of vertebrate miR-7 sequences annotated in miRBase 18, with vertical lines indicating Watson-Crick base pairs. (B) Genomic context and sequence conservation of the linc-birc6 (megamind) lincRNA gene. As in (A), except the region is aligned to fish genomes in the whole-genome alignments, and the consensus logo is for the RNA sequence inferred from 75 sequences from 47 vertebrate genomes. An alternative isoform of the zebrafish RNA retains the first intron (Figure S3E).

Figure 5. The Importance of linc-oip5 (cyrano) for Proper Embryonic Development

(A) In situ hybridization showing cyrano expression in the CNS and notochord of zebrafish embryos at 72 hpf. (B) Gene architecture of cyrano, showing the hybridization sites of the RNA-blot probe and MOs (red boxes). (C) RNA blot monitoring cyrano accumulation in wild-type embryos (48 hpf) that had been injected with the indicated MOs. To control for loading, the blot was re-probed for β-actin mRNA. (D) Embryos at 48 hpf that had been either injected with the indicated MO or co-injected with the splice-site MO and mature mouse cyrano RNA. (E) Brain ventricles after injection with the indicated reagents, visualized using a red fluorescent dye injected into the ventricle space at 28 hpf. (F) Embryos at 48 hpf that had been injected with the indicated reagents. NeuroD-positive neurons in the retina and nasal placode were marked with GFP expressed from the neurod promoter (Obholzer et al., 2008). Near absence of NeuroD-positive neurons in the retina (dotted line) and enlargement of the nasal placode (arrow) are indicated. (G) Frequency of morphant phenotypes in injected embryos (Table S5). (H) Schematic of DNA point substitutions in the cyrano conserved site. (I) Architecture of a hybrid transcript containing the cyrano conserved segment in the context of linc-birc6 (megamind) flanking sequences.

Figure 6. The Importance of linc-birc6 (megamind) for Proper Brain Development

(A) In situ hybridization showing megamind expression in the brain and eyes of zebrafish embryos at 28 hpf. (B) Gene architecture of megamind, showing the hybridization sites of the MOs (red boxes) and RT-PCR primers (arrows). (C) Semi-quantitative RT-PCR of mature megamind in embryos at 72 hpf that had been injected with the indicated MOs. β-actin mRNA was used as a control. (D) Brain ventricles after injection with either the indicated MOs or co-injected with the splice-site MO and mature mouse megamind RNA, visualized using a red fluorescent dye injected into the ventricle space at 28 hpf. An expanded midbrain ventricle (arrow) and abnormal hindbrain hinge point (asterisk) are indicated. (E) Embryos at 48 hpf that had been injected with the indicated reagents. Abnormal head shape and enlarged brain ventricles are indicated (arrow). (F) Embryos at 48 hpf that had been injected with the indicated reagents. NeuroD-positive neurons in the retina and nasal placode were marked with GFP expressed from the neurod promoter (Obholzer et al., 2008). Near absence of NeuroD-positive neurons in the retina and tectum (arrows) is indicated. (G) Frequency of morphant phenotypes in injected embryos (Table S5). (H) Schematic of DNA point substitutions in the megamind conserved segment. (I) Architecture of a hybrid transcript containing the megamind conserved segment in the context of cyrano flanking sequences.

Figure 7. lincRNA Conservation Patterns

(A) The human MALAT1 locus and the orthologous locus in zebrafish. Protein-coding genes are in green and lincRNA genes are in blue. Arrows indicate direction of transcription, black triangles indicate canonical poly(A) sites and white triangles indicate 3′ termini obtained by RNAse P cleavage (Wilusz et al., 2008). The conservation plot is relative to the zebrafish locus and based on the 8-genome alignment. (B) The zebrafish linc-epb4.1l4 gene showing homology to the 3′UTR of an mRNA expressed in human, mouse, chicken and other amniotes. Gray boxes indicate two deeply conserved regions. The repeats track indicates all the repetitive elements predicted by RepeatMasker, taken from the UCSC genome browser. The conservation plot is as in (A).