Introduction

Leucine-rich repeat proteins (LRRs) constitute a large and widespread superfamily found in all eukaryotes (Kobe and Deisenhofer 1994; Kajava 1998). Their defining feature is their namesake domain composed of short, tandemly repeated leucine-rich motifs, each 18–29 amino acids in length. Differences in length and consensus sequence of the unit motif form the basis for classification of LRRs within the superfamily (Claudianos and Campbell 1995; Kajava 1998). While found on diverse proteins, LRR domains serve a common purpose, forming platforms that mediate specific protein–protein interactions important for function (Kobe and Deisenhofer 1994; Kobe and Kajava 2001).

Many LRR proteins take part in cell signaling pathways, often in developmental contexts. One example for this is the “Ras-Group” LRRs found in animals and fungi (Claudianos and Campbell 1995; Buchanan and Gay 1996). Some members of this class are directly involved in Ras-mediated signal transduction, such as C. elegans SUR8 and human ERBIN, which mediate Ras interactions with larger regulatory complexes (Sieburth et al. 1998; Sternberg and Alberola-Ila 1998; Li et al. 2000; Dai et al. 2006), and mammalian Rsu/Rsp-1, which can suppress Ras-mediated transformation (Cutler et al. 1992; Dougherty et al. 2008). Another Ras-group LRR, FLI-1, binds transcription complexes and takes part in hormone-induced chromatin remodeling in mammals (Lee et al. 2004; Jeong et al. 2009), indicating that Ras-group LRRs may have diverse biochemical functions.

In plants as in other eukaryotes, LRRs function in a wide range of developmental processes (Vernon and Forsthoefel 2002). Prominent examples of major plant LRR families include the LRR-receptor kinases, which feature extracellular N-terminal LRR domains important for ligand recognition in numerous developmental and physiological pathways (Morris and Walker 2003; Dievart and Clark 2004; Morillo and Tax 2006; De Smet et al. 2009). Some F-box proteins that mediate signal-responsive protein degradation also feature LRR domains (Hellmann and Estelle 2002; Somers and Fujiwara 2009), and two different classes of disease resistance proteins, NBS-LRRs and cf-type receptors, feature LRR domains involved in pathogen recognition or response (Fluhr 2001; Belkhadir et al. 2004; McHale et al. 2006; Eitas and Dangl 2010).

Plant intracellular Ras-group-related LRRs (PIRLs) are a small protein class featuring a conserved internal LRR domain comprised of unit motifs most closely related to those found in animal and fungal Ras-group LRR proteins (Forsthoefel et al. 2005). The LRR domains are flanked by divergent hydrophilic terminal regions. PIRLs are a plant-specific protein family: no close homologs with similarity beyond the LRR motifs are identified in other eukaryotes. Beyond the LRR domain, PIRLs lack other features found in some major LRR classes in plants, such as signal or transmembrane sequences, kinase domains, or nucleotide binding sequences (Morris and Walker 2003; Morillo and Tax 2006; McHale et al. 2006, Eitas and Dangl 2010). Based on this combination of features, PIRLs constitute a novel class of plant LRRs related to Ras-group proteins by virtue of their LRR motif sequence. The Arabidopsis thaliana genome contains nine PIRL genes that fall into three subfamilies, the largest of which, sub-family I, consists of PIRL1, PIRL2, PIRL3, and PIRL9. Within this group, PIRL1 and PIRL9 are the most closely related, with products sharing 77% amino acid identity. These sequence relationships and the widespread expression of PIRL subfamily I mRNAs (Forsthoefel et al. 2005) make for potential functional redundancy between these genes.

The structural relationship between PIRLs and Ras-group LRRs suggests potential functions for the PIRLs in plant development, and we have undertaken a systematic reverse genetic investigation of Arabidopsis PIRLs using T-DNA knockout mutants to investigate this hypothesis. Here, we report that the PIRL1 and PIRL9 genes are genetically redundant and required for differentiation of microspores into pollen, a plant-specific developmental process crucial for reproduction. Loss of PIRL1 and PIRL9 gene functions results in transmission failure specifically through the male gametophyte due to arrest of microspore development just prior to the first pollen mitosis. These results establish that PIRL1 and PIRL9 have redundant functions essential at a key transition point early in the development of the male gametophyte, and they provide the first functional information on this novel class of plant LRR proteins.

Materials and methods

Plant materials

Plants were grown on soil from seed under 16 h light/8 h dark conditions as described by Cushing et al. (2005). In initial work with prospective T-DNA-insertion lines from the Wisconsin collections (Sussman et al. 2000), seeds were plated in culture on Murashige and Skoog medium supplemented with 1% sucrose, following surface sterilization as previously described (Vernon and Meinke 1994; Vernon et al. 2001), in the presence of 50 μg mL−1 kanamycin, to identify plants with T-DNA inserts.

Mature pollen for viability staining or scanning electron microscopy (SEM) was collected on slides from open flowers from plants that had been previously genotyped by PCR (see genetic analysis, below). Pirl1/PIRL1;pirl9 plants, which produce 50% double-mutant pollen, were used for analysis of the pirl1;pirl9 pollen phenotypes unless otherwise indicated in “Results”. For characterization of microspore and pollen development, developing flowers were excised from wild-type controls or previously genotyped plants segregating for single- or double-mutant pollen, and subjected to clearing or anther-squashing treatments as described below.

Mutant identification, genetics, and molecular characterization

Sequence data for PIRL mRNA and proteins were previously deposited in Genbank and protein databases (Forsthoefel et al. 2005), and the gene nomenclature is registered at TAIR (http://www.arabidopsis.org). Nucleotide accessions: AY849571 (PIRL1); AY849579 (PIRL9). Protein accessions: AAW57410 (PIRL1) and AAW57418 (PIRL9). TAIR Atg loci designations: At5g05850 (PIRL1) and At3g11330 (PIRL9).

The pirl1-1 and pirl9-1 knockout lines were identified among pooled lines provided by the University of Wisconsin Knockout Facility, using the prescribed PCR screening and Southern blot strategy (Krysan et al. 1999; Sussman et al. 2000). T-DNA-specific primers were those dictated by the knockout facility protocols: JL202 (left-border specific), CATTTTATAATAACGCTGCGGACATCTAC; XR2 (right-border specific), TGGGAAAACCTGGCGTTACCCAACTTAAT. Gene-specific primers for knockout screens were: AGATCATTCTCAAGCCCATAAAACATTTG (PIRL1-RG, forward), GCAGGAAAACTCAAGACTGAAGATCACTT (PIRL1-RG, reverse), AACTTGACATCAGAGAAAACGCAGAGAAG (PIRL9-RG, forward), and AAACGAGAAATTTCAACCTCTTTGGATTC (PIRL9-RG, reverse). Plants containing the mutant alleles were sequentially backcrossed to wild type (ecotype Wassilewskija, WS) three times prior to phenotype analysis or further genetic experiments.

Both pirl1-2 alleles and the pirl9-2 allele were identified in the publically available sequence-indexed Salk insertion mutant collection (Alonso et al. 2003; ecotype Columbia), and seeds were obtained from the Arabidopsis Biological Resource Center (ABRC, Ohio State University). Salk designations for these lines were S072332 and S146770 for PIRL1 and PIRL9, respectively. Detailed information on T-DNA insert positions and orientations for these alleles is available at the Arabidopsis Information Resource (TAIR, http://www.arabidopsis.org). T-DNA insert locations in all lines were confirmed by PCR of T-DNA–plant junction fragments using genomic DNA preparation as described by Tax and Vernon (2001) and PCR as described by Cushing et al. (2005). Primers used for T-DNA–plant junction amplification were the LBb1 T-DNA left-border primer (GCGTGGACCGCTTGCTGCAACT) in combination appropriate gene-specific primers: PIRL1 forward primer, CCTCAATCTCCCAATCGTCTCTTCC; reverse primer, GCAGCGGCAGGAAAACTCAAGAC; PIRL9 forward primer, TCGTCACTCAGATGCCTCAC; reverse primer, CACCTTCCTTCACCACTTCC. Specificity and product identity for these gene-specific primers had been previously verified by product sequencing (Forsthoefel et al. 2005).

Genotyping to identify mutant homozygotes, heterozygotes, and carriers of both mutations was done by PCR, using gene-specific forward and reverse primer combinations to detect wild-type alleles, and T-DNA:PIRL primer combinations to detect the presence of mutant alleles. DNA was isolated from individual leaves as described by Tax and Vernon (2001), and subjected to PCR as described by Cushing et al. (2005), using the primers described above for verification of T-DNA insert location. Analysis of pirl1 and pirl9 mutant allele segregation and transmission was done using this genotyping procedure on populations of seedlings produced by either self-fertilization or by controlled crosses between genotypes, as indicated for each experiment in the “Results” section.

Crosses to qrt1 were carried out by crossing pirl1;pirl9/PIRL9 individuals to homozygous qrt1 plants grown from qrt1 seed (Preuss et al. 1994) obtained from the ABRC (Ohio State University). The pirl parent served as female for these crosses to insure transmission of both mutant alleles to 50% of the F1s. F1 individuals were genotyped by PCR to identify individuals heterozygous for both pirl knockout alleles. These plants were then selfed, F2 seed collected, and individuals homozygous for qrt1 (25%) identified in the F2 by visual screening of pollen. Qrt1 individuals were subjected to PCR genotyping to identify qrt1;pirl1/PIRL1;pirl9 plants for pollen analysis. Pollen tetrads were subjected to Alexander’s staining, then quantified by bright field microscopy to determine the percentage of aborted pollen and quantify its distribution in tetrads.

Sequence analyses and amino acid sequence predictions were done with empirically determined cDNA sequences (Forsthoefel et al. 2005), using MacVector software (Accelrys, San Diego, CA, USA). Protein domain predictions were carried out using the Simple Modular Architecture Research Tools (http://smart.embl-heidelberg.de/; Letunic et al. (2009). RNA isolation and RT-PCR to determine null status of mutant alleles were carried out as described by Cushing et al. (2005), using the gene-specific forward and reverse primer pairs described above. Reactions were carried out for 35 cycles (94°C for 45 s, 58°C for 60 s, and 72°C for 120 s). PCR results were visualized on 1% agarose gels.

Pollen development and microscopy

For pollen SEM, mature flowers from 3–10 PCR-genotyped plants of each genotype were collected and dried for at least 2 days. Pollen was transferred to a stub and sputter-coated with gold–palladium at 18 mA for 120 s under partial vacuum with Argon using a Pelco Sputter Coater Model 3 (Ted Pella, Redding, CA, USA). Samples were viewed in a JSM-T300 scanning electron microscope (JEOL USA, Peabody, MA, USA), and images were collected on film, or they were viewed and digitally imaged using an FEI Quanta200 scanning electron microscope with variable vacuum and Peltier cold stage capabilities (FEI, Hillsboro, OR, USA).

Viability staining was carried out using a 1:50 dilution of Alexander’s stain stock solution [10 mL 95% ethanol, 5 mL 1% malachite green in 95% ethanol, 5 g of phenol, 5 mL 1% acid fuchsin in H2O, 0.5 mL 1% orange G in H2O, 2 mL glacial acetic acid, 25 mL glycerol, and 50 mL H2O (Alexander 1969; Johnson-Brousseau and McCormick 2004)]. For samples, plants were genotyped using the PCR protocol described above, and stained according to the method of Henry et al. (2005), using 3–5 flowers excised from 3–9 separate plants of each indicated genotype. Pollens were observed by bright field microscopy, and staining was quantified manually.

For microscopy of developing microspores and pollen, flowers of various developmental stages were excised from pirl1/PIRL1;pirl9 plants at the pedicel, dissected to expose anthers and treated as follows. For fluorescence microscopy, anthers were removed, gently squashed under coverslips in 25 μl 4′,6-diamidino-2-phenylindole (DAPI) solution [0.1 M sodium phosphate (pH 7), 1 mM EDTA, 0.1% Triton X-100, 0.5 μg mL−1 DAPI, high grade], and viewed by bright field and UV epi-illumination on a Leica DMIRB microscope equipped with a Coolsnap camera (Photometrics, Tucson, AZ, USA). For callose staining, anthers were placed on slides in 25 μl 0.1% aniline blue in 0.1 M phosphate buffer (pH 8.6) and similarly viewed by bright field and UV epi-illumination. Anther developmental stages were designated according to the conventions described by Sanders et al. (1999). Histological and transmission electron microscopy (TEM) samples were prepared by the Franceschi Microscopy and Imaging Center at Washington State University (Pullman, WA, USA). Inflorescence buds were excised from PCR-genotyped pirl1/PIRL1;pirl9 plants segregating for double-mutant microspores. Samples were fixed, dehydrated, and embedded according to Trump et al. (1961), sectioned at 1 μm, and stained for light microscopy with toluidine blue (0.5% in 1% borax) for 0.5–2 min, or for TEM with 2% uranyl acetate for 8 min, followed by Reynolds lead for 8–10 min.

Results

Identification of pirl1 and pirl9 T-DNA knockout alleles

PIRL1 and PIRL9 comprise the most closely related gene pair in the Arabidopsis PIRL family. The predicted products have similar domain organization, with a characteristic PIRL LRR region consisting of nine tandem leucine-rich motifs (Forsthoefel et al. 2005), flanked by 204 amino acid N-terminal and 76 amino acid C-terminal regions, both of which are hydrophilic and feature charged residues and some low-complexity sequences (Fig. 1). A potential coiled coil is predicted N-terminal to the leucine-rich repeats. PIRL1 and PIRL9 are encoded by the unlinked At5g05850 and At3g11330 loci, respectively. The structural annotations of both genes have been empirically confirmed, and mRNAs for both are widely expressed during plant development, based on RNA blots, reverse-transcription PCR (RT-PCR), and Genevestigator data (Forsthoefel et al. 2005; Zimmermann et al. 2004).

Fig. 1
figure 1

PIRL1 and PIRL9 gene structures, knockout alleles, and predicted protein features. a Predicted PIRL1 and PIRL9 protein domain architecture. Dark boxes Leu-rich motifs constituting the LRR domain; white hydrophilic N-terminal region; lightly shaded region predicted coiled-coil domain. Length in amino acids is indicated at right (PIRL1/PIRL9). Amino acid sequences and alignments for the PIRL protein family have been described previously (Forsthoefel et al. 2005). NCBI protein accessions and TAIR locus designations: PIRL1, AAW57410 and At5g05850; PIRL9, AAW57418 and At3g11330. b Gene maps and T-DNA insertion alleles used in this study. Exons are shaded, with coding regions in black; introns are white. Length in nucleotides is indicated at right. Arrowheads mark confirmed T-DNA insert positions, with insertion site nucleotide positions indicated in parentheses. Black arrows mark positions of primers used in c. c Confirmation of knockout status by RT-PCR. PolyA RNA prepared from pirl1 or pirl9 single-mutant homozygote seedlings was subjected to RT-PCR to detect the presence of PIRL transcripts, and products were resolved on 1% agarose gels. Knockout mutants from which RNA was isolated are indicated above each respective gel. G, genomic PCR product shown as control for primer efficacy; P1 or P9, PIRL1 or PIRL9 RT-PCR amplification products (lacking in corresponding homozygous mutants); plus amplification of transcripts from the non-disrupted PIRL locus, included as positive controls for RT-PCR efficacy

Full size image

Two knockout alleles for each locus were identified using the Wisconsin and Salk Arabidopsis T-DNA insertion mutant collections (Sussman et al. 2000; Alonso et al. 2003). Wisconsin alleles were designated pirl1-1 and pirl9-1, and Salk alleles pirl1-2 and pirl9-2. Gene maps showing T-DNA positions are shown in Fig. 1b. Insert positions were confirmed by PCR and sequencing to verify T-DNA insertion site and structure and to ensure that additional T-DNA-associated chromosomal rearrangements were not present, as is occasionally the case with T-DNA insertion lines (Tax and Vernon 2001). T-DNAs in all alleles were located within PIRL transcription units. Mutants were subjected to serial backcrossing to wild type to remove the influence of any potential secondary mutations. RT-PCR was used to confirm knockout status for each of the four mutant alleles using RNA from single-mutant homozygotes. Expression of mRNAs from disrupted loci was not detected, consistent with the internal location of T-DNA inserts (Fig. 1c).

Genetic analysis reveals a role for PIRL1 and PIRL9 in the male gametophyte

Pirl1 and pirl9 homozygous mutants were recovered routinely and resembled wild type. PCR analysis of mutant allele transmission in F1 progeny of selfed heterozygotes and of heterozygotes crossed to wild type further demonstrated that single-mutant homozygotes are viable, and that the individual knockout alleles exhibit normal segregation and transmission (Table 1). To investigate possible functional redundancy between these related genes, we then attempted construction of double mutants. However, double pirl1;pirl9 mutants could not be identified in progeny populations of selfed pirl1/PIRL1;pirl9 or pirl1;pirl9/PIRL9 plants (n = 142; χ2 = 47.3; P ≫ 0.005), suggesting lethality or gametophytic transmission failure of the pirl1;pirl9 mutant combination. The possibility of embryo lethality was ruled out through observation of siliques and developing seeds using established methods for analysis of embryo-defective mutants (Meinke 1995; Vernon and Meinke 1994).

Table 1 Normal genetic transmission of pirl1 and pirl9 single knockout alleles and viability of single-mutant homozygotes
Full size table

To investigate the possibility of gametophytic transmission failure, reciprocal crosses to wild type were carried out with pirl1/PIRL1;pirl9 plants, which produce 50% pirl1;pirl9 double-mutant gametophytes. F1 progeny were genotyped by PCR to assess allele transmission; results are shown in Table 2. The pirl1;pirl9 allele combination was transmitted normally through the female parent, resulting in approximately 50% of F1 progeny containing mutant alleles for both pirl1 and pirl9 when wild-type plants were used as pollen donors. However, when pirl1/PIRL1;pirl9 plants were used as pollen donors, no pirl1 mutant allele transmission was observed; all F1 progeny contained a wild-type PIRL1 allele. This male-specific transmission failure was observed with both the pirl1-1;pirl9-1 and the pirl1-2; pirl9-2 allele combinations (Table 2), verifying that it was due to PIRL gene disruption.

Table 2 Male-specific gametophytic transmission failure of the pirl1,pirl9 double-mutant allele combination
Full size table

Inviability of pirl1;pirl9 pollen

The observed pirl1;pirl9 transmission failure could have been due to a defect in pollen development and viability, or to flaws in subsequent steps of pollen function, such as pollen tube elongation, guidance, or fertilization itself. To investigate the developmental basis of transmission failure, mature pollen produced by wild type and pirl1/PIRL1;pirl9 plants was examined by SEM. Plants segregating for the double-mutant allele combination produced approximately 50% grossly distorted and shrunken pollen (Fig. 2), a frequency corresponding to the expected meiotic segregation of the pirl1;pirl9 double-mutant genotype. In contrast, wild-type plants and single-mutant homozygotes produced deformed pollen at much lower frequencies, with single mutants (which generate 100% mutant pollen) only producing 6–8% abnormal pollen (Fig. 2f).

Fig. 2
figure 2

Severe morphological defects in prospective pirl1;pirl9 double-mutant pollen. Dehisced pollen was collected from wild type, pirl single mutants, or pirl1/PIRL1;pirl9 plants and viewed by SEM. Plant genotypes were determined by PCR as described in “Materials and methods”. a Pollen from a pirl1/PIRL1;pirl9 plant, showing prospective double-mutant pollen segregating with phenotypically normal grains. Bar 10 μm. b, c Wild type and prospective double-mutant pollen, respectively; bar 5 μm. d, e Similar exine patterning on wild type (d) and prospective double-mutant pollen (e), despite morphologic deformity. f Frequency of distorted pollen produced by wild type, pirl single mutants (which produce 100% single-mutant pollen), and pirl1/PIRL1;pirl9 plants (which produce 50% pirl1;pirl9 pollen). Mean values (±SE) derived from 3–10 plants of each genotype are plotted. n = 597, 128, 88, 1,341, respectively

Full size image

In some Arabidopsis mutants, pollen lethality has been associated with defective exine patterning or synthesis (Paxson-Sowders et al. 2001; Dong et al. 2005; Guan et al. 2008). However, SEM indicated that despite severe morphological defects, exine patterning on the prospective pirl1;pirl9 pollen resembled that of wild type.

Alexander’s staining was used to assess viability of pirl1;pirl9 mutant pollen. Approximately 42–44% of pollen produced by pirl1/PIRL1;pirl9 plants was unambiguously aborted based on this staining (Fig. 3), which provides a conservative estimate of pollen viability (Dafini et al. 2005; Baez et al. 2002). This corresponds to the predicted segregation of the double-mutant genotype in pollen produced by such parents. The high frequency of pollen inviability suggested that the pollen transmission failure observed in reciprocal crosses resulted from a highly penetrant defect in pollen development, rather than later defects in pollen tube elongation, guidance, or fertilization.

Fig. 3
figure 3

Inviability of prospective pirl1;pirl9 double-mutant pollen. Mature pollen from flowers excised from parent plants of indicated genotypes were subjected to Alexander’s staining to assess viability. Genotypes were determined by PCR as described in “Materials and methods”. Homozygous pirl1 or pirl9 single mutants produce 100% single-mutant pollen, pirl1/PIRL1;pirl9 (p1/P1;p9) plants produce 50% pirl1;pirl9 double-mutant pollen. a Photograph of pollen from a pirl1/PIRL1;pirl9 plant, illustrating segregation of inviable pollen (blue-green). Bar 10 μm. b Quantification of Alexander staining results. Mean values (±SE) derived from 3–10 plants of each genotype are plotted. n values: 968, 544, 747, 1,622, 364, 2,378, 1,103, respectively

Full size image

The pirl1 and pirl9 mutations affect pollen development after meiosis

Double-mutant allele transmission and phenotype segregation suggested pirl1 and pirl9 act after meiosis. To confirm this, pirl1 and pirl9 were crossed into a qrt1 background (Preuss et al. 1994), and segregation of the pollen-defective phenotype was evaluated in pollen tetrads produced by qrt1;pirl1/PIRL1;pirl9 individuals. Results are shown in Fig. 4. Tetrads produced by qrt1 control plants, or qrt1 plants homozygous for either only pirl1 or pirl9, produced only a background frequency of abnormal pollen, and the rare aborted pollen were observed mainly as single aborted grains in an occasional tetrad. In contrast, qrt1;pirl1/PIRL1;pirl9 plants overwhelmingly produced tetrads with two normal and two aborted pollen grains, consistent with the meiotic segregation of pirl1,pirl9 genotype (Fig. 4b). These results indicated that the pirl1 and pirl9 mutations act after meiosis to affect microspore or pollen development.

Fig. 4
figure 4

Post-meiotic segregation of the pirl1;pirl9 mutant phenotype in a qrt1 background. pirl1-1 and pirl9-1 mutant alleles were crossed into qrt1, and plants of indicated genotypes were identified in the F2 by PCR as described in “Materials and methods”. Segregation of aborted pollen produced by parent plants of each genotype was determined by Alexander’s stain. pirl1/PIRL1;pirl9 (p1/P1;pirl9) plants were predicted to produce tetrads with two aborted pollen if the phenotype were caused by post-meiotic expression of the pirl1;pirl9 allele combination. a Representative tetrads produced by a qrt1;PIRL1;PIRL9 control plant. b Tetrads produced by a qrt1;pirl1/PIRL1;pirl9 plant that segregates for 50% pirl1;pirl9 pollen. c Comparison of the distribution of aborted pollen in tetrads produced by control qrt1 plants, pirl1 and pirl9 single mutants, and qrt1;pirl1/PIRL1;pirl9 plants. Bar colors represent tetrads with different numbers of aborted pollen, for each genotype: white tetrads with zero aborted pollen; lightest blue shading tetrads with one aborted; medium (red) shading tetrads with two aborted; darkest blue shading tetrads with three aborted pollen. n values, number of tetrads: 411, 346, 468, and 890, respectively

Full size image

Arrested growth and differentiation of pirl1;pirl9 microspores

To determine when the PIRL1 and PIRL9 genes are first required, we examined pollen development in pirl1/PIRL1;pirl9 anthers, which segregate 50% for pirl1;pirl9 pollen. Anther sections stained with toluidine blue were examined first to determine when mutants first became morphologically distinct from segregating normal pollen. Results are shown in Fig. 5. Double mutants could not be distinguished from wild-type microspores in tetrads viewed just after meiosis, or in early free microspores. However, differences in size and morphology were detected in anthers during stage 10 when microspores normally undergo the first pollen mitosis and initiate vacuolar fission (Regan and Moffatt 1990; Sanders et al. 1999; Yamamoto et al. 2003). First, double mutants appeared smaller than phenotypically normal pollen segregating within the same anthers, and they retained the large vacuole, single large nucleus, and homogenous toluidine blue staining characteristic of uninucleate microspores in stage nine anthers (Regan and Moffatt 1990; Sanders et al. 1999). As anthers approached maturity, normal pollen continued to enlarge while double mutants remained significantly smaller, lost cytoplasm, and became increasingly misshapen (Fig. 5d, e). Similar developmental arrest was observed with TEM (Fig. 5j–m), which showed arrested microspores that failed to undergo vacuolar fission and appeared to have thin intine compared to normal mitotic pollen segregating within the same locule. Following developmental arrest, cytoplasm in double mutants appeared grainy, detached, and eventually degenerated as anther development progressed (Fig. 5l, m).

Fig. 5
figure 5

Arrested development of pirl1;pirl9 microspores. Microspores and pollen at different developmental stages were viewed in anthers from pirl1/PIRL1;pirl9 plants, which produce pollen segregating 50% for the pirl1;pirl9 genotype. Arrows mark representative arrested pollen. a–i Anther sections stained with toluidine blue and visualized by bright field microscopy. a–e Sections of pirl1/PIRL1;pirl9 anthers; scale bars 10 μm. f–i sections of WT controls from approximately equivalent anther stages, shown at the same scale. Stages were designated as described by Sanders et al. (1999). a, f Tetrads just after meiosis; b, g free microspores shortly after separation from tetrads, with mutants not yet morphologically distinguishable from normal siblings. c, h Locules from mitotic stage 10–11 anthers. Arrows in the pirl1/PIRL1;pirl9 anther indicate arrested microspores, which retain the large vacuoles characteristic of younger uninucleate microspores (Regan and Moffatt 1990), while phenotypically normal pollen segregating in the same anther are undergoing vacuolar breakdown. d Locule from a later mitotic stage anther. Some double mutants (arrows) are becoming deformed; one still resembles an arrested vacuolated microspore. e, i Locules from maturing stage 12–13 anthers; double mutants are empty, deformed, and clearly inviable (e.g. e, arrow). j–m TEM of arrested pirl1;pirl9 microspores within pirl1/PIRL1;pirl9 anthers. j A developmentally arrested double-mutant microspore (at center-left) in a mitotic stage anther, still featuring a large vacuole, the characteristic three-lobed morphology of uninucleate microspores from younger anthers, and an abnormally thin intine layer. Neighboring grains are phenotypically normal PIRL1;pirl9 pollen segregating within the locule. Arrows indicate intine. k Phenotypically normal pollen in a stage 12 anther, illustrating appearance after characteristic vacuolar fission. l An arrested pirl1;pirl9 microspore from same locule as k, illustrating the remnant large vacuole. Cytoplasm has begun degenerating and appears grainy. m A maturing post-mitotic anther with normal pollen (top) and an aborted pirl1;pirl9 mutant (bottom). Scale bars j–l 5 μm; m 10 μm

Full size image

To more precisely define the developmental arrest of pirl1;pirl9 pollen and further confirm that observed defects parallel the segregation of the double-mutant genotype, developing pirl1;pirl9 pollen segregating in tetrads in a qrt1 background were stained with DAPI and observed at different developmental stages. Representative tetrads are shown in Fig. 6. In tetrads from stage 9 anthers, nuclear staining patterns were indistinguishable between double-mutants and normal sibling microspores segregating within the same tetrad (Fig. 6f). However, as anther development progressed to the mitotic stages the double-mutant phenotype consistently became apparent. Double mutants from bicellular-stage anthers remained uninucleate, while normal pollen segregating within the same tetrads featured both a large vegetative nucleus and a clear, brightly focused generative nucleus (Fig. 6g, h). Coinciding with these differences in DAPI staining, mutant members of these bicellular-stage tetrads also appeared slightly smaller than their phenotypically normal siblings, reflecting very recent growth arrest (evident in bright field images; Fig. 6b, c). This size difference became more pronounced in early tricellular stage tetrads as normal pollen continued to increase in size and approached maturity, while mutants began losing nuclear content and becoming distorted (Fig. 6d, i). Mutants in tetrads with mature tricellular pollen shortly before anthesis appeared shrunken and lacked DAPI staining. Altogether, in pirl1/PIRL1;pirl9 tetrads segregating for the pirl1;pirl9 genotype, no mutant microspores were observed beyond the uninucleate stage (n > 198). Therefore, PIRL1 and PIRL9 are first required in developing pollen during anther stage 9–10, after exine has been established and microspores become vacuolated, but before mitosis and differentiation into functional tricellular pollen are initiated.

Fig. 6
figure 6

Developmental arrest of mutant microspores before the first pollen mitosis. Tetrads produced by qrt1;pirl1/PIRL1;pirl9 plants (segregating for 50% pirl1;pirl9 double mutants) were stained with DAPI, and viewed with phase contrast (a–e) or fluorescence (f–j). Representative tetrads from anthers at different developmental stages are shown. a, f Uninucleate microspores shortly before pollen mitosis I; double mutants are not yet distinguishable by size or nuclear content. b, c, g, h Tetrads from anthers at the bicellular stage. Double mutants (white arrows) are now distinguishable, appearing slightly smaller in tetrads viewed with phase contrast (b, c) and resembling the DAPI staining characteristic of uninucleate microspores (g, h). d, i Tetrad with two larger pollen at the early trinucleate stage. Double mutants remain arrested at microspore size, one with a single, smeared DAPI pattern (arrow), one lacking DNA. e, j A tetrad from a maturing anther, with viable pollen at the late trinucleate stage and double-mutant pollen that now lack DAPI staining

Full size image

During normal pollen development, a transient callose disk forms between the vegetative nucleus and newly formed generative nucleus just after the first pollen mitosis (Park and Twell 2001). We used aniline blue, a fluorescent callose-binding stain, to determine whether such a disk forms in pirl1;pirl9 double mutants. Results of representative tetrads observed at just this point in development are shown in Fig. 7. We consistently did not observe callose walls in double-mutant pollen, while the walls were clearly visible in mitotically competent normal pollen segregating within the same tetrads (Fig. 7e–h). As with DAPI staining, these internal differences became evident just at the point in development at which size differences between double mutant and normal pollen became apparent (Fig. 7a–d, bright field images), suggesting that developmental arrest occurs in pirl1;pirl9 microspores just prior to the first pollen mitosis.

Fig. 7
figure 7

Pre-mitotic arrest of pirl1;pirl9 microspores visualized by aniline blue staining. Tetrads produced by qrt1;pirl1/PIRL1;pirl9 plants (segregating for 50% pirl1;pirl9 double mutants) were stained with aniline blue and visualized at the start of the binucleate stage to assess formation of the transient callose disks that form between pollen nuclei after pollen mitosis I (Park and Twell 2001). a–d Representative bicellular stage tetrads just after pollen mitosis I, viewed with phase contrast. e–h The same tetrads as a–d, viewed with fluorescence. Callose disks can be seen in two phenotypically normal bicellular pollens in each tetrad. Arrows indicate positions of double-mutant microspores, which lack staining and appear very slightly smaller than normal siblings, reflecting recent developmental arrest. All pictures are at the same magnification; bar 10 μM

Full size image

Discussion

A new class of LRRs associated with pollen development

PIRL1 and PIRL9 are members of a novel, plant-specific class of LRR proteins whose leucine-rich domains are related to those of Ras-group LRRs that function in cell signaling and gene regulation in animals. Here, we have shown that PIRL1 and PIRL9 are essential for the differentiation of microspores into pollen, a process specific to plant development. Segregation of defective pollen in qrt1 tetrads demonstrated that the pirl1;pirl9 phenotype arises after meiosis, thus indicating that these genes are active and essential during the haploid phase. This was further confirmed by the lack of any obvious defects in early microspores, which were indistinguishable from developmentally normal microspores segregating within the same anthers. Mutation of both pirl1 and pirl9 consistently caused developmental arrest at the late microspore stage, just before the first pollen mitosis.

Different types of LRR proteins, in several model species, have been previously associated with pollen development. Examples include LRR-receptor kinases such as tomato LePRK1, 2, and 3, as well as related proteins in maize, petunia, and Arabidopsis (Kim et al. 2002; Muschietti et al. 1998). A secreted pollen-specific LRR, SHY, was identified in petunia and tomato, and has been implicated in signaling important for pollen tube growth, perhaps serving as a ligand for PRK receptors (Guyon et al. 2004; Tang et al. 2004). Another class of pollen-specific LRRs identified in tomato, maize, and Arabidopsis, the PEX proteins, share conserved LRR domains and are associated with pollen tube cell walls (Stratford et al. 2001). All of these examples of pollen-related LRRs are expressed late in pollen development and are primarily associated with pollen tube elongation and guidance (Kim et al. 2002; Guyon et al. 2004). PIRL1 and PIRL9 represent a new class of pollen-associated LRRs critical at a much earlier point in pollen development.

The PIRL1 and PIRL9 genes form a related gene pair within the Arabidopsis PIRL family, and our results indicate that they are functionally redundant. As single mutations, pirl1 and pirl9 alleles showed normal transmission and resulted in no obvious developmental phenotypes, except for a slight increase in pollen inviability that was only ~1–7% above the frequency observed for wild type (Fig. 3). Disruption of both genes together, however, had a highly penetrant impact on pollen development, causing consistent arrest of microspores. There may be a slight asymmetry in the redundant relationship between these two PIRLs: pirl9 single mutants produced a slightly higher percentage of inviable pollen. However, that frequency, approximately 6%, was little above background and ultimately had no discernible effect on observed transmission frequency.

A number of forward genetic approaches have been employed to identify genes involved in pollen development (Feldmann et al. 1997; Boavida et al. 2009; Procissi et al. 2001; Johnson et al. 2004; Howden et al. 1998; McCormick 2004). PIRL1 and PIRL9 would not be identified in such screens due to their genetic redundancy. Furthermore, microarrays have been employed to identify genes important in pollen development (Pina et al. 2005; Honys and Twell 2004; da Costa-Nunes and Grossniklaus 2003). While microarray data indicate expression of both of these genes in anthers and developing pollen (Zimmermann et al. 2004; Winter et al. 2007; https://www.genevestigator.ethz.ch; http://bar.utoronto.ca/efp/cgi-bin/efpWeb.cgi), an essential role in pollen development might not be predicted a priori from those data because PIRL1 and PIRL9 expression is not pollen-specific. RNA blotting, RT-PCR, and microarray data indicate that both genes are also widely expressed in the sporophyte (Forsthoefel et al. 2005; Zimmermann et al. 2004; https://www.genevestigator.ethz.ch). The results reported here therefore provide an example of how in-depth genetic and reverse-genetic investigation of specific gene families can contribute to functional genomic studies of pollen development, as well as other plant developmental processes.

Features of the pirl1;pirl9 mutant phenotype

Several features of the pirl1;pirl9 phenotype are notable. The first is the consistent early stage of developmental arrest. This early arrest accounts for the morphological severity and the high frequency of transmission failure: pre-mitotic arrest does not allow for production of sperm nuclei and development of functional pollen. Many male gametophyte mutations identified to date disrupt early pollen development, but do not result in consistent pre-mitotic arrest. For example, sidecar pollen, gem, duo, mud, and gum mutations affect pollen mitoses, post-mitotic cell fate, or organization of the male germ unit, but all allow development to progress to the bicellular stage or beyond (Chen and McCormick 1996; Park et al. 1998; Lalanne and Twell 2002; Durbarry et al. 2005; Brownfield et al. 2009; Borg et al. 2009). Other mutants, such as raring-to-go and many of the hapless mutants, affect later pollen stages, causing abnormal pollen tube germination, growth or guidance (Johnson and McCormick, 2001; Johnson et al. 2004). Some previously identified mutations that do disrupt earlier (microspore) development do not disrupt gametophytic gene activity, but rather are male-sterile mutations that result in even earlier defects in all microspores of the mutant anther, regardless of microspore genotype (e.g., Regan and Moffatt 1990; Taylor et al. 1998). These are therefore genetically distinct from the pirl mutations described here.

We observed a lack of vacuolar fission and a reduced intine layer in arrested double-mutant microspores from mitotic anthers. However, we do not necessarily propose a causal relationship between either of these features and developmental arrest: either of these could be consequences rather than primary causes of arrest. Based on both histology and TEM, it seems likely that pirl1;pirl9 developmental arrest occurs prior to the break up of the vacuole that normally accompanies mitotic stages in Arabidopsis anthers (Yamamoto et al. 2003; Regan and Moffatt 1990). Likewise, it is possible that the observed reduced intine could be an early consequence, rather than source, of developmental arrest in pirl1;pirl9 microspores, which initiate internal degeneration after development halts (a process that has been observed with some other pollen-lethal knockout mutants; Leon et al. 2007; Lee et al. 2008). It is difficult to determine the chronological relationship between ultrastructural features in the mutant microspores, given the technical limitations of TEM for viewing large numbers of segregating pollen during the transient developmental time window between the appearance of phenotypic differences and the rapid degeneration of the cytoplasm.

Another notable feature of the pirl1;pirl9 phenotype is that it is specific to development of the male gametophyte; the double-mutant allele combination was transmitted normally through the female gametophyte in reciprocal crosses. Similar male-specific gametophytic lethality has been observed for other Arabidopsis mutants, such as (for example) MAP3Kε knockout mutants (Chaiwongsar et al. 2006) and the duo pollen mutants (Durbarry et al. 2005). In fact, a large percentage of gametophyte mutants identified in forward screens were defective specifically in the male gametophyte (Boavida et al. 2009; Procissi et al. 2001; Johnson et al. 2004). Together, these results indicate, not surprisingly, that there are substantial differences in the genetic requirements for Arabidopsis male and female gametophyte development. The pirl1;pirl9 knockout phenotype is consistent with this and indicates that male-specific gene functions contribute to microspores and early stages of pollen differentiation, not just to later, more obvious pollen-specific processes such as pollen tube growth and guidance.

The PIRLs may contribute to one of several cellular processes critical for the dynamic development of microspores and differentiation of pollen. For example, they may be involved in early events in mitosis or initiation of vacuolar break up. A role in microspore growth does not seem likely since mutant microspores appear to grow normally until just prior to pollen mitosis I, and they undergo proper vacuolization during the pre-mitotic growth stage (Regan and Moffatt 1990). Given the consistent, specific timing of arrest of pirl1;pirl9 microspores, it is possible that these PIRL genes may have a regulatory function in the microspore–pollen transition because a more general housekeeping function might be expected to result in earlier microspore growth defects and/or more variable timing of arrest. A regulatory function, either in signaling or gene regulation, would be consistent with known functions for animal Ras-group LRRs (Dai et al. 2006; Sieburth et al. 1998; Li et al. 2000; Jeong et al. 2009; Lee et al. 2004; Cutler et al. 1992). However, we cannot entirely rule out that PIRLs take part in a fundamental cellular process that somehow has an acute impact on microspore development just before pollen mitosis I.

While our results indicate that PIRL1 and PIRL9 first become essential just prior to pollen mitosis I, they do not rule out additional roles for these genes later in development, either in pollen or in the sporophyte. In fact, their expression in mitotic pollen as well as sporophytic organs (Forsthoefel et al. 2005; Zimmermann et al. 2004; Winter et al. 2007; http://bar.utoronto.ca/efp/cgi-bin/efpWeb.cgi) suggests that both of these PIRLs may have additional functions later in plant development. This is not surprising: while the Arabidopsis pollen transcriptome is relatively rich in pollen-specific mRNAs (Honys and Twell 2003; 2004), it still shares at least 60% overlap with the sporophyte transcriptome, including some genes of developmental significance (Borg et al. 2009; Honys and Twell 2003). The pollen lethality of the pirl1;pirl9 double-knockout phenotype prevents straightforward genetic dissection of potential additional roles in the sporophyte. Future investigation of PIRL1 and PIRL9 using inducible knockdowns in the sporophyte is one promising strategy to investigate this.