INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Fibroblasts microinjected with human immunodeficiency virus type 1 protease (HIV-1 PR) examined at the light microscopic level were found to exhibit rapid changes in nuclear morphology and chromatin organization, followed by disruption of actin-containing stress fibers and an eventual collapse of cytoplasmic vimentin intermediate filaments (IFs) (Höner et al., 1991). Although the effects on the cytoplasmic cytoskeletal structures are readily explained by the cleavage of their integral or associated proteins (Shoeman et al., 1990a, 1991, 1993), the events giving rise to the nuclear alterations have remained an enigma. These nuclear changes occur most rapidly at the lowest concentrations of HIV-1 PR tested (Höner et al., 1991) and closely resemble those described for a variety of tissues examined in HIV-1-infected individuals (Shoeman et al., 1992, for discussion and references).

Because several IF subunit proteins are excellent substrates for HIV-1 PR (Shoeman et al., 1990a) and because the nuclear matrix core filaments are morphologically indistinguishable from the cytoplasmic IFs (Jackson and Cook, 1988; He et al., 1990; Wang and Traub, 1991; Padros et al., 1997), it seemed likely that one or more components of the nuclear matrix might be cleaved by HIV-1 PR and be responsible for the alterations in nuclear structure. Although it was possible to visualize alterations in residual cellular and nuclear structures in HIV-1 PR-treated cells, it was not possible to correlate the cleavage of a nuclear protein (e.g., NuMA; the nuclear mitotic apparatus protein) with the changes in nuclear architecture. Instead, we were able to show, by a combination of cell biological and biochemical approaches, that the activity of the isolated head domain of vimentin (either produced by cleavage by HIV-1 PR or purified biochemically) is both necessary and sufficient to perturb nuclear shape and disrupt chromatin organization.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Proteins and Peptides

Vimentin was purified from mouse Ehrlich ascites tumor cells (Nelson et al., 1982). Mouse vimentin has 97% sequence identity to human vimentin and yields products similar to those obtained from human vimentin when cleaved by HIV-1 PR (Shoeman et al., 1990a). T-vimentin, which lacks the first 70 amino acid residues of vimentin, was prepared as described (Traub et al., 1992a). The amino-terminal vimentin peptide NT1, containing residues 1-96, was prepared from mouse vimentin as previously described (Traub et al., 1992b). Peptide R23R, whose sequence is identical to that of mouse vimentin (vim) residues 22 to 44 (i.e., vim_22-44), peptide R25R (vim_44-68), peptide P410 (vim_3-22) peptide P411 (vim_25-44), and peptide P412 (vim_84-103) were synthesized and purified (>95%) by Neosystem (Strasbourg, France) or Eurogentec (Seraing, Belgium). Peptides were dissolved either in 10 mM K-phosphate, pH 7.5, at a concentration of 2 mg/ml or in 10 mM 3-(N-morpholino)ethanesulfonic acid (MOPS), pH 7 at 0.2 mM for microinjection, as indicated in results. Vimentin deletion mutant proteins, lacking sequences as indicated within the N-terminal head domain, were produced by conventional recombinant DNA technology and purified from Escherichia coli as described (Shoeman et al., 1999). For the initial experiments, purified recombinant HIV-1 PR provided by S. Roy (Hoffmann-La Roche, Nutley, NJ) was used. For the majority of the experiments, recombinant HIV-1 PR prepared in this laboratory from E. coli bearing the plasmid pPTN (Graves et al., 1988; used with permission of Dr. M.C. Graves, Hoffmann-La Roche) according to a protocol supplied by S. Roy was used (details available on request). The HIV-1 PR was dialyzed against 50 mM K-phosphate, pH 7.5, 10% (vol/vol) glycerol and stored in aliquots at 196°C. As in previous experiments (Shoeman et al., 1990a), this preparation was ~10% pure and control bacterial extract, a corresponding chromatography fraction from E. coli cells lacking the HIV-1 PR expression vector, was used as indicated for microinjection controls. Both preparations were used at a protein concentration of 70 µg/ml unless indicated otherwise. Alternatively, 10 mM MOPS, pH 7.0 or 0.1% bovine serum albumin in Ca²⁺- and Mg²⁺-free phosphate-buffered saline (Höner et al., 1991) were used as controls. Limit digests of vimentin and mutant vimentin proteins were produced by incubating the proteins for 3-18 h with an excess of the HIV-1 PR (i.e., the amount of HIV-1 PR, determined by titration, necessary to ensure complete cleavage of the individual proteins in 1 h under standard conditions, was added to the sample mixtures that were then incubated at 37°C for 3 h or overnight). The terminal peptides produced by treatment with HIV-1 PR (which cleaves wild-type vimentin after residues 51, 60, 92, and 422 [Shoeman et al., 1990a]) of vimentin and the mutant vimentin proteins were purified by reverse phase high performance liquid chromatography (HPLC) on a C18 column (Wang et al., 2000). OD_{220 nm} peak fractions were concentrated by drying in a vacuum centrifugal concentrator and were resuspended in 20 µl of 0.1% trifluoroacetic acid. The mass and thus identity of each purified peptide was determined by MALDI-TOF mass spectrometry, using sinapinic acid as a matrix, performed on a Shimadzu MALDI IV instrument (Shimadzu, Duisburg, Germany), in linear, high mass, positive mode with delayed extraction, by using oxidized bovine insulin B chain and bovine ubiquitin (both from Sigma, Deisenhofen, Germany) as mass standards, essentially as described (Wang et al., 2000). Observed mass/charge (m/z) values were accurate to >0.2%, based on the predicted values, except for peptide vimentin_1-(25-63)-92 (0.5%). The amount of each peptide recovered from the HPLC column was determined by UV-spectroscopy in a microcuvette (1-cm path length, illuminated volume ~5 µl), based on an E_{235 nm} of 1.334 for a 1-mg/ml solution of purified NT1 in 0.1% trifluoroacetic acid. The concentration of NT1 was determined as described (Shoeman et al., 1999). The trifluoroacetic acid was removed by vacuum centrifugal concentration and the peptides were resuspended in 10 mM MOPS, pH 7.0 to give a final concentration of 0.2 mM. An aliquot of NT1 was dialyzed against distilled water and then processed as the other peptides to give a final concentration of 0.2 mM in 10 mM MOPS, pH 7.0 buffer (0.2 mM NT1 peptide is equivalent to 2 mg/ml). The cleavages of vimentin proteins or NuMA by HIV-1 PR could be inhibited by pepstatin A, an inhibitor of aspartyl proteases (Seelmeier et al., 1988). For experiments to determine the cellular distribution of NT1 after microinjection, a mutant vimentin protein, containing a cysteine in place of serine₁ as the amino terminal residue, was produced by recombinant DNA technology and partially purified as described (Beuttenmüller et al., 1994). This cys₁-vimentin was labeled with fluorescein-5-maleimide (Molecular Probes, Eugene, OR), essentially according to Molecular Probe's instructions, and digested with endoproteinase Lys-C (Roche Molecular Biochemicals, Mannheim, Germany) to generate a fluorescein-labeled cys₁-NT1 molecule (F-cys₁-NT1). The F-cys₁-NT1 was purified by HPLC on a C18 column and identified by MALDI-TOF, as described above. A vimentin_1-104 NT-green fluorescent protein fusion protein expression vector, in which the first 104 codons of the mouse vimentin cDNA was fused in-frame to the 5' end of a green fluorescent protein open reading frame, also was transfected into SW 13 [vimentin] cells to localize the vimentin_1-104 NT peptide (plasmid and results kindly provided by Dr. U. Niesel, this institute). In initial experiments, substances injected into SW 13 cells subjected only to propidium iodide labeling (see below) were supplemented with FITC-dextran (20,000 average molecular weight; Sigma) at 1 mg/ml to permit unequivocal identification of microinjected cells. For later experiments (i.e., those reported on in Figure 6 and Table 1), the dextran was omitted.

View this table: [in this window] [in a new window]   Table 1.  Statistical evaluation of the effect of microinjected substances on the distribution of chromatin in the nuclei of SW 13 [vimentin] cells

Cells, Microinjection, and Gel Electrophoresis

Primary human skin fibroblasts (HSF cells) (Höner et al., 1991) and cells of the F8 subclone of the human adrenal cortex carcinoma cell line SW 13 (Beuttenmüller et al., 1994) were cultured as previously described. The HSF cells possess vimentin IFs, whereas the SW 13 F8 cells are devoid of vimentin (or any other known cytoplasmic) IFs. Additional SW 13 cell lines (SW 13 [vimentin] and SW 13 T3 M [vimentin⁺]) (Sarria et al., 1990, 1992) were kindly provided by Dr. Robert Evans (University of Colorado Health Sciences Center, Denver, CO) and were grown as described. Cells for microinjection were grown on Eppendorf Cellocate coverslips (Eppendorf, Hamburg, Germany); the uninjected cells in surrounding quadrants of the coverslips served as additional controls. Microinjection was performed essentially as previously described (Höner et al., 1991), except that some of the initial experiments were carried out with a Zeiss AIS automatic injection system (Zeiss, Oberkochen, Germany). Typically, ~20 (manual injection) to 100 or more (AIS injection) cells were injected for each test substance in each experiment. Syringe loading of HSF and SW 13 cells (5 × 10⁵ cells) was performed as described (Clark and McNeil, 1992). Cells were incubated for 15-60 min (as indicated) at 37°C after microinjection or syringe loading, before further treatment for microscopy or gel electrophoresis. Where indicated, vimentin was introduced into SW 13 [vimentin] cells by microinjection of purified vimentin protein (2 mg/ml in 10 mM Na-phosphate, pH 7.0) 24 h before the injection of HIV-1 PR or the control bacterial extract. For some experiments, cells were permeabilized with either digitonin or Triton X-100 and the residual structures obtained were incubated with HIV-1 PR before analysis by gel electrophoresis, electron microscopy, or indirect immunofluorescence microscopy.

One- and two-dimensional gel electrophoresis was performed as described (Höner et al., 1991) and according to the manufacturer's instructions (Bio-Rad, Munich, Germany). Gels were scanned and analyzed with a Pharmacia Imagemaster DTS scanner and software (Amersham-Pharmacia Biotech, Braunschweig, Germany).

Antibodies

Polyclonal goat anti-vimentin antibody was affinity-purified against mouse vimentin, as previously described (Hartig et al., 1997). A monoclonal mouse anti-human NuMA antibody was purchased from Serva (Heidelberg, Germany). Appropriate, FITC-labeled secondary antibodies were from either Sigma, Nordik (Tilburg, Holland), or Dakopatts (Glostrup, Denmark).

Other Materials

Bovine serum albumin, pepstatin A, pluronic F68, propidium iodide, 4', 6-diamidino-2-phenylindole (DAPI), FITC-dextrans, and RNase A were purchased from Sigma. Antibed (diethyl diglycol distearate) was purchased from EM (Chestnut Hill, MA). Vectashield antifading agent was purchased from Serva. Other materials were reagent grade or as described in the references and were generally obtained from either Merck (Darmstadt, Germany), Pharmacia, Roth (Karlsruhe, Germany), or Sigma.

Preparation of Cells for Confocal Laser Scanning (CLS) Microscopy and Electron Microscopy

Cells were fixed in paraformaldehyde and glutaraldehyde, permeabilized with Triton X-100, and incubated with appropriate primary and secondary antibodies by using standard procedures. For CLS microscopy, the permeabilized cells were incubated in phosphate-buffered saline containing 10 U/ml RNase A for 60 min at 37°C and then mounted in embedding medium containing antifading agents and 2 µg/ml propidium iodide. Chromatin also was visualized with DAPI (Shoeman et al., 1999) in cells stained with multiple antibodies for indirect immunofluorescence. Where indicated, some preparations were treated simultaneously with DAPI and RNase (using the same procedure as for propidium iodide staining of chromatin). Alternatively, for electron microscopy, resinless sections (Capco et al., 1984; Fey et al., 1986) were prepared from cells embedded in agarose, extracted with buffers CSK I and CSK II (with or without appropriate nuclease treatment to reveal the nuclear matrix), fixed in glutaraldehyde, dehydrated, and transferred to antibed. Sections of ~300-nm thickness were prepared for electron microscopy by using standard procedures. Antibed was extracted and the sections were critical point dried in CO₂. The sections were sputtered with tungsten at an angle of 5° and then coated with carbon. Sections were viewed and photographed on a Zeiss EM902 electron microscope.

Confocal Laser Scanning Microscopy

CLS microscopy was initially performed on a Zeiss IM35 inverted microscope equipped with a Leica Lasertechnik CLS unit and an argon/krypton ion laser (Leica Lasertechnik, Heidelberg, Germany). Most images were produced using a 63× objective and were scanned using line averaging (8 or 16 scans/line) with an image Z plane spacing of 0.5 µm. For later experiments (Figures 5 and 6), a Leica TCS NT microscopy system with an argon/krypton laser, and an argon UV laser (Leica Microsystems Heidelberg GmbH, Heidelberg, Germany) was used. Images were obtained with either a 63 or 100× objective and were scanned using frame averaging (4 scans/frame) with an image Z plane spacing of 0.243 µm. Images were processed using Application Visualization System software from Advanced Video Systems (Waltham, MA) on a Silicon Graphics Indigo 2 workstation, as described (Hartig et al., 1997). Where only one optical section is presented, it is the equatorial section from a serial section series and/or is representative of the entire series. In one case, pairs of images obtained from the red (propidium iodide) and green (FITC) channels are presented from the same series of optical sections in red and green pseudo colors. Figures were printed using Adobe Photoshop 5.02 software (Adobe Systems, San Jose, CA) and an HP 2000C inkjet printer (Hewlett Packard, Palo Alto, CA). Each experiment reported here was independently repeated two or more times, except for the data presented in Table 1 for which the statistics are provided; for Figures 1-5, the selected images in the figures are typical of the results observed with the entire series of cells treated. The data presented in Table 1 are summaries of the analysis of the entire series of sections obtained from >300 microinjected cells (5.9 gigabytes of data), with at least 10 cells analyzed for each peptide or treatment. Although necessarily subjective in nature, assignments as "normal" or "abnormal" were made by two independent observers and were conservative in nature (i.e., "intermediate" conditions were scored as positive for the MOPS buffer and as negative for the peptides). CLS microscopy parameters, such as photomultiplier gain and offset, pinhole diameter, and laser power, were held constant within any one experiment. Each set of data from each microscopy system was subjected to identical graphic analysis and printing.

View larger version (175K): [in this window] [in a new window]   Figure 1.   Resinless thick sections of HSF cells syringe-loaded with control bacterial extract (a and d) or HIV-1 PR at 35 µg/ml (b and e) or 70 µg/ml (c). The cells were incubated for 30 min at 37°C after syringe loading, embedded in agarose, and prepared for electron microscopy as described in MATERIALS AND METHODS. In cells treated with the control bacterial extract (a and d), the nucleus (N) is uniformly spherical or ovoid, possesses a well-defined nuclear lamina (arrow in a-c, white letter L in d and e) and is filled with an extensive and dense matrix of fibers and filaments. In cells treated with HIV-1 PR, the nuclear periphery is lobed and invaginated (b) and becomes considerably thicker than in control cells (compare L in d and e). In cells subjected to more extensive treatment with HIV-1 PR (c), the nuclear matrix is lost and the nuclear periphery becomes even more prominent, presumably due to the collapse of nuclear matrix remnants onto the lamina. Bar, 1 µm (a-c); 0.1 µm (d and e).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Microinjection of the HIV-1 PR into fibroblast cells results in changes readily detectable at the light microscopic level, including alterations in nuclear shape and chromatin organization, dissolution of stress fibers, and a collapse of the vimentin IF network (Höner et al., 1991; our unpublished results). Electron microscopy of resinless thick sections of HSF cells (Figure 1) after syringe loading revealed several HIV-1 PR-dependent effects. The most dramatic is a shift from a spherical or ovoid nuclear morphology in cells loaded with the control bacterial extract (Figure 1a) to one with multiple lobes or deep invaginations in cells loaded with the HIV-1 PR (Figure 1b). After syringe loading with the HIV-1 PR at a concentration of 35 µg/ml, the nuclear matrix collapsed, with individual elements noticeably thicker than in control-treated cells (Figure 1b), or, at a concentration of 70 µg/ml, was lost from the preparation such that only peripheral rudiments remained (Figure 1c). The extreme nuclear periphery, corresponding to the region where the nuclear lamina and tightly bound chromatin are expected to be associated with the nuclear envelope, generally contained more electron dense material and was generally thicker in cross section in the HIV-1 PR-treated cells (L in Figure 1e) relative to control bacterial extract-treated cells (L in Figure 1d). The changes in nuclear shape and chromatin organization also were clearly detectable using CLS microscopy of propidium iodide-stained, microinjected cells. In contrast to the relatively uniform staining of nuclear chromatin seen in optical sections of HSF cells microinjected with control bacterial extract (Figure 2A), wholesale condensation of chromatin and multiple nuclear membrane invaginations were visible in cells microinjected with HIV-1 PR (Figure 2B). Because the morphology of the control and HIV-1 PR-treated nuclei of the HSF cells observed in CSL microscopy was so similar to that observed in electron microscopy, further microscopic investigations were performed using CSL microscopy. It should be noted that the effects on nuclear chromatin organization readily visible in the propidium iodide-stained cells (Figure 2) were not apparent when similarly treated cells were stained with DAPI by using the standard protocol. Inclusion of a RNase treatment step in the DAPI staining protocol yielded results similar to those seen with propidium iodide staining (our unpublished results).

View larger version (53K): [in this window] [in a new window]   Figure 2.   Distribution of nuclear chromatin in HSF cells injected with either control bacterial extract (A) or HIV-1 PR (B). After microinjection, the cells were incubated for 15 min at 37°C and fixed; DNA was stained with propidium iodide and the preparations were subjected to CLS microscopy. Note the wholesale condensation of chromatin and multiple nuclear membrane invaginations in the HIV-1 PR-injected cell, similar to that seen in the electron micrographs of syringe-loaded fibroblasts (Figure 1, b and c). Bar, 10 µm.

Although the cleavage of vimentin was readily detectable in these cells, our attempts to identify a major nuclear structural protein (via 1D- and 2D-gel electrophoresis) whose cleavage by the HIV-1 PR might be responsible for these alterations in nuclear architecture have, to date, been unsuccessful (our unpublished results). The nuclear lamin A, B, and C proteins were not detectably cleaved (our unpublished results). We have previously reported preliminary results (Shoeman et al., 1996) and hereby confirm that NuMA is cleaved by treatment with HIV-1 PR (our unpublished results); however, the cleavage of NuMA proved to be a surrogate marker for, rather than the cause of, the nuclear alterations. Detailed analysis of the effect of the HIV-1 PR on nuclear organization were hampered not only by a lack of antibodies or other reagents reacting specifically with the core filaments of the nuclear matrix but also by the fact that these structures were unaffected by direct treatment of either digitonin- or Triton X-100-permeabilized cells with HIV-1 PR (our unpublished results). As in previous experiments (Höner et al., 1991), extensive cleavage of vimentin was consistently observed (our unpublished results). We therefore chose to investigate whether the cleavage of vimentin was correlated with any of the nuclear alterations observed in these cells.

Changes in nuclear architecture and chromatin distribution were observed in SW 13 T3 M [vimentin⁺] cells after microinjection of HIV-1 PR (Figure 3C), compared with cells injected with control solution (Figure 3A), although the changes in nuclear shape were less dramatic than those observed with the HSF cells, in part due to the more irregular shape of the nuclei in SW 13 cells compared with HSF cells. No reproducible nuclear changes were observed in any of the SW 13 [vimentin] cells (regardless of origin) after treatment with HIV-1 PR (Figure 3D) or bacterial control extract (Figure 3B), either after microinjection (Figure 3) or syringe loading (our unpublished results).

View larger version (94K): [in this window] [in a new window]   Figure 3.   Presence of the cytoplasmic IF protein vimentin correlates with the occurrence of nuclear aberrations in SW 13 cells after microinjection of the HIV-1 PR. After microinjection of either SW 13 T3 M [vimentin+] cells (A and C) or SW 13 [vimentin] cells (B and D), the cells were incubated for 30 min at 37°C and fixed; DNA was stained with propidium iodide and the preparations were examined via CLS microscopy. (A and B) Distribution of chromatin in the cells of the cell lines SW 13 T3 M [vimentin+] and SW 13 [vimentin], respectively, after microinjection with control bacterial extract control buffer. Equatorial optical sections of the chromatin distribution of a SW 13 T3 M [vimentin+] cell (C) and of a SW 13 [vimentin] cell (D) after microinjection with HIV-1 PR are presented. Although the effect of the action of HIV-1 PR on nuclear chromatin organization is dramatic, the effect on nuclear shape is less obvious, partially due to the more irregular shape of the nuclei of SW 13 cells in comparison to fibroblasts. Bar, 10 µm.

A second type of experiment was performed to confirm whether the presence of vimentin plays a role in the HIV-1 PR-mediated nuclear changes. This approach involved the microinjection of vimentin, before the microinjection of the HIV-PR or vimentin peptides into SW 13 [vimentin] cells. Preinjection of vimentin into SW 13 [vimentin] cells led to the formation of a cytoplasmically extended IF network surrounding the nucleus, both of which were not changed in their organization and structure after microinjection of control bacterial extract (our unpublished results). However, microinjection of HIV-1 PR into such cells brought about substantial alterations in the structure of the vimentin IF network and in nuclear chromatin staining and distribution (our unpublished results). Nuclear shape was not affected. The same changes in nuclear chromatin staining and distribution were observed in SW 13 [vimentin] cells after the injection of a mixture of vimentin peptides, produced by the cleavage of vimentin to completion by the HIV-1 PR in vitro (Figure 4B), whereas these effects could not be seen in control SW 13 [vimentin] cells microinjected with HIV-1 PR (Figure 4A). As illustrated in Figure 4C, even the isolated amino-terminal peptide of vimentin, NT1, was able to induce structural changes in the nuclei of SW 13 [vimentin] cells. Microinjection of the vimentin NT1 peptide into SW 13 T3 M [vimentin⁺] cells resulted in condensation of chromatin and alterations in nuclear shape (Figure 4D), as well as in a perturbation of the endogenous vimentin IF network (our unpublished results). In control experiments (Figure 5), F-cys₁-NT1 (m/z 10,705) was found in both the cytoplasm and nucleus after microinjection into SW 13 [vimentin] cells (Figure 5A). Likewise, both a vimentin_1-104 NT-GFP (green fluorescent protein) fusion protein (M_r ~38,000) and native GFP were found in both the cytoplasm and nucleus of transiently transfected cells (our unpublished results). T-vimentin remained in the cytoplasm of injected cells and had no effect on chromatin distribution (Figure 5B). FITC-dextran, 70,000 MW, remained in the cytoplasm and had no effect on chromatin distribution (Figure 5C), whereas FITC-dextran, 4000 MW, was found in both the cytoplasm and nucleus and likewise had no effect on chromatin distribution (our unpublished results).

View larger version (91K): [in this window] [in a new window]   Figure 4.   Microinjection of the mixture of vimentin cleavage products produced in vitro by the action of HIV-1 PR or a defined vimentin peptide encompassing the entire amino-terminal head domain into SW 13 [vimentin] cells also produces nuclear chromatin condensation and redistribution, but has no effect on nuclear shape. CLS microscopy optical sections of the chromatin distribution of a cell microinjected with HIV-1 PR (7 µg/ml) as a control (A) and a cell microinjected with a mixture of vimentin cleavage products (produced from overnight digestion of vimentin at 0.2 mg/ml) (B) demonstrate that the vimentin peptides, and not the HIV-1 PR, are directly responsible for the nuclear effects. Microinjection of the isolated amino-terminal peptide of mouse vimentin, NT 1 (vimentin residues 1-96), into SW 13 [vimentin] cells resulted in nuclear alterations (C), similar to those seen with the vimentin cleavage product mixture or in SW 13 T3 M [vimentin+] cells injected with HIV-1 PR (Figure 2). Microinjection of NT 1 into SW 13 T3 M [vimentin+] cells and incubation for 30 min at 37°C also resulted in condensation of chromatin and alteration in nuclear shape (D) and a perturbation of the vimentin IF network (our unpublished results). Injected cells were incubated for 30 min at 37°C before fixation, staining with propidium iodide and preparation for CLS microscopy. Bar, 10 µm (A-D).

View larger version (56K): [in this window] [in a new window]   Figure 5.   Intracellular distribution of vimentin peptides and control substances after microinjection into SW 13 [vimentin] cells. F-cys1-NT1 is found in both the cytoplasm and nucleus (A, fluorescein channel). T-vimentin (B, composite panel of both fluorescein channel [green] and propidium iodide channel [red]), visualized with an anti-vimentin antibody and an FITC-labeled second antibody, remains in the cytoplasm (fluorescein channel) and has no effect on nuclear structure or chromatin distribution (propidium iodide channel). FITC-dextrans of molecular weight 20-70,000 or higher remained in the cytoplasm and had no effect on nuclear architecture (C, composite of both fluorescein channel [green] and propidium iodide channel [red] after microinjection of MW 70,000 FITC-dextran). Smaller dextrans (i.e., MW 4000 FITC-dextran) were distributed throughout the cell, i.e., readily entered the nucleus, but had no effect on nuclear architecture (our unpublished results). Bar, 10 µm (A-C).

The results presented in the first five figures are typical results from the evaluation of >150 individual cells. To get a more statistically significant overview and to delineate those residues of the vimentin peptides responsible for the effects observed, a large number of SW 13 [vimentin] cells were microinjected with various purified peptides, each at a final concentration of 0.2 mM, or control substances and subjected to CLS microscopy and analysis. A panel of typical images obtained in this study is presented in Figure 6 and the tabulated data for the 346 cells analyzed is presented in Table 1. The nuclei of cells injected with buffer (Figure 6A and Table 1) were not detectably different from uninjected controls. Microinjection of the carboxy-terminal peptide released from the primary cleavage of vimentin by HIV-1 PR (Shoeman et al., 1990a), vimentin_423-465, had no effect on chromatin (Figure 6H and Table 1). All three amino-terminal peptides, produced by the secondary cleavage of vimentin (Shoeman et al., 1990a), vimentin_1-51, vimentin_1-60, and vimentin_1-92, each individually affected chromatin distribution in SW 13 [vimentin] cells (Figure 6, E-G, and Table 1) as efficiently as the mixture produced by action of the HIV-1 PR on vimentin (Figure 4). Internal peptides, vimentin_17-60 and vimentin_17-92, produced from the vimentin 17 mutant protein, also affected chromatin distribution in SW 13 [vimentin] cells (Figure 6, J and K, and Table 1). The results obtained with the other deletion peptides and the smaller synthetic peptides were somewhat more complicated (Figure 6 and Table 1). Two different regions appear equally able to elicit chromatin aberrations: one region corresponds to part or all of the two DNA-binding wings of vimentin (residues 27-62) and another distal to that (within residues 68-92). Although restraint should be exercised to avoid overinterpreting the data obtained with the mutant peptides (which do not normally occur in nature), it appears as if the first 25 amino acid residues may, in combination with the deletion of the middle of the head domain, inhibit the ability of the distal region to affect chromatin organization, perhaps by interacting with residues between amino acids 63 and 68 (compare the activities of the 1-(25-63)-92 and 1-(25-68)-92 peptides; Table 1). HIV-1 PR had no effect on chromatin in SW 13 [vimentin] cells (Figure 6B and Table 1), but had a dramatic effect on chromatin organization in SW 13 T3 M [vimentin⁺] cells (Figure 6C and Table 1). Together, these results suggest that residues in two regions of the vimentin head domain, i.e., vimentin_17-60 and vimentin_68-92, are those primarily responsible for the effects observed on chromatin organization.

View larger version (140K): [in this window] [in a new window]   Figure 6.   Typical optical sections of cells microinjected with the peptides and substances listed in Table 1. All peptides were used at a concentration of 0.2 mM in 10 mM MOPS, pH 7. All panels show SW 13 [vimentin] cells, except for C in which a SW 13 T3 M [vimentin+] cell is shown. Substances injected were MOPS buffer (A); HIV-1 PR, 10 µg/ml in 10 mM MOPS, pH 7 (B); HIV-1 PR, 10 µg/ml in 10 mM MOPS, pH 7, microinjected into an SW 13 T3 M [vimentin+] cell (C); vimentin peptide NT1 (vimentin1-96) (D); vimentin1-51 (E); vimentin1-60 (F); vimentin1-92 (G); vimentin423-465 (H); vimentin17-51 (I); vimentin17-60 (J); vimentin17-92 (K); vimentin32-51 (L); vimentin32-60 (M); vimentin32-92 (N); vimentin1-(25-38)-51 (O); vimentin1-(25-38)-60 (P); vimentin1-(25-38)-92 (Q); vimentin1-(25-63)-92 (R); vimentin1-(25-68)-92 (S); vimentin1-(44-68)-92 (T); peptide R23R (vimentin22-44) (U); peptide R25R (vimentin44-68) (V); peptide P410 (vimentin3-22) (W); peptide P411 (vimentin24-44) (X); and peptide P412 (vimentin83-103) (Y). Bar, 10 µm (A-Y).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

HIV-1 PR introduced into HSF cells results in dramatic changes in nuclear shape and organization, as well as in alterations of the cytoskeleton, which have been described at the level of conventional light microscopy (Shoeman et al., 1990a, 1993; Höner et al., 1991). Although many mammalian cells are known to contain deep tubular invaginations of the nuclear envelope (Fricker et al., 1997), some of which are associated with perinuclear rings of IFs (Kamei, 1994), and nuclei of the cell line SW 13 are generally irregular in shape (Sarria et al., 1994), the invaginations described in this article in HSF and SW 13 T3 M [vimentin⁺] cells correlate with treatment with the HIV-1 PR. Given the propensity for HIV-1 PR to cleave IF proteins (Shoeman et al., 1990a), we had originally thought that the nuclear matrix core filaments, which are morphologically indistinguishable from the cytoplasmic IFs (Jackson and Cook, 1988; He et al., 1990; Wang and Traub, 1991; Padros et al., 1997), or other components of the nuclear matrix, such as the intranuclear matrix protein, which is related to the nuclear lamins (Menz et al., 1996), would be extensively cleaved by HIV-1 PR. This could not be shown so far. Results of experiments not presented here have shown that these intranuclear filament systems are unaffected by treatment of detergent-permeabilized cells with HIV-1 PR. Although the NuMA protein is cleaved by HIV-1 PR, the facts that the HIV-1 PR treatment of SW 13 [vimentin] cells (whose nuclei contain NuMA) has no influence on nuclear organization, whereas microinjection of vimentin peptides does, demonstrate that the cleavage of NuMA (or any other as-yet-unidentified nuclear component) occurs incidental to the events that ultimately perturb nuclear architecture.

Comparison of results obtained after microinjection of HIV-1 PR into SW 13 [vimentin] and SW 13 T3 M [vimentin⁺] cells or into SW 13 [vimentin] cells previously injected with vimentin provided the first indication that the cleavage products of vimentin might be the effector molecules responsible for the changes in nuclear organization. This was largely substantiated by the microinjection of a mixture of vimentin peptides (produced by limit digestion with HIV-1 PR) into SW 13 [vimentin] cells: alterations in chromatin distribution and condensation were seen, but nuclear morphology was unaffected. Critically important to the interpretation and conclusions of this study are the data obtained from control experiments: the organization of the nuclei of cells lacking vimentin remains unaffected by treatment with HIV-1 PR. Because the limit digest of vimentin by HIV-1 PR produces a mixture of several peptides (Shoeman et al., 1990a,b), corresponding roughly to the head domain, the rod domain and the tail domain, and because it has been shown that the various domains of vimentin can interact with themselves and some other related proteins (Traub et al., 1992b; Meng et al., 1996) as well as with histones (Traub et al., 1986), it was of interest to define which peptide(s) of this mixture was responsible for the changes observed. The amino-terminal peptide NT1 was originally chosen because it was available in sufficient quantity and purity, is almost identical to the largest amino-terminal peptide produced by HIV-1 PR, and because it encompasses a domain of vimentin known to be involved in interactions with a variety of compounds, including nucleic acids, lipids, and coiled-coil protein structures (Traub et al., 1992b and references therein). Changes in chromatin distribution were seen after microinjection of NT1 into SW 13 [vimentin] cells. The changes were identical to those produced by microinjection of the HIV-1 PR-vimentin cleavage products. However, other peptides are also produced (Shoeman et al., 1990a). Therefore, the 16 vimentin peptides liberated after HIV-PR treatment of wild-type and deletion mutant vimentin (Table 1) were all purified by HPLC and identified by their masses in MALDI-TOF analysis. Interestingly, only peptides derived from the N-terminal head domain of vimentin had an effect on nuclear architecture (Figure 6). F-cys₁-NT1 alone, or as a fusion protein with green fluorescent protein, was found both in the cytoplasm and in the nucleus of microinjected or transfected cells. Because all of the other peptides tested (Table 1) are smaller than NT1 (m/z values range from 2041 to 9739; NT1 is m/z 10,283), it is likely that they, too, can enter the nucleus unimpeded by the nuclear pores. Larger vimentin peptides, such as T-vimentin did not enter the nucleus and, importantly, had no effect on nuclear architecture. If the site of action of these peptides is cytoplasmic (i.e., the nuclear effect is an indirect one), it might be expected that T-vimentin should have had an effect on nuclear chromatin organization (because it possesses residues 71-103 of the head domain and thus has at least one of the active domains [see Table 1, peptide 412, vimentin_83-103]). Because it did not, it is likely that the effect of the vimentin peptides is exerted within the nucleus directly on the nuclear matrix or chromatin itself. Taken together, these results suggest that the head domain of vimentin alone is sufficient to perturb chromatin organization in vivo and, furthermore, that the head domain must be liberated from the IF to exert this activity. At least some of the effect of vimentin peptides may be a result of direct interaction of these peptides with nuclear DNA: the DNA binding domain of vimentin has been localized to the middle of the head domain (Shoeman et al., 1999) and, in fact, tyrosine residues Y₂₉, Y₃₇, and Y₅₂ can be photo-cross-linked to DNA bound to vimentin (Wang et al., 2000). The amino-terminal polypeptides of vimentin also may interact with and perturb the 10-nm filament network of the nucleus, thus directly causing the changes in chromatin organization observed. Alternatively, the head domain of vimentin may function as a competitive inhibitor for the lamin B receptor and/or the nuclear lamina, both of which probably provide binding sites for chromatin at the nuclear envelope/periphery (Höger et al., 1991; Luderus et al., 1994; Pyrpasopoulou et al., 1996). Disturbance of nuclear lamina organization by a dominant negative mutant lamin protein has been shown to affect the distribution of several replication factors and to inhibit DNA synthesis (Spann et al., 1997).

It is intriguing that only changes in chromatin distribution but no reproducible alterations in nuclear shape were detected in SW 13 [vimentin] cells, regardless of the peptides injected. On the other hand, alterations in both chromatin organization and nuclear shape, as well as perturbations in the cytoplasmic vimentin IF network, were observed in SW 13 T3 M [vimentin⁺] cells after microinjection not only of HIV-1 PR but also of NT1. Sarria et al. (1994) have previously shown that the nuclei of SW 13 cells are more smooth and regular in appearance when an intact vimentin IF network is present and, furthermore, that disruption of the vimentin IF network by coexpression of a truncated vimentin in these cells gives rise to major alterations in nuclear shape. The conclusion that can be made from these results based on the two disparate approaches is that, when present, an intact vimentin IF cytoplasmic network plays an important role in the establishment and maintenance of a regular nuclear morphology. Because no effect on nuclear shape was observed in SW 13 [vimentin] cells, preinjected with vimentin before the injection of HIV-1 PR, it may be possible that the IFs, when present, must engage in temporally or spatially specific interactions with the nucleus before an effect can be observed.

In a more global sense, electron microscopy of whole-mount preparations has revealed a continuum of filaments, extending from the outside of the cell to the interior of the nucleus (French et al., 1989; Carmo-Fonseca and David-Ferreira, 1990). Maniotis et al. (1997) have shown that this continuum exists in living cells and that IFs play an important role in the transduction of mechanical signals from the extracellular matrix to the nuclear interior. These data provide support for models of gene regulation (Lelievre et al., 1996; Ingber, 1997) that postulate physical connections between the surface of cells and the nucleus, disruption of which may play a role in transformation and carcinogenesis. The methods and reagents used in this study lend themselves to more detailed investigations in this direction and in the general field of chromatin and nuclear matrix interactions and organization. It will be of interest to identify the target(s) with which the vimentin peptides interact; hopefully, such information might provide insight into the normal arrangement of chromatin and the nuclear matrix. Because the head domain peptides of vimentin do not contain any free amino groups, it was not possible to label them with amine reactive reagents or to fix these peptides with aldehydes for localization via indirect immunofluorescence. If problems of poor immunogenicity and fixation can be successfully addressed (perhaps by site-directed mutagenesis), immunoelectron microscopy might prove suitable to localize the vimentin peptides in treated cells and thus provide an explanation for how the changes in nuclear organization are brought about. The results of experiments designed to localize the amino terminal peptides in microinjected cells were particularly disappointing due to their ambiguity: NT1-GFP fusion protein was found in the nucleus, but showed no ability to perturb chromatin and, furthermore, GFP alone also was found in the nucleus.

What is the relevance of these observations with respect to infection by HIV-1 and other related retroviruses? Vimentin may participate in the uptake of HIV-1 preintegration complexes into the nuclear compartment (Thomas et al., 1996) and appears to be important for the cytoplasmic localization of the HIV-1 Vif protein, which regulates viral infectivity by controlling either virus maturation and/or interaction with the cytoskeleton after virus entry (Karczewski and Strebel, 1996). It has been shown that vimentin is cleaved in vivo by the HIV-1 PR after microinjection (Höner et al., 1991), as well as within HIV-1-infected cells (Lindhofer et al., 1993; Konvalinka et al., 1995). A genetic dimer of HIV-1 PR is active when expressed in host cells and exerts a cytopathic or toxic effect (Kräusslich, 1991). Up to 50% of the HIV-1 PR activity of cytopathic strains of HIV-1 is not associated with virions and gives rise to inappropriate processing of the viral polyproteins (Kaplan and Swanstrom, 1991); presumably, susceptible cellular proteins also are cleaved under these conditions, although this was not addressed in this study. In this respect, it is well known that HIV-1 (and many other retroviruses) is not very efficient in terms of packaging viral proteins into intact virions (Brown et al., 1996). HIV-1 PR plays a crucial role in the early phase of viral replication (Nagy et al., 1994) because PR inhibitors effectively block viral replication in a single cycle of infection, altering the stability of unintegrated viral cDNA and affecting the proper formation of the preintegration complex and/or its transport to the nucleus.

We have previously proposed that the HIV-1 PR may play a direct role in cytopathogenesis associated with infection by HIV-1 (Shoeman et al., 1992, 1993). Although HIV-1 cytopathogenesis is most often correlated with syncytia formation, it is noteworthy that treatment of HIV-1-infected macrophage cultures in vitro with a PR inhibitor abolished cytopathogenesis but not syncytia formation (Bergamini et al., 1996). It is unclear how much (or how little) PR activity is actually present in HIV-1-infected cells. Apparently, a lower threshold exists (Rosé et al., 1995), below which mutations in viral polyprotein cleavage sites are selected for to compensate for the lowered PR activity of the mutant PRs expressed when HIV-1 strains are serially passaged in the presence of PR inhibitors (Croteau et al., 1997). It will be of interest to see whether such mutant PRs also show reduced cleavage of cellular protein substrates and thus reduced contributions to cytopathogenesis. In HIV-1-infected individuals, condensation or degeneration of nuclear chromatin has been described in cells from a variety of tissues (Shoeman et al., 1992, 1993 for references and discussion). Furthermore, complications of infection with HIV-1 include an unexplained increase in incidence of cancer, particularly lymphomas. Attempts at identifying an additional agent responsible for non-Hodgkin's lymphoma in a large HIV-1-infected cohort have failed (Armenian et al., 1996), raising the possibility that HIV-1 itself is directly responsible. We have previously described a model that proposes a role for vimentin (and other IF subunit proteins) in the global regulation of gene expression (Traub and Shoeman, 1994). If this model is correct, then the ability of the HIV-1 PR to liberate amino-terminal peptides from vimentin provides a direct means for HIV-1 to interfere with host cell gene expression. Although it is not possible to "prove" that the cleavage of vimentin by HIV-1 PR makes "sense," there is no doubt that it occurs: indeed, native vimentin is one of the best substrates for HIV-1 PR (Shoeman et al., 1990a) and it contains cleavage sites that fit theoretical models very well (Chou et al., 1996). The liberation of the head domain of vimentin by HIV-1 PR may be an important aspect of the HIV-1 replication cycle because it perturbs IF organization and the ability of IFs to interact with important viral proteins or events (Thomas et al., 1996). An added consequence of this cleavage would be the perturbation of nuclear chromatin organization by the vimentin head domain peptides. These changes may be more subtle and long term in affected cells of HIV-1-infected individuals because the total amount of active HIV-1 PR per cell (but not the concentration in areas of viral entry or budding; see Shoeman et al., 1990b) is probably lower than that used in these cell biology experiments. This proposal is supported by our previous study in which we found changes in nuclear chromatin organization at levels of HIV-1 PR activity that were not sufficient to cause detectable changes in the vimentin IF network (Höner et al., 1991); the amino-terminal peptides of vimentin affect nuclear architecture, and presumably nuclear function, long before the organization of the cytoplasmic IF network is disturbed. It is now clear that this apparently paradoxical situation is due to the exquisite sensitivity of the nuclear events and the relative resistance of the IF network to the presence of the vimentin head domain peptides. The collapse of the IF network after the removal of the terminal domains by action of the HIV-1 PR would be expected to be a "late" event in these experiments because it has been shown that the fraction of N-terminally truncated vimentin molecules that can be tolerated in a normal IF network is ~25% (Andreoli and Trevor, 1994). In light of the results presented in this article and the ability of HIV-1 PR to degrade myofibrils when applied exogenously (Shoeman et al., 1993), it becomes especially important to consider treating HIV-1-infected individuals with HIV-1 PR inhibitors to not only block viral replication but to also prevent the HIV-1 PR from contributing to cytopathogenesis and carcinogenesis.