Differential Glycosylation of the Cas-Br-E Env Protein Is Associated with Retrovirus-Induced Spongiform Neurodegeneration

Abstract

The wild mouse ecotropic retrovirus, Cas-Br-E, induces progressive, noninflammatory spongiform neurodegenerative disease in susceptible mice. Functional genetic analysis of the Cas-Br-E genome indicates that neurovirulence maps to the env gene, which encodes the surface glycoprotein responsible for binding and fusion of virus to host cells. To understand how the envelope protein might be involved in the induction of disease, we examined the regional and temporal expression of Cas-Br-E Env protein in the central nervous systems (CNS) of mice infected with the highly neurovirulent chimeric virus FrCas^E. We observed that multiple isoforms of Cas-Br-E Env were expressed in the CNS, with different brain regions exhibiting unique patterns of processed Env glycoprotein. Specifically, the expression of gp70 correlated with regions showing microglial infection and spongiform neurodegeneration. In contrast, regions high in neuronal infection and without neurodegenerative changes (the cerebellum and olfactory bulb) were characterized by a gp65 Env protein isoform. Sedimentation analysis of brain region extracts indicated that gp65 rather than gp70 was incorporated into virions. Biochemical analysis of the Cas-Br-E Env isoforms indicated that they result from differential processing of N-linked sugars. Taken together, these results indicate that differential posttranslational modification of the Cas-Br-E Env is associated with a failure to incorporate certain Env isoforms into virions in vivo, suggesting that defective viral assembly may be associated with the induction of spongiform neurodegeneration.

The appearance of spongiform neurodegeneration in the mammalian central nervous system (CNS) represents a unique pathologic picture typically associated with infection by either unconventional protein infectious agents (prions) or retroviruses. While little is known at the cellular level about how prions induce vacuolar lesions, a detailed picture of how retroviruses induce spongiform pathology is emerging from the analysis of murine leukemia virus (MuLV) models. The best studied of the neurovirulent murine retroviruses is the wild mouse ecotropic virus, Cas-Br-E, which was discovered by Gardner and coworkers in a population of feral mice (13). CNS infection by this virus results in vacuolar changes in motor areas from the cortex through the spinal cord and is manifest clinically, first, as tremulous paralysis of the hindlimbs, progressing to the forelimbs, with associated wasting and eventual death. The appearance and severity of clinical signs and lesions correlates with the level of Cas-Br-E virus infection in the CNS (5), although neurodegeneration cannot occur prior to the 2nd postnatal week no matter how significant the viral load (25). Interestingly, the primary degenerating elements, the motor neurons, are not infected, indicating that neurologic disease is mediated by an indirect mechanism (16, 19, 24). While multiple CNS cell populations are infected, it is microglial infection which specifically correlates with regions of motor neuronal degeneration in vivo (1, 2, 16, 24). Furthermore, CNS transplantation of Cas-Br-E-infected microglia alone is sufficient to induce spongiform neuropathology (26).

Since genetic mapping analysis has demonstrated that the primary determinants for neurovirulence reside within the env gene (8, 30, 31, 38), much of the focus on mechanisms of MuLV-induced neuropathogenesis have centered on the viral envelope protein, the membrane-associated surface glycoprotein which mediates virus binding and entry into the cell. Interestingly, however, neither the expression of high levels of Cas-Br-E envelope protein alone nor production of replication-restricted Cas-Br-E virus is capable of precipitating acute pathological changes in the brain, when either protein or virus is expressed from cells of neuroectodermal origin (27). Rather, our results indicate that late Cas-Br-E virus replication events within the bone marrow-derived microglia are required for inducing neurodegenerative disease. These findings raise the possibility that a unique neurotoxic Env protein is generated upon microglial infection. In this regard, additional Cas-Br-E envelope protein isoforms have been observed when the Cas-Br-E virus spreads to microglia from transplanted Cas-Br-E-infected neural stem cells (27). The unique envelope isoforms observed within the CNS may either be byproducts of the coincident neurodegenerative process or represent Env synthetic events within microglia involved in the precipitation of neuropathogenesis.

How envelope protein synthesis in microglia could be involved in the induction of neurodegeneration is not yet known. Understanding the neuropathogenic process may come from understanding Env biosynthesis. Analysis of MuLV retroviral Env protein synthesis and trafficking in cells in culture (reviewed in reference 10) indicates that envelope is synthesized in the rough endoplasmic reticulum as a precursor protein, where it has its amino-terminal signal sequence cleaved off, undergoes disulfide bonding, obtains multiple asparagine-linked high-mannose sugars, and oligomerizes, prior to transport to the Golgi apparatus. In the Golgi apparatus, the high-mannose sugars are modified to complex type, and the precursor protein polypeptide backbone is cleaved to give rise to the surface-expressed domain (SU) and the transmembrane-associated domain (TM); then the complex is transported to the plasma membrane. SU and TM remain associated by way of noncovalent interactions and in some instances, a disulfide bond (15). Upon assembly into virions and release from the cell, the carboxy-terminal cytoplasmic tail of the TM is cleaved, which renders the SU-TM complex within the virus fusigenic (33). Functional mapping studies have demonstrated that the receptor binding function of envelope resides in the N-terminal half of SU (17), while TM contains the regions responsible for oligomerization, membrane attachment, and fusigenicity.

In an effort to understand whether Env protein synthesis could be involved in neuropathogenesis, Wong and coworkers compared envelope synthesis and processing between nonneurovirulent wild-type Moloney murine leukemia virus (MoMuLV) and a temperature-sensitive MoMuLV mutant, ts1, which is neurovirulent. Interestingly, in a spleen-derived cell line (TB cells), the cleavage of the envelope precursor protein (gpr80) to SU and TM (gp70 plus p15E) was demonstrated to be inefficient for the neurovirulent ts1 virus. Similar results were also observed in primary cultures of CNS glia (36) and, more recently, in immortalized astrocyte cultures (22). Significantly, this inefficient envelope precursor protein processing genetically segregates with the ability to cause clinical neurological disease in vivo (37, 38).

Analysis of Cas-Br-E envelope synthesis in primary CNS glial cultures indicates normal Cas-Br-E Env processing and virus production. Interestingly, however, infected microglia isolated from the mixed glial cultures appear incapable of processing the Cas-Br-E envelope precursor (gpr85) to SU (gp70) and TM (p15E) (23). This defect is associated with intracellular budding of viral particles, similar to that seen in human immunodeficiency virus-infected monocytes in vitro (14). In contrast, defective budding of virus into microglia in vivo has yet to be reported. However, what has been observed is the appearance of a unique CNS-specific envelope protein isoform which has a smaller apparent molecular size (65 kDa) than that noted in infected spleen (70 kDa) (6, 23, 25). Whether this unique envelope protein species is related to the defective envelope processing noted in vitro and/or to neuropathogenesis in vivo has not been addressed.

Therefore, to understand how envelope protein synthesis might be involved in the induction of neurodegeneration, we examined the expression of envelope protein in the brains of mice infected with the highly neurovirulent chimeric Cas-Br-E retrovirus, FrCas^E. This virus contains the Cas-Br-E env gene in the background of Friend leukemia virus, clone FB29 (30). It was used because it infects the CNS at high levels and induces neurodegenerative disease with a rapid and stereotypic progression (6, 24, 30). Our results revealed the presence of both regional and temporal Cas-Br-E envelope protein isoform heterogeneity. This envelope heterogeneity could be accounted for by unique N-linked Env glycosylation events as a result of the regional and temporal infection of different CNS cell types. Furthermore, our analysis specifically implicates non-virion gp70 envelope isoforms as being associated with the appearance of spongiform neuropathology.

Differential expression of Cas-Br-E envelope protein in a region- and time-specific manner.

We have previously reported that envelope protein expression in mice infected with the highly neurovirulent chimeric Cas-Br-E retrovirus, FrCas^E, differs qualitatively between the brain and spleen (6, 23, 25). Extracts from both brain and spleen show significant levels of the envelope precursor protein, gpr85; however, the two tissues differ in the nature of the proteolytically processed envelope proteins expressed. Splenic infection by FrCas^E is characterized by expression of a single proteolytically processed envelope isoform referred to as gp70. In contrast, brain extracts are characterized by the appearance of multiple proteolytically processed envelope protein isoforms, with the dominant species migrating with an apparent molecular size of 65 kDa. Because FrCas^E infection of the brain occurs in diverse areas and in a variety of cell types (6, 24), our prior analyses of whole-brain homogenates constituted a “biochemical average” of the envelope proteins being expressed (6, 23, 25). This is a significant issue because the neuronal infection noted in the FrCas^E disease model is not accompanied by cytopathic changes and occupies physically distinct regions of the brain. We reasoned, then, that dissection of regions of high neuronal infection away from the regions of high microglial infection (areas with extensive pathology) should reveal features of Cas-Br-E Env expression which are relevant to neuropathogenesis.

Thus, various regions were dissected from the brains of FrCas^E-infected mice at 14 days postinoculation (dpi), the time when infected mice initially expressed clinical neurologic disease. A diagram indicating the gross brain dissection is shown in Fig. 1A, with regions roughly corresponding to the spinal cord, the medulla and pons (referred to collectively as the brainstem), the colliculus, the thalamus, the cerebellum, the olfactory bulb, and the neocortex which includes the cerebral cortex, hippocampus, and striatum). Extracts from the dissected CNS regions and the spleen were evaluated for Cas-Br-E Env expression by Western immunoblotting using monoclonal antibody 697 (28) as described previously (6, 27). The results, shown in Fig. 1B, reveal the presence of multiple isoforms of the Cas-Br-E Env in a region-specific manner. Of particular note were the two areas characterized by very high neuronal infection and little extravascular microglial infection, namely, the olfactory bulb and the cerebellum. The Env proteins observed in these regions included gpr85 precursor protein and a processed Env protein with an M_r of 65 kDa. Notably absent or in vanishingly small quantities was protein migrating on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels with a mobility consistent with gp70. In contrast, samples from the brainstem, spinal cord, colliculus, thalamus, and neocortex contained additional envelope protein species which migrated between the gpr85 precursor and the 65 kDa protein isoform. These intermediate Env proteins are collectively referred to as gp70's. Interestingly, the brain regions where gp70's were observed constitute the regions where microglial infection predominates and spongiform neurodegeneration occurs. Thus, the results specifically implicate the gp70 isoforms in the pathogenic process.

FIG. 1.

Regional and temporal expression of Cas-Br-E envelope protein in the brain. (A) Saggital diagram of the mouse brain. Gray shading indicates regions where pathology arises after infection with the FrCas^E virus. Heavy dark lines approximate where brains were dissected for regional Cas-Br-E envelope protein analysis by Western immunoblotting. (B) Cas-Br-E envelope protein immunoblot on dissected brain regions, compared with spleen and purified virus made in dunni fibroblasts. Spleen and CNS samples were generated from FrCas^E-infected mice at 14 dpi, and protein sample loads and blot exposure times were adjusted to provide for resolution of the multiple envelope protein isoforms. (C) Time course of CNS expression of the different Cas-Br-E envelope protein isoforms. Samples from two different brains were run for each time point to show the limited variability that occurs during CNS infection. Equivalent sample loads were used for each time course; however, different exposures were used for the different regions to reveal the details of envelope expression. Note that no virus control was loaded for NCX or BS. Abbreviations: OB, olfactory bulb; NCX, neocortex; COL, colliculus; THAL, thalamus; PONS, pons region of the brainstem; MED, medulla oblongata; CB, cerebellum; SC, spinal cord; SPL, spleen; U, uninfected total brain; V, virus from cultured dunni cells; BS, brainstem (pons and medulla together).

Because of the complexity of Cas-Br-E envelope expression observed, the reproducibility of the results was evaluated by Western blot analysis of brains of at least four different mice in four separate experiments. Each experiment reproduced the same region-specific envelope isoform expression patterns shown in Fig. 1 and 2, with little mouse-to-mouse variation. These results are consistent with the stereotypic nature of CNS infection by FrCas^E (6, 24, 30) and suggest that the observed protein heterogeneity was unlikely to be due to random virus mutation and selection.

FIG. 2.

Sucrose gradient sedimentation analysis of envelope protein in FrCas^E-infected tissues. (A) Western blot of sucrose gradient fractions after sedimentation of extracts from different CNS regions and the spleen. Fractions were loaded from left to right, starting with the higher-density fractions on the left. Downward-pointing arrows indicate the fractions with densities of 1.14 to 1.16 g/ml, correlating with fractions expressing the infectivity peak (see panel B). Note that for CNS extracts the 65-kDa protein is the predominant isoform observed associated with virion density. (B) Sedimentation behavior of infectivity in the different CNS regions. Peak infectivity for all regions occurs at 1.14 to 1.16 g/ml, consistent with that noted for virions made in culture and coincident with the 65-kDa protein peaks noted in panel A. The asterisk indicates that virus titers equal to or greater than 300 focus-forming units (FFU)/10 μl could not be distinguished from one another in this analysis. Therefore, these points were placed at 300. Abbreviations: NCX, neocortex; BS, brainstem; CB, cerebellum; SC, spinal cord; SPL, spleen; V, virus from cultured dunni cells.

Our previous analysis of the neurodegenerative disease induced by FrCas^E has demonstrated that CNS infection and the induction of neuropathology proceed in a highly predictable fashion (6). Infection of spleen, bone marrow, and other peripheral tissues is observed within the first few days after intraperitoneal inoculation. Initial CNS infection is apparent by 8 dpi and is associated with vascular elements (endothelia and pericytes). Virus appears to spread from the vasculature to the CNS parenchyma, infecting postnatally mitotic neurons by 10 dpi and infecting parenchymal microglia by 12 dpi. Spongiform pathology arises in multiple CNS areas as early as 10 dpi, while clinical neurological signs of disease first become apparent at 14 to 15 dpi, when spongiform changes are quite extensive.

Therefore, to evaluate whether the appearance of certain Cas-Br-E envelope protein isoforms correlated with the infection of specific CNS cell types and the appearance of status spongiosis, a region-specific kinetic analysis of envelope expression was performed. The results for representative CNS regions and the spleen are shown in Fig. 1C. Because protein equivalents were analyzed in this set of experiments, relative quantitative assessments were made throughout the time course as a means to understand the role of the different isoforms in the pathogenic process. Envelope protein expression in the spleen was observed as early as 4 dpi (not shown), and the pattern observed did not change over the course of the analysis, i.e., equal levels of gpr85 precursor protein and proteolytically processed gp70 were observed throughout the time course. In contrast, protein expression in the CNS initially appeared around 8 dpi in all regions examined and changed as the disease progressed. Specifically, some gp70 envelope isoforms arose first, followed by the appearance of the gpr85 envelope precursor and then the smaller-molecular-size 65-kDa proteins. As the time course progressed, the precursor protein diminished while the gp70's and smaller-molecular-size 65-kDa species increased or remained constant. The gpr85 precursor decrease appeared both in regions with abundant pathology (spinal cord, brainstem, and neocortex) and those without pathology (cerebellum). The envelope protein expression pattern noted in the cerebellum was unique in that it initially appeared as gp70 proteins but quickly shifted to exclusively the precursor protein and 65-kDa isoforms. In contrast, the gp70 class of envelope isoforms peaked at 12 to 14 dpi in the spinal cord, brainstem, and neocortex and then diminished by 16 dpi. Whether this decrease in gp70 Env expression by 16 dpi reflects gp70 conversion to another isoform or degradation is not known, but it correlates with the onset of clinical signs of paralysis and the appearance of extensive spongiform neuropathology in these regions (see reference 6).

The apparent association of the gp70 envelope isoforms with the onset of neuropathology is of considerable interest. We have previously noted that high-level CNS expression of Cas-Br-E gp70 (alone or in the context of a replication-restricted virus) was not sufficient to induce neuropathology in susceptible mice when it was expressed from neural stem cells (27). In these experiments only one processed Env isoform was noted, and it resolved as gp70. In contrast, neuropathology was observed when transplanted neural stem cells served as platforms for delivering Cas-Br-E virus to microglia. In these experiments, where both neural stem cells and microglia were infected, envelope expression was characterized by the presence of multiple gp70 isoforms rather than the singular gp70 species noted in the former experiments. Interestingly, no 65-kDa Env proteins were observed in either of these experiments. Thus, the appearance of multiple gp70 isoforms coincident with microglial infection and pathology noted herein further supports the idea that unique gp70 isoforms play an important role in spongiform neurodegeneration.

Incorporation of CNS-derived Cas-Br-E envelope proteins into virions.

We previously reported that infection of the CNS by FrCas^E appeared to result in viral particle production by all infected CNS cell types upon examination by electron microscopy (24). Given that multiple isoforms of Cas-Br-E envelope protein were observed in various brain regions of FrCas^E-infected animals, we were interested in determining which, if any, of these species were incorporated into virions. Failure of envelope to be incorporated into virions might suggest specific intracellular modifications which could be important for pathogenesis. Therefore, freshly dissected tissue was homogenized by 20 strokes of a Wheaton homogenizer in 10 volumes of 150 mM NaCl–50 mM Tris-HCl (pH 7.4)–0.1 mM EDTA at 4°C. Homogenates were sedimented at 10,000 × g for 10 min at 4°C to pellet nuclei and large cellular fragments. Supernatants (1-ml samples) were further sedimented over 10 ml of 20 to 60% sucrose gradients at 38,000 rpm in a Beckman SW41 rotor for 3 h. Sucrose gradient fractions were collected (0.5 ml each), measured for density by directly weighing 50-μl samples (n = 3), and tested for Gag and envelope protein by dot immunoblot assays on nitrocellulose membranes with rabbit anti-p30 Gag protein and Cas-Br-E-specific anti-gp70 envelope monoclonal antibody 697, respectively (not shown). Fractions positive for Gag and/or envelope proteins were subjected to SDS-PAGE and Cas-Br-E Env protein immunoblot analysis. The results, shown in Fig. 2A, indicate that the 65-kDa protein is the primary CNS envelope protein found sedimenting at a density consistent with that of virions made in cultured fibroblasts (1.14 to 1.16 g/ml). The gp70 envelope protein isoforms from the brain appeared at lower densities, suggesting that they were not associated with virions but rather were associated with other cellular fractions. Interestingly, the gp70 peak noted for extracts taken from the spleens of FrCas^E-infected animals was rather small compared to the level of total gp70 in the tissue extract. This suggests that most of the spleen-associated gp70 protein was not associated with virions. This contrasts remarkably with the results obtained from the cerebellum, where the vast majority of the 65-kDa protein sediments with virions. Thus, the spleen results suggests that either virus is quickly transported away from the spleen parenchyma into the blood once gp70 is incorporated into virus or the infection of cells in the spleen is not highly productive. Since spleen cells from infected mice readily score positive in infectious-center assays (4), these results suggest that cultured target cells may aid the infectious process. A similar phenomenon was observed for Cas-Br-E-infected microglia in culture, where the cells produce little free virus but score 100% positive by infectious-center assay (23). It will be of interest in the future to compare virus production from the spleen after infection with nonneurovirulent viruses to evaluate whether the envelope protein is the primary determinant responsible for the nonproductive phenotype. It may be that the virus replication process occurring in spleen cells closely mimics that occurring in CNS microglia.

To determine whether infectivity cosedimented with the 65-kDa envelope protein peak in the various brain regions, sucrose gradient fractions were tested by a focus assay for infectious virus (5). Figure 2B indicates that infectivity in all CNS regions sedimented with a density consistent with virions generated from tissue culture fibroblasts. Thus, the 65-kDa protein, which dominates the peak infectivity fraction, appears fully capable of mediating viral infection. Perhaps more interesting, however, is the lack of correlation between gp70 and infectivity in the various brain regions. These results suggests that the specific gp70 Env isoforms are not efficiently incorporated into virions, and this may be critical for the induction of neurodegeneration. This could be due to a failure of envelope to traffic to the sites of viral assembly at the plasma membrane, or it could result from an altered envelope structure which precludes virus incorporation. Whether these intermediate isoforms are capable of binding to the ecotropic receptor, MCAT-1, remains to be determined. If receptor binding were significantly altered, then altered trafficking or degradation of such an Env-receptor complex could be involved in alterations in microglial function and subsequent neuropathogenesis.

Differential glycosylation accounts for the complex envelope protein expression patterns in the CNS.

In order to understand what accounted for the appearance of the multiple CNS Env isoforms, protein extracts from the spleen and representative brain regions (6) were treated with recombinant peptide N-glycosidase F (PNGase F; peptide-N⁴-(N-acetyl β-glucosaminyl)asparagine amidase; EC 3.5.1.52 and EC 3.2.2.18; Genzyme) to remove all asparagine-linked (N-linked) sugars and leave the protein backbone. Because the Cas-Br-E envelope protein has 7 potential N-linked glycosylation sites (32), this approach could determine whether the multiple Cas-Br-E Env isoforms observed were due to alterations in the protein backbone or to posttranslational sugar modifications. The deglycosylation reactions were carried out on 10% tissue homogenates from which nuclei had been removed. Samples were denatured by boiling in the presence of 0.1% SDS–1 mM dithiothreitol–200 mM Tris-HCl (pH 8.0) for 5 min and then were incubated with PNGase F (1.3 U/50 μl of extract) for 16 h at 37°C after addition of Triton X-100 to a final concentration of 0.5%. Reactions were terminated by the addition of SDS-PAGE sample buffer and boiling. Immunoblotting of the deglycosylated tissue extracts after separation by SDS–10% PAGE (Fig. 3A) showed that the envelope protein profiles from different brain regions and the spleen were rendered indistinguishable by PNGase F treatment. These results suggest that the multiple gp70 and gp65 isoforms of Cas-Br-E envelope appeared as a result of differential N-linked glycosylation.

FIG. 3.

Glycosylation analysis of Cas-Br-E envelope protein from FrCas^E-infected mice 14 dpi. (A) Effect of PNGase F treatment of detergent extracts from spleen and dissected CNS regions. Representative extracts were incubated overnight with (+) or without (−) PNGase F, followed by Western immunoblot analysis with the Cas-Br-E envelope monoclonal antibody 697 (6). Overnight incubation of envelope extracts in deglycosylation buffer without PNGase F (upper panel) results in reduced envelope isoform resolution by SDS-PAGE and Western blot analysis (compared to Fig. 1A and B). However, despite the extended 37°C incubation, no envelope proteolysis was apparent. The lower panel shows that PNGase F treatment effectively eliminated the envelope isoform heterogeneity observed in the untreated samples. Protein loads were adjusted to give comparable signals for envelope protein. Abbreviations: SPL, spleen; TB, total brain; CB, cerebellum; SC, spinal cord; NCX, neocortex; BS, brainstem. (B) Comparison between treatment of spleen SPL, SC, and CB envelope extracts with endo H and PNGase F as detected by Western blotting. Endo H treatment of extracts from the brain (SC and CB) resulted in Env protein shifts to lower-molecular-weight proteins, whereas endo H treatment of spleen only shifted the gpr85 precursor protein. Because endo H treatment fails to shift any of the Envs to the fully deglycosylated 55-kDa form, the Env proteins all traffic through the Golgi apparatus, where some N-linked sugars are modified to complex type. As shown for SPL and CB, incomplete deglycosylation of envelope protein was sometimes observed in PNGase F-treated samples; however, this could be eliminated by a more prolonged primary digestion or a secondary digestion with PNGase F. V, Cas-Br-E virus supernatant from cultured dunni cells. (C) Endo H (top panel) and PNGase F (bottom panel) treatment of virion-enriched sucrose gradient fractions from either the infected CB or tissue-cultured dunni fibroblasts (TC). Despite the partial purification away from other cellular proteins which could be inhibiting the endo H, both gp65 and gp70 were only partially susceptible to endo H treatment. This result indicates that some high-mannose sugars are processed to the complex form in the Golgi apparatus, while others are not. Treatment of a mixture of cerebellar (gp65) and tissue culture (gp70) virus fractions with PNGase F yields a single band migrating at approximately 55 kDa, further confirming that the protein mobility differences reside in the N-linked sugar modifications.

Since the CNS envelope heterogeneity occurs in protein which has been proteolytically processed to SU and TM, the differential processing most likely arises during the conversion of high-mannose oligosaccharides to complex sugars in the Golgi compartment of various infected cell types. To evaluate the extent to which Cas-Br-E Env N-linked sugars are converted from high-mannose to complex types in the spleen and CNS, we compared extracts from the spleen, spinal cord, and cerebellum, treated either with endoglycosidase H (endo H; Boehringer Mannheim) (overnight, according to the manufacturer's instructions) or PNGase F, to remove either high-mannose N-linked sugars only or all N-linked sugars, respectively. The results, shown in Fig. 3B, indicate that endo H treatment of spleen and CNS extracts was able to deglycosylate the envelope precursor protein (gpr85), shifting it to approximately 70 kDa. However, differential deglycosylation was noted for the gp70 and gp65 isoforms of Cas-Br-E envelope in the spleen and the CNS. Specifically, the gp70 in spleen did not appear to be shifted significantly after endo H treatment, although the gpr85 band was shifted completely. In contrast, the gp70 and gp65 proteins in the spinal cord and the gp65 in the cerebellum showed a significant shift after endo H treatment. The resistance of spleen gp70 suggests that all the high-mannose sugars are modified to complex forms, while the partial susceptibility to endo H treatment of gp70 and gp65 in the CNS extracts indicates that only some high-mannose sugars are modified. Unfortunately, differences in endo H sensitivity between the gp70 and gp65 envelope protein isoforms from the spinal cord could not be discerned due to the lack of resolution after endo H treatment.

To evaluate whether Env protein incorporated into virions had distinct susceptibility to endoglycosidases F and H, compared to Env found in brain extracts, virus-enriched sucrose gradient fractions from the cerebellum or cultured dunni fibroblasts were subjected to deglycosylation. The results, shown in Fig. 3C, demonstrated that endo H treatment of the gp65 and gp70 viral Envs did not produce proteins with similar mobilities. The results specifically indicated that virion Env from the cerebellum is more susceptible to endo H than virion Env from cultured dunni fibroblasts. This suggests that some of the N-linked sugars on Cas-Br-E Env are processed to the complex type in the cerebellum while others are not. Since PNGase F treatment of gp65 and gp70 resulted in proteins with similar mobilities, this result corroborates the data generated with CNS extracts and further suggests that the Env peptide backbones are not differentially altered. Thus, the partial endo H resistance of CNS gp70 and gp65 envelope isoforms indicates that these proteins traffic through the Golgi apparatus, where some high-mannose sugars are converted to the complex type in a cell type-specific manner.

To exclude the possibility that the differences in glycosylation noted were not due to the in vivo selection of glycosylation mutants by certain infected cell types, sucrose gradient-purified virions from the cerebellum containing only gp65 (see Fig. 2A) were used to infect NIH 3T3 fibroblasts at a multiplicity of infection of 1 for 1 h. Twenty-four hours later, supernatants from these cells were examined by Western blotting for Env. The results showed that only gp70 and not gp65 was produced in these cells (not shown). Furthermore, inoculation of gp65-containing cerebellar virions into neonatal animals resulted in the same kinetic Env isoform expression pattern and neurodegenerative disease as inoculation of NIH 3T3 fibroblast-derived gp70-containing virions (not shown). These results support the idea that Env heterogeneity is unlikely to be due to mutations in the env gene but is due rather to differential cellular glycosylations. What specific modifications are made and at which glycosylation sites remain to be determined.

The results presented here provide a new framework for understanding how the Cas-Br-E viral envelope protein may be involved in the induction of spongiform neurodegeneration. In particular, our present analysis has revealed that the multiple Env isoforms observed in the infected CNS fall into three distinct classes. First, there is a gp70 class of protein, which shows a great deal of heterogeneity depending on the CNS region in which it is found. Unlike gp70 noted in cultured cell lines, it does not appear to be readily incorporated into virions. This protein is found specifically in CNS regions where microglial infection predominates and spongiform pathological changes are observed. These observations suggest a direct association between the gp70 isoforms and neurodegenerative disease. Second, there is a gp65 class of Cas-Br-E Env which appears specifically associated with regions of neuronal infection; in the regions where it was observed, it was the only Env isoform clearly associated with virus particles. Because gp65 proteins were found in regions with and without pathological changes, it seems unlikely that they play a causal role in disease induction beyond promoting virus spread to microglia. The third class of Env protein observed in the CNS was the gpr85 precursor class of protein. This Env protein class was observed in all infected CNS regions. Unlike the data generated for microglia in culture (23), no obvious Env precursor accumulation was noted for this protein in degenerating regions. This finding suggests that Env precursor processing is unlikely to be causal in the neurodegenerative process. Given that previous studies indicate that late virus replication events occurring in microglia are responsible for inducing neurodegeneration (27), our present results suggest that microglial production of unique gp70 isoforms via differential glycosylation could be the critical event in the induction of spongiform neuropathology. This could be due to the production of a uniquely neurotoxic Env or to some alteration of microglial physiology resulting from the inability to assemble the microglial gp70 isoform into virions. Alternatively, the unique Env glycosylation noted may be a response to other virus-induced changes and thus may reflect the neurodegenerative changes taking place in the brain. To resolve this issue will require specific mutagenesis of the Env protein glycoslyation sites.

Not surprisingly, the Cas-Br-E envelope protein has a unique set of glycosylation sites compared to other MuLVs (20, 32). Specifically, Cas-Br-E Env is missing the gs3 and gs5 MuLV consensus glycosylation sites and contains an additional nonconsensus glycosylation site at Asn 186 (Fig. 4). Functional dissection analysis of ecotropic retroviral envelope proteins indicates that the receptor binding domain is found in the N-terminal half of gp70 (17), with receptor specificity being determined by sequences in the variable domains VRA and VRB. Although the consensus glycosylation sites do not appear to directly interfere with purported receptor contact regions in VRA or VRB (11), functional studies on the Friend MuLV envelope protein indicate that when the gs1 and gs2 consensus glycosylation sites are removed, the envelope protein shows a weaker interaction with the receptor (indicated by a reduced ability to mediate superinfection interference) (3). This is consistent with reports showing that glycosylation inhibitors (e.g., tunicamycin) alter SU processing and intracellular transport (29, 35) and eliminate superinfection interference (34). Since removal of N-linked sugars on the ecotropic receptor does not diminish envelope binding (9, 21), the implication is that the loss of N-linked glycosylation on the envelope protein results in weakened Env-receptor interactions.

FIG. 4.

Schematic representation of the Cas-Br-E glycosylation sites in relation to the functional domains in MuLV Env. Small asterisks, consensus MuLV glycosylation sites; light asterisks, absence of a consensus glycosylation site in the Cas-Br-E Env sequence; large asterisk, additional nonconsensus site specific for Cas-Br-E Env (18). VRA and VRB, variable MuLV Env domains believed to be involved in receptor specificity and thus in binding (7, 17). Arrows indicate sites of proteolytic cleavage which occur, from left to right, in the endoplasmic reticulum, the Golgi apparatus, and the budding virion.

In addition to affects on receptor binding properties, Kayman et al. have also demonstrated that mutation of the gs4 or gs5 glycosylation site of Friend MuLV prevents the cleavage of the gpr85 precursor protein to gp70 and p15E (20). In the case of the gs4 mutation in Friend MuLV, envelope incorporation into virions is not observed. Similar gs4 mutations in MoMuLV had the same effect (12). However, for the Friend MuLV gs5 mutation, the failure to cleave gpr85 did not affect virus assembly or infectivity when tested in cell culture. In a somewhat related fashion, the absence of the gs5 site does not appear to affect proteolytic processing of Cas-Br-E envelope protein in fibroblast or astrocyte cultures (23). However, we have observed a lack of proteolytic processing of Cas-Br-E envelope protein in virus- and Env-infected primary microglial cultures (23). Whether this is due to the presence or absence of specific glycosylation sites has not been tested.

As demonstrated in this report, defective Cas-Br-E Env proteolytic processing does not appear to be occurring in the CNS regions where microglial infection predominates. Several possibilities exist which could explain the lack of consistency between in vitro and in vivo analyses. For example, proteolytic cleavage of envelope precursor protein in microglia in vivo could be facilitated by acquiring the necessary protease from another CNS cell type through endocytosis. Proteolytic cleavage of Env might in turn facilitate cell surface budding of virions. In in vitro experiments, seeding cleavage-defective FrCas^E-infected microglia as infectious centers demonstrated that they were readily capable of infecting cells with which they came in direct physical contact, despite the fact that they do not release virions when cultured alone (23). Furthermore, our present observations indicate that even though proteolytic processing of Cas-Br-E envelope protein does occur in regions of the CNS where microglial infection is high, we noted that the gp70 proteins associated with these regions were not readily incorporated into sedimenting virion particles. Support for the idea that the gp70 isoforms arise from microglial infection comes from our previous experiments where transplanted neural progenitor cells were used as platforms to infect CNS microglia at high levels. In these experiments only the gpr85 precursor protein and multiple gp70 isoforms were evident (27). These observations suggest that the gp65 class of virion protein observed in infected brains is not derived from microglial cells. More likely, the gp65 virion protein comes from infection of the neurons and/or vascular cells. Taken together, the in vivo results presented here indicate that Env cleavage alone may not be sufficient for effective Env incorporation into particles. This process may also require appropriate Env sugar processing in certain cell types.

The results presented here suggest several possibilities with regard to the mechanism of spongiform neuropathology induction. One possibility is that unique Cas-Br-E Env protein folding or glycosylation in microglia results in a gain of function. In this scenario, either the unique Env is neurotoxic or its expression within microglia induces the production of a microglial neurotoxin. However, since neither microglial activation nor inflammation is associated or required for disease induction (6, 24, 26), a novel mechanism for the production and release of such a toxin would need to be established. An alternative hypothesis is that unique Cas-Br-E Env folding or glycosylation results in a loss of microglial function. In this regard it should be noted, however, that Cas-Br-E infection of microglia in vitro is not associated with detectable cell death. Thus, in this scenario microglia must be providing some critical neuronal support function which is disrupted by Cas-Br-E Env expression. This disruption could be due to the presence of Env alone or to altered virus assembly and protein trafficking, as noted in microglia in vitro. Addressing these issues directly will require regulated expression of one or more retroviral genes within microglia in the intact CNS.