Cowpea Mosaic Virus 32- and 60-Kilodalton Replication Proteins Target and Change the Morphology of Endoplasmic Reticulum Membranes

Construction of the transient expression vectors.

Plasmid pMON999 is a plant expression vector that contains the enhanced cauliflower mosaic virus (CaMV) 35S promoter, followed by a multiple cloning site and the terminator of the nopaline synthase gene (43). pMON-HA is derived from pMON999 and contains the coding region of the hemagglutinin (HA) epitope tag (YPYDVPDYA) between the promoter and terminator (44). The individual CPMV RNA1 proteins were PCR amplified with gene-specific primers (Table 1) and cloned in these vectors. Because the individual proteins are expressed in a polyprotein, the primers were designed to add a start codon at the 5′ end and a stop codon at the 3′ end of each gene when appropriate. Furthermore, the primers contained restriction sites to facilitate cloning. pM-HA-32K was constructed by amplification of the 32K coding region from pTB1G (14) with primers EcoRI_32K_F and BamHI_32K_R and insertion in pMON-HA. pM-HA-60K was constructed by PCR amplification of the 60K coding region from pTB1G with EcoRI_60K_F and BamHI_60K_R, and pM-HA-GFP was constructed by amplification of the green fluorescent protein (GFP) coding region from pM19GFP10 (16) with primers EcoRI_GFP_F and BamHI_GFP_R. pM-60K was constructed by amplification of the 60K coding region from pTB1G (14) with primers XbaI_60K_F and BamHI_60K_R and insertion in pMON999. pM-32K was constructed by insertion of an XbaI-SstI fragment of pTB32*, which contains the 32K coding sequence followed by a stop codon and the SstI restriction site (47), into pMB200K. pMB200K contains the entire RNA1-encoded 200K polyprotein followed by the SstI restriction site in pMON999 (43). The XbaI restriction site is present internally in the coding sequence of 32K.

Construction of TRV expression vectors.

Tobacco rattle virus (TRV) vectors carrying CPMV RNA1-encoded proteins were constructed with TRV-GFPc. This plasmid contains cDNA from TRV RNA2 and an engineered subgenomic promoter followed by the GFP gene (23). First this plasmid was modified to contain a start codon followed by the coding sequence of the HA epitope tag and a multiple cloning site. This was achieved by replacement of the NcoI fragment released from TRV-GFPc with a PCR fragment amplified from pM-HA-32K with primers NcoI_HA_F and SstI_TAA_32K_R digested with the same enzyme. This intermediate plasmid was designated TRV-HA-Δ32K and contains the SalI, PstI, and EcoRI sites directly following the coding sequence for the HA epitope. These restriction sites and the EcoRI, SstI, or KpnI present at the 3′ end can be used to insert a gene of choice as an N-terminal fusion with the HA tag.

TRV-HA-24K was constructed by PCR amplification of the 24K coding sequence from pTB1G with primers NcoI_24K_F and SstI_TAA_24K_R. This PCR product was digested with NcoI (filled in with Klenow polymerase) and SstI and cloned in TRV-HA-Δ32K digested with EcoRI (filled in with Klenow polymerase) and SstI. TRV-HA-32K was constructed by PCR amplification of the 32K coding sequence from pTB1G with primers NcoI_32K_F and SstI_TAA_32K_R. This PCR product was digested with HindIII and SstI and was cloned in TRV-HA-Δ32K digested with HindIII and SstI. TRV-HA-60K was constructed by PCR amplification of the 60K coding sequence from pMON-HA-60K with primers NcoI_60K_F and EcoRI_TAA_60K_R. This PCR product was digested with EcoRI and cloned in TRV-HA-Δ32K digested with EcoRI. To construct TRV-HA-110K, the 110K coding sequence was PCR amplified from pTB1G with primers NcoI_110K_F and SstI_TAA_110K_R, digested with NcoI and SstI, and cloned in TRV-GFPc digested with the same enzymes, resulting in clone TRV-110K. Subsequently TRV-HA-24K was digested with KpnI, and the released fragment was replaced with the KpnI fragment of TRV-110K to construct TRV-HA-110K. The fusion constructs of the different viral proteins with GFP were made by using the SalI restriction site directly downstream of the HA epitope sequence. The GFP coding sequence was PCR amplified from pM19GFP10 (16) with primers SalI_GFP5_F and SalI_GFP5_R. For the construction of TRV-HA-VAP27, the complete coding sequence of the 27-kDa VAMP-associated protein (VAP27) was released as an EcoRI fragment from pGAD10-VAP27, a clone isolated from an Arabidopsis thaliana cDNA library. This fragment was ligated in TRV-HA-Δ32K digested with the same enzyme.

Transfection of cowpea protoplasts and infection of plants.

Cowpea mesophyll protoplasts were prepared and transfected by polyethylene glycol-mediated transformation of plasmid DNA as described previously (43). Five-week-old wild-type N. benthamiana plants or N. benthamiana plants carrying the mGFP5-ER transgene (34) were dusted with carborundum and infected with TRV RNA1 together with transcripts of the TRV RNA2 recombinants as described previously (23).

Subcellular fractionation and immunoblotting.

Protoplasts transfected with the transient expression vectors were collected 16 h posttransfection and centrifuged at 30,000 × g, and proteins from the supernatant and pellet fraction were electrophoresed in 10% polyacrylamide gels and immunoblotted as described previously (5) with the mouse monoclonal antibody anti-HA (clone 12CA5; Boehringer) recognizing the HA epitope tag.

For immunoblot analysis of the proteins produced by the TRV expression vector, infected leaves were collected 1 day postinfection and ground with a mortar and pestle with 2 ml of extraction buffer (50 mM Tris-acetate [pH 7.4], 10 mM potassium acetate, 1 mM EDTA, 5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride per g of fresh weight). After extraction, the homogenates were centrifuged, electrophoresed, and immunoblotted as described earlier (5) with anti-HA or anti-GFP (Clontech). In some cases, homogenates were not centrifuged, but total protein was extracted with sample buffer (10% glycerol, 5% β-mercaptoethanol, 2% sodium dodecyl sulfate, 0.01% bromophenol blue, 75 mM Tris-HCl, pH 6.8), boiled for 3 min, and loaded on the gel.

Fluorescence microscopy.

A Zeiss LSM 510 confocal microscope was used to obtain images. GFP fluorescence was observed with standard settings (excitation wavelength, 488 nm; emission band-pass filter, 505 to 550 nm). Red chlorophyll fluorescence was detected with an excitation wavelength of 543 nm and an emission long-pass filter of 560 nm. To observe GFP fluorescence in living epidermal cells, large samples (2 by 2 cm) were cut from N. benthamiana leaves infected with the TRV recombinants 1 day postinfection. The samples were placed on a glass slide, and after a drop of water was applied on top of the samples, they were covered with a coverslip. The samples were directly analyzed by confocal laser scanning microscopy.

Expression of individual CPMV RNA1-encoded proteins in plant cells.

In previous studies several individual CPMV proteins were transiently expressed in cowpea protoplasts by using the plant expression vector pMON999, which carries the enhanced CaMV 35S promoter (30, 43). Those studies concerned the viral RdRp and therefore were focused on the C-terminal 110K region of the 200K polyprotein. In the present study the proteins contained in the N-terminal region were included also. For this purpose, the coding sequences for the 32K proteinase cofactor and the 60K nucleoside triphosphate (NTP) binding protein were inserted into the multiple cloning site of pMON999 to produce pM-32K and pM-60K. The constructs were used to transfect protoplasts, which were then harvested 16 h posttransfection. Homogenates of the protoplasts were subjected to centrifugation at 30,000 × g, crudely separating soluble proteins from membrane-bound proteins. It was shown previously by immunoblotting procedures that 110K and 87K accumulate to high levels in protoplasts and are found mainly in the soluble fraction (S30) (43). 32K and 60K were not detectable in either S30 or the pellet fraction (P30) with anti-32K and anti-VPg sera (data not shown). The failure to detect 32K and 60K might be due to low expression levels of these proteins and/or to low titers of the antisera. To avoid the latter possibility, constructs were made with the HA epitope tag fused to the N terminus of 32K and 60K (Fig. 1B). As a positive control, the HA epitope was also fused to GFP. Western blot analysis of extracts from transfected protoplasts with mouse monoclonal antibodies raised against the HA epitope revealed that HA-GFP accumulated to readily detectable levels and was present mainly in the S30 fraction (Fig. 2). HA-60K accumulated to much lower levels and was present mainly in the P30 fraction (Fig. 2), whereas 32K was not detectable in either S30 or P30.

To confirm the results obtained in the protoplast system, the individual CPMV RNA1-encoded proteins were expressed in plants by using the TRV expression system (23). TRV does not replicate and spread in cowpea plants, which are the natural host of CPMV. However, in N. benthamiana, which is also a systemic host for CPMV, the TRV-based expression system was shown previously to give high levels of foreign protein expression (23). Since TRV does not cause visible morphological changes to the host endomembrane system as observed by electron microscopy (9) and is not related to CPMV, a TRV infection probably does not interfere with proper targeting of the CPMV proteins. Vector TRV-GFPc contains the coding sequence of GFP in a cDNA clone of TRV RNA2 in which GFP can be replaced with other genes of choice (Fig. 1C) (23). To facilitate immunodetection of the CPMV proteins, TRV-GFPc was modified to encode the HA epitope followed by a multiple cloning site. Subsequently, the coding sequences for the 24K proteinase, the 32K proteinase cofactor, the 60K NTP binding protein, and the 110K polymerase were each inserted into the TRV vector immediately downstream of the HA tag, thereby replacing the GFP gene (Fig. 1C). In vitro transcripts of these constructs were combined with TRV RNA1 and used as inoculum on N. benthamiana. Homogenates of infected leaves were separated into P30 and S30 fractions. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis was performed on samples of these fractions followed by Western blot analysis with mouse monoclonal antibodies recognizing the HA epitope present at the N terminus of the proteins. 24K and 110K were present predominantly in the soluble fraction, whereas 60K was present in the crude membrane fraction (Fig. 3A). We were unable to detect 32K in either fraction (Fig. 3A). To confirm that membrane-associated proteins are localized to the pellet during fractionation, cDNA of an A. thaliana integral membrane protein, VAP27, was cloned into TRV as a fusion with the HA epitope tag. VAP27 is a SNARE-like protein that is inserted posttranslationally into the ER membrane via a C-terminal transmembrane domain (37). Homogenates of TRV-VAP27-infected leaves were fractionated, and like 60K, VAP27 was detected predominantly in the P30 fraction (Fig. 3A).

These data suggest that 60K associates with membranes when expressed in the absence of other CPMV-encoded proteins whereas 24K and 110K behave as cytoplasmic, soluble proteins. We were unable to detect the 32K protein by immunoblotting either expressed in N. benthamiana leaves with the TRV expression system or transiently expressed in cowpea protoplasts.

Intracellular localization of the replication proteins.

To visualize and characterize the protein-membrane interaction in live cells, TRV expression vectors encoding fusions of the CPMV proteins with GFP were constructed. First it was established by immunoblotting that the constructs were properly expressed. Homogenates of infected leaves were separated by centrifugation in the P30 and S30 fractions. By using anti-GFP serum, GFP-24K was found present mainly in the S30 fraction (Fig. 3B). In contrast to native 32K, GFP-32K was detectable by immunoblotting and was found to be present mainly in the P30 fraction (Fig. 3B). Expression of GFP-60K and GFP-110K was detected in the leaf homogenates (Fig. 3B); however, after fractionation in the P30 and S30 fractions no specific bands could be visualized for unknown reasons (data not shown).

The intracellular localization of the fusion proteins was examined in infected leaf epidermal cells 1 day postinoculation (dpi) by confocal laser scanning microscopy. Cells containing GFP-24K (data not shown) or GFP-110K (Fig. 4A) displayed a pattern of fluorescence similar to that of nonfused GFP (Fig. 4B), and fluorescence was present diffusely in the cytoplasm and in the nucleus. Since GFP is excluded from the vacuole that takes up more than 90% of the cell volume in epidermal cells, cytoplasmic GFP is confined to a small layer between the vacuole and the cell wall (the cortex) and to a small region near the nucleus (Fig. 4B). The occurrence of GFP in the nucleus is in line with earlier reports that indicate that GFP alone or fused to a soluble protein accumulates partly in the nucleus, probably by means of passive diffusion through the nuclear pore (46). The fluorescence pattern of GFP-60K differed from that of nonfused GFP. GFP-60K fluorescence was most obvious in a ring surrounding the nucleus, in small rings surrounding spherical organelles in the cytoplasm, and in one or several spherical aggregates often near the nucleus, while no fluorescence was present in the nucleus (Fig. 4C). The fluorescent ring around the nucleus was reminiscent of what we have observed in transgenic N. benthamiana plants expressing ER-GFP (7), a GFP fusion protein that contains an N-terminal cleavable signal peptide and a C-terminal HDEL retention signal and therefore is targeted to the lumen of the ER (18) (compare Fig. 4C with Fig. 5B). It thus appears that GFP-60K localizes to the ER membranes that are contiguous with the nuclear envelope. Alternatively, 60K is present in the nuclear envelope itself. Red chlorophyll autofluorescence was observed in the small spherical organelles (Fig. 4C, inset), suggesting that GFP-60K is also present in the outer membrane of plastids. The nature of the spherical aggregates near the nucleus (Fig. 4C, arrow) is unknown, but it might be a structure derived from ER membranes, as pointed out below. Also the fluorescence observed with GFP-32K differed markedly from that for nonfused GFP. The fluorescence was most obvious at the cortex of the infected cells and was present in a clear reticulate pattern (Fig. 4D) closely resembling the pattern observed with ER-GFP in transgenic plants (compare Fig. 4D with Fig. 5A). It thus appears that GFP-32K localizes to the cortical ER network. In some cells the reticulate pattern was less bright and GFP-32K was additionally present in large highly fluorescent aggregates at the cortex, which may be derived from ER membranes (Fig. 4E). No fluorescence of GFP-32K was observed in the nuclear envelope (data not shown), which is in contrast to the fluorescence observed in transgenic plants that express ER-GFP or in TRV-GFP-60K-infected plants.

These data corroborate the cell fractionation data and indicate that individually expressed 24K and 110K are cytosolic, soluble proteins and that 32K and 60K associate with membranes mainly derived from the ER.

Expression of 32K and 60K modifies ER morphology.

CPMV infection induces an extensive rearrangement of ER membranes, a process independent of RNA2-encoded proteins (7). To investigate whether a specific RNA1-encoded protein is responsible for these membrane modifications, the TRV constructs each carrying an individual RNA1-encoded protein (not fused to GFP) were used to infect N. benthamiana plants transgenic for GFP targeted to the lumen of the ER (18, 34). At 1 dpi the morphology of the ER membranes in infected epidermal cells was observed and compared to that of mock-infected cells. TRV infection itself had no influence on ER membrane morphology, since, in plants infected with TRV, ER-GFP fluorescence in epidermal cells did not differ from that in mock-infected cells and fluorescence was detected in the typical cortical ER network (Fig. 5A) and the nuclear envelope (Fig. 5B). In TRV-24K- and TRV-110K-infected cells the fluorescence pattern of ER-GFP was also indistinguishable from that of mock-infected cells (data not shown). In contrast, the ER in cells infected with TRV-32K or TRV-60K differed radically from the ER in mock-infected cells. In TRV-32K-infected cells, ER-GFP fluorescence was present in large amorphous aggregates that were connected with the cortical ER network (Fig. 5C) and in spherical aggregates (Fig. 5D) which were often near the nucleus. While in TRV-60K-infected cells these spherical aggregates were also found near the nucleus (Fig. 5E), cortical ER aggregates were not observed (data not shown). Both types of aggregates resembled the regions of proliferative ER membranes induced in CPMV-infected cells, although the extent of ER proliferation in CPMV-infected cells was generally larger (Fig. 5F). The ER modifications seen with TRV-32K and with TRV-60K may be artifacts that occur upon expression of any ER membrane-associated protein with the TRV-based expression system. This was tested by expressing the integral ER membrane protein VAP27 with the TRV vector. When TRV-VAP27 was used to infect ER-GFP transgenic plants, no changes were found in the ER-GFP fluorescence pattern from that of mock-infected cells (data not shown).

From these data we conclude that expression of 32K and 60K induced drastic morphological changes in the ER, resembling CPMV-induced ER membrane proliferation, whereas 24K and 110K expression did not influence ER morphology.

Expression of 32K and 60K induces necrosis.

During the course of the experiments a severe effect of prolonged expression of 32K and 60K on cell viability was observed. TRV infection caused no visible symptoms on inoculated leaves of N. benthamiana, but a slight curling of the systemically infected leaves was apparent 4 dpi. Infection of leaves with TRV-GFP allowed us to monitor the spread of TRV from the initial infection sites to the systemic leaves by using a handheld UV lamp. One day after infection GFP fluorescence was visible as green spots on inoculated leaves. With time these spots grew larger, and 3 dpi fluorescence was observed in large patches and in the major veins of the inoculated leaves as well as in the veins of the systemically infected leaves (Fig. 6A ′). In contrast to TRV and TRV-GFP (Fig. 6A), infection with TRV-32K or TRV-60K triggered formation of necrotic spots on the inoculated leaves. The first sign of lesion formation occurred from approximately 2 dpi onward when infected tissue developed a glassy appearance. At 3 dpi the developing lesions were more distinct from the surrounding tissue, appearing more clearly necrotic and collapsed (Fig. 6B and C). The necrotic tissue was typically confined to the inoculated leaves, but occasionally some necrosis was observed in the upper, noninoculated leaves at 4 to 6 dpi (data not shown). This suggests that movement of the recombinant TRV RNA2 is severely inhibited. Also the TRV construct carrying the GFP-32K fusion produced formation of necrotic spots, but TRV-GFP-60K did not trigger necrosis (data not shown). Infection with TRV-24K, TRV-110K, TRV-GFP-24K, TRV-GFP-110K, and TRV-VAP27 did not result in the formation of necrotic lesions, although a slight yellowing of infection sites was observed for TRV-110K at 6 dpi (data not shown). These results suggest that prolonged expression of 32K or 60K or the fusion of 32K with GFP from the TRV expression vector leads to cell death. This was rather surprising, since CPMV does not cause necrosis on N. benthamiana plants and moves systemically to the upper leaves and high amounts of 32K and 60K accumulate during the infection. To determine whether 32K and 60K expression also produced necrosis in other host plants, Nicotiana clevelandii, Nicotiana tabacum, and Nicotiana glutinosa plants were infected with TRV-32K, TRV-60K, and TRV-GFP as control. In all these Nicotiana species, necrosis was observed at 2 to 3 dpi on inoculated leaves infected with TRV-32K and TRV-60K, whereas leaves infected with TRV-GFP were symptomless (Fig. 6D to G and data not shown).