Defining ohnolog pairs on the L- and S-subgenomes
The goldfish genome consists of fifty chromosomes, which may be divided into 25 chromosomes (L-chromosomes and S-chromosomes), with every group similar to the chromosomes of the 2 progenitor species concerned in Cs4R, an allotetraploid event11,23 (Supplementary Fig. 1a, left panel). We termed the ohnologs positioned on the L-chromosomes the L-ohnologs and people positioned on the S-chromosomes the S-ohnologs. The chromosome set consisting of the 25 L-chromosomes is known as the L-subgenome, and that consisting of the 25 S-chromosomes is known as the S-subgenome23. We newly outlined high-quality ohnolog pairs in goldfish to research preservation and divergence of L-ohnolog and S-ohnolog gene expressions utilizing the scRNA-seq dataset (Supplementary Fig. 1a, proper panel). We first recognized the ohnolog candidates of 23,438 genes on the L-chromosomes and 20,666 genes on the S-chromosomes. Subsequently, we carried out a reciprocal BLAST evaluation between these two goldfish gene teams and all zebrafish genes (23,651 genes). Consequently, we recognized 11,444 ohnolog pairs (22,888 goldfish genes) with a excessive diploma of confidence (Supplementary Fig. 1b–f; Supplementary Knowledge 1). Thus, these 11,444 ohnolog pairs in goldfish have 11,444 corresponding zebrafish orthologs, positioned on their orthologous chromosomes (Supplementary Fig. 1c–f), which permit the inference of the organic features of goldfish ohnologs by referring to earlier research utilizing zebrafish. The 11,444 ohnolog pairs recognized have been improved in contrast with the 5404 ohnolog pairs that have been analyzed in our earlier study23.
Era of the single-cell transcriptomic atlas of the goldfish retina
To generate the single-cell retinal transcriptomic atlas, we used the retina from Wakin, a typical goldfish pressure, that retained wild goldfish options apart from its physique coloration13. We ready single-cell suspensions from the dissected retina through enzymatic digestion. We ready scRNA-seq libraries utilizing the ten× Genomics Chromium system, and sequenced them (Fig. 1a). After the info preprocessing and high quality management, 22,725 cells have been retained for downstream evaluation (n = 3). We projected this information set into the two-dimensional area utilizing the uniform manifold approximation and projection (UMAP) methodology. For this evaluation, we used the publicly accessible scRNA-seq information of the zebrafish retina as a reference42. Subsequently, we recognized 12 cell clusters from the goldfish retina (Fig. 1b). Based mostly on the gene expression of the identified cell-type markers of the vertebrate retina36,37,38,42,43, we recognized 12 forms of retinal cells, together with seven forms of neurons (rod photoreceptor cells [5842 cells, 25.7%], cone photoreceptor cells [911 cells, 4.0%], bipolar cells [7080 cells, 31.2%], GABAergic amacrine cells [1064 cells, 4.7%], glycinergic amacrine cells [895 cells, 3.9%], horizontal cells [668 cells, 2.9%], and retinal ganglion cells [338 cells, 1.5%]) and 5 forms of non-neuronal cells (Müller glia [2720 cells, 12.0%], microglia [2393 cells, 10.5%], retinal pigmented epithelial cells [181 cells, 0.8%], oligodendrocytes [554 cells, 2.4%], and vascular endothelial cells [79 cells, 0.3%]) (Fig. 1b, c). For instance, we recognized 895 glycinergic amacrine cells, the glycinergic interneurons concerned in visible sign transduction44, which confirmed robust expression of the solute provider household 6 member 9 (slc6a9), a glycine transporter (Fig. 1c). We discovered 2720 Müller glia with apolipoprotein Eb (apoeb) expression (Fig. 1c), which is secreted into the vitreous and transported into the optic nerve45. Then, we examined the extent to which the retinas of every of the three Wakin goldfish analyzed right here contained every cell sort. We discovered that the three samples contained the seven forms of neurons and the 5 forms of non-neuronal cells, suggesting the experiment’s reproducibility (Fig. 1b, Supplementary Fig. 2). Subsequent, we questioned whether or not there have been any variations within the cell forms of the retina between goldfish and zebrafish, a cyprinid teleost species with out Cs4R. We in contrast the cell forms of the goldfish retina with these of the zebrafish retina and located the 12 retinal cell sorts in each species (Supplementary Fig. 3). To verify whether or not the composition of every cell-type within the scRNA-seq evaluation of goldfish displays tissue cell composition, we measured cell-type ratios by a special methodology. The vertebrate retina has three nuclear layers and two plexiform layers the place synaptic connections are formed35. The outer nuclear layer (ONL) consists of the photoreceptor our bodies, and the interior nuclear layer (INL) consists of the cell our bodies of bipolar, horizontal, and amacrine cells. In the meantime, the ganglion cell layer (GCL) is especially composed of ganglion cell our bodies. Within the vertebrate retina, every cell-type kinds a layered construction, and the variety of cells in every layer may be counted to find out the approximate variety of cell sorts. Accordingly, we ready frozen sections from Wakin retinal tissues, stained with DAPI to stain the nuclei of all retinal cells and counted the variety of cells within the totally different layers of the retina (Supplementary Fig. 3c). The outcomes confirmed 33.1 ± 6.0% of ONL, 53.6 ± 7.8% of INL, and 13.2 ± 2.1% of GCL (n = 3). These values are roughly according to the results of our scRNA-seq evaluation (29.7% ONL, 54.7% INL, and 1.5% GCL). The variety of cells within the GCL in scRNA-seq appears to be smaller than anticipated from the frozen part experiment. This may be defined as a result of the GCL additionally accommodates some amacrine cells. We in contrast the cell composition of the zebrafish and goldfish retinas (Supplementary Fig. 3d). The numbers in goldfish and zebrafish are roughly related, nonetheless, essentially the most vital distinction is the 25.7% rod photoreceptor cells in goldfish in comparison with only one.0% in zebrafish. Earlier histological research recommend that the ratio of photoreceptor cells within the retina needs to be related between these two species46. Nonetheless, in zebrafish, it’s a lot decrease than in tissue sections. Since photoreceptors have lengthy outer segments and connecting cilia, they could be extra vulnerable to variations in experimental circumstances. Variations in filtering and annotating cells throughout information evaluation may contribute to the variations within the variety of rod cells47. No main variations have been famous amongst cones, bipolar, and amacrine, suggesting that there have been no crucial points within the scRNA-seq evaluation carried out on this research. This outcome means that the retinal cell inhabitants and transcriptomic profile of goldfish and zebrafish are related, and that the single-cell transcriptome information of the 2 species are comparable for additional subgenome evaluation.
Fig. 1: scRNA-seq evaluation of the Wakin goldfish retina. a The retinas from three Wakin goldfish people have been enzymatically dissociated, and single cells have been remoted, adopted by scRNA-seq library preparation. The picture was taken by the authors. All figures have been produced by the authors. b UMAP plot exhibiting the mobile composition of the retina from Wakin goldfish (n = 3 biologically impartial samples). The 22,725 cells have been projected right into a two-dimensional area by UMAP. The cells have been labeled into 12 cell sorts. c The expression of the cell-type-specific marker genes within the 12 goldfish retinal cell sorts is proven. RPE retinal pigmented epithelial cells, RGC retinal ganglion cells, HC horizontal cells, AC amacrine cells, BC bipolar cells, MG Müller glia, V/E cells vascular endothelial cells. Full measurement picture
Uneven subgenome expression on the mobile stage
We beforehand carried out a bulk RNA-seq evaluation within the seven tissues of the goldfish and recognized a world gene expression bias towards the L-ohnologs over the S-chromosomes (S-ohnologs) in all tissues analyzed11,23. Right here, we first examined whether or not the whole gene expression of every cell was biased towards the L-subgenome or the S-subgenome utilizing the 11,444 ohnolog pairs. We evaluated the sum of the L-ohnolog and S-ohnolog expression ranges for every cell. The sum of the L-ohnolog and that of the S-ohnolog expression ranges confirmed a powerful optimistic and vital correlation (Pearson’s correlation coefficient = 0.99, P < 1.0e–15, Fig. 2a). This implies that each the L-ohnologs and S-ohnologs globally contribute to shaping the transcriptome of every cell. The sum of the L-ohnolog expression ranges was considerably greater than that of the S-ohnolog expression ranges (P < 1.0e–15, Wilcoxon take a look at; Fig. 2b). The sum of the L-ohnolog expression ranges was greater than that of the S-ohnolog expression ranges in 22,365 (98.4%) of the 22,725 cells analyzed (pink space in Fig. 2c). This bias in gene expression towards the L- over the S-ohnologs was noticed in all 12 retinal cell sorts (Fig. second, left panel). The common ratio of the sum of the L-ohnolog and the S-ohnolog expression ranges for every cell sort ranged from 1.12 within the retinal ganglion cells to 1.23 within the rod photoreceptor cells (Fig. second, proper panel). The goldfish genome continues to be within the means of rediploidization, and each L-ohnologs and S-ohnologs nonetheless take part in shaping the transcriptome, as Cs4R is an evolutionarily latest occasion (~14 MYA)11. On the identical time, the L-ohnologs and S-ohnologs are within the gradual means of uneven subgenome expression on the mobile level23. Fig. 2: Expression bias towards the L-ohnolog over the S-ohnolog in goldfish retinal cells. a Scatter plot of the whole gene expression of L-ohnologs and S-ohnologs in cells. The x axis signifies the whole gene expression of L-ohnologs and the y axis signifies that of S-ohnologs. The dots are coloured in keeping with cell sorts. The overall gene expression of L-ohnologs and S-ohnologs confirmed a optimistic and vital correlation (Pearson’s correlation coefficient = 0.99, P < 1.0e–15). The pink line signifies y = x. b Boxplots of the whole gene expression of L-ohnologs andS-ohnologs in cells. The median of the whole gene expression of L-ohnologs was considerably greater than that of S-ohnologs (P < 1.0e–15, Wilcoxon test). The ends of the box are the 25 and 75% quantiles. The horizontal line in the box indicates the median. The lines extending from the top and bottom of the box represent the minimum and maximum values. c Distribution of the log2-transformed fold changes of the total gene expression of L-ohnologs and that of S-ohnologs in all 22,725 cells. Values higher than 0 (L-ohnolog > S-ohnolog) are coloured in pink, and values decrease than 0 (L-ohnolog < S-ohnolog) are coloured in blue. The pink line signifies x = 0. d Distribution of the log2-transformed fold adjustments of the whole gene expression of L-ohnologs and that of S-ohnologs in every cell sort. The left panel represents the UMAP plot exhibiting the log2-transformed fold change of the whole gene expression of L-ohnologs and that of S-ohnologs in every cell. Cells coloured in pink point out that the whole gene expression stage of L-ohnologs is greater than that of S-ohnolog, whereas the cells coloured in blue present the alternative consequence. The suitable panel represents boxplots of the log2-transformed fold change within the complete gene expression of L-ohnologs and S-ohnologs in every cell sort. The horizontal line signifies a price of zero. The ends of the field are the 25% and 75% quantiles. The horizontal line within the field signifies the median. The traces extending from the highest and backside of the field characterize the minimal and most values. Full measurement picture Ohnolog pairs displaying an expression bias towards the L- or S-ohnolog Subsequent, we targeted on particular person ohnolog pairs. We carried out a two-dimensional projection of the dataset by UMAP independently based mostly on the person gene expression of the 11,444 L-ohnologs or the 11,444 S-ohnologs, respectively. We discovered that each units of ohnologs confirmed related clusters based mostly on the seven forms of neurons and the 5 forms of non-neuronal cells recognized when analyzed utilizing each L-ohnologs and S-ohnologs (Supplementary Fig. 4a). This outcome signifies that genes on each subgenomes contribute to the characterization of a cell-type-specific transcriptome and are according to the discovering that ohnologs confirmed a major correlation in complete gene expression (Fig. 2a). Subsequently, we calculated the typical expression ranges in all cells for every gene and in contrast them in all ohnolog pairs. The fold change between the L-ohnolog and S-ohnolog confirmed a broad distribution, with a median of 1.1 (L-ohnolog/S-ohnolog), indicating a world expression bias towards the L-ohnolog (P < 1.0e–15, Wilcoxon take a look at; Supplementary Fig. 4b). In all 12 cell sorts, we discovered that the gene expression ranges of the L-ohnolog have been considerably greater than these of the S-ohnolog (Supplementary Fig. 4b). We plotted the L/S expression ratio for every ohnolog pair in every cell-type (Supplementary Fig. 4c). The L/S peak is centered at zero, suggesting that the majority ohnolog pairs present no biased expression. This contrasts with the right-shifted peak noticed in Fig. 2c (L/S of complete gene expression). This outcome means that ohnolog pairs present globally L-biased when it comes to complete gene expression, however not at most particular person gene ranges. For every ohnolog pair, we in contrast the typical gene expression between the L-ohnolog and the S-ohnolog in all 22,725 cells. We recognized 5123 (44.8%) ohnolog pairs during which the typical gene expression within the 22,725 cells was considerably totally different between the L-ohnolog and the S-ohnolog (Fig. 3a); 2690 (23.5%) ohnolog pairs confirmed biased expression towards the L-ohnolog over the S-ohnolog, whereas 2433 (21.3%) ohnolog pairs confirmed biased expression towards the S-ohnolog. The variety of ohnolog pairs with biased expression towards the L-ohnolog was considerably greater than that of the ohnolog pairs with biased expression towards the S-ohnolog (P < 5.0e–4, Binomial take a look at). Subsequent, we looked for ohnolog pairs among the many 11,444 ohnolog pairs that have been considerably extra extremely expressed in a number of particular cell sorts in comparison with different cell sorts (Fig. 3a). We discovered that 1070 ohnolog pairs (9.3%) confirmed cell sort particular expression profiles (cell-type specificity of the whole expression of L- and S-ohnologs) (Fig. 3a). Of those, L-ohnologs confirmed greater gene expression than S-ohnologs in 260 ohnolog pairs (Fig. 3a, b; Supplementary Knowledge 2), and S-ohnologs confirmed greater gene expression than L-ohnologs in 245 ohnolog pairs (Fig. 3a, b; Supplementary Knowledge 3). The remaining 10,374 ohnolog pairs (90.7%) confirmed ubiquitous expression patterns (Fig. 3a). Amongst them, 2430 ohnolog pairs confirmed greater gene expression for the L-ohnolog than the S-ohnolog (Fig. 3a, c; Supplementary Knowledge 4), and 2188 ohnolog pairs confirmed greater gene expression for the S-ohnolog than the L-ohnolog (Fig. 3a, c; Supplementary Knowledge 5). We carried out an enrichment take a look at to find out whether or not L/S-biased genes present extra cell-type particular expression than a random set of genes. The outcome confirmed no vital distinction between them (Fisher’s actual take a look at, P = 0.093). To characterize the features of those genes with ubiquitous expression patterns, we carried out purposeful enrichment evaluation on these gene units (Supplementary Fig. 5, 6). Within the former gene set (2430 ohnolog pairs), genes associated to features similar to neuron growth (GO database ID, GO:0048666), mobile response to progress issue stimulus (GO:0071363), and enzyme-linked receptor protein signaling pathway (GO:0007167) have been considerably enriched (Supplementary Fig. 5; Supplementary Knowledge 6). Within the latter gene set (2188 ohnolog pairs), genes associated to features similar to receptor metabolic course of (GO:0043112), protein acylation (GO:0043543), and mesenchyme growth (GO:0060485) have been considerably enriched (Supplementary Fig. 6; Supplementary Knowledge 7). Fig. 3: Ohnolog expression profiles amongst retinal cell sorts. a Venn diagram exhibiting the connection between ohnolog pairs with expression ranges which are biased towards the S- or L-ohnolog, and ohnolog pairs with cell-type-specific expression. b Ohnologs with cell-type-specific expression. The left panel exhibits the gene expression of the ohnolog pairs during which each the L-ohnolog and S-ohnolog confirmed comparable cell-type-specific expression. The suitable panel exhibits the gene expression of the ohnolog pairs during which both the L-ohnolog or S-ohnolog confirmed considerably greater expression than the opposite. c Ohnologs with ubiquitous expression. The left panel exhibits the ohnolog pairs during which each the L-ohnolog and S-ohnolog are ubiquitously expressed. The suitable panel exhibits the gene expression of the ohnolog pairs during which both the L-ohnolog or S-ohnolog confirmed ubiquitous expression and considerably greater expression than the opposite. Full measurement picture Choice of ohnolog pairs retaining unbiased expression on the single-cell-resolution stage Sustaining gene dosage stability is crucial for sure forms of ohnolog pairs after WGD2. For instance, homeobox genes have a tendency to take care of their expression stage after WGD as a result of the change of their dosage stability impacts embryonic sample formation and survival2,48. To determine ohnolog pairs with unbiased expression, we statistically in contrast the gene expression between the L-ohnolog and S-ohnolog in all ohnolog pairs for every cell sort, and looked for ohnolog pairs during which there was no apparent distinction in gene expression between the L-ohnolog and S-ohnolog in any cell sort. The outcomes confirmed that 611 (5.3%) ohnolog pairs had no vital distinction in gene expression between the L-ohnolog and S-ohnolog in any cell sort (Fig. 4a, Supplementary Knowledge 8). These genes included a number of transcriptional components, similar to distal-less homeobox genes (dlx3b, dlx4a, dlx4b, dlx5a, and dlx6a). In 762 (6.7%) ohnolog pairs, a major distinction in gene expression was discovered between the L- and S-ohnologs solely in a single cell sort (Fig. 4a, Supplementary Knowledge 9). Within the remaining 10,071 (88.0%) ohnolog pairs, the L-ohnolog and S-ohnolog confirmed vital variations in gene expression in a minimum of two cell sorts (Fig. 4a). It's recommended that evolutionary constraints acted on the 611 ohnolog pairs and not using a vital distinction in gene expression between the L-ohnolog and S-ohnolog in any cell sort. To additional receive organic insights from the checklist of ohnolog pairs that confirmed no vital distinction in gene expression between the L-ohnolog and S-ohnolog in any cell sort, we carried out a purposeful enrichment evaluation. We discovered that these ohnolog pairs have been considerably associated to 295 organic classes, together with cytoplasmic ribosomal proteins (WikiPathways ID, WP324), chordate embryonic growth (GO:0043009), and cell destiny dedication (GO:0045165) (Fig. 4b; Supplementary Knowledge 10). A few of these 295 organic classes shared related ohnolog pairs. We generated an enrichment community based mostly on their membership similarities the place nodes represented organic classes and edges represented membership similarities with statistical significance. The ensuing enrichment community contained 20 clusters. Notably, we discovered one massive complicated cluster consisting of 12 clusters (enclosed clusters, Fig. 4c). The organic classes forming the 12 clusters have been composed of a number of transcription issue genes, together with dlx genes, homeobox protein NK-2 homolog (nkx2) genes, and forkhead field (fox) genes. These 12 clusters have been broadly associated to embryonic growth and contained organic classes similar to chordate embryonic growth (GO:0043009), formation of main germ layer (GO:0001704), central nervous system growth (GO:0007417), and cell destiny dedication (GO:0045165). Fig. 4: Useful enrichment evaluation of the ohnolog pairs with retained unbiased expression. a Of the 11,444 ohnolog pairs analyzed, 611 (5.3%) confirmed retained unbiased expression in every cell sort (orange). For the rest ofthe ohnolog pairs, expression biases towards one of many ohnologs have been detected in a single (yellow) or extra (blue) cell sorts. b Useful enrichment evaluation of the 611 ohnolog pairs that confirmed retained unbiased expression in every cell sort. The x axis represents the damaging log10-transformed P worth based mostly on the accumulative hypergeometric distribution67. The y axis represents the 20 organic classes. Deeper coloration of the bar plot means smaller P worth. c Community illustration of the statistically enriched classes within the purposeful enrichment evaluation of the ohnolog pairs with retention of unbiased expression. The nodes characterize the enriched classes and the sides are outlined based mostly on the similarities amongst their gene memberships. The title of the cluster is adopted from the title of the cluster with the smallest P worth among the many organic classes contained in that cluster. The dotted line highlights the big complicated cluster consisting of 12 clusters. The node measurement is proportional to the variety of enter ohnolog pairs grouped into every class. Full measurement picture Identification of ohnolog pairs exhibiting diversified expression patterns (sub/neo-functionalization) Beforehand, we carried out bulk RNA-seq on seven goldfish tissues (coronary heart, muscle, bone, mind, eye, gill, and tail fin) and located that 0.46% ohnolog pairs confirmed sub-functionalization and three.78% ohnolog pairs confirmed neo-functionalization11. We examined the expression sample of ohnologs with cell-type-specific expression to determine ohnolog pairs with sub- or neo-functionalization on the single-cell-level decision. First, we chosen the cell sort particular ohnologs among the many 11,444 ohnolog pairs by looking for ohnologs with considerably greater expression in a selected cell-type (cell-type specificity of particular person expression of L- or S-ohnolog). We recognized 632 ohnolog pairs during which each the L- and S-ohnologs confirmed cell-type-specific expression patterns (Fig. 5a; Supplementary Fig. 7, Supplementary Fig. 8; Supplementary Knowledge 11). Subsequent, we in contrast the expression patterns between the L-ohnologs and the S-ohnologs and divided them into two teams. The primary group contained the ohnolog pairs during which each the L- and S-ohnologs exhibited the same expression patterns (326 ohnolog pairs, 2.8%; Supplementary Fig. 7). The second group included the ohnolog pairs during which the L- and S-ohnologs confirmed totally different expression patterns (306 ohnolog pairs, 2.7%; Supplementary Fig. 8). The second group was most probably to comprise the sub/neo-functionalized ohnolog pairs (Supplementary Fig. 8). For instance, cyclin-dependent kinase 5 regulatory subunit 2b (cdk5r2b) is particularly expressed in zebrafish cone photoreceptor cells (Fig. 5a). Within the goldfish retina, the L-ohnolog was expressed within the rod photoreceptor cells and the retinal pigmented epithelial cells, along with the cone photoreceptor cells, whereas the S-ohnolog of cdk5r2b was particularly expressed within the cone photoreceptor cells, as noticed within the zebrafish retina (Fig. 5a). This implies that the L-ohnolog cdk5r2b skilled neo-functionalization, whereas its S-ohnolog preserved the unique gene perform. The solute provider household 3 member 2a (slc3a2a) gene, which encodes an amino acid transporter, was expressed in Müller glia and microglia within the zebrafish retina, whereas the L-ohnolog of this gene was expressed in Müller glia and the S-ohnolog of this gene was expressed in microglia (Fig. 5a). This implies that the 2 ohnologs of slc3a2a have undergone sub-functionalization. We examined whether or not the sub/neo-functionalized gene cluster discovered on scRNA-seq information (306 ohnolog pairs) overlapped with a beforehand reported sub/neo-functionalized gene cluster present in bulk RNA-seq of tissue (368 ohnolog pairs11). This evaluation confirmed that solely 4 genes (oxr1b, rgs16, nat16, and tubb2) overlapped in each teams, indicating that the sub/neo-functionalized genes discovered by scRNA-seq are considerably totally different from these obtained by bulk RNA-seq evaluation. This means that scRNA-seq evaluation is beneficial for figuring out sub/neo-functionalized genes. Moreover, we carried out purposeful enrichment evaluation on the 326 ohnolog pairs during which each the L- and S-ohnologs displayed related expression patterns (Supplementary Fig. 7; Supplementary Fig. 9a; Supplementary Knowledge 12) and the 306 ohnolog pairs during which the L- and S-ohnologs confirmed totally different expression patterns (Supplementary Fig. 8; Supplementary Fig. 9b; Supplementary Knowledge 13). We discovered that each units of ohnolog pairs have been considerably associated to organic classes similar to cytoplasmic ribosomal proteins (WP324; Supplementary Fig. 9). Notably, the classes of phototransduction (KEGG Pathway, dre04744) or cone photoresponse restoration (GO:0036368) have been present in solely the 326 ohnolog pairs during which each the L- and S-ohnologs exhibited related expression patterns (Supplementary Fig. 9). Per this, rhodopsin (rho) and opn1lw1 confirmed non-divergent expression in rods and cones, respectively (Supplementary Fig. 10a). Regardless of the low expression ranges, each L/S-ohnologs of opn1mw4 and opn1sw2 are expressed in cones. Fig. 5: Ohnolog pairs during which the L-ohnolog and S-ohnolog present totally different expression profiles. a Violin plots exhibiting the gene expression of ohnolog pairs during which the L-ohnolog (pink) and S-ohnolog (blue) exhibit totally different expression profiles. The scRNA-seq information of the zebrafish retina42 are proven on the left, for reference. b The cabp5a ohnologs exhibit a diversified expression sample within the bipolar cells of the UMAP cluster. The pink or blue arrowheads point out the enriched cells expressing cabp5aL or cabp5aS, respectively. Full measurement picture We recognized 326 ohnolog pairs during which each the L- and S-ohnologs exhibited related expression patterns (Supplementary Fig. 7); nonetheless, these ohnolog pairs might have diversified expression patterns in several subtypes of cells. To handle this risk, we looked for ohnolog pairs exhibiting diversified expression patterns within the UMAP cluster of every retinal cell sort. On this evaluation, we recognized a minimum of 4 ohnolog pairs, together with calcium binding protein 5a (cabp5a), 5b (cabp5b), retinoschisin 1a (rs1a), and prostaglandin-endoperoxide synthase 2b (ptgs2b) that confirmed diversified expression patterns within the UMAP clusters (Fig. 5b and Supplementary Fig. 10b). For instance, cabp5aL expression was enriched in sure subpopulations of the bipolar cell cluster (pink arrowheads in Fig. 5b), whereas cabp5aS expression was enriched within the totally different subpopulations of the bipolar cell cluster (blue arrowheads in Fig. 5b). These outcomes present that ohnologs are differentially expressed even on the subtype stage of retinal cells. Dedication of the open chromatin areas (OCRs) of particular person cells by scATAC-seq Chromatin accessibility within the promotor areas dynamically controls gene expression49. Subsequently, we questioned whether or not a divergent evolution of the L-ohnologs and S-ohnolog can also be noticed within the accessibility of promoter areas in single-cell decision utilizing single-cell Assay for Transposase-Accessible Chromatin sequencing (scATAC-seq). The scATAC-seq expertise profiles the accessible chromatin areas utilizing a genetically engineered hyperactive DNA transposase (Tn5) that cleaves and tags OCRs with single-cell resolution50. We carried out scATAC-seq on the retina of a single Wakin particular person and mapped OCRs in 19,750 cells. We recognized 245,817 OCRs throughout the genome, of which 132,681 (54.0%) have been positioned within the L-subgenome and 113,136 (46.0%) have been positioned on the S-subgenome (Supplementary Knowledge 14). To comprehensively quantify the promoter accessibility of the genes in every cell, we counted the variety of reads from OCRs positioned on the gene physique and the area as much as 2 kb upstream of the transcriptional begin web site in every of the 11,444 ohnolog pairs (22,888 genes). We recognized the seven forms of neurons that have been additionally recognized by the scRNA-seq evaluation, and the three forms of non-neuronal cells, together with Müller glia, microglia, and oligodendrocytes (Fig. 6a). These ten cell sorts had OCRs within the promotor areas of the identified cell-type marker genes (Fig. 6b). The promotor areas of L-ohnologs confirmed greater total accessibility than these of S-ohnologs in all cell sorts (Fig. 6c, left panel). Furthermore, for every cell sort, the promotor areas of L-ohnologs confirmed greater total accessibility than these of S-ohnologs (Fig. 6c, proper panel). Fig. 6: scATAC-seq evaluation of the goldfish retina. a UMAP plot exhibiting the mobile composition of the Wakin goldfish retina. The 19,750 cells have been projected right into a two-dimensional area by UMAP. b Promoter accessibility of the cell-type-specific marker genes. c Promoter accessibility of ohnologs. The left panel exhibits boxplots of the whole variety of open chromatin reads within the promotor areas in every cell. The suitable panel exhibits boxplots of the log2-transformed fold change of the whole variety of open chromatin reads within the promotor areas of L-ohnologs and S-ohnologs in every cell sort. The ends of the field are the 25 and 75% quantiles. The horizontal line within the field signifies the median. The traces extending from the highest and backside of the field characterize the minimal and most values. d Pearson correlation coefficient matrix between noticed cell sorts for every of the gene units utilizing the info from scRNA-seq and scATAC-seq evaluation. The correlations between scRNA-seq and scATAC-seq are considerably excessive between the corresponding cell sorts for any gene set (left panel, 2690 ohnolog pairs confirmed biased expression towards the L-ohnolog over the S-ohnolog; center panel, 2433 ohnolog pairs confirmed biased expression towards the S-ohnolog over the L-ohnolog; proper panel, 306 ohnolog pairs with sub/neo-functionalization). e Accessible chromatin panorama of every cell sort. The x axes characterize chromosomal positions (bp) and the y axesrepresent chromatin accessibilities. The peaks of scATAC-seq are highlighted by the black traces. The genome annotation relies on the NCBI Gnomon genome annotation of RefSeq goldfish genome meeting (GCA_003368295.1). Full measurement picture Then, we targeted on the potential regulatory areas underlying the expression patterns discovered on scRNA-seq evaluation. We in contrast vital peaks of OCRs within the scATAC-seq information with mRNA expression patterns on scRNA-seq information. To judge the connection between these two strategies, we calculated a Pearson correlation coefficient matrix between the noticed cell sorts for every gene set. We discovered considerably excessive correlations between the corresponding cell sorts for any gene set in scRNA-seq and scATAC-seq (Fig. 6d), suggesting that the gene expression profiles in retinal cell sorts from our scRNA-seq evaluation and the OCR profiles in retinal cell sorts scATAC-seq information have primarily related options. Utilizing these information, we predicted the candidate regulatory areas of the bias-expressed genes (L-bias 2690 ohnolog pairs, 6083 OCRs, Supplementary Knowledge 15; S-bias 2443 ohnolog pairs, 5810 OCRs, Supplementary Knowledge 16). These information present a complete view of the evolution of potential gene regulatory areas after WGD of the goldfish ancestor genome. Subsequently, we evaluated the accessibility of the gene our bodies and promoter areas of the genes that confirmed distinct expression patterns in scRNA-seq. We discovered that the accessibility of the gene our bodies and promotor areas exhibited a sample that was much like that of genes with distinct expression patterns measured by scRNA-seq (Fig. 6e). For instance, each ohnologs of ubiquitin c (ubc) have been ubiquitously expressed in scRNA-seq (Fig. 3c, left panel), and the promoter accessibility of each ohnologs confirmed an analogous sample (Fig. 6e). The L-ohnolog of aldolase, fructose-bisphosphate aa (aldoaa) was ubiquitously expressed (Fig. 3c, proper panel), and the S-ohnolog of proteasome 26 S subunit ATPase 6 (psmc6) was ubiquitously expressed (Fig. 3c, proper panel), and an analogous bias was noticed within the promoter accessibility (Fig. 6e). The arrestin 3a (arr3a), chemokine (C-X-C motif) receptor 4b (cxcr4b), and calcium-binding protein 5b (cabp5b) genes confirmed distinct expression patterns in cone photoreceptors, microglia, and bipolar cells, respectively (Fig. 3b), in addition to totally different promoter accessibility (Fig. 6e). These outcomes recommend that the bias towards L-ohnologs over S-ohnologs and distinct expression patterns noticed in scRNA-seq are additionally current in promoter accessibility on the particular person cell stage. Subsequent, we targeted on the evolution of regulatory areas for cell-type-specific ohnolog pairs. We recognized OCRs for sub/neo-functionalization genes (306 ohnolog pairs, 849 OCRs, Supplementary Knowledge 17). These information units are subsequently candidates for cell type-specific, subgenome-evolved regulatory areas. We additional used this information set to seek for Otx2/Crx binding sequences (Fig. 7a) related to photoreceptor/bipolar cell-specific gene expression within the retina. Otx2 and Crx are important transcription components for retinal growth and maintenance51. This evaluation revealed that Otx2/Crx binding sequences are considerably enriched within the OCRs of those cell sorts (P < 1.0e–15, Wilcoxon take a look at, Fig. 7b). We described 89 OCRs containing Otx2/Crx binding websites for regulation of ohnolog pairs with photoreceptor/bipolar cell-specific expression (Supplementary Knowledge 18). These OCRs are candidate regulatory areas for cell sort specificity of differentially expressed ohnolog pairs between L/S-subgenomes. For instance, scRNA-seq evaluation confirmed that prph2b, a photoreceptor particular glycoprotein, is extra extremely expressed within the L-subgenome than within the S-subgenome (Fig. 7c). Per this, scATAC-seq evaluation revealed vital OCRs with Otx2/Crx binding websites within the L-subgenome however not within the S-subgenome (Fig. 7c). Accordingly, our scATAC-seq information present info on gene regulatory areas necessary for the evolution of gene expression in retinal ohnolog pairs.
Leave a Reply