A PERK-Like Receptor Kinase Interacts with the Geminivirus Nuclear Shuttle Protein and Potentiates Viral Infection

ABSTRACT The nuclear shuttle protein (NSP) from bipartite geminiviruses facilitates the intracellular transport of viral DNA from the nucleus to the cytoplasm and acts in concert with the movement protein (MP) to promote the cell-to-cell spread of the viral DNA. A proline-rich extensin-like receptor protein kinase (PERK) was found to interact specifically with NSP of Cabbage leaf curl virus (CaLCuV) and of tomato-infecting geminiviruses through a yeast two-hybrid screening. The PERK-like protein, which we designated NsAK (for NSP-associated kinase), is structurally organized into a proline-rich N-terminal domain, followed by a transmembrane segment and a C-terminal serine/threonine kinase domain. The viral protein interacted stably with defective versions of the NsAK kinase domain, but not with the potentially active enzyme, in an in vitro binding assay. In vitro-translated NsAK enhanced the phosphorylation level of NSP, indicating that NSP functions as a substrate for NsAK. These results demonstrate that NsAK is an authentic serine/threonine kinase and suggest a functional link for NSP-NsAK complex formation. This interpretation was corroborated by in vivo infectivity assays showing that loss of NsAK function reduces the efficiency of CaLCuV infection and attenuates symptom development. Our data implicate NsAK as a positive contributor to geminivirus infection and suggest it may regulate NSP function.

The nuclear shuttle protein (NSP) from bipartite geminiviruses facilitates the intracellular transport of viral DNA from the nucleus to the cytoplasm and acts in concert with the movement protein (MP) to promote the cell-to-cell spread of the viral DNA. A proline-rich extensin-like receptor protein kinase (PERK) was found to interact specifically with NSP of Cabbage leaf curl virus (CaLCuV) and of tomato-infecting geminiviruses through a yeast two-hybrid screening. The PERK-like protein, which we designated NsAK (for NSP-associated kinase), is structurally organized into a proline-rich N-terminal domain, followed by a transmembrane segment and a C-terminal serine/threonine kinase domain. The viral protein interacted stably with defective versions of the NsAK kinase domain, but not with the potentially active enzyme, in an in vitro binding assay.

In vitro-translated NsAK enhanced the phosphorylation level of NSP, indicating that NSP functions as a substrate for NsAK. These results demonstrate that NsAK is an authentic serine/threonine kinase and suggest a functional link for NSP-NsAK complex formation. This interpretation was corroborated by in vivo infectivity assays showing that loss of NsAK function reduces the efficiency of CaLCuV infection and attenuates symptom development. Our data implicate NsAK as a positive contributor to geminivirus infection and suggest it may regulate NSP function.
The family Geminiviridae represents a group of plant DNA viruses characterized by a single-stranded circular genome and unique twinned icosahedral particles (17,35). Species of this family are taxonomically grouped into four genera according to the insect vector, host range, and genomic organization, which can be in either single or double configuration. In the bipartite geminiviruses, the virus-encoded replication, encapsidation, and movement functions are segregated into two genomic components, designated DNA-A and DNA-B. DNA-A has the potential to code for five gene products (AV1, AC1, AC2, AC3, and AC4), which are involved in replication of the viral genome (AC1, or Rep, and AC3, or REn), transactivation of viral genes (AC2, or TrAP), and encapsidation of the viral genome (AV1, or CP) (9,10,11,23,44,45). The DNA-B component encodes two proteins, the nuclear shuttle protein (NSP, or BV1) and the movement protein (MP, or BC1), which act in a cooperative manner to mediate the intra-and intercellular trafficking of viral DNA (14,26).
Like any other plant virus, geminiviruses rely on the host cellular machinery to perform their basic replication and movement functions toward the establishment of a productive infection. Accordingly, the geminivirus infection cycle has been shown to depend extensively on virus-host intermolecular interactions, which are required either for basic compatibility (1,3,4,6,24,25,27,29,40,41) or for modulation of virus infection by subverting defense responses (13,19,28,48,49). The discovery of each new viral-protein-interacting host factor has provided insights into the molecular bases of pathogenicity (51). In the case of geminiviruses, their potential as a model system to elucidate basic cellular components of the plant replication and transport machinery has become increasingly apparent (18,26,35). Particularly interesting is the unique aspect of the geminivirus infection cycle that requires a movement function to facilitate the transport of viral DNA from the nucleus, where viral replication takes place, to the cytoplasm (14,26).
Our understanding of geminivirus movement in infected plants has advanced with the genetic and biochemical characterization of the movement proteins from Squash leaf curl virus (SLCV) and Bean dwarf mosaic virus. From these investigations, it has been conceptually established that NSP shuttles viral DNA between the nucleus and the cytoplasm, whereas MP is predominantly involved in mediating the cell-to-cell movement of viral DNA across the wall (14). However, two distinct mechanisms for cell-to-cell trafficking of viral DNA, which basically differ in the nature of the viral DNA complex transported into the adjacent uninfected cells, are supported by the relevant data (30,37). In the first model, NSP facilitates the intracellular movement of the viral genome from the nucleus to the cytoplasm, where it is replaced by MP, which transports the viral DNA to adjacent cells via plasmodesmata (30,34). The second model, as in the case of the phloemlimited SLCV, holds that MP facilitates NSP-mediated intracellular transport of viral DNA from the nucleus to the cytoplasm and then mediates the transport of the NSP-DNA complex to adjacent cells via endoplasmic-reticulum-derived tubules induced by the viral infection (20,36,37,38,50). Regardless of the mechanism of geminivirus cell-to-cell trafficking, the fundamental role of NSP in virus movement pre-dicts that this viral protein may interact with host factors in different subcellular compartments. Accordingly, NSP has been shown to interact with an Arabidopsis thaliana nuclear acetylase, designated nuclear shuttle protein interactor (AtNSI), as well as with plasma membrane receptor kinases, designated NIKs (NSP-interacting kinases), from tomato, soybean, and Arabidopsis (28,29). In Arabidopsis, NSP interacts with three members of the LRR-receptor-like kinase (RLK) family, NIK1, NIK2, and NIK3, which have been shown to be authentic serine/threonine kinases with biochemical properties consistent with a receptorsignaling function (13).
Despite the interaction of NSP with acetylase and receptorlike kinases, NSP does not function as a substrate for either of these host enzymes. In fact, AtNSI does not acetylate NSP, but rather, it acetylates CP (29). During rolling-circle replication of viral DNA, CP is thought to mediate a single-stranded DNA (ssDNA) partition by sequestering viral ssDNAs away from the replication pool and making them available for NSP binding and export. The cycling of CP between the sequestered ssDNA pool and the replication pool has been proposed to be modulated by AtNSI through interaction with NSP (29). Consistent with this model, overexpression of AtNSI enhances Cabbage leaf curl virus (CaLCuV) infection, and disruption of the NSP capacity to bind AtNSI generates a mutant virus defective for infection (4,29). In the case of the receptor-like kinase NIK, NSP inhibits its kinase activity to suppress a NIK-mediated antiviral response (13). Accordingly, inactivation of NIK genes increases the susceptibility of Arabidopsis mutants to virus infection (13). Here, we extend the characterization of the putative basic network of NSP-interacting host proteins and provide evidence for the biological significance of complex formation. We show that NSP also interacts with a proline-rich extensin-like receptor kinase (PERK)-like protein from Arabidopsis that belongs to the receptor-like serine/threonine kinase superfamily and is here referred to as NsAK (for NSP-associated kinase). NSP-NsAK complex formation is not virus specific, because NSPs from Tomato golden mosaic virus (TGMV) and Tomato crinkle leaf yellows virus (TCrLYV) (accession number AY090556) also interact with the NsAK protein. In contrast to the receptor-like kinase NIK, the NsAK protein seems to potentiate infectivity and may be a positive regulator of NSP function, as disruption of NsAK expression attenuates virus infection.

MATERIALS AND METHODS
Yeast two-hybrid screen. An Arabidopsis thaliana cDNA library was prepared from mRNA isolated from aerial tissues and fused to the GAL4 activation domain in the Leu Ϫ pEXAD502 vector (Invitrogen Life Technologies, Inc.). The pBD-NSPCLCV clone, which contains the GAL4 DNA-binding domain fused to NSP sequences from CaLCuV, has been previously described (13). The yeast reporter strain MaV203 (MAT␣ leu2-3,112 trp1-901 his3200 ade2-101 gal4 gal80 SPAL10::URA3 GAL1::lacZ HIS3UAS GAL1::HIS3-LYS2 can1R cyh2R) is deficient in producing tryptophan, leucine, and uracil (Trp Ϫ Leu Ϫ Ura Ϫ ). Mav203 cells were transformed sequentially with pBD-NSPCLCV and 25 g of pEXAD502 cDNA libraries, along with 3 mg of salmon sperm carrier DNA, using the lithium acetate/polyethylene glycol method. Transformants were plated on synthetic dropout medium lacking Trp, Leu, Ura, and His but supplemented with 25 mM 3-aminotriazole and cultured for 3 to 5 days at 30°C. The interactions were further confirmed by measuring ␤-galactosidase activities from yeast extracts with o-nitrophenyl ␤-D-galactopyranoside, as described previously (46). Approximately 5 ϫ 10 6 transformants were obtained, as estimated based on the number of transformants growing on the synthetic-dropout-Trp Ϫ Leu Ϫ plate.
Plasmid DNA was recovered from yeast and transformed into Escherichia coli strain XL-1 Blue (Stratagene) by electroporation.
Plasmids. A full-length NsAK cDNA (U17522) was obtained from the Arabidopsis Biological Resource Center. All the other recombinant plasmids were obtained through the GATEWAY system (Invitrogen). Briefly, the specified DNA fragments were amplified by PCR with appropriate extensions and introduced by recombination into the entry vector pDONR201 and then transferred to the appropriate destination vector. The full-length NsAK cDNA, the NsAK C-terminal kinase domain (KD) (encoding amino acids 238 to 652), and its truncated version (⌬KD 392-652 ) were amplified by PCR from U17522 cDNA and cloned into pDONR201. The resulting clones, pDON-NsAK, pDON-KDNsAK, and pDON-⌬KDNsAK, respectively, were used to transfer the respective fragments into expression vectors.
For in vitro transcription and translation of proteins, the NsAK coding region and its KD 238-652 and ⌬KD 392-652 fragments were transferred from the entry vector to the T7 RNA polymerase-dependent transcription vector pDEST14 to generate pH-NsAK, pH-KDNsAK, and pH-⌬KDNsAK, respectively. Glutathione S-transferase (GST)-fused and His-tagged truncated NsAKs were generated by transferring KD and ⌬KD sequences (as defined above) from pDONR201 to the bacterial expression vector pDEST15 (GST fusions) or pDEST17 (His tags).
The NsAK coding region was also transferred from the entry vector to the binary vector pK7WG2 to yield pK7-NsAK, which harbors the NsAK cDNA under the control of the Cauliflower mosaic virus 35S promoter. The NsAK coding region was also cloned into pK7FWG2 and pCTAPi via recombination. The resulting constructs, pK7F-NsAK and pK7C-NsAK, respectively, contained a GFP gene or a ProtA tag domain, TAP (32), fused in frame after the last codon of NsAK cDNA, under the control of the Cauliflower mosaic virus 35S promoter.
In vitro protein-protein binding assay. To express intact and truncated versions of NsAK in vitro, 1 g of recombinant plasmid containing the appropriate insert in pDEST14 (as described above) was used in an in vitro transcription and translation system supplemented with [ 35 S]methionine and T7 RNA polymerase (Promega). The 35 S-labeled proteins were incubated for 1 h at 4°C with 50 l of glutathione-Sepharose beads to which purified GST or GST-NSP had been adsorbed. The beads were pelleted by centrifugation and washed five times with 1 ml of 50 mM Tris-HCl, pH 7.5, 120 mM NaCl, 0.1% Nonidet P-40. Bound proteins were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on an 8 to 20% polyacrylamide gel and visualized by fluorography.
Purification of GST-NSP fusion proteins. The expression plasmid pGST-NSPCLCV (13), containing the GST sequence fused to the NSP coding region from CaLCuV, was transformed into E. coli strain BL21, and the synthesis of the recombinant protein was induced by 0.5 mM isopropyl-␤-D-thiogalactopyranoside for 12 h at 20°C. The recombinant GST-NSP fusion protein was affinity purified using GST-Sepharose beads (Amersham), according to the manufacturer's instructions.
Phosphorylation assay. For the phosphorylation assay, 2 g of recombinant plasmid containing the appropriate insert in pDEST14 (as described above) was used in an in vitro transcription and translation system supplemented with T7 RNA polymerase (Promega). An aliquot (15 l) of the reaction mixture was incubated alone or with GST-NSP or GST for 30 min at 25°C in 30 l of kinase buffer containing 18 mM HEPES (pH 7.4), 10 mM MgCl 2 , 10 mM MnSO 4 , 1 mM dithiothreitol, 10 M ATP, and 5 Ci [␥-32 P]ATP. Phosphoproteins were resolved by SDS-PAGE. The gel was stained with Coomassie brilliant blue to verify protein loading, dried, and subjected to autoradiography. Incorporated radioactivity in protein bands was quantified by phosphorimaging. The highest-molecular-mass phosphoprotein band was used as an internal control to normalize the radioactive signals.
RT-PCR. Total RNA was extracted from the plant tissues using TRIzol (Invitrogen). Reverse transcription (RT)-PCR assays were performed with NsAKor NSP-specific primers as described previously (5). PCR was also carried out with nucleoporin-like gene-specific primers to assess the quantity and quality of the cDNA. The PCR comprised 30 cycles of 30 s at 94°C, 1 min at 55°C, and 2 min at 72°C.
Plant material, growth conditions, and genotyping. The Columbia (Col-0) ecotype of Arabidopsis thaliana was used as the wild type. The nsak mutant was from the SALK Institute (SALK_008504) (2). Seeds were surface sterilized and cold treated at 4°C for 2 days in the dark and then exposed to white light. Seedlings were grown at 22°C on plates containing Murashige-Skoog medium for 3 weeks and then transferred to soil. Plants were grown in a growth chamber at NsAK-Rvs (5Ј-CCTCCAATGGCGATTCCTACC-3Ј) primers and, for detection of the nsak alleles, the NsAK-Fwd primer was used in combination with the SALK LBa1 (5Ј-TGGTTCACGTAGTGGGCCATCG-3Ј) primer.

CaLCuV inoculation and analysis of infected plants.
Arabidopsis thaliana plants at the seven-leaf stage were inoculated with plasmids containing partial tandem repeats of CaLCuV DNA-A and DNA-B by biolistic delivery as de-scribed previously (39). Total nucleic acid was extracted from systemically infected leaves, and viral DNA was detected by PCR with DNA-A-or DNA-Bspecific primers (33).

Identification of a PERK-like kinase domain that interacts with NSP.
To identify host proteins that interact with NSP, we performed yeast two-hybrid screens and identified several NSP-interacting proteins, including the NIKs already described (13,28). We describe here the characterization of NsAK, a PERK-like kinase with a transmembrane receptor configuration (Fig. 1). From a total of 5 ϫ 10 6 independent double transformants that were assayed for Ura and His prototrophy and ␤-galactosidase activity, three positive clones were found that encoded a truncated kinase domain from the Arabidopsis gene At3g24550, designated NsAK (NSP-associated kinase). The deduced NsAK protein contains 652 amino acid residues (calculated M r , 69,271; pI, 8.51), and it is structurally related to the RLK superfamily. Its receptor configuration consists of an N-terminal proline-rich region, followed by a putative hydrophobic membrane-spanning stretch of 23 amino acid residues and a C-terminal serine/threonine kinase catalytic domain with all 11 subdomains typical of the eukaryotic protein kinase family (Fig. 1). Based on the proline-rich modular organization of its N-terminal domain, NsAK falls into the PERK-like subfamily of the RLK family (42). Among the members of the PERK-like RLK subfamily from Arabidopsis, NsAK is most closely related to the At3g24400-encoded product (63% identity), followed by At3g24540 (62%) and At1g52290 (54%), whose functions are unknown. Extending the analysis to general plant proteins, NsAK was found to be most closely related to the Brassica napus protein PERK1 (BsPERK1; GenBank accession number AY028699; 85% sequence identity) (Fig. 1). It also shares significant conservation of primary structure with the Oryza sativa homolog (Os-PERK1; GenBank accession number NP_913464; 57% identity). Both proteins belong to the PERK-like class of putative receptor-like serine/threonine protein kinases of unknown function.
We also evaluated the capacity of NsAK to interact with NSP from the tomato-infecting geminiviruses TGMV and TCrLYV by the two-hybrid system. The interactions were monitored for histidine and uracil prototrophy and further confirmed by determining ␤-galactosidase activity in yeast protein extracts ( Table 1). The truncated kinase domain of NsAK also interacted with NSP from TGMV and TCrLYV, indicating that NsAK-NSP complex formation is not virus specific.
NSP interacts stably with an inactive kinase domain of NsAK in vitro and may serve as an NsAK substrate. Specific interaction between NSP and NsAK was also assayed by GST pull-down assays. Purified GST-NSP or GST was incubated with the in vitro-translated [ 35 S]Met-labeled intact NsAK, as well as its KD or the inactive kinase domain with the nucleo-tide binding site deleted (⌬KD), as indicated in Fig. 2. The resulting complexes were isolated on glutathione-Sepharose beads. While the C-terminal ⌬KD-NsAK, in which the critical serine/threonine kinase subdomains I and II were deleted, interacted stably and strongly with NSP (lane 8), interaction of the viral protein with either the intact NsAK protein (lane 2) or its potentially active kinase domain (lane 5) was barely detected by our in vitro assay. This scenario for NSP-NsAK complex formation resembles that of enzyme-substrate interaction under a high catalytic efficiency and may indicate that NsAK phosphorylates the viral protein. In this case, inability of the kinase domain to perform catalysis would trap the enzymesubstrate complex in a nonproductive interaction, increasing the stability of the complex formation, as in the case of the ⌬KDNsAK truncated protein. Consistent with this notion, truncated NsAK polypeptides, which were also produced by the in vitro-transcribed and -translated reaction mixtures of KDNsAK and ⌬KDNsAK, bound to GST-NSP (lanes 5 and 8), but not to GST alone, indicating that NSP specifically and stably interacted with defective versions of truncated NsAK, but not with the potentially active NsAK kinase domain.
To directly assess the capacity of NsAK to phosphorylate NSP, we attempted to express the NsAK kinase domain in E. coli using several bacterial expression systems. We also tried to overexpress NsAK in Arabidopsis both as intact and as GFPand TAP-tagged fusion proteins. Except for the ⌬KD-NsAK inactive truncated kinase domain, which was produced in E. coli, all the potentially active recombinant kinases failed to be expressed either in E. coli or in plants. Given that both fulllength NsAK and its kinase domain were efficiently transcribed and translated in vitro (Fig. 2), the transcription/translation system was then adjusted to perform in vitro phosphorylation assays (Fig. 3A). Several endogenous reticulocyte proteins were phosphorylated by the reticulocyte kinases (lane 1), and so was the GST-NSP fusion protein (lane 4), which displayed a threefold-increased level of 32 P labeling compared to the GST-NSP-comigrating 52-kDa-band background (Fig. 3B, compare  bars 1 and 4). Nevertheless, inclusion of in vitro-transcribed and -translated NsAK in the phosphorylation reaction promoted a threefold increase in the 32 P labeling of GST-NSP (bar 2) compared to the reticulocyte kinase-phosphorylated GST-NSP signal (bar 4). Likewise, in the presence of KDNsAK, the GST-NSP-incorporated radioactive signal was twice as high as the reticulocyte kinase-phosphorylated GST-NSP signal (bars 4 and 5) and 14-fold higher than the corresponding 52-kDa-band background (bar 1). The incorporation The bait proteins were expressed as GAL4 DNA-binding domain fusions, and the prey proteins were expressed as GAL4 activation domain fusions in yeast. ⌬KD corresponds to the truncated C-terminal kinase domain of NsAK. SBP is an unrelated sucrose-binding protein from soybean used as a control in the two-hybrid assays. the GST-NSP-phosphorylated protein from the phosphorylation reaction mixture further confirmed these results (Fig. 3C).

The NsAK transcripts are expressed ubiquitously in Arabidopsis.
To determine the expression pattern of NsAK mRNA, we performed semiquantitative RT-PCR with total RNA from various Arabidopsis tissues (Fig. 4A). The NsAK mRNA was detected in all tissues investigated, and its steady-state level showed only minor variations among the different tissues. When normalized to the control band, it appeared that NsAK mRNA accumulation predominated in flowers. All other tissues showed intermediate mRNA levels.
Recently, it has been demonstrated that the geminivirus protein REn induces expression of its partner, a NAC protein (40). It was of interest to determine the effect that NSP expression might have on NsAK transcript accumulation. Total RNA from leaves of NSP-expressing Arabidopsis was analyzed by semiquantitative RT-PCR (Fig. 4B). Expression of NSP from CaLCuV (lanes 2 and 3) or TCrLYV (lanes 4, 5, and 6) did not induce NsAK transcript accumulation in transgenic leaves. Likewise, geminivirus infection had no apparent effect on NsAK expression in Arabidopsis leaves and roots (Fig. 4C).

Inactivation of the NsAK gene results in attenuation of symptoms and reduces the efficiency of viral infection.
To examine directly the biological significance of NAK-NSP interactions in vivo, we identified a transfer DNA (T-DNA) insertional mutation in the At3g24540 (NAK) gene (Fig. 5A). RT-PCR was performed on leaf RNA samples from wild-type (Col-0) and nsak knockout (KO) lines. With the gene-specific primers, we detected no accumulation of the corresponding transcript in the respective homozygous T-DNA insertion mutant, confirming that it was a null allele (Fig. 5B).
Wild-type Col-0 and nsak mutant plants were inoculated with CaLCuV DNA-A and DNA-B. Both Col-0 and nsak KO lines developed typical CaLCuV symptoms with distinct intensities. In fact, disease symptoms varied in severity from extreme stunting with severe chlorosis in Col-0 lines to mild stunting with epinasty and moderate chlorosis in nsak lines (Fig. 6A). The accumulation of viral DNA was detected in all symptomatic plants (Fig. 6B). The attenuated infection in nsak KO lines was reflected during the course of infection. As judged by symptom appearance, inactivation of NsAK alleles reduced the efficiency of virus infection in comparison with Col-0 plants that displayed a higher infection rate (Fig. 6C). In other experiments, the infectivity data, expressed as the number of days postinoculation required to get 50% infected plants, further confirmed the results (Fig. 6D).

DISCUSSION
Begomovirus NSP shuttles the viral DNA between the nucleus and the cytoplasm and cooperates with MP to move the viral DNA cell to cell across the wall. This facilitated trafficking of NSP indicates that the viral protein interacts extensively with host factors to usurp the endogenous host transport machinery. We report here the identification of NsAK, a novel cellular interaction partner for NSP. Interaction of an inactive kinase domain of NsAK was identified in the yeast two-hybrid system and further substantiated in an in vitro interaction assay. The latter observation argues for a direct interaction between the two proteins. However, we were unable to detect in vitro a stable interaction between NSP and an intact NsAK protein. The simplest explanation for the failure to detect interaction between the intact proteins is that the in vitrosynthesized active NsAK folded in a closed conformation that prevented NSP binding. Although we were unable to completely rule out this possibility, we provided two lines of evidence that argue against this hypothesis. First, disruption of NsAK expression indicated that NsAK is required for maximal viral infectivity. This observation suggests that any conformational constraint on NsAK function would be alleviated upon infection. Furthermore, the in vitro-transcribed and -translated NsAK was able to phosphorylate NSP, confirming that binding between the proteins might have occurred. More likely, a high catalytic efficiency (K cat /K m ) of NsAK with NSP as a substrate prevented the formation of a sufficiently stable enzyme-substrate complex to be detected in our assay. Among many other possible reasons, this might also explain why we were not able to demonstrate an interaction in vivo by a classical coimmunoprecipitation assay. NSP has also been shown to bind specifically to three members of the RLLII-RLK family, designated NIK, through their serine/threonine kinase domains. However, it binds to both active and defective kinases, and it is not phosphorylated by the proteins, but rather, it inhibits the kinase activity (13). Loss of NIK gene function enhances susceptibility to geminivirus infection, suggesting that NSP acts as a virulence factor to suppress NIK-mediated antiviral defenses. Our results suggest that binding of NSP to the NsAK kinase domain results in a productive enzyme-substrate interaction that, in contrast to the NSP-NIK interaction, may regulate NSP function. Consistent with this hypothesis, NSP from SLCV has been shown to be posttranslationally modified by phosphorylation (31). In addition, we provided evidence for an in vivo functional link between NSP and NsAK, as loss of NsAK function led to attenuation of viral infection. This nsak mutant phenotype implicates NsAK as a positive contributor to geminivirus infection, which most likely acts through regulation of NSP function. Cell-to-cell trafficking of plant viruses has been demonstrated to be regulated by the phosphorylation state of viral MPs that are phosphorylated by plasmodesma-associated protein kinases (7,47). In the case of NSP, however, as a facilitator of both intracellular and MP-mediated intercellular transport of viral DNA, a phosphorylation event appears equally likely to occur in the nucleus, nuclear envelope, cytoplasm, or plasma membrane. Thus, the determination of the precise localization of NsAK will provide valuable information about the functional significance of the interaction between NsAK and NSP. Protein sequence analysis revealed that NsAK belongs to the PERK-like RLK subfamily and, as such, is structurally organized into an N-terminal proline-rich domain and a serine/ threonine C-terminal domain. We provided evidence that NsAK is an authentic serine/threonine kinase, as the in vitrotranslated protein or its carboxyl domain enhances the phosphorylation level of NSP in an in vitro translation mixture background (Fig. 3). This is consistent with the structural-motif prediction for NsAK, as its C-terminal region contains all 11 of the conserved subdomains of protein kinases, in addition to specific signatures of serine/threonine kinases (8,16). These include the putative active site HrDvKssNxLLD in subdomain VIb; the DFG motif in subdomain VII, which may chelate Mg 2ϩ ions; and the highly conserved APE motif in subdomain VIII, which is assumed to be involved in the recognition of substrates and in autophosphorylation. As a member of the PERK-like RLK subfamily, the NsAK N-terminal portion is rich in proline and shares sequence similarity with the extensin family of cell wall proteins (42). The Brassica napus PERK-like RLR homolog was localized to the plasma membrane and has been shown to be rapidly induced by mechanical stresses (43). As a wound-induced and plasma membrane-localized member of the RLK family that has been fundamentally implicated in a wide range of signal transduction pathways, it has been proposed that the Brassica napus homolog is involved in transducing the wounding signal (42,43). Various wounding stimuli are often generated during the process of plant pathogen infection. This is particularly true in the case of the insect-transmitted geminiviruses. We found here, however, that the NsAK transcript level was not altered during the onset of geminivirus infection, although the biolistic method of inoculation generates a wound stimulus by mechanically injuring the leaves. The rather constitutive and ubiquitous expression of NsAK argues for a housekeeping function of this functional serine/threonine kinase.
Even though a complete knockout of nsak T-DNA insertional mutation resulted in attenuation of geminivirus infection, it did not totally block the infection, as would be expected if NsAK had an essential role in regulating NSP function. However, this result was not surprising, because the Arabidopsis genome encodes 14 NsAK homologs from the PERK-like RLK subfamily (42,43), and some of them might partially replace NsAK function during geminivirus infection. If such a strict NSP dependence on a specific member of a gene family in fact exists, it is reasonable to assume that NSP evolved to accommodate a certain degree of promiscuity to recruit the pivotal kinase activity to its function.
The specific interactions of NSP with members of the RLLII-RLK subfamily as a suppressor of kinase activity and antiviral defenses (13), and with a member of the PERK-like-RLK subfamily as a kinase substrate (this work), may indicate that the binding of NSP to serine/threonine kinase domains is an inherent property of the viral protein to cover both virulence and basic compatibility functions. Functionally, the binding of NSP to NIK1 (RLLII-RLK) antagonizes its binding to NsAK (PERK-like), as NSP inhibits the NIK kinase activity but acts as a NsAK substrate. The site of NSP binding to NIK1 was mapped to an 80-amino-acid-residue stretch that encompasses the putative active site for serine/threonine kinases (subdomain VIb, HrDvKssNxLLD) and the activation loop (subdomain VII, DFGAk/rx, plus subdomain VIII, GtxGyiaPEY) (13). As a potential substrate for NsAK, one may predict that NSP associates with NsAK through its substrate binding site, which conceptually overlaps the corresponding NSP-interacting region on NIK (16,22). Nevertheless, the presence of the conserved subdomains Vb, VII, and VIII per se does not fulfill the requirement for NSP-specific binding to kinases, because NSP does not interact stably with other receptor-like serine/ threonine kinases that contain these conserved subdomains, such as BRLI (for brassinosteroid insensitive 1) and SERK (for somatic embryogenesis receptor-like kinase) (13). More likely, discrete contacts with divergent residues of the catalytic domains of RLKs confer the required specificity for binding. While the identification of NsAK-NSP interaction sites would allow the construction of mutants with impaired binding capacities, the predicted functional redundancy of the PERKlike RLK gene family members and the antagonistic effect of NSP binding to NIK and to NsAK complicate the use of these mutant proteins to evaluate the significance of NsAK-NSP interaction in the geminivirus infection cycle. In contrast, the identification and subsequent targeting of the NSP phosphorylation sites in a mutagenesis-based approach will allow us to handle the NSP-NsAK interaction separately and hence to elucidate the role of phosphorylation in NSP function.