Barley AGO4 proteins show overlapping functionality with distinct small RNA-binding properties in heterologous complementation

Abstract

Barley AGO4 proteins complement expressional changes of epigenetically regulated genes in Arabidopsis ago4-3 mutant and show a distinct affinity for the 5' terminal nucleotide of small RNAs, demonstrating functional conservation and divergence. The function of Argonaute 4 (AGO4) in Arabidopsis thaliana has been extensively characterized; however, its role in monocots, which have large genomes abundantly supplemented with transposable elements (TEs), remains elusive. The study of barley AGO4 proteins can provide insights into the conserved aspects of RNA-directed DNA methylation (RdDM) and could also have further applications in the field of epigenetics or crop improvement. Bioinformatic analysis of RNA sequencing data identified two active AGO4 genes in barley, HvAGO4a and HvAGO4b. These genes function similar to AtAGO4 in an Arabidopsis heterologous complementation system, primarily binding to 24-nucleotide long small RNAs (sRNAs) and triggering methylation at specific target loci. Like AtAGO4, HvAGO4B exhibits a preference for binding sRNAs with 5' adenine residue, while also accepting 5' guanine, uracil, and cytosine residues. In contrast, HvAGO4A selectively binds only sRNAs with a 5' adenine residue. The diverse binding capacity of barley AGO4 proteins is reflected in TE-derived sRNAs and in their varying abundance. Both barley AGO4 proteins effectively restore the levels of extrachromosomal DNA and transcript abundancy of the heat-activated ONSEN retrotransposon to those observed in wild-type Arabidopsis plants. Our study provides insight into the distinct binding specificities and involvement in TE regulation of barley AGO4 proteins in Arabidopsis by heterologous complementation.

Keywords: Arabidopsis thaliana; Hordeum vulgare; ARGONAUTE 4 (AGO4); Heat stress; RNA-directed DNA methylation; Small RNAs; Transposable elements.

Fig. 1

A Phylogenetic tree of AGO4-clade protein sequences from Arabidopsis thaliana, Oryza sativa subsp. japonica and Hordeum vulgare was inferred using Maximum Likelihood method (1000 bootstrap repetitions) and JTT matrix-based model. Barley AGO4-like translated protein sequences were marked with a red square. B Visualization of the HvAGO4a and HvAGO4b gene structure, including the 5′ and 3′ UTR. Solid rectangles and lines indicate exons of the coding regions and introns, respectively. The UTR regions are marked with empty shapes. Within the coding region PAZ and PIWI domains were labeled with red and blue, respectively. The scale bar represents 500 bp. C Protein alignment of the PIWI domain region involved in the 5′ sRNA anchoring of the Arabidopsis, rice, and barley AGO4 proteins. Alignment was performed with ClustalW and was visualized using ESPript. The vertical black arrows and the black line indicate the amino acids directly involved in the binding domain, while the red arrow indicates the AA showing variations between Arabidopsis (dicot) and the monocots. D Transcript abundancy of putative AGO4-clade genes in H. vulgare cv. Golden Promise (developing inflorescences) expressed in TPM (calculated using Salmon). Error bars represent the mean ± SD, n = 3 (color figure online)

Fig. 2

A AtAGO4 and HA-tagged barley AGO4 gene expression levels determined by RT-qPCR of T1 mixed-stage inflorescences. Data were normalized using AtUBC9 and AtACT2. For individual primer pairs used to detect the three different AGO4 genes see Table S1. The average of 3 independent biological replicates was calculated and statistically significant differences from Columbia wild type (Col WT) are indicated with asterisks (Anova one-way with Dunnett’s post-hoc test, * < 0.05, ** < 0.01, *** < 0.001 and **** < 0.0001). Error bars represent the mean ± SD, n = 3. B HA-HvAGO4 protein level in mixed-stage inflorescences of T1 transgenic plants. To quantify the HA-HvAGO4, volume intensity of each sample was referred to the corresponding actin signal and was presented as the ratio of HA and Actin signal

Fig. 3

A AtSN1 expression levels determined by RT-qPCR analysis of T2 mixed-stage inflorescences. Data were normalized using AtUBC9 and AtACT2. B Methylation levels of AtSN1 locus determined using Chop-qPCR on digested and undigested DNA. Digestion was performed with MspJI, modification-dependent restriction endonuclease, and data were normalized to AtSN1 level from the undigested DNA and then the reciprocal was calculated to show the relative methylation levels. C AtROS1 expression levels determined by RT-qPCR analysis of T2 mixed-stage inflorescences. Data were normalized using AtUBC9 and AtACT2. All the results show the average of 3 independent biological replicates and were statistically analyzed against Columbia wild type (Col WT) and significant differences are indicated with asterisks (Anova one-way with Dunnett’s post-hoc test, * < 0.05, ** < 0.01, *** < 0.001 and **** < 0.0001). Error bars represent the mean ± SD, n = 3

Fig. 4

Sequencing of the HA-HvAGO4A- and HA-HvAGO4B-associated sRNA pools of Arabidopsis complementation plants. A Size distribution profile of filtered sRNA-IP data sets derived from the mean of 3 independent biological replicates. AtAGO4 IP raw data were retrieved from Sigman et al. (2021). Error bars represent the mean ± SD, n = 3. B Percentage of sRNA-IP read distribution based on their origin. C Percentage of the sRNA read distribution according to the 5′ nucleotide identity. D Graphical representation of the 24-nt long sequences conservation of nucleotides using sequence logos. The graphs represent only one data set per type; in particular, graphs of AtAGO4 #1, HvAGO4A #5, and HvAGO4B #17 are shown here. Maximum value in bits is 2 on the Y axis. Higher value for a nucleotide indicates a higher conservation

Fig. 5

A Venn diagram showing the number of TEs where the amount of sRNAs show at least twofold statistically significant (p-value < 0.05) change compared to AtAGO4. B Percentage of the AtSN1-derived sRNAs in the categories based on 5′-end nucleotide. Error bars represent the SD of the three sequenced lines. C SRNAs mapped on AtSN1 genomic locus. The reads are merged from the three independent biological replicate IP datasets. The quantity of mapped reads (counts) is written on the left side of the image inside the square brackets. Different colors indicate individual IP datasets: AtAGO4 (blue), HA-HvAGO4A (green) and HA-HvAGO4B (orange) (color figure online)

Fig. 6

A Relative copy number of ONSEN extrachromosomal DNA (ecDNA) in non-treated (NT) and heat stress (HS—24 h at 37 °C) 1-week-old Arabidopsis seedlings. Data were normalized using AtUBC9. B Relative expression of ONSEN before and after the heat stress activation measured by RT-qPCR and normalized on AtUBC9 and AtPP2AA3. Statistically significant differences compared to Col WT are indicated by asterisks (Anova one-way with Dunnett’s post-hoc test, * < 0.05, ** < 0.01, *** < 0.001 and **** < 0.0001). Error bars represent the mean ± SD, n = 3