Mutation of Capsid Protein Phosphorylation Sites Abolishes Cauliflower Mosaic Virus Infectivity

Abstract

The cauliflower mosaic virus (CaMV) capsid protein is derived by bidirectional processing of the precapsid protein (CP56). We expressed several derivatives of CP56 in Escherichia coli and used them as substrates for virus-associated kinase and casein kinase II purified from plant cells. Three serine residues located at the N terminus of the mature viral protein CP44 were identified as phosphorylation targets. A mutation of one of them in the viral context had little or no effect on viral infectivity, but a mutation of all three serines abolished infectivity. The mapping of phosphorylation sites in CP44, but not CP39 or CP37, and immunodetection of the Zn finger motif in CP44 and CP39, but not CP37, support the model that CP39 is produced from CP44 by N-terminal processing and CP37 is produced from CP39 by C-terminal processing. We discuss the possible role of phosphorylation in the processing and assembly of CaMV capsid protein.

The structural proteins of retroviruses and pararetroviruses are modified by proteolysis, phosphorylation (5, 6), ubiquitination (23, 18), glycosylation (3, 19), meristylation, and other reactions. In many cases, these modifications have been shown to control assembly, genome packaging and release, and intracellular transport and distribution (2). The best studied of the plant pararetroviruses is cauliflower mosaic virus (CaMV). Its precapsid protein (CP56) is 489 amino acids long (Fig. 1A). Several processing steps lead to three main capsid protein species, CP44, CP39, and CP37 (named after their respective mobilities, in kilodaltons, on sodium dodecyl sulfate [SDS]-polyacrylamide gel electrophoresis [PAGE]), that share the same central sequence but differ in their termini (1, 4, 16). This is in contrast to the structural proteins of retroviruses, which represent different, nonoverlapping domains (MA, CA, and NC) of the precursor Gag protein, or to the hepatitis B virus (HBV) C protein, which is not cleaved.

FIG. 1.

(A) Schematic representation of the CaMV precapsid (CP56) and capsid proteins. Only the N terminus of CP44 (amino acid 77) is known exactly. The proposed N and C termini for CP39 and CP37 are based on the present study. The positions of the cloned sequences are indicated below. Clones were obtained from PCR amplification as described previously (1). All clones were verified by the dideoxynucleotide sequencing method. NLS, nuclear localization signal; ZnF, zinc finger. (B, upper panel) Phosphorylation assay. CP derivatives were overexpressed in E. coli strain BL21(DE3) cells by using pET3d vector, as described by Chapdelaine and Hohn (1). Bacterial inclusion bodies were isolated, solubilized in 8 M urea, and dialyzed against 10 mM Tris-Cl, pH 7.5. After dialysis, soluble proteins were separated by SDS-PAGE. Overexpressed proteins were visualized in 1 M KCl, electro-eluted in SDS buffer with a Bio-Rad model 422 electro-eluter according to the manufacturer's specifications, and dialyzed against 10 mM Tris-Cl, pH 7.5. Proteins were quantified by staining SDS-PAGE gels with Coomassie blue dye. Reaction mixtures (30 μl each) contained 20 ng of the CP derivatives [except for CP(77-265), of which only 5 ng was used], 15 μg of CaMV particles as a source of virus-associated kinase, and 2.5 μCi of [γ-³²P]ATP (3,000 Ci/mmol) in a solution containing 20 mM Tris-Cl, 10 mM MnCl₂, 0.8 mM dithiothreitol, and 0.25 mM EDTA (16) and were incubated at 37°C for 3 h. Phosphorylation was stopped by the addition of SDS loading buffer, and samples were separated by SDS-12% PAGE. Gels were dried and exposed for autoradiography. (Lower panel) Duplicate gels were transferred onto a nitrocellulose membrane (Schleicher & Schuell) for immunodetection (ECL system; Amersham) with anti-p37 antibodies (1). Note that the (smaller) amount of protein used for CP(77-265) was not sufficient to give a band.

CP44, but not CP39 or CP37, contains an acidic N-terminal sequence, which also includes serine, threonine, and proline residues, a combination that frequently renders proteins unstable (12). The acidic domain is followed by a short nuclear localization signal (13), an assembly domain (1), and a large cluster of basic amino acids involved in nucleic acid binding (1, 4), including a Zn finger motif specifying interaction with a packaging signal located on the CaMV RNA leader sequence (7). CP44 starts with amino acid 77 of the precursor protein (16); however, neither the N termini of CP39 and CP37 nor the C termini of any of these proteins are precisely known. Additional domains characterized by acidic residues, serine, and threonine and leading to protein instability make up the N and C termini of CP56 (12).

A virion-associated enzyme (17) with casein kinase II (CK II) properties (16) phosphorylates CP44, but not CP39 or CP37. In the experiments leading to these results, substrates and enzymatic activity in the same preparation were evaluated. In the present study, CaMV CP derivatives were expressed in Escherichia coli to separate the enzymatic activity and substrates. Serine residues S82, S86, and S88 were identified as the main targets on CP44 of the virion-associated kinase, and we report that Arabidopsis thaliana CK II has the same activity as the virion-associated kinase (13). At least one of the three serines, but not a specific one, is required for infectivity. Furthermore, we show that the Zn finger motif is included in CP44 but missing in CP37.

Sequences comprising precursor capsid protein amino acids 1 to 265, 77 to 265, and 99 to 265 [CP(1-265), CP(77-265), and CP(99-265), respectively] (Fig. 1A) were cloned in pET3d (Stratagene) and overexpressed in E. coli, and the corresponding proteins were purified. After incubation with viral capsids and [γ-P³²]ATP, the mixtures were separated on SDS-polyacrylamide gels. Positions and relative quantities of the CP derivatives were determined by Western blotting, and their phosphorylation was determined by autoradiography (Fig. 1B). CP(77-265), but not CP(99-265), became phosphorylated, showing that the phosphorylation targets of CP44 are located between positions 77 and 99. This sequence includes three serines (S82, S86, and S88) and one threonine (T81). Replacement of any two of the three serines by alanines led to strong [CP(77-265//S82,88A) and CP(77-265//S86,88A)] and moderate [CP(77-265//S82,86A)] reduction of the labeling, while replacement of all three serines CP(77-265//S82,86,88A) abolished most of it. This identifies the three serines as equal contributors to the phosphorylation of CP44. The residual labeling in the triple mutant is probably due to T81 phosphorylation (see below).

CP(1-265) became more efficiently labeled than CP(77-265) (Fig. 1B), indicating that additional phosphorylation targets are located in the preprotein region of this polypeptide [note that only 5 ng of CP(1-265) was used in the assay, compared to 20 ng of the other polypeptides]. The excess of CP derived from the CaMV virions over the substrate polypetides was about 1,000-fold. In our gels, phosphorylation of the virion-derived CP44 also became visible. The specific labeling of the latter was much weaker than that of the polypetides produced in E. coli, most likely because the virion proteins were extensively phosphorylated prior to the assay.

When the sequence of the A. thaliana CK became available (13), we cloned and expressed the two subunits in E. coli and reconstituted and purified the enzyme. The resulting preparation phosphorylated its own α subunit, as well as CP(1-265) and, to a lesser extent, CP(1-265//S82,86,88A), the result of the triple S→A change introduced into CP(1-265) (Fig. 2). Furthermore, a fragment encompassing amino acids 99 to 489 [CP(99-489)] became phosphorylated. This indicates that the virion-associated kinase is in fact CK II and confirms the presence of additional phosphorylation targets in the N- and C-terminal preprotein domains.

FIG. 2.

Phosphorylation assays of CP(1-265), CP(1-265//S82,86,88A), and CP(99-489) with A. thaliana CK II. Note also the autophosphorylation of CK II α. The phosphorylation assay was carried out as described in the legend for Fig. 1, except that recombinant CK II from A. thaliana was purified and used as described by Klimczak et al. (13).

To evaluate the importance of the three serines in CP44, the triple and also a single serine replacement was introduced into cloned CaMV [CaMV-CP56(S82,86,88A) and CaMV-CP56(S82A)]. Plants were inoculated at the three-leaf stage with 20 μg of wild-type (four plants) or mutant (eight plants each) viral DNAs which had been excised from the plasmid vector and religated. In all plants, the original CaMV clone and the single replacement led to normal mosaic symptoms after 16 days without any noticeable difference; the triple replacement mutant did not yield symptoms in any of the eight inoculated plants, even after a prolonged incubation of 27 days. A second passage with CaMV- and CaMV-CP56(S82A)-infected plant extracts yielded symptoms already after 10 days, while extracts of plants that had been inoculated with the triple replacement mutant again remained healthy. To test for a possible silent infection, protein extracts from plants inoculated with CaMV-CP56(S82,86,88A) or CaMV-CP56(S82A) were analyzed by Western blotting with anti-CP37 antibodies. No positive signal could be detected from the triple mutant, while extracts from the single mutant showed strong signals at the positions of CP44, CP39, and CP37 (Fig. 3A). To show that symptoms in CaMV-CP56(S82A)-infected plants were not due to reversions to the wild type, extracts from four of these plants were used for recloning the CP sequence by PCR and sequencing it. In all cases the CP sequence including the S82A mutation was retained. These experiments strongly suggest that phosphorylation of at least one of the serines S82, S86, and S88 is required for infectivity, although we cannot formally exclude the possibility that the amino acid substitutions rather than the lack of phosphorylation caused the lethality.

FIG. 3.

(A) Test for capsid proteins in plants inoculated with CaMV mutants. Results are shown from Western blotting with rabbit anti-CP37 (1) of total viral proteins extracted from plants inoculated with mutant virus DNAs (10). Ten milligrams of total protein was loaded for each sample. (B) Absence of the Zn finger motif in CP37. Results are shown from Western blotting of purified CaMV with rabbit anti-CP37 and anti-Zn finger immunoglobulin G (anti-ZF). For virus purification, 5 g of tissue was homogenized in an extraction buffer (0.2 M Tris-Cl [pH 7.0], 20 mM EDTA, and 1.5 M urea). A one-tenth volume of 20% Triton X-100 was added. Extracts were incubated at 40°C with gentle stirring for 2 h, filtered through four layers of cheesecloth, and cleared by centrifugation at 17,000 × g for 10 min. Supernatants were underlaid with a 15% sucrose cushion, and virus particles were pelleted at 22,000 rpm in an SW27 rotor for 2 h. Pellets were resuspended in 500 μl of storage buffer (0.1 M Tris-Cl [pH 7.4], 2.5 mM MgCl₂). Rabbit anti-Zn finger immunoglobulin G serum was obtained by injection of the peptide NIEGHYANECPN conjugated to bovine serum albumin (Readysysteme AG, Bad Zurzach, Switzerland).

The CK II phosphorylation target consensus sequence is S/TXXD/E, with additional D or Es at positions −1, +1, +2, +4, and +5 increasing its quality and R, K, or Hs at position −1 decreasing its quality. Serine is the better target residue, and aspartic acid at +3 is the better activator (20). According to these criteria, a total of eight possible targets are located within CP44 (Fig. 4). Since CP39 and CP37 are not phosphorylated (16) and are known to include the sequence from positions 122 to 400, the five targets located within this sequence are apparently not used at all or are very inefficiently used. T126, T160, T175, and T210 are in fact poor targets because they are preceded by basic amino acids. Also, S347 is relatively weak, since the site is not supported by an additional acidic amino acid. According to these criteria, CP44 phosphorylation is due mainly to phosphorylation of S82 and S88, because they are preceded by acidic amino acids. In addition to the acidic amino acid at position +3, further acidic amino acids are present in key positions. However, S86 was also efficiently phosphorylated, although the acidic amino acid at +3 is missing. In this case, acidic amino acids are present at positions −1, +1, +2, +4, and +5. Most of the phosphorylation of CP(77-265) was abolished when these three serines were mutated to alanines. A weak residual activity is probably due to phosphorylation of T81, which is in good context for CK II. Several very good additional phosphorylation sites can be predicted for the CP precursor, namely, S4, S25, T61, T62, S66, and S68 at its N-terminal sequence and T472, S473, S481, T482, and S486 at its C-terminal sequence (Fig. 4). At least one each, but probably more, in the N- and C-terminal preprotein domains is responsible for the phosphorylation of the preprotein sequences that we had observed.

FIG. 4.

Possible Ser/Thr kinase target sites are shown in grey boxes, with those predicted specifically as strong targets for CK II being in white letters. The three serines identified in this study as targets are underlined. Important motifs in the N- and C-terminal regions of CP44, i.e., the nuclear localization signal (RKRK) (14) and the Zn finger motif, are boxed. The N- and presumptive C-terminal sequences of the precursor protein not present in CP44 are shown separately.

Phosphorylation of CP44, but not of CP39 or CP37, suggests that CP39 is generated by N-terminal cleavage, and later, there is a possibility that CP37 is generated by C-terminal cleavage. Furthermore, the mobility of recombinant CP derivatives (1) indicates that C-terminal processing of CP44 or CP39 into CP37 is likely to remove the Zn finger motif (amino acids 414 to 427). We tested this hypothesis by Western blotting of capsid proteins with antibodies raised against a synthetic peptide representing this motif. CP44 and CP39, but not CP37, were clearly recognized by anti-Zn finger antibodies (Fig. 3B), showing that the Zn finger is at least partly removed in CP37.

Deciphering the processing steps of the CaMV capsid protein has turned out to be a challenging task. It remains an open question whether CP44 or CP37 is the mature virus capsid protein. Here we provided evidence that the processing of CP44 into CP37 implies the removal of the phosphorylation sites and of the Zn finger motif, both of which are essential for infectivity (7, 16). We assume that the removed polypeptides are required early in infection, e.g., in the assembly and/or reverse transcription process. It has been shown previously (8) that fast- and slow-migrating forms of the virus contain almost exclusively CP44 and CP37, respectively. The conversion from a fast- to a slow-migrating form is rapid and probably cooperative since, as we pointed out, only a limited amount of intermediate forms are observed.

The removal of 4 to 5 kDa of protein from the CP44 N terminus to yield CP39 indicates that the cleavage site is located between amino acids 100 and 110. This cleavage occurs in an acidic environment. C-terminal cleavage to generate CP37, on the other hand, occurs in the Zn finger motif or immediately upstream of it, i.e., in a basic environment. This finding raises the question of whether only one protease is involved in the processing of CP56, whether two different proteases are involved, or whether self-cleavage occurs. It has been shown that the viral protease is involved in the N-terminal processing of CP56 (24). The instability in protoplasts of CP56 mutants containing the phosphoserines (12) may hint that a host protease is involved in CP44 processing. However, this processing may also result from degradation of the host machinery (12). CaMV CP56 processing certainly merits further attention.

Although phosphorylation of at least one of the serines S82, S86, and S88 seems to be essential for virus infectivity, their role in assembly remains to be determined. HBV is a pararetrovirus that shares with CaMV similar strategies of replication (21). Phosphorylation of the core protein of HBV was proposed to be involved in the docking of the virus particles to the nuclear pore (11), exclusion of the virus particles from membrane binding (15), and packaging of the viral genomic RNA (6). We have shown that the phosphorylation of CaMV CP44 is not important for nuclear targeting when the protein is expressed transiently in plant protoplasts (14). However, it is possible that, as for HBV, phosphorylation plays a role in the docking of the virus particle to the nuclear pore complex (12). It is also possible that phosphorylation regulates the RNA or DNA binding activity of CP56 during the replication cycle. In this scheme, the acidic N and C termini may be removed once assembly is completed. Such regulation may occur intra- or intermolecularly. This hypothesis is attractive since the strong nucleic acid binding activity of CP56 (1, 9) needs to be tightly regulated to allow the specific packaging of the RNA pregenome, notwithstanding the timing between packaging and reverse transcription (7).

The Zn finger motif is the most well-conserved sequence among retroelements, and it has been shown to be involved in specific RNA binding and to be essential for replication (22). We have shown here that this motif is still located at the C terminus of CaMV CP44 and is cleaved off in CP37, the main capsid protein in CaMV. The smaller cleavage product has not been detected in virus preparations; we assume that it is released from CaMV particles after cleavage. This previous assumption suggests that the motif also plays a role as part of CP44, presumably in reverse transcription. In retroviruses, the RNA binding nucleocapsid protein is maintained within the virus shell because reverse transcription occurs only after infection. The absence of the Zn finger motif from mature CaMV particles indicates that the motif has no function after reverse transcription and assembly have been completed. Furthermore, the Zn finger might be involved in interaction with the transactivator/viroplasmin (TAV, encoded by open reading frame VI). We have found that the introduction of mutant DNA lacking the Zn finger motif into CP56 weakens the interaction with TAV (9), which also implies that TAV's interaction with CP37 is weaker than that with CP44. This weaker interaction might allow the release of virus particles from the inclusion bodies and concomitant transport to the nucleus or to neighboring cells.