Identification of SFBB-Containing Canonical and Noncanonical SCF Complexes in Pollen of Apple (Malus × domestica)

Gametophytic self-incompatibility (GSI) of Rosaceae, Solanaceae and Plantaginaceae is controlled by a single polymorphic S locus. The S locus contains at least two genes, S-RNase and F-box protein encoding gene SLF/SFB/SFBB that control pistil and pollen specificity, respectively. Generally, the F-box protein forms an E3 ligase complex, SCF complex with Skp1, Cullin1 (CUL1) and Rbx1, however, in Petunia inflata, SBP1 (S-RNase binding protein1) was reported to play the role of Skp1 and Rbx1, and form an SCFSLF-like complex for ubiquitination of non-self S-RNases. On the other hand, in Petunia hybrida and Petunia inflata of Solanaceae, Prunus avium and Pyrus bretschneideri of Rosaceae, SSK1 (SLF-interacting Skp1-like protein1) is considered to form the SCFSLF/SFB complex. Here, we isolated pollen-expressed apple homologs of SSK1 and CUL1, and named MdSSK1, MdCUL1A and MdCUL1B. MdSSK1 was preferentially expressed in pollen, but weakly in other organs analyzed, while, MdCUL1A and MdCUL1B were almost equally expressed in all the organs analyzed. MdSSK1 transcript abundance was significantly (>100 times) higher than that of MdSBP1. In vitro binding assays showed that MdSSK1 and MdSBP1 interacted with MdSFBB1-S 9 and MdCUL1, and MdSFBB1-S 9 interacted more strongly with MdSSK1 than with MdSBP1. The results suggest that both MdSSK1-containing SCFSFBB1 and MdSBP1-containing SCFSFBB1-like complexes function in pollen of apple, and the former plays a major role.


Introduction
Self-incompatibility (SI) is a widespread genetic mechanism to prevent self-fertilization and promote outcrossing in angiosperms. In Rosaceae, Solanaceae and Plantaginaceae, gametophytic selfincompatibility (GSI) is controlled by a single S locus with multiple haplotypes. If the haploid pollen tube has an S haplotype in common with one of the two S haplotypes of the diploid pistil, the pollen tube is recognized as self and rejected [1]. The S haplotype contains at least two genes, S-RNase and F-box protein encoding gene SLF/SFB/SFBB that control pistil and pollen specificity, of apple (MdSBP1) was identified for the first time in Rosaceae [31]. MdSBP1 includes a RING-HC domain and interacts with S-RNase, as for SBP1 homologs in Solanaceae; however, it still remains unclear whether MdSBP1 is a component of an SCF-like complex involved in GSI. In Rosaceae, both two putative E3 ligase complexes, i.e., SBP1-containing SCF-like complex and SSK1containing SCF, exist in pollen within a species have not been reported. The functions of SBP1 and SSK1 are an intriguing issue to be addressed in order to understand the biochemical mechanism of self/non-self S-RNase recognition by E3 ligase complexes.
As the first step toward understanding the molecular mechanism of the S-RNase-based GSI of Rosaceae, we aimed to identify putative members of the GSI-related E3 (-like) complex(es) of apple. Here, we isolated apple homologs of SSK1 and CUL1s from pollen RNA by RT-PCR, and named them MdSSK1, MdCUL1A and MdCUL1B. Then, we examined the binding of MdSSK1 and MdSBP1 with MdCUL1s and MdSFBB1-S 9 , a candidate for pollen S [14,16]. In vitro binding assays showed that both MdSSK1 and MdSBP1 interacted with MdSFBB1-S 9 and MdCUL1, suggesting that both MdSSK1 and MdSBP1 would form SCF SFBB (-like) complexes with MdSFBB and MdCUL1 in pollen of apple. We discuss the putative functions of the two types of SCF SFBB

Expression Patterns of MdSSK1 and MdCUL1s
RT-PCR analysis revealed that MdSSK1 was preferentially expressed in pollen ( Figure 3A). Using 25 cycles of PCR amplification, MdSSK1 seemed to be specifically expressed in pollen, and with 30 cycles, signals of MdSSK1 were observed strongly in pollen, but weakly in other organs analyzed. RT-PCR analysis showed that MdCUL1s were expressed in all organs analyzed ( Figure 3A). To compare the expression levels of MdSSK1 and MdSBP1 in pollen, absolute qRT-PCR was performed. The result showed that MdSSK1 transcript abundance was significantly (.100 times) higher than that of MdSBP1 (P,0.05) ( Figure 3B).

Interactions of MdSFBB1-S 9 with MdSSK1 and MdSBP1
We examined the interaction of MdSFBB1-S 9 with MdSSK1 and MdSBP1 using an in vitro binding assay. In the tribe Pyreae of Rosaceae, many pollen S candidate genes (SFBB) were identified [13][14][15][16][17]36]. Among SFBB genes, the Pyrus pyrifolia SFBB1-S 4 (PpSFBB1-S 4 /S 4 F-box0) gene was most strongly supported as a pollen S by mutant analysis. S 4sm pollen lacking PpSFBB1-S 4 of a Japanese pear mutant 'Osa-Nijisseiki' [37] was rejected by the pistil harboring not only the self S 4 but also the non-self S 1 haplotype, suggesting that PpSFBB1-S 4 would be required for degradation of non-self S 1 -RNase [16,19]. Because MdSFBB1-S 9 is a probable ortholog of PpSFBB1-S 4 , we used MdSFBB1-S 9 for protein-protein interaction analyses. It was reported that the interaction between F-box protein and Skp1 of the SCF complex is mediated through the F-box motif of the F-box protein [38]; therefore, we used the part of the protein (amino acid residues 1-61), N-terminal region containing F-box motif of MdSFBB1-S 9 named MdSFBB1-S 9 -N, for the binding assay in addition to fulllength MdSFBB1-S 9 . MBP-fused MdSSK1 (MBP: MdSSK1), MBP-fused MdSBP1 (MBP: MdSBP1) and MBP (negative control) proteins were expressed in E. coli and reacted with amylose resin. The recombinant protein-bound beads were then incubated with a crude extract of E. coli expressing GST-fused and FLAG-tagged MdSFBB1-S 9 (GST: MdSFBB1-S 9 : FLAG) or GST-fused MdSFBB1-S 9 -N (GST: MdSFBB1-S 9 -N: FLAG). Eluted proteins were separated by SDS-PAGE and detected using anti-FLAG antibody. The results showed that both MdSSK1 and MdSBP1 interact with MdSFBB1-S 9 and MdSFBB1-S 9 -N ( Figure 4A). Because MdSFBB1-S 9 and MdSFBB1-S 9 -N seemed to interact more strongly with MdSSK1 than with MdSBP1, a competitive pull-down assay between the recombinant proteins was conducted. GST: MdSFBB1-S 9 : FLAG, GST: MdSFBB1-S 9 -N: FLAG and GST (negative control) proteins were reacted with Glutathione Sepharose 4B and incubated with a protein mixture of MBP: MdSSK1 and MBP: MdSBP1. The result revealed that MdSSK1 exhibits a stronger interaction affinity to MdSFBB1-S 9 and MdSFBB1-S 9 -N than MdSBP1 ( Figure 4B). Because MdSFBB1-S 9 -N seemed to interact more strongly with MdSSK1 and MdSBP1 than MdSFBB1-S 9 , this possibility was examined by a competitive pull-down assay. MBP: MdSSK1, MBP: MdSBP1 and MBP (negative control) proteins were reacted with amylose resin and incubated with an equal amount protein mixture of GST: MdSFBB1-S 9 : FLAG and GST: MdSFBB1-S 9 -N: FLAG. Taken into account the calculated molecular mass of GST: MdSFBB1-S 9 : FLAG and GST: MdSFBB1-S 9 -N: FLAG, 74 kDa and 35 kDa, respectively, 4.5 mg of GST: MdSFBB1-S 9 : FLAG and 2.1 mg of GST: MdSFBB1-S 9 -N: FLAG were used. The result showed that MdSFBB1-S 9 -Nhad higher affinity than MdSFBB1-S 9 for binding to MdSSK1 and MdSBP1 ( Figure 4C).

Interactions of MdCUL1s with MdSSK1 and MdSBP1
To examine the binding of MdCUL1s with MdSSK1 and MdSBP1, in vitro binding assays were conducted. GST: MdSSK1, GST: MdSBP1 and GST (negative control) proteins were reacted with Glutathione Sepharose 4B. MdCUL1A: FLAG and MdCUL1B: FLAG proteins were expressed using wheat germ extracts and incubated with the protein-bound Glutathione Sepharose 4B. The results showed that MdSSK1 interacts with MdCUL1A, but not with MdCUL1B, whereas, MdSBP1 interacted with both MdCUL1A and MdCUL1B ( Figure 5A, B).

MdSSK1 and MdSBP1 may form Canonical and Noncanonical SCF Complexes, Respectively, with MdSFBB1 and MdCUL1 in Pollen of Apple
We isolated an SSK1 homolog (MdSSK1) of apple. MdSSK1 was preferentially expressed in pollen, but weakly in other organs analyzed, whereas, MdSBP1 was almost equally expressed in all organs analyzed [31]. The expression pattern of MdSBP1 was the same as Solanaceous SBP1 homologs [29,[39][40][41], suggesting that MdSBP1 may also be involved in general cellular functions besides pollination [31]. The expression pattern of MdSSK1 was slightly different from other SSK1 genes of Rosaceae, Solanaceae and Plantaginaceae. The SSK1 genes of Pyrus bretschneideri of Rosaceae, Petunia hybrida and Petunia inflata of Solanaceae and Antirrhinum hispanicum of Plantaginaceae were reported to show pollen or anther-specific expression patterns [24,25,27,42]. In Prunus of Rosaceae, PavSSK1 was expressed strongly in pollen and anthers, but weakly in styles, suggesting that PavSSK1 serves as an adaptor for not only PavSFB but also PavSLFL1 expressed in pollen, anthers and styles [26]. The expression pattern of MdSSK1 suggests that MdSSK1 mainly, but not exclusively, functions in pollen.
In vitro binding assays revealed that both MdSSK1 and MdSBP1 interact with MdSFBB1-S 9 and MdSFBB1-S 9 -N. MdSSK1 and MdSBP1 interacted more strongly with MdSFBB1-S 9 -N than with MdSFBB1-S 9 . Given that MdSFBB1-S 9 -N almost corresponds to the F-box domain, this is possibly because MdSSK1 and MdSBP1 interact with MdSFBB1-S 9 through the F-box domain of MdSFBB1-S 9 , and the conformation of bacterially expressed MdSFBB1-S 9 was different from that of the native state, affecting binding of the F-box motif with SSK and SBP1. The interaction between the F-box protein and Skp1 of the SCF complex is known to be mediated through the F-box motif [38]. The finding that a truncated SLF of Petunia inflata (PiSLF 2 ) without the F-box domain expressed in S 2 S 3 plants did not cause breakdown of SI in S 3 pollen [42], whereas, the full-length PiSLF 2 did [43], suggests that the Fbox domain of PiSLF 2 is required for GSI [42]. The F-box domain of PiSLF 2 was reported to interact with PiSBP1 in a yeast twohybrid assay [42]. These findings are consistent with our data in apple proteins that the F-box domain of MdSFBB1-S 9 may be important for binding to MdSSK1 and MdSBP1.
MdSSK1 interacted with MdCUL1A but not with MdCUL1B, whereas, MdSBP1 interacted with both MdCUL1A and MdCUL1B by in vitro binding assays. The N-terminal domain (NTD) of human CUL1 is reported to bind the human Skp1 [33]. The C-terminal domains (CTDs) of MdCUL1A and MdCUL1B are fairly well conserved, however, their NTDs are very different ( Figure S4). Divergence at the NTDs of MdCUL1A and MdCUL1B may be responsible for the difference in the affinities of the proteins with MdSSK1. A pollen-expressed CUL1 gene of Solanum is considered to be involved in unilateral interspecific incompatibility (UI) and SI [44,45]. Generally, UI occurs in crosses between self-incompatible (SI) and self-compatible (SC) species. Pollen of SI species are compatible with the SC pistil, but not vice versa (SI 6SC rule) [46]. The pollen of SI species of Solanum express functional CUL1 genes, whereas, SC species shared the same loss-of-function mutations, even though the pollen fertilities of SC species are normal [44]. CUL1-reduced pollen of transgenic plants obtained by introducing LAT52-CUL1-RNAi to SI S. arcanum was selectively eliminated on non-transgenic SI pistils, but it was not rejected on S-RNase-deficient SC pistils [45]. The results suggest that the functions of CUL1 genes of Solanum species might have diverged evolutionarily, and SI species of Solanum shared CUL1 specialized for degradation of S-RNases in addition to Putative Functions of MdSSK1-containing SCF SFBB and MdSBP1-containing SCF SFBB -like Complexes in Apple The results of protein-protein interaction analyses suggest that MdSSK1 and MdSBP1 form canonical and noncanonical SCFlike complexes, respectively, with MdSFBB1-S 9 and MdCUL1 within pollen of apple; however, the functions of the two SCF SFBB1 (-like) complexes in GSI of apple are unclear at present. In Petunia inflata, PiSBP1 interacted with PiSLF1-S 2 and PiCUL1-G, suggesting that PiSBP1 would be a component of a noncanonical E3 ligase complex, which interacts with non-self S-RNases to ubiquitinate them for degradation [29]. Structural similarity of PiSBP1 and the apple homolog MdSBP1 [31], together with the results of in vitro binding assays of this study, may suggest that MdSBP1 forms noncanonical SCF complex like PiSBP1. In Petunia hybrida, functional analyses of PhSSK1 using RNAi plants (LAT52-PhSSK1-RNAi) revealed that a substantial reduction of PhSSK1 in transgenic pollen reduced cross-pollen compatibility, although the transgenic plants retained SI [25],   suggesting that the PhSSK1-containing SCF complex is involved in degradation of non-self S-RNases. MdSSK1-and MdSBP1containing SCF SFBB1 (-like) complexes may also be involved in the degradation of non-self S-RNases in pollen of apple. In vitro binding assays showed that MdSFBB1-S 9 and MdSFBB1-S 9 -N interacted more strongly with MdSSK1 than with MdSBP1, suggesting that, in apple pollen, MdSSK1-containing SCF SFBB complexes plays a major role, which is likely to be in the degradation of non-self S-RNases. The higher abundance of MdSSK1 transcripts than MdSBP1 transcripts would also support the idea. Recent co-immunoprecipitation assays using pollen extracts of Petunia inflata detected PiSSK1 but not PiSBP1 as the co-purified protein with PiSLF [28]. Like in the case of apple, this is possibly because PiSSK1 have higher affinity than PiSBP1 for binding to PiSLF, and/or PiSSK1 is more abundant than PiSBP1 in pollen.
Recent studies provide evidence that SBP1 may have other functions besides self/non-self discrimination in S-RNase-based GSI. SBP1 of Petunia hybrida, PhSBP1, could be a candidate for the non-allele-specific inhibitor of all S-RNases because it showed no polymorphism in different S alleles [30,39]. SBP1 of Nicotiana alata, NaSBP1, was reported to interact with the C-terminal domain of pistil arabinogalactan proteins (AGPs), transmitting tract-specific glycoprotein (TTS) and 120-kDa glycoprotein (120K), suggesting that binding between NaSBP1 and the pistil AGPs may contribute to signaling and trafficking processes inside pollen tubes [41]. MdSBP1 and solanaceous SBP1 homologs were expressed in all tissues examined, and the proteins included the same proteinprotein interaction domains, RING-HC finger motif and coiledcoil region [31], suggesting that MdSBP1 may also function besides self/non-self discrimination like as SBP1 homologs of Solanaceae.
In S-RNase-based GSI, two different systems are proposed, 'collaborative non-self recognition system by multiple factors' for Solanaceae [18] and rosaceous tribe Pyreae [16,19], and 'self recognition system by a single factor' for Prunus of Rosaceae [16,20,21]. The 'collaborative non-self recognition system by multiple factors' is consistent with 'competitive interaction (CI)' known to be a phenomenon that coexistence of different pollen S alleles in a pollen grain causes breakdown of pollen SI function [43,47]. For example, a tetraploid plant with S 1 S 1 S 2 S 2 genotype produces three S-genotypes of pollen (S 1 S 1 , S 2 S 2 and S 1 S 2 ), and S 1 S 1 and S 2 S 2 pollen are rejected by the self pistil, but S 1 S 2 pollen is accepted. In S 1 S 2 heteroallelic pollen, pollen S 1 and S 2 proteins would target non-self S 2 -RNase and S 1 -RNase, respectively. The CI phenomenon was reported in the tribe Pyreae of Rosaceae, Petunia of Solanaceae and Antirrhinum of Plantaginaceae, but not in Prunus of Rosaceae [6,43,[48][49][50][51][52]. In Prunus, most pollen-part selfcompatibility (SC) mutants encode a truncated SFB protein or lack the SFB gene [20,21,[53][54][55][56]. These findings suggest that the species of Prunus exhibit the 'self recognition system by a single factor' [16,20,21]. This model postulates that, in Prunus, non-self S-RNase is inactivated by an unidentified 'general inhibitor', while self S-RNase is protected by SFB. The protected self S-RNase would degrade RNA in a self pollen tube to prevent growth. It seems that the pollen S functions of the tribe Pyreae and Prunus of Rosaceae are different, although SSK1 homologs of the tribe Pyreae and Prunus are suggested to form similar SCF SFBB/SFB complexes [26,27]. Further biochemical characterization and comparative analyses of the functions of SSK1-and SBP1containing SCF(-like) complexes in S-RNase-based GSI plants would shed light on the difference in the two self/non-self recognition systems of S-RNase-based GSI.

Plant Materials
Leaves and floral organs of apple cultivar 'Kitaro' (S 3 S 9 ) were collected in spring, frozen in liquid nitrogen, and stored at 280uC until use.

Isolation of cDNA Sequences
RNA was isolated from the leaves and floral organs of apple as described by [10]. Total RNA samples were treated with DNaseI (Nippongene). cDNA was synthesized from the treated RNA as described by [10], and used for RT-PCR.
The amino acid identities among SSK1 or CUL1 proteins were analyzed using GENETYX-MAC (version 17; Genetyx). The amino acid sequences of SSK1 or CUL1 proteins were aligned using Clustal W [57]. A neighbor-joining tree was constructed [58] based on the alignment using MEGA ver. 5.05. [59].

RT-PCR and Quantitative Real-time PCR (qRT-PCR)
The expression levels of MdSSK1, MdCUL1A and MdCUL1B were analyzed by RT-PCR with gene-specific primers for MdSSK1 Data were collected using ABI PRISM 7000 sequence detection system (Applied Biosystems) in accordance with the instruction manual. The cDNA sequences of the two genes were cloned into vector pEU3-NII (Toyobo). The plasmid DNA containing the two genes was used to generate standard curves for absolute quantification. C T values for each sample were converted into absolute copy numbers (x) using the standard curves (x = (y intercept -C T )/slope).
To produce MBP-fused MdSBP1 protein, the full-length coding sequence of MdSBP1 was cloned into vector pColdIIMBP [31] to construct pColdMBPMdSBP1. The BamHI-XbaI fragment of MdSBP1 released from the pColdMBPMdSBP1 construct was cloned into vector pColdIVGST to make pColdGSTMdSBP1 for expression of GST-fused MdSBP1 protein (GST: MdSBP1).

Pull-down Assays
Constructs, except for pEU3MdCUL1AFLAG and pEU3Md-CUL1BFLAG, were introduced into BL21 (DE3) pLysS (Novagen) and cultured and induced as described in [31]. pColdIIMBP and pColdIVGST were also transferred to BL21 (DE3) pLysS for the expression of MBP and GST, respectively, as negative controls in the pull-down assay. MBP: MdSSK1, MBP: MdSBP1 and MBP were extracted from bacteria by sonication, and reacted with amylose resin (New England BioLabs) in binding buffer [31]. Crude proteins of GST: MdSFBB1-S 9 : FLAG and GST: MdSFBB1-S 9 -N: FLAG were extracted from bacteria and incubated with protein-bound amylose resin at 4uC for 2 hours. For a competitive pull-down assay of MdSSK1 and MdSBP1 with MdSFBB1-S 9 and MdSFBB1-S 9 -N, a recombinant protein mixture of GST: MdSFBB1-S 9 : FLAG and GST: MdSFBB1-S 9 -N: FLAG were incubated with protein-bound amylose resin. Taken into account the calculated molecular mass of GST: MdSFBB1-S 9 : FLAG and GST: MdSFBB1-S 9 -N: FLAG, 74 kDa and 35 kDa, respectively, 4.5 mg of GST: MdSFBB1-S 9 : FLAG and 2.1 mg of GST: MdSFBB1-S 9 -N: FLAG were used. The beads were washed five times with washing buffer [31], and the proteins were eluted from the beads using maltose-containing native elution buffer (20 mM Tris-HCl pH 7.5, 0.2 M NaCl, 1 mM EDTA, 10 mM maltose). The eluted proteins were separated by SDS-PAGE and detected using an anti-FLAG M2 monoclonal antibody (SIGMA).
For the next competitive pull-down assay of MdSFBB1-S 9 and MdSFBB1-S 9 -N with MdSSK1 and MdSBP1, GST: MdSFBB1-S 9 : FLAG and GST: MdSFBB1-S 9 -N: FLAG were reacted with Glutathione Sepharose 4B (GE Healthcare). Equal amounts of recombinant protein mixture of MBP: MdSSK1 (15 mg) and MBP: MdSBP1 (15 mg) were incubated with protein-bound Glutathione Sepharose 4B. The beads were washed five times with washing buffer [31], and the proteins were eluted from the beads using glutathione-containing native elution buffer (50 mM Tris-HCl pH 8.0, 10 mM reduced glutathione). The eluted proteins were separated by SDS-PAGE and detected using an anti-MBP monoclonal antibody (HRP-conjugated) (New England BioLabs).