Abstract
Small RNAs play important roles during plant development by regulating transcript levels of target mRNAs, maintaining genome integrity, and reinforcing DNA methylation. Dicer-like 5 (Dcl5) is proposed to be responsible for precise slicing to generate diverse 24-nt phased, secondary small interfering RNAs (phasiRNAs), which are exceptionally abundant in meiotic anthers of maize, rice, and other grasses 1. The importance and functions of these phasiRNAs remain unclear. Here, we used the Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) Cas9 system to mutate Dcl5. We report that dcl5 mutants have few or no 24-nt phasiRNAs, develop short anthers and defective tapetal cells, and exhibit temperature-sensitive male fertility. We propose that DCL5 and 24-nt phasiRNAs are critical for fertility under growth regimes for optimal yield.
Introduction
Three major classes of endogenous small RNAs (sRNAs) exist in plants: microRNAs (miRNAs), heterochromatic small interfering RNAs (hc-siRNAs), and phased, secondary small interfering RNAs (phasiRNAs). From extensive sRNA sequencing in plants, numerous loci generating phasiRNAs have been reported; in grasses, phasiRNAs are enriched in flowers, particularly male reproductive organs 1–4. PhasiRNA production is initiated by miRNA-mediated cleavage of RNA polymerase II transcripts of two classes of PHAS loci. Subsequently, the 3’ portion of cleaved transcripts is converted to double-stranded RNA, a substrate for precise chopping by DICER-LIKE 4 (DCL4) yielding 21-nt products; a distinct, proposed role for DCL5 is in the generation of 24-nt phasiRNAs 2–4. The 21-nt phasiRNAs are highly abundant during initial cell fate setting in maize anthers 1 and are important for male fertility in rice 5,6. The 24-nt phasiRNAs accumulate coincident with meiotic start, peak during meiosis, and persist at lower levels afterwards; this pattern has generated speculation that they regulate meiosis 7,8.
In angiosperms, five DCLs have been described and partially characterized 9. DCL1 is important for miRNA biogenesis, as illustrated by regulation of meristem determinacy in maize inflorescences 10. In Arabidopsis thaliana, DCL2 processes viral- and transgene-derived siRNAs 11; DCL3 produces 24-nt hc-siRNAs, which then direct DNA methylation of target loci 12. In many plants including Oryza sativa (rice), DCL4 generates 21-nt trans-acting siRNAs and phasiRNAs 3. Although the functions of these four DICER-LIKE genes are well conserved in flowering plants, these genes have partially overlapping functions 11,13. A fifth, more recently discovered gene, Dcl5, is monocot-specific 9. To better understand the function of DCL5 and its proposed role in 24-nt phasiRNA biogenesis 1, we characterized maize dcl5 mutants generated using a high efficiency CRISPR-mediated gene editing system 14.
Results
Loss-of-function mutants of the maize Dcl5 gene confer male-sterility
We picked four alleles for analysis (Fig. 1A), and from T0 plants we developed stable lines that show Mendelian inheritance of each dcl5 allele (Fig. 1B, Table S1, and Fig. S1). dcl5-1 and dcl5-4 are frameshift mutants with transcript levels down-regulated to about a third of their wild type siblings; in contrast, dcl5-2 (3 bp deletion) and dcl5-3 (12 bp deletion, 1 bp substitution) have similar or higher transcript levels compared to control siblings (Fig. 1B and Fig. S2). Wild type and heterozygous dcl5-1//Dcl5 plants greenhouse-grown at 32° C maximum day/21° C minimum night were identical in whole plant architecture, anther morphology, and fertility. Under the same conditions, homozygous dcl5-1 plants were male sterile (Fig. 1C). Compared to their fertile siblings, sterile dcl5-1 plants lacked visible differences in tassels and spikelets, however, the sterile anthers were shorter, contained shrunken pollen, and did not exert from the spikelets (Fig. 1C and Fig. S3). Sporadically, a few anthers exerted and shed viable pollen. Under field conditions, all four alleles showed equivalent levels of male sterility. No genomic editing can be detected in the homologous sequence in Dcl3 which is the most likely potentially off-target locus (Fig. S4). We concluded that Dcl5 is required for robust male fertility.
dcl5-1 mutants exhibit tapetal defects
To investigate whether dcl5-1 plants display defects in anther cell patterning, we used confocal microscopy and found normal somatic layer architecture without defects in initial cell differentiation (Fig. S5). In addition, chromosome pairing, alignment, and meiotic progression were normal in dcl5-1 meiocytes (Fig. S6). The dcl5-1 meiocytes routinely complete meiosis and produce haploid gametophytes. Transmission electron microscopy was utilized to visualize nuclei and other cell organelles at higher resolution. During mid-meiosis (2 to 2.5 mm anthers), normal tapetal cells were densely packed with dark-staining materials and were mostly binucleate (Fig. 2A and Fig. S7). In contrast, the tapetal cells in dcl5-1 were pale and mostly mononucleate, and many cell organelles were not clearly resolved (Fig. 2B and Fig. S7). Whole mount, fluorescence microscopy was used to quantify binucleate status in dcl5-1 and fertile anthers. In 1.5 mm anthers (prophase I), there were five-fold fewer binucleate tapetal cells in dcl5-1, and a significant difference persisted at 2.5 mm (meiosis II) (Fig. 2 and Fig. S8). In extensive confocal microscopic examination of dcl5-1 anthers, we observed no significant or consistent differences in cell number or volume compared to fertile siblings. Collectively, these observations indicate that tapetal development is delayed or arrested and that the dcl5-1 tapetal cells are therefore likely defective in conducting post-meiotic functions supporting pollen maturation.
Dcl5 is exclusively expressed in the meiocytes and tapetal cells at the onset of meiosis
To obtain a comprehensive spatiotemporal profile of maize Dcl5 expression, we queried microarray, published and newly generated RNA-seq (Table S2), and proteomics data. Dcl5 transcripts are highly enriched in tassels, cobs, embryos, and seeds 15,16, and in fertile anthers. In contrast, DCL5 protein is low in 1.0 mm, pre-meiotic anthers and extremely high in 2.0 mm mid-meiosis anthers, but is undetectable in ear, embryo, or endosperm 17. Microarray analysis of laser-microdissected anther cell types 18,19 and in situ hybridization analysis of anther lobes 1,20 established that Dcl5 transcripts are highly enriched in the tapetum and are present at lower levels in pre-meiotic pollen mother cells and meiocytes. RNA-seq confirmed Dcl5 expression in isolated maize meiocytes, however, quantitatively there are even higher levels in whole anthers 21, as confirmed by newly-generated anther data (Table S2). Therefore, Dcl5 is expressed much more highly in one or more somatic cell types than in meiocytes 21. Integrating prior observations that 24-nt phasiRNA biogenesis is contingent on a normal tapetum 1, with the peak of Dcl5 and DCL5 expression in meiotic anthers, we conclude that the 24-nt phasiRNA pathway is localized in tapetal cells. We hypothesize that this localization and timing is of functional importance for the redifferentiation of tapetal cells into secretory cells.
24-phasiRNAs abundance is greatly reduced in dcl5 mutants. To investigate if 24-nt phasiRNA biogenesis is affected in dcl5 mutants, sRNA libraries were constructed from anthers or spikelets of each dcl5 allele (Table S2). Previously, we found that 24-nt phasiRNAs were readily detected from 176 24-PHAS loci in W23 inbred anthers 1. The 24-nt sRNAs produced from all these loci were reduced dramatically in plants homozygous for each dcl5 allele compared to fertile siblings even though dcl5-2 and dcl5-3 encode proteins that lack only a few amino acids (Fig. 3). Other sRNAs, including 24-nt hc-siRNAs and 21-nt phasiRNAs were retained (Fig. S9). We conclude that Dcl5 is required for 24-nt phasiRNA biogenesis and that the dcl5-2 and dcl5-3 mutations define amino acids essential for DCL5 function. Analysis of 24-PHAS precursors in RNA-seq anther libraries from the same plants showed greater or nearly the same levels of precursor abundance in each of the dcl5 mutant alleles 4 (Fig. S10). Therefore, the absence of functional DCL5 severely disrupts 24-nt phasiRNA precursor processing with a modest impact on their accumulation.
Analysis of these RNA-seq data to characterize downstream transcriptional pathways impacted in dcl5 mutants was uninformative, consistent with prior results that failed to identify mRNA targets for 24-nt phasiRNAs 1. The transcriptional changes in dcl5-1 mutants were minimal with many of these impacted transcripts being downstream effects of the arrested tapetal development (Table S3) and proximal genes were not impacted; alterations in the transcript population are likely indirect and reflective of defective tapetal development.
dcl5-1 mutants are temperature-sensitive male sterile
dcl5-1 field-grown plants exhibited variable fertility under typical summer conditions in Stanford, CA: 25/18° C interrupted by multi-day heat waves exceeding 32° C. To test whether temperature is a restrictive condition, three controlled greenhouse regimes were chosen -- 28/22° C, 26/20° C, and 23/20° C (14 h day/10 h night temperature) -- and maintained within narrow limits over the life cycle (Fig. 4). At 28/22° C, dcl5-1 mutants had shorter anthers that never exerted, while heterozygous siblings developed normally. In contrast, under both the 26/20° C and 23/20° C regimes, dcl5-1 mutants were partially to fully fertile; pollen was viable based on Alexander staining despite a prolonged life cycle and delayed flowering date under cooler temperatures (Fig. S11). We also compared dcl5-1 transcript levels, 24-nt phasiRNA production, and PHAS precursor accumulation in dcl5-1 plants under restrictive and permissive temperature and we observed no significant differences (Fig. 3, Fig. S2 and Fig. S10). Our interpretation is that the functions of Dcl5 and 24-nt phasiRNAs are dispensable for maize male fertility at low temperatures (23/20° C) but are required at higher temperatures. Cool temperatures slow the pace of development and may allow alternative pathways independent of the 24-nt phasiRNAs to support tapetal cell redifferentiation.
Maize tassels contain anthers representing six days of development 22, and we wondered if sporadic anther exertion in dcl5-1 plants reflected a short phenocritical period when some anthers experienced permissive conditions. Homozygous dcl5-1 plants were greenhouse-grown at 28/22° C until the tassel inflorescence formed (~30 days), then plants were moved into two walk-in chambers: permissive 23/20° C or restrictive 28/22° C regimes. In the next three weeks, 14 sets of three plants were swapped between the two regimes for 3, 6, 9, 12, 15, 18 or 21 days, and then all plants finished their life cycle in the 28/22° C greenhouse (Fig. S12). Plants in the 23/20° C regime from the start of meiosis through the release of mononucleate microspores (~9 day period, swaps 7 to 11, Fig. S12) were fully fertile. Sample plants were assessed during the 6th through 9th days to check anther staging in the main tassel spike, confirming that this interval corresponds to meiosis and post-meiotic stages in the most mature part of the tassel. These stages also encompass the initiation, peak, and continued presence of 24-nt phasiRNAs in normal anthers 1. Plants in the permissive temperature for fewer than nine days during this interval had reduced fertility; we hypothesize that this extended interval is required to allow all anthers to proceed through the phenocritical period to ensure full fertility. The temperature swap experiment demonstrates that there is a short period in which DCL5 is required to “buffer” development at elevated temperatures. Surprisingly, the restrictive temperature is similar to what is considered optimal day temperature for the U.S. corn belt; historical data indicate that growth between 20° to 29° C generates optimal yield, with higher yields associated with warmer temperatures within this range 23,24.
Discussion
Maize DCL5 is essential for robust male fertility: dcl5 plants are temperature-sensitive during meiosis, show arrested tapetal development at this stage, and lack 24-nt phasiRNAs. In rice, perturbed 21-nt, premeiotic phasiRNAs confer photoperiod and temperature-sensitive male fertility 25–27. Molecular and in situ analyses indicate that the 24-nt phasiRNA biogenesis factors and these sRNAs are located primarily in the tapetum 1. For maize, we conclude that 24-nt phasiRNAs play a direct role in tapetal redifferentiation and only an indirect role in meiocytes or gametophytes. We hypothesize that the 24-nt phasiRNAs are essential for tapetal redifferentiation for biosynthesis and secretion of materials that support haploid microgametophytes. The molecular mechanisms by which 21- and 24-nt reproductive phasiRNAs act remain unclear.
Our study in maize establishes that DCL5 plays a unique role, having subfunctionalized from DCL3 for 24-nt phasiRNA biogenesis in anthers. Dcl5 transcripts are still expressed in diverse organs, however, protein and critical functions are restricted to anthers. Expression analysis found just 23 genes differentially expressed in our dcl5-1 mutants, but all of these are likely reflective of arrested tapetal development. That the thousands of 24-nt phasiRNAs fail to impact mRNA abundances significantly indicates that 24-nt phasiRNAs are distinct from tasiRNAs which do have specific mRNA targets and modulate transcript abundances. DCL5 and therefore the 24-nt phasiRNAs are dispensable at cool growing temperatures for maize, however, DCL5 is required for robust fertility at temperatures for optimal yield in the U.S. corn belt. As climate disruption, including more extensive heat waves, and a general warming trend increase field temperatures, buffering maize tapetal development from adverse effects may be possible by enhancing 24-nt phasiRNA functions. The 24-nt phasiRNAs may permit development during other untested environmental challenges. Given the high degree of variation typically observed across maize inbreds for many traits, the dcl5 mutant phenotype may vary across inbreds. More generally, it is unknown how many crop plants utilize reproductive phasiRNAs and whether their tapetal health and functions could benefit by introduction or optimization of these or other sRNA pathways 28–30.
Materials and Methods
gRNA construction and transformation
To increase the mutagenesis frequency in the targeted region, two target sites separated by 76 bp were selected for guide RNA (gRNA) construction, designed using the CRISPR Genome Analysis Tool as previously described 14. Rice U6 small nuclear RNA gene promoters (PU6.1 and PU6.2) were used to drive the expression of each gRNA. Guide RNAs were first cloned into a pENTR vector using BtgZI and BsaI (New England BioLabs) restriction enzymes. These constructs were transferred using Gateway recombination to the binary vector containing the Cas9 gene driven by the maize Ubiquitin 1 promoter and the bar gene driven by 35S promoter 14. The final construct was transformed into Agrobacterium tumefaciens for delivery into HiII maize immature embryos, followed by screening of positive calli as previously described 14.
Characterization of transformants
DNA samples were extracted from Basta®-resistant calli and screened for mutations in the target region by PCR using primers Dcl5_F1 (5’-ATCTAGATCTCCAGACCATTGAACCCTGTC-3’) and Dcl5_R1 (5’-GTATTCTAGACTTGAATAACCTGCTCTTTG-3’). Positive calli were transferred to regeneration media; multiple plants regenerated from the same callus were considered to be biological clones 14. No chimeric plants were detected, consistent with Cas9 action in a single cell. Leaves from mature T0 plants were genotyped for a second time using PCR amplification and Sanger sequencing to confirm the mutations. Over several generations families were established that segregated 1:1 for dcl5//+ fertile: dcl5//dcl5 homozygous sterile individuals using standard maize genetic procedures.
Microscopy and imaging
For anther wall structure analysis with confocal microscopy, the anthers of precisely measured sizes were dissected and stored in 70% ethanol for fixation. Fixed anthers were then stained by propidium iodide and visualized on a Leica SP8 confocal microscope as previously described 18. Meiocytes were extruded and stained by DAPI (4’,6-diamindino-2-phenylindole) following published protocols 31. The Alexander’s staining solution 32 was used to test the viability of the pollen grains in the permissive conditions. TEM procedures were followed as described 33, summarized as follows: anthers were dissected and fixed in fresh 0.1 M PIPES buffer (pH 6.8). A 2% osmium tetroxide stain was applied with brief washes, followed by dehydration in an acetone gradient. Specimens were infiltrated using a gradient of Spurr’s resin (Sigma-Aldrich) and embedded. The sections were cut using a Leica UCT ultramicrotome, stained in urayl and lead salts, and observed using a Zeiss TEM instrument with a LEO 912 AB energy filter.
ScaleP clearing of maize anthers
Maize anthers were fixed directly into 4% paraformaldehyde in phosphate-buffered saline (pH 7.4), and then prepared following the ScaleP protocol 34. After 8 h KOH treatment, anthers were stained for a week at room temperature with 1 mM Calcofluor white for imaging of cell walls and 5 μM Syto13 (ThermoFisher) for imaging nuclei. Samples were then cleared for three days with ScaleP solution consisting of 6 M urea, 30% (v/v) glycerol, and 0.1% (v/v) Triton X-100 in sterile water. Cleared samples were then imaged with a Zeiss 880 multi photon confocal microscope with a 40× LD C-Apochromat water lens (numerical aperture of 1.1; working distance of 0.62 mm). The 3D rendering was performed with Amira (FEI, ThermoFisher) software. Tapetal cells were classified as mononucleate or binucleate, and then artificially colored.
sRNA-seq and RNA-seq library construction and sequencing
Total RNA for sRNA-seq and RNA-seq libraries was isolated using the PureLink Plant RNA Reagent (ThermoFisher) following the manufacturer’s instructions. Total RNA quality was assessed by denaturing agarose gel electrophoresis and quantified by the Qubit RNA BR Assay Kit (ThermoFisher). For library preparation, 20 to 30 nt RNAs were excised from a 15% polyacrylamide/urea gel, and ~25 ng of sRNA used for library construction with the TruSeq Small RNA Prep Kit (Illumina) following the manufacturer’s instructions. For RNA-seq, 2 μg of total RNA was treated with DNase I (New England BioLabs) and then cleaned with RNA Clean and Concentrator-5 (Zymo Research). The TruSeq Stranded Total RNA with RiboZero-Plant Kit (Illumina) was used for library construction with 500 ng of treated RNA, following the manufacturer’s instructions. Sequencing in single-end mode on an Illumina HiSeq 2500 (University of Delaware) yielded 51 bp reads for both sRNA-seq and RNA-seq.
sRNA-seq data were processed as previously described 35. Briefly, we first used Trimmomatic version 0.32 to remove the linker adaptor sequences 36. The trimmed reads were then mapped to version 4 of the B73 maize genome using Bowtie 37. Read counts were normalized to 20 million to allow for the direct comparison across libraries. phasiRNAs were designated based on a 24-nt length and mapping coordinates within the previously identified 176 24-PHAS loci 1, updated to version 4 of the B73 genome using the assembly converter tool 38. If a 24-nt phasiRNA did not uniquely map to the genome, we divided the abundance equally to each location to which the read mapped, i.e. a hits-normalized abundance. These hits-normalized abundances were then summed for each of the 176 loci to calculate the 24-nt phasiRNA abundances.
RNA-seq libraries were trimmed as above, and mapped to version 4 of the B73 genome using Tophat version 2.0.12 39. For differential expression analysis of genes, we assembled the transcripts using the Cufflinks package 40 and raw gene expression levels were quantified using featureCounts version 1.5.0 41 to generate count tables that were imported into R for statistical analysis 42. Weakly expressed transcripts (fewer than 1 read) were filtered out, and the remaining transcripts were normalized using DEseq2 to finally identify a set of differentially expressed genes for anthers homozygous for each mutant allele 43. Due to a lack of a replicate for the dcl5-1 restrictive library, we calculated dispersions using the mean of libraries to identify differentially expressed transcripts without a complete set of replicates.
For differential expression analysis of the PHAS precursors, we were unable to assemble the transcripts that mapped to these regions because of the absence of annotation in any feature files. Instead, we conducted this analysis in parallel to the analysis of the sRNA data. We first normalized the mapped fragmented reads as counts per 15 million, and applied the hits-normalized abundance approach described above. Using the annotated 24-PHAS loci (above), we then identified the summed abundance of RNA-seq transcript fragments that mapped to these regions to identify a representative abundance for each 24-PHAS precursor. Replicate libraries were averaged together prior to the creation of the heatmap, and we added 1 to all abundances in order to prevent log2(0) error while also giving the added benefit of removing negative log transformed values. Boxplots were generated for each replicate’s log transformed values.
Funding
This work was supported by U.S. National Science Foundation Plant Genome Research Program (NSF-PGRP) award #1649424. Author contributions: H.Z., B.C.M, and V.W. conceived of the project, H.Z., C.T., and K.H. performed and interpreted experiments, R.Z. analyzed sequencing data, H.Z. and C.T. wrote the manuscript with editing by V.W., B.C.M, and with input from all co-authors.
Competing Interests
V.W. and B.C.M declare that a patent application was filed concerning the use of Dcl5 for modulating male fertility.
Data and materials availability
Correspondence and requests for materials should be addressed to walbot{at}stanford.edu. Raw data were submitted to GEO under GSE122449 (including sRNA-seq as GSM3466686 to GSM3466700 and RNA-seq data as GSM3466701 to GSM3466714), and the processed data are available via our maize genome browser at https://mpss.danforthcenter.org.
Acknowledgments
The dcl5 CRISPR mutants were generously generated by Dr. Bing Yang’s group at Iowa State University. TEM images were produced and imaged by Dr. Howard Berg at the Donald Danforth Plant Science Center. We thank the greenhouse staff at the Donald Danforth Plant Science Center for facilitating the temperature swap experiment, Sandra Mathioni for the preparation of small RNA and RNA-seq libraries, and Jeff Caplan for advice on microscopy.