Download The Diversity of Self-Incompatibility Systems in Flowering Plants

Review Article
The Diversity of Self-Incompatibility Systems in Flowering Plants
S. J. Hiscock and S. M. McInnis
School of Biological Sciences, University of Bristol, Bristol, UK
Received: September 2, 2002; Accepted: January 15, 2003
Abstract: Flowering plants are the most successful group of
land plants and dominate the earths vegetation with around
300 000 species. This success is, in part, the consequence of a
set of unique reproductive innovations that evolved with the
flower. Most notable of these innovations were the closed carpel and double fertilization. Closed carpels permitted the evolution of effective mechanisms for pollen selection and discrimination, while double fertilization leading to endosperm formation allowed for more efficient utilization of resources because
reserves are only allocated to the seed after fertilization. This
review will focus on the most important and best understood
mechanism of pollen discrimination, self-incompatibility (SI), a
genetically determined pollen recognition system that prevents
self-fertilization and fertilization by other individuals with the
same incompatibility phenotype. In recent years much progress
has been made towards elucidating the molecular mechanisms
of SI operating in three distinct SI systems found in the Brassicaceae, Solanaceae and Papaveraceae, respectively. More recent molecular data obtained from the Poaceae, Convolvulaceae and Asteraceae, however, suggest that other molecular
mechanisms of SI exist. A survey of classical genetic studies of
SI predicts yet further potential molecular mechanisms of SI.
We discuss the evolutionary implications of this apparent diversity in molecular pathways leading to SI and stress the need for
more molecular studies of different SI systems.
Key words: Self-incompatibility, S genes, S-RNase, S receptor
kinase, S-protein.
Introduction
Most flowering plants have co-sexual (bi-sexual) flowers containing male reproductive structures (the stamens, consisting
of anther and filament) and female reproductive structures
(carpels/pistils, consisting of stigma, style and ovary). Co-sexual flowers greatly increase the efficiency of insect pollination
because deposition of cross pollen on stigmas and removal of
self-pollen from anthers are accomplished during one insect
visit. Co-sexuality, however, also increases the risk of self-pol-
Plant biol. 5 (2003) 23 ± 32
Georg Thieme Verlag Stuttgart ´ New York
ISSN 1435-8603
lination leading to self-fertilization and its detrimental consequence, inbreeding depression. Nevertheless, many co-sexual
species are able to prevent self-fertilization because they are
ªself-incompatibleº (SI). SI is a genetically determined pre-zygotic barrier to fertilization by self- or self-related pollen that
eliminates any risk of inbreeding and therefore optimizes the
potential for out-breeding afforded by insect pollination. It
has been suggested that SI was a feature of the earliest flowering plants that may have been a significant factor in their extraordinary evolutionary success (Whitehouse, 1950). Indeed
today, it is estimated that SI is present in around 60 % of flowering plants (see Hiscock and Kües, 1999).
Darwin (1877) was aware of the phenomenon of SI when he
described ªillegitimateº and ªlegitimateº matings between
the two different flower forms (morphs) in Primula. Primula
has two flower forms, ªpinº flowers have long styles such that
the stigma is placed above the anthers, and ªthrumº flowers
with short styles, and anthers placed above the stigma. ªLegitimateº matings (resulting in seed set) are only possible between pin and thrum flowers, never between pin and pin or
between thrum and thrum ± these being ªillegitimateº matings. This type of SI, where morphological features of the flower differ between incompatibility types is known as heteromorphic SI, in contrast to so-called homomorphic SI where
there are no such differences in floral morphology between
incompatibility types. The Primula system with two flower
morphs is known as distyly and is the most widespread form
of heteromorphic SI (reviewed by Barrett and Cruzan, 1994).
Less common is tristyly where three flower morphs co-exist
as in Lythrum (Lythraceae), again, first described by Darwin
(1877). Heteromorphic SI has a scattered distribution among
flowering plants, being recorded in 25 families and 155 genera
(Ganders, 1979) but will not be dealt with in this review (but
see Barrett and Cruzan, 1994, for review). Instead we will focus
on homomorphic SI systems as more is known about their molecular genetics and cell biology.
The Genetics of SI
SI (whether homomorphic or heteromorphic) is usually controlled by a single S (self-incompatibility) locus. In homomorphic SI systems S is multiallelic, whereas in heteromorphic
systems S is always diallelic (reviewed by de Nettancourt,
1977). In most cases, if pollen and pistil share a common S allele, the pollen is rendered incompatible; exceptions only be-
23
24
Plant biol. 5 (2003)
ing possible where dominance interactions occur between S
alleles (see below). Classical genetic studies identified two distinct genetic forms of SI: gametophytic (GSI) and sporophytic
(SSI). In GSI the incompatibility phenotype of the pollen is determined by its own haploid genome, whereas in SSI the incompatibility phenotype of the pollen is determined by the
diploid genome of the plant (sporophyte) that produced it. This
important genetic difference has significant consequences for
the behaviour of S alleles in the two SI systems. In GSI, S alleles
are expressed co-dominantly in the pistil whereas in SSI systems, complex dominance interactions are possible between S
alleles. In SSI, dominance interactions between S alleles can act
independently in pollen and pistil and can allow successful
matings to occur between individuals that share one recessive
S allele. This means that it is theoretically possible to have S
homozygotes within populations of SSI species, a phenomenon
that is theoretically impossible under GSI. Distyly and tristyly
are both under sporophytic control with the two S alleles, S
and s showing complete dominance (for a review of SI genetics
see de Nettancourt, 1977). Within a given angiosperm family,
homomorphic SI is always of one type (Hiscock and Kües,
1999). Thus, GSI is found typically in the Solanaceae, Rosaceae,
Scrophulariaceae and Papaveraceae, whereas SSI is found in
the Brassicaceae, Asteraceae and Convolvulaceae. Over the last
20 years, research into the molecular basis of GSI and SSI has
focused principally on three ªmodelº SI systems and within
these systems generally just one or a few species, often chosen
because of their economic importance. The three model systems are a) SSI in the Brassicaceae (Brassica oleracea, B. rapa
[syn. campestris] and more recently Arabidopsis lyrata); b) SRNase-mediated GSI (principally in species from the Solanaceae, such as Nicotiana, Petunia, Solanum and Lycopersicon,
but also in species from the Rosaceae, i.e., Pyrus, Prunus and
Malus and Scrophulariaceae, i.e., Antirrhinum); c) GSI in Papaver (Papaveraceae) which involves an S-protein not related
to S-RNases (for comprehensive reviews see Hiscock and Kües,
1999; McCubbin and Kao, 2000). Evidence from research into
these model SI systems indicates that different genes must
regulate the expression of SI in pollen and pistil. Nevertheless,
so far, only in Brassica have both the pollen and pistil S genes
been characterized and shown to encode different proteins
(see Schopfer and Nasrallah, 2000). A minimum of two genes
is thus predicted to reside at the S-locus, so variants of the Slocus are now better referred to as S-haplotypes rather than
S-alleles ± the term ªalleleº now only being used to describe
variants of the same S gene.
Despite the understandable emphasis of molecular research
on the three model SI systems there have been some molecular
studies of SI in species from other families. These studies predict further distinct systems of SI. Following a review of SI in
the three model systems we will discuss these new molecular
findings and highlight genetic evidence from the literature
that together predict a greater diversity for molecular mechanisms of SI than was hitherto thought.
Molecular Studies of SI in Model Systems
a) Sporophytic SI in the Brassicaceae
The SSI system in the Brassicaceae is the only SI system for
which both pollen and stigma S-specificity determinants have
been identified. Early molecular studies on Brassica oleracea
S. J. Hiscock and S. M. McInnis
and B. rapa (syn. campestris) identified two stigma-expressed
genes at the S locus, the S Locus Glycoprotein (SLG) and the S
Receptor (serine-threonine) Kinase (SRK) (Nasrallah et al.,
1994). SLG encodes a secreted glycoprotein that accumulates
within the cell wall of stigmatic papillar cells, while SRK encodes a membrane-spanning receptor kinase that localizes to
the plasma membrane of stigmatic papillae. SRK and SLG both
typically have conserved regions and regions of extreme amino acid variability (hypervariable regions) that may be important for allelic specificity. Interestingly, SLG and the receptor
domain of SRK have very similar amino acid sequences, particularly when they are from the same haplotype, where they
are characteristically ~ 90 % identical (Kusaba et al., 1997). This
led to the suggestion that SRK and SLG may interact with the
putative pollen ligand in a concerted manner and were therefore likely to have co-evolved (Kusaba et al., 1997). However,
subsequent studies showed that this close within-haplotype
sequence similarity between SRK and SLG did not hold for all
S haplotypes. The amino acid sequences of some SLGs and
SRKs were clearly more similar between S haplotypes than
within the same S haplotype, indicating that recombination
can take place between S haplotypes and that strict sequence
conservation of related pairs of SRK and SLG is not essential for
SSI (Suzuki et al., 2000). Subsequently it was demonstrated
that SLG is dispensable for functional SI in Brassica even
though its presence produces a ªstrongerº SSI response, as
measured by the extent of incompatible pollen tube development (Takasaki et al., 2000). Even though the evolutionary
and functional relationship between SRK and SLG is obscure,
amino acid sequence similarities allow them to be assigned to
one of two S haplotype classes (class I and class II) first defined
on the basis of their dominance relationships (discussed in Kusaba et al., 1997, 2000). Class I haplotypes are all dominant or
co-dominant, whereas class II haplotypes are recessive in pollen to the class I haplotypes. Sequences of SRKs and SLGs are
clearly conserved within each class but diverge significantly
between classes ± class I SLGs and SRKs showing only ~ 65 %
amino acid sequence similarity to class II SRKs and SLGs, compared to the ~ 80 % similarity typical of within-class comparisons (Kusaba et al., 2000).
Recently the pollen determinant of SSI was independently
identified in Brassica rapa as a small (6 kD) cysteine-rich peptide located in the pollen coating. Schopfer et al. (1999) named
the peptide S Cysteine Rich protein (SCR), and Takayama et al.
(2000) named it SP11 (S Pollen protein 11). Transgenic gain-offunction experiments demonstrated that SCR/SP11 was necessary and sufficient for pollen S-specificity and in situ hybridization studies confirmed sporophytic expression of SCR/SP11
in the anther tapetum (Takayama et al., 2000), the predicted
site of expression for sporophytically derived proteins destined for the pollen coating (Stephenson et al., 1997). Gametophytic SCR/SP11 expression was also detectable in developing
pollen grains (Takayama et al., 2000) but this was later found
to be true only for dominant (class I) alleles; for recessive
(class II) alleles only sporophytic expression could be detected
(Shiba et al., 2002). Interestingly, in heterozygotes with dominant and recessive SCR/SP11 alleles, expression of the recessive
allele was undetectable (Shiba et al., 2002; Kusaba et al., 2002),
suggesting that allelic dominance is a consequence of suppressed expression of recessive alleles in the presence of dominant
ones. A number of SCR/SP11 alleles have now been characterized, revealing more extreme sequence polymorphism than
Self-Incompatibility in Plants
Plant biol. 5 (2003)
was previously observed for either SLGs or SRKs. Typical pairwise amino acid sequence comparisons of SCR/SP11 alleles
range from ~ 30 ± 55 % amino acid identity (Watanabe et al.,
2000). All SCR/SP11 proteins possess 8 conserved cysteine residues that presumably form four disulphide bridges in the mature protein suggesting that, despite their highly divergent
amino acid sequences, the overall tertiary structure of allelic
forms of SCR/SP11 may be similar.
With both male and female determinants of SSI identified in
Brassica, the biochemical mechanism of pollen recognition
and rejection can potentially be investigated. Hypothetical
models, based on analogies with receptor kinase signalling
systems in animals, predicted an S haplotype-specific protein±protein interaction between SCR/SP11 and the extracellular receptor domain of SRK leading to dimerization of activated
SRKs and autophosphorylation on serines and threonines
within the kinase domain (Fig. 1, Schopfer and Nasrallah,
2000). Recently, this model was confirmed independently by
Kachroo et al. (2001) and Takayama et al. (2001). Kachroo et
al. (2001) used in vitro ªpull downº assays to show haplotypespecific binding between tagged versions of SRK6 and SCR6.
Takayama et al. (2001) demonstrated similar S-haplotypespecific binding of a radio-iodinated synthetic form of S8-SP11
and SRK8. Interestingly, neither Takayama et al. (2001) nor
Kachroo et al. (2001) could demonstrate any significant affinity between SP11/SCR and SLG. However, Takayama et al.
(2001) did detect the presence of SLG8 in SP118/SRK8 complexes, suggesting a haplotype-specific interaction between
SRK and SLG before or during the SRK-SP11 interaction. This
may explain why SLG apparently enhances S-haplotypespecific rejection of pollen but is itself a non-essential component of the SI rejection process (Takasaki et al., 2001). Takayama et al. (2001) also showed that formation of the SRK8/
S8-SP11/SLG8 complex resulted in specific autophosphorylation of serine and threonine residues in the kinase domain of
SRK8, the presumed cue for further downstream intracellular
signalling events leading to pollen arrest.
Progress has already been made in identifying components of
this signalling cascade using the yeast two-hybrid assay system
with SRK as ªbaitº (Bower et al., 1996; Gu et al., 1998). ARC1 is
an arm-repeat protein required for the SI response in Brassica
that shows a specific phosphorylation-dependent interaction
with the kinase domain of SRK (Gu et al., 1998; Stone et al.,
1999). How ARC1 mediates further downstream signalling
events remains to be determined. Two thioredoxin-h proteins,
THL1 and THL2 also interact with the kinase domain of SRK in
a phosphorylation-independent manner (Bower et al., 1996).
THL1 may act as a negative regulator of SRK because THL1 can
inhibit the kinase activity of SRK in the absence of ªactivatingº
components of the pollen coating (presumably SCR/SP11 of the
same haplotype) (Cabrillac et al., 2001). However, the requirement of THL1 for regulated autophosphorylation of SRK during
its binding to SCR/SP11 is inconclusive (Takayama et al., 2001).
The quest to identify further downstream components of the SI
signalling system in the Brassicaceae has been given a boost by
the recent engineering of SI lines of Arabidopsis thaliana. Orthologues of SRK and SCR were isolated from Arabidopsis lyrata, a close SI relative of A. thaliana and transformed into A.
thaliana (Nasrallah et al., 2002). Transfer of the A. lyrata SRK
and SCR genes into A. thaliana was sufficient to impart a full
Fig. 1 Model for the mechanism of SSI in Brassica. Pollen from an
S1S2 individual is inhibited on an S1S3 stigma as a consequence of an
S-haplotype-specific interaction between male (pollen) and female
(stigma) products of the S1 haplotype. The pollen protein SCR1 is recognized by and binds to the extracellular receptor domain of SRK1
thereby inducing dimerization of SRK1 and autophosphorylation on
serine and threonine residues in the kinase domain. Activation of
the SRK initiates an intracellular signalling cascade within the stigmatic papilla cell that leads to localized rejection of the pollen. While
just SCR and SRK are required for S-haplotype specific pollen rejection, the strength of the SI response (in terms of numbers of incompatible pollen grains failing to germinate) is enhanced by the presence of SLG from the same haplotype. SLG shares approximately
90 % amino acid sequence identity with the receptor domain of SRK.
This sequence identity may be important the association of SRK and
SLG within the ªreceptor complexº. Signalling downstream of SRK
has yet to be characterized, however ARC1, an arm-repeat protein,
that binds to the kinase domain of SRK in a phosphorylation-dependent manner is essential for SI, but as yet its function is unknown.
SI phenotype to this normally self-compatible species, indicating that all the signalling components downstream of SRK that
are required for SI are present and functional in A. thaliana.
This finding offers endless possibilities for complete dissection
of the SSI signalling pathway in the Brassicaceae using Arabidopsis.
b) RNase-mediated gametophytic SI
In species from the Solanaceae, Rosaceae and Scrophulariaceae, the female determinant of GSI is a secreted ribonuclease
(S-RNase) that accumulates to very high concentrations in the
transmitting tract of stigma and style (McClure et al., 1989).
S-RNase alleles have been characterized from many species:
Nicotiana, Petunia, Solanum, Physalis and Lycopersicon (Solana-
25
26
Plant biol. 5 (2003)
S. J. Hiscock and S. M. McInnis
Fig. 2 Cytotoxic/inhibitor model for pollen
rejection during the GSI response in species
from the Solanaceae, Rosaceae and Scrophulariaceae (based on Luu et al., 2000). In
these species, S-RNases are the female (pistil) determinant of GSI. The GSI response occurs later than the SSI response and pollen
tubes grow into the style before being arrested. Secreted S-RNases (S1 and S2) are
taken up non-specifically by the growing
pollen tubes. Once inside the pollen tube,
S-RNases are recognized by a general RNase
inhibitor and inhibited. If, however, the SRNase has the same S genotype as the pollen tube, an S-specific RNase interactor
binds to the S-RNase and prevents interaction with the general S-RNase inhibitor. This
S-specific interaction retains the enzymic activity of the S-RNase which proceeds to degrade rRNA within the pollen tube, thereby
preventing protein synthesis and further pollen tube growth.
ceae), Malus, Pyrus and Prunus (Rosaceae) and Antirrhinum
(Scrophulariaceae) and found to be highly polymorphic. Typical amino acid sequence identities between S-RNases from
species in the Solanaceae are around 50 % (McCubbin and Kao,
2000). Conserved and hypervariable regions can be identified
within the S-RNase protein and site-directed mutagenesis, and
domain swapping experiments indicate that the hypervariable
regions are important for S allele-specific pollen recognition
and rejection (Matton et al., 1997; Zurek et al., 1997). Interestingly, Solanaceous S-RNases are frequently more similar between species than they are within species indicating that SRNases within this family are extremely ancient and that they
diversified before speciation (McCubbin and Kao, 2000). More
recently, a phylogenetic analysis of all available S-RNase sequences suggested that S-RNases from the Solanaceae, Scrophulariaceae and Rosaceae have a monophyletic origin, even
though these three families diverged from their common ancestor at least 110 million years ago (Igic and Kohn, 2001).
Transgenic loss-of-function and gain-of-function experiments
demonstrated that S-RNases are required for allele-specific
pollen rejection in Nicotiana (Murfett et al., 1994) and Petunia
(Lee et al., 1994), and analyses of mutant S-RNases in Petunia
and Lycopersicon indicate that ribonuclease activity is essential
for pollen tube rejection (Royo et al., 1994; Huang et al., 1994).
Nevertheless, other stylar components also appear to be required for S-RNase-mediated inhibition of pollen tube growth
(reviewed in McClure et al., 2000) particularly a small asparagine-rich protein, HT. The gene encoding HT is not linked to the
S locus but seems to act as an essential co-factor for SI in Nicotiana (McClure et al., 1999) and Lycopersicon (Kondo et al.,
2002). Indeed a picture is emerging for the existence in vivo
of the S-RNase as a large protein complex consisting of the SRNase (probably as a dimer or oligomer), HT protein, NaTTS
protein (the Nicotiana alata homologue of a N. tabaccum
Transmitting Tissue-Specific arabinogalactan glycoprotein),
NaMG-15 protein (the N. alata homologue of a N. tabaccum
stylar glycoprotein, PELPIII) and p11 an 11 kD copper-binding
phytocyanin (McClure et al., 2000; Cruz-Garcia et al., 2003).
The functional significance of this S-RNase complex is obscure,
but it may be required for uptake of S-RNases by pollen tubes
(Cruz-Garcia et al., 2003).
The male determinant of GSI in pollen remains elusive. It is
clear that S-RNases play no part in the expression of SI in pollen (Dodds et al., 1999) so, as was shown for SSI in the Brassicaceae, two different genes at the S-locus must encode the
male and female determinants of SI. Current models for GSI in
the Solanaceae, Rosaceae and Scrophulariaceae evoke a cytotoxic response in which S-RNases enter incompatible pollen
tubes and degrade ribosomal RNA and mRNA thereby preventing further pollen tube growth (Fig. 2; Golz et al., 2000;
McCubbin and Kao, 2000). Such a model predicts two possible
identities for the pollen S-protein. Either it could be an Sspecific RNase translocator or it could be an S-specific RNase
inhibitor ± specific in the sense that if the S-RNase and S-inhibitor are encoded by the same S-haplotype, the S-RNase inhibitor is ineffectual. The translocator hypothesis now seems the
less likely hypothesis because S-RNases have been shown to
be taken up in vivo with equal ease by both compatible and incompatible pollen tubes (Luu et al., 2000). Moreover, direct
evidence for the inhibitor model has come from studies of pollen-part mutants (PPMs) ± self-compatible plants defective in
pollen SI function but not affected in pistil SI function (de Nettancourt, 1977; Golz et al., 2000). PPMs generated by X-ray and
gamma mutagenesis were shown to possess a duplicated copy
of the S locus translocated to another chromosome or, more
usually, carried on a small additional chromosomal fragment
(ªcentric fragmentº) generated by the mutagenesis (Golz et
al., 1999, 2001). A proportion of pollen grains from the PPMs
will thus carry two different S alleles and hence, two different
S-RNase inhibitors capable of inhibiting both pistil S-RNases,
thereby accounting for the self-compatible phenotype. The
well-known occurrence of GSI breakdown in tetraploids (de
Nettancourt, 1977; Golz et al., 2000) is also consistent with
the inhibitor model because, provided plants are S heterozygotes, their diploid pollen will carry two different S-RNase inhibitors.
Self-Incompatibility in Plants
One current variation on the ªinhibitor modelº suggests that
there may be two pollen components involved in GSI ± a general RNase inhibitor (RI) that can inactivate any S-RNase and
an S-allele-specific product that maintains the activity of a
specific S-RNase by blocking RI binding (Luu et al., 2000,
Fig. 2). A possible candidate for such a general inhibitor of SRNases has been identified recently in Petunia hybrida by its
ability to bind S-RNases in the yeast two-hybrid system (Sims
and Ordanic, 2001). This novel protein, PhSBP1 (Petunia hybrida S-RNase Binding Protein 1) cannot be the pollen S determinant because it does not vary between S haplotypes and its
interaction with S-RNases is not haplotype-specific. However, the strong binding of PhSBP1 to S-RNases suggests that
it plays a general role in the SI response. Interestingly, PhSBP1
contains a C-terminal RING-HC finger domain and RING domain proteins are known to act as E3 ubiquitin ligases that target specific proteins for ubiquitination and subsequent degradation (Joazeiro and Weissmann, 2000). Sims and Ordanic
(2001) therefore speculate that PhSBP1 may have the role of
an E3 ubiquitin ligase in pollen tubes by targeting all S-RNases
for degradation via the ubiquitin pathway. The pollen S determinant would then be a protein that disrupts the interaction
between PhSBP1 and S-RNases in an S-allele-specific manner
as predicted by Luu et al. (2000) (Fig. 2). A strong candidate
for the pollen S determinant has recently been identified by
Lai et al. (2002) working on Antirrhinum (Scrophulariaceae).
This pollen-expressed gene, AhSLF-S2 (Antirrhinum hispanicum S-Locus F-box-S2), exhibits characteristics expected for
the pollen S determinant ± tight linkage to the S locus and allelic polymorphism (Yongbiao Xue, pers. commun.). AhSLF-S2
encodes a putative F-box protein that is expressed specifically
in pollen and in the anther tapetum. Interestingly, F-box proteins, like RING domain proteins, are known to recruit specific
substrates into the ubiquitin-mediated protein degradation
pathway (Craig and Tyers, 1999). However, F-box proteins require their substrates to be phosphorylated and contain a lysine residue for ubiquitin ligation. All S-RNases contain a conserved lysine residue in their third conserved (C3) domain and
a pollen-derived Ca2+-dependent protein kinase has been
shown to phosphorylate S-RNases in vitro in a non-haplotypespecific manner (Kunz et al., 1996). These observations suggest
that the conditions necessary for F-box-mediated degradation
of S-RNases could be met in pollen tubes in vivo. Recently three
more AhSLF alleles have been cloned, AhSLF-S1, AhSLF-S4 and
AhSLF-S5, and like AhSLF-S2, each is tightly linked to a corresponding S-RNase gene encoding the S1-RNase, S4-RNase
and S5-RNase, respectively. AhSLF alleles share ~ 90 % amino
acid sequence identity, and each allele is situated alongside
1 ± 2 paralogous AhSLF copies (30 ± 40 % amino acid identity)
at the S locus (Yongbiao Xue, pers. comm.). By virtue of this allelic polymorphism, AhSLF could function as both the S-RNase
inhibitor and the determinant of S-allele specificity (Lai et al.,
2002) making it the strongest candidate for the pollen determinant yet identified. Alternatively AhSLF may determine pollen S specificity by interacting with the general S-RNase inhibitor (PhSBP1?) according to the model of Luu et al., 2000. Notwithstanding these two possibilities, the fact that two proteins
involved in the ubiquitin-mediated protein degradation pathway have been implicated in the pollen part of the GSI reaction
bodes well for molecular confirmation of the inhibitor model
of S-RNase-mediated GSI.
Plant biol. 5 (2003)
c) GSI in Papaver
Even though the genetics of GSI is identical in Papaver rhoeas
(common poppy) and species from the Solanaceae, Rosaceae
and Scrophulariaceae, S-RNases do not regulate the GSI response in Papaver (reviewed by Jordan et al., 2000 a). Instead
a novel ~ 15 kD stigma-specific secreted protein (S-protein)
has been shown to induce S-haplotype-specific arrest of pollen
tube growth in vitro (Franklin-Tong et al., 1988). The S-protein
gene has been cloned and five alleles characterized (Foote et
al., 1994; Walker et al., 1996; Kurup et al., 1998). Like other Sproteins, Papaver S-proteins are highly polymorphic, sharing
between ~ 51 ± 64 % amino acid identity, and some are glycosylated. However, the proteins are predicted to possess a virtually identical secondary structure consisting of a series of bstrands linked by hydrophilic (probably surface) loops. Using
this secondary structure prediction in combination with sequence comparisons, Ride et al. (1999) identified a large gene
family in Arabidopsis with homology to Papaver S-proteins. So
far over 50 S-protein homologues (SPHs) have been identified
in the Arabidopsis genome but surveys of predicted open-reading frames suggest there may be as many as 100 SPHs in Arabidopsis (Ride et al., 1999; Jordan et al., 2000 a). The function(s)
of SPHs in Arabidopsis has yet to be determined but preliminary expression studies of two SPH genes revealed fairly universal expression (albeit variable) in different tissues, suggesting that these SPH genes have a general role in cell development and differentiation (Jordan et al., 2000 a).
In Papaver an in vitro bioassay for SI has allowed biochemical
dissection of the signalling events that accompany the arrest
of incompatible pollen tubes and eventual pollen cell death.
Challenge of growing pollen tubes with incompatible S-protein (purified or recombinant) initiates a rapid transient increase in cytosolic free Ca2+ (Franklin-Tong et al., 1993) ahead
of tube arrest. Increasing the Ca2+ concentration in growing
pollen tubes with Ca2+ ionophores (such as A23187, or mastoporan) or by the release of caged Ca2+ can mirror this effect, indicating that GSI in Papaver is mediated by a Ca2+-dependent
signal transduction pathway (Franklin-Tong et al., 1993, 1995).
Ratio-imaging of intracellular free Ca2+ in pollen tubes challenged by incompatible S-protein demonstrated that the tipfocussed Ca2+ gradient essential for continuous pollen tube
growth becomes dissipated, such that Ca2+ levels fall at the
tube tip but increase rapidly in subapical regions (FranklinTong et al., 1997). These events coincide with dramatic changes
in the F-actin cytoskeleton, such that after 10 ± 15 min no actin
microfilament bundles can be seen within pollen tubes (Snowman et al., 2000). More recent studies have shown that this
phenomenon is associated with a large and sustained depolymerization of F-actin (Snowman et al., 2002).
The in vitro GSI response is also accompanied by rapid phosphorylation and dephosphorylation of specific pollen proteins.
Two proteins showing hyperphosphorylation in pollen tubes
after challenge with incompatible S-protein are p26 and p68
(Rudd et al., 1996). Phosphorylation of p26 is Ca2+- and calmodulin-dependent, whereas phosphorylation of p68 (which becomes phosphorylated later than p26) is Ca2+-independent
(Rudd et al., 1997). The gene encoding p26 has been cloned
and shown to encode an active inorganic pyrophosphatase
(Rudd and Franklin-Tong, 2003) but the role of this enzyme in
the GSI response is at present unclear. Recent studies have re-
27