Nuclear transport of granzyme B (fragmentin-2). Dependence of perforin in vivo and cytosolic factors in vitro.

Cytotoxic T and natural killer cells are able to kill their target cells through synergistic action of the pore-forming protein perforin and the serine protease granzyme B, resulting in very distinctive nuclear changes typical of apoptosis. Whereas perforin acts at the membrane, granzyme B appears to be both capable of entering the cytoplasm of target cells and accumulating in isolated nuclei. In this study we examine nuclear transport of fluoresceinated granzyme B both in vivo in intact cells in the presence of perforin and in vitro in semi-permeabilized cells using confocal laser scanning microscopy. Granzyme B alone was observed to enter the cytoplasm of intact cells but did not accumulate in nuclei. In the presence of sublytic concentrations of perforin, however, it accumulated strongly in intact cell nuclei to levels maximally about 1.5 times those in the cytoplasm after about 2.5 h. In vitro nuclear transport assays showed maximal levels of nuclear and nucleolar accumulation of granzyme B of about 2.5- and 3-fold those in the cytoplasm. In contrast to signal-dependent nuclear accumulation of SV40 large tumor antigen (T-Ag) fusion proteins in vitro, nuclear/nucleolar import of granzyme B was independent of ATP and not inhibitable by the non-hydrolyzable GTP analog GTPγS (guanosine 5′-O-(3-thiotriphosphate)). Similar to T-Ag fusion proteins, however, granzyme B nuclear and nucleolar accumulation was dependent on exogenously added cytosol. Specific inhibitors of granzyme B protease activity had no effect on nuclear/nucleolar accumulation, implying that proteolytic activity was not essential for nuclear targeting. The results imply that granzyme B (32 kDa) may be transported from the cytoplasm to the nucleus through passive diffusion and accumulate by binding to nuclear/nucleolar factors in a cytosolic factor-mediated process. Active and passive nuclear transport properties were normal in the presence of unlabeled granzyme B, implying that the nuclear envelope and pore complex are not granzyme B substrates.

tegrally involved in apoptotic cell death induced in target cells upon their exposure to the contents of lysosome-like cytoplasmic granules (or cytolytic granules) found in cytotoxic T-lymphocytes and natural killer cells (4 -6). Granzyme B acts in conjunction with a second cytolytic granule component, perforin, which can induce complement-like pores within target cell membranes (7,8). As a serine protease, granzyme B is an "Asp-ase," cleaving at aspartic acid residues. The involvement of serine proteases in cytolysis was first suspected on the grounds that serine proteases inhibitors could block cytotoxic T-lymphocyte activity as measured by chromium release (9,10). Later experiments showed that granzyme B can specifically induce rapid DNA fragmentation, but only when cells are simultaneously exposed to sublytic quantities of perforin (11). This implies that granzyme B is the causative agent triggering apoptosis, but that perforin is required in order for it to gain access to its ligand/substrate. Apoptosis is distinguished by very distinctive nuclear changes, whereby the chromatin becomes condensed, the DNA is cleaved into a nucleosomal ladder, the lamina disassembles, and the nucleus ultimately breaks up (12)(13)(14). The signal initiating apoptosis clearly has to be communicated in some form or other to the nucleus; significantly, we have demonstrated that granzymes can be detected in the nuclear lysates of human cytotoxic T-lymphocytes and are recoverable from the nuclei in an active form, in stark contrast to non-granzyme serine proteases, which appear to be confined to the cytoplasm of cells that synthesize them (15). Furthermore, using cell fractionation and confocal laser scanning microscopy (CLSM) 1 (16), we have recently shown that the nuclei of a variety of cells are specifically able to sequester human granzyme B following their release from cytosolic granules, but are unable to accumulate non-granzyme serine proteases, and that free granzyme B can accumulate within nuclei and nucleoli. Pinkoski et al. (17) have reported similar results in terms of granzyme B association with the nucleus and nucleolus using in situ binding assays and immunoelectron microscopy studies of cells expressing the mouse granzyme B cDNA. The implication is that targeting of granzyme B to the nucleus/nucleolus may be a key step in linking granzyme B's proteolytic activity to the nuclear changes of apoptosis.
Precisely scheduled nuclear import of different types of signaling molecules is a key event in the regulation of gene expression and initiation of longer term phenotypic changes such as proliferation and differentiation (see Refs. 18 and 19). Constituents of particular signaling pathways that can translocate * This work was supported by a Clive and Vera Ramaciotti Foundation grant (to D. A. J.) and a project grant and senior research fellowship from the National Health and Medical Research Council of Australia (to J. A. T.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
In the present study we examine the nuclear transport of granzyme B in vivo in intact rat hepatoma cells in the presence of sublytic concentrations of perforin, and nuclear and nucleolar transport of granzyme B in vitro in a semi-permeabilized cell nuclear transport assay system in order to gain insight into its mechanism. We find that granzyme B alone is able to enter intact cells, but requires perforin to accumulate in nuclei, accumulating to levels about 1.5 times those in the cytoplasm within about 2.5 h. In vitro results show that, in contrast to nuclear localization sequence (NLS)-dependent accumulation of SV40 large tumor antigen (T-Ag) fusion proteins, nuclear and nucleolar import of granzyme B is not energy-dependent and not inhibitable by a non-hydrolyzable GTP analog. Similar to T-Ag fusion proteins, however, it is dependent on exogenously added cytosol. Specific inhibitors of granzyme B protease activity have no effect on the nuclear/nucleolar import, implying that proteolytic activity is not essential for nuclear targeting. Taken together, the results suggest that granzyme B may be transported from the cytoplasm to the nucleus through a passive diffusion process and accumulates through binding to nuclear and nucleolar factors. The cellular components mediating nuclear protein import, as well as the nuclear envelope and nuclear pore complex, do not appear to be direct targets of granzyme B proteolytic activity.

MATERIALS AND METHODS
Chemicals and Reagents-Specific antibodies to nucleolin (C23) active in both Western blotting with HTC cell extracts and immunofluorescence with mechanically perforated HTC cells 2 were provided by Dr. Harris Busch (Baylor College, Houston, TX). Isopropyl-1-thio-␤-D-galactopyranoside was from Boehringer Mannheim, and fluorescein isothiocyanate (FITC) and 5-iodacetamidofluorescein from Molecular Probes. Other reagents were from the sources described previously (16, 40 -42).
Cell Culture-Cells of the HTC rat hepatoma tissue culture cell line, a derivative of Morris hepatoma 7288C, were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum as described previously (40,43).
Protein Purification and Fluorescent Labeling-Immunoaffinity purification of human granzyme B from nuclear extracts of YT cells was performed as described (16,44). It was free of granzyme A and Met-ase activities and perforin, as routinely demonstrated by Western blotting and functional assays (16,44). Granzyme B was labeled with FITC as described previously (16) and retained Ͼ75% of its proteolytic activity. Protein concentrations were determined using the dye binding assay of Bradford (45) with bovine serum albumin as a standard, or by measuring absorbance at 280 nm and using the extinction coefficient of 1.20 for human granzyme B (46).
Assay of Proteolytic Activity-Microtiter assays were used to measure the proteolytic activity of unlabeled and labeled granzyme B as described previously in detail (16).
Perforin Sensitivity-Rat perforin was isolated from RNK-16 cells as described (11). A "sublytic" dose of perforin was defined as that producing less than 10% specific 51 Cr release in a 4-h assay at 37°C. This was determined experimentally for HTC cells and all other cell types used.
Intact Cell Experiments in the Presence of Perforin-HTC cells were grown on coverslips to about 50% confluence, washed twice with phosphate-buffered saline, and the coverslip then placed onto a 6-l drop of 10 mM Hepes, pH 7.2, 0.1% bovine serum albumin, 1 mM CaCl 2 , 25 mM NaCl (HBSC buffer), containing either 17 g/ml FITC-labeled granzyme B or 20-kDa FITC-labeled dextran, with or without 2830 units/ml perforin. Perforin was added just prior to commencement of the incubation to prevent polymerization and inactivation in the presence of free Ca 2ϩ . The slide was sealed with nail polish, prior to incubation at 37°C and imaging of fluorescence by CLSM at various times using a Bio-Rad MRC-600 CLSM.
In Vitro Nuclear Transport Assay-Analysis of nuclear import kinetics at the single cell level using mechanically perforated HTC cells in conjunction with CLSM was as described previously (40,(47)(48)(49) and in detail for FITC-labeled granzyme B (16). Incubations were carried out in a final volume of 5 l of transport mix, containing 45 mg/ml cytosolic extract (untreated reticulocyte lysate; Promega catalog no. L415A (150 mg/ml)), an ATP regenerating system (0.125 mg/ml creatine kinase, 30 mM creatine-phosphate, 2 mM ATP), and transport substrate (30 g/ml FITC-labeled granzyme B or 0.2 mg/ml 5-iodacetamidofluorescein-labeled SV40 T-Ag fusion protein or FITC-labeled dextran).
In experiments where the ATP dependence of transport was tested, apyrase pretreatment was used to hydrolyze ATP present in cytosolic extracts (10 min at room temperature with 800 units/ml) and perforated cells (15 min at 37°C with 0.2 units/ml) (50). In vitro transport assays were then performed in the absence of the ATP regenerating system. In experiments where the dependence of transport on the GTP-binding protein Ran/TC4 (51) was tested, cytosolic extract was treated with 850 M GTP␥S (non-hydroyzable GTP analog) for 5 min at room temperature, prior to use in the in vitro assay.
In experiments where protease inhibitors (stock solutions of 20 mM in dimethyl sulfoxide (Me 2 SO)) were used, FITC-labeled granzyme B (2 M) was incubated with a 100-fold excess of the inhibitor (final concentration of 200 M) for 30 min at room temperature, prior to assay in the in vitro transport system as above. The inhibitors used were the specific granzyme B inhibitors t-butyloxycarbonyl-Ala-Ala-Asp-CH 2 F and Val-Ala-Asp-CH 2 F, and the chymotrypsin inhibitor succinyl-Phe-Leu-Phe-CH 2 Cl. As a control, labeled granzyme B was similarly incubated with an equivalent final concentration of Me 2 SO for 30 min at room temperature, and assayed (the final concentration of Me 2 SO in the assay was 0.5%).

Nuclear Localization of Granzyme B in Intact Cells in the Presence of Sublytic Concentrations of Perforin-We have
shown recently that immunopurified granzyme B incubated with YAC-1 nuclei can localize in both the nucleoplasm and nucleolus (16). In order to carry out transport studies both in intact cells and in vitro in semi-permeabilized cells, granzyme B was labeled with FITC resulting in less than 25% loss of proteolytic activity. We first tested whether nuclear uptake of fluoresceinated biologically active granzyme B could be detected in living HTC rat hepatoma cells in the presence of perforin at 37°C. CLSM was used to both visualize (Fig. 1A) and quantify (Fig. 1B) the kinetic process. HTC cells were chosen for this study because they are adherent cells, and hence easier to image in terms of subcellular localization at the nucleolar resolution level, and have been used extensively in in vivo and in vitro nuclear transport studies (Refs. 16, 40, 41, 43, and 49; see below). In addition, HTC cells were found to be relatively resistant to lysis by perforin (data not shown), so that we anticipated that the temporal resolution of cellular uptake and transport would be easier to perform than in other cells. Similar experiments were subsequently carried out with cells in suspension, including mouse FDC P1 cells, 3 results being identical to those with those with HTC cells except that the kinetics of nuclear uptake were much faster, concomitant with an increased sensitivity of FDC P1 cells to perforin (results not shown).
Granzyme B alone was able to enter the cytoplasm of intact HTC cells but did not accumulate in nuclei ( Fig. 1B and data not shown). In the presence of sublytic concentrations of perforin, it was taken up by the cells more rapidly (easily detectable within 2 min; see Fig. 1B, inset), and to a greater extent (2-3 times greater than in its absence up to about 150 min) as indicated by measurements of cytoplasmic fluorescence (Fc) (Fig. 1B, left panel). More significantly, granzyme B accumulated strongly in the cell nuclei in the presence of perforin, to levels maximally about 1.5 times those in the cytoplasm, with maximal accumulation being attained within about 2-2.5 h (Fig. 1B, right panel). Incubation of cells for longer (greater than 5 h) periods with granzyme B in the presence of perforin resulted in detectable nucleolar accumulation (data not shown), which was occasionally difficult to distinguish from relatively highly fluorescent apoptotic bodies also present at higher frequency after 5 h of incubation. Cells that were incubated for over 5 h with granzyme B and perforin at room temperature showed essentially no nuclear localization of granzyme B (data not shown), implying that nuclear targeting was temperature-dependent.
The results thus showed that, while granzyme B is able to

FIG. 1. Nuclear uptake of granzyme B in the presence of sublytic concentrations of perforin in intact HTC cells.
A, visualization of uptake of FITC-labeled granzyme B or a 20-kDa dextran in the absence or presence of sublytic concentrations of perforin in intact HTC cells. Cells were grown for 2 days on coverslips, washed, and then mounted in HBSC buffer (see "Materials and Methods") containing either FITC-labeled granzyme B (17 g/ml) or a 20-kDa dextran (left panel) in the presence of unlabeled perforin (2830 units/ml) and incubated for the times indicated at 37°C. Fluorescence was visualized using CLSM. Later time points (ϳ5-6 h) in the case of granzyme in the presence of perforin indicated nucleolar localization (data not shown). Arrows indicate nuclear perimeter, whereas arrowheads indicate cell perimeter. B, kinetics of cytoplasmic uptake (left) and nuclear accumulation (right) of granzyme B and/or a 20-kDa dextran in the absence or presence of sublytic concentrations of perforin in intact HTC cells. Cells were treated as in A, and image analysis was performed to quantitate fluorescence in various subcellular compartments. Results represent at least 45 separate measurements for each of nuclear (Fn) and cytoplasmic (Fc) fluorescence, respectively, and 12 separate measurements of autofluorescence, with the S.E. not greater than 8% the value of the mean. Specific cytoplasmic fluorescence (Fc) is the fluorescence quantitated in the cytoplasm subsequent to the subtraction of autofluorescence, the inset focusing on the first 24 min to emphasize the rapidity of granzyme B uptake in the presence of perforin. Fn/c is the nuclear to cytoplasmic ratio (after the subtraction of autofluorescence), representing the n-fold accumulation in the nucleus. The Fc value of 1.8 -2.0 for the 20-kDa dextran in the presence of perforin remained constant up to 6 h at 37°C. enter cells in its absence, perforin induces much higher cytoplasmic uptake and is necessary for nuclear accumulation. In contrast to granzyme B, a FITC-labeled 20-kDa dextran did not accumulate appreciably in intact cells in the presence of perforin even up to 6 h ( Fig. 1 and data not shown). This result shows that sublytic concentrations of perforin do not result in the formation of large membrane pores enabling the direct entry of macromolecules such as cytolytic granule components into the cell. The implication is that granzyme B is taken up by a specific process, and that a membrane signaling event effected by perforin is required to bring about nuclear accumulation once inside the target cell.
Reconstitution of Nuclear and Nucleolar Import of Granzyme B in Vitro-To initiate examination of the mechanistic basis of granzyme B accumulation in the nucleus and nucleolus in the absence of perforin, we set out to reconstitute granzyme B nuclear and nucleolar transport in vitro using our previous system of mechanically perforated HTC cells used in conjunction with the technique of CLSM (see Refs. 40,47,and 49). This system relies on the addition of an ATP regenerating system and exogenous cytosol for active, NLS-dependent nuclear protein import (40) (see "Materials and Methods" for experimental details). Granzyme B was tested for nuclear transport properties in mechanically perforated HTC cells and found to show rapid accumulation in both nuclei and nucleoli (Figs. 2 and 3A, top left panels). This nuclear and nucleolar accumulation was in stark contrast to the exclusively nuclear (nucleoplasmic) accumulation (and nucleolar exclusion) of the SV40 T-Ag-NLScontaining fusion protein T-Ag-CcN-␤-Gal (see Fig. 4, middle left panel). The ratio of nucleolar to nuclear fluorescence at steady state was 0.37 Ϯ 0.01 (mean Ϯ S.E., n ϭ 17) for T-Ag-CcN-␤-Gal, compared with a ratio of about 1.2 for granzyme B (see Table I). Results for 20-and 70-kDa FITC-labeled dextrans are also shown for comparison in Figs. 2 (bottom right panel) and 4 (top left panel), respectively. The former showed equilibration between nucleoplasm and cytoplasm but exclusion from the nucleolus, with a ratio of nucleolar to nuclear fluorescence of 0.23 Ϯ 0.09 (mean Ϯ S.E., n ϭ 12) at steady state, whereas the latter was excluded from the nucleus. These results with respect to nuclear entry/exclusion are consistent with the functional molecular weight cut-off for nuclear import of ϳ45 kDa (see Ref. 19), above which proteins require an NLS in order to be targeted to the nucleus, and below which proteins can diffuse freely between nuclear and cytoplasmic compartments. We have previously shown that the non-granzyme serine protease chymotrypsin (ϳ30 kDa) (16), as well as the Pseudomonas aeruginosa protein azurin (9 kDa) 2 are able to enter nuclei, but accumulate in neither the nucleus nor nucleolus in our in vitro transport system; similar results have been reported for other small proteins such as calmodulin (17 kDa) (52). Granzyme B has a molecular mass of ϳ32 kDa and is thus below the molecular mass cut-off for an NLS requirement, but clearly accumulates in nucleus and nucleolus, in contrast to the 20-kDa dextran, which simply equilibrates between the nuclear and cytoplasmic compartments (a ratio of nuclear to cytoplasmic fluorescence, Fn/c, of near 1.0; see also Table I). This implies that granzyme B accumulates in both nucleus and nucleolus through a directed and specific process. Quantitative analysis indicated that the maximal levels of nuclear (Fn/ c(max)) and nucleolar (Fnu/c(max)) accumulation of granzyme B were about 2.5-and 3-fold, respectively, that in the cytoplasm (Fig. 3A, top left panel ; Table I), with half-maximal levels achieved within about 0.58 and 0.92 min, respectively.
Lack of Dependence of Granzyme B Nuclear/Nucleolar Import on ATP-Nuclear protein import has been shown to be an active process, dependent on ATP (Refs. 50 and 53; see also Ref. 19). We tested whether this characteristic also applied to nuclear and nucleolar accumulation of granzyme B (Figs. 2 and 3) by carrying out apyrase pretreatment of both exogenously added cytosol and perforated cells and omitting the ATP-regenerating system in the assay itself (see "Materials and Methods"). Surprisingly, this did not affect nuclear or nucleolar import significantly (Fn/c(max) and Fnu/c(max) of 2.58 and 3.02, respectively, in the absence of ATP, compared with values of 2.53 and 3.0 in its presence) (Fig. 2, middle right panel; Fig.  3A, bottom right panel; Table I). It was concluded that both nuclear and nucleolar accumulation of granzyme B was independent of ATP. This was in stark contrast to the results for T-Ag-CcN-␤-Gal ( Fig. 3B; Fig. 4, bottom right panel; Table I) in contrast to results for T-Ag-CcN-␤-Gal (Table I) and active NLS-dependent nuclear protein import in general (Ref. 51; see also Ref. 19), nuclear and nucleolar transport of granzyme B was not significantly inhibited by the non-hydrolyzable GTP analog GTP␥S (Table I).
Dependence of Granzyme B Nuclear/Nucleolar Import on Cytosolic Factors-Active, NLS-dependent nuclear protein import has been shown to be dependent on cytosolic factors, a number of which have recently been identified (Refs. 51, 54, and 55; see also Ref. 19). We tested whether this was also the case for granzyme B nuclear/nucleolar import (Figs. 2 and 3) by performing transport assays in the absence of exogenously added cytosol (see "Materials and Methods"). Nuclear and nucleolar accumulation of granzyme B was strongly dependent on the addition of exogenous cytosol, since its absence resulted in greatly reduced maximal levels of nuclear and nucleolar accumulation (Fig. 2, middle left panel; Fig. 3A, bottom left panel; Table I). The maximum levels of nuclear and nucleolar accumulation were 1.16 and 1.45, respectively, in the absence of exogenously added cytosol, compared with values of 2.53 and 3.0 in its presence (Fig. 3A and Table I). This indicated that in the absence of cytosol, equilibration between the cytoplasm and

FIG. 3. In vitro nuclear/nucleolar import kinetics of granzyme B (A) and the SV40 T-Ag fusion protein T-Ag-CcN-␤-Gal (B).
Nuclear/nucleolar transport was measured in mechanically perforated HTC cells using CLSM as described under "Materials and Methods" (16,40,(47)(48)(49) in the presence (unless otherwise indicated) and absence of exogenously added cytosol and/or an ATP-regenerating system (see "Materials and Methods") as shown. In the case of granzyme B, kinetics are also shown in the presence of cytosol and an ATP-regenerating system, together with the granzyme B protease inhibitor Val-Ala-Asp-CH 2 F (200 M; top right panel). Measurements are from a single typical experiment (see Table  I Table I). nucleoplasm occurred, with essentially no accumulation in nucleus or nucleolus. Nuclear and nucleolar transport of granzyme B could be concluded to be dependent on cytosolic factors. As expected, nuclear transport of T-Ag-CcN-␤-Gal was also dependent on the addition of exogenous cytosol ( Fig. 3B; Fig. 4, bottom left panel; Table I).
A number of approaches were used in order to try to identify the cytosolic factor(s) essential for granzyme B nuclear/nucleolar accumulation. Among several candidate molecules examined, nucleolin (C23) seemed the most interesting since it is known to shuttle between nucleolus and cytoplasm (56) and has been reported to bind to and act as a substrate for granzyme A (57). It has also been reported to be able to specifically bind to NLS-carrying proteins (58), and, with the granzyme B primary sequence containing several sequences resembling NLSs (data not shown), we hypothesized that granzyme B accumulation in the nucleus and nucleolus might be through binding of these sequences to nucleolin. To test for this possibility, specific antibodies to nucleolin (confirmed to be active by both Western blotting and immunofluorescence experiments) were used to pretreat both cytosol and perforated cells, and nuclear/nucleolar protein import examined in mechanically perforated HTC cells (data not shown). A reduction in maximal nuclear import of only 4.4% was observed due to the anti-C23 antibody, compared with that after identical treatment with a control antibody, implying that C23 is unlikely to play an active role in granzyme B nuclear/nucleolar accumulation (see also Ref. 17).
Passive and Active Nuclear Import in the Presence of Granzyme B-Since one of the results of apoptosis is disintegration of nuclear structure, ultimately leading to breakdown of the nuclear envelope, we examined whether granzyme B's proteolytic activity, instrumental in triggering the distinctive nuclear changes of apoptosis, is directed toward components of the nuclear pore complex or the nuclear envelope. To this end, we examined the passive and active transport properties of the nuclei of perforated HTC cells in our in vitro nuclear transport assay system in the presence of 1 M granzyme B (unlabeled). We compared the nuclear import kinetics of T-Ag-CcN-␤-Gal, as well as the nuclear exclusion properties of the nuclear envelope with respect to a 70-kDa dextran, in the absence or presence of granzyme B (Figs. 4, 5, and 6A; Table I). There was no significant difference in either the maximal level of nuclear accumulation or the rate of nuclear import of T-Ag-CcN-␤-Gal in the presence of granzyme B (Fn/c(max) of 6.48; k ϭ 0.06), or in its absence (Fn/c(max) of 6.39; k ϭ 0.07) (Fig. 4, Table I, indicate that the nuclear pore complex and nuclear envelope are not direct targets of proteolysis by granzyme B. The results also indicate that the cytosolic components of the nuclear import system (51,54,55) remain fully functional in the presence of granzyme B, implying that the nuclear changes elicited during apoptosis are not mediated directly by granzyme B activity.
Transport of Granzyme B in the Presence of Inhibitors of Proteolytic Activity-As a first step toward establishing the mechanistic basis of granzyme B's nuclear/nucleolar accumulation, we examined whether the proteolytic activity of granzyme B was essential for its subcellular targeting. We accordingly pretreated FITC-labeled granzyme B with either of the granzyme B-specific inhibitors t-butyloxycarbonyl-Ala-Ala-Asp-CH 2 F and Val-Ala-Asp-CH 2 F, which reduce granzyme B activity by over 50 and 90%, respectively at 200 M (results not shown), as well as with the chymotrypsin inhibitor succinyl-Phe-Leu-Phe-CH 2 Cl, which does not inhibit granzyme B. The pretreated granzyme B was then tested for nuclear and nucleolar import activity (Figs. 2 and 3A, top right panel in both cases). In no case was a significant reduction in nuclear or nucleolar import efficiency observed (Figs. 2 and 3), e.g. the maximal levels of nuclear and nucleolar accumulation in the presence of Val-Ala-Asp-CH 2 F were 90.1 and 94.3%, respectively, those of untreated granzyme B (Fig. 6B and Table I).
The results for the various inhibitors, summarized in Fig. 6B and Table I, imply clearly that proteolytic activity of granzyme B is unlikely to be essential to its nuclear and nucleolar localization (see, however, Ref. 17). DISCUSSION The intact cell experiments of the present study clearly indicate that the cytolytic granule component serine protease granzyme B can enter intact cells and accumulate in the nucleus in the presence of perforin. Granzyme B would thus appear to be a protein able to be transported all the way from the external plasma membrane surface to the cell nucleus, in comparable fashion to constituents of viral (36 -38), steroid hormone (39), and other signal transduction pathways (20). Granzyme B itself appears to be able to enter cells by an unknown mechanism, but perforin, which does not enter cells (59), is essential for granzyme B's nuclear accumulation. The specificity of granzyme B's uptake and nuclear accumulation in the presence of perforin is demonstrated by the fact that a FITC-labeled 20-kDa dextran does not accumulate in intact cells in the presence of perforin (Fig. 1). This result indicates that the concentrations of perforin used did not result in the formation of large membrane pores enabling the direct entry of granzyme B into the cell. The nature of the membrane signaling event effected by perforin can be surmised in light of the fact that nuclear accumulation of granzyme B can occur in vitro in the presence of exogenous cytosol and absence of perforin.
One possibility is that the cytosolic factors present in the perforated cell nuclear import assay system simulate the membrane signaling event elicited by perforin in intact cells. Alternatively, granzyme B taken up by intact cells may be in a vesicular form and thus not have direct access to the cytoplasm; perforin-induced signaling may be necessary to enable it to access the cytoplasm and thus facilitate subsequent transport to the nucleus.
Our semi-intact cell studies represent, to our knowledge, the first mechanistic examination of the basis of nucleolar protein  h ND, not determined.

FIG. 5. In vitro nuclear import kinetics of the SV40 T-Ag fusion protein T-Ag-CcN-␤-Gal and nuclear exclusion of a 70-kDa FITC-labeled dextran in vitro in the absence (left) and presence (right) of exogenously added granzyme B (unlabeled) as indicated.
Measurements were performed in mechanically perforated HTC cells in the presence of exogenously added cytosol, and an ATP-regenerating system (see "Materials and Methods"). Results are shown for a single typical experiment (see Table I for pooled data), where each point represents the average of at least eight separate measurements for each of Fn and Fc, respectively, with autofluorescence subtracted.
transport in vitro. As demonstrated here, nuclear and nucleolar transport of granzyme B in vitro is dependent on cytosolic factors. However, in contrast to the transport of other nuclear proteins, it appears to be independent of ATP and is not inhibited by a non-hydrolyzable GTP analog. Granzyme B nuclear transport can thus be concluded to be independent of the GTPbinding protein Ran/TC4 (51). This study also establishes that the proteolytic activity of granzyme B is not essential for targeting to the nucleus or nucleolus; that granzyme activity is essential in inducing apoptosis, however, is shown by the observation that protease inhibitors "loaded" into the cytoplasm of target cells specifically block apoptosis (see Ref. 60).
The detailed mechanism by which granzyme B accumulates within the nucleus and nucleolus is unclear. Proteins smaller than ϳ45 kDa are able to diffuse freely into and out of the nucleus, whereas larger proteins require an NLS that interacts directly with the NLS-binding karyopherin/importin subunits (54,55) in order to cross the nuclear envelope (see Ref. 19). Granzyme B is about 32 kDa in molecular mass and hence should not require an NLS to enter the nucleus. A number of small proteins, however, have been shown to exhibit nuclear accumulation (see Ref. 20), including histone 2B (61), and the 13-kDa cyclin-dependent kinase subunit p13 suc1 (52). Of these, a functional NLS has only been defined for histone 2B (61). The fact that granzyme B nuclear import is dependent on cytosolic factors implies that the karyopherin/importin proteins (54,55) may be involved. NLS-dependent protein transport, however, is known to be dependent on both ATP and the monomeric GTPbinding protein Ran/TC4, which does not appear to be the case for granzyme B as shown here, implying rather that granzyme B is not accumulated in the nucleus by a "classical" active, NLS-dependent nuclear protein import pathway (see Ref. 19).
Based on the experimental evidence of this study, it seems plausible to suggest that granzyme B may enter the nucleus and nucleolus by passive diffusion from the cytoplasm, and probably accumulates through binding to target molecules in both nucleoplasm and nucleolus (see also Pinkoski et al. (17)). The rapidity of granzyme B nuclear and nucleolar import (steady state achieved within about 3 and 6 min, respectively; see The fact that nuclear import of granzyme B reaches steady state before nucleolar import implies that nuclear accumulation precedes nucleolar transport (see also Ref. 16), which is also consistent with the intact cell experiments (see "Results"). Nucleolar protein localization of the rRNA gene transcription factor mouse upstream binding factor (62), chicken nucleolin (63), and the yeast NLS-binding protein Nsr1 (64) has been shown to be largely mediated by binding in the nucleolus (which is not a membrane-enclosed structure), and it seems quite likely that granzyme B accumulation in the nucleus and nucleolus is through a similar mechanism. Our experiments using specific antibodies to nucleolin (C23), which is known both to shuttle between nucleolus and cytoplasm (56) and to bind granzyme A (57), indicate that nucleolin is unlikely to play a role in nuclear/nucleolar accumulation of granzyme B (see also Ref. 17). It seems reasonable to speculate, however, that a protein with similar properties may play an integral role in granzyme B nuclear import, either in mediating binding to a nuclear/nucleolar site, or as a specific transport molecule ferrying granzyme B from the cytoplasm to the nucleus/nucleolus, which may well constitute the mechanistic basis of the dependence on cytosolic factors for granzyme B nuclear/nucleolar import. In this regard, a strong candidate for the latter would appear to be a 69-kDa cytosolic protein (which interestingly is also recognized by a proteolytically inactive derivative of granzyme B) identified by Pinkoski et al. (17) using in situ binding and other techniques, whereas a further 80-kDa nuclear protein recognized by granzyme B that is strongly associated with heterochromatin (see Ref. 17) may well constitute the binding site for granzyme in the nucleus. Our in vitro transport system in conjunction with cytosolic fractionation approaches should enable these and other putative factors responsible for granzyme B nuclear and nucleolar localization to be identified and characterized definitively.
The cellular consequences of exposure to perforin and granzyme B are nuclear changes resulting in apoptosis. Whereas perforin acts exclusively at the target cell membrane (59) to provide an apparently additional but essential signal in intact FIG. 6. A, maximal nuclear accumulation of the SV40 T-Ag fusion protein T-Ag-CcN-␤-Gal and a 70-kDa FITC-labeled dextran in vitro in the absence and presence of exogenously added granzyme B (30 g/ml). Measurements were performed in mechanically perforated HTC cells in the presence of exogenously added cytosol, and an ATP-regenerating system as described under "Materials and Methods" (see Fig. 5 and Table I), with the S.E. indicated. B, maximal nuclear and nucleolar accumulation of FITC-labeled granzyme B in vitro in the absence and presence of protease inhibitors (200 M). Measurements were performed in mechanically perforated HTC cells in the presence of exogenously added cytosol, and an ATP-regenerating system as described under "Materials and Methods" (see Fig. 3A, top panels, and Table I), with the S.E. indicated. cells, this and other recent studies (16,17) indicate that granzyme B can target to the nucleus and nucleolus both in vitro and in intact cells in the presence of perforin, implying that granzyme B is the cytolytic granule component that almost certainly initiates DNA fragmentation. The distinctive changes in the nucleus that occur during apoptosis include disassembly of the nuclear lamina, and dissolution of the nuclear envelope (12)(13)(14). We show here that the active transport and passive permeability properties of the nucleus are completely retained in the presence of unlabeled granzyme B, indicating intactness both of the nuclear envelope and the cellular nuclear transport system. The cellular components mediating nuclear protein import, as well as the nuclear envelope and nuclear pore complex, can accordingly be concluded not to be targets of the proteolytic activity of granzyme B. This confirms that the nuclear changes elicited during apoptosis are downstream events within the signaling pathway initiated by granzyme B action on a probably nuclear or nucleolar target(s). One such candidate target is the apoptotic protease CPP32, the precursor of the protease that cleaves the nuclear protein poly(ADP-ribose) polymerase (PARP). Proteolysis of PARP is known to be an early event of apoptosis in intact cells (65), and it is significant that granzyme B has recently been shown to be capable of activating CPP32 in vitro (66). A further potentially critical substrate may be the 80-kDa nuclear protein associated with heterochromatin identified by Pinkoski et al. (17) mentioned above.
We have found that the non-Asp-ase serine proteases granzyme A and Met-ase (cleaving at methionine residues) are also capable of nuclear targeting (see Refs. 15 and 16). 2 In this context, it is interesting that proteases with specificities other than Asp-ase have been shown to be capable of inducing programmed cell death in collaboration with perforin (67,68). It seems likely that granzymes may feed into a common protease cascade at an upstream (such as PARP) or downstream nuclear substrate that is susceptible to proteases of different specificities (see Ref. 69). Future work in this laboratory is focused on determining the precise mechanistic role of granzyme B and other granzymes in eliciting the primary apoptotic signal, as well as trying to elucidate the exact signaling role of perforin in the process. In vitro reconstitution of the complete apoptotic signaling pathway will be a first step in this direction.