Generation and characterization of mutants of tomato spotted wilt virus

In nature, tospoviruses like tomato spotted wilt virus (TSWV) are exclusively transmitted by thrips species (Sakimura, 1962) producing numerous enveloped virions during infection, which accumulate in the cisternae of the endoplasmatic. reticulum. system (Kitajima, 1965; Milne, 1970; Ie, 1971). Under experimental conditions however, it is common practice to maintain the virus by mechanical inoculation onto susceptible host plants.

Repeated passages of animal viruses, certainly at high inoculum densities results in the generation of defective mutants which co-replicate with the wild type virus interfering in their multiplication (Lazzarini et al ., 1981). This phenomenon is reported for only a few plant viruses (Morris & Knorr, 1990). As defective viral mutants lack one or more genetic functions but are still capable to co-replicate, they may constitute useful tools to study viral genes and viral protein functions.

This thesis reports studies aimed to generate and characterize defective forms of TSWV by their biological, serological and genetic properties. The understanding of these characteristics indeed, can help the elucidation the multiple events which take place during the infection process, and ultimately provide new ways to control the virus.

The first attempts to characterize TSWV defective forms showed that non-enveloped virus isolates are generated upon mechanical inoculation (Ie, 1982). These mechanically transmitted isolates were considered to represent morphologically defective forms of TSWV, lacking the lipid envelope though still being infectious (Ie, 1982). Further analysis revealed that these morphologically defective isolates failed to produce detectable amounts of the envelope glycoproteins G1 and G2 (Verkleij & Peters, 1983).

The results reported in this thesis show that during a series of mechanical transfers of TSWV, actually two distinct types of mutants are generated (Chapter 4). Firstly, starting with a wild type, the Dutch isolate of TSWV NL-04, a morphologically-defective isolate was obtained, which had lost its ability to produce the membrane-glycoproteins and, as a consequence, was not able to form enveloped particles. The appearance of such isolates could be followed by ELISA tests and readily detected by electron microscopy (Chapter 3 and 4). Secondly, starting from various TSWV wild type-isolates, defective mutants were obtained that had accumulated deleted forms of the large (L) RNA segment that most likely represented defective interfering (DI) RNAs, since they replicated more rapidly than full-length L RNA and their appearance was often associated with symptom attenuation in host plants (Chapter 4).

After elucidation of the nucleotide sequence of the M RNA of two tospoviruses, i.e. TSWV and Impatiens necrotic spot virus (INSV) (Kormelink et al ., 1992; Law et al. , 1992), the genetic nature of the morphological defectiveness of the envelope-deficient isolates could be identified by comparing their M RNA sequences with those of wild type isolates (Chapter 6). Comparison of (partial) M RNA sequences of several TSWV and INSV isolates revealed that the accumulation of point mutations in the G2/G1 ORF in this RNA may have been the causal event that led to the generation of envelopedeficient isolates during mechanical transmission. It was found that an envelope-deficient isolate (US-01) of INSV had acquired an extra nucleotide in this gene, causing a frameshift and consequently the loss of the putative signal peptide of the glycoprotein precursor (Chapter 6). In this case, the envelope deficiency may be explained by a blockage of the trans-membrane transport, and hence further maturation of the glycoprotein precursor. For the isolate NL-04 of TSWV the morphological defect seems to be caused by the accumulation of point mutations in the glycoprotein precursor rather than deletion of the hydrophobic signal sequence. The accumulation of point mutations in these isolates may either result in dysfunctional glycoproteins which do not become inserted in viral envelopes, or in stop codons though not detected in the sequenced part of the G2/G1 gene, but which are possibly present (further downstream), in the non analyzed part of this gene. The presence of these stop codons would then lead to premature termination of the G2/G1 precursor.

The generation and characterization of the envelope-deficient isolate of TSWV NL-04 has already lead to two important conclusions, with respect of the possible involvement of the glycoproteins during the tospovirus infection cycle. First, it has been observed that this defective-isolate is still able to infect leaf tissues at a similar rate as to that of wild type infection. Thus the glycoproteins (and lipid envelope) are neither essential for cellto-cell movement of the virus, nor for long-distance transport in the plant. The conclusion seems therefore to justify that tospoviruses, during the development of infection, are transported mainly as free nucleocapsid complexes. Secondly, the presence of lipid enveloped (and glycoproteins) appeared to be essential for successful thrips transmission, since sofar, the envelope-deficient isolate of NL-04 fails to show vector transmissibility (Wijkamp, personal communication). This finding strongly suggests a role of G1 and/or G2 glycoproteins in the virus-vector interactions. It is noteworthy in this context, that the primary structure of these proteins contain a so-called "cell attachment site (RGD) at the N- terminus of G2 (Kormelink et al. , 1992) which may be involved in the recognition of a receptor in the thrips midgut.

The results reported in this thesis also show that, tospoviruses generate DI RNAs during sequential passages of the virus at high multiplicity of infection (Chapters 4 and 5). DI RNAs have frequently been described in either positive or negative-stranded
animal virus systems (Holland, 1985; Nayak et al. , 1990), but have been reported for only few positive-strand plant viruses (Morris & Knorr, 1990). Therefore, the DI RNAs of TSWV represent the first fully characterized DI molecules among negative-strand RNA viruses infecting plants. The DI molecules are found to interfere strongly with the replication of the wild type virus genome resulting in less severe, attenuated symptoms in host plants (Chapter 4). Characterization of the DI RNAs occurring with four distinct tospovirus isolates revealed that they are exclusively derived from the (viral polymerase-encoding) L RNA segment (therefore represent "typical" DI RNAs, Holland, 1985; Nayak et al. , 1990) by deletion of approximately 5.4 (60%) to 7.0 (80%) kilobases of the standard genomic L RNA while both genomic termini are retained. The presence of both 5' and the 3' termini in the DI molecules of TSWV indicate that these regions possess the essential signals for genome replication and possibly other regulatory functions (Chapters 4 and 5). Furthermore, these defective molecules are able to be encapsidated and incorporated into enveloped particles (Chapter 4).

Short repeated nucleotide sequences were identified at the junction sites of various DI molecules and a possible model for their generation is proposed, in which the viral polymerase can jump across secondary structures where these repeated internal sequences are located.
Due to their replicative advantage over wild type RNAs, the TSWV DI RNAs may constitute a powerful tool to study the genome information required for viral replication, encapsidation and packaging into virus particles, and to unravel the RNA replication process. Infectious transcripts from cloned DI DNA copies, provide a useful approach among positive-strand RNA viruses infecting either animal or plant systems (Makino et al., 1991; van der Most et al., 1992; Hagino-Yamagishi et al., 1990; Burgyan et al., 1992; Morris & Knorr, 1990). For negative-strand RNA viruses, however, the results have been very limited thus far. Recently, with the development of the vaccinia T7-expression system (Fuerst et al., 1987) infectious rhabdovirus VSV-DI transcripts could be produced and has opened new ways for using these defective molecules (Pattnaik et al., 1992). This system, however, requires the expression of all viral replicatory proteins, an approach which is not available yet for TSWV.

Alternatively, transgenic DI-expressing plant systems could be used, not only to study the viral RNA replication process, but also as an approach towards engineered protection of host plants against virus infection. Due to their features, transgenetically
expressed DI transcripts could potentially protect plants against the disease symptoms of TSWV, infection even though the transformed plants are only "tolerant", rather than "immune". The results described in Chapter 7 demonstrate that engineered protection to TSWV (as reported using the nucleoprotein gene, Gielen et al., 1991; de Haan et al., 1992) can also be obtained by transforming tobacco plants with defective interfering (DI) L RNA copies. Transgenic plants expressing viral DI RNA sequences were obtained, which upon virus inoculation, showed a pronounced delay in symptom expression in 9
out of 20 lines tested, while in 3 of these 9 lines (VC4, VC7 and V7) part of the progeny plants was completely protected. Since the type of protection obtained appeared to be rather than tolerance, it is proposed that the protection observed in the Dl
expressing transgenic plants is based on anti-sense inhibition and not in the ability of the DI transcripts to co-replicate along with the infecting virus. Northern analyses on challenged, protected plants are required to verify whether the DI-specific transcripts,
though containing extensive non-viral sequences at both ends, have indeed been unable to replicate in the transgenic plants. To obtain genuine "DI-mediated" protection and not only anti-sense inhibition, transgenic DI-cDNA cassettes should probably be provided with a trimmed CaMV promoter (upstream) and a ribozyme sequence (e.g. a hepatitis delta type ribozyme) downstream as to obtain transgenic DI transcripts without any extra, non-viral nucleotides which may block their potential replicative ability.

Since DI RNAs of tospoviruses may be generated in any host system that permits the growth of the wild type virus, it is likely that they are not only generated under laboratory conditions, but also in natural infections. The presence of such molecules in enveloped particles (Chapter 4), indeed indicates that tospovirus DI RNAs can potentially be transmitted by thrips vectors and hence, "survive" in nature. If DI RNAs indeed do occur in natural infections, their presence during virus infection may serve to attenuate the pathogenic effects of the parental virus infection. Their occurrence may also have been advantageous for the evolution of tospovirus and for the establishment of their current, impressive host range.