|Home | About | Journals | Submit | Contact Us | Français|
The DA strain and other members of the TO subgroup of Theiler's murine encephalomyelitis virus (TMEV) induce a persistent central nervous system infection associated with an inflammatory white matter demyelinating disease. TO subgroup strains synthesize an 18-kDa protein, L*, out of frame with the polyprotein from an initiation codon 13 nucleotides downstream from the polyprotein's AUG codon. We previously generated a mutant virus from our infectious DA full-length clone that has a change of the L* AUG codon to ACG (with no change in the polyprotein's amino acid sequence). Studies of this mutant virus showed that L* was key to the TO subgroup phenotype because the mutant had a decreased ability to persist and demyelinate. This work was initially called into question because a similar mutant derived from a different full-length DA infectious clone persisted and demyelinated similarly to wild-type DA virus (O. van Eyll and T. Michiels, J. Virol. 74:9071-9077, 2000). We now report that (i) the sequence of the L* coding region differs in the two infectious clones, resulting in a Ser or Leu as the predicted amino acid at position 93 of L* (with no change in the polyprotein's amino acid sequence), (ii) the difference in this amino acid is key to the phenotypic differences between the two mutants, and (iii) the change in amino acid 93 may affect L* phosphorylation. It is of interest that this amino acid only appears critical in determining the virus phenotype when L* is present in a significantly reduced amount (i.e., following translation from an ACG initiating codon).
The DA strain and other members of the TO subgroup of Theiler's murine encephalomyelitis virus (TMEV) induce an acute subclinical gray matter central nervous system (CNS) infection in SJL mice that is followed by a chronic inflammatory white matter demyelinating disease; virus persists in the CNS for the life of the mouse (for a review, see reference 16). The DA virus demyelinating disease serves as a model of multiple sclerosis since the two are similar in white matter pathology and both disease processes appear to be immune mediated. In contrast, the GDVII strain and other members of the GDVII subgroup of TMEV cause an acute fatal neuronal infection and fail to persist.
The study of the TMEV model system is an especially attractive one for molecular pathogenesis studies because TMEV is a relatively simple virus with only four structural proteins in the virion; the nucleotide sequence and deduced amino acid sequence are known for several TMEV strains; the three-dimensional crystal structure of two TMEV strains has been solved; infectious TMEV clones are available, easing the preparation of recombinant or mutated viruses; TMEV-neutralizing monoclonal antibody sites and T-cell epitopes are known; and the experimental and natural host for TMEV, the mouse, is easily manipulated and is well studied immunologically and genetically. An understanding of the determinants of the TMEV disease phenotypes (neurovirulence, virus persistence, restricted infection, demyelination) may be valuable in clarifying the pathogenesis of picornaviral diseases, as well as and non-virus-induced CNS diseases such as amyotrophic lateral sclerosis and multiple sclerosis.
The genome of all picornaviruses contains a long open reading frame that is translated into a polyprotein. One remarkable feature of the TO subgroup strains is that an 18-kDa protein, L*, is synthesized out of frame with the polyprotein from an initiation codon that is 13 nucleotides (nt) downstream from the polyprotein's AUG codon (8) (Fig. (Fig.1A1A).
We previously engineered a mutant virus, called DAL*-1, which had a change of the L* AUG codon to ACG (but with no change in the amino acid sequence of the polyprotein) (4) from our infectious clone pDAFL3 (17) (Fig. (Fig.1B).1B). The DAL*-1 virus had a decreased ability to persist and demyelinate, suggesting that L* is key to the distinctive TO subgroup phenotype. Subsequent studies showed that L* interfered with virus clearance by CD4+ T cells, allowing the virus to persist (10). This work was initially called into question by Michiels and colleagues (21) because a mutant DA virus, OV48, with the same mutation as DAL*-1 (as well as an additional replacement of an AUG codon, in the fifth codon of the L* reading frame, to ACG) had no effect on the ability of the DA virus to persist or demyelinate; however, this mutant was engineered from a different full-length DA clone, pTMDA1 (11). In a subsequent study, they further investigated the importance of L* in virus persistence and demyelination by demonstrating that an L* mutant that contained a stop codon in the L* sequence (with no change in the polyprotein's amino acid sequence) had a significant decrease in virus persistence and demyelination (22). We suspected that the contrasting disease phenotypes induced by the DAL*-1 and OV48 viruses might have resulted from a sequence difference(s) in L*. In the present study, we demonstrate that a change in L* of nucleotide (nt) 1356/amino acid 93 led to the change in phenotype of DAL*-1 versus OV48.
BHK (baby hamster kidney) cells were used for transfection, plaque assays, and the growth of viruses as previously described (4). P388D1 cells, a mouse macrophage cell line, were propagated on RPMI 1640 medium containing 15% fetal bovine serum, 2 mM l-glutamine, and 0.01% gentamicin.
The template used to construct the wild-type and mutant DA viruses was either the full-length infectious cDNA clone of the DA strain known as pDAFL3 (17) or another full-length infectious cDNA clone of DA known as pTMDA1 (11, 12). The DAFL3 virus is a wild-type DA virus prepared from pDAFL3. The TMDA virus (a gift of T. Michiels) is a wild-type DA virus prepared from pTMDA1. The DAL*-1 virus (4) is a DAL*-1 virus that was engineered from pDAFL3 with an ACG codon replacing the L* AUG codon. The OV48 virus (22) (a gift of T. Michiels) is an OV48 virus that was engineered from pTMDA1 with an ACG codon replacing the L* AUG codon, as is the case with DAL*-1, but also with replacement of a second AUG with ACG in the fifth codon of L*. The DAL*-1S93L virus is identical to DAL*-1, except for a change of the C at nt 1358 to a U. In order to generate the DAL*-1S93L virus, a QuikChange XL site-directed mutagenesis kit (Stratagene, La Jolla, CA) was used on a subgenomic plasmid that included the L region; the forward primer used was GGTAACGAAGGGGTCATTATTAACAACTTCTATTCCAATCA, and the reverse primer was TGATTGGAATAGAAGTTGTTAATAATGACCCCTTCGTTACC. The mutated plasmid was digested with MscI and NotI, and the fragment containing the mutation was ligated into MscI/NotI-digested DAL*-1. The OV48L93S virus is identical to OV48, except for a change of the U at nt 1358 to a C. A method similar to that described above was used in order to generate the OV48L93S virus, except with different forward (GGTAACGAAGGGGTCATTATTAACAACTTCTATTCCAATCA) and reverse (TGATTGGAATAGAAGTTGTTAATAATGACCCCTTCGTTACC) primers. The DAL*stop virus has an A rather than a U at nt 1122, creating a stop codon in the coding region of L* and predicting the synthesis of a 14-amino-acid-long L* with no change in the polyprotein's coding region. Mutations were introduced by a two-step PCR amplification of pDAFL3 with forward primer 5′-CGCGTACGAGCGGTAAGTC-3′ and reverse primer 5′-GCAGCGGAACCCCCCTCC-3′. A PmlI/XhoI-digested fragment of the amplified fragment containing the engineered mutations was ligated to an XhoI/PmlI fragment containing the rest of the pDAFL3 clone.
As previously described, viruses were generated from transfection of BHK-21 cells with RNA derived from the in vitro transcription of XbaI-linearized plasmids using T7 RNA polymerase (17). To confirm the presence of the engineered mutation, a region of the viral RNA between nt 867 and 1607 was amplified by reverse transcriptase PCR (RT-PCR) using forward primer CTCTAGTGAACCCTTGAATGGC and reverse primer CAGAGTCGACCAACAGTAGATTGGG, and the amplified region from nt 867 to nt 1500 was then sequenced using the forward primer.
Weanling SJL mice (Jackson Laboratory, Bar Harbor, ME) were inoculated intracerebrally (i.c.) with 0.03 ml containing the designated amount of virus. Mice were sacrificed at various times, and CNS tissue was harvested. CNS tissue was processed for quantitation of the amount of virus genome or infectious virus (see below). In some cases, animals were perfused with 4% paraformaldehyde 6 weeks postinfection (p.i.), the time when DA-induced demyelination is robust; spinal cord sections were paraffin embedded and stained with hematoxylin and eosin. The amount of white matter pathology was graded prominent if white matter abnormalities (disruption of the white matter, inflammation) were seen in many areas of the spinal cord.
CNS tissue was harvested from mice at 7 days p.i. and homogenized in Hanks balanced salt solution. BHK cells in 35-mm-diameter dishes were adsorbed in duplicate for 1 h with infected cell lysates or serially diluted CNS homogenates. Cells were incubated for 2 days at 37°C in 5% CO2. The monolayers were then fixed and stained with bromophenol blue, and the plaques were counted.
To quantitate the amount of virus genome, RNA was extracted from homogenates of brains and spinal cords of mice that had been inoculated with wild-type or mutant viruses at 1 and 6 weeks p.i. using the RNeasy MIDI KIT (Qiagen, Valencia, CA). Real-time quantitative PCR (qPCR) was performed using the Superscript III Platinum two-step qPCR kit with SYBR green (Invitrogen, Carlsbad, CA). A region of the viral RNA between nt 1485 and 1684 was amplified using forward primer TACTATGGCACCTCTCCTCTTGGA and reverse primer CAGCCGCAAGAACTTTATCCGTTG. The hypoxanthine phosphoribosyltransferase (HPRT) gene was used as a housekeeping gene for normalization and determination of the quality of the total RNA. Forward primer GCGTCGTGATTAGCGATGATGAACC and reverse primer GCCTCCCATCTCCTTCATGACATCT were used to amplify a 158-nt fragment of the HPRT mRNA; this primer pair did not amplify HPRT genomic DNA. Serial dilutions of the pDAFL3 plasmid and pHPRT were used to extrapolate the amount of virus genome present in the spinal cords and brains of infected mice. An ABI PRISM 7700 Sequence Detection System (Applied Bioscience, Foster City, CA) was used to run the samples.
P388D1 cells were infected at a multiplicity of infection (MOI) of 10. The cells were harvested at 12 h p.i., fixed with 4% paraformaldehyde for 10 min at 37°C, and then permeabilized with 100% ice-cold methanol for 30 min at 4°C. The cells were blocked with phosphate-buffered saline-0.5% bovine serum albumin and incubated for 1 h with biotinylated anti-VP1 monoclonal antibody (GDVIImAb2, which reacts against all TMEV strains ) and rabbit anti-cleaved caspase 3 (Asp175) antibody (Alexa Fluor 488 conjugate; Cell Signaling Technologies, Danvers, MA). Cells were then washed and incubated with streptavidin-allophycocyanin (BD Pharmingen, San Jose, CA) and fluorescein isothiocyanate-conjugated goat anti-rabbit immunoglobulin G (IgG [H+L]; Chemicon International, Temecula, CA). Samples containing 105 cells were sorted using a BD FACSCanto flow cytometer (BD Bioscience) and analyzed using the Flowjo flow cytometry analysis software (Tree Star Inc., Ashland, OR).
BHK cells were infected at an MOI of 1 and harvested at 16 h p.i. Cells were then lysed (50 mM β-glycerophosphate, pH 7.3, 1.5 mM EGTA, 1.0 mM EDTA, 0.1 mM sodium vanadate, 1.0 mM benzamidine, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 2.0 μg/ml pepstatin A, 1.0 mM dithiothreitol). Immunoprecipitation was performed in lysis buffer (20 mM HEPES, pH 7.5, 150 mM NaCl, 10% glycerol, 1% Triton X-100 containing a protease inhibitor cocktail [Sigma]) and immunoprecipitated with rabbit anti-L* antibody (19) (courtesy of Y. Ohara) and 50 μl of protein G-Sepharose 4B (Invitrogen, Carlsbad, CA; Santa Cruz Biotechnologies, Santa Cruz, CA). Immunoprecipitated proteins were subjected to 12% sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) and Western blotted with anti-L* antibody, followed by horseradish peroxidase-conjugated anti-rabbit IgG antibody (Cell Signaling Technologies, Beverly, MA).
BHK cells were infected at an MOI of 5 with various viruses or transfected with a pcDNAL*-myc/his construct and incubated in phosphate-free DMEM (Invitrogen, Carlsbad, CA). Three hours later, 1 mCi 32Pi (Perkin-Elmer, Boston, MA) was added to the medium. Cells were harvested 13 h later and lysed (20 mM HEPES, pH 7.5, 150 mM NaCl, 10% glycerol, 1% Triton X-100) containing phosphatase and a protease inhibitor cocktail (Sigma). Immunoprecipitation was performed in lysis buffer with protease and a phosphatase inhibitor cocktail (Sigma) using anti-L* antibody (in the case of infected cells) or rabbit anti-myc antibody (Cell Signaling Technologies) (in the case of transfected cells) and 50 μl of protein G-Sepharose 4B. The immunoprecipitated proteins were subjected to 15% SDS-PAGE. The gels were then dried and exposed to X-ray film for autoradiography.
As noted above, mutant viruses generated from pDAFL3 and pTMDA1 that had an ACG codon replacing the L* AUG initiation codon exhibited very different phenotypes. Following DAL*-1 virus infection, mice had little evidence of virus persistence or demyelination, while infection with OV48 (which is similar to DAL*-1 but is derived from a different parental wild-type clone, pTMDA1) led to levels of persistent virus and demyelination that were similar to those of the wild-type parental virus. Because of these differences, we wondered whether there might be a sequence difference in the L* coding region between pDAFL3 and pTMDA1 that was responsible for the different phenotypes.
When we compared the L* sequences of the two full-length infectious clones, we found that pDAFL3 had a C while pTMDA1 had a T at nt 1356, resulting, respectively, in Ser or Leu as the predicted amino acid at position 93 in the L* coding region (Fig. (Fig.1B),1B), which is the sole difference between DA L* and TMDA L*. There was no change in the polyprotein's amino acid sequence despite this change in the L* coding region because the L* sequence is out of frame with the polyprotein and because this nucleotide change (which is in the second base of the L* codon) is in the third base of the leader coding region. As one would expect, the L* sequence difference seen in pDAFL3 versus pTMDA1 was also present in viruses that were generated from these full-length clones.
We questioned whether the difference in phenotype seen following infection with the DAL*-1 virus versus the OV48 virus was a result of the sequence difference in L*. In order to clarify the importance of this sequence difference, we investigated a number of viruses (Fig. (Fig.1),1), (i) two wild-type viruses, DAFL3 and TMDA, (ii) two L* mutant viruses generated from DAFL3 and OV48, respectively, DAL*-1 and OV48, which had an ACG codon replacing the L* AUG initiation codon, and (iii) two L* mutant viruses derived from the latter two mutants with an additional change at nt 1356: DAL*-1S93L (which has a C instead of a T at nt 1356) and OV48L93S (which has a T instead of a C at nt 1356).
The wild-type and mutant viruses described above were inoculated i.c. into weanling mice that were sacrificed at 6 weeks p.i. Because the amount of infectious virus in the spinal cord at 6 weeks p.i. was predicted to be very low and possibly undetectable, the amount of DA genome was analyzed by qPCR rather than attempting virus isolation and quantitation.
The amount of virus genome at 6 weeks following infection with the wild-type and mutant viruses was evaluated by qPCR (Fig. (Fig.2).2). There were large amounts of virus genome in the spinal cord following infection with the two wild-type viruses, DA and TMDA. As previously described, we found that (i) mice infected with the DAL*-1 virus had substantially less virus genome than did those infected with the corresponding parental wild-type virus (P ≤ 0.05) (4) and (ii) mice infected with the OV48 virus, which was derived from TMDA, had levels of virus genome that were comparable to those of mice infected with the corresponding parental wild-type virus (21). Of interest, the spinal cords from mice inoculated with the DAL*-1S93L virus (which is the DAL*-1 virus with a second mutation at nt 1356) had an increased level of virus genome compared to mice inoculated with the DAL*-1 virus (P ≤ 0.01) and a level similar to that seen with the wild-type DA virus. Similarly, the spinal cords from mice inoculated with the OV48L93S virus (which is OV48 with an additional mutation at nt 1356) had a decreased level of virus genome compared to those infected with OV48 and TMDA (P ≤ 0.01), comparable to that seen with the DAL*-1 virus. These studies suggested that there was a decrease in virus genome that persisted for 6 weeks p.i. in the case of DA mutant viruses that had an ACG codon replacing the L* AUG codon and a T at nt 1356.
Although nt 1356 had a clear influence on virus genome persistence seen 6 weeks following infection with mutant viruses that had an ACG codon replacing the L* AUG codon, there was no significant difference seen in the amount of virus genome present at 6 weeks p.i. when the two wild-type viruses were compared (even though the DA virus has a C at nt 1358 while TMDA has a T). In other words, the influence of the L* sequence difference on genome persistence is only apparent when L* is synthesized in a significantly reduced amount (i.e., following translation from an ACG initiating codon in the DAL*-1 or OV48 virus) and not when it is synthesized more abundantly, as in the wild-type virus.
The histopathology of the mouse spinal cord following infection with the wild-type and mutant viruses was evaluated at 6 weeks p.i. (Table (Table1).1). As expected, the spinal cords of mice infected with the two wild-type viruses had evidence of inflammation and disruption of the white matter that was prominent in virtually all of the inoculated animals. Also, as previously described (4), the spinal cords of mice infected with the DAL*-1 virus had only rare evidence of pathology, such as a single focus of perivascular lymphocyte cuffing. In contrast, the spinal cords of mice infected with the OV48 virus had evidence of white matter disease similar to that seen after infection with the respective parental wild-type virus (21). The spinal cords of mice infected with the DAL*-1S93L virus (in which the C at nt 1356 of the DAL*-1 virus was changed to a T) led to enhanced white matter pathology at 6 weeks p.i., similar to that seen with the wild-type virus. The OV48L93S virus (in which the T at nt 1356 was changed to a C) had minimal evidence of white matter pathology, similar to that seen with the DAL*-1 virus. We assume that the reason that there are differences in the histopathology of the different mutant viruses is related to the amount of virus that persists (Fig. (Fig.2),2), since the degree of histopathology in the different virus groups was correlated with the amount of virus genome. The importance of virus persistence to demyelination is clear since DA-induced demyelination in susceptible mouse strains is always associated with the presence of virus in the CNS, while mouse strains resistant to demyelination clear the virus. The amount of viral antigen detected by immunohistochemical staining with an anti-VP1 monoclonal antibody correlated with the pathology and had a similar localization in the different neural cell types in the spinal cord (data not shown).
Thus, DA L* nt 1356/amino 93 is key to the phenotypic differences seen following inoculation of the DAL*-1 versus the OV48 virus, with respect to both the amount of virus genome and the white matter pathology seen at 6 weeks p.i.
We questioned whether the reason for the decrease in virus genome and attenuated disease following infection with the DAL*-1 or OV48L93S virus compared to the wild-type parental virus was related to a decrease in the initial growth of virus after infection. For this reason, we performed qPCR and infectivity assays on brains harvested at 1 week p.i. with the two wild-type parental viruses and the four L* mutant viruses. Figure Figure3A3A shows that there was a similar amount of virus genome in the CNS 1 week after infection with each of these viruses. To confirm this finding, 10% brain homogenates were assayed for infectious virus at 1 week p.i. Consistent with the results of the qPCR study, the amount of infectious virus was similar at 7 days p.i. with each of these viruses (Fig. (Fig.3B).3B). These findings suggest that the effect of nt 1356/amino acid 93 on the level of virus genome at 6 weeks p.i. with DAL*-1 and OV48 occurs subsequent to 1 week p.i. The findings also demonstrate that the reason that there is substantially less virus genome in mice infected with L* mutant viruses is not that the viruses do not initially grow well in the mouse CNS, since there are no differences in the levels of virus genome and infectious virus at 1 week p.i. with either the wild-type viruses or the L* mutant viruses.
Macrophages are important cells in TMEV-induced demyelinating disease since they are one of the main sites for virus persistence and because apoptosis of these cells may play a role in virus persistence (18, 20). We previously found that L* has antiapoptotic activities in cultured macrophages that might facilitate the growth of DA in these cells (6, 19); however, it remained unclear whether this activity of DA L* was correlated with its effect on virus persistence. The availability of a number of L* mutant viruses that induced various levels of persistence at 6 weeks p.i. (Fig. (Fig.2)2) allowed us to test whether the ability of the virus to persist was correlated with antiapoptotic activity in infected cultured macrophages.
In order to determine whether the antiapoptotic activity of L* was correlated with virus persistence, we measured the amount of cleaved caspase 3 following the infection of a mouse macrophage cell line, P388D1. Almost all of the cells were infected at 12 h p.i., as demonstrated by the presence of DA virus VP1 antigen (Fig. (Fig.4A).4A). As previously described (6), wild-type DA virus-infected cells had lower levels of apoptosis than DAL*-1 virus-infected cells (P ≤ 0.05) (Fig. (Fig.4B).4B). The level of apoptosis seen with DAL*-1S93L was similar to that seen with DAL*-1, despite the change in nt 1356/amino acid 93. Similar levels of apoptosis were seen with TMDA, OV48, and OV48L93S. These results suggest that a change in nt 1356/amino acid 93 does not affect the antiapoptotic activity of L*, and therefore, the antiapoptotic activity of L* is not correlated with its effect on virus persistence and demyelination. The reason for the differences between the antiapoptotic activity of DAL*-1 and DAL*-1S93L and that of OV48 and OV48L93S remains unclear.
A previous study found that a small amount of L* was produced in viruses derived from TMDA with an ACG initiation codon (22). We questioned whether the difference in the phenotypes of the DAL*-1 and OV48 viruses might result because these viruses synthesize different amounts of L*. Figure Figure55 shows a Western blot analysis of lysates of cells infected with similar amounts of two wild-type parental viruses, the four L*-1 mutant viruses, and the DAL*stop virus; because L* levels are low following infection, the lysates were immunoprecipitated with anti-L* antiserum and then immunostained with anti-L* antiserum. As expected, (i) no L* was apparent following infection with the DAL*stop virus (Fig. (Fig.5,5, lane 8) and (ii) compared to infection with the wild-type parental viruses (Fig. (Fig.5,5, lanes 2 and 5), decreased amounts of L* were made following infection with mutant viruses that had an ACG codon as their L* initiation codon (Fig. (Fig.5,5, lanes 3, 4, 6, and 7). There were only slight inconsistent differences in the amount of L* found in lysates infected with the four L* mutants, suggesting that a Leu or Ser at amino acid 93 does not have a reproducible effect on the amount of L*. These results indicate that the difference in the phenotypes of the L* mutants is not due to a variation in the amount of L* that is present in the infections.
The finding that the only difference between the L* proteins of the two wild-type parental viruses is a Ser at position 93 made us question whether this amino acid is phosphorylated. For this reason, BHK-21 cells radiolabeled with 32P were infected with the DA, TMDA, and DAL*stop viruses. Because levels of L* are low, cell lysates were immunoprecipitated with anti-L* antibody and the immunoprecipitated proteins were then subjected to SDS-PAGE. Autoradiograms showed that there was an immunoprecipitated protein with the predicted electrophoretic mobility of L*, 18 kDa, in DA-infected lysates (Fig. (Fig.6A,6A, lane 2); this band was not seen in lysates from DAL*stop virus infections (Fig. (Fig.6A,6A, lane 3). Of interest, there tended to be less intense radiolabeling of the L* protein in the immunoprecipitated TMDA virus-infected cell lysates (Fig. (Fig.6A,6A, lane 4). These results suggest that (i) DA and TMDA L* is phosphorylated during infection and (ii) there is a difference in the levels of phosphorylation of the L* protein of the DA and TMDA viruses, perhaps because a putative additional phosphorylation site at the Ser at position 93 in DA was changed to Leu in the case of TMDA. It is also possible that Ser93 of DA L* is not phosphorylated, but that another amino acid(s) in L* is phosphorylated and that this phosphorylation is influenced by another viral protein that differs in sequence in DA versus TMDA; however, this possibility is less likely since Ser93 is one of the highest predicted sites for phosphorylation (1). In order to confirm the phosphorylation of L*, 32P-radiolabeled cells were transfected with myc-tagged L*. The lysates were immunoprecipitated with anti-myc antibody, and the immunoprecipitated proteins were subjected to SDS-PAGE. An autoradiogram of these lysates showed a radiolabeled band with the predicted mobility of L*myc (Fig. (Fig.6B,6B, lane 1).
The DA strain and other members of the TO subgroup of TMEV provide a valuable model system to explore the pathogenesis of a virus-induced immune-mediated demyelinating disease (for a review, see reference 16). We previously found that there is an alternative initiation site in the RNA genome of TO subgroup strains (but not GDVII subgroup strains) that is used to synthesize an 18-kDa protein, L*, out of frame with the polyprotein (8). Mice infected with a DA L* mutant virus (derived from the pDAFL3 infectious clone) that has a replacement of the L* AUG initiation codon with an ACG codon (4) or has a stop codon in the L* reading frame (22) with no change in the predicted amino acid sequence of the polyprotein have minimal evidence of virus persistence or white matter disease. We subsequently showed that L* was important in interfering with the clearance of virus by CD4+ T cells (10). Our studies with the DAL*-1 virus were initially called into question because the OV48 virus, which has the same mutation in the L* initiation codon as DAL*-1 (but is derived from a different full-length infectious clone of DA, pTMDA1) persisted and demyelinated to a similar level as the wild-type TMDA virus (21). In the present study, we report a sequence change in the L* coding region so that the DAFL3 virus has a C at nt 1356 and a Ser at amino acid 93 while the TMDA virus has a T at nt 1356 and a Leu at amino acid 93. Interestingly, all nine TO subgroup strains of TMEV, except for DA, have a T as nt 1356 and a Leu as amino acid 93 (13).
We investigated the importance of the L* sequence change in determining the phenotypes of the DAL*-1 and OV48 viruses. We analyzed the properties of two wild-type parental viruses, DAFL3 and TMDA, and four mutant viruses derived from these two parental viruses; the mutant viruses had replacements of the L* AUG initiation codon with an ACG codon and a varying nucleotide (T or C at 1356, giving a Ser or Leu at amino acid position 93). A mutant virus that had a replacement of the L* AUG codon along with C at nt 1356/Ser at amino acid 93 had lower levels of virus genome persisting at 6 weeks p.i. and was not able to induce significant white matter pathology compared to the respective wild-type virus and the same mutant virus with T at nt 1356/Leu at amino acid 93. Conversely, a mutant virus that had an ACG codon replacing the L* AUG initiation codon along with T at nt 1356/Leu at amino acid 93 behaved similarly to the respective wild-type DA virus, with persistence and the ability to induce white matter pathology. These findings demonstrate that nt 1356/amino acid 93 affects the function of L* in the DAL*-1 and OV48 viruses.
We previously reported that L* has an antiapoptotic activity and fosters DA virus grown in infected cultured macrophage cell lines (6, 19). Additional evidence supporting this activity of L* has been previously described (7). We questioned whether the influence of L* on virus persistence and white matter pathology was related to its antiapoptotic activity in macrophages. Our studies, however, failed to find any correlation between the antiapoptotic activity of L* and its ability to persist and induce white matter pathology. This lack of correlation is not surprising since Van Eyll and Michiels found that OV48 persisted and induced demyelination similarly to the TMDA wild-type virus, although the two viruses had different effects with respect to facilitating the infection of macrophages (21). Of note, a recent publication (5) reported that another protein of TMEV, L, whose sequence overlaps that of L*, results in apoptosis in transfected cells. It may be that the antiapoptotic effect of L* in the DA and TMDA viruses is influenced by sequence differences in DA versus TMDA in L or another viral protein or the 5′ untranslated region (UTR).
How can a nucleotide or amino acid modify the function of L* if L* lacks an AUG initiation codon? We corroborated the previous findings of van Eyll and Michiels (21) showing that small amounts of L* are synthesized despite the replacement of the L* AUG initiation codon with an ACG codon; the ability of an ACG codon to be used as an initiation codon, albeit inefficiently, has been reported in the case of other proteins, including viral proteins (2). We found that similar low levels of L* are synthesized in BHK cells from viruses that have an ACG codon used as the L* initiation codon and either T or C at nt 1356; it may be, however, that differences in L* production occur in other cells, such as neural cells, depending on the presence of this nucleotide difference.
Interestingly, our results showed that the sequence difference at nt 1356/amino acid 93 is only critical to the function of L* with respect to virus persistence and the resultant white matter pathology when this sequence is present in a virus RNA genome which has an ACG codon replacing the L* initiating AUG codon; i.e., when small amounts of L* are made. Once large amounts of L* are made, as in the case of the DA and TMDA viruses, the sequence difference at nt 1356/amino acid 93 no longer significantly influences virus persistence or the disease phenotype.
A question that arises from our study is whether the effect of the L* sequence difference on virus persistence and demyelination is at the nucleotide or amino acid level. The fact that a stop codon in the L* sequence (with no change in the amino acid sequence of the polyprotein) causes a decrease in virus persistence and demyelinating activity of the virus (22) suggests that L* amino acid residue 93, rather than the nucleotide sequence, is the critical determinant influencing the activity of L*.
Although we identified the importance of nt 1356/amino acid 93 in DA persistence and demyelination, the reasons for this effect remain unclear. The fact that a serine is present at residue 93 in the DA, but not the TMDA, virus raised the possibility that phosphorylation of L* at this amino acid is important in the function of L*. Our studies demonstrated that L* is phosphorylated and also suggested that the L* in DA may be more phosphorylated than TMDA. It is possible that an enhanced phosphorylation of L* in DA changes its activity when only a small amount of L* is produced, as in the case of the DAL*-1 mutant virus. Another possibility, perhaps related to the phosphorylation of amino acid 93, is that the presence of Ser versus Leu affects the binding of L* to other partners and that this interaction is important for the function of L*.
Of interest was our finding that the level of virus genome at 1 week p.i. is far less than that seen at 6 weeks p.i. despite the fact that there is 105 PFU of virus/ml in the CNS at 1 week p.i. and (as previously reported [3, 15]) barely detectable levels of virus in the CNS at 6 weeks p.i. A high number of TMEV viral RNA copies in the face of restricted virus gene expression has been previously noted in the CNS of demyelinated animals (9, 20); the reason for this restriction remains unclear. One of the possible reasons for the restriction could be that oligodendrocyte- or macrophage-specific RNA binding proteins direct translation initiation to the AUG start codon for L* rather than the polyprotein, limiting the amount of infectious virus that can be produced.
This work was supported by grants from the National Institutes of Health (R.P.R.) (1RO1 463 NS37958-07), the National Multiple Sclerosis Society (R.P.R.), and the Multiple Sclerosis Foundation (S.S.).
Published ahead of print on 18 November 2009.