In human skin fibroblasts microinjected with purified human immunodeficiency virus type 1 protease (HIV-1 PR), stress fibers were lost and alterations in nuclear morphology and condensation of nuclear chromatin were observed. Thereafter, the vimentin intermediate filament (IF) network collapsed. No effect was seen on the microtubules. While complicated by loss of affected cells from the substratum, a minimum estimate of the proportion of cells demonstrating these effects is 50%. Observation of single cells demonstrated that these effects were largely irreversible and were steps leading to the death of the HIV-1 PR-injected cells. After microinjection of various dilutions of the HIV-1 PR, it was observed that the changes in nuclear morphology and chromatin condensation were detectable under conditions where little or no effect was observed on both stress fibers and the IF network. Proteins of cells labelled with [35S]methionine and microinjected with either HIV-1 PR or BSA were subjected to two-dimensional gel electrophoresis. The major differences in the gel patterns were a diminution in the amount of vimentin and the appearance of novel products comigrating with cleavage products obtained after treatment of vimentin with HIV-1 PR in vitro. Thus, the HIV-1 PR is capable not only of cleaving IF subunit proteins in vivo, but also can catalyze alterations in other cellular structures.

Retroviruses encode a variety of structural proteins and enzymes necessary for their unique life cycles in the form of precursors or polyproteins. As reviewed by Skalka (1989) and Kay and Dunn (1990), one of these proteins is a protease, belonging to the aspartyl protease family, possessing the specificity necessary to cleave correctly the viral polyproteins into the individual, functional proteins. In the life cycle of the human immunodeficiency virus type 1 (HIV-1), this maturation process is intimately linked with the budding process that releases virus particles from the surface of the infected cell. The particles released contain polyproteins and are noninfectious; during and after release, mature proteins are produced by proteolysis and the final capsid and internal structures are formed, giving rise to infectious virus particles (Gelderblom et al. 1988). A variety of studies have demonstrated that the viral PR is required for both polyprotein processing and infectivity of the virus particle (Kohl et al. 1988, and reviewed by Kràusslich and Wimmer, 1988). No host protease has been identified that can substitute for the viral PR and, until recently, dogma has stipulated that the viral PR is specific for the viral polyproteins. In another retrovirus, equine infectious anemia virus, it has been demonstrated that this PR may be involved in the initiation of infection, presumably by opening the capsid to allow the release of nucleic acid from its interior (Roberts and Oroszlan, 1989). A similar role has been postulated for the HIV-1 PR and evidence has been presented at recent meetings demonstrating that this enzyme can also cleave the nucleocapsid protein of the mature viral capsid (Oroszlan, 1990). Grewe et al. (1990) have shown that the capsid of HIV-1 disintegrates within minutes following viral-cell membrane fusion, releasing the viral ribonucleoprotein into the cytoplasm. Thus, the viral PR may play an important role not only in the budding (maturation) process, but also in the initiation of infection.

Our initial interest in this field was prompted by diverse observations that suggested that intermediate filament (IF) subunit proteins might be involved in the infection process and/or modified by HIV-1 during the life cycle. We have previously shown that purified HIV-1 PR can cleave the IF subunit proteins vimentin, desmin and glial fibrillary acid protein in vitro and that microinjection of this PR resulted in a collapse and rearrangement of the vimentin IF network in many of the injected cells (Shoeman et al. 1990a,b). While our previous in vitro studies have yielded considerable biochemical details of the effects of HIV-1 PR on IF subunit proteins, they have not been able to resolve several key questions: does HIV-1 PR cleave vimentin in vivo? Is vimentin the only protein substrate or, as might be expected, are other structures modified and thus play a role in the alterations observed after microinjection of this enzyme? We therefore chose to microinject purified HIV-1 PR into human skin fibroblasts, to assay for alterations in cell morphology using fluorescence microscopy and to detect cleavage of host cell proteins by comparing protein patterns after two-dimensional (2-D) gel electrophoresis. We report here that vimentin is indeed cleaved by HIV-1 PR in vivo and represents the major host cell protein cleaved. Furthermore, we observed rapid alterations in stress fiber distribution, nuclear morphology and nuclear chromatin condensation, prior to the effects observed later on the vimentin IF network.

BSA, Hoechst 33258, fluorochrome-conjugated phalloidin and fluorochrome-conjugated anti-goat Ig antibodies were purchased from Sigma (Deisenhofen, FRG). HIV-1 PR (>90% pure, concentration=20/IM) was kindly supplied by S. Roy (Hoffmann-La Roche Inc., Nutley, NJ 07110, USA). BSA was added to the PR solution to a final concentration of 1 mg ml−1 (to reduce the extent of self-digestion by the PR) and this preparation was dialysed against CMF-PBS (Ca2+-free, Mg2+-free phosphate buffered saline; 140 mM NaCl, 2.6 mM KC1, 8mM Na2HPO4 and 1.4 mM KH2PO4, pH7.2). BSA, at lmgml−1 in CMF-PBS, was employed as a control solution for microinjection. The activity of the HIV-1 PR preparations was monitored using vimentin as a substrate in an in vitro assay as previously described (Shoeman et al. 1990a). Human skin fibroblasts (Hika cells) were cultured as previously described (Willingale-Theune et al. 1989). Microinjection, formalin fixation and analysis by immunofluorescence microscopy were performed as previously described (Hôner et al. 1988; Shoeman et al. 1990a). A conditioned medium for metabolic labelling of these cells was prepared by incubating monolayers for about 12 h in methionine-free minimum essential medium (MEM) (Flow Laboratories, UK) supplemented with 1/zgml−1 methionine. For gel analysis, 200 cells were cultured for 2 h in 30 /μI of this conditioned medium additionally supplemented with 9.25 MBqml−1 of [36S]methionine (in vivo cell labelling grade, specific activity >37TBqmmol−1, Amersham, Braunschweig, FRG). This medium was replaced with normal, conditioned MEM medium lacking [35S]methionine 30 min prior to microinjection. Approximately 200 cells, injected with either HIV-1 PR or BSA, were resuspended in sample buffer and 2-D gel electrophoresis was performed in a BioRad Mini Protean 2-D system (BioRad, Munich, FRG) according to the manufacturer’s instructions. In some cases, the samples were supplemented with purified vimentin digested with HIV-1 PR m vitro to localize the position of the vimentin cleavage products in the 2-D gel patterns. The gels were stained with silver, then impregnated with Amplify (Amersham), dried and subjected to fluorography with Kodak X-Omat film at -70°C for 4-5 weeks.

Human skin fibroblasts possess a well-delineated vimentin IF network and pattern of stress fibers and thus represent a good model system to study the effect of agents on their structure. Microinjected human skin fibroblasts demonstrated time- and HIV-1 PR-dependent changes in morphology (Fig. 1). At various times after injection of either HIV-1 PR, or BSA as a control, the cells were fixed, permeabilized and treated with 3 reagents labelled with different fluorochromes: phalloidin to decorate filamentous actin (f-actin), an anti-vimentin antibody and Hoechst 33258 dye to label DNA (chromatin). Cells microinjected with BSA (Fig. 1) were indistinguishable from uninjected cells, retaining a well-spread vimentin IF network, pronounced stress fibers (bundles of f-actin) and a normal nuclear morphology at all time points. As can be seen in Fig. 1D, in selected cells displaying typical effects, stress fibers were no longer detectable at 15 min after injection of HIV-1 PR and the chromatin in these affected cells appeared to have condensed somewhat. The distribution of the vimentin IF network was largely normal, except that in some cells there were indications of retraction of the network from the extremities of the cell periphery. Two hours after microinjection, these changes became much more dramatic (data not shown). No stress fibers were detected. Fewer foci of condensed chromatin were observed, but they were much larger than observed at 15 min and encompassed most of the detectable DNA. The vimentin IF network collapsed into a large clump with a perinuclear localization. At this time point, many of the cells were less well spread and had begun to round up. After 4h, all of these effects were observed to be more extreme, especially the rounding up of the cells (Fig. 1G,H). As a consequence of their reduced contacts with the substratum, many of the microinjected and affected cells were lost to analysis (see below). Particularly impressive was the extent of DNA condensation and reduction in size and collapse of the nucleus in these cells. At 20 h after microinjection (Fig. 1I,J), some cells were observed that gave the impression of having partially recovered from these changes: while they were spread normally, their vimentin IF network was collapsed adjacent to the nucleus rather than well spread in the cytoplasm and they had few, but detectable, stress fibers. No specific effect on the microtubule system was observed in either HIV-1 PR- or BSA-injected cells (data not shown). The microtubule network remained well spread and filled the cytoplasm even in those cells that extensively rounded up as a consequence of microinjection of HIV-1 PR (data not shown).

Fig. 1.

Indirect immunofluorescence microscopy of human skin flbroblasts processed at various times after microinjection of either HIV-1 PR (C,D,G-J) or BSA as a control (A,B>E,F). Each row of photographs contains images of the same cells treated with an antibody to vimentin (A,C,E,G,I), phalloidin to visualize f-actin (B,D,F,H,J) or Hoechst 33258 to stain DNA (inserts in B,D,F,H,J) at the indicated times post-injection (A-D: 15min; E-H: 4h; I,J: 20 h). Bars, 20μm.

Fig. 1.

Indirect immunofluorescence microscopy of human skin flbroblasts processed at various times after microinjection of either HIV-1 PR (C,D,G-J) or BSA as a control (A,B>E,F). Each row of photographs contains images of the same cells treated with an antibody to vimentin (A,C,E,G,I), phalloidin to visualize f-actin (B,D,F,H,J) or Hoechst 33258 to stain DNA (inserts in B,D,F,H,J) at the indicated times post-injection (A-D: 15min; E-H: 4h; I,J: 20 h). Bars, 20μm.

To rule out a selection bias in the analysis of these cells, 4 independent experiments employing large numbers of cells (about 100-200 per time point in each experiment) were evaluated from 0.5 to 20 h after microinjection (Fig. 2). The number of cells found in the marked microinjection fields did not differ between HIV-1 PR- or BSA-microinjected cells for the first hour following microinjection (Fig. 2A); thereafter, the recovery of HIV-1 PR-microinjected cells declined steadily. Between 8 and 20 h, the percentage of cells found within the fields containing cells originally injected with HIV-1 PR increased, either as a result of cell division in minimally affected cells or because of infiltration of the fields by neighboring, uninjected cells. Stress fibers were affected in about 60% of these cells, a maximum value reached about 1 h post-injection (Fig. 2D). The distribution of the vimentin IF network was affected in a maximum of about 50% of HIV-1 PR-microinjected cells. The kinetics of this effect were clearly different from those of the effect on stress fibers, reaching a maximum after 2h. In contrast to the well spread, flattened morphology normally exhibited by these fibroblasts, many of the HIV-1 PR-injected cells rounded up and were easily detached from the substratum. The maximum percentage of cells (about 40%) with a round morphology was observed at 5h after microinjection. Although not statistically evaluated, most of the cells demonstrating one effect at 1-5 h post-injection also demonstrated the other 2 effects. All three changes declined in frequency after about 8h (Fig. 2). In contrast, 90% or more of those cells injected with BSA showed no changes in these three parameters (Fig. 2).

Fig. 2.

(A-D) Statistical analysis of 4 independent microinjection experiments. Each point represents results obtained from 200-500 cells. The continuous lines connect points from cells injected with HIV-1 PR and the broken lines connect those injected with BSA. The bars show one standard deviation above and below the mean values. As indicated, panel A shows the cell recovery rate (number of cells observed divided by number of cells originally injected, x 100%), panel B shows the percentage of cells with a collapsed vimentin IF network, panel C shows the percentage of cells with a rounded morphology and panel D shows the percentage of cells having modified or lacking detectable stress fibers. (E,F) Statistical analysis of another microinjection experiment demonstrating the effect on chromatin condensation (panel E) and the correlation of the effects on chromatin and the vimentin IF network within individual cells (panel F). Each point represents analysis of 100-200 cells per time point. The percentages in (B-F) were calculated relative to the number of cells actually recovered at each time point.

Fig. 2.

(A-D) Statistical analysis of 4 independent microinjection experiments. Each point represents results obtained from 200-500 cells. The continuous lines connect points from cells injected with HIV-1 PR and the broken lines connect those injected with BSA. The bars show one standard deviation above and below the mean values. As indicated, panel A shows the cell recovery rate (number of cells observed divided by number of cells originally injected, x 100%), panel B shows the percentage of cells with a collapsed vimentin IF network, panel C shows the percentage of cells with a rounded morphology and panel D shows the percentage of cells having modified or lacking detectable stress fibers. (E,F) Statistical analysis of another microinjection experiment demonstrating the effect on chromatin condensation (panel E) and the correlation of the effects on chromatin and the vimentin IF network within individual cells (panel F). Each point represents analysis of 100-200 cells per time point. The percentages in (B-F) were calculated relative to the number of cells actually recovered at each time point.

Data from another experiment correlating the kinetics of alterations in the vimentin IF network and changes in chromatin condensation are presented in Fig. 2E,F. An HIV-1 PR-dependent condensation of chromatin was observed in more than 70% of the cells at 15 min postinjection, reaching a maximum of 80% at 30 min (Fig. 2E). The collapse of the vimentin IF network was observed in only 20% of the cells 30 min post-injection and reached a peak value of 50-60% several hours later, similar to the effect seen in the experiments reported in Fig. 2B. After 2-4 h, more than 90% of those cells demonstrating chromatin condensation also had an aberrant vimentin IF network (Fig. 2F). As in the other experiments, the microinjection of BSA was largely without effect.

Microinjection of various amounts of HIV-1 PR (achieved by varying the concentration of PR) demonstrated that the chromatin condensation, stress fiber breakdown and collapse of the vimentin IF network are all independent effects (Fig. 3). Chromatin condensation occurred most rapidly and at the lowest concentration of HIV-1 PR; collapse of the vimentin IF network took the longest time and the highest concentration of PR, while the effect on stress fibers was intermediate.

Fig. 3.

Differential sensitivity of cellular effects to microinjection of HIV-1 PR. As indicated, the percentage of cells showing effects on chromatin condensation, stress fibers and vimentin IF distribution were analyzed at 15 min (A) and 4 h (B) after injection of 3 different amounts of HIV-1 PR (realized by microinjection of a constant volume (10−7μl) of various concentrations of HIV-1 PR). Each point was derived from the analysis of 100-200 cells.

Fig. 3.

Differential sensitivity of cellular effects to microinjection of HIV-1 PR. As indicated, the percentage of cells showing effects on chromatin condensation, stress fibers and vimentin IF distribution were analyzed at 15 min (A) and 4 h (B) after injection of 3 different amounts of HIV-1 PR (realized by microinjection of a constant volume (10−7μl) of various concentrations of HIV-1 PR). Each point was derived from the analysis of 100-200 cells.

A number of individual cells were observed over a period of 24 h in phase contrast microscopy following microinjection. Most (80%) were observed to round up at some time following microinjection of HIV-1 PR; many went on to die or detach from the substratum (data not shown). The percentage of affected cells (80%) obtained in the single cell experiments was much higher than in the experiments described in Fig. 2C (peak value of 40%), since the single cell results represent the sum total of affected cells over 24 h. The values reported in Fig. 2C represent only those, fewer, cells affected at a single time point. Conceptually, the 80% value obtained from the single cell experiments represents the summation of affected cells reported at all time points in Fig. 2C. It is not possible to calculate this sum percentage accurately from the type of results presented in Fig. 2C, since some rounded cells were lost and others would undoubtedly have been present at multiple time points (especially at the earlier time points) had they been assayed with a vital assay rather than being fixed for immunofluorescence microscopy. With this caveat in mind, the two seemingly discordant values (80% total versus 40% peak value) are indeed mutually compatible. Loss of rounded cells from the substratum due to mechanical stress means that the values presented in Figs 2 and 3 are minimum estimates of the effect of the HIV-1 PR, since those cells most affected would be expected to be rapidly lost.

Two major differences were found between the patterns of [35S]methionine-labelled proteins separated by 2-D gel electrophoresis following microinjection of HIV-1 PR and BSA (Fig. 4). Beginning at 30 min and most pronounced at 4h post-injection, the amount of intact vimentin was reduced and, concomitantly, lower molecular weight peptides appeared in extracts of cells injected with HIV-1 PR. These peptides co-migrated with the authentic primary and secondary vimentin cleavage products produced in in vitro reactions (Shoeman et al. 1990a), whose positions are indicated by arrows in Fig. 4. No other consistent changes were observed in the roughly 100 specific proteins resolved on these gels during the first 30 min following microinjection. At later time points, some proteins of higher Mr were reduced in intensity, while several proteins of lower Afr increased in intensity (Fig. 4C). The reduced amount of radioactivity found in the 4 h sample is most likely due to the above mentioned loss of cells following HIV-1 PR injection.

Fig. 4.

2-D gel electrophoresis and fluorography of r36S]methionine-labelled proteins from cells microinjected with either BSA (A) or HIV-1 PR (B,C) and incubated for 30 min (B) or 4 h (A,C). In panel A, the positions of actin (A), a-actinin (aA), tubulin (T) and vimentin (V) are indicated. In panel B, the positions of the vimentin cleavage products (arrows) are indicated; they were determined by adding in vitro cleavage products to the labelled proteins and staining with silver prior to autoradiography (panel B, insert).

Fig. 4.

2-D gel electrophoresis and fluorography of r36S]methionine-labelled proteins from cells microinjected with either BSA (A) or HIV-1 PR (B,C) and incubated for 30 min (B) or 4 h (A,C). In panel A, the positions of actin (A), a-actinin (aA), tubulin (T) and vimentin (V) are indicated. In panel B, the positions of the vimentin cleavage products (arrows) are indicated; they were determined by adding in vitro cleavage products to the labelled proteins and staining with silver prior to autoradiography (panel B, insert).

We chose to employ cultured human skin fibroblasts in our studies since these cells grow well, have a well-spread and flattened morphology and possess particularly impressive networks of intermediate filaments, stress fibers and microtubules. Thus they represent an ideal model system for investigating the effect of microinjected HIV-1 PR on cytoskeletal structures and their subunit proteins. These same cytoskeletal structures are found in T4 lymphocytes, which are normally regarded as the major cell type targeted by HIV-1 during infection of humans. However, T4 lymphocytes have a rounded morphology with only a thin rim of cytoplasm surrounding the nucleus, making the detailed study of cytoskeletal structures with the light microscope particularly difficult. While it was not our intention to establish a system that perfectly mimics the natural life cycle of HIV-1, it is not at all unreasonable to study the effect of HIV-1 PR on human skin fibroblasts since lentiviruses have a known affinity for fibroblasts (discussed by Rosenberg and Fauci, 1991), HIV-1 can naturally infect skin cells, notably Langerhans cells (Braathen, 1988; Stingl et al. 1990) and some strains of HIV-1 can infect CD4-negative human foreskin fibroblasts grown in tissue culture (Tateno et al. 1989).

Our data show that vimentin is cleaved by HIV-1 PR in vivo following its microinjection, since the peptides produced co-migrate with authentic vimentin cleavage products obtained in vitro using purified components and are clearly distinct from those produced by cellular proteases. Vimentin (and IF proteins in general) is exquisitely sensitive to calcium-activated neutral thiol proteases (calpains) (Vorgias and Traub, 1986; Traub et al. 1988); the major calpain cleavage sites in vimentin are located in the head and rod domains (Fischer et al. 1986). This enables us to rule out both artifactual degradation of vimentin by calpains during sample preparation, and the known cell cycle-specific cleavage of vimentin (which is presumably calpain-modulated) (Bravo and Celis, 1980; Bravo et al. 1982) as explanations for the cleavage of labelled vimentin seen in Fig. 4. Furthermore, this cleavage was indeed HIV-1 PR-dependent, since microinjection of BSA was without effect (thereby ruling out an activation of cellular proteases via the trauma associated with microinjection). Cleavage of vimentin was clearly evident (perhaps 5-10% of the total) 30 min post-injection and continued until about 50% was cleaved after 4h. While both the kinetics and magnitude of this process support the idea that the cleavage of vimentin itself is sufficient for the observed effects on the IF network, it seems rather unlikely that this cleavage of vimentin is responsible for the more rapid effects seen in chromatin condensation and disappearance of stress fibers. Preliminary data in support of this have been obtained from microinjection of HIV-1 PR into SW13 cells, a human adenocarcinoma cell line lacking detectable vimentin (Paulin-Levasseur et al. 1989), since stress fibers disappear and chromatin condensation also occurs (data not shown). The identity of the additional proteins that change in intensity at the later time points (Fig. 4C) are not known; however, proteins with similar mobilities in 2-D gels are likewise observed to change when Triton X-100-resistant cytoskeleton preparations from these cells are treated with HIV-1 PR in vitro (data not shown). We are currently trying to identify these proteins.

Piedimonte et al. (1990) reported the activation of a protease during HIV infection of CEM cells (a CD4positive lymphocyte cell line). They concluded, on the basis of preliminary inhibitor studies and published observations, that this protease activity was due to induction or activation of a host cell protease. Using in vitro cleavage assays, we have previously shown that the IF subunit proteins vimentin, desmin and glial fibrillary acidic protein (Shoeman et al. 1990a), as well as several other cytoskeletal proteins including actin, a--actinin, spectrin and tropomyosins, are also substrates for HIV-1 PR (Shoeman et al. 1991). Analysis of sequence homology suggests that other cytoskeletal proteins such as dystrophin, villin and microtubule-associated protein 2 (MAP-2) are potential substrates for HIV-1 PR (Shoeman et al. 1991). Wallin et al. (1990) have independently shown that HIV-1 PR can cleave in vitro MAP-1 and MAP-2 proteins and, furthermore, that this cleavage interferes with the assembly of microtubules in vitro. Tomasselli et al. (1991a,b) have shown that other non-viral proteins are also cleaved by HIV-1 PR: calmodulin, actin, Alzheimer’s amyloid precursor protein, troponin C and pro-interleukin 1/3. Detailed analysis of all of these data has enabled Poorman et al. (1991) to construct a cumulative cleavage site specificity model for HIV-1 and HIV-2 PRs. Their model proposes that no absolute specificity exists; rather, specificity may result from either a few strong interactions or several moderate interactions between the substrate and HIV-1 PR. We have also observed that cleavage of vimentin by HIV-1 PR affects its polymerization properties in vitro (Shoeman et al. 1990b). These diverse observations provided a rational basis for the more detailed investigation of the phenomenon of altered IF distribution in cells microinjected with HIV-1 PR reported in our original study (Shoeman et al. 1990a).

Since no other major labelled proteins (Fig. 4) in these cells were detectably cleaved by HIV-1 PR, the question of what other proteins are cleaved or what structures are affected by the PR (perhaps by direct binding without cleavage) remains open. The distribution of microtubules was unaffected by microinjection of HIV-1 PR, suggesting that the cleavage of MAP proteins, as found by Wallin et al. (1990),in vitro, either does not occur in these cells in vivo, or is undetectable by the relatively insensitive assay of immunofluorescence microscopy. In the light of our in vitro observations (Shoeman et al. 1991), we thought it likely that the cleavage of actin might be responsible for the disappearance of the stress fibers. Yet, f-actin, as measured by the binding of fluorescent phalloidin, remains in these cells and no major decline in the actin content of these cells is detectable by gel electrophoresis. This would suggest that another protein(s), instrumental in the crosslinking of f-actin into bundles to form stress fibers, may be affected by the PR and may have been missed in our 2-D gel analysis if it is only a minor component and/or if it contains little or no methionine. Likely candidates include a-actinin and tropomyosins (both of which are also substrates for HIV-1 PR in vitro (Shoeman et al. 1991)), myosin, vinculin and talin. These proteins are variously involved in stabilizing, crosslinking or anchoring the actin microfilaments into stress fibers or to the components of the membrane cytoskeleton (reviewed by Lazarides, 1976; Niggli and Burger, 1987). Since HIV-1 PR-injected cells eventually round up and detach from the substratum, adhesion plaques (which contain talin, vinculin, integrins and the extracellular matrix proteins fibronectin and vitronectin (reviewed by Burridge et al. 1988)), or the attachment sites of microfilaments to them, represent likely targets for this enzyme. In this regard, it has been reported (Brands et al. 1990) that adhesion plaques and cell morphology are affected in cells infected with another retrovirus, Rous sarcoma virus, an observation not incompatible with our speculation that this element of the cytoskeleton is a target for modification by retroviral PRs. It should be noted, however, that current thinking holds the phosphorylation of cytoskeletal proteins by a viral protein kinase responsible for the alterations in adhesion plaques (reviewed by Kellie, 1988). This same dogma applies to the changes seen in vimentin phosphorylation and C-terminal cleavage in Moloney murine sarcoma virus-infected cells (Singh and Arlinghaus, 1989); we have previously discussed that these data are not only compatible with, but are perhaps best explained by, a processing of the mouse vimentin by the viral PR (Shoeman et al. 1990a,b).

Preliminary results obtained in vitro (Shoeman et al. 1991 and manuscript in preparation) suggest that the susceptibility of actin (like vimentin) depends on its polymerization state: f-actin is much more resistant to cleavage by HIV-1 PR than g-actin (monomeric actin). Since actin-containing structures are quite dynamic, it may be possible that the cleavage of a very few molecules at the growing end of actin filaments, where monomers are being added, is sufficient to disrupt the normal organization or interaction of the microfilaments with other cellular components. That is, cleavage of relatively few actin molecules may suffice to cause dramatic changes in a cell, while remaining essentially undetectable by our 2-D gel analysis. An as yet unproven cleavage of actin in vivo might also partially explain the observed chromatin condensation, since actin is present within the nucleus (Douvas et al. 1975) and is associated with DNA and the nuclear matrix (Nakayasu and Ueda, 1986; Valkov et al. 1989). In the specialized situation found in human spermatozoa, decondensation of chromatin was found to be caused by polymerization of nuclear actin (Jamil, 1984). It is attractive to speculate that the reverse might also occur: breakdown of nuclear actin polymers may cause chromatin condensation.

The effect on the vimentin IF network in vivo agrees quite well with the observations made in vitro (Shoeman et al. 1990b). Primary cleavage (which results in the removal of the tail domain) of the vimentin subunits of preformed IFs had little effect on filament morphology, and preformed IFs retained a percentage of the polymerizationincompetent secondary cleavage products lacking both head and tail domains. IFs formed from the purified primary cleavage product did, however, exhibit an increased tendency to aggregate laterally and form clumps. Thus, it might be expected that, although many or even most of the vimentin molecules are cleaved by HIV-1 PR in these microinjected cells, the overall network would remain, perhaps as an aggregate of filaments with lateral associations, and only those interactions of the IF network with cellular structures that depend on the tail domain (initially) or the head domain (later time points or locally) would be affected. The tail domain of vimentin has been implicated in the interaction of IFs with the nucleus and other membrane-bound organelles (Fey et al. 1984; Georgatos and Blobel, 1987a,b;Georgatos et al. 1987; Katsuma et al. 1987) and the head domain can interact with negatively charged lipid bilayers (Perides et al. 1987) and probably directly with the plasma membrane as well.

While the experiments described here were not designed to prove the role of HIV-1 PR in host cell pathogenesis, the relative amounts and activities of vimentin and HIV-1 PR in in vitro and in vivo experiments compare favorably to those possibly found in cells actively producing virus or within the local environment of a virus during infection (Shoeman et al. 1990b). Rough calculations yield a turnover number for vimentin in the microinjected cells of 0.06 per min, a value about 100-fold lower than observed in vitro (Shoeman et al. 1990a) with purified components at a lower, more optimal pH (pH 6.5). While no accurate estimates of the number of active PR molecules per HIV-1 virion nor of the amount of active PR released in the cell during the life cycle have been published, it is pertinent to note that Kaplan and Swanstrom (1991) have found that in lytically infected cells up to about 50% of HIV-1 Gag polyprotein is appropriately processed and accumulates within the cytoplasm. They speculate that this activation of HIV-1 PR within the cytoplasm might result in cleavage of cellular proteins. Our results suggest that the cleavage site specificities, activities and amounts of HIV-1 PR are sufficient to cause local perturbations in the host cell cytoskeleton during the infection process.

We have shown that vimentin can be cleaved in vivo in human skin fibroblasts microinjected with HIV-1 PR and conclude that the changes in the IF network in these cells are a direct consequence of the cleavage of the IF subunit protein. We have not been able to identify biochemically the targets affected by HIV-1 PR that are responsible for the changes observed in stress fibers, nuclear morphology and chromatin condensation. This implies that the affected proteins may be regulatory elements in the cytoskeleton and karyoskeleton and/or present in amounts too low to detect in our system. A number of observations from studies of many retroviruses are compatible with, but not proof of, the idea that retroviral PRs, in general, may indeed affect host cell proteins during the viral life cycle. In this regard, it is of interest to note that either the active PR dimer itself or polyprotein cleavage products are toxic to cells transfected with a vector coding for an HIV-1 polyprotein containing a single chain dimer of HIV-1 PR (Krâusslich, 1991). Furthermore, although the etiology is not known, condensation or degeneration of nuclear chromatin similar to that described in this study has been found in cells from a variety of tissues in HIV-1 infected individuals: kidney (Genderini et al. 1990; Rao, 1991), gastrointestinal tract (Dobbins and Weinstein, 1985; Francis, 1990), joint endothelial cells (Bentin et al. 1990), peripheral lymphocytes (Junca et al. 1989), Schwann cell of the peripheral nervous system (Gibbels and Diederich, 1988) and skeletal (Simpson et al. 1990) and cardiac muscle (Cenacchi et al. 1990). We are currently trying to identify the proteins whose modification leads to the unexplained changes presented here and are employing, in addition to microinjection, molecular genetics to enable the expression of the HIV-1 PR in a larger number of cells in vivo.

We thank S. Roy, E. Russoman and S. Monkarsh (Hoffmann-La Roche, Inc., Nutley, NJ 07110, USA) for providing pure HIV-1 PR, Mrs Margot Bialdiga and Mrs Ulrike Traub for providing tissue culture cells and Mrs Heidi Klempp for secretarial work.

Bentin
,
J.
,
Feremans
,
W.
,
Pasteels
,
J.-L.
,
Hauzeur
,
J.-P.
,
Menu
,
R.
and
Appelboom
,
T.
(
1990
).
Chronic acquired immunodeficiency syndrome-associated arthritis: a synovial ultrastructural study
.
Arthritis and Rheumatism
33
,
268
273
.
Braathen
,
L. R.
(
1988
).
Langerhans cells and HIV infection
.
Biomedicine and Pharmacotherapy
42
,
305
308
.
Brands
,
R.
,
De Boer
,
A.
,
Feltkamp
,
C. C.
and
Roos
,
E.
(
1990
).
Disintegration of adhesion plaques in chicken embryo fibroblasts upon Rous sarcoma virus-induced transformation: different dissociation rates
Expl Cell Res
.
186
,
138
148
.
Bravo
,
R.
and
Celis
,
J. E.
(
1980
).
A search for differential polypeptide synthesis throughout the cell cycle of HeLa cells
.
J. Cell Biol
84
,
795
802
.
Bravo
,
R.
,
Small
,
J. V.
,
Fey
,
S.
,
Larsen
,
P. M.
and
Celis
,
J. E.
(
1982
).
Architecture and polypeptide composition of HeLa cytoskeletons: Modification of cytoarchitectural polypeptides during mitosis
.
J. molec. Biol
.
154
,
121
143
.
Burridge
,
K.
,
Fath
,
K.
,
Kelly
,
T.
,
Nuckolls
,
G.
and
Turner
,
C.
(
1988
).
Focal adhesions: Transmembrane junctions between the extracellular matrix and the cytoskeleton. A
.
Rev. Cell Biol
.
4
,
487
525
.
Cenacchi
,
G.
,
Re
,
M. C.
,
Furuni
,
G.
,
La Placa
,
M.
,
Binetti
,
G.
,
Rapezzi
,
C.
,
Alampi
,
G.
and
Magnani
,
B.
(
1990
).
Human immunodeficiency virus type 1 antigen detection in endomyocardial biopsy: an immunomorphological study
.
Microbiologica
13
,
145
149
.
Dobbins
,
W. O.
and
Weinstein
,
W. M.
(
1985
).
Electron microscopy of the intestine and rectum in acquired immunodeficiency syndrome
.
Gastroenterology
88
,
738
749
.
Douvas
,
A. S.
,
Harrington
,
C. A.
and
Bonner
,
J.
(
1975
).
Major nonhistone proteins of rat liver chromatin: Preliminary identification of myosin, actin, tubulin and tropomyosin
.
Proc natn. Acad. Sci. U.S.A
.
72
,
3902
3906
.
Fey
,
E. G.
,
Won
,
K. M.
and
Penman
,
S.
(
1984
).
Epithelial cytoskeletal framework and nuclear matrix - intermediate filament scaffold: Three dimensional organization and protein composition
.
J. Cell Biol
.
98
,
1973
1983
.
Fischer
,
S
,
Vandekerckhove
,
J.
,
Ampe
,
C.
,
Traub
,
P.
and
Weber
,
K.
(
1986
).
Protein-chemical identification of the major cleavage sites of the Ca2+ proteinase on murine vimentin, the mesenchymal intermediate filament protein
.
Biol. Chem. Hoppe-Seyler
367
,
1147
1152
.
Francis
,
N.
(
1990
).
Light and electron microscopic appearances of pathological changes in HTV gut infection
.
Bailliere’s Clinical Gastroenterology
4
,
495
527
.
Gelderblom
,
H. R.
,
Ôzel
,
M.
,
Haussmann
,
E. H. S
,
Winkel
,
T.
,
Pauu
,
G.
and
Koch
,
M. A.
(
1988
).
Fine structure of human immunodeficiency virus (HIV), immunolocalization of structural proteins and virus-cell relation
.
Micron and Microscópica
19
,
41
60
.
Genderini
,
A.
,
Bertani
,
T.
,
Bertoli
,
S.
,
Scorza
,
D.
and
Barbiano di Belgiojoso
,
G.
(
1990
).
HIV-associated nephropathy: A new entity. A study of 12 cases
.
Nephrol. Dial Transpl. Suppl
.
1
,
84
87
.
Georgatos
,
S. D.
and
Blobel
,
G.
(
1987a
).
Two distinct attachment sites for vimentin along the plasma membrane and the nuclear envelope in avian erythrocytes: a basis for a vectorial assembly of intermediate filaments
.
J. Cell Biol
.
105
,
105
115
.
Georgatos
,
S. D.
and
Blobel
,
G.
(
1987b
).
Lamin B constitutes an intermediate filament attachment site at the nuclear envelope
.
J. Cell Biol
.
105
,
117
127
.
Georgatos
,
S. D.
,
Weber
,
K.
,
Geisler
,
N.
and
Blobel
,
G.
(
1987
).
Binding of two desmin derivatives to the plasma membrane and the nuclear envelope of avian erythrocytes: Evidence for a conserved sitespecificity in intermediate filament membrane interactions
.
Proc. natn. Acad. Sci. U.S.A
.
84
,
6780
6784
.
GibbeiS
,
E.
and
Diederich
,
N.
(
1988
)
Human immunodeficiency virus (HlV)-related chronic relapsing inflammatory demyelinating polyneuropathy with multifocal unusual onion bulbs in sural nerve biopsy
.
Acta Neuropathol
.
75
,
529
534
.
Grewe
,
C.
,
Beck
,
A.
and
Gelderblom
,
H. R.
(
1990
).
HIV. early virus-cell interactions
.
J. Acquired Immune Deficiency Syndromes
3
,
965
974
.
HOner
,
S.
,
Citi
,
S.
,
Kendrick-Jones
,
J.
and
Jockusch
,
B. M.
(
1988
).
Modulation of cellular morphology and locomotory activity by antibodies against myosin
.
J. Cell Biol
.
107
,
2181
2189
.
Jamil
,
K.
(
1984
).
Decondensation of human spermatozoal chromatin by nuclear actin polymerization
.
Archives of Andrology
13
,
137
146
.
Junca
,
J.
,
Clotet
,
B.
and
Milla
,
F.
(
1989
).
Multilobated nuclei in peripheral blood lymphocytes from HIV+ patients
.
Bnt. J. Haematol
.
73
,
275
276
.
Kaplan
,
A. H.
and
Swanstrom
,
R.
(
1991
).
Human immunodeficiency virus type 1 Gag proteins are processed in two cellular compartments
.
Proc. natn. Acad. Sci. U.S.A
.
88
,
4528
4532
.
Katsuma
,
Y.
,
Swierenga
,
H. H.
,
Marceau
,
N.
and
French
,
W.
(
1987
).
Connections of intermediate filaments with the nuclear lamina and the cell periphery
.
Biol. Cell
59
,
193
204
.
Kay
,
J.
and
Dunn
,
B. M.
(
1990
).
Viral proteinases: weakness in strength
.
Biochim. biophys. Acta
1048
,
1
18
.
Kellie
,
S.
(
1988
).
Cellular transformation tyrosine kinase oncogenes, and the cellular adhesion plaque
.
BioEssays
8
,
25
30
.
Kohl
,
N. E.
,
Emini
,
E. A.
,
Schleif
,
W. A.
,
Davis
,
L. J.
,
Heimbach
,
J. C.
,
Dixon
,
R. A. F.
,
Scolnick
,
E. M.
and
Sigal
,
I. S.
(
1988
).
Active human immunodeficiency virus protease is required for viral infectivity
.
Proc. natn. Acad. Sci. U.S.A
84
,
8903
8906
.
Krausslich
,
H -G.
(
1991
).
Human immunodeficiency virus protease dimer as a component of the viral polyprotein prevents particle assembly and viral infectivity
.
Proc. natn. Acad. Sci. U.S.A
.
88
,
3213
3217
.
Krausslich
,
H.-G.
and
Wimmer
,
E.
(
1988
).
Viral proteases
.
A. Rev. Biochem
.
57
,
701
754
.
Lazarides
,
E.
(
1976
).
Actin, a-actinin, and tropomyosin interaction in the structural organization of actin filaments in nonmuscle cells
.
J. Cell Biol
.
68
,
202
219
.
Nakayasu
,
H.
and
Ueda
,
K.
(
1986
).
Preferential association of acidic actin with nuclei and nuclear matrix from mouse leukemia L5178Y cells
.
Expl Cell Res
.
163
,
327
336
.
Niggli
,
V.
and
Burger
,
M. M.
(
1987
)
Interaction of the cytoskeleton with the plasma membrane
.
J. Membr. Biol
.
100
,
97
121
.
Oroszlan
,
S.
(
1990
).
The protease of human retroviruses: biosynthesis, structure and function
.
Proceedings of the 20th FEBS Meeting, Budapest/Hungary, August 19-24, 1990
, p.
237
.
Paulin-Levasseur
,
M.
,
Scherbarth
,
A.
,
Giese
,
G.
,
Rôser
,
K.
,
Bohn
,
W.
and
Traub
,
P.
(
1989
).
Expression of nuclear lamins in mammalian somatic cells lacking cytoplasmic intermediate filament proteins
.
J. Cell Sci
.
92
,
361
370
.
Perides
,
G.
,
Harter
,
C.
and
Traub
,
P.
(
1987
).
Electrostatic and hydrophobic interactions of the intermediate filament protein vimentin and its amino terminus with lipid bilayers
.
J. biol. Chem
.
262
,
13 742
13749
.
Piedimonte
,
G.
,
Petronini
,
P. G.
,
Guetard
,
D.
,
Favier
,
V.
,
Borghetti
,
A. F.
and
Montagnier
,
L.
(
1990
).
Protease activation during HIV infection in a CD4-positive cell line
.
Aids Res. Human Retroviruses
6
,
251
260
.
Poorman
,
R. A.
,
Tomasselli
,
A. G.
,
Heinrikson
,
R. L.
and
Kezdy
,
F. J.
(
1991
).
A cumulative specificity model for proteases from human immunodeficiency virus type-1 and type-2, inferred from statistical analysis of an extended substrate data base
.
J. biol. Chem
.
266
,
14554
14561
.
Rao
,
T. K. S.
(
1991
).
Human immunodeficiency virus (HIV) associated nephropathy. A
.
Rev. Med
.
42
,
391
401
.
Roberts
,
M. M.
and
Oroszlan
,
S.
(
1989
).
The preparation and biochemical characterization of intact capsids of equine infectious anemia virus
.
Biochem. biophys. Res. Commun
.
160
,
486
494
.
Rosenberg
,
Z. F.
and
Fauci
,
A. S.
(
1991
).
Immunopathology and pathogenesis of human immunodeficiency virus infection
.
Pediatr. infect. Dis. J
.
10
,
230
238
.
Shoeman
,
R. L.
,
Hôner
,
B.
,
Stoller
,
T. J
,
Kesselmeier
,
C.
,
Miedel
,
M. C.
,
Traub
,
P.
and
Graves
,
M. C.
(
1990a
).
Human immunodeficiency virus type 1 protease cleaves the intermediate filament proteins vimentin, desmin and glial fibrillary acidic protein
.
Proc. natn. Acad. Sci. U.S.A
.
87
,
6336
6340
.
Shoeman
,
R. L.
,
Kesselmeier
,
C.
,
Mothes
,
E.
,
HOner
,
B.
and
Traub
,
P.
(
1991
).
Non-viral cellular substrates for human immunodeficiency virus type 1
.
FEBS Lett
.
278
,
199
203
.
Shoeman
,
R. L.
,
Mothes
,
E.
,
Kesselmeier
,
C.
and
Traub
,
P.
(
1990b
).
Intermediate filament assembly and stability m vitro: Effect and implications of the removal of head and tail domains of vimentin by the human immunodeficiency virus type 1 protease
.
Cell Biol. Int. Rep
.
14
,
583
594
.
Simpson
,
D M.
,
Bender
,
A. N.
,
Farraye
,
J.
,
Mendelson
,
S. G.
and
Wolfe
,
D. E.
(
1990
).
Human immunodeficiency virus wasting syndrome may represent a treatable myopathy
.
Neurology
40
,
535
538
.
Singh
,
B.
and
Arlinghaus
,
R. B.
(
1989
).
Vimentin phosphorylation by p37mos protein kinase in vitro and generation of a 50 kDa cleavage product in V-mos-transformed cells
.
Virology
173
,
144
156
.
Skalka
,
A. M.
(
1989
).
Retroviral proteases: First glimpses at the anatomy of a processing machine
.
Cell
56
,
911
913
.
Stingl
,
G.
,
Rappersberger
,
K.
,
Tschachler
,
E.
,
Gartner
,
S.
,
Groh
,
V.
,
Mann
,
D. L.
,
Wolff
,
K.
and
Popovic
,
M.
(
1990
).
Langerhans cells in HIV-1 infection
.
J. Am. Acad, Dermatol
.
22
,
1210
1217
.
Tateno
,
M.
,
Gonzalez-Scarano
,
F.
and
Levy
,
J. A
(
1989
).
Human immunodeficiency virus can infect CD-I-negative fibroblastoid cells
.
Proc. natn. Acad. Sci. U.S.A
.
86
,
4287
4290
.
Tomasselli
,
A. G.
,
Howe
,
W. J.
,
Hui
,
J. O.
,
Sawyer
,
T. K.
,
Reardon
,
I M.
,
DeCamp
,
D. L.
,
Craik
,
C. S.
and
Heinrikson
,
R. L.
(
1991a
).
Calcium-free calmodulin is a substrate of proteases from human immunodeficiency virus-1 and virus-2
.
Prot. Struct. Fund. Genet
.
10
,
1
9
.
Tomasselli
,
A. G.
,
Hui
,
J. O.
,
Adams
,
L.
,
Chosay
,
J.
,
Lowery
,
D.
,
Greenberg
,
B.
,
Yem
,
A.
,
Deibel
,
M. R.
,
Zurcherneely
,
H.
and
Heinrikson
,
R. L.
(
1991b
).
Actin, troponin-C, Alzheimer amyloid precursor protein and pro-interleukin If) as substrates of the protease from human immunodeficiency virus
.
J. biol. Chem
.
266
,
14548
14554
.
Traub
,
P.
,
Scherbarth
,
A.
,
Willincale-Theune
,
J.
,
Paulin-Levasseur
,
M.
and
Shoeman
,
R.
(
1988
).
Differential sensitivity of vimentin and nuclear lamins from Ehrlich ascites tumor cells toward Ca2+-activated neutral thiol proteinase
.
Eur. J. Cell Biol
46
,
478
490
.
Valkov
,
N. I.
,
Ivanova
,
M. I.
,
Uscheva
,
A. A.
and
Krachmarov
,
C. P.
(
1989
).
Association of actin with DNA and nuclear matrix from Guenn ascites tumour cells
.
Molec. cell. Biochem
.
87
,
47
56
.
Vorgias
,
C. E.
and
Traub
,
P
(
1986
).
Efficient degradation in vitro of all intermediate filament subunit proteins by the Ca2+-activated neutral thiol proteinase from Ehrlich ascites tumor cells and porcine kidney
.
Biosci. Rep
.
6
,
57
63
.
Wallin
,
M.
,
Deinum
,
J.
,
Goobar
,
L.
and
Danielson
,
U. H.
(
1990
).
Proteolytic cleavage of microtubule-associated proteins by retroviral proteinases
.
J. gen. Virol
.
71
,
1985
1991
.
Willingale-Theune
,
J.
,
Schweiger
,
M.
,
Hirsch-Kauffmann
,
M.
,
Meek
,
A. E.
,
Paulin-Levasseur
,
M.
and
Traub
,
P.
(
1989
).
Ultrastructure of Fanconi anemia fibroblasts
.
J. Cell Sci
.
93
,
651
665
.