(Received for publication, July 3, 1996, and in revised form, October 24, 1996)
From the Department of Microbiology and Molecular Genetics, Markey
Center for Molecular Genetics, and Vermont Cancer
Center, University of Vermont, Burlington, Vermont 05405
We addressed the question of whether furin is the endoprotease primarily responsible for processing the human immunodeficiency virus type I (HIV-I) envelope protein gp160 in mammalian cells. The furin-deficient Chinese hamster ovary (CHO)-K1 strain RPE.40 processed gp160 as efficiently as wild-type CHO-K1 cells in vivo. Although furin can process gp160 in vitro, this processing is probably not physiologically relevent, because it occurs with very low efficiency. PACE4, a furin homologue, allowed processing of gp160 when both were expressed in RPE.40 cells. Further, PACE4 participated in the activation of a calcium-independent protease activity in RPE.40 cells, which efficiently processed the gp160 precursor in vitro. This calcium-independent protease activity was not found in another furin-deficient cell strain, 7.P15, selected from the monkey kidney cell line COS-7.
Previous studies have demonstrated the importance of endoproteolytic processing for the biological activation of many proteins in mammalian cells (1, 2). Many viruses also require processing of their envelope proproteins by host cell proteolytic enzymes for pathogenicity, as exemplified by the human immunodeficiency virus type I (HIV-I)1 (3, 4). Recently a novel family of endoproteases, similar to the KEX2 protease from Saccharomyces cerevisiae, has been shown to play a role in proprotein processing in mammalian cells (5). This family includes furin, PACE4, PC1/PC3, PC2, PC4, and PC5/PC6 (6, 7, 8, 9, 10, 11).
We have previously reported results of our studies of mammalian cell strains that are deficient in endoproteolytic processing (12, 13, 14). The mutant cell strain RPE.40 does not produce active furin (15, 16), a subtilisin-like eukaryotic endoprotease. Furin is a 90-kDa membrane-bound, CA2+-dependent serine protease that catalyzes cleavage of a wide variety of precursor proteins at sites marked by the sequences -Arg-X-Lys/Arg-Arg- (1) or -Arg-X-X-Arg- (17, 18). RPE.40 cells were selected for resistance to Pseudomonas exotoxin A following ethylmethane sulfonate mutagenesis of Chinese hamster ovary cells (CHO-K1) (12). RPE.40 cells, which fail to process Pseudomonas exotoxin A, are also impaired in the proteolytic processing of viral glycoproteins of Sindbis virus (SV), Newcastle disease virus (NDV), and of the insulin- and lipoprotein receptor-related protein (LRP) proreceptors. Expression of mouse furin in RPE.40 cells restores processing of viral envelope glycoprotein precursors and sensitivity to SV and NDV, and also restores processing of the insulin proreceptor and binding of insulin, and sensitivity to Pseudomonas exotoxin A (13, 14, 15, 19, 20, 21). Analysis of fur cDNAs showed that RPE.40 cells are diploid at the fur locus and have a Cys (TGC) to Tyr (TAC) mutation in codon 196 of one allele. In addition, pre-mRNAs transcribed from the second allele are spliced incorrectly (16). These results confirmed that RPE.40 cells do not produce a functional furin endoprotease.
A number of recent studies have focused on the identification of the endoprotease which processes the precursor of the HIV-I envelope glycoprotein, gp160. gp160 is cleaved in mammalian cells to produce a large external glycoprotein (gp120) and a smaller transmembrane protein (gp41) (4). This processing is required for viral pathogenicity. Recently, furin was shown to process gp160 when it was coexpressed with the viral protein in CV-1 cells (22). It has also been suggested that PACE4, a closely related endoprotease found in most mammalian cells, may process gp160 (23). However, the role of furin, or of other subtilisin/kexin-like protease(s) such as PACE4, in the processing of gp160 glycoproteins in vivo is unclear.
In this report, we present evidence of endogenous proteolytic activity in the furin-deficient cell strain RPE.40 that processed gp160 in vivo and in vitro. We have determined that furin can process gp160, but that this ability is probably not physiologically relevent. We have also examined the role of PACE4 in processing gp160 when expressed in RPE.40 cells and a second furin-deficient cell strain, 7.P15, selected from the monkey kidney cell line COS-7. Evidence presented suggests PACE4 participates in the activation of a calcium-independent protease activity in RPE.40 cells, but not in 7.P15 cells, which processes the gp160 precursor efficiently in vitro.
Soybean trypsin inhibitor,
leupeptin, phenylmethylsulfonyl fluoride,
(2S,3S)trans-epoxysuccinyl-L-leucyl-amido-3-methylbutane ester (E64d), bovine serum albumin, pepstatin, aprotinin, EDTA, and
iodoacetamide were purchased from Sigma. Dithiothreitol was purchased
from U. S. Biochemical Corp. Purified recombinant human furin was a
gift of Dr. Gary Thomas (Vollum Institute, Oregon Health Sciences
University, Portland, OR). The PACE4 cDNA in pBluescript SK was a
gift from Dr. Steven Smeekens and the Chiron Corp.
CHO-K1, LoVo human colon carcinoma cells, and COS-7 monkey kidney cells were obtained from the American Type Culture Collection (Rockville, MD). The isolation and characterization of the mutant cell strain, RPE.40, has been described (11). Mutant cell strain 7.P15 was selected from COS-7 cells, also on the basis of resistance to Pseudomonas exotoxin A. Somatic cell hybridization studies determined that it was of the same complementation group as RPE.40. Cells were cultured in DME/F-12 (Sigma) containing 5% fetal bovine serum at 37 °C, in an atmosphere of 5% CO2 in air. Anthrax toxin protective antigen (PA) was the gift of Stephen Leppla (NIDR, National Institute of Health, Bethesda, MD).
Preparation of Recombinant Human Adenovirus Type 5 (Ad5env) Viral StocksThe recombinant adenovirus type 5 encoding the entire HIV-I env gene was a gift from Dr. Norman Salzman of Georgetown University School of Medicine (24, 25). Confluent monolayers of Hela cells in tissue culture plates (100 mm) were infected at a multiplicity of infection of 10 plaque-forming units/cell for 1 h. The inoculum was removed, 10 ml of growth medium (DME/F-12 supplemented with 5% fetal bovine serum) were added, and the cells were incubated for 24 h. The medium was aspirated, fresh medium (6 ml) was added, and the cells were incubated overnight. The cells were disrupted with two cycles of freeze-thawing, and cell debris was removed with low speed centrifugation. The concentration of the virus was titered by plaque assay on monolayers of Hela S3 cells.
Preparation of Cell Membrane FractionsCell membrane fractions were prepared as described previously (21).
Immunoprecipitation of HIV-I gp160A total of 2.2 × 106 cells were plated on tissue culture dishes (60 mm) 24 h prior to the experiment. The cells were infected with Ad5env at a m.o.i of 10. After 1-h incubation, the inoculum was removed, and fresh DME/F12 medium (5 ml) supplemented with 5% dialyzed fetal bovine serum was added to the plate and incubated for 19 h. The cells were labeled for 2 h with Tran35S-label (ICN Biomedical) at 70 µCi/ml. The cells were then lysed in a lysis buffer (50 mM Tris-HCl, 1.0% Triton X-100, 0.5% deoxycholate, 0.1% SDS, 0.15 mM NaCl) and precleared with 25 µl of 50% protein A-Sepharose beads (Sigma) by rocking overnight at 4 °C. Cleared supernatants were transferred to clean tubes. Appropriate amounts of polyclonal human HIV-I pooled serum were added to the cleared lysates and rocked overnight at 4 °C. Protein A-Sepharose beads were added, and the samples were rocked for 2 h at 4 °C. Sepharose beads were washed three times with lysis buffer. Proteins were analyzed by SDS-polyacrylamide gel electrophoresis (PAGE), under reducing conditions, using 7.5% resolving gels. Gels were dried, and autoradiographs were prepared using Kodak X-Omat XAR-5 film.
Preparation and Purification of gp160 ProteinHIV-I gp160
glycoproteins were immunoprecipitated from LoVo cell lysates infected
with Ad5env in the presence of 0.25 µM A23187, a
calcium-specific ionophore (26). The gp160 glycoprotein was recovered
on a ProsieveTM gel system (FMC). Initially, proteins were
suspended in SDS sample buffer and boiled for 3 min. The samples were
then run on 3% resolving agarose gels at 15 mA. The region of the gel
which contained the gp160 protein was identified and removed with a
spatula. The gel slices were added to 9 volumes of extraction buffer
(50 mM Tris-HCl, 1 mM EDTA at pH 8.0) and
melted by heating to 70 °C. The mixture was then allowed to freeze
at 70 °C for 1-2 h and then thawed on ice. The mixture was
centrifuged at 13,000 × g for 10-20 min at 4 °C.
The supernatant containing the protein (gp160) was removed. About 2,000 cpm (gp160) were used in the in vitro assay.
Different concentrations of cell membrane fractions or purified furin were added to a final volume of 30 µl in buffered reaction mixtures containing 3 mM CaCl2 and 0.1% Triton X-100. Reaction mixtures containing 2,000 cpm 35S-labeled gp160 were incubated for 16 h at 30 °C. Reaction mixtures containing 125I-PA were incubated similarly for 4 h. SDS sample buffer was added to quench the reaction, and proteins were resolved and analyzed by SDS-PAGE electrophoresis and autoradiography. The percent of cleavage of gp160 and PA was determined using a GS-250 molecular imager (Bio-Rad).
Transfection of CellsTransfection procedures were as described previously (15). RPE.40 cells were also transfected with PACE4 cDNA subcloned into the mammalian expression vector pSVL (Pharmacia Biotech Inc.). Positive transfectants were screened for expression of PACE4 by determining sensitivity to Sindbis virus and ability to process pro-von Willebrand factor (27). All clones were maintained in selective medium containing G418.
Expression and cleavage of gp160 in CHO-K1 and RPE.40 cells
was examined by immunoprecipitation of
[3H]glucosamine-labeled proteins from lysates of
recombinant Ad5env-infected cells with a polyclonal human
anti-HIV-I serum pool. Analysis by SDS-PAGE and autoradiography showed
that gp160 and its cleavage products, gp120 and gp41, were present in
the lysates. The cleavage of gp160 in the furin-deficient RPE.40 cells
was equal to that in the wild-type CHO-K1 cells (Fig.
1).
Overexpression of Furin in CHO-K1 and RPE.40 Cells Does Not Increase Cleavage of gp160
To determine the effect of
coexpressing furin on the processing of gp160, CHO-K1 (K1.fur5e), and
RPE.40 (40.fur13d) cells, stably transfected and expressing mouse furin
(15), were infected with recombinant Ad5env followed by labeling with
[3H]glucosamine. gp160 and its cleavage products were
then analyzed by autoradiography following immunoprecipitation and
SDS-PAGE. Expression of the recombinant mouse furin in RPE.40 and
CHO-K1 cells did not increase processing of gp160 (Fig.
2) beyond that observed in cells expressing only gp160
(compare to Fig. 1). The pulse-chase analysis of both CHO-K1 and RPE.40
produced similar results (Fig. 2).
Expression of Recombinant Furin Does Not Increase the Processing of gp160 by Cell Membrane Fractions in Vitro
We have previously quantified the relative furin protease activity of CHO-K1 and RPE.40 cell membrane fractions using the protective antigen component of anthrax toxin (PA) as a substrate in vitro (17). RPE.40 cell membrane fractions express no furin activity. Cell membrane fractions derived from furin-transfected RPE.40 cells (40.fur13d) have a higher (75-fold) protease activity than CHO-K1 cell membrane fractions (21).
In vitro, cleavage of purified gp160 by CHO-K1 and RPE.40
cell membranes, and by membranes from furin-transfected RPE.40 cells, generated two products with approximate molecular masses of 120 and 41 kDa (Fig. 3). There was no correlation between the furin activity expressed in the cell membranes and the amount of gp160 processed.
HIV-I gp160 Is Resistant to Cleavage by Purified Recombinant Furin
35S-Labeled gp160 was treated with furin, and
the cleavage products were analyzed as described under "Experimental
Procedures." The gp160 protein was highly resistant to cleavage by
furin (Fig. 4A). The low sensitivity of gp160
to furin can be compared to the high sensitivity of anthrax toxin PA, a
substrate shown to be processed by furin (17) (Fig. 4B).
Serine Protease Inhibitors and Chelators Do Not Reduce the Enzymatic Cleavage of gp160 by RPE.40 Cell Membrane Fractions
The effects of divalent cations and protease inhibitors on cleavage of gp160 in vitro by RPE.40 membrane fractions are outlined in Table I. The presence of divalent cations, EDTA, and EGTA were relatively ineffective in suppressing the processing of gp160 glycoproteins. Although this indicated Ca2+-independent processing of gp160, some increase in activity was observed when Ca2+ or Mg2+ was added. Inhibitors of both cysteine proteases and aspartate proteases, as well as reducing agents and heavy metal ions were inhibitory; however, serine protease inhibitors had no effect (Table I).
|
PACE4 processes subtrates at multibasic cleavage
sites, similar to those recognized by furin (28). To determine if
expression of PACE4 affects processing of gp160, an in vivo
assay was performed. RPE.40 cells were transfected with cDNA
encoding PACE4. RPE.40 and RPE.40 expressing PACE4 (40.P4.T34) were
infected with Ad5env and metabolically labeled with
Tran35S-label, and the gp160 products were
immunoprecipitated and analyzed by SDS-PAGE. Expression of PACE4 did
not increase the cleavage of gp160 by RPE.40 cells (Fig.
5).
Overexpression of PACE4 in RPE.40 but Not in 7.P15 Cells Increases the Efficiency of Processing of gp160 in Vitro
The effect of
expression of PACE4 on the processing of gp160 by membrane fractions
in vitro was also examined. RPE.40 cells were transiently
transfected with a plasmid containing a PACE4 cDNA insert, and the
cell membrane fractions were isolated and used in the in
vitro endoprotease assay. Incubation of gp160 with these membrane
fractions resulted in complete cleavage at a membrane concentration of
10 µg per reaction (Fig. 6A, lane
3). The protease activity was still evident at a concentration of
1 µg of cell membranes per reaction (data not shown). No
membrane-associated endoproteolytic activity was observed when gp160
was incubated with 10 µg of RPE.40 membrane fractions (lane
2).
Surprisingly, cleavage of gp160 by the cell membrane fractions was only partially reduced in the presence of chelators (lanes 4 and 5), indicating that the membrane-associated protease activity has a low requirement for calcium or magnesium. This suggested that the protease activity in RPE.40 cell membranes was not due to PACE4 alone, which requires calcium,2 but rather involved a Ca2+-independent endoproteolytic activity.
To further investigate whether PACE4 processed gp160, we used another furin-deficient mutant cell strain, 7.P15. We have determined that 7.P15 cells, derived from COS-7 cells, are in the same complementation group as RPE.40. 7.P15 cells are unable to process pro-von Willebrand factor. 7.P15 cells were transiently transfected with cDNA encoding PACE4, and cell membranes were prepared to analyze cleavage of gp160 in vitro. To ensure that the transfected 7.P15 cells were expressing PACE4, they were tested for ability to process pro-von Willebrand factor, which is processed by PACE4 (27). Expression of PACE4 in 7.P15 cells resulted in efficient processing of pro-von Willebrand factor in vivo and in vitro (data not shown). However, 10 ug of membranes from 7.P15 cells expressing PACE4 produced low detectable levels of cleaved gp160 products. Addition of EDTA and EGTA to the reaction mixtures inhibited this processing, confirming that this protease activity requires Ca2+ (Fig. 6B). These results reveal that there is a calcium-independent protease activity activated by PACE4 in RPE.40 cells, but not in 7.P15 cells, which is responsible for gp160 cleavage in vitro.
Cleavage of the HIV-I gp160 envelope glycoprotein to a gp120-gp41 heterodimer is essential for its activation (4). Previously, it has been suggested that the mammalian endoprotease furin is important for this cleavage (31, 32). In this report, we have presented evidence that furin is not the major processing enzyme of gp160 glycoproteins in vivo. We have reported previously that the mutant cell strain RPE.40, derived from CHO-K1 cells, does not produce functional furin endoprotease (14, 15). Nonetheless, the gp160 protein was cleaved as efficiently by RPE.40 cells as by the wild-type CHO-K1 cells. Our results are reminiscent of LoVo cells, derived from human colon carcinoma cells. LoVo cells have a frameshift and a Trp to Arg mutation within the homo B domain that disrupts furin activity (33, 34). LoVo cells, like RPE.40 cells, are still able to cleave gp160 into the gp120 and gp41 products (23).
Even when furin was overexpressed in RPE.40 and CHO-K1 cells, it did not increase the proteolytic cleavage of gp160, either in vivo or in vitro. Our results are consistent with the fact that gp160 is highly resistant to cleavage by purified furin; in vitro, gp160 was approximately 1000 times less sensitive to furin than anthrax toxin PA, a substrate known to be cleaved by furin (21). These data lead us to question the idea that furin is the primary protease involved in the cleavage of gp160 in vivo.
We have demonstrated that RPE.40 cells and RPE.40 membrane fractions fail to cleave selected proprotein precursors cleaved by the wild-type cells and their membrane fractions in vivo and in vitro (12, 13, 14, 15, 19, 20, 21). However, both CHO-K1 and RPE.40 cells do have an endogenous activity that cleaves gp160. This is consistent with the processing that is observed in virally infected human T-cells and lymphoid cell lines expressing gp160 (35). The endogenous proteolytic activity expressed in purified RPE.40 membrane fractions was not suppressed in the presence of EDTA or EGTA, nor in the absence of either Ca2+ or Mg2+. However, there was an increase in protease activity in the presence of Ca2+ or Mg2+. Thus, the in vitro studies identified both calcium-dependent and calcium-independent activities that process gp160 in vitro.
PACE4 is an endoprotease which recognizes the same -Arg-X-Lys/Arg-Arg- cleavage motif as furin, and which is widely distributed in mammmalian cells (28, 29, 30). Therefore, we investigated the potential role of PACE4 in processing gp160. Cell membrane fractions from RPE.40 cells transiently transfected with PACE4 processed gp160 at a significantly higher level than did membranes from nontransfected RPE.40 cells. However, an increase in processing was not observed when the gp160 was incubated with membranes from another furin-deficient strain, 7.P15, which was also expressing PACE4. This suggests that expression of PACE4 in RPE.40 cells leads to activation of Ca2+-independent protease activity which is not present in COS-7 cells. Recently, a calcium-independent protease was purified from T lymphocytes that efficiently processes gp160 glycoproteins in vitro (36). The relationship between this protease and the PACE4-activated protease in RPE.40 cells is unknown.
Although processing of gp160 was increased in vitro by membranes from cells transiently transfected with PACE4, processing was not increased in stably transfected RPE.40 cells in vivo. The inefficiency of cleavage may represent a saturation of the cleaving capacity in the secretory pathway of these cells, and a portion of the expressed gp160 may not reach the secretory pathway in which the endoprotease is concentrated (35). However, the role of PACE4 in processing of gp160 in vivo is still uncertain.
Previous studies have implicated furin as playing a major role in the processing of gp160 in human cells (22, 31, 32). Although gp160 contains a potential cleavage site for furin (-Arg-Glu-Lys-Arg-), we have presented evidence that gp160 is cleaved by furin with very low efficiency in vitro. We propose that the ability of the furin-negative cell strain RPE.40 to cleave gp160 may be due primarily to the activity of a calcium-independent protease enzyme, and to a smaller extent a calcium-dependent enzyme which requires a furin-related protease such as PACE4 for activation in vitro. To our knowledge, this is the first evidence for gp160 processing enzyme(s) that requires endoproteolytic processing by a subtilisin-related proprotein convertase, suggesting that cleavage of gp160 is not simple as previously suggested. The significance of this result to HIV-I infectivity in vivo is uncertain. Nevertheless, it is clear that more than one cellular protease can cleave gp160, and that further studies of processing-deficient cell strains such as RPE.40 and LoVo, which are deficient in specific endoproteases, are needed to identify potential protease inhibitors for medical applications.