A small asparagine-rich protein required for S-allele-specific pollen rejection in Nicotiana

Abstract

Although S-locus RNases (S-RNases) determine the specificity of pollen rejection in self-incompatible (SI) solanaceous plants, they alone are not sufficient to cause S-allele-specific pollen rejection. To identify non-S-RNase sequences that are required for pollen rejection, a Nicotiana alata cDNA library was screened by differential hybridization. One clone, designated HT, hybridized strongly to RNA from N. alata styles but not to RNA from Nicotiana plumbaginifolia, a species known to lack one or more factors necessary for S-allele-specific pollen rejection. Sequence analysis revealed a 101-residue ORF including a putative secretion signal and an asparagine-rich domain near the C terminus. RNA blot analysis showed that the HT-transcript accumulates in the stigma and style before anthesis. The timing of HT-expression lags slightly behind S_C10-RNase in SI N. alata S_C10S_C10 and is well correlated with the onset of S-allele-specific pollen rejection in the style. An antisense-HT construct was prepared to test for a role in pollen rejection. Transformed (N. plumbaginifolia × SI N. alata S_C10S_C10) hybrids with reduced levels of HT-protein continued to express S_C10-RNase but failed to reject S_C10-pollen. Control hybrids expressing both S_C10-RNase and HT-protein showed a normal S-allele-specific pollen rejection response. We conclude that HT-protein is directly implicated in pollen rejection.

Self-incompatibility (SI) systems are the best understood pollen rejection mechanisms. Many plants have SI systems in which the specificity of pollen rejection is controlled by a single locus, referred to as the S-locus (1, 2). In the Solanaceae, pollen is rejected when the single S-allele in a haploid pollen tube matches either of the two S-alleles in the diploid pistil. In the pistil, the products of the S-locus are secreted ribonucleases called S-RNases (3). Each S-allele encodes a different S-RNase that contains the specificity determinants for S-allele-specific recognition by pollen. RNase activity is required for pollen rejection (4), and it is generally accepted that S-RNases act as S-allele-specific cytotoxins that inhibit growth of pollen bearing a matching S-allele (5, 6). The nature of the specificity determinant in pollen is not known, but it is distinct from S-RNase.

By definition, the S-locus encodes the determinants of allelic specificity. However, other loci are also required for pollen rejection. In SI Brassica, an aquaporin gene has been shown to be required for pollen rejection (7). In Brassica and in Papaver, gene products that bind to S-proteins have been identified, but it is not yet known whether they play a direct role in pollen rejection (8–10). In early studies in the Solanaceae, East demonstrated a requirement for multiple loci in SI and also suggested that such factors may interact differently with different S-alleles (11). Recently, Bernatzky et al. (12) generated self-compatible (SC) Lycopersicon esculentum lines containing SI Lycopersicon hirsutum chromosome fragments bearing the S-locus, providing further genetic evidence that the S-RNases alone are not sufficient for SI. Similarly, an S-allele from SC Petunia hybrida cv. Strawberry Daddy was shown to be functional when crossed into SI Petunia inflata, suggesting that a factor from the SI background could complement a factor missing in cv. Strawberry Daddy (13). By using plant transformation, we showed that, when S-RNase is expressed in transgenic SC Nicotiana plumbaginifolia, it does not cause S-allele-specific pollen rejection. However, when it is expressed in (N. plumbaginifolia × SC Nicotiana alata) hybrids, both S-allele-specific pollen rejection and a type of interspecific pollen rejection occur normally (14). Thus, when present in trans, factors from the SC N. alata background allow the S-RNase transgene to function in SI and interspecific pollen rejection. The identities and functions of these factors are not known.

We used a differential cDNA-cloning approach to identify putative non-S-RNase factors required for S-allele-specific pollen rejection in the style. Here, we report cloning a cDNA encoding a small asparagine-rich protein expressed late in style development. Antisense transformed (N. plumbaginifolia × SI N. alata S_C10S_C10) hybrids showing reduced expression of this protein accept N. alata S_C10 pollen even though they still express S_C10-RNase.

Materials and Methods

Plant Materials and Transformation.

Nicotiana longiflora (inventory no. TW79, accession no. 30A) was obtained from the U.S. Tobacco Germplasm Collection (Crops Research Laboratory, Oxford, NC). All other plant materials have been previously described (14–16). For the antisense experiments, transgenic N. plumbaginifolia plants were generated by Agrobacterium-mediated transformation of leaf explants (17).

cDNA Cloning.

Polyadenylated RNA was prepared from mature N. plumbaginifolia and SC N. alata cv. Breakthrough styles. cDNA libraries of stylar SC N. alata sequences were then prepared in λ-gt-10 (Promega) and λ-ZIPLOX (Life Technologies, Grand Island, NY) by following recommended procedures. The λ-gt-10 library was screened by differential hybridization with ³²P-labeled cDNA prepared from N. plumbaginifolia and SC N. alata style RNA. DNA sequencing and RNA blot analyses showed that the HT cDNAs were incomplete. The λ-ZIPLOX library was therefore screened to obtain a full-length clone.

Antisense Experiments.

The HT-cDNA was engineered to create a SacI site 27 bp upstream of the initiation codon and a BamHI site 146 bp past the terminator by PCR with the synthetic oligonucleotides GCTTGGATCCTTATTACAAACAAAGTGGAAATTAACATAACG and GTCAGGAGCTCGAAAATTTATAAGATAATTCGTCCAAATGGC. The product was digested and ligated to the cauliflower mosaic virus 35S promoter and nos 3′ sequences from pAGUS1 (15, 18). The HT-antisense construct was recloned in pZP122 (19), transferred to Agrobacterium tumefaciens GV3101, and used to transform N. plumbaginifolia. (N. plumbaginifolia × SC N. alata) hybrids previously transformed with pSC109617 (14), and expressing S_C10-RNases, were also transformed with the HT-antisense construct. Transformants were regenerated on gentamycin or gentamycin plus kanamycin. Twelve independent, doubly transformed (N. plumbaginifolia × SC N. alata) hybrids were analyzed as primary transformants. Then, 26 independently transformed N. plumbaginifolia lines were crossed to untransformed N. alata S_C10S_C10 and second-generation transgenic (N. plumbaginifolia × SI N. alata S_C10S_C10) hybrids were analyzed for HT-expression and pollination phenotype (15).

RNA and Protein Blot Analysis.

Organs were collected, frozen in liquid nitrogen, and stored at −70°C until needed. Total RNAs were prepared and separated in 2% agarose formaldehyde gels as described (5). RNAs were blotted onto Hybond N+ (Amersham) and stained with methylene blue to check for equal loading. Blots were hybridized to ³²P-labeled HT- or S_C10-RNase cDNAs. Stringent washes were performed in 0.15 × SSPE [standard saline phosphate/EDTA (0.18 M NaCl/10 mM phosphate, pH 7.4/1 mM EDTA)], 1% SDS, 1% powdered milk, and 4 mg/ml salmon testis DNA (Sigma, D-3159) at 68°C. Autoradiographs were prepared with Kodak XRP-5 films (Rochester, NY) at −70°C. In the experiment shown in Fig. 4, the results were quantitated with a Molecular Dynamics model 400A PhosphorImager and were normalized to the highest signal.

Figure 4.

HT-transcript accumulation in SI N. alata S_C10S_C10. Developing styles (i.e., stigma plus style) were dissected from buds of the indicated sizes. (Upper) Autoradiographs. (Lower) Signals quantified with a PhosphorImager and normalized to the highest signal. Total RNAs (5 μg per lane) were blotted and probed with ³²P-labeled HT- or S_C10-RNase probes.

To generate the HT-antiserum, the HT-cDNA was engineered to create EcoRI and BamHI sites flanking the nonreptitive part of the mature-coding sequence by using the synthetic oligonucleotides GAAGGATCCAGGGATATGGTTGATCCTTCAATATCATT and GAGGAATTCTTAACCCTTTTGGCATTTGCAAGCTGCAC. The ORF was cloned into pGEX2TK (Amersham), and a glutathione S-transferase-HT fusion was purified from Escherichia coli with glutathione agarose (Sigma, G-4510). After further purification by SDS/PAGE, the fusion protein was injected into a rabbit to raise an anti-HT serum (20). For protein blot analysis, styles were weighed, homogenized in SDS-loading buffer (0.2 M Tris⋅HCl, pH 6.8/0.5 M DTT/4% SDS/25% glycerol; 10 μl per mg of freshweight), boiled, and centrifuged. Proteins were separated in 10% Tris-Tricine gels (21) and blotted onto Nitrobind (Micron Separations, Westborough, MA) by using a Bio-Rad Transblot SD semi-dry electroblotting apparatus. Blots were treated with the rabbit HT-antiserum or a mouse monoclonal anti-S_C10-RNase Ab (14), and immune complexes were detected by using alkalinephosphatase-conjugated secondary antibodies and nitroblue tetrazolum/BCIP (5-bromo-4-chloro-3-indolyl phosphate) (20).

Pollination Phenotypes.

Emasculated flowers were pollinated 1 day after petal opening with pollen from SI N. alata S_C10S_C10 or S₁₀₅S₁₀₅ as described (14). Styles were harvested after 72 hr, stained with decolorized aniline blue, and examined by epifluorescence (22). Whenever possible, the total number of pollen tubes penetrating to the base of the style was counted. However, when >≈50 pollen tubes were present, it was not possible to obtain accurate counts. Styles with >150 pollen tubes were scored as highly compatible (+++).

Results

The SC N. alata cultivar Breakthrough does not express an S-RNase, but our previous results show that it does express non-S-RNase factors required for pollen rejection (14). To clone these factors, a cDNA library was prepared by using mature style mRNA. Several sequences were identified in a differential screen as being highly expressed in SC N. alata but showing little or no signal with a N. plumbaginifolia probe. As a secondary screen, putative non-S-RNase factor clones were tested for hybridization to RNA from SI N. alata, SC N. alata, N. plumbaginifolia, and also N. longiflora, a second SC species that lacked factors required for S-allele-specific pollen rejection (B.M, unpublished data). Putative non-S-RNase factors are expected to be expressed in both accessions of N. alata but show little or no expression in N. plumbaginifolia or N. longiflora. Fig. 1 shows RNA blot results for a clone designated HT. The 600-nt HT-transcript is present in style RNA from SI N. alata S₁₀₅S₁₀₅ and SC N. alata but cannot be detected in either N. plumbaginifolia or N. longiflora, even after extended exposures.

Figure 1.

HT-transcript accumulation in different genetic backgrounds. The 600-nt HT-transcript is only visible in SI N. alata S₁₀₅S₁₀₅ and SC N. alata, which both express a full complement of the non-S-RNase factors required for S-allele-specific pollen rejection. N. plumbaginifolia and N. longiflora represent backgrounds deficient in one or more such factors. Polyadenylated RNA from mature styles (200 ng/lane) was blotted and hybridized with ³²P-labeled HT-cDNA.

Fig. 2 shows the predicted 101-residue amino acid sequence of the HT-protein. A putative signal sequence is present at the N terminus, including a basic residue at position 4 followed by a series of hydrophobic residues. The SignalP algorithm (23) predicts the highest probability of signal sequence cleavage before Arg-24, leaving a mature polypeptide of 8.6 kDa. The mature HT-protein is predicted to be fairly acidic (i.e., calculated pI = 3.76 for residues 24–101, Fig. 2). In part, this is due to a striking stretch of 20 asparagine and aspartic acid residues near the C terminus. The asparagine-rich domain is flanked on each side by three cysteine residues that could be involved in disulfide bonding or posttranslation modification, but it is not yet known whether such modifications exist. The sequence does not contain potential N-glycosylation sites. Aside from homologies to the repetitive asparagine-rich domain, there were no significant matches to the HT-protein sequence in the databases (i.e., blast (24) score >30; search conducted in Feb. 1999).

Figure 2.

Predicted amino acid sequence of HT-protein. The predicted secretion signal sequence is lowercase.

Fig. 3 shows total RNA blot analysis of HT-transcript expression in various organs of SC N. alata. Anthers and pistils were removed from buds ranging from 0.5 to 5.5 cm. In SC N. alata cv. Breakthrough, the 0.5- to 1-cm stage includes anthers in tetrads and anthesis occurs in buds ≈5.5 cm in length. HT-transcript is not detectable at any stage of anther development. In the pistil (i.e., including the stigma and the style, but not the ovary), a low level of HT-transcript is first detected in 2- to 3.5-cm buds, and the amount of expression increases dramatically at maturity. Fig. 3 shows the highest level of HT-transcript present in anthesis-stage pistils. Stigmas and styles were dissected from anthesis-stage pistils, and Fig. 3 shows that HT-transcript is expressed in both organs, with a slightly higher level in the style. Fig. 3 shows no HT-transcript in four nonsexual organs: petals, sepals, stems, and leaves.

Figure 3.

HT-transcript accumulation in SC N. alata cv Breakthrough. Developing anthers (lanes 1–4) and styles (i.e., stigma plus style, lanes 5–8) were dissected from buds of the indicated sizes. G + Y, styles at anthesis; G, isolated stigmas at anthesis; Y, isolated styles at anthesis; Pt, petals; Sp, sepals; St, stems; Lf, leaf. Total RNAs (10 μg/lane) were blotted and hybridized with ³²P-labeled HT-cDNA.

A similar experiment was performed to examine HT-transcript expression in SI N. alata S_C10S_C10. RNA blots were prepared and hybridized to S_C10-RNase or HT-cDNA probes. Fig. 4 shows that HT-transcript accumulation lags behind S_C10-RNase transcript accumulation. Whereas S_C10-RNase transcript is first detectable in 1- to 2-cm buds, HT-transcript is first detectable at the next stage, in 2.0- to 3.5-cm buds (Fig. 4 Upper). The hybridization results were quantified by using a PhosphorImager and normalized to the signal from mature style RNA (i.e., 5.5-cm anthesis, Fig. 4). The most critical stage is when the buds are 2.0–3.5 cm long because this is when the style becomes competent to support pollen tube growth but rejects S_C10-pollen poorly. At this stage, Fig. 4 shows that S_C10-RNase transcript levels are already 67% of their highest level, whereas HT-transcript levels are only 4% of the level in mature styles. By the next stage, when the pistil is fully competent to reject S_C10-pollen, HT-transcript levels show an eightfold increase (i.e., to 32% of the mature level, Fig. 4).

An antisense construct was prepared to test whether HT-protein is required for S-allele-specific pollen rejection. Initially, the construct was transformed into (N. plumbaginifolia × SC N. alata) hybrids expressing S_C10-RNase from pSC109617 (14). The results with these doubly transformed hybrids were promising. Controls showed normal S-allele-specific rejection of S_C10-pollen, but some antisense transformed plants showed changes in pollination phenotype (data not shown). To confirm these results with second generation plants, the HT-antisense construct was transformed into N. plumbaginifolia and then crossed with SI N. alata S_C10S_C10. Because N. plumbaginifolia is easier to transform than N. alata, this approach affords an opportunity to analyze a greater number of independently transformed lines than would be available from direct transformation of N. alata. We have previously used this approach to show that S-RNase antisense constructs suppress S-RNase expression and prevent S-allele-specific pollen rejection in the resulting hybrid [i.e., the second generation after transformation, (15)].

To facilitate analysis of the transformed plants, an HT-antiserum was prepared to a glutathione S-transferase-HT-protein fusion. The antiserum reacts with several stylar polypeptides that migrate with apparent M_r from 10 to 18 kDa. To confirm that these polypeptides correspond to HT-protein, they were partially purified from SI N. alata S_A2S_A2 and S_C10S_C10 material, blotted onto poly(vinylidene difluoride) and subjected to N-terminal sequencing. A band with apparent M_r of ≈18 kDa gave the sequence RDMVDPSISL, corresponding to the putative N terminus predicted by the SignalP algorithm (Fig. 2). Two slightly faster migrating bands isolated from SI N. alata S_C10S_C10 material, both gave the sequence KIGGKVGMFF, corresponding to residues 52–61 in Fig. 2. Thus, HT-protein may be subject to processing or degradation, but it is clear that the HT-antiserum is specific to HT-protein.

The HT-antiserum was used to examine HT-protein levels in hybrids transformed with the antisense construct. The results show that suppression of HT-protein expression interferes with S-allele-specific pollen rejection. Fig. 5 shows protein blot analysis of style extracts from SI N. alata S_C10S_C10, N. plumbaginifolia, an untransformed (N. plumbaginifolia × SI N. alata S_C10S_C10) hybrid, and hybrid progeny from six independent antisense HT-transformed N. plumbaginifolia lines. The positive controls, SI N. alata S_C10S_C10 and the untransformed hybrid, both express HT-protein and S_C10-RNase. N. plumbaginifolia, the negative control, does not express either protein and does not show S-allele-specific pollen rejection. The transformed hybrids all express S_C10-RNase, but five (i.e., 9.1.2, 28.1.4, 33.1.6, 36.1.3, and 44.1.1) show no detectable HT-protein expression, and hybrid 22.1.3 shows partial suppression.

Figure 5.

Protein blot analysis of HT-antisense transformed (N. plumbaginifolia × SI N. alata S_C10S_C10) hybrids. SI N. alata S_C10S_C10 and the untransformed (N. plumbaginifolia × SI N. alata S_C10S_C10) hybrid are positive controls showing expression of S_C10-RNase, HT-protein, and a normal S-allele-specific pollen rejection phenotype. Untransformed N. plumbaginifolia is a negative control expressing neither protein and accepts both types of pollen. The transformed hybrids represent progeny from six independently transformed HT-antisense N. plumbaginifolia lines. Style extracts equivalent to 0.33 mg (anti S_C10-RNase) or 2.5 mg of freshweight (anti-HT) were blotted and immunostained. Pollination phenotypes were tested using pollen from N. alata S_C10- or S₁₀₅-pollen (Table 1) and are summarized here as +, compatible; −, incompatible; or +/−, semi-compatible.

Table 1 contains the results from at least five pollinations on each transgenic hybrid (results also summarized in Fig. 5). Fig. 6 shows examples of the style squashes used to score the pollination phenotypes. Because pollen tubes become crowded and tangled, styles with >150 pollen tubes at the bottom of the style were scored as “uncountable” or highly compatible (cf. S₁₀₅-pollen results in Fig. 6; +++, Table 1). Untransformed control (N. plumbaginifolia × SI N. alata S_C10S_C10) hybrids showed a normal S-allele-specific pollen rejection response. Thirteen pollinations with S₁₀₅-pollen were scored as +++ (i.e., uncountable), but only two pollen tubes were observed in 10 styles pollinated with S_C10-pollen. Broadly, the number of S_C10-pollen tubes that penetrate to the base of the style is inversely correlated with the level of HT-protein expression seen in Fig. 5. For example, the partially suppressed hybrid consistently showed several (i.e., 11–27, Table 1 and Fig. 6) S_C10-pollen tubes at the base of the style 72-h after pollination. Two fully suppressed hybrids (i.e., 28.1.4 and 33.1.6, Fig. 5) were scored as uncountable in every pollination. Three others (i.e., 9.1.2, 36.1.3, and 44.1.1; Fig. 5) showed a reduced number of S_C10-pollen tubes in at least some pollinations. Thus, the ability to specifically reject S_C10-pollen is suppressed in these transformed hybrids. When HT-protein levels are reduced, pollination with S_C10-pollen comes to resemble pollination with S₁₀₅-pollen (Fig. 6).

Table 1.

Pollen tubes at the base of the style 72 h postpollination in antisense-HT-transformed (N. plumbaginifolia × SI N. alata S_C10S_C10) hybrids

Pistil	S105-pollen positive control	SC10-pollen test
22.1.3	+++, +++, +++, +++, ∼150	27, 19, 11, 15, 19, 27
9.1.2	+++, +++, +++, +++, +++	+++, +++, +++, 34, 26
28.1.4	+++, +++, +++, +++, +++, +++	+++, +++, +++, +++, +++
33.1.6	+++, +++, +++, +++, +++, +++	+++, +++, +++, +++, +++, +++
36.1.3	+++, +++, +++, +++, +++	+++, +++, +++, 45, 32
44.1.1	+++, +++, +++, +++, +++	+++, 100, 80, 29, 22

Figure 6.

Pollen tube staining results for HT-antisense transformed (N. plumbaginifolia × SI N. alata S_C10S_C10) hybrids. Hybrids were pollinated with S_C10- or S₁₀₅-pollen and prepared for epifluorescence microscopy after 72 hr. As shown in the diagram, the field of view is at or very near the base of the style. No pollen tubes are visible at the base of the style after pollinating untransformed hybrids with S_C10-pollen but a fragment of epidermis (ep) is shown for orientation. Pollen tubes (pt) appear as fibers with brightly stained callose plugs (arrows). The untransformed control hybrid shows normal S-allele-specific pollen rejection because no S_C10-pollen tubes are evident but many S₁₀₅-pollen tubes can be seen at the base of the style. Approximately 20 S_C10-pollen tubes are visible in the partially suppressed hybrid (22.1.3). Numerous S_C10-pollen tubes are visible in fully suppressed hybrids such as 33.1.3, indicating loss of S-allele-specific pollen rejection.

Discussion

Because N. plumbaginifolia and N. alata are closely related, and have similarities in their pollination behavior, we reasoned that most of the genes expressed in their styles would be highly homologous. We therefore used a differential screen to identify candidate non-S-RNase factors expressed in SC N. alata, but not in N. plumbaginifolia, that might be required for S-allele-specific pollen rejection. Among the selected cDNAs, only the HT clone showed a qualitative difference in expression between N. plumbaginifolia and SC N. alata. Further experiments showed that the 600-nt HT-transcript is expressed in SI N. alata S₁₀₅S₁₀₅ but not in a SC N. longiflora accession (Fig. 1). Expression in the SI background is consistent with a role in S-allele-specific pollen rejection. N. longiflora was tested because, like N. plumbaginifolia, it appears to lack one or more factors required for S-RNase-dependent pollen rejection (B. Mou and B. McClure, unpublished data). The accession used here expresses an S-RNase-like protein, but it is SC and also accepts pollen from N. plumbaginifolia. However, (N. longiflora × SC N. alata) hybrids reject pollen from N. longiflora and from N. plumbaginifolia, suggesting that the S-RNase-like protein can function when expressed in conjunction with factors from SC N. alata. This result is identical to the behavior of transgenic N. plumbaginifolia and (N. plumbaginifolia × SC N. alata) hybrids expressing S-RNase (14). Thus, the results in Fig. 1 show that HT-transcripts are present in two N. alata accessions known to possess a full complement of factors required for S-allele-specific pollen rejection and are not expressed in two SC species known to be defective for one or more such factors.

The 101-residue polypeptide predicted from the HT-cDNA sequence contains a stretch of 20 asparagine and aspartate residues near the C terminus. Asparagine-rich domains occur in genes from diverse organisms, but no general function has been ascribed to them. Some “nonclassical” arabinogalactan protein genes contain asparagine-rich domains (25, 26). Interestingly, the asparagine-rich domains inferred from arabinogalactan protein-cDNA sequences were not present in the isolated proteins, suggesting that they were removed in vivo. Unfortunately, aside from homologies to the repetitive asparagine-rich domain, there are no clearly homologous sequences in the databases to suggest a function for the HT-protein. However, it is noteworthy that HT-protein is predicted to be acidic, so it could interact with basic proteins such as S-RNase, although no such direct interaction has yet been detected.

The HT-protein was identified in style extracts by using an HT-antiserum raised against a glutathione S-transferase-HT fusion expressed in E. coli. The most prominent immunoreactive species in style extracts migrate with an apparent M_r of ≈18 kDa. This is considerably larger than the size predicted from the cDNA sequence, but N-terminal sequencing confirmed that these polypeptides are derived from the HT sequence. The aberrant migration may be due to posttranslational modification. Sequence from the 18-kDa band matched the N-terminal sequence predicted by the SignalP program suggesting that the HT-protein is secreted (i.e., RDMVDPSISL, Fig. 2). However, the HT-protein is unstable in style extracts (B.M., unpublished data), and faster migrating bands showed sequences derived from internal HT-protein regions (i.e., KIGGKVGMFF, Fig. 2). Thus, the conclusion that the HT-protein is secreted should be regarded as tentative until it is confirmed by another method.

The pattern of HT-transcript expression is consistent with a role in S-allele-specific pollen rejection. In SC N. alata, the transcript is not detectable in anthers or in nonsexual organs, but it is expressed in mature stigmas and the styles (Fig. 3). In SI N. alata S_C10S_C10, HT-transcript accumulation lags slightly behind S_C10-RNase transcript accumulation (Fig. 4). The flowers in this SI accession are slightly larger than in cv. Breakthrough, tetrads are visible in anthers from 0.8-cm buds, and the flowers open when they are 6- to 7-cm long. Bud-selfing is only successful in buds 2.5-cm long, suggesting that the pistil has just become competent to support pollen tube growth. Slightly longer buds reject S_C10-pollen. At this critical time (i.e., 2.0- to 3.5-cm buds, Fig. 4), S_C10-RNase expression has reached two-thirds of its maximum level, but HT-transcript levels are very low. The subsequent rapid increase in HT-transcript is strongly correlated with the onset of S-allele-specific pollen rejection (Fig. 4).

An antisense construct was prepared from the HT-cDNA, and its effect on pollination phenotype was tested in (N. plumbaginifolia × SI N. alata S_C10S_C10) hybrids. These hybrids are sterile so pollination phenotypes were assessed by examining style squashes stained with decolorized aniline blue. HT-protein and S_C10-RNase levels were monitored in control and transgenic hybrids by immunostaining style extracts. Untransformed hybrids express both proteins and show S-allele-specific pollen rejection; few S_C10-pollen tubes penetrate to the base of the style, but the hybrids are compatible with S₁₀₅-pollen (Fig. 5, Table 1). Suppressed HT-expression is correlated with loss of the ability to reject S_C10-pollen in the transgenic hybrids (Fig. 5, Table 1). S_C10-RNase levels in the transgenic hybrids were similar to the levels in the untransformed control. Therefore, we conclude that loss of S-allele-specific pollen rejection in the transgenic hybrids is due to reduced expression of HT-protein.

Although these results implicate HT-protein in S-allele-specific pollen rejection, its precise function is unknown. Clearly, it is incapable of causing pollen rejection on its own because it is expressed in SC N. alata. The variability in S_C10-RNase expression in transgenic hybrids with suppressed HT-expression (Fig. 5) is intriguing. However, HT-protein cannot be required for S-RNase expression per se because it is possible to achieve high level S-RNase expression in N. plumbaginifolia, which does not express detectable HT-protein. Moreover, HT-transcript accumulation begins after S_C10-RNase transcript (Fig. 4). We have not been able to purify enough HT-protein to quantify expression at the protein level. However, even allowing for the fact that it appears to be unstable in style extracts (B. McClure, unpublished data), it is unlikely that HT-protein is stoichiometric with S-RNase, which is expressed at near millimolar levels (27). These observations suggest that it is unlikely that HT-protein and S-RNase form a complex that is active in pollen rejection. It remains possible that HT-protein and S-RNase interact in some other way. It is also possible that HT-protein interacts only with pollen tubes, perhaps facilitating S-RNase uptake.

Although its exact function is unknown, our data show that HT-protein is implicated in S-allele-specific pollen rejection. We are currently working toward identification of further non-S-RNase factors required in style-part SI functions. It is likely that several such factors exist. However, we still know nothing about the pollen-part specificity determinant (pollen-S), or whether additional factors, beyond the specificity determinant, may be required for pollen SI functions. As more factors in the SI pathway are characterized, it should be possible to define the functions of individual factors more precisely.

Abbreviations

SI

self-incompatible

S-RNases

S-locus RNases

SC

self-compatible