The SIDECAR POLLEN gene encodes a microspore-specific LOB/AS2 domain protein required for the correct timing and orientation of asymmetric cell division


*For correspondence (fax +82 53 958 6880 or +44 116 252 2791; e-mail or


Cellular patterning and differentiation in plants depend on the balance of asymmetric and symmetric divisions. Patterning of the male gametophyte (pollen grains) in flowering plants requires asymmetric division of the microspore followed by a symmetric division of the germ cell to produce three highly differentiated cells: a single vegetative cell and two sperm cells. In Arabidopsis sidecar pollen (scp) mutants a proportion of microspores first divide symmetrically, and then go on to produce ‘four-celled’ pollen with an extra vegetative cell; however, details of the timing and origin of phenotypic defects in scp and the identity of the SCP gene have remained obscure. Comparative analysis of the original hypomorphic scp-1 allele and a T-DNA-induced null allele, scp-2, revealed that in the absence of SCP, microspores undergo normal nuclear positioning, but show delayed entry into mitosis, increased cell expansion and alterations in the orientation of nuclear division. We identified the SCP gene to encode a male gametophyte-specific LATERAL ORGAN BOUNDARIES DOMAIN/ASYMMETRIC LEAVES 2-like (LBD/ASL) protein that is expressed in microspore nuclei in a tightly regulated phase-specific manner. Therefore, our study demonstrates that the correct patterning of male gametophyte depends on the activity of a nuclear LBD/ASL family protein that is essential for the correct timing and orientation of asymmetric microspore division.


The life cycle of flowering plants alternates between the familiar diploid sporophyte generation and a highly reduced haploid gametophyte generation that produces the male and female gametes required for double fertilization. The unique cellular patterning and differentiation of angiosperm gametophytes also ensures the transport and delivery of sperm cells to the embryo sac (via the pollen tube), and the attraction of the pollen tube to the micropyle (via cells within the embryo sac). Thus, the correct developmental patterning of the gametophytes is a vital step in gamete production and union, and therefore is necessary for flowering plants to proliferate.

The male gametophyte consists of three cells, a vegetative cell and two sperm cells, which are produced from a haploid microspore via two rounds of post-meiotic mitosis (reviewed in McCormick, 2004; Borg et al., 2009). The first mitosis, which occurs in unicellular microspores, is intrinsically asymmetric, producing two unequal daughters, the vegetative and germ cell, with distinct cellular identities and biological functions. The second mitosis of the germ cell produces twin sperm cells, whereas the vegetative cell differentiates as a ‘companion’ cell to support the delivery of the sperm cells to the female gametophyte via the pollen tube. Asymmetric microspore division is thus a key event in the patterning of male gametophyte development.

The ontogeny of haploid microspores reflects a series of cellular events involving polarity establishment, control of division orientation and cell specification. Polarity establishment and the orientation of division are achieved in two sequential steps (Chen and McCormick, 1996; Twell et al., 1998; Figure S1). Firstly, the microspore nucleus migrates towards the future germ cell pole, concomitant with small vacuoles coalescing into a single large vacuole. Secondly, the mitotic spindle assembles near the germ cell pole with the spindle axis orientated perpendicular to the microspore wall. This regular pattern implies strict genetic mechanisms that ensure the correct temporal and spatial execution of asymmetric microspore division.

Although asymmetric division is a universal instrument to generate cellular diversity for higher-level patterning in all multicellular eukaryotes, the underlying mechanisms are seemingly different in animal and plants (Abrash and Bergmann, 2009). In animal cells, the spindle midzone plays a major role, the orientation of which is known to be controlled by several mechanisms, including ‘centrosome retention’, ‘spindle pole capture’ and ‘differential pulling forces’ (reviewed in Cowan and Hyman, 2004; Doe, 2008; Yamashita et al., 2007). In plant somatic cells, a plant-specific preprophase band of microtubules (PPB) influences spindle orientation and determines division orientation (Muller et al., 2009); however, the PPB does not form at microspore division. This implies different regulators between animal and plant cells, and even within plants between somatic and male gametophytic cells. For example, the PAR protein and Wnt/β-catenin signalling mechanisms underlying polarization and asymmetric division in animals (reviewed in Goldstein and Macara, 2007; Knoblich, 2008; Gretchen and Sokol, 2007) have not been detected in plants (Abrash and Bergmann, 2009).

To date, the number of genes or mutants with demonstrated or implicated functions in microspore division and division asymmetry control is limited. Mutants in this class include sidecar pollen (scp; Chen and McCormick, 1996), gemini pollen 1 (gem1) (Park et al., 1998; Twell et al., 2002) and gem2 (Park et al., 2004). gem1 and gem2 mutations cause a proportion of microspores to divide equally, because of incomplete or unstable nuclear migration (Park et al., 1998, 2004). In contrast to gem1 and gem2 that also affect female gametophytic development, scp phenotypes are specific for the male gametophyte (Chen and McCormick, 1996), highlighting the potential value of scp in uncovering the mechanisms determining microspore development and asymmetric division.

It has been reported that scp microspores divide symmetrically and prematurely into two undetermined microspores, and later one of the two daughters undergoes two rounds of normal mitotic division (Chen and McCormick, 1996). However, the possible scenarios for SCP function originally proposed have not been investigated further. Here, we examined the origin of scp mutant phenotypes in detail using two independent alleles: the original scp (termed scp-1) identified from a mutant pool generated by fast neutron in the Nossen (No-0) background (Chen and McCormick, 1996) and a T-DNA insertion allele, scp-2, isolated in this study. Our results show that nuclear migration occurs normally in scp microspores, but that the onset of mitosis is significantly delayed. Similar to previous findings, division orientation and cellular patterning are also disturbed in scp. We identified the SCP gene to encode LBD27/ASL29, a member of the plant-specific LBD/ASL family that is specifically and dynamically expressed in microspore nuclei. Our results clarify the ontogeny of scp defects and demonstrate the requirement for an LBD/ASL protein in microspore development and asymmetric division.


To obtain further insight into the role of SCP we carried out comparative phenotypic and genetic analyses using two independently identified scp mutant alleles. Progeny from scp mutant plants in the No-0 background (Chen and McCormick, 1996) was designated as scp-1, and a T-DNA insertion allele was identified here as scp-2 (see below for details).

Phenotypic and genetic analyses of scp mutants

Figure 1 shows mature pollen phenotypes in wild-type and scp mutant plants observed after 4′,6-diamidino-2-phenylindol (DAPI)-staining. In wild-type plants, essentially all mature pollen grains (98.9%, n = 1342) were tricellular (Figure 1a,k), whereas in scp mutants a range of aberrant pollen phenotypes were observed (Figure 1b–j,k). Heterozygous and homozygous scp-1 plants showed 17.8% (n = 8133) and 48.1% (n = 2443) aberrant pollen, respectively (Figure 1k), which was lower than the respective frequencies (55 and 80%) previously reported (Chen and McCormick, 1996).

Figure 1.

 Mature pollen phenotype in the wild type (a) and scp mutants (a–j). Mature pollen grains from freshly dehiscing anthers were stained with 4′,6-diamidino-2-phenylindol (DAPI) and viewed with fluorescence microscopy.
(a) Normal tricellular pollen.
(b–e) Four-nucleate pollen, an extra vegetative-like nucleus was either separated in different cellular compartment (b) or in the same cytoplasm (c–e).
(f) Tricellular pollen in larger size than those in the wild type.
(g–h) Binucleate pollen with (g) or without (h) an anucleate cell.
(i) Uninucleate pollen with a single fusion nucleus.
(j) Aborted pollen.
(k) The percentages of pollen showing normal tricellular (TC) and aberrant pollen phenotypes at mature pollen stage in the wild type and scp mutants. The extracellular compartment is marked with a pair of arrow heads. All images were taken at the same magnification. Scale bar: 10 μm.

The distribution of pollen grains among several aberrant phenotypic classes was scored. Aberrant pollen classes in homozygous scp-1 plants were aborted pollen (21.6%, n = 2443; Figure 1j), pollen with two nuclei or a single fusion nucleus (21.3%, n = 2443; Figure 1g–i), larger tricellular pollen (3.0%, n = 2443; Figure 1f) and pollen containing an extra vegetative-like nucleus (2.2%, n = 2443; Figure 1b–e). Interestingly, the four-nucleate pollen grains were found in four-celled (Figure 1b,c) and three-celled (Figure 1d,e) structures. The examples in Figure 1(c–e) show that the extra vegetative-like nucleus is present within the same cytoplasm of the tricellular pollen with or without an anucleate compartment. Similarly, two vegetative-like nuclei were present within the same cytoplasm with or without an anucleate compartment (Figure 1g,h).

In the previous analysis of homozygous scp-1, aborted pollen (Figure 1j, 53%) and extracelled pollen (Figure 1b, 27%) represented major abnormal classes. Mutant pollen with two nuclei or a single fusion nucleus (Figure 1h,i) and a type of four-nucleate pollen (Figure 1e) were observed infrequently, at approximately 1%, and other types of four-nucleate pollen (Figure 1c,d), larger tricellular pollen (Figure 1f) and binucleate pollen with an anucleate compartment (Figure 1g) were not detected (Chen and McCormick, 1996). Despite these differences, scp-1 phenotypes were reproducible between individuals, and were stable in backcrossed generations.

The T-DNA-induced allelic mutant scp-2 identified in this study (see below for details) showed a similar range of mutant pollen phenotypes (Figure 1b–j), but with increased penetrance when compared with scp-1. The average frequency of aberrant pollen was 36.9% (n = 5087) in heterozygous and 88.0% (n = 684) in homozygous scp-2 mutant plants (Figure 1k).

Consistent with the reduced expressivity of scp-1 in our study and the higher frequency of aberrant pollen in scp-2, we observed 80.7% (n = 291) and 24.3% (n = 373) genetic transmission of scp-1 and scp-2 through the male, respectively (Table 1). On the other hand, genetic transmissions of scp-1 and scp-2 through the female were completely normal (Table 1).

Table 1.   Genetic transmission of scp alleles. The numbers of wild-type and mutant plants among test-cross progenies are shown. The transmission efficiency (TE = mutant/wild-type progeny × 100) through the male and female are shown.
CrossesWild-type plantsscp mutant plantsTotalTE (%)
+/+ × scp-1/+16113029180.7
scp-1/+ × +/+128129257100.8
+/+ × scp-2/+3007337324.3
scp-2/+ × +/+19618838495.9

These results using two independent scp alleles, scp-1 and scp-2, confirmed that scp mutants are male gametophytic, and are distinguished from the wild type in both cell number and cellular pattern.

scp delays the timing of microspore division and alters division orientation

In order to identify the primary defects in scp mutants, we examined microspores and developing pollen dissected from the DAPI-stained anthers of serially staged bud samples from wild-type and heterozygous scp plants. In the wild type (Figure 2a–f), early microspores with a central nucleus (Figure 2a) become polarized after nuclear migration (Figure 2b). Asymmetric division gives rise to early-bicellular pollen, with the germ cell attached to the pollen radial wall (Figure 2c). Vacuole morphogenesis coincides with microspore polarization and asymmetric division (Figure 2a–c). A single large vacuole is obvious in polarized microspores (Figures 2b and S1b), and is reduced in size in early bicellular pollen after division (Figures 2c and S1c) through a vacuolar invagination process (Yamamoto et al., 2003). At later bicellular stages, the germ cell undergoes morphological transformation, involving progressive elongation of the germ cell nucleus (Figure 2d,e), followed by germ cell division, to produce tricellular pollen (Figure 2f).

Figure 2.

 Pollen development in wild-type and scp plants.
(a–f) 4′,6-diamidino-2-phenylindol (DAPI)-stained developing spore/pollen from the wild type. Image of early microspore (a), polarized microspore (b), early bicellular (c), mid-bicellular (d), late bicellular (e) and tricellular (f) stages.
(g–l) Aberrant pollen phenotypes that first appear in scp mutants when the wild types are at the mid-bicellular stage (d).
(g) Uninucleate pollen remaining at extended polarized microspore stage.
(h, i) Mutant early bicellular pollen grains: larger in size and vacuolated, compared with the wild type.
(j–l) Mutant pollen with two equally DAPI-stained nuclei, which are either clearly separated in different cytoplasms (j) or remain in the same cytoplasm (k, l).
(m) The percentages of developing spore/pollen at typical stages contained in single anthers from serially staged buds in wild-type and heterozygous scp-1 and scp-2 plants. The nascent cell walls produced after microspore division are marked with a pair of arrowheads. All images were taken at the same magnification. Scale bar: 10 μm. v, vacuole.

In scp mutants, microspores at early and polarized stages did not differ from those in the wild type (Figure 2a,b,m), indicating that nuclear migration and positioning are not affected by scp mutations. However, by analysing spores from successive bud stages in heterozygous scp plants, it was clear that polarized microspores extend across four successive bud stages, rather than being restricted to two bud stages, as in the wild type (Figure 2m). This clearly indicates that scp microspores are delayed in entering mitosis. For instance, in heterozygous scp-2 plants, when more than half of the pollen population (56.5%, n = 177) reached mid-bicellular stage, 16.9% remained as polarized microspores with a single large vacuole (Figure 2g,m). These delayed microspores increased in size similar to wild-type pollen at mid/late bicellular stages (Figure S2). In the same population, a further 26.6% of microspores completed mitosis and displayed the range of phenotypes shown in Figure 2(h–l). Abnormal pollen grains shown in Figure 2(h,i) are similar to wild-type pollen at the early bicellular stage, except they are larger in size and often highly vacuolated (compared with Figure 2c). Thus a proportion of scp microspores undergo relatively normal but delayed division relative to the wild type. It is evident that the majority of the delayed microspores that completed mitosis result in pollen with two similar DAPI-stained nuclei (Figure 2j–l). Although the mitosis occurs at the pollen radial wall, as in the wild type, the positioning of the two nuclei appears to be significantly altered. Whereas in normal early bicellular pollen only the germ cell nucleus is positioned against the pollen wall, in mutants (Figure 2i–l) both daughter nuclei remain close to the wall, implying defects in the orientation of the division. Two diffusely DAPI-stained nuclei were found either separated in two cells (Figure 2j), or in the same cell with or without an anucleate compartment (Figure 2k,l), showing that cytokinesis does not always separate the cytoplasm precisely between the two nuclei in scp mutants. In addition, these mutant pollen grains remain highly vacuolated, suggesting delayed differentiation or incomplete specification that may contribute to ‘secondary’ asymmetric division. A schematic for the ontogeny of pollen development in scp mutants is presented in Figure S3. Our observations further suggest that the asymmetric maintenance of microspore cell fate in scp does not depend on complete cytokinesis separating two daughter nuclei.

Division orientation defects in scp microspores were found to be associated with abnormally orientated anaphase mitotic figures in scp. We restricted our analysis to microspores observed in axial (Figure 3a–e) or in radial view (Figure 3f–j) to enable the unambiguous scoring of the orientation of anaphase pairs. Anaphases in wild-type microspores (n = 45) were divided between the two types shown in Figure 3b (axial view) and Figure 3g (radial view). In contrast, seven anaphase pairs out of 56 observed in scp-1 and nine out of 28 in scp-2 showed different arrangements from those in the wild type. Multiple examples of severely altered chromosome alignment in scp are shown (Figure 3d,i) that are likely to be responsible for subsequent mutant pollen phenotypes (Figure 3e,j) respectively. Our observations show that in addition to the delay in the onset of mitotic entry, the control of division orientation is commonly disturbed by scp mutations.

Figure 3.

 Chromosome alignments at anaphase during microspore division. Visual distribution pattern of apertures are used as an orientation reference of the individual spore viewed. DAPI-stained developing spores were viewed in pollen axial view (a–e) or radial view (f–j).
(a–c, f–h) Wild-type spores representing three consecutive stages: before (a, f), during (b, g) and after (c, h) asymmetric mitosis.
(d, e, i, j) Unambiguously aberrant types during (d, i) and after (e, j) asymmetric mitosis in scp mutants. Black dots mark the germ cell pole, black arrowheads mark pollen apertures and an arrow marks the pollen axis. All images were taken at the same magnification. Scale bar: 10 μm. v, vacuole.

SCP encodes a member of the LBD/ASL family proteins

The scp locus was reported to reside near markers nga6 and BGL1 on chromosome 3 (Chen and McCormick, 1996). For map-based cloning of SCP we analysed F2 recombinants generated by crossing scp-1 (No-0) and the wild type (Columbia-0, Col-0), and narrowed down the scp locus to a approximately 130-kb region containing 30 predicted genes (Figure 4a). Sequencing of several candidate genes in this region using scp-1 heterozygous and homozygous plants identified a 20-bp deletion in the open reading frame of At3g47870, creating a premature stop codon at amino acid position 264 (Figures 4b and S4). Sequence analyses further revealed SCP polymorphisms between Col-0 and No-0, which result in amino acid substitutions at positions 12 and 155 (Figure S4). It was also found that SCP in No-0 encodes a 363-aa protein, which is 35 residues longer than Col-0 SCP, as a consequence of a single nucleotide polymorphism at the stop codon of Col-0 SCP (Figure S4).

Figure 4.

 Identification of the SCP gene.
(a) A schematic of the positional cloning of the SCP gene. The positions of molecular markers used, the physical distances between the markers and the numbers of recombinants per 644 plants analyzed are shown.
(b) RT-PCR analysis using bud RNAs from wild-type, No-0 (No) and Col-0 (Col), and homozygous scp mutants. The positions of two mutations are shown on the SCP gene structure, with a gene model containing two exons (bar) and a single intron (line). The set of primers used is shown in arrows. Histone H3 (At5g10980) was used as a control for equal loading.
(c) Complementation analysis in scp-2 mutant. The percentages of aberrant pollen grains at the mature pollen stage were counted in nine non-transformed heterozygous scp-2 and 34 transgenic plants from heterozygous scp-2 dipped with a complementation vector.

We searched for allelic mutants and found similar scp mutant pollen phenotypes in a T-DNA insertion line, SALK_019070, which we termed scp-2 (Figure 4b). Analyses of flanking sequences revealed that the T-DNA right border was inserted at position +567 bp in the SCP open reading frame, producing a premature stop codon at amino acid position 190 (Figure S4). In scp-1 homozygotes, mutant transcripts harbouring the 20-bp deletion were detected by sequencing RT-PCR products, whereas in scp-2 homozygotes no expression was detected (Figure 4b), in accord with the increased penetrance of scp-2.

To further confirm the identity of At3g47870 as SCP, we introduced a 3.8-kb fragment containing the wild-type At3g47870 locus into heterozygous scp-2 plants. scp-2 plants harbouring this genomic fragment showed significantly reduced frequencies of abnormal pollen, compared with non-transgenic scp-2 heterozygotes (Figure 4c). Thus, the loss of At3g47870 is responsible for the scp mutant pollen phenotypes.

BLAST searches revealed that SCP corresponds to LBD27/ASL29, a member of the LBD/ASL family characterized by the LOB/AS2 domain (Iwakawa et al., 2002; Shuai et al., 2002). The LOB/AS2 domain is approximately 100-aa long, located in the N-terminal region of all LBD/ASL family members, and consists of a cysteine repeat, an invariable glycine residue and leucine-zipper-like sequences (Iwakawa et al., 2002; Shuai et al., 2002). These typical signatures are shown in Figure S4.

Apart from signature sequences in the N-terminal LOB/AS2 domain, LBD/ASL family proteins generally do not share a high degree of homology, and are highly variable beyond the LOB/AS2 domain, even within the Arabidopsis LBD/ASL family (Matsumura et al., 2009). The highest homology found in BLASTp searches in Arabidopsis protein databases using the entire 328-aa SCP protein sequence is 42% identity and 58% similarity to LBD22/ASL30. When the N-terminal 142-aa region of SCP including the LOB/AS2 domain was used, LBD22/ASL30 was again detected as the closest, with slightly higher levels of 50% identity and 68% similarity. BLASTp searches in an extended protein database including all non-redundant Genbank coding sequence detected several proteins from different plant species showing some significant homologies (Figure S5; Table S1). Their overall homology with the 328-aa SCP protein sequence are 38–44% identity and 52–58% similarity, which is similar to that of LBD22/ASL30, but their similarity is detected in extended regions beyond the LOB domain. Moreover, when the N-terminal 142-aa sequence was used, their homologies were much higher, 63–68% identity and 80–85% similarity, showing that these proteins share higher homology with SCP than with any other proteins, including the entire LBD/ASL family in Arabidopsis. This suggests conservation of SCP protein structure and function in distant plant species.

Expression analysis of SCP

Among the six different tissue samples we examined – seedling, root, leaf, stem, bud and open flower –SCP transcripts were only detected in young buds and at trace levels in open flowers by RT-PCR (Figure 5a). Similarly, RNA from mixed microspores and developing pollen (Figure 5a, Is) showed a much higher expression than mature pollen (Figure 5a, Mp). RT-PCR analysis of RNA samples enriched for unicellular, bicellular, tricellular and mature pollen stages showed that SCP transcript levels persist from microspore to tricellular pollen stages, but are clearly reduced in mature pollen.

Figure 5.

 Expression analysis of the SCP gene.
(a) RT-PCR analysis in the wild type. RNA samples were extracted using 1-week-old seedlings (Sd), roots (Rt), leaves (Lf), stems (St), bud clusters (Bd), open flowers (Fl), isolated spore mixture (Is) and mature pollen (Mp), either from Col-0 or Ler ecotypes. Isolated spore mixtures were further separated into unicellular microspore (Uc)-, bicellular pollen (Bc)-, or tricellular pollen (Tc)-enriched fractions from Landsberg erecta (Ler). A negative control for RT-PCR reaction was performed without reverse transcriptase (−). The Histone H3 gene (At5g10980) was used as a control for the input RNAs used.
(b–j) GFP::GUS expression driven by the control of SCP promoter sequences.
(b–e) Whole inflorescence and 1-week-old seedlings from ProSCP1-GFP::GUS plants containing a approximately 1.7 kb upstream sequence (b, d) or ProSCP2-GFP::GUS plants with approximately 1.0 kb (c, e) were assayed for GUS expression.
(f–j) GFP visualization in a hemizygous ProSCP1-GFP::GUS plant. Whole anther (f) and microspores with a central nucleus (g) and polarized nucleus (h), bicellular pollen (i) and tricellular pollen (j). DAPI counter-stainings of the same spores were shown in the lower panel (g–j). Scale bars: 100 μm (f) and 10 μm (g–j).

For more detailed analysis of SCP expression we generated two SCP promoter GFP::GUS fusion constructs, ProSCP1-GFP::GUS and ProSCP2-GFP::GUS, using fragments of 1701 and 1005 bp, respectively. Histochemical analysis of whole inflorescence samples from approximately 50 different T1 plants for each construct revealed that 43 of 49 ProSCP1-GFP::GUS plants (Figure 5b) and 40 of 50 ProSCP2-GFP::GUS plants showed GUS expression (Figure 5c). GUS activity was restricted to anthers, with staining first detected in young buds, and increasing in older buds and open flowers (Figure 5b,c). SCP promoter activity appears to be specific to the male gametophyte as no GUS expression was found in seedlings from approximately 20 independent lines for both constructs (Figure 5d,e), and GFP signals were restricted to pollen inside the anther locules (Figure 5f). GFP signals were detectable as early as the unicellular microspore stage (Figure 5g, arrow), and increased throughout pollen development (Figure 5h–j).

Subcellular localization of SCP

To determine the subcellular localization and expression profile of the SCP protein, we transformed plants with a construct, ProSCP1-SCP::mRFP1, containing the promoter and the coding region of SCP fused to the red fluorescent protein variant mRFP1. In contrast to the expression of GUS::GFP in ProSCP1-GFP::GUS lines, these lines did not show mRFP1 signals in the mature pollen. However, when we dissected anthers at earlier stages, 36 of 40 ProSCP1-SCP::mRFP1 T1 lines showed mRFP1 fluorescence specifically at the microspore stage (Figure 6). SCP::mRFP1 signals were not detectable at tetrad (data not shown) and early released microspore stages (Figure 6a), but were present in the central nucleus of microspores before and after the obvious thickening of the exine (Figure 6b,c). At polarized microspore stage the nuclear signals became much weaker (Figure 6d), and were no longer detectable by the early bicellular stage (Figure 6e).

Figure 6.

 Subcellular localization of SCP protein during pollen development. Using approximately 1.7 kb of the native SCP promoter, SCP::mRFP1 (a–e) and H2B::mRFP1 (f–l) fusion protein expressions were examined in the wild type. Freshly released developing spores and pollen from serially staged buds were viewed under epifluorescence with an RFP filter (upper panels) or bright field (lower panels).
(a–c, f–h) Microspores with a central nucleus at three consecutive stages.
(d, i) Microspores at polarized stage.
(e, j) Early bicellular pollen immediately after asymmetric microspore division. A pair of arrowheads depict the nascent germ cell wall.
(k) Mid-bicellular pollen.
(l) Tricellular pollen. All images were taken at the same magnification. Scale bar: 10 μm. (Note that signals on the pollen wall are autofluorecence produced at the UV wavelength used.)

To test the function of the SCP::mRFP1 fusion protein we generated F1 plants heterozygous for scp-2 and hemizygous for ProSCP1-SCP::mRFP1 at a single locus. F1 plants with ProSCP1-SCP::mRFP1 showed reduced frequencies of mutant pollen (25–28%) compared with those without (35–41%), showing that the SCP::mRFP1 fusion protein can rescue scp-2 defects. In ProSCP1-H2B::mRFP1 lines in which Histone2B::mRFP1 is driven by the same SCP promoter, mRFP1 signals were first detected in microspores at the same stage as in ProSCP1-SCP::mRFP1 lines (Figure 6g), but expression persisted through all subsequent stages (Figure 6h–l). As SCP transcript levels are maintained from microspore to tricellular pollen stages (Figure 5a), our results thus suggest that the transient expression of the SCP::mRFP1 protein during a restricted phase of microspore development is controlled by post-transcriptional and/or post-translational mechanisms.


In this study we have detailed the ontogeny of scp mutant pollen phenotypes, and have shown that a loss of SCP function leads to delayed microspore division and defects in division orientation. We identified the SCP gene to encode a male gametophyte-specific member of the LBD/ASL family that is expressed in a tightly regulated phase-specific manner in microspores. Collectively, our data highlight the role of SCP as an essential regulator of microspore development, asymmetric division and patterning of the male gametophyte.

SCP function in division asymmetry is independent of nuclear positioning

In eukaryotes, the determination of cell fate and morphogenesis can be achieved by ‘regulated’ division asymmetry (Rhyu and Knoblich, 1995). Thus dividing cells must operate mechanisms for positioning and/or orientating their division plane in a manner that is coordinated and responsive to cell fate-specifying factors (Abrash and Bergmann, 2009). Microspore division is intrinsically asymmetric and the hypothetical cell fate specifying factors are proposed to be distributed in a radial gradient (Twell et al., 1998). With such intrinsic cues being differentially distributed, division asymmetry in microspores is regulated in two steps, nuclear migration to the germ cell pole and control of spindle orientation, which together dictate the eccentric site of division and the division plane, ensuring asymmetric division (Twell et al., 1998). The first step involves microtubule-dependent nuclear migration (Brown and Lemmon, 1991; Oh et al., 2010a,b), a process that requires GEM1, a plant member of the MAP215 family of microtubule-associated proteins, also known as MICROTUBULE ORGANIZATION 1 (MOR1) (Park et al., 1998; Whittington et al., 2001; Twell et al., 2002). The tobacco orthologue of MOR1/GEM1 is required during both steps, and knockdown phenotypes can uncouple nuclear migration and division orientation (Oh et al., 2010a,b). Our results show that nuclear migration occurs normally in scp microspores, but that abnormal pollen with two equal nuclei at the germ cell pole result from altered division orientation, indicating a role for SCP in the second step of asymmetric division (Figure S3). In addition, the fact that a fraction of scp binucleate pollen are able to undergo the secondary asymmetric division, which later leads to four-nucleate (or four-celled) pollen, indicates that deregulated division asymmetry in a polarized position can prolong microspore cell fate. Other genes that are likely to play roles in microspore division asymmetry in Arabidopsis include γ-tubulin genes, TUBG1 and TUBG2, (Pastuglia et al., 2006) and γ-tubulin targeting factor NEDD1 (Zeng et al., 2009). Microspores deficient in these proteins divide into binucleate pollen with equal nuclei, which is attributed to disorganized (Pastuglia et al., 2006) or elongated (Zeng et al., 2009) spindles. In these studies nuclear migration, the onset of mitosis and division orientation in mutant microspores were not clearly delineated, making it difficult to draw comparisons with the analysis of scp described here. However, it is evident that their functions are not likely to overlap with SCP in that only the scp-induced defects allow two subsequent rounds of mitoses, the secondary asymmetric division and the tertiary symmetric division, resulting in four-nucleate (or four-celled) pollen.

SCP, a member of the LBD/ASL family of putative transcription factors

Map-based cloning revealed that SCP encodes the LBD27/ASL29 protein, a member of the plant-specific LBD/ASL family. LBD/ASL family proteins are widely conserved in angiosperms, with 43 LBD/ASL genes present in Arabidopsis (Iwakawa et al., 2002; Shuai et al., 2002), and 35 in each of two rice subspecies (Yang et al., 2005). Compared with the large number of Arabidopsis LBD/ASL family genes, to date only a small fraction have implicated functions. ASYMMETRIC LEAVES 2 (AS2) (Iwakawa et al., 2002) regulates the establishment of leaf polarity through an interaction with AS1 (Xu et al., 2003), and acts to repress the expression of KNOX transcription factors in lateral organs (Lin et al., 2003). The LOB gene identified from an enhancer trap line showing GUS reporter activity in a variety of lateral organ boundaries (Shuai et al., 2002) was suggested to be a transcription factor post-translationally regulated by interacting with basic helix-loop-helix (bHLH) family transcription factors in yeast (Husbands et al., 2007). In addition, LBD36/ASL1 was reported to control proximal–distal patterning in petals (Chalfun-Junior et al., 2005), and LBD30/ASL19, also known as JAGGED LATERAL ORGANS, was found to be essential for embryo development (Borghi et al., 2007). More recently, Lee et al. (2009) showed that LBD18/ASL20 regulates lateral root formation with a combination of LBD16/ASL18 downstream of ARF7 and ARF9, and Rubin et al. (2009) showed that three genes, LBD37ASL39, LBD38/ASL40 and LBD39/ASL41, repress anthocyanin synthesis and affect additional nitrogen responses in Arabidopsis. These results suggest that LBD/ASL family proteins play a variety of roles as developmental and metabolic regulators in the sporophyte.

In the gametophyte, one study in maize has identified a key role for the LOB domain protein indeterminate gametophyte 1 in female gametophyte development (Evans, 2007). ig1 mutations result in a prolonged phase of free nuclear divisions that lead to a range of phenotypes, such as extra egg cells, polar nuclei and synergids. ig1 does not affect the male gametophyte, and normally acts to repress proliferation. The identification of SCP extends the role of LBD/ASL family proteins to male gametophyte development, suggesting that specific LOB domain proteins have been recruited to regulate cell proliferation required for subsequent patterning of the male and female gametophytes.

Possible targets of SCP as a putative transcription factor

As LBD/ASL family members have been suggested to be transcription factors (Husbands et al., 2007; Rubin et al., 2009), our results are also in favour of SCP being a putative transcription factor. First, the SCP::mRFP1 fusion protein is localized to the nucleus and expression is restricted to a short period of microspore development before nuclear migration (Figure 6). Given that scp mutant microspores undergo normal nuclear migration but display defects at later stages, SCP expression at earlier microspore stages may not control the later process directly, but rather functions indirectly as a transcription factor to provide the necessary downstream proteins.

If SCP is a transcription factor it may target genes in which functions are directly involved in controlling the correct onset and division orientation. Such functions may include prospindle assembly/orientation in polarized microspores. For instance, a recent study identified the plant TPX2 to regulate prospindle assembly before nuclear envelope breakdown (Vos et al., 2008). In the report, living Tradescantia virginiana stamen hair cells injected with TPX2-specific antibodies blocked nuclear envelope breakdown with cell cycle arrest or delay at late prophase or prometaphase, and prospindle formation. Whether or not the plant TPX is also involved in prospindle formation during asymmetric microspore division is not known. However, it will not be surprising if there are specific molecular players to regulate and ensure asymmetric spindles to be correctly assembled and orientated according to intrinsic cues differentially distributed before the mitotic onset and nuclear envelope breakdown in polarized microspores. In acentrosomal higher plants, the nuclear envelope is proposed to act as a microtubule-organizing centre for spindle organization (Meier and Brkljacic, 2009). It has also been reported that a prospindle forms from aster-like microtubules nucleated at the nuclear envelope (Stoppin et al., 1994, 1996; Canaday et al., 2000), and that future spindle poles are marked by two opposing polar caps with microtubule plus ends formed at the nuclear periphery (Chan et al., 2005). In somatic plant cells, the PPB encircling the nucleus prior to the mitotic onset cooperatively influences prospindle formation, whereas in polarized microspores where the PPB is absent the nucleus envelope may play roles in mitotic onset and division orientation. This may demand that regulatory networks controlling nuclear envelope breakdown, the onset of mitosis and division orientation should be more closely linked to each other in polarized microspores than in other types of plant cells. On the other hand, it is also possible that the action of SCP on division orientation may act through genes in which functions are not directly related to asymmetric division. For example, the prolonged large vacuoles observed in scp microspores may interfere with the coordination of cytoplasmic and cortical domains, and polarized microtubule arrays (Oh et al., 2010a,b). In this view, it can be postulated that SCP may target genes that are not directly required for division asymmetry, but may interfere with juxtaposed processes like vacuole fission that may indirectly affect later processes such as the onset of mitotic entry and division orientation. The identification of the SCP gene now provides a significant opportunity to uncover a largely unknown molecular network governing asymmetric microspore division that is required to establish the male germline in flowering plants.

Experimental Procedures

Plant materials and growth conditions

Arabidopsis plants were grown at 22°C with an 18-h photoperiod in a controlled-environment growth room. Mutant alleles for the SCP gene used in this study were scp-1 in No-0 and scp-2 (SALK_019070) in Col-0 backgrounds. Arabidopsis plants were transformed with Agrobacterium tumefacienes (GV3101) following a floral-dipping protocol (Clough and Bent, 1998). Transgenic plants were selected on plant media supplemented with 110 mg L−1 gentamycin, 50 mg L−1 kanamycin or 20 mg L−1 hygromycin.

Genetic analysis and positional cloning of SCP

Genetic transmission analyses were performed as previously described (Park et al., 1998). For positional cloning of the SCP locus, mapping populations were generated using scp-1 mutants. Homozygous scp-1 plants in the No-0 background were crossed as a female with pollen from wild-type Col-0 to obtain F1 plants. Self-progenies from the F1 plants were used for fine mapping as F2 populations. Genomic DNAs were extracted from leaf tissues of individual F2 plant, homogenized using TissueLyser (Qiagen, Molecular markers to detect recombinants include existing ones from TAIR (, and new ones developed during the chromosome walk based on the sequence polymorphism found between No-0 and Col-0 (Table S2). T-DNA insertion alleles available for the 30 predicted genes that reside in the approximately 130-kb region were obtained from TAIR ( to examine pollen morphology. The T-DNA insertion sites were verified by sequencing DNA samples amplified with specific primer sets for the T-DNA borders and the gene of interest. In parallel, several putative candidates among approximately 30 predicted genes were sequenced from heterozygous scp-1 mutants to find detrimental sequence changes. Nucleotide changes were confirmed by sequencing DNA samples from No-0 wild-type and heterozygous and homozygous scp-1 plants.

Vector construction

For ProSCP-GFP::GUS fusion vector construction, fragments containing the 1701- and 1005-bp upstream sequences from the start codon of the SCP gene were amplified with a high-fidelity KOD DNA polymerase (Novagen,, and were introduced into the pKGWFS7 plant destination vector (Karimi et al., 2005) via the pDONR201 donor vector using Gateway Technology (Invitrogen, For complementation vector construction, a genomic fragment of 3858-bp from the bacterial artificial chromosome (BAC) clone T23J7, containing a 2231-bp 5′ upstream region, a 987-bp coding region and a 640-bp 3′ untranslated region (UTR), was introduced into the pPZP221 vector using the BamHI and HindIII restriction enzyme digestions. For subcellular protein localization analysis, ProSCP1-SCP::mRFP1 vector was constructed using fragments of a 1701-bp promoter, a full-length SCP coding region and mRFP1 sequences that were PCR-amplified with primers listed in Table S3. As a control, ProSCP1-H2B::mRFP1 vector was constructed by replacing the coding region of SCP in the ProSCP1-SCP:mRFP1 with that of Histone 2B. Primers used for various vector constructions can be found in Table S3.

RT-PCR analysis

Mature pollen and developing spores enriched at different stages were isolated as described by Honys and Twell (2004). Total RNAs were extracted using RNeasy Plant Mini kit (Qiagen). Samples of 0.9 μg of total RNA were reverse-transcribed using an Improm II cDNA synthesis kit (Promega,, following the manufacturer’s instructions. A 1-μl volume of single-strand cDNA samples was used for PCR amplification under 95°C for 30 sec and 30 cycles of 94°C for 15 sec, 55°C for 30 sec and 72°C for 30 sec. A Histone H3 gene (At5g10980) was used to control for RNA input and integrity.


Mature pollen samples were stained in DAPI solution, as described by Park et al. (1998). For developing spores at different stages, whole inflorescences were prefixed in a 3:1 mixture of ethanol:acetic acid prior to the DAPI staining. For fluorescent protein visualization, mature pollen and developing spores were mounted either in 0.3 m Mannitol or DAPI solution. Microscopy imaging was carried out using a Nikon Eclipse 80i microscope (Nikon, under either bright-field or epifluorescence with 40×, 60× and 100× objectives. Images were captured using a ProgRes MFcool camera (Jenoptik, equipped with a ProgRes CapturePro 2.5 software (Jenoptik, and processed using Photoshop v8 (Adobe, Histochemical staining for GUS activity was carried out as described by Honys et al. (2006) in a solution containing 1–2 mm X-gluc (5-bromo-4-chloro-indoyl β-d-glucuronide) and 0.5–1 mm K3Fe(CN)6 at 37°C for 1–2 days. Stained tissues were subsequently cleared in 70% ethanol and viewed under a stereomicroscope. Images of seedlings and whole inflorescences after GUS staining were captured using a ProgRes C3 camera (Jenoptik,


The authors thank Prof. Sheila McCormick (University of California, Berkeley, USA) for the kind gift of the original scp-1 seeds and Yeun-Ju Lee and Young-Sun Lee for technical assistance. This work was supported by the Korea Research Foundation Grant KRF-C00849 to SKP and BBSRC funding to DT. SAO was supported by the Korea Research Foundation Grant KRF-F00008.