Effects of Mutant Ubiquitin on ts1 Retrovirus-Mediated Neuropathology

ts1 is a temperature-sensitive mutant of Moloney murine leukemiavirus that induces a rapid spongiform encephalopathy in miceinfected as newborns. The pathological features include theformation of ubiquitinated inclusions resembling Lewy bodies.To determine how perturbation of the ubiquitin-proteasome pathwaymight affect ts1-mediated neurodegeneration, the virus was introducedinto transgenic mice in which the assembly of ubiquitin chainswas compromised by the expression of dominant-negative mutantubiquitin. The onset of symptoms was greatly delayed in a transgenicmouse line expressing K48R mutant ubiquitin; no such delay wasobserved in mice expressing a wild-type ubiquitin transgeneor K63R mutant ubiquitin. The extended latency was found tocorrelate with a delayed increase in viral titers. Pathologicalfindings in K48R transgenic mice at 60 days were found to besimilar to those in the other strains at 30 days, suggestingthat while delayed, the neurodegenerative process in K48R micewas otherwise similar. These data demonstrate the sensitivityof retroviral replication to the partial disruption of ubiquitin-mediatedproteolysis in vivo, a finding that may have therapeutic potential.

INTRODUCTION

Top

Introduction

Moloney murine leukemia virus (Mo-MuLV) is a C-type retrovirusthat typically induces T-cell lymphomas in susceptible miceafter a latency of several months (reviewed in reference 12).The neuropathogenic strain ts1 was identified from a group oftemperature-sensitive mutants of the TB variant of Mo-MuLV (22,46). Mice infected as neonates with the ts1 virus develop aneuroimmunodegenerative disease which invariably progressesfrom slightly impaired locomotion through ataxia to paraparesisand ultimately hind limb paralysis (1, 2, 33, 34). The spongiformlesions induced by ts1 occur in the lumbar spinal cord, brainstem, thalamus, and cerebellum but rarely occur in the cerebrum(34). The lesions are accompanied by the formation of eosinophilic,Lewy body-like inclusions in the nuclear perikarya, and affectedregions of the central nervous system (CNS) show intense immunoreactivityfor ubiquitin (35). The spongiform neurodegeneration has beendocumented in the BALB/c and CFW/D laboratory mouse strainsbut is most rapid in the highly susceptible FVB/N strain (44).

The neurovirulence of ts1 is attributable to a single aminoacid substitution in gPr80^env, the envelope polyprotein precursorof Mo-MuLV. In ts1, the substitution of isoleucine for valineat position 25 of gPr80^env is thought to lead to misfoldingof the protein and inefficient transport from the endoplasmicreticulum (ER) to the Golgi apparatus (37), where the polyproteinwould normally be processed by proteolytic cleavage. As a C-typeretrovirus, ts1 does not infect nonmitotic cells, such as neurons,and so the neuronal loss must somehow be mediated through theinfection of astrocytic and/or microglial cells (52). The mechanismof the indirect neuropathogenesis of ts1 is currently unknown.

The ubiquitin-proteasome system is responsible for the regulatedelimination of normal cellular protein substrates as well asdamaged or misfolded proteins (reviewed in reference 6). Thereare at least three reasons to suspect that interference withubiquitin-mediated proteolysis might influence ts1 neurovirulence.The first is that ubiquitin is thought to play a role in retroviralbudding (26), probably through covalent modification of theGag protein (p12Gag in the case of Mo-MuLV [23, 24]). Decreasingthe levels of free ubiquitin has been shown to dramaticallyaffect the budding of Rous sarcoma virus, an avian C-type retrovirus(26). The second is that the V25I substitution in ts1 gPr80^envprobably generates a burden of misfolded protein within theER (37, 50) that would normally be eliminated by the ER-associateddegradation (ERAD) pathway, a system that depends on the ubiquitin-mediatedproteolysis of proteins removed from the ER through a pore complex(reviewed in reference 18). The third is that the lesions inducedby ts1 are intensely immunoreactive for ubiquitin (35), suggestingthat components of the ubiquitin-proteasome system are involvedin the response to viral infection. For these reasons, we soughtto determine how an impaired ubiquitin-proteasome system mightinfluence the onset and/or severity of ts1-mediated neurodegeneration.

Tsirigotis et al. previously reported the creation of a transgenicmouse model in which epitope-tagged ubiquitin of human originwas expressed in most cells and tissues and throughout development(39). Tsirigotis et al. also described a site-directed mutantform of ubiquitin (K48R) that, by interfering with chain assemblyin a dominant-negative fashion, could stabilize cellular substratesof the ubiquitin-proteasome pathway (40). The K48R mutant ubiquitinwas found to sensitize murine neuroblastoma cells to protein-damagingagents, presumably by interfering with the ability of cellsto eliminate damaged protein through ubiquitin-mediated proteolysis(40). A second mutant form of ubiquitin used in the currentstudy (K63R) is known to interfere with the assembly of an alternateubiquitin chain that functions in DNA repair (14), ribosomalmodification (32), and NF- ${kappa}$ B signaling (9). In the current work,we used novel transgenic mouse lines expressing these mutantubiquitin isoforms as hosts for ts1 virus infection and demonstrated,for the first time, that mutant ubiquitin profoundly affectsretroviral replication and associated pathology in vivo.

MATERIALS AND METHODS

Top

Materials and Methods

Generation of transgenic mouse lines. A transgenic mouse line expressing wild-type human ubiquitinas a linear fusion with enhanced green fluorescent protein (GFP)was described previously (39). New lines expressing K48R orK63R mutant ubiquitin were generated by using plasmids derivedfrom the original construct by site-directed mutation (theseplasmids were previously used in cell culture studies [40]).Microinjection, embryo culturing, and embryo transfer were doneas described for the wild-type construct (39). Founder miceexpressing ubiquitin-GFP fusion proteins were identified byfluorescence as follows. Mice were illuminated with a Volpi4000 light source fitted with a 470-nm excitation filter andwere observed through a hand-held 470-nm band-pass filter (ChromaTechnology) as previously described (39). Fluorescent mice werebred with FVB/N animals to obtain F₁ litters. Based on a surveyof tissue fluorescence, progeny of selected founders were chosenfor further breeding and establishment of homozygous lines.The current work was based on the novel K48R mutant line designated279 and the K63R mutant line designated 75.

Inoculation and monitoring of mice. Stocks of the ts1 retrovirus were prepared by passing culturesupernatants of TB producer cells (45) through 0.2-µm-pore-sizefilters. Stocks were stored at -80°C in 10% dimethyl sulfoxide.Newborn pups were infected by intraperitoneal injection of 0.1ml of a viral stock (10⁷ PFU/ml; measured by a focus assay ofa thawed aliquot of the filtered frozen stock) on the firstor second day postpartum. Mice were monitored daily for signsof weight loss, dehydration, and lethargy and were euthanatizedwhen there was evidence that loss of hind limb motor functionwas interfering with feeding and/or hydration.

Western analysis. Brain tissues from age-matched, nontransgenic mice or homozygousmice expressing wild-type, K48R mutant, or K63R mutant ubiquitinwere homogenized in phosphate-buffered saline (PBS) containing1% NP-40, 20% glycerol, and the following protease inhibitors:1 mM phenylmethylsulfonyl fluoride, 5 µg of leupeptin/ml,2 µg of aprotinin/ml, 200 µM NaF, and 200 µMsodium pyrophosphate. The homogenates were sonicated three timeson ice by using a Fisher Sonic Dismembrator (30 s per sonicationat setting 35, with 30 s between sonications) and spun at 20,800x g for 30 min. The supernatants were divided into two fractions.One fraction was precleared overnight by incubation in the presenceof GammaBind beads (Amersham Pharmacia Biotech) prewashed threetimes with PBS. The soluble fractions were recovered, and theproteins from both fractions (precleared and uncleared) werequantified by using the Bradford protein assay (Bio-Rad). Fortymicrograms of cytoplasmic protein (20 µg from each fraction)was resolved on a two-phase (15 and 8%) sodium dodecyl sulfate-polyacrylamidegel and electroblotted onto a Hybond C nitrocellulose membrane(Amersham Pharmacia Biotech). The membrane was stained withPonceau S prior to immunoblotting with the appropriate primaryantibody. Proteins were visualized by a horseradish peroxidasemethod with an enhanced chemiluminescence kit (Kirkegaard &Perry Laboratories). The specificities and sources of the antibodiesused in the Western analysis were as follows: mouse monoclonalanti-RGS-His epitope (Qiagen), rabbit polyclonal anti-GFP (SantaCruz Biotechnology Inc.), and rabbit antiubiquitin (Dako Corp.).

Immunohistochemical analysis. Brains and spinal cords from ts1-infected or uninfected controlmice were fixed in phosphate-buffered 10% formalin for 1 to2 days. Tissues were embedded in paraffin and cut at a thicknessof 5 µm. For immunostaining, deparaffinized sections wereheated in a solution of 10 mM sodium citrate (pH 6.0) in a 700-Wmicrowave oven for 10 min. Endogenous peroxidase activity wasblocked by incubation in methanol containing 3% hydrogen peroxidefor 20 min. Sections were washed with 0.1 M PBS (pH 7.4) andincubated for 30 min with 1.5% normal goat serum to block nonspecificbinding. They were then incubated overnight at 4°C withrabbit anti-PGP 9.5 antibody (Chemicon) diluted 1:6,000 in 50mM Tris-HCl- 1.5% NaCl- 1% normal goat serum. After being washedwith PBS, the sections were exposed to biotinylated goat anti-rabbitimmunoglobulin G diluted 1:200 in PBS- 1.5% normal goat serum.The reaction product was visualized by the avidin-biotin-peroxidasemethod with an ABC Elite kit (Vector Laboratories). Cell nucleiwere counterstained with hematoxylin.

Measurement of viral titers. Viral titers in infected mouse tissues were determined by usingpreviously published methods (47). Briefly, tissues were weighedand then homogenized for 30 s in 2 ml of alpha minimal essentialmedium by using a Brinkmann/Kinematica Polytron PT 10/35 deviceoperating at approximately 12,000 rpm. Homogenates were passedthrough 0.45- and 0.2-µm-pore-size filters, dimethyl sulfoxidewas added to 10%, and homogenates were divided into aliquotsand stored at -80°C. Samples were diluted in alpha minimalessential medium containing 3% heat-inactivated newborn calfserum and 3 µg of Polybrene/ml and incubated with monolayersof 15F cells (51) as previously described. Foci were countedat 5 days postinfection (p.i.).

RESULTS

Top

Results

Characterization of mutant transgenic lines. Transgenic mice expressing the K48R or K63R mutant form of ubiquitinwere found to be outwardly normal in physical appearance andhave shown no indication of reduced viability to date. Althoughthe numbers of animals reaching geriatric age are still relativelysmall, both K48R mice and K63R mice have lived beyond 2 yearsof age in our colony and have yet to display signs of pathologybeyond the normal spectrum of illnesses characteristic of agingmice of the FVB/N background strain, such as a high frequencyof lung cancer (21). The fertility of the K48R strain is somewhatreduced, as evidenced by smaller and more infrequent litters,and an investigation into the physiological basis of this observationis ongoing.

Cells process naturally occurring fusions of ubiquitin and ribosomalsubunit peptides very efficiently (10) through the action ofubiquitin carboxyterminal hydrolase enzymes (19). The premiseof our strategy is that cells will process a ubiquitin-GFP fusionin the same manner; thereafter, the free ubiquitin will be availablefor conjugation to protein, whereas the GFP moiety will serveas an easily detectable marker for transgene expression. Tsirigotiset al. previously reported that for both wild-type ubiquitinand mutant ubiquitin, this strategy works well in transfectedcells (40); wild-type ubiquitin also functions as expected intransgenic mice (39). In establishing the utility of transgeniclines expressing mutant ubiquitin for the purpose of investigatingthe role of ubiquitin in virus-mediated pathogenesis, thereare two issues of paramount importance. The first is the properprocessing and hence functional availability of ubiquitin fromthe ubiquitin-GFP fusion. The second is the expression of transgene-derivedubiquitin in appropriate target tissues. To confirm that themutant ubiquitin-GFP fusions were properly processed in vivo,lysates were prepared from brain tissues and analyzed by Westernblotting. With an antibody specific for the hexahistidine epitopeat the amino terminus of the transgene-derived ubiquitin, thetagged ubiquitin could be detected in both monomeric form andthe form of higher-molecular-weight conjugates in all threelines of ubiquitin transgenic mice but not in nontransgeniccontrol mice (Fig. 1B). Consistent with previous observationsin transfected cell lines (40), the highest intensity of epitopetag immunoreactivity was observed with the K48R mutant. Theincreased immunoreactivity cannot be attributed simply to higherlevels of transgene expression in the K48R mice, because suchdifferences were not reflected in GFP levels, either by directobservation of fluorescence or by immunoblotting (Fig. 1D).The addition of transgene-derived ubiquitin did not seem togreatly elevate the total levels of ubiquitin in brain lysates(Fig. 1C), suggesting that endogenous ubiquitin levels probablyexceed the levels of transgene-derived ubiquitin.

View larger version (62K):
[in this window]
[in a new window]
FIG. 1. Expression and processing of transgene-derived ubiquitin. (A) General design of ubiquitin expression cassettes. The human ubiquitin C promoter (UbC) was used to drive the expression of hexahistidine-tagged human ubiquitin (Ub) fused in frame to enhanced GFP (EGFP). The simian virus 40 polyadenylation signal (pA) was placed downstream. (B) Processing and conjugation of transgene-derived ubiquitin in the brains of transgenic mice. Transgene-derived ubiquitin was detected by using an antibody specific for the hexahistidine eptitope tag. The presence of the low-molecular-weight band (arrow) confirmed that the Ub-enhanced GFP fusion had been processed to provide free monomers, which were not detected in brain lysates from nontransgenic control mice (nontg). The high-molecular-weight smear (particularly evident in brain lysates from K48R transgenic mice) corresponded to epitope-tagged ubiquitin conjugated to substrates. wt, wild type. (C) Analysis of total ubiquitin in brain lysates. The addition of transgene-derived ubiquitin was not found to greatly increase the overall levels of the ubiquitin monomer (arrow) or to grossly affect the levels of ubiquitin conjugates (smear). (D) Comparison of transgene expression levels. With a GFP-specific antibody, only processed enhanced GFP could be detected (open arrowhead), confirming that the ubiquitin moiety was rapidly released from the fusion. The levels of transgene expression appeared comparable in the K48R and K63R transgenic lines, whereas expression in the wild-type ubiquitin transgenic line may have been slightly lower. The simultaneous use of a ß-actin-specific antibody confirmed equal loading of lysates (closed arrowhead).

The ts1 retrovirus generates spongiform changes predominantlyin the brain stems and spinal cords of FVB/N mice. As a C-typeretrovirus, ts1 does not infect terminally differentiated neuronsbut generates pathological effects in the brain through infectionof glial supporting cells (34). For our model to be meaningful,it was necessary to demonstrate the expression of the transgenein appropriate target cell populations. To establish which lineagesin the CNS expressed the transgene, brains were removed, cultured,and examined by fluorescence microscopy. Primary cerebellarcultures from the wild-type ubiquitin transgenic line revealedthat the transgene was expressed in both neuronal and gliallineages (Fig. 2A to F). To confirm that the transgene was expressedin relevant anatomical locations, in vivo frozen sections wereprepared and observed for fluorescence. As was previously observedin the wild-type ubiquitin transgenic line (39), there was intensefluorescence throughout the CNS, including the spinal cord,in the K48R and K63R transgenic lines; an example is shown inFig. 2G. The pattern of fluorescence that we have observed atthe gross and microscopic levels has not been found to varyamong the ubiquitin transgenic lines, and although primary cultureshave not been generated from the mutant lines, they are notexpected to differ with respect to transgene expression.

View larger version (150K):
[in this window]
[in a new window]
FIG. 2. Expression of ubiquitin transgenes in primary cultures and in vivo. (A) Low-power phase-contrast image of a cerebellar culture from a 7-day-old wild-type ubiquitin-enhanced GFP transgenic mouse on the third day of culturing. (B) Same field as in A under UV illumination showing uniform fluorescence of most to all cells in the culture. (C to F) High-power magnification of primary cultures demonstrating fluorescence of cells displaying neuronal and astrocytic morphologies. Images C and F were intentionally overexposed to reveal fluorescence of the neuronal processes. Images D and E were from the same field at different focal planes, revealing cells with neuronal morphology on top of cells with astrocytic morphology. (G) Frozen section of spinal cord from a K63R transgenic mouse. Uniform fluorescence confirmed the expression of the transgene throughout the cord. P, pial surface; VH, ventral horn; VLS, ventral longitudinal sulcus.

Time course of ts1-mediated neurodegeneration in infected mice. Mice were infected with the ts1 retrovirus through the intraperitonealroute on the first or second day postpartum and were monitoredclosely for changes in gait, grasping response following elevationby the tail, deterioration in grooming, weight loss, and othersigns of debility. In the FVB/N strain, there is typically deteriorationin the health of the animals within 3 weeks p.i., with the incidenceof hind limb paralysis reaching 100% by 34 days (44). The findingsfor FVB/N mice in the current study were in good agreement withthe previous findings (Fig. 3) in that most animals requiredeuthanasia by day 34, although one animal survived to day 60(at which time the experiment was terminated). Mice that werehomozygous for the wild-type ubiquitin-GFP transgene were foundto be very similar to the FVB/N control mice with respect tothe onset and progression of disease, as were mice that werehomozygous for the K63R mutant ubiquitin transgene. Remarkably,the onset and progression of disease symptoms were manifestedvery differently in mice homozygous for the K48R mutant ubiquitintransgene. These animals demonstrated a significant delay inthe onset and progression of neurodegeneration, with half ofthe mice surviving at 60 days p.i. Several of the survivingK48R mice were largely or entirely asymptomatic at 60 days p.i.,a phenomenon not observed in the other transgenic strains orin the nontransgenic control mice.

View larger version (18K):
[in this window]
[in a new window]
FIG. 3. Survival of ts1-infected mice. Kaplan-Meier plot demonstrating the enhanced survival of infected mice expressing K48R mutant ubiquitin compared to other transgenic strains of mice or nontransgenic FVB/N control mice. The experiment was terminated at 60 days, at which time approximately half of the K48R mice were still viable and largely asymptomatic. wt, wild type; Ub, ubiquitin.

Pathological findings. Sections from the brains and spinal cords of the three transgenicstrains as well as the nontransgenic control mice were stainedwith hematoxylin-eosin or were stained immunohistochemicallywith an antibody to the neuronal marker PGP 9.5 (38). Examinationof ts1-infected mice that were not transgenic (Fig. 4B) or thatexpressed transgene-derived wild-type (Fig. 4C) or K63R mutant(Fig. 4F) ubiquitin at or about 30 days p.i. revealed pathologicalfindings similar to those previously reported for mice of theFVB/N strain (35), with spongiform degeneration occurring mostprominently in the brain stem and in the ventral horns of thelumbar spinal cord. Remaining neurons appeared pyknotic, andthere was considerable gliosis not accompanied by inflammatorycell infiltration. Less severe spongiform encephalopathy wasobserved in the cerebella of the infected mice. At 30 days p.i.,the brains and spinal cords of mice expressing the K48R mutantform of ubiquitin were found to be indistinguishable from thoseof uninfected FVB/N control mice (Fig. 4D and A, respectively).Spinal cords from infected K48R mutant mice at 60 days p.i.showed obvious disease progression approximating that of theother strains at 30 days p.i., with similar spongiosis of theneuropil and gliosis (Fig. 4E). These findings were in closetemporal agreement with the onset of paralysis in infected K48Rmice (Fig. 3).

View larger version (160K):
[in this window]
[in a new window]
FIG. 4. Neuropathology of spinal cords. Neurons were visualized by immunohistochemical analysis with an antibody to PGP 9.5 (UCHL1). Counterstaining for nuclei was done with hematoxylin. (A) Spinal cord from an uninfected FVB/N mouse, showing normal morphology and cellularity. (B) Spinal cord from an infected FVB/N mouse at 30 days p.i., showing spongiform lesions. (C) Spinal cord from an infected wild-type ubiquitin-enhanced GFP transgenic mouse at 31 days p.i. (D) Spinal cord from an infected K48R mutant mouse at 30 days p.i. No pathological changes were evident. (E) Spinal cord from an infected K48R mutant mouse at 60 days p.i. Spongiform changes were similar to those observed in nontransgenic mice at 30 days p.i. (F) Spinal cord from an infected K63R mutant mouse at day 29 p.i. Spongiform changes were evident. Bars, 50 µm.

Viral titers. To determine how viral replication was affected by the expressionof wild-type or mutant ubiquitin isoforms, titers were establishedfor two tissues: the spleen, a tissue rich in lymphocytes andas such an important site of viral amplification, and the brainstem, one of the tissues most affected by the ts1 retrovirus(34, 35). Titers in the spleens of nontransgenic FVB/N micerose quickly, reaching 10⁷ infectious units per g of tissueby 10 days p.i. and increasing by an additional 0.5 log unit(Fig. 5A). Titers in the wild-type ubiquitin-GFP transgenicmice and K63R mutant mice were identical to those in the nontransgeniccontrol mice at early time points and were only marginally lowerby 30 days p.i. Titers in the K48R mutant mice were approximately0.5 log unit lower than those in the nontransgenic control micefrom the earliest time point through 30 days p.i., reachinglevels similar to those observed in the other transgenic miceonly after 60 days p.i.

View larger version (16K):
[in this window]
[in a new window]
FIG. 5. Measurement of viral titers in two tissues. (A) Titers (y axis) in the spleens of K48R mice were found to be lower than those in the other lines and rose to levels equivalent to those in nontransgenic FVB/N mice only after 60 days. Standard errors were calculated based on titers determined for three infected animals of each strain at each time point. wt, wild type. (B) As expected, titers (y axis) in the brain were initially much lower than those in the spleen. As in the spleen, viral titers in K48R mice did not reach levels comparable to those in the other lines until 60 days p.i., when the experiment was terminated.

In the brain stems, the difference in titers between K48R mutantmice and control mice was more pronounced. A 1 log unit differencewas observed between the viral titers in the lysates of K48Rbrain tissue and those in the nontransgenic control sample (Fig.5B). The titers measured in the wild-type ubiquitin-GFP transgenicmice and the K63R mutant mice were intermediate between theK48R mutant mouse and nontransgenic mouse values. The titersin the brain stems of K48R mice were more than 1 log unit lowerthan those in nontransgenic control mice at 30 days p.i. andwere comparable by day 60 p.i. to the titers in the other transgenicstrains but were still below those measured in nontransgeniccontrol mice. Statistical analysis by Student's t test of viraltiters at day 30 p.i. confirmed a significant difference betweenK48R mutant mice and nontransgenic FVB/N control mice (P = 0.0068).A significant difference was also established between K48R miceand transgenic mice expressing wild-type human ubiquitin (P= 0.0161). The difference between the K48R and K63R mutant transgeniclines was not significant.

The measured viral titers are consistent with the pathologicalfindings shown in Fig. 4, wherein spongiform lesions in K48Rmice at day 60 p.i. were similar to those in the other miceat day 30 p.i., and suggest that a threshold titer must be attainedbefore fulminant disease is manifested.

DISCUSSION

Top

Discussion

We have demonstrated a remarkable delay in the onset of ts1-mediatedspongiform neurodegeneration in mice expressing K48R mutantubiquitin, doubling the onset of paralysis from approximately30 days to 60 days p.i. (Fig. 3). This delay was not observedin mice expressing a K63R mutant ubiquitin transgene, nor wasit observed in mice expressing a wild-type ubiquitin transgene.Our analysis of histological sections (Fig. 4) suggests thatthe spongiform disease in K48R mice, while delayed, is otherwisequalitatively similar to the established pattern of neurodegenerationseen in ts1-infected mice of the FVB/N strain (44). The ubiquitin-proteasomepathway is present in many cellular systems, and the in vivoexpression of mutant ubiquitin might alter the host responseto viral infection in a manner that would delay the onset ofsymptoms (for example, by modulating cytokine release or otheraspects of innate immunity). We have not yet determined theextent to which innate immunity is affected in our transgeniclines and cannot exclude the possibility of other pleiotropiceffects of mutant ubiquitin, but there is clearly an effecton viral replication. One parsimonious explanation for the delayin disease onset in mice expressing K48R (but not K63R) mutantubiquitin is the impaired replication of the ts1 virus in vivo.Concomitant with the observed delay in disease onset in K48Rmice was a lag in viral replication evidenced by direct measurementof viral titers in infected tissues (Fig. 5). Titers in K48Rmice appeared to take twice as long to reach the threshold levelof virus required for spongiform alterations to occur. Thisinterpretation is in agreement with the longer disease latencyobserved when smaller amounts of ts1 were injected (22) or withthe lower titers and delayed disease observed when ts1 was injectedafter the newborn stage (43).

There are two subcellular locations at which retroviral replicationand the ubiquitin system must intersect: the plasma membraneand the ER. The budding of retroviruses from the plasma membranehas been speculated to occur through a process that drives endocytosisin reverse, usurping cellular constituents for this purpose(reviewed in references 3 and 42). There is growing evidencethat this process involves the ubiquitin E3 ligase Nedd4 (15,36, 48) and variant E2-like protein TSG101 (8, 13, 41) actingthrough interactions (28) with late domain motifs (11) in viralGag proteins to ligate ubiquitin. Although the role of the ubiquitinationof retroviral Gag proteins at the plasma membrane has been moreextensively investigated in human immunodeficiency virus (27)and Rous sarcoma virus (26) than in Mo-MuLV, the p12Gag proteinof Mo-MuLV is monoubiquitinated (23, 24), and it seems likelythat modification of Mo-MuLV p12 allows it to take advantageof the endosomal machinery in a fashion similar to that of theother viruses. The addition of proteasome inhibitors has beenshown to interfere with the budding of Rous sarcoma virus (26)and human immunodeficiency virus (29), an effect that can beat least partially reversed by providing Gag as a linear fusionwith ubiquitin (26). A possible explanation for the sensitivityof viral assembly to proteasome inhibitors is that the inductionby these agents of the accumulation of polyubiquitinated substratesdepletes free pools of ubiquitin, pools that are therefore unavailablefor membrane-associated events. This model is supported by datashowing that the proteasome inhibition or blockage of assemblywas suppressed by the addition of an exogenous source of ubiquitin(26). We do not believe that the impediment imposed on ts1 viralreplication in mice expressing K48R mutant ubiquitin was theconsequence of depleted ubiquitin pools; our data suggest thatalthough there was an accumulation of ubiquitinated substratesin K48R mouse tissues (Fig. 1B), this accumulation was not accompaniedby a depletion of free ubiquitin reserves (Fig. 1C). We currentlyhave neither evidence nor reason to suspect that the assemblyof ts1 Gag constituents at the membrane is altered in our transgenicmice, but the possibility remains open.

A second major intersection of retroviral replication and theubiquitin-proteasome system exists at the membrane of the ER.Proteins unable to fold correctly in the ER are extracted fromthe ER for degradation in proteasomes associated with the cytoplasmicaspect of the ER membrane (18), a process that has been designatedERAD. Membrane-associated proteins otherwise destined for theplasma membrane may be degraded by this route (17, 49), whichdepends on the assembly of K48-linked chains on these substratesby ER membrane-associated ubiquitinating enzymes (20). Retroviralenvelope proteins are translated and processed and must correctlyoligomerize within the oxidizing environment of the ER (reviewedin reference 7). In cells expressing K48R mutant ubiquitin,there is a general deficit in the proteasomal degradation ofmisfolded or damaged protein substrates in the cytoplasm (40).Because the ERAD pathway makes use of ubiquitin chains assembledthrough K48 linkages, it is likely that K48R mutant ubiquitinwould also impair ERAD directly, but even were this not thecase, an indirect effect would be expected as the consequenceof proteasomal occupation with an increased burden of cytosolicas opposed to ER resident substrates. We postulate that in thepresence of K48R mutant ubiquitin, there may be a global impedimentto the production of fully processed transmembrane proteinsin the ER and that the production of mature, correctly assembledretroviral surface and transmembrane proteins may be particularlysensitive to this effect. The problem may be compounded by thepropensity of the ts1 gPr80^env protein to misfold (37, 50) andthereby add to the ERAD burden. Our data are entirely consistentwith such a model in that we observed a considerable delay inthe increase in viral titers and the onset of ts1-mediated neuropathogenicsymptoms in K48R mice but no qualitative difference in the diseaseonce it had been initiated. To test the hypothesis, we are developingreagents that will allow us to evaluate envelope protein productionin cells expressing various forms of ubiquitin while monitoringERAD efficiency.

The infection of glial cells by the ts1 retrovirus (34) inducesthe expression of a group of inflammatory cytokines emblematicof NF- ${kappa}$ B activation (5). Intriguingly, in a previous study, thecytopathic effects of ts1 on cultured primary astrocytes werefound to correlate closely with the accumulation of viral envelopeproteins in the ER (31), and ER stress is known to activatethe NF- ${kappa}$ B pathway (reviewed in reference 25). It has been convincinglydemonstrated that I ${kappa}$ B degradation (a key step in NF- ${kappa}$ B activation)is triggered by ts1 infection (16); therefore, in addition tothe effects of mutant ubiquitin expression on retroviral replication,we must consider its possible effects on activation of the NF- ${kappa}$ Bpathway. NF- ${kappa}$ B activation is dependent on two types of ubiquitinchains. K63-linked chains are required for initiating the kinasecascade that leads to the phosphorylation of inhibitors of NF- ${kappa}$ B(9). K48R-linked chains are required for the subsequent degradationof I ${kappa}$ B proteins via the proteasome (4), as well as for the limitedproteasome-mediated processing of NF- ${kappa}$ B proproteins to theiractive forms (30). If I ${kappa}$ B phosphorylation was impaired in theK63R mutant mice, the effect was insufficient to affect diseaseonset or progression (Fig. 2). The K48R mice did demonstratea considerable delay in disease onset, but there did not seemto be an obvious difference in the extent of spongiform lesionsin spinal cords at later time points (Fig. 4). Our data do notexclude a K48R mutant ubiquitin-mediated perturbation of I ${kappa}$ Bdegradation, but at this point there is no reason to implicateone in the delayed progression of disease in ts1-infected K48Rmutant mice. Further experiments will be required to determinewhether our observations are solely attributable to impairedts1 replication or whether more subtle effects of the ubiquitinpathway are playing some role.

ACKNOWLEDGMENTS

We are indebted to William Staines, University of Ottawa, forestablishing primary cerebellar cultures from the transgenicmice and to John Woulfe, Ottawa Hospital, for many useful comments.

This work was supported by NIH grant ES 05022 to K.R.R. andby grant MT-15134 from the Canadian Institute of Health Researchand grant 77369 from the National Cancer Institute of Canadato D.A.G.

FOOTNOTES

* Corresponding author. Mailing address: Centre for Cancer Therapeutics, Ottawa Regional Cancer Centre, 503 Smyth Rd., Ottawa, Ontario, Canada K1H 1C4. Phone: (613) 737-7700, ext. 6896. Fax: (613) 247-3524. E-mail: Doug.Gray{at}orcc.on.ca.

REFERENCES

Top