Stylar glycoproteins bind to S-RNase in vitro


(fax 1 573 882 5635; e-mail


S-RNases determine the specificity of S-specific pollen rejection in self-incompatible plants of the Solanaceae, Rosaceae, and Scrophulariaceae. They are also implicated in at least two distinct types of unilateral interspecific incompatibility in Nicotiana. However, S-RNase itself is not sufficient for most types of pollen rejection, and evidence for its direct interaction with pollen tubes is limited. Thus, non-S-RNase factors also are required for pollen rejection. As one approach to identifying such factors, we tested whether SC10-RNase from Nicotiana alata would bind to other stylar proteins in vitro. SC10-RNase was immobilized on Affi-gel, and binding proteins were analyzed by SDS-PAGE and immunoblotting. In addition to SC10-RNase and a small protein similar to lily chemocyanin, the most prominent binding proteins include NaTTS, 120K, and NaPELPIII, these latter three being arabinogalactan proteins previously shown to interact directly with pollen tubes. We also show that SC10-RNase and these glycoproteins migrate as a complex in a native PAGE system. Our hypothesis is that S-RNase forms a complex with these glycoproteins in the stylar ECM, that the glycoproteins interact directly with the pollen tubes and thus that the initial interaction between the pollen tube and S-RNase is indirect.


It is generally thought that the lipids, sugars, proteins, and glycoproteins that are secreted onto the stigma surface and into the extracellular matrix (ECM) of the transmitting tract allow the pistil to recognize and reject undesirable pollen while supporting the growth of compatible pollen. However, there are few components of the ECM that have been characterized in detail and a very limited number of pollen-pistil recognition mechanisms that are understood at the molecular level.

S-RNase-dependent pollen rejection mechanisms are the most widespread pollen rejection systems that have been described (Igic and Kohn, 2001). S-RNase dependent self-incompatibility (SI) occurs in the Solanaceae, Rosaceae, and Scrophulariaceae. These families have gametophytic SI systems, where a single multi-allelic S-locus controls the specificity of pollen rejection (Kao and Tsukamoto, 2004). Pollen is rejected if its S-haplotype is matched by either of the two S-haplotypes present in the diploid pistil.

Each S-haplotype encodes a distinct S-RNase that determines specificity on the style side (Lee et al., 1994; McClure et al., 1989, 1990; Murfett et al., 1994). S-RNase is expressed in cells of the stigma, style and ovary and is secreted into the ECM (Anderson et al., 1989; Cornish et al., 1987). S-RNase ribonuclease activity is required for pollen rejection (Huang et al., 1994). Moreover, pollen RNA is degraded in incompatible pollination but this does not occur in compatible pollinations (McClure et al., 1990). Such results are the basis for models that suggest that S-RNases act as S-specific cytotoxins (Gray et al., 1991; Kao and Tsukamoto, 2004; McClure et al., 1990). It is not known how S-RNase gains access to the pollen tube cytoplasm.

Recent results show that S-locus F-box protein genes (SLF/SLB genes) determine S-specificity on the pollen side, and thus, correspond to pollen-S (McClure, 2004). F-box proteins are well known as components of SCF E3 ubiquitin ligase complexes, and it is possible that SLF forms such a complex in pollen (Qiao et al., 2004). As compatible pollen tubes do not show RNA degradation, it has long been thought that pollen-S inhibits S-RNase providing resistance to its cytotoxic action (Kao and Tsukamoto, 2004). A common model is that SLF participates in an SCF complex that tags non-self-S-RNase for degradation and thus provides a form of resistance (Kao and Tsukamoto, 2004; Qiao et al., 2004; Ushijima et al., 2004). However, both self- and non-self-S-RNases have been detected in Solanum chacoense pollen tubes (Luu et al., 2000) and the biochemical basis of SLF action remains uncertain.

S-RNase-dependent pollen rejection is mechanistically diverse. Apart from its role in S-specific pollen rejection, S-RNases have also been implicated in at least two different types of interspecific unilateral incompatibility (UI) in Nicotiana (Murfett et al., 1996). SI N. alata rejects pollen from SC N. plumbaginifolia through a mechanism that closely resembles SI (Murfett et al., 1996). This type of UI is referred to as ‘factor dependent’ as, like SI, it requires both S-RNase and other stylar factors (McClure et al., 2000). Pollen from Nicotiana tabacum is also sensitive to S-RNase-dependent rejection, but the mechanism is distinct (Murfett et al., 1996). Both these UI systems show less specificity than S-specific pollen rejection (Beecher, 1999; Beecher and McClure, 2001). For example, most S-RNases cause rejection of N. plumbaginifolia pollen, whereas each S-RNase causes rejection of a single N. alata S-haplotype. In addition, a chimeric SA2-/SC10-RNase that could not cause S-specific pollen rejection retained inhibitory activity toward N. tabacum pollen. However, RNaseI from Escherichia coli, a non-S-RNase, could not substitute for S-RNase in either UI mechanism, suggesting that S-RNases are specially adapted for pollen rejection (Beecher, 1999; Beecher and McClure, 2001).

S-RNases are expressed as abundant components of the ECM that forms the pathway for pollen tube growth from the stigma to the ovary (Anderson et al., 1989; Cornish et al., 1987). Typically, they are basic glycoproteins of 30–35 kDa that can accumulate to near-millimolar concentrations (Jahnen et al., 1989). S-RNases are often extremely polymorphic. Only about 40 residues distributed among five conserved regions are highly conserved. Outside these conserved regions, the level of identity between allelic S-RNases varies such that overall identity ranges from 40% to >95%. Crystal structures for two S-RNases have been reported. The structures show that the conserved regions form the hydrophobic core of the protein and the active site (Ida et al., 2001). Variable regions are on the surface where they can interact with factors from the pollen or the pistil.

A catalytically active S-RNase is critical for pollen rejection (Huang et al., 1994), but non-S-RNase factors are also required. Such factors can be grouped according to their functions (McClure et al., 2000). Group 1 factors directly affect expression of S-RNase. Tsukamoto et al. (1999) described a group 1 factor in Petunia axillaris that specifically affects expression of S13-RNase. Group 2 factors are required for pollen rejection, but do not affect S-RNase expression per se. For example, HT-B-protein, a group 2 factor, is a small asparagine-rich protein required for S-RNase-dependent pollen rejection but not for S-RNase expression (Kondo et al., 2002; McClure et al., 1999; O'Brien et al., 2002). Group 1 and 2 factors function only in pollen rejection and, therefore, defects in such factors lead to compatibility of otherwise incompatible pollinations. S-RNase may also interact with stylar factors that have broader functions. Factors that function in pollen rejection and have other roles as well (e.g., as structural proteins, support pollen tube growth) are referred to as group 3 factors (McClure et al., 2000).

Abundant glycoprotein components of the stylar ECM in Nicotiana interact directly with pollen tubes. A transmitting tract-specific (TTS) glycoprotein in N. tabacum is deglycosylated by pollen tubes and associates with the pollen wall and vegetative cell membrane (Cheung et al., 1993, 1995; Wu et al., 1995). The 38 kDa TTS polypeptide backbone contains a variable amount of arabinogalactan, and the glycoprotein runs a smear from about 50–100 kDa in SDS-PAGE. Purified TTS enhances pollen tube growth in vitro, and antisense inhibition of TTS expression reduces pollen tube growth in vivo. A similar low-salt extractable glycoprotein – NaTTS – has been described in N. alata (Wu et al., 2000). A high-salt extractable glycoprotein with the same polypeptide backbone but with subtly different properties has also been described (Sommer-Knudsen et al., 1996, 1998). Pistil extensin-like protein III (PELPIII) from N. tabacum, another abundant glycoprotein of the stylar ECM, localizes to the pollen callose wall and plugs (Goldman et al., 1992; de Graaf et al., 2003). Like TTS, PELPIII is an arabinogalactan protein (AGP); the 78 kDa polypeptide is modified with both arabinogalactan and tetra-arabinosyl moieties. Unlike TTS, PELPIII does not affect pollen tube growth in vitro and antisense inhibition does not affect growth in vivo (Bosch, 2002). The 120 kDa glycoprotein (120K) is one of the most abundant components of the transmitting tract matrix in N. alata (Lind et al., 1994). While its properties are very similar to PELPIII, EM-level immunolocalization studies have shown that 120K is taken into pollen tubes (Lind et al., 1996). NaTTS, 120K, and NaPELPIII (i.e., the N. alata ortholog of PELPIII) are chimeric AGPs – they possess homologous cysteine-rich C-terminal domains of about 138 amino acids that are unlikely to be glycosylated as well as highly glycosylated N-terminal segments. NaPELPIII and 120K are also regarded as hybrid AGPs as they contain both arabinogalactan and extensin-like glycan (Schultz et al., 2002). It is not known whether glycosylated N-terminal segments or the C-terminal segments mediate interactions with pollen tubes.

Our goal was to identify stylar S-RNase binding proteins as candidates for non-S-RNase factors involved in SI. We hypothesized that components of the ECM bind to S-RNase and play a role in pollen rejection. To test this, we prepared affinity matrices by immobilizing SC10-RNase on Affi-gel. The pollen-interacting glycoproteins are prominent among the proteins retained from style extracts and are thus candidates for non-S-RNase SI factors involved in S-specific pollen rejection.


SC10-RNase was purified from SI N. alataSC10SC10 and immobilized on Affi-gel-10. In vitro binding assays were performed by passing style extracts over this SC10-RNase Affi-gel matrix (McClure et al., 2000). Unbound material was washed through, and bound material was eluted at pH 2.8 and analyzed by SDS-PAGE. Buffer conditions were chosen to minimize potential non-specific interactions. SC10-RNase is a basic protein (pI = 8.5), so binding studies were usually carried out at elevated pH (pH 8.8) to minimize ion exchange effects. To minimize retention of aggregated proteins, buffers containing 1% Tween-20 were used in washing steps.

Figure 1 shows that binding is specific and is not pH dependent. Equal amounts of extract from SI N. alataSC10SC10 were applied to either SC10-RNase Affi-gel or to control matrices prepared with RNaseI from E. coli, bovine serum albumin (BSA), or ethanolamine (McClure et al., 2000). Each matrix contained approximately 400 μg of immobilized protein per ml of resin. RNaseI is a member of the RNase Rh/T2/S family of ribonucleases. It has similar size and charge properties to S-RNases (30 kDa, pI = 8.3), but it is not functional in pollen rejection (Beecher and McClure, 1999; Beecher et al., 1998). BSA and ethanolamine were used as further controls for non-specific binding. The three control columns bind little protein at either pH 6.0 or pH 8.8 (unbound, UB versus bound, B; Figure 1a,b). Similar results were obtained at pH 5.2 (data not shown). SC10-RNase-Affi-gel retains two prominent stainable bands, aside from SC10-RNase itself (arrows p11, and HMW; Figure 1), and a variable number of minor bands.

Figure 1.

Specific protein binding to SC10-RNase Affi-gel. Style extracts (200 mg FW ml−1) from SI Nicotiana alataSC10SC10 were prepared and tested for binding in either pH 8.8 binding buffer (a), or pH 6 binding buffer (b). Extracts were applied to 0.5 ml of Affi-gel with 400 μg ml−1 immobilized SC10-RNase, RNaseI, or BSA, or to Affi-gel blocked with ethanolamine. CE, crude extract. UB, unbound proteins. B, bound proteins eluted with low pH buffer after washing with binding buffers containing 1% Tween-20. Major binding proteins are indicated with arrows.

The low molecular weight S-RNase binding protein, designated p11, is a member of the plantacyanin family of copper containing proteins (McClure et al., 2000). p11 was purified by ion exchange chromatography and the N-terminal sequence was determined. This sequence was used to design primers for amplifying the corresponding cDNA. Ultimately, a full-length cDNA was cloned from SI N. alata. Figure 2 is an alignment of the mature p11 sequence; a basic copper binding protein (CBP) from cucumber whose three-dimensional structure is known; and chemocyanin, a lily protein recently shown to act as a pollen chemoattractant (Kim et al., 2003). The four residues that form the copper ligands in CBP are marked with asterisks (i.e., H39, C79, H84, and Q89 in Figure 2). Methionine and glutamine are both common at position 89. Although it is likely that p11 is a true CBP, we have not directly tested this feature.

Figure 2.

Alignment of p11-like sequences. Sequences of cucumber basic protein, CBP (Murata et al., 1982); p11; chemocyanin, Lily; and a consensus sequence, Con are shown. Signal sequences are omitted. Underlining, N-terminal sequence of p11. CBP sequences in β-sheet conformation (arrow), α-helix conformation (swirl), or a disulfide bond (S-S) in the crystal structure are shown. Amino acids bound to the copper ion in CBP are indicated by asterisks. Known CBPs accommodate either M or Q at the position equivalent to M-89 in CBP. Some p11 clones contained a single base change resulting in a polypeptide four residues shorter (i.e., terminating at the same position as CBP and chemocyanin).

The high molecular weight (HMW, Figure 1) species runs as a relatively tight band in the 12.5% Tris-tricine gels used to resolve all the major binding proteins. However, the appearance of this band depends on gel conditions; it runs as a smear in lower percentage gels and, as described below, contains several glycoproteins. In 7–10% gels, the HMW S-RNase binding species runs as a smear from about 70 kDa to >120 kDa. This pattern is similar to the TTS protein described in N. tabacum and NaTTS described in N. alata. Figure 3 confirms that NaTTS is a component of the HMW S-RNase binding fraction.

Figure 3.

Sequential extraction of high molecular weight SC10-RNase binding proteins.
(a, b) SI Nicotiana alataSC10SC10 styles were ground under liquid nitrogen and subjected to seven sequential extractions. Lanes 1–4, extractions with 0.05 m NaCl; lanes 5–7, extractions with 0.4 m NaCl. Extracted proteins were subjected to SDS-PAGE and (a), silver stained; or (b), blotted to nitrocellulose and immunostained with anti-TTS.
(c) Pooled extracts were desalted and tested for binding to SC10-RNase-Affi-gel. The protein blot was immunostained with anti-TTS. P, pooled extract; UB, unbound, B, bound.

Some transmitting tract proteins including S-RNase, 120K, and NaTTS are extractable under low-salt conditions (Cheung et al., 1995; Lind et al., 1994; Wu et al., 1995, 2000) while others are extractable under high-salt conditions (Sommer-Knudsen et al., 1996, 1998). To test the solubility characteristics of the S-RNase binding proteins, we performed four extractions with low-salt buffer (LSB) containing 0.05 m NaCl, followed by three extractions with high-salt buffer (HSB) containing 0.4 m NaCl. The protein concentrations were 1.00, 0.42, 0.22, and 0.19 mg ml−1 for the 0.05 m NaCl extracts, and 0.68, 0.33, and 0.26 mg ml−1 for the extractions in 0.4 m NaCl. Figure 3(a) shows that most of the SC10-RNase and 120K were extracted in the first two low-salt extractions. Little additional SC10-RNase or 120K was extracted with high salt. The protein blot in Figure 3(b) was immunostained with an anti-TTS antibody. The results show anti-TTS-reactive protein was extracted under both low- and high-salt conditions; more of this protein was recovered after high-salt extraction. However, the low-salt extractable material shows slower migration and is more diffuse than the high-salt extractable material (Figure 3b), suggesting a higher degree of glycosylation. Fractions were pooled and tested for binding to SC10-RNase-Affi-gel. Figure 3(c) shows that anti-TTS-reactive material from both low- and high-salt pools bound to the affinity matrix. Only the low-salt extractable form of this protein would be recovered under the conditions normally used to prepare extracts for S-RNase binding assays (i.e., 0.05 m Tris pH 8.8, 0.05 m NaCl, 1% 2-mercaptoethanol). Thus, based on its solubility, mobility, and antibody reactivity, one of the SC10-RNase binding proteins is NaTTS.

TTS proteins from N. tabacum and N. alata interact directly with pollen tubes, but other pollen-interacting style glycoproteins including the 120K and NaPELPIII have been described. We prepared a panel of highly specific antibodies to test whether these glycoproteins also bind SC10-RNase. The N-terminal domains are more divergent than the C-terminal domains, but they are not good targets for antibody production because they are heavily glycosylated. Figure 4(a) shows an alignment of the conserved C-terminal cysteine-rich domains of the glycoproteins and the regions selected for anti-peptide antibody production are indicated. To test the specificity of the antibodies, the C-terminal domains of NaTTS, 120K, and NaPELPIII were expressed in E. coli as GST fusions. Figure 4(b) shows that each antibody recognizes only its target protein. Two different NaPELPIII antibodies (i.e., G498 and M1019) were used in this study; both show equivalent specificity.

Figure 4.

Glycoprotein antibody specificity.
(a) Alignment of the C-terminal cysteine-rich regions of pollen-interacting glycoproteins NaTTS, NaPELPIII, and 120K. Sequences used to generate antipeptide antibodies are in bold.
(b) C-terminal regions are expressed as GST fusions blotted to nitrocellulose and stained as indicated at left. Lane 1, vector only; lane 2, NaTTS fusion; lane 3, 120K fusion; lane 4, NaPELPIII fusion. Some cleavage of the fusion protein occurred in each case.

We devised a more stringent binding assay to test whether the three pollen-interacting glycoproteins bind to SC10-RNase. In these experiments, crude style extracts were prepared and passed over SC10-RNase-Affi-gel. Bound proteins were eluted and then reapplied to the column. Figure 5 shows that all three glycoproteins, SC10-RNase, and p11 are present in both the first and second bound fractions of such a bind–rebind experiment. Excess protein is present in the extracts applied to the column, and only a portion of the binding proteins is retained on the first pass. However, little of the protein present in the first bound fraction fails to bind when reapplied to the column (B1 versus UB2 and B2, Figure 5). As binding efficiency is similar in the first and second rounds, we conclude that factors present in the unbound fraction are not required for binding. Further studies focused on the HMW glycoprotein fraction as this constitutes the most abundant bound fraction.

Figure 5.

Pollen-interacting glycoproteins bind to SC10-RNase Affi-gel. Extracts from 2.5 g Nicotiana alataSC10SC10 (CE) were passed over SC10-RNase Affi-gel in a bind–rebind assay. UB1, unbound proteins from the first pass; B1, bound proteins from the first pass; UB2, unbound proteins from the second pass; B2, bound proteins from the second pass. Loading based on fresh weight. Silver stain and glycoprotein blots: CE, UB1, 0.8 mg FW/lane; B1, UB2, B2, 37 mg FW/lane; SC10-RNase and p11 blots: CE, UB1, 0.5 mg FW/lane; B1, UB2, B2, 18 mg FW/lane. The G498 NaPELPIII Ab was used in this experiment.

Figure 6 shows results from a 2-D gel system devised as a further means to test for interactions between SC10-RNase and NaTTS, 120K, and NaPELPIII. The experimental approach is diagrammed in Figure 6(a). Our hypothesis is that SC10-RNase exists in a complex with the pollen-interacting glycoproteins. Thus, we used a native gel system suitable for basic proteins such as SC10-RNase, NaTTS, 120K, and NaPELPIII in the first dimension followed by Tris-tricine SDS-PAGE in the second dimension (Schuster, 1971). Purified SC10-RNase migrates about halfway through the native gel and appears as a spot near the middle of the second dimension gel (Figure 6b). In contrast, when proteins eluted from SC10-RNase Affi-gel are resolved in this system, most of the SC10-RNase, NaTTS, 120K, and NaPELPIII protein comigrate near the top of the native gel (Figure 6c, filled arrowheads). A variable amount of each glycoprotein migrates farther into the gel (Figure 6c, open arrowheads). The S-RNase smearing in Figure 6(c) is typical and may be due to poor dissociation from the glycoproteins. The results are consistent with a large SC10-RNase glycoprotein complex migrating near the top of the native gel. It is also possible that complexes are formed with individual glycoproteins. On some gels, we have observed multiple S-RNase spots and glycoproteins migrating with slightly different mobilities. However, the results in Figure 6 are more typical; among nine experiments, three showed multiple S-RNase spots at the left of the denaturing gel and six showed results like those shown in Figure 6(c).

Figure 6.

Two-dimensional gel analysis of SC10-RNase-glycoprotein complexes.
(a) Schematic of the separation. Vertical hash marks, glycoproteins; horizontal hash marks, SC10-RNase; cross hash, complexes. Proteins were separated in a native gel system for cationic proteins (top). Native gel strips were positioned with the cathode end to the right for electrophoresis through denaturing gels.
(b) Purified SC10-RNase (3 μg) runs about half way through the native gel (open arrowhead).
(c) Proteins eluted from SC10-RNase Affi-gel in a bind–rebind experiment (i.e., equivalent to 35 mg FW, fraction B2 in Figure 5). Free glycoproteins run farther into the native gel (open arrowheads) than the complexes (filled arrowheads). The M1019 NaPELPIII Ab was used in this experiment.

Figure 7 shows that pollen-interacting glycoproteins from self-compatible N. plumbaginifolia and N. tabacum also bind to SC10-RNase. Figure 7(a) shows extracts prepared and applied to SC10-RNase Affi-gel in a one-pass binding experiment. The immunostained blots show that all three pollen-interacting glycoproteins are present in the bound fractions. We previously described expression of S-RNases in the self-compatible species N. plumbaginifolia and N. tabacum (Beecher and McClure, 2001; Beecher et al., 1998; Murfett et al., 1996). These plants do not display S-specific pollen rejection, but they do gain the ability to reject pollen from SC species such as N. tabacum and N. glutinosa. As this is the most general type of S-RNase-dependent pollen rejection system known, we tested whether complexes could be detected in such transgenic plants. Extracts from transgenic N. tabacum expressing SC10-RNase behaved very similarly to extracts from SI N. alata SC10SC10 (Figure 7b versus Figure 5). A portion of each of the pollen-interacting glycoproteins bound to the matrix (CE versus UB1 and B1, Figure 7b), and nearly all of the protein in the B1 fraction was retained in the B2 fraction (B1 versus UB2 and B2, Figure 7b).

Figure 7.

Pollen-interacting glycoproteins from self-compatible species bind SC10-RNase.
(a) Extracts from Nicotiana plumbaginifolia and N. tabacum were passed over SC10-RNase Affi-gel in a one-pass experiment. Proteins were immunostained as shown. CE, crude extract; UB, unbound; B, bound. Nicotiana alata CE, UB, 0.8 mg FW, B, 28 mg FW; N. tabacum TTS and PELPIII blots CE, UB, 0.5 mg FW, B, 12.5 mg FW, 120 kDa blot, CE, UB, 1 mg FW, B, 25 mg FW; N. plumbaginifolia CE, UB, 1 mg FW, B, 20 mg FW.
(b) Extract from N. tabacum expressing SC10-RNase was passed over SC10-RNase Affi-gel in a bind–rebind experiment as described in Figure 5. CE, UB1, 0.8 mg FW/lane; B1, UB2, B2, 37 mg FW/lane. The G498 NaPELPIII Ab was used in these experiments.


The column-based assay shows that style proteins specifically bind S-RNase in vitro. Columns with immobilized BSA or RNaseI from E. coli were very poor binding substrates compared to immobilized SC10-RNase. Ion exchange effects or non-specific protein aggregation cannot explain binding. Under our standard binding conditions, both BSA and RNaseI are more highly charged than SC10-RNase, yet they bind very little protein. Moreover, S-RNase binding occurs over a broad pH range; a result inconsistent with ion exchange behavior.

There are two prominent classes of S-RNase binding proteins in style extracts of SI N. alataSC10SC10 aside from S-RNase itself: the HMW class, migrating near 100 kDa, consisting of at least three pollen-interacting glycoproteins; and p11, a putative CBP similar to chemocyanin (Kim et al., 2003). Additional minor bands are also present in the bound fraction, but have not been extensively characterized. Some of these minor species appear to be degradation products of other proteins.

p11 is similar to the plantacyanins, a group of blue CBPs (Nersissian et al., 1998). Although we did not directly test whether p11 binds copper, it is interesting to note that low levels of copper improve the growth and morphology of Nicotiana pollen tubes grown in vitro (Read et al., 1993). p11 is also similar to lily chemocyanin (Kim et al., 2003). Chemocyanin lacks the methionine or glutamine residues that usually form the fourth copper ligand in blue CBPs but it has been shown to be a chemoattractant for lily pollen tubes. We have not tested whether p11 is a chemoattractant for N. alata pollen tubes.

The HMW glycoproteins are clearly the most abundant stylar S-RNase binding proteins. Immunological and biochemical results confirm that these correspond to NaTTS, 120K, and PELPIII, three proteins that individually interact with pollen tubes (Bosch, 2002; de Graaf, 1999; Lind et al., 1994; Wu et al., 2000). It is possible that other stylar proteins may also bind S-RNase. Identification of a TTS-like protein in the bound fraction is of particular interest. TTS associates with pollen tubes in vitro, and it is possible that the pollen tubes use TTS as a growth substrate in vivo (Cheung et al., 1995; Wu et al., 1995, 2000). A similar protein, designated galactose-rich style glycoprotein (GaRSGP), has been identified in N. alata, but it is reportedly extractable only with high salt (Sommer-Knudsen et al., 1998). In our experiments, style extracts normally were prepared under low-salt conditions (i.e., 0.05 m NaCl). Our results show anti-TTS-reactive material is extracted under both low- and high-salt conditions (Figure 2b; Wu et al., 2000), and that both forms bind SC10-RNase-Affi-gel (Figure 2c). We refer to the HMW anti-TTS-reactive S-RNase binding protein as NaTTS because our extracts contain the low-salt extractable form and because the disperse molecular weight of this material resembles TTS from N. tabacum.

Both the column binding assay and the 2-D gel results are consistent with formation of complexes between S-RNase and other stylar proteins. The stringent bind–rebind assay showed the three pollen-interacting glycoproteins and p11 bind to SC10-RNase Affi-gel (Figure 5). The 2-D gel analysis of material eluted from SC10-RNase Affi-gel shows that S-RNase and the glycoproteins migrate to similar positions in a native gel. In contrast, free S-RNase migrates to a different position, as do free glycoproteins (open arrows, Figure 6b). These results are consistent with formation of a complex but the resolution is not sufficient to determine its architecture. For example, as glycoproteins and SC10-RNase show similar mobilities in the native dimension, we cannot distinguish between a single large complex and separate complexes.

The pollen-interacting glycoproteins NaTTS, 120K, and NaPELPIII are candidates for SI factors. They are clearly not products of the S-locus and are not likely to be required for S-RNase expression. Each glycoprotein interacts with pollen tubes (Cheung et al., 1995; de Graaf, 1999; Lind et al., 1996; Wu et al., 1995, 2000), suggesting they have roles apart from S-specific pollen rejection. Our hypothesis is that S-RNase is bound to these glycoproteins in the ECM and that the initial interaction between S-RNase and pollen tubes may be an indirect consequence of this interaction. This hypothesis helps explain some enigmatic features of S-RNase-based pollen rejection. For example, extremely high (i.e., approximately millimolar concentrations) levels of S-RNase are required for pollen rejection. This is unexpected for a highly specific direct interaction with the pollen tube. Furthermore, S-RNases are cytotoxins directed against pollen. The species-level benefits of outcrossing aside, there is considerable selection on the pollen to evade the action of S-RNase. The long-term stability of SI suggests that S-RNase-based pollen rejection systems are linked to a critical pollen function such that pollen experiences a strong over-riding selection that ultimately drives its interaction with S-RNase. It has been suggested that the S-RNase binding glycoproteins, and TTS in particular, support growth of compatible pollen tubes (Wu et al., 1995). Thus, we propose that S-RNase-based pollen rejection exploits the interaction with these glycoproteins to drive pollen tubes into association with S-RNase.

In addition to their well-known role in SI, S-RNases also are implicated in inter-specific pollen rejection. An indirect interaction with pollen tubes also makes sense in this context as pollen from other species clearly derives no benefit from interaction with S-RNase. For instance, S-RNase is implicated in the unilateral interspecific incompatibility between N. alata and its SC relative N. tabacum. We previously showed that while transgenic N. plumbaginifolia or N. tabacum plants expressing S-RNase do not display S-specific pollen rejection, they do reject N. tabacum pollen in an S-RNase-dependent manner (Beecher and McClure, 2001; Murfett et al., 1996). This is the simplest example of a non-S-specific pollen rejection mechanism that requires S-RNase. A non-S-RNase, RNaseI from E. coli, could not substitute for S-RNase (Beecher and McClure, 2001). However, all the factors other than S-RNase that are required for this type of pollen rejection clearly must be expressed in N. plumbaginifolia and N. tabacum. Here, we showed that complexes form between SC10-RNase and the glycoproteins present in these SC species and that transgenic N. tabacum expressing SC10-RNase also shows evidence of complexes (Figure 7b). Thus, S-RNase interacts with glycoprotein components of the ECM that are thought to support compatible pollen tube growth in both SI and SC species. Interaction with these glycoproteins may explain why pollen from SC species is subject to S-RNase-dependent pollen rejection.

This hypothesis is also consistent with current models for pollen-part function in SI. A pollen receptor for S-RNase has never been detected. This is expected if S-RNase binds directly to stylar glycoproteins and its initial interaction with the pollen tube is, therefore, indirect. The popular models postulate that pollen-S acts as an inhibitor of non-self-S-RNase (Kao and Tsukamoto, 2004). It is possible that SLF fulfills this role by tagging non-self-S-RNase for degradation (Qiao et al., 2004; Sijacic et al., 2004; Ushijima et al., 2004). In any case, this interaction between S-RNase and SLF is thought to take place after S-RNase enters the pollen tube. Thus, the pollen-interacting glycoproteins could play a role in delivery of S-RNase to SLF.

Experimental procedures

Plant materials

All plant materials have been described previously (Murfett et al., 1994, 1996).

Preparation of Affi-gel matrices

SC10- and SA2-RNases were purified by FPLC cation exchange chromatography as described (Murfett et al., 1994) or using a two-column procedure as follows. Styles (i.e., SA2SA2 or SC10SC10) were ground under liquid nitrogen and homogenized in 0.05 m MES, 0.05 m NaCl, 0.05 m sodium ascorbate, 0.005 m EDTA pH 6.5 (5 ml g−1 FW). Extracts were filtered through miracloth, clarified by centrifugation at 10 500 g (Sorvall HB-6 rotor, Kendro, Langenselbold, Germany) and the oil pellicle removed. Extracts were applied to CM-Sepharose (Sigma, St Louis, MO, USA) and the column was washed with 0.05 m MES, 0.175 m NaCl pH 6.5. Proteins eluted with 0.35 m NaCl were diluted with nine volumes 0.05 m HEPES pH 8.0 and applied to SP-Sepharose. The column was washed with 0.05 m HEPES pH 8.0 and then eluted with 0.06, 0.12, 0.24, and 0.72 m NaCl in the same buffer. Fractions containing S-RNase were identified by SDS-PAGE.

For each different matrix, Affi-gel-10 (Bio-Rad, Hercules, CA, USA) was washed with cold water and mixed with SC10-RNase, RNaseI (Promega M4261, Madison, WI, USA), or BSA (Sigma A-3059, St. Louis, MO, USA) in 0.1 m MOPS pH 7.5. After gentle rotation at 4°C overnight, remaining active esters were blocked by addition of ethanolamine. The ethanolamine control matrix was prepared in the same way but without any added protein. Matrices were transferred to small columns and run through washing cycles with binding and elution buffer prior to use. Matrix preparations that ‘leaked’ S-RNase were discarded. Binding efficiencies were approximately 80%. Matrices used in Figures 1 and 3 each contain about 0.4 protein ml−1 resin; those used in Figures 5, 6, and 7 contain about 3 mg SC10-RNase ml−1.

Affi-gel binding

Affi-gel matrices were equilibrated with binding buffer [BB; 0.05 m Tris, 0.05 m NaCl, 1% (v/v) 2-mercaptoethanol pH 8.8] in small columns. For binding assays, crude extracts were prepared from N. alataSC10SC10 styles in BB (200 mg FW ml−1), clarified by centrifugation, and the oil removed. Extracts were applied, and unbound material removed by washing with 10 column volumes of BB containing 1% Tween-20 (Sigma P1379, St. Louis, MO, USA). Proteins were eluted with 3 column volumes of 0.01 m glycine, 0.01 m NaCl pH 2.8 and acetone precipitated. For bind–rebind experiments, pellets were resuspended in BB, reapplied to the column, and eluted. Binding assays conducted at pH 5.2 and pH 6.0 were performed similarly except that the BBs contained 0.05 m sodium acetate or 0.05 m MES, respectively. For the 2-D gel analyses shown in Figure 6, the bind–rebind procedure was scaled up for 2.5 g preparations of N. alataSC10SC10 styles and a 3 ml SC10–RNase Affi-gel column.

Sequential extractions

For the experiment shown in Figure 3, 200 mg SI N. alata SC10SC10 styles were ground under liquid nitrogen and homogenized in LSB [0.05 m sodium acetate, 0.05 m NaCl 1% (v/v) 2-mercaptoethanol]. After centrifugation at 13 000 g for 15 min at 4°C, the supernatant was carefully removed. The pellet was resuspended in 1 ml LSB, and the procedure repeated. After a total of four extractions in LSB, the pellet was similarly extracted three times with HSB [0.05 m sodium acetate, 0.4 m NaCl 1% (v/v) 2-mercaptoethanol]. Portions of each supernatant were analyzed in 12.5% Tris-tricine gels and silver stained or blotted to nitrocellulose for immunostaining. LSB and HSB extracts were separately pooled and desalted into BB by passage over Sephadex G-25 (PD-10; Amersham Pharmacia, Piscataway, NJ, USA). Binding to SC10-RNase-Affi-gel was performed as described earlier.

Purification and cloning of p11

p11 was purified from SC N. alata and from both SI N. alataSC10SC10 and SA2SA2. For the experiment shown in Figure 4, 1 g SC N. alata styles were homogenized in 0.05 m Tris, 0.05 m NaCl, 1% 2-mercaptoethanol pH 8.8. The extract was centrifuged at 10 000 g for 10 min at 4°C. The supernatant was passed over S-Sepharose, and the unbound (acidic) fraction collected. Basic proteins were eluted with 0.05 m Tris, 0.5 m NaCl pH 8.8. Acidic and basic proteins were exchanged into BB and bound to SC10-RNase-Affi-gel as described above. In the experiment shown in Figure 5, 10 g SI N. alataSA2SA2 styles were homogenized in 0.05 m Tris, 1% 2-mercaptoethanol pH 8.8, and acidic proteins removed by adsorption onto Q-Sepharose (Sigma, St. Louis, MO, USA). Basic proteins were then adsorbed onto S-Sepharose, eluted batchwise with 0.05 m sodium acetate, 1 m NaCl pH 5.0, and precipitated from saturated ammonium sulfate. After centrifugation, the pellet was resuspended in 0.05 m sodium acetate pH 5.2, desalted on Sephadex G-25, and applied to a Mono-S FPLC column. The column was developed with a pH gradient (A = 0.05 m sodium acetate pH 5.2, B = 0.01 m sodium carbonate pH 11.5; 40 ml gradient, 1 ml fractions). For N-terminal sequencing, p11 was prepared similarly from 10 g of SI N. alataSC10SC10 styles. Fraction 39 was acetone precipitated, separated in a 12.5% Tris-tricine gel, and blotted onto PVDF (Immobilon-Psq; Millipore, Billerica, MA, USA). The blot was stained with Amido Black, and the center of the p11 band was cut out for sequencing. A single sequence [AIYNVGDGNG(C)TFGVS(N)(N)] was observed, but residues 11, 17, and 18 were ambiguous.

Style RNA from SI N. alataSC10SC10 was used as template to prepare cDNA primed by the oligonucleotide GGCCACGCGTCGACTAGTACT17. A 279-bp cDNA was amplified using the oligonucleotide GGCCACGCGTCGACTAGTAC and a degenerate oligonucleotide [GAGAGAATTCATITA(T/C)AA(T/C)GTIGGIGATGG] designed based on the p11 peptide sequence IYNVGDG. A style cDNA library was prepared from SC N. alata RNA in lambda Ziplox (Life Technologies) and screened with the 279 bp 3′ RACE product. Six independent clones were obtained and sequenced.

2-D gel analysis

The first-dimension discontinuous native gel system for cationic proteins was modified from Schuster (1971). The 4% stacking gel buffer was 0.125 m potassium acetate pH 4.7. The 7.5% separating gel buffer was 0.15 m potassium acetate pH 6.7. The tank buffer was 0.07 mβ-alanine, 0.012 m acetic acid adjusted to pH 4.5 with KOH. Samples were applied in 10% sucrose, 0.1% methylene blue. Electrophoresis was carried out until the methylene blue tracking dye had run 5 cm through the separating gel toward the cathode (approximately 200 V/h, Bio-Rad mini-Protean II apparatus, 0.75 mm gels). First-dimension gels were cut into strips, washed in water, soaked in SDS loading buffer and placed on top of standard 7 or 10% Tris-tricine gels for glycoproteins or S-RNase, respectively.

Protein blot analysis

120K is a highly basic protein that does not migrate toward the anode under standard blotting conditions. The following alkaline blotting procedure was developed and used for glycoprotein analysis. After electrophoresis, gels were soaked in freshly prepared alkaline blot buffer (0.05 m Na3PO4, 0.01 m H3BO3, 0.0025% SDS, 0.001 m DTT, 20% methanol) two times for 5–10 min each. Gels were placed onto nitrocellulose membranes with filter paper soaked in alkaline blot buffer on both the anode and cathode sides. These were sandwiched between filter papers soaked in transfer buffer (0.025 m Tris, 0.19 m glycine, 0.0025% SDS, 20% methanol); transfer was accomplished at 7–9 V for 1 h (Bio-Rad semidry blot apparatus). The anti-TTS antibody was a gift from Prof. Alice Cheung, University of Massachusetts, Amherst. The NaTTS, 120K, and NaPELPIII antibodies were produced against synthetic peptides (NTKKTLVEQGKTC amide, NNARKANVQTC amide, and PPKQPITPAVV amide, respectively) as ovalbumin (Biosource International, Hopkinton, MA, USA) or MAP conjugates (Genemed Synthesis, San Francisco, CA, USA), and were affinity purified prior to use. Blots were developed with alkaline phosphatase-conjugated secondary antibodies and visualized with NBT/BCIP as previously described (Murfett et al., 1996).


We thank Prof. Alice Cheung, University of Massachusetts-Amherst, for supplying the anti-TTS antibody and for helpful suggestions. We thank Melody Kroll for assistance with manuscript preparation. Peptide sequencing was preformed by Dr Gautam Suarath, University of Nebraska. Felipe Cruz-Garcia was supported by Universidad Nacional Autonoma de Mexico grant DGAPA IN211702 and the University of Missouri Molecular Biology Program. This work was supported by the University of Missouri Food for the 21st Century Program, and National Science Foundation Grants 96-04645 and 99-82686.