Journal of Protein Chemistry, Vol. 18, No. 1, 1999
`
`Properties of Soluble Fusions Between Mammalian Aspartic
`Proteinases and Bacterial Maltose-Binding Protein
`
`Deepali Sachdev1 and John M. Chirgwin1,2
`
`Received September 4, 1998
`
`The mammalian aspartic proteinases procathepsin D and pepsinogen form insoluble inclusion bod-
`ies when expressed in bacteria. They become soluble but normative when synthesized as fusions
`to the carboxy terminus of E. coli maltose-binding protein (MBP). Since these nonnative states of
`the two aspartic proteinases showed no tendency to form insoluble aggregates, their biophysical
`properties were analyzed. The MBP portions were properly folded as shown by binding to amylose,
`but the aspartic proteinase moieties failed to bind pepstatin and lacked enzymatic activity, indi-
`cating that they were not correctly folded. When treated with proteinase K, only the MBP portion
`of the fusions was resistant to proteolysis. The fusion between MBP and cathepsin D had increased
`hydrophobic surface exposure compared to the two unfused partners, as determined by bis-ANS
`binding. Ultracentrifugal sedimentation analysis of MBP-procathepsin D and MBP-pepsinogen
`revealed species with very large and heterogeneous sedimentation values. Refolding of the fusions
`from 8 M urea generated proteins no larger than dimers. Refolded MBP-pepsinogen was proteo-
`lytically active, while only a few percent of renatured MBP-procathepsin D was obtained. The
`results suggest that MBP-aspartic proteinase fusions can provide a source of soluble but normative
`folding states of the mammalian polypeptides in the absence of aggregation.
`
`KEY WORDS: Aspartic proteinases; protein folding; protein fusions.
`
`1. INTRODUCTION
`
`Cathepsin D, pepsin, and renin are members of the as-
`partic proteinase family and share structural features.
`They are synthesized in the mammalian endoplasmic re-
`ticulum (ER)3 with propeptides of about 45 amino acids.
`The signal peptides that are required for translocation
`across the ER membrane are removed cotranslationally,
`and as the polypeptide grows in length it is believed to
`
`1 Research Service, Audie L. Murphy Memorial Veterans Administra-
`tion Hospital, and Departments of Biochemistry and Medicine, Uni-
`versity of Texas Health Science Center at San Antonio, Texas
`78284-7877.
`2 To whom correspondence should be addressed, at Division of En-
`docrinology and Metabolism, Department of Medicine, University of
`Texas Health Science Center, San Antonio, Texas 78284-7877; e-
`mail: chirgwin@uthscsa.edu.
`3 Abbreviations: ER, endoplasmic reticulum; bis-ANS, l,l'-bis(4-ani-
`lino)naphthalene-5,5'-disulfonic acid; CpD, cathepsin D; MBP, mal-
`tose-binding protein; Pgn, pepsinogen; PMSF, phenylmethane
`sulfonyl flouride.
`
`interact with folding catalysts in the lumen of the ER
`(Nilsson and Anderson, 1991). Procathepsin D is pro-
`teorytically processed via several steps to the mature
`form on the way to and within the lysosome (Hasilik
`and Neufeld, 1980; Hasilik, 1992; Delbruck et al.,
`1994). Procathepsin D is N-glycosylated, while glyco-
`sylation of pepsinogen and prorenin is variable and spe-
`cies specific.
`When expressed in Escherichia coli, the aspartic
`proteinases accumulate in inclusion bodies. Pepsinogen
`is readily renatured after solubilizing inclusion bodies in
`urea and refolding in vitro (Lin et al., 1989, 1995; Cot-
`trell et al., 1995). In contrast, procathepsin D forms
`mostly insoluble aggregates during refolding, with only
`a few percent recovery of native protein (Conner and
`Udey, 1990; Scarborough and Dunn, 1994). This has
`made it difficult to study recombinant procathepsin D.
`The aspartic proteinases are bilobed molecules
`which show a conserved pattern of secondary structure,
`with the active-site cleft located between the two lobes.
`
`127
`
`0277-8033/99/0100-0127$16.00/0 C 1999 Plenum Publishing Corporation
`
`GNE 2007
`Page 1
`
`

`
`128
`
`Sachdev and Chirgwin
`
`The successful prediction and modeling of the structures
`of members of the family based on known three-dimen-
`sional coordinates of other members (Koelsch et al.,
`1995) suggest that a basic folding pathway is probably
`shared by the members of the family. The aspartic pro-
`teinases have a high proportion of B-sheet structure,
`which may favor inclusion body formation when they
`are expressed in E. coli.
`We have previously investigated the bacterial ex-
`pression of the aspartic proteinases as fusion proteins
`with E. coli partners such as maltose-binding protein
`(MBP) and thioredoxin. When the bacterial partners
`were fused to the C-terminus of aspartic proteinases, the
`fusion proteins accumulated in insoluble inclusion bod-
`ies (Sachdev and Chirgwin, 1998a). However, when
`MBP was moved to the N-terminus of the aspartic pro-
`teinase, the MBP-aspartic proteinase fusion proteins
`were expressed at high levels in the cytosol of E. coli
`as soluble and stable proteins (Sachdev and Chirgwin,
`19986). However, the soluble fusions lacked detectable
`proteinase activity, suggesting that the aspartic protein-
`ase portions were nonnative. Therefore, we undertook
`physicochemical characterization of these fusions, both
`as isolated from the bacterial cytosol and after refolding
`in vitro.
`
`2. EXPERIMENTAL PROCEDURES
`
`2.1. General Methods
`
`All reagents were of analytical grade and obtained
`from Sigma (St. Louis, MO) or New England Biolabs
`(Beverly, MA). All DNA constructs have been described
`previously (Sachdev and Chirgwin, 19986). The enzy-
`matic activities of the aspartic proteinases were deter-
`mined by digestion of labeled hemoglobin into peptides
`soluble in 3% trichloroacetic acid (Sachdev and Chirg-
`win, 1998a). Human cathepsin D, a two-chain glycopro-
`tein, was purified from human placentas by pepstatinyl
`agarose affinity chromatography. Sodium dodecyl sul-
`fate-polyacrylamide gel electrophoresis (SDS-PAGE)
`was carried out on 1.5-mm-thick, 12.5% or 10% acry-
`lamide slab gels according to Laemmli (1970). Molec-
`ular weight standards were bought from Sigma (St.
`Louis, MO). Gels were stained with 0.1% Coomassie
`brilliant blue R250 in 40% methanol and 10% acetic
`acid. Gels were destained with 40% methanol and 10%
`acetic acid. Protein concentrations were determined by
`the Bradford method using Bio-Rad (Richmond, CA)
`protein assay dye reagent.
`
`2.2. Pepstatinyl Agarose Binding
`MBP-procathepsin D and MBP-pepsinogen were
`expressed in E. coli and purified to greater than 90%
`homogeneity as described (Sachdev and Chirgwin,
`19986). Refolded MBP-procathepsin D and MBP-pep-
`sinogen, 50-75 ug, were adjusted to 0.1 M sodium ac-
`etate (pH 3.5), 0.1% Brij 35, and 1 M NaCl. Pepstatinyl
`agarose (Pierce, Rockford, IL) was washed three times
`with the same buffer. The protein was mixed with 50 ul
`of a 50% slurry of pepstatinyl agarose overnight at 4°C.
`The resin was pelleted by centrifugation in a microcen-
`trifuge at full speed for 1 min. The supernatant was
`saved as the unbound fraction. The agarose was washed
`3 X with 1-ml aliquots of buffer containing 0.1 M ac-
`etate (pH 3.5), 0.1% Brij 35, and 0.35 M NaCl. The
`bound fraction was eluted with 2 X 500-ul aliquots of
`50 mM Tris (pH 8.6), 0.1% Brij 35, and 1 M NaCl. The
`unbound fraction, wash fraction, and bound fraction
`were acetone precipitated prior to SDS-PAGE analysis
`to reduce volume. They were analyzed by SDS-PAGE
`on 10% polyacrylamide gels followed by Coomassie
`blue staining. To control for nonspecific binding, assays
`were also performed in parallel in the presence of sol-
`uble pepstatin added to the samples prior to mixing with
`pepstatinyl agarose.
`
`2.3. Protease Digestion
`Ten ug each of MBP-procathepsin D and MBP-
`pepsinogen in 0.1 M phosphate buffer (pH 7.4) and 5
`mM EDTA was digested with 0.01 ug (0.1% w/w) or
`0.1 ug (1% w/w) proteinase K for 1 hr at 22°C in 0.1
`M phosphate buffer (pH 7). Proteolysis was stopped by
`adding 50 ug of PMSF from a 1 mg/ml stock and boiling
`the sample. As a control to ensure that the proteinase K
`was completely inactivated, aliquots of each fusion were
`also analyzed by adding 10 ug of fusion protein to sim-
`ilarly inactivated proteinase K. Laemmli sample buffer
`(2 X concentrated) was added to all the samples, and
`they were analyzed by SDS-PAGE.
`
`2.4. Bis-ANS Binding to MBP-Cathepsin D
`
`Fluorescence of l,l'-bis(4-anilino)naphthalene-5,5'-
`disulfonic acid (bis-ANS) was measured using an SLM
`Model 500C fluorometer at an excitation wavelength of
`397 nm after adding 10 uM bis-ANS to 1 mg/ml protein.
`Solutions were mixed thoroughly after addition of bis-
`ANS and the emission spectra recorded over the range
`400-600 nm. bis-ANS binding to placental cathepsin D
`
`GNE 2007
`Page 2
`
`

`
`Aspartic Proteinases
`
`129
`
`and to MBP was compared to its binding to the fusion
`protein, MBP-cathepsin D.
`
`2.7. Solubility of MBP-Procathepsin D at various
`urea concentrations
`
`2.5. Sedimentation Velocity Analysis by Analytical
`Ultracentrifugation
`
`Sedimentation velocity analysis of the various pro-
`teins was performed in a Beckman XL-A analytical ul-
`tracentrifuge. The protein concentrations were between
`0.2 and 0.4 mg/ml. The temperature was kept constant
`at 25°C. The rotor speed varied from 8000 to 30,000
`rpm depending on the sample. The scans were analyzed
`by the method of van Holde and Weischet (1978) using
`the UltraScan ultracentrifuge data collection and analysis
`program (B. Demeler, Department of Biochemistry, Uni-
`versity of Texas Health Science Center). Data were cor-
`rected for buffer density and viscosity. The samples were
`dialyzed against either 0.1 M phosphate buffer (pH 7.0)
`and 5 mM EDTA, or 10 mM Tris (pH 7.4) and 100 mM
`NaCl. Refolded fusions were analyzed in the former buf-
`fer. All of the sedimentation velocity runs were per-
`formed at the Center for Analytical Ultracentrifugation
`of Macromolecular Assemblies, Department of Bio-
`chemistry. The analysis of the sedimentation velocity
`runs is shown as the integral distribution of s20,w for the
`protein. The y axis indicates the fraction of sample with
`S2o,w values less than or equal to the values on the x axis
`(s20,w values).
`
`2.6. Refolding of Purified MBP-Aspartic Proteinase
`Fusion Proteins
`
`Fusion proteins were refolded as described by Ku-
`helj et al. (1995), who described the successful refolding
`from bacterial inclusion bodies of procathepsin B, a ly-
`sosomal cysteine proteinase. MBP-procathepsin D,
`MBP-cathepsin D,
`and MBP-pepsinogen were
`expressed as soluble proteins in E. coli, purified by am-
`ylose resin affinity chromatography, and equilibrated in
`0.1 M phosphate (pH 7.0), 5 mM EDTA to a protein
`concentration of 500 ug/ml. They were denatured by ad-
`dition of solid urea to 8 M and incubation for 4-6 hr at
`room temperature. The denatured proteins were refolded
`by dialysis for 15 hr against 500 volumes of refolding
`buffer, 0.1 M phosphate (pH 7.0), 5 mM EDTA, and 5
`mM cysteine, at 4°C. The cysteine was removed by di-
`alysis for 2 hr against 500 volumes of the same buffer
`without reducing agent. After dialysis, the protein was
`transferred to a polypropylene tube and centrifuged at
`6000 X g for 10 min to remove particulates.
`
`MBP-procathepsin D and unfused MBP were ex-
`pressed, purified by affinity chromatography on amylose
`resin, and equilibrated in 50 mM HEPES (pH 7.6) and
`1 mM EDTA to a protein concentration of 1 mg/ml.
`They were denatured by addition of solid urea to 8 M
`and incubated for 4 hr at 21°C. Equal amounts of each
`protein in 8 M urea were diluted to successively lower
`concentrations of urea (7, 6, 5, 4, 3, 2, 1 M) buffered
`with 50 mM HEPES (pH 7.6) and 1 mM EDTA and
`incubated at room temperature for 2 hr. The samples were
`centrifuged at full speed in a microfuge. The pellet and
`supernatant for each urea concentration were analyzed by
`SDS-PAGE, followed by Coomassie blue staining, to de-
`termine if the proteins remained soluble or were aggre-
`gating at intermediate concentrations of urea.
`
`2.8. Amylose Resin Binding of Refolded MBP-
`Aspartic Proteinase Fusions
`
`Ten ug each of refolded MBP-aspartic proteinase
`fusion proteins was mixed with 25ul of washed amylose
`resin (New England Biolabs, Beverly, MA) in a 1.8-ml
`microfuge tube and incubated on ice for 1 hr. The tube
`was centrifuged for 1 min and the supernatant saved as
`unbound fraction. The resin was washed with 3 X 1-ml
`aliquots of 20 mM Tris-Cl (pH 7.4), 200 mM NaCl, 1
`mM EDTA, and 0.02% Tween-80 buffer by resuspen-
`sion and centrifugation. Bound protein was eluted with
`100ul of the above buffer containing 10 mM maltose,
`and the unbound and bound fractions were analyzed by
`SDS-PAGE.
`
`3. RESULTS
`
`3.1. Folded State of Aspartic Proteinase Portions of
`Fusion Proteins
`
`We previously reported that three aspartic protein-
`ases, procathepsin D, pepsinogen, and prorenin, when
`expressed as fusions with bacterial MBP at the N-ter-
`minus accumulated as soluble proteins in the cytoplasm
`of E. coli even when the propeptides of the aspartic pro-
`teinases were deleted (Sachdev and Chirgwin, 1998b).
`These proteins, when expressed unfused, accumulate as
`insoluble inclusion bodies in E. coli. Although the fu-
`sions to MBP solved the problem of insolubility, they
`did not yield active enzymes. Therefore, we analyzed the
`protein folding states of several of the fusions.
`
`GNE 2007
`Page 3
`
`

`
`130
`
`Sachdev and Chirgwin
`
`Fig. 2. Proteinase K susceptibility of E. cofr-expressed MBP-aspartic
`proteinases. MBP-procathepsin D, MBP-pepsinogen, and placental
`cathepsin D were treated with 0.1 or 1% (w/w) proteinase K for 1 hr
`at room temperature. After proteinase K treatment, samples were
`boiled with Laemmli sample buffer and analyzed by SDS—PAGE fol-
`lowed by Coomassie blue staining. Lanes 1-4: placental cathepsin D;
`lanes 5-8: MBP-procathepsin D; lanes 9-12: MBP-pepsinogen. Lanes
`1, 5, and 9 show samples with no proteinase K treatment. Lanes 2, 6,
`and 10 were treated for 1 hr with 0.1% proteinase K. Lanes 3, 7, and
`11: 1 hr treatment with 1.0% proteinase K. Lanes 4, 8, and 12 are
`control reactions of the three proteins added to inactivated proteinase
`K.
`
`buffer. The unbound and bound fractions were concen-
`trated by acetone precipitation and analyzed by SDS-
`PAGE. All of the assays were also performed after
`preincubating the samples with 1 uM soluble pepstatin
`A before mixing with resin. This ensured that any bind-
`ing was specific. Neither MBP-procathepsin D nor
`MBP-cathepsin D bound significantly to pepstatinyl
`agarose at pH 3.5 (Fig. 1A). Lanes 1 and 6 are the start-
`ing materials used in the assay. MBP-procathepsin D
`appeared in the pepstatinyl agarose-unbound fraction
`(lane 3). No protein was in the bound fraction in lane 2.
`MBP-cathepsin D also did not bind to pepstatinyl aga-
`rose (lane 7). Mature cathepsin D isolated from human
`placentas was included as a positive binding control
`(Fig. 1C, lane 2). MBP-pepsinogen also failed to bind
`pepstatinyl agarose (Fig. 1B, lane 2). MBP-procathepsin
`D, MBP-cathepsin D, and MBP-pepsinogen all re-
`mained stable and soluble even at very high concentra-
`tions (not shown), indicating they were not grossly
`aggregated.
`The conformations of the various fusion proteins
`were probed using susceptibility to proteinase K diges-
`tion. MBP-procathepsin D and MBP-pepsinogen were
`treated with 0.1% and 1% proteinase K for 1 hr and the
`results analyzed by SDS-PAGE (Fig. 2). Purified pla-
`
`Fig. 1. Pepstatinyl agarose binding of MBP-aspartic proteinases. Fifty
`ug of each protein was mixed with pepstatinyl agarose for 1 hr at 4°C.
`The unbound supernatants and bound fractions were analyzed by
`SDS-PAGE on 10% gels and visualized by Coomassie blue staining.
`Assays were also done in the presence of soluble 1 uM pepstatin A
`added to the samples before incubation with pepstatinyl agarose as
`controls to ensure specific binding. (A) MBP-procathepsin D and
`MBP-cathepsin D. (B) MBP-pepsinogen. (C) Placental cathepsin D.
`Lanes 1 and 6 are starting materials used in the assay, lanes 2 and 7
`are bound fractions, lanes 3 and 8 are the unbound fractions, lanes 4
`and 9 are bound fractions in the presence of soluble pepstatin A added
`to the samples before incubation with pepstatinyl agarose, and lanes
`5 and 10 are the unbound fractions in the presence of added soluble
`pepstatin A. In panel C only the heavy chain of two-chain mature
`cathepsin D is visible on the gel.
`
`We first tested whether the aspartic proteinase moi-
`eties were native as judged by binding to immobilized
`pepstatin. Pepstatin is a transition state analog for as-
`partic proteinases (Marciniszyn et al., 1976) and binds
`to the properly folded active site clefts of these enzymes.
`Mature cathepsin D can bind to pepstatin at pH 5.3 or
`3.5, but procathepsin D binds only at pH 3.5 (Conner,
`1989). Therefore, 50 ug of purified fusion protein was
`mixed with pepstatinyl agarose at pH 3.5. The super-
`natant after centrifugation was saved as the unbound
`fraction. The bound fraction was eluted with high-pH
`
`GNE 2007
`Page 4
`
`

`
`Aspartic Proteinases
`
`131
`
`cental cathepsin D was the control (lanes 1-4) and was
`resistant to both 0.1% (lane 2) and 1% (lane 3) protein-
`ase K. Lanes 5-8 are MBP-procathepsin D and lanes 9-
`12 are MBP-pepsinogen. Lanes 1, 5, and 9 are samples
`of the various proteins with no proteinase K treatment.
`Control reactions of the three proteins added to inacti-
`vated proteinase K showed no proteolysis (compare
`lanes 4 vs. 1 for the 31—kDa heavy chain of placental
`cathepsin D, lanes 8 vs. 5 for MBP-procathepsin D, and
`lanes 12 vs. 9 for MBP-pepsinogen) indicating that the
`proteinase K inactivation at the end of the incubations
`was complete. MBP-pepsinogen was digested to a lower
`molecular weight species (the size expected for the MBP
`partner) by 1% proteinase K (lane 11). MBP-procathep-
`sin D was partially resistant to 0.1% proteinase K (lane
`6). However, it was digested to a lower molecular
`weight species by 1% proteinase K (lane 7), correspond-
`ing to the size expected for MBP.
`We next examined whether one of the fusion pro-
`teins displayed increased solvent exposure of hydropho-
`bic residues compared
`to
`its unfused component
`proteins. bis-ANS binds to hydrophobic surfaces, and its
`fluorescence spectrum can be used to probe the hydro-
`phobic surface exposure of a protein. In Fig. 3, 1 mg/ml
`each of MBP-cathepsin D, placental cathepsin D, or
`MBP was mixed with 10 uM bis-ANS and the fluores-
`cence spectra recorded. The binding of bis-ANS to
`MBP-cathepsin D (filled diamonds) was compared to its
`binding to MBP (open squares) and to placental cathep-
`sin D (filled circles). If the folded state of the fusion
`protein were a composite of the folded states of the two
`individual fusion partners, the fluorescence spectrum of
`bis-ANS binding to MBP-cathepsin D would be the av-
`erage of the spectra of placental cathepsin D and MBP;
`i.e., the spectrum for MBP-cathepsin D would be be-
`tween that of placental cathepsin D and MBP. However,
`the fusion protein MBP-cathepsin D, even at half the
`molar concentration of unfused MBP and placental ca-
`thepsin D, showed greater bis-ANS fluorescence, indi-
`cating substantially increased hydrophobic exposure.
`Since a sufficient quantity of native procathepsin D was
`unavailable, we analyzed only MBP-cathepsin D, and
`not MBP-procathepsin D, in order to make a direct com-
`parison between bis-ANS binding to a fusion protein and
`to its individual components.
`
`3.2. Hydrodynamic Properties of Purified MBP
`Fusions
`Sedimentation velocity analysis of the MBP-aspar-
`tic proteinases was undertaken to characterize physically
`
`the soluble fusion proteins. The sedimentation velocity
`runs are shown as the integral distribution of sedimen-
`tation values (S20,w) for the protein. A single pure-com-
`ponent system showing ideal behavior will appear as a
`vertical line intercepting the x axis at the appropriate s20, w
`value. Sedimentation velocity analysis of the soluble
`MBP-aspartic proteinases MBP-procathepsin D, MBP-
`cathepsin D, and MBP-pepsinogen revealed that all of
`them were large species with s20,w values greater than
`50S. van Holde and Weischet analysis also indicated that
`they were heterogeneous in size distribution. Figure 4A
`shows the analysis of MBP (filled squares), which is 42
`kDa and can exist as a monomer or a dimer, and pla-
`cental cathepsin D (open triangles) of ~45 kDa; both
`showed a homogeneous distribution of species. Figure
`4B shows the distribution of s20,w for MBP-procathepsin
`D (filled squares) and MBP-cathepsin D (open trian-
`gles), both of which show a heterogeneous population
`of species between 50S and 125S. Figure 4C shows the
`distribution for MBP-pepsinogen, which demonstrates
`s20,w values of 50-140S. The data indicate that these
`large, multimeric species were not in dynamic equilib-
`rium with smaller species, such as monomers.
`
`3.3. Properties of Fusion Proteins after Refolding
`In Vitro
`
`Since several aspartic pro-proteinases, such as pep-
`sinogen, are efficiently renatured from urea-denatured
`inclusion bodies, we wondered whether a cycle of de-
`naturation and refolding would alter the properties of the
`fusion proteins described above. We asked if the con-
`formation reached by refolding in vitro was the same as
`the folded state assumed by the MBP-aspartic protein-
`ases in the bacterial cytosol, or if denaturation followed
`by refolding in vitro would yield properly folded MBP-
`aspartic proteinases. The fusion proteins were denatured
`by incubation with 8 M urea for 4 hr at room tempera-
`ture and refolded in vitro by dialysis to remove the urea,
`following a protocol for procathepsin B (Kuhelj et al.,
`1995). All of the MBP-aspartic proteinases remained
`soluble. After refolding, the proteins were mixed with a
`50% v/v slurry of amylose resin in a microfuge tube for
`1 hr and the bound and unbound fractions analyzed by
`SDS-PAGE, as shown in Fig. 5. Solubly expressed
`MBP bound amylose resin (lane 2). Lane 3 shows the
`amount of refolded MBP used in the assay. MBP, after
`denaturation and refolding in vitro, bound amylose resin
`(lane 5), indicating that it has folded into its native con-
`formation. MBP-procathepsin D expressed in E. coli
`also bound amylose resin (lane 7). The amount of re-
`
`GNE 2007
`Page 5
`
`

`
`132
`
`Sachdev and Chirgwin
`
`Fig. 3. Fluorescence emission spectra of bis-ANS binding to MBP-cathepsin D compared to placental cathepsin
`D and MBP. One mg/ml each of MBP, placental cathepsin D, and MBP-cathepsin D was mixed with 10 uM
`bis-ANS. The protein solutions were mixed thoroughly after addition of bis-ANS. Fluorescence was measured
`using an SLM Model 500C fluorometer at an excitation wavelength of 397 nm, and the emission spectra were
`recorded over the range 400-600 nm.
`
`folded MBP-procathepsin D used in the assay is shown
`in lane 8. Refolded MBP-procathepsin D also bound to
`amylose resin (lane 10), indicating that the MBP partner
`was refolded to its native state. Refolded MBP-pepsin-
`ogen also bound amylose resin (data not shown).
`The sedimentation velocity analysis of refolded
`MBP-procathepsin D and MBP-pepsinogen is shown in
`Fig. 6. MBP-procathepsin D and MBP-pepsinogen were
`much smaller species after refolding. They had a distri-
`bution indicating a probable two-component (monomer-
`dimer) system. The s20,w value for refolded MBP-pro-
`cathepsin D ranged from 4S to 11S. In the case of MBP-
`pepsinogen, the 520, w value was 4-6S and the distribution
`was more homogeneous than for MBP-procathepsin D.
`Refolding produced an homogeneous distribution of
`much smaller species. Analysis of refolded MBP-ca-
`thepsin D (not shown) also indicated the presence of
`species with s20,w value ranging from 7S to 10S, consis-
`tent with a monomer-dimer distribution. These results
`indicate that the folding pathway leading to soluble
`MBP-aspartic proteinases in the bacterial cell may be
`different from that followed in vitro.
`
`3.4. Resistance to Aggregation of MBP Fusion
`Proteins
`
`The known aspartic proteinase structures are char-
`acterized by a high proportion of B-sheet. Since B-sheets
`show a propensity to form insoluble aggregates by in-
`termolecular polymerization of folding intermediates, we
`tested whether the fusion proteins showed any tendency
`toward insoluble aggregation at intermediate concentra-
`tions of denaturant. On refolding in vitro following a
`protocol for procathepsin B (Kuhelj et al., 1995), all of
`the MBP-aspartic proteinases remained soluble. Upon
`stepwise dilution of 8 M urea solutions of MBP-pro-
`cathepsin D from 8 to 1 M, the protein remained soluble
`at all concentrations tested (Fig. 7A), as did unfused
`MBP (Fig. 7B). (Data for other fusions not shown.)
`
`3.5. Sequence-Specific Restoration of Aspartic
`Proteinase Activity upon Refolding
`
`It is well established that pepsinogen can efficiently
`renature in vitro, whereas the recovery of successfully
`
`GNE 2007
`Page 6
`
`

`
`Aspartic Proteinases
`
`133
`
`Fig. 4. Sedimentation velocity analysis of the E. co/i-expressed MBP-
`aspartic proteinases. (A) MBP and placental cathepsin D, (B) MBP-
`procathepsin D and MBP—cathepsin D, (C) MBP—pepsinogen. Anal-
`ysis was performed in a Beckman XL-A analytical ultracentrifuge. The
`samples were prepared by dialysis against 0.1 M phosphate (pH 7.0),
`100 mM NaCl. The sedimentation velocity run was performed at 21°C
`and at 8000 rpm for MBP-aspartic proteinases and 40,000 rpm for
`native cathepsin D. Forty scans were collected and analyzed by the
`method of van Holde and Weischet. All data were corrected for buffer
`density and viscosity. The analysis of the sedimentation velocity runs
`is shown as the integral distribution of s20,w for the protein. The y axis
`indicates the fraction of sample with s20,W values less than or equal to
`the values on the x axis (s20,w values).
`
`Fig. 5. Amylose resin binding. Refolded fusion proteins were mixed
`with amylose resin in a microfuge tube, and the bound and unbound
`fractions were analyzed by SDS-PAGE with Coomassie blue staining.
`Lanes 1-5 are MBP and lanes 6-10 are MBP-procathepsin D. Lanes
`1 and 6: unbound fraction; lanes 2 and 7: bound fraction; lanes 3 and
`8: protein refolded in vitro, amount used in assay; lanes 4 and 9:
`protein refolded in vitro, unbound fraction; lanes 5 and 10: protein
`refolded in vitro, bound fraction.
`
`(Fig. 8), which was comparable to that of placental ca-
`thepsin D and was inhibited completely by 1 uM soluble
`pepstatin. When MBP-procathepsin D was denatured
`and refolded, it demonstrated very little enzymatic activ-
`ity, but the amount was significant compared to that of
`the protein prior to refolding, and it was inhibited by
`pepstatin. The specific activity of the refolded protein
`was only a few percent of the native enzyme control,
`consistent with that recovered during refolding in vitro
`of unfused procathepsin D (Conner and Richo, 1992;
`Scarborough and Dunn, 1994; Beyer and Dunn, 1996).
`Thus, fusion to MBP prevented the aggregation during
`
`refolded procathepsin D is only a few percent. There-
`fore, we tested whether fusion to MBP changed these
`results. When MBP-pepsinogen was denatured in 8 M
`urea and refolded, it demonstrated enzymatic activity
`
`GNE 2007
`Page 7
`
`

`
`134
`
`Sachdev and Chirgwin
`
`Fig. 6. Sedimentation velocity analysis of refolded MBP—procathepsin
`D and MBP-pepsinogen. Analyses were done as described for Fig. 4.
`The samples were prepared by dialysis of refolded MBP-procathepsin
`D and MBP-pepsinogen against 0.1 M phosphate (pH 7.0), 5 mM
`EDTA. The sedimentation velocity runs were performed at 21°C and
`at 20,000 rpm.
`
`refolding of procathepsin D, but did not alter the frac-
`tional folding to the native state.
`
`4. DISCUSSION
`
`Overexpression of proteins in the bacterial cytosol
`often leads to aggregation and formation of inclusion
`bodies. This is a general problem with many heterolo-
`gous proteins expressed in E. coli and with some E. coli
`proteins themselves. The aspartic proteinases, when ex-
`pressed in E. coli, accumulate in inclusion bodies.
`Therefore, we expressed them as fusions to the E. coli
`proteins MBP and thioredoxin. When MBP or thiore-
`doxin was present at the C-terminus of the aspartic pro-
`teinases, the fusions still accumulated in inclusion bodies
`even when the cells were grown at reduced temperature
`(Sachdev and Chirgwin, 1998a). However, when MBP
`was moved to the N-terminus, procathepsin D and pep-
`sinogen fusions were expressed solubly in the bacterial
`cytosol. These fusions offer the opportunity to test recent
`suggestions about the independent folding of multiple
`domains in proteins (Netzer and Haiti, 1997) and the
`inhibition of aggregation/inclusion body formation
`(Betts et al., 1997).
`The MBP-aspartic proteinase fusions as isolated
`from E. coli are soluble, and the MBP is folded cor-
`rectly, since it binds to immobilized amylose. However,
`the fusions did not bind to pepstatinyl agarose (Fig. 1),
`
`Fig. 7. Solubility of refolded MBP-procathepsin D. (A) MBP-proca-
`thepsin D, (B) MBP. MBP-procathepsin D and MBP were denatured
`in 8 M urea, followed by dilution of the urea to between 1 and 8 M.
`The samples were centrifuged, and the pellet and supernatant fractions
`for each urea concentration were analyzed by SDS-PAGE followed
`by Coomassie blue staining to determine resistance to aggregation. For
`each concentration of urea indicated at the top of the figure, p = pellet
`and s = supernatant.
`
`confirming their lack of enzymatic activity (Sachdev and
`Chirgwin, 1998b). Susceptibility of these fusions to di-
`gestion by proteinase K (Fig. 2) and to V8 protease (data
`not shown) also confirmed that the MBP partner of the
`fusion proteins was folded into a conformation resistant
`to proteolysis. However, the aspartic proteinase partners
`were not resistant to proteolysis, unlike mature placental
`cathepsin D control. Nevertheless, the nonnative confor-
`mations of the aspartic proteinase portions of the MBP
`fusions were stable and soluble even at very high con-
`centrations. These results suggest that they were not
`grossly unfolded, since they showed no propensity to
`aggregate.
`Although the aspartic proteinase portions of the
`MBP-aspartic proteinase fusions are nonnative, they are
`
`GNE 2007
`Page 8
`
`

`
`Aspartic Proteinases
`
`135
`
`a posttranslational event and thus does not provide an
`explanation for the difference in solubilities between
`MBP-aspartic proteinases compared to the insoluble as-
`partic proteinase-MBP fusions expressed in E. coli
`(Sachdev and Chirgwin, 1998a).
`Our results suggest that MBP, when at the N-ter-
`minus of fusions, may provide a general protective effect
`against insoluble aggregation. The robust folding of
`MBP, which readily occurs in vivo from separately ex-
`pressed fragments (Betton and Hofhung, 1994), may
`contribute to this protective function. Fusion to MBP
`completely blocked formation of insoluble aggregates at
`intermediate urea concentrations (Fig. 7)—conditions
`which trap aggregation-prone folding intermediates of
`other proteins (London et al., 1974; Jaenicke, 1987). The
`mechanism whereby fusing MBP with an aggregation-
`prone aspartic proteinase results in a highly soluble pro-
`tein chimera is unclear, but of substantial theoretical and
`practical interest.
`Our results show that refolding in vitro yielded
`states different from those isolated directly from E. coli.
`After refolding, the proteins remained soluble (not
`shown) and bound to amylose (Fig. 5), as expected from
`the known robustness of MBP folding (Liu et al., 1988,
`1989). MBP shows no tendency to aggregate either in
`vitro or in vivo during folding either in the cytoplasm or
`the periplasm (Betton et al., 1993; Betton and Hofnung,
`1994), unlike B-lactamase, which forms inclusion bodies
`in both compartments upon overexpression (Bowden et
`al., 1991). Thus, it appears that MBP does not fold along
`a pathway involving aggregation-prone intermediates.
`After refolding in vitro, the fusion proteins were no
`larger than dimers (Fig. 6). This indicates that the large
`oligomers isolated from E. coli are a nonequilibrium
`folding state. There was no evidence for stable associ-
`ation of bacterial chaperones, such as groEL, from gels
`of material purified under nondenaturing conditions. Our
`results do not support the model of Netzer and Hartl
`(1997), which proposes that bacterial protein folding oc-
`curs posttranslationally.
`Refolded pepsinogen was active, indicating that
`folding of the aspartic proteinase is not hindered by
`MBP, since unfused pepsinogen from E. coli gives sim-
`ilar refolding results (Lin et al., 1989, 1995). However,
`MBP-procathepsin D could not be renatured beyond a
`few percent.
`It is generally thought that protein folding is a ki-
`netic competition between on-pathway reactions, leading
`to the folded state, and off-pathway reactions, leading to
`aggregation (Jaenicke, 1987, 1991; Zettlmeissl et al.,
`1979; Zetina and Goldberg, 1980; Jaenicke and Seckler,
`1997). During refolding in vitro, the yield of any native
`
`Fig.