PROTEIN EXPRESSION AND PURIFICATION 6, 727–736 (1995)

Overexpression in Escherichia coli, Folding, Purification,and Characterization of the First Three Short ConsensusRepeat Modules of Human Complement Receptor Type 1

Ian Dodd,*,1 Danuta E. Mossakowska,* Patrick Camilleri,† Margaret Haran,† Preston Hensley,‡Elizabeth J. Lawlor,§ Diane L. McBay,* Wendy Pindar,* and Richard A. G. Smith*Departments of *Protein Chemistry, §Molecular and Cellular Biology, and †Analytical Sciences, SmithKline BeechamPharmaceuticals, Yew Tree Bottom Road, Epsom, Surrey, KT18 5XQ, United Kingdom; ‡Department of MacromolecularSciences, SmithKline Beecham Pharmaceuticals, 709 Swedeland Road, King of Prussia, Pennsylvania 19406-0939

Received April 18, 1995, and in revised form June 29, 1995

tration-dependent inhibition of complement-mediatedlysis of sensitized sheep red blood cells. q 1995 AcademicWe have developed a simple expression, isolation,Press, Inc.and folding protocol for an SCR oligomer comprising

the first three SCRs of complement receptor Type 1(C3b/C4b receptor, CD35). A T7 RNA polymerase ex-pression system in Escherichia coli was used to ex-press the oligomer as inclusion bodies. The oligomer The complement system in man is a complex arraywas recovered from solubilized inclusion bodies using of interacting proteins involved in a number of physio-batch adsorption on SP–Sepharose. The oligomer was logical events. Activation of the complement systemfolded by one-step dilution in 20 mM ethanolamine/1 leads to a variety of responses that include increasedmM EDTA supplemented with 1 mM GSH/0.5 mM GSSG. vascular permeability, chemotaxis of phagocytic cells,The folded material was processed to a concentrated activation of inflammatory cells, opsonization of foreign(ú20 mg/ml), usable product of greater than 98% purity particles, cell lysis, and, in pathological states, tissueusing a combination of ultrafiltration, ammonium sul- damage [for review, see ref. (1)]. Thus, regulators offate treatment, hydrophobic interaction, and size-ex- complement activation, including complement receptorclusion chromatography. The yield of folded material Type 1 [C3b/C4b receptor, CD35 (CR1)] (2),2 can playvaried between 6 and 15 mg/liter culture. The oxida- a key role in controlling many of the events associatedtion states of the 12 cysteine residues in SCR(1–3) were

with the immune responses.identified by HPLC of peptide fragments from a trypticCR1 is a Type 1 integral membrane protein founddigest using dual UV/fluorescence detection, collection

on a variety of cell types, notably erythrocytes (3) andof selected peaks, and N-terminal sequencing. Thisperipheral blood lymphocytes (4). It is a single chainmethodology confirmed the expected location of disul-glycoprotein (Mr160–250 kDa, depending on allotype)fide bridges. Equilibrium and velocity sedimentationand the extracellular portion of the most common allo-studies are interpreted in terms of a single sedi-type comprises 30 modules known as short consensusmenting species with molecular weights of 21,629 andrepeats (SCR) or complement control protein motifs.21,063 by these respective techniques. These valuesEach SCR contains 60 to 70 amino acids and in eachcompare to the predicted molecular weight, from

amino acid composition, of 21,817. The hydrodynamic SCR about 29 of the average 65 residues are conservedproperties of the molecule indicate that it is asymmet-ric with an axial ratio of 1:5.2 or equivalent dimensionsof 21 1 110 A. SCR(1–3) has an unusual CD spectrum

2 Abbreviations used: CR1, complement receptor Type 1; SCR,short consensus repeat; GSH, reduced glutathione; GSSG, oxidizedexhibiting a broad maximum at 220–230 nm and a min-glutathione; HPLC, high pressure liquid chromatography; NMR, nu-imum at 190 nm. There was little evidence of classical clear magnetic resonance spectroscopy; PMSF, phenylmethylsulfo-

secondary structure. The product exhibited concen- nylfluoride; IPTG, isopropylthio-b-D-galactoside; SDS–PAGE, so-dium dodecyl sulfate–polyacrylamide gel electrophoresis; TFA, tri-fluoroacetic acid; HIC, hydrophobic interaction chromatography;PTH, phenylthiohydantoin.1 To whom correspondence should be addressed.

7271046-5928/95 $12.00Copyright q 1995 by Academic Press, Inc.All rights of reproduction in any form reserved.

/ m3971$0527 10-30-95 09:01:05 pepa AP-PEP

DODD ET AL.728

(5). Each SCR contains four conserved cysteine resi- sitized sheep erythrocytes were obtained from Dia-medix Corporation, Miami, FL. Human serum from vol-dues and the first and third and the second and fourth

cysteines are covalently linked through disulfide unteers was prepared by standard techniques (18) toprovide a source of complement and was stored atbonds (6).

The SCR motif is widely distributed; it has been iden- 01967C.tified in a number of proteins of the complement system Construction of holding plasmid pDB1010-D11 en-and in noncomplement system proteins including the coding SCR(1–2). Oligonucleotides coding for SCR(1–2)IL2 receptor, b2 glycoprotein I, Factor XIII, thyroid were ligated in a unique fashion via complementaryperoxidase, and some cell adhesion proteins. Approxi- 8-bp overhangs between the pairs of oligonucleotides.mately 140 SCR modules have now been described. Oligonucleotide design was based on the known cDNA

The functional roles of SCR modules vary and de- sequence for CR1 except that codons were modified totailed structural models are essential to an under- allow for optimal expression in E. coli (see text) andstanding of both the variable ligand binding and (prob- are detailed elsewhere (12). The DNA was amplified byably less variable) architectural aspects of this module. PCR and a band of approximately 400 bp was isolatedA number of groups have constructed proteins compris- using agarose gel electrophoresis and a NA45 DEAEing one or more SCR motifs, including from CR1 (e.g., membrane. The DNA was then cut with NdeI and Hin-7, 8). The 1H-NMR structure of the 5th and the 16th dIII before ligating into pBROC413 that had been cutSCR modules from Factor H have been reported (9,10). with the same enzymes to yield pDB1011-D11. The vec-In order to understand SCR structure and function in tor was transformed into E. coli HB101 made compe-more detail further examples are needed. tent with calcium chloride. The cDNA sequence was

As part of an ongoing investigation into the biological confirmed by restriction mapping and by sequencing ofactivities of the CR1 molecule (11–13) we have ex- both strands across the gene coding for SCR(1–2).pressed a number of SCR constructs derived from CR1 Construction of holding plasmid pBROC435 encod-in an Escherichia coli inclusion body system with the ing SCR 3. Three pairs of oligonucleotides encom-potential to produce large amounts of protein both for passing the SCR3 coding sequence were synthesizedstructural studies and functional activity determina- and annealed as pairs and the middle pair was kinasedtions. A major challenge in producing multidisulfide thus allowing the three pairs to be ligated together viaproteins in this way is to overcome the problems associ- 8-bp overlapping sequences. The 5* end of this trimericated with isolation of correctly folded protein. molecule was designed to be complementary to NdeI-This paper describes the expression, folding, purifi- digested DNA and the 3* end to HindIII-digested DNA.cation to homogeneity, and physicochemical and initial This enabled the trimer to be cloned into NdeI/HindIII-biological characterization of the first three SCRs digested pBROC413 generating pBROC435. The iden-[SCR(1–3)] of CR1. tity of pBROC435 was checked by restriction enzymeThis is believed to be the first reported successful analysis and confirmed by DNA sequencing.expression of folded and substantially active short

Construction of plasmid pDB1013 encoding SCR(1–consensus repeat modules from CR1 in a prokaryotic3). The plasmid encoding SCR(1–3) was constructedsystem.from pBROC435 and pDB1010-D11 as follows. Bothwere cut with EcoRI and HindIII and the appropriate

MATERIALS AND METHODS bands were purified on NA45 DEAE membranes. TheSCR3 coding unit was ligated into pDB1010-D11 toMaterials. E. coli BL21(DE3) (14) was from Dr. A.generate pDB1013 which was then transformed intoShatzman, SmithKline Beecham. Plasmid pT7-7 wascalcium chloride-competent E. coli HB101. The re-a gift to SmithKline Beecham from Dr. Stan Tabor andsulting colonies were analyzed by miniplasmid prepa-has been described previously (15). pBROC413 hasration of DNA followed by restriction mapping. Thebeen described before (16). Oligonucleotides were syn-peptide sequence encoded by pDB1013 corresponds tothesized using a 381A synthesizer (Applied BioSys-residues Q1 to K196 of CR1 with an additional N-termi-tems) or a Gene Assembler Plus (Pharmacia LKB) ac-nal methionine.cording to the manufacturer’s protocols and are de-

scribed in detail elsewhere (12) and are based on the Expression of SCR(1–3). pDB1013 was trans-formed into calcium chloride-competent E. colipublished sequence of CR1 (17). [Note that ref. (12) is

a patent and is freely available.] NA45 DEAE mem- BL21(DE3) and resulting colonies were analyzed byrestriction digestion of miniplasmid DNA preparations.branes were from Anderman, Kingston-Upon-Thames,

UK. SP–Sepharose FF and Sephadex G25 were from Single colonies were inoculated into universals con-taining 10 ml of NCYZM medium (see below) and 50Pharmacia LKB, Milton Keynes, UK. Ethanolamine

was from BDH, Poole, UK. Toyopearl Butyl 650M was mg/ml ampicillin. Overnight cultures (377C, typically 5ml) were used to inoculate 2-liter conical flasks con-from Tosohass, Philadelphia, PA. Rabbit antibody sen-

/ m3971$0527 10-30-95 09:01:05 pepa AP-PEP

PROKARYOTE EXPRESSION OF SCR(1–3) OF HUMAN CR-1 729

taining 500 ml of NCYZM medium, 150 mg/ml ampicil- was then ultrafiltered using a YM10 membrane to afinal retentate volume of about 10 ml; this retentatelin; cultures were grown at 377C, 220 rpm to A600 0.5

AU. Cultures were induced with 1 mM isopropylthio- was usually cloudy. It was mixed with 9 vol 0.1 M

NaH2PO4/1 M (NH4)2SO4, pH 7.0 (buffer A), at roomb-D-galactoside (IPTG) and allowed to grow a further3 h under the same conditions. The cultures were cen- temperature and immediately centrifuged at 4000 rpm

for 20 min. The supernatant was further purified bytrifuged (Ç8000g/10 min) and the supernatants werediscarded. The cell pellets were stored at 0407C. chromatography on a column (Vt Ç 12 ml) of Toyopearl

Butyl 650M equilibrated with buffer A and developedNCYZM medium is 1% Bactotryptone, 0.5% Bactoyeast extract, 0.5% NaCl, 0.1% casamino acids, 0.2% with a linear gradient of 100% buffer A to 100% 0.1 m

NaH2PO4, pH 7.0. All the chromatography was at roomMgSO4r7H2O, pH 7.0.temperature at approximately 30 cm h01. A major A280Isolation of solubilized inclusion bodies. Frozen cellpeak was eluted during the gradient. Fractions span-pellet from 4 liters of culture was allowed to thaw atning the peak were analyzed by SDS–PAGE followed47C for 2 h and resuspended in 50 mM Tris/50 mMby protein staining. Selected fractions (represented byNaCl/1 mM EDTA/0.1 mM PMSF, pH 8.0 (130 ml). Thelane 5 in Fig. 3) were pooled and buffer-exchanged intosuspension was transferred to a glass beaker and soni-50 mM formic acid using Sephadex G25 at 50 cm h01

cated (Heat Systems–Ultrasonics W380; 70 W, 50 1at 47C. The eluate from the column was monitored at50% pulse, pulse time 5 s). The sonicate was immedi-280 nm and the V0 fraction was collected and lyophi-ately centrifuged (6000g/47C/10 min) and the superna-lized in aliquots.tant was discarded. The pellet containing the inclusion

Disulfide bridge position determination. Lyophi-bodies was resuspended in 20 mM Tris/8 M urea/50 mMlized SCR(1–3) (1.7 mg) was reconstituted in 3.6 ml2-mercaptoethanol/1 mM EDTA/0.1 mM PMSF, pH 8.50.1 M NH4HCO3, pH 8.3. TPCK-treated trypsin (0.1 mg/(400 ml) and left static at room temperature (Ç237C)ml; 0.8 ml) was added and the mixture was incubatedfor 1–1.5 h to facilitate solubilization. The resultingat 377C for 16 h. An aliquot of 0.1 ml was then removedsolution was grayish-white in color and viscous.and mixed with 0.005 ml 2-mercaptoethanol and the

Preliminary purification using SP–Sepharose. To incubation was continued for a further 15 min. The twothe viscous solution was added SP–Sepharose FF (Ç40 products provided the nonreduced and reduced prepa-ml gel bed equivalent) that had been water washed and rations, respectively. Peptides were resolved on a C18suction dried. The mixture was swirled vigorously and

m-Bondapak column (150 1 3.9 mm) at ambient tem-left static for 1–2 h at room temperature. The superna- perature using a 0.1% TFA/0 to 63% acetonitrile gradi-tant was decanted, sampled, and discarded. The re- ent. Eluate was monitored by UV (at 220 nm) or bymaining slurry was resuspended to a uniform suspen- tryptophan fluorescence (lex 274 nm, lem 345 nm). Peaksion and poured into a glass jacket (id, 41.5 mm) and samples were evaporated to dryness and sequenced us-allowed to settle into a packed bed. The column was ing the pulsed liquid sequencer (Applied Biosystems,equilibrated with 0.02 M Tris/8 M urea/0.05 M 2-mer- Model 477A) using TFA and polybrene-coated What-captoethanol/0.001 M EDTA, pH 8.5, at 47C. when the man GF/C disks as support matrix. Residues were iden-A280 of the eluate had stabilized at baseline, the buffer tified using an online PTH-AA analyzer (Applied Bio-was changed to equilibration buffer additionally con- systems, Model 120A) essentially according to the man-taining 1 M NaCl. A single A280 peak was eluted by the ufacturer’s instructions.1 M NaCl-containing buffer, in a volume equivalent to Analytical ultracentrifugation. Equilibrium and ve-approximately 1 Vt. The protein concentration of the locity sedimentation experiments were performed withsolution was estimated by A280 determination of a sam- a Beckman XL-A analytical ultracentrifuge (Beckmanple that had been buffer-exchanged (Sephadex G25) Instruments, Spinco Division, Palo Alto, CA). Doubleinto 50 mM formic acid, using a molar extinction coeffi- sector cells with charcoal-filled epon centerpieces andcient of 25,000 cm01 (determined by amino acid analy- sapphire windows were used. Sedimentation experi-sis). Products were stored at 0407C. ments were performed at 257C in a buffer of 50 mM

Folding and further purification. The SP–Sepha- sodium phosphate, pH 7.4. The initial protein concen-rose-purified product was diluted with 0.02 M Tris/8 M trations were A280 Å 0.33 or 1.0 1 1005 M [0.2 mg/mlurea/1 M NaCl/0.05 M 2-mercaptoethanol, pH 8.5, to 2 assuming e Å 33,720 M01 cm01 or 1.55 (mg/ml)01 cm01,mg/ml. Thirty milliliters was added gradually over a both determined from amino acid composition] for equi-1-min period to 930 ml freshly prepared, cold 0.02 M librium experiments and A280 Å 0.80 or 2.4 1 1005 Methanolamine/1 mM EDTA with continuous swirling, (0.52 mg/ml) for velocity experiments. Note that theand left static for 1 h at 47C. Reduced glutathione molar extinction coefficient is different from that deter-(GSH) was added to 1 mM and oxidized glutathione mined for the SP–Sepharose product described earlier.(GSSG) was added to 0.5 mM. The solution was clear It is probable that the differently folded states of the

two products account for the difference.and was left static approx 2–37C for 3 d. The solution

/ m3971$0527 10-30-95 09:01:05 pepa AP-PEP

DODD ET AL.730

Equilibrium sedimentation data were analyzed us-ing nonlinear least-squares methods (19–21) under thecontrol of a modified version of IGOR (WaveMetrics,Lake Oswego, OR). Data sets were collected afterreaching equilibrium, 18–24 h, at a rotor speed of30,000 rpm. Equilibrium was established by determin-ing that scans taken 4 h apart were superimposable.

Sedimentation velocity data were analyzed using theprogram VELGAMMA (Beckman Instruments). Eigh-teen data sets were collected, 200 s apart, starting fromthe time the rotor reached 60,000 rpm.

CD spectroscopy. The CD spectrum was taken us-ing a JASCO 710 spectropolarimeter, using a 0.1-mm

FIG. 1. SDS–PAGE of IPTG induction of SCR(1–3). E. coli whole-pathlength cell. SCR(1–3) was was 2.4 1 1005 M in 50cell pellets were fully solubilized under reducing conditions and ana-mM sodium phosphate, pH 7.4. lyzed by SDS–PAGE followed by protein staining. Lanes 1, 5, and

SRBC hemolytic assay. Diluted human serum (50 9, pDB1013 180, 90, and 0 min after induction. Lanes 2,6, and 10,pBROC413 180, 90, and 0 min after induction. Lanes 3,7, and 11,ml) was preincubated in the presence of incrementalno plasmid 180, 90, and 0 min after induction. Lanes 4 and 8 areconcentrations of SCR(1–3) (50 ml) for 15 min, 377C,molecular weight markers. The arrow marks the position ofbefore adding 100 ml SRBC. Incubation was for a fur- SCR(1–3).

ther 60 min, 377C, after which samples were clarifiedby centrifugation and lysis was assessed spectrophoto-metrically at 410 nm. Inhibition was expressed as a

sorption using sulfopropyl (SP)–Sepharose was identi-fraction of the total cell lysis such that IH50% representsfied as a rapid, facile method (Fig. 2, main panel) ofthe concentration of inhibitor required to give 50% inhi-producing 70 to 80% pure target (Fig. 2, lane 6). Ap-bition of lysis.proximately 20 to 30% of the target protein was presentSodium dodecyl sulfate–polyacrylamide gel electro- in the unadsorbed fraction (Fig. 2, lane 4), probablyphoresis. SDS–PAGE used the Novex (R & D Sys- due to insufficient SP–Sepharose matrix being usedtems, Oxford, UK) gel systems (typically the 4–20% for the adsorption.acrylamide gels) and commercial protein molecular The unfolded protein could be maintained in solutionweight standards (LMW kit, Pharmacia). Gels were under acidic conditions. Buffer exchange of a samplestained with Page Blue 83 (BDH). of the SP–Sepharose product into 50 mM formic acidallowed the determination of protein content by amino

RESULTS acid analysis. With reference to the A280 of the samesolution, the molar extinction coefficient was calculatedThe strategy adopted for the prokaryotic expression

of the SCR(1–3) oligomer was to utilize a T7 RNA poly- to be Ç25,000. The SP–Sepharose product was stablefor several months at 0407C. Based on the amino acidmerase promoter system in E. coli. A synthetic cDNA

encoding SCR(1–3) was constructed in plasmid analysis, yields at the SP–Sepharose stage were ap-proximately 30 to 60 mg/liter of culture. MicrotiterpBROC413 under the control of the F10 promoter.

IPTG induction of E. coli BL21(DE3):pDB1013 was plate-based sequential dilutions and simultaneous tur-bidity measurements were used to identify a bufferused to induce expression of SCR(1–3) (Fig. 1). Elec-

tron microscopy studies of the cultured recombinant E. that maintained the oligomer in solution, under condi-tions that should allow folding of the protein to occur.coli cells, 3 h after IPTG induction, clearly showed the

presence of inclusion bodies (M. Newman, personal Ethanolamine at 20 mM, pH 10.0, was shown to be asuitable buffer. Dilution into another alkaline buffer,communication). Subsequent isolation studies showed

that the majority of the SCR(1–3) protein could be pre- 50 mM borate, pH 10.0, led to immediate precipitation.Further investigation identified a folding protocolcipitated from sonicated E. coli cells using a relatively

low (6000g/10 min) centrifugal force (Fig. 2, lanes 2,3), based on one-step dilution into 20 mM ethanolamine,using the redox couple oxidized and reduced glutathi-supporting the conclusion that the protein was present

as inclusion bodies. one. GSH and GSSG respective concentrations of 1 and0.5 mM were shown to be optimal. Folding was carriedThe 6000g precipitate was solubilized in an 8 M urea/

50 mM 2-mercaptoethanol-containing buffer over ap- out using final concentrations of protein, 62 mg/ml; 2-mercaptoethanol, 1.6 mM; urea, 0.25 M. Elevated (am-proximately 1 h at room temperature to yield a viscous

solution. Since the presence of nontarget proteins bient) temperature promoted aggregation so foldingwas carried out at approximately 2 to 37C.might interfere with correct folding of the target, par-

tial purification was considered desirable. Batch ad- In order to formulate the dilute, folded protein solu-

/ m3971$0527 10-30-95 09:01:05 pepa AP-PEP

PROKARYOTE EXPRESSION OF SCR(1–3) OF HUMAN CR-1 731

FIG. 2. Preliminary isolation of SCR(1–3) and analysis by SDS–PAGE. Recombinant cell pellet was sonicated, the sonicate pellet wassolubilized, and SCR(1–3) was purified using batch adsorption on SP–Sepharose. After 1 h static at room temperature the slurry (matrix)was poured into a glass jacket and developed as a normal chromatography. The main figure shows the A280 trace of the final rinsing of thecolumn and the peak eluted by the 1 M NaCl-containing buffer (marked by the arrow). The inset shows analysis of the key fractions bySDS–PAGE followed by staining for protein using Page Blue 83. The lanes are 1, marker proteins; 2, sonicate supernatant; 3, solubilizedsonicate pellet; 4, SP–Sepharose unadsorbed fraction; 5, SP–Sepharose fraction 1 (neat); 6, SP–Sepharose fraction 2 (1/5, 1/15, and 1/45dilutions). All samples were treated with 5% (v/v) 2-mercaptoethanol.

tion into a product suitable for further evaluation the by SDS–PAGE was Ç25,000 under reducing condi-tions andÇ20,000 under nonreducing conditions. Anal-solution was ultrafiltered and the retentate was mixed

with 9 volumes of 1 M ammonium sulfate-containing ysis by electrospray mass spectrophotometry indicateda molecular mass of 21,816 (A. West, personal commu-buffer. This treatment invariably led to significant pre-

cipitation. The suspension was clarified by centrifuga- nication). This compares well with the theoretical(chemical average) mass of 21,817.tion and the supernatant was applied to a Toyopearl

Butyl HIC column which was then developed with a An additional purity check was carried out using C8reverse-phase HPLC; this showed the product wasdecreasing ammonium sulfate gradient. Two A280 peaks

were identified, one in the unadsorbed fraction and a §98% pure (data not shown).From the expected disulfide-bridged tryptic frag-major one at approximately 0.2 M ammonium sulfate.

Analysis by SDS–PAGE/protein staining suggested ments of SCR(1–3), three of the four fragments withone or more disulfide bonds contain a tryptophan resi-that the unadsorbed material was nonproteinaceous,

the second peak was target protein, and the tail con- due. The characteristic fluorescence of tryptophan facil-itated the selection of peptides that contained one ortained impurities. Typical results are shown in Fig.

3. The major SCR(1–3) fractions were pooled, buffer- more cysteine residues. Only in the case of the seconddisulfide bridge was resort made to UV detection asexchanged into the volatile buffer 50 mM formic acid

using Sephadex G25, and lyophilized in aliquots. An the corresponding peptide fragment does not contain atryptophan residue. It was not possible to sequence theoverall summary of the results is given in Table 1.

Aliquots of the lyophilizate were resolubilized in ei- longer peptides completely as peak heights of the PTH-amino acids were reduced to near background levelsther 50 mM formic acid or a neutral buffer for various

assays. The protein was soluble to at least 20 mg/ml after about cycle 15 (Table 2). In the case of the intacttryptic digest, the simultaneous sequence of more thanin these buffers. Protein estimation based on amino

acid analysis indicated yields of 6 to 15 mg SCR(1–3) one N-terminal was a good indication of the presenceof two or more peptides covalently linked by disulfideper liter of culture. The molecular weight of the protein

/ m3971$0527 10-30-95 09:01:05 pepa AP-PEP

DODD ET AL.732

FIG. 3. Ammonium sulfate treatment and HIC isolation of SCR(1–3) and analysis by SDS–PAGE. Folded SCR(1–3) that had beenultrafiltered and mixed with 1 M ammonium sulfate-containing buffer was spun (4000g/20 min), the supernatant was chromatographed onToyopearl Butyl (main figure) and developed with a decreasing ammonium sulfate gradient (hatched line). The inset shows analysis of thekey fractions by SDS–PAGE. The resolubilized ammonium sulfate precipitate is shown in lane 10. The supernatant (lane 2) was chromato-graphed. The unadsorbed fraction is shown in lane 3. The remaining tracks (lanes 4–9) span the A280 peak eluted by the decreasingammonium sulfate gradient. Note that lane 5 is a pool of the three major A280 fractions and is diluted 1/5. Lanes 4, 6–9 are single fractionsand are undiluted. All samples were prepared in the absence of reducing agent.

bridges. The presence of two peptides in peptide group equation, the molecular weight may be determined fromthe hydrodynamic data. This yields a value of 21,0637 suggests that disulfide number 5 is resistant toward

reduction. Da, which is very close to the value determined fromequilibrium sedimentation and that predicted fromResults of replicate equilibrium ultracentrifugation

experiments are shown in Fig. 4. The value of the molec-ular weight determined from simultaneous analysis ofthe three data sets was 21,629 //0 256 (1.2%). Data TABLE 2were analyzed in terms of various models for self-associ- N-Terminal Sequence Data for Resolved Peptidesation (19,22), however, there were no increases in the of Tryptic Digestquality of fit, suggesting that these models were notrequired and that the molecule was a monomer. Using Peptide Sequence Disulfide

group Amino acid sequence position bridgevelocity sedimentation, the sedimentation coefficient,s20,wÅ 2.05 S, and the translational diffusion coefficient,

1 NSVWTGAK 48–55 —D20,w Å 8.59 1 1007 cm2s01, were determined (results2 (1) LIGSSSATCIISGDTVIWDN 96–115 C63–C104

not shown). Using these two values and the Svedberg (2) YSCTK 88–92 and(3) SCR 62–64 C90–C120

3 (1) IPCGLPPTITNG 123–134(2) VFELVGEPSI 163–172 C125–C174TABLE 1(3) CNPGSGG 154–160

Purification of SCR(1–3) (4) WSGPAPQCIIPN 184–195 C154–C1914 (1) MQCNAPEW 0–7

(2) PFSIICL 40–46 C2–C45TotalSCR(1–3) protein Estimated 5 (1) PTNLTDEFEFPIGTY 13–27

(2) CR 58–59 C32–C58Stage Volume (mg)a (mg) purityb

6 NSVWTGAK 48–55 —7 (1) IPCGLPPTIT 123–132 C125–C174E. coli Culture 4 liters NDc ND —

SP–Sepharose (2)CTSNDDQVGIWSGPAPQCII 174–1938 MQCNAPEWLPF 0–10 C2–C45Product 40 ml 120–140 ND 70–80%

Toyopearl 9 PTNLTDEFEPIGTY 13–27 C32–C58Butyleluate 15 ml 24–60 24–60 ú98%

G25 eluate 20 ml 24–60 24–60 ú98% Note. SCR(1–3) digested with trypsin was resolved using reverse-phase HPLC and monitored using UV or fluorscence. Peptide groups1 to 5 were resolved from nonreduced protein. Peptide groups 6 to 9a Ranges given correspond to values obtained for multiple batches

of 4 liters each. were from reduced protein. The numbering system used the conven-tional CR1 system, and the N-terminal methionine residue is calledb Based on SDS–PAGE and/or HPLC determinations.

c Not determined. residue 0.

/ m3971$0527 10-30-95 09:01:05 pepa AP-PEP

PROKARYOTE EXPRESSION OF SCR(1–3) OF HUMAN CR-1 733

FIG. 4. Equilibrium analytical ultracentrifugation of SCR(1–3). Three replicate experiments were analyzed, yielding a single sedimentingspecies of 21,629 { 256 (1.2%) Da. The rotor speed was 30,000 rpm and the initial protein concentration was 1 1 1005 M (A280 Å 0.33). Thesolvent was 50 mM sodium phosphate, pH 7.4. The three data sets were offset 0.10 A280 arbitrarily for visualization purposes. The threepanels above the data set are the residuals for the fits to the three data sets in reverse order, i.e., the upper set of residual is for the fit tothe lowest data set.

amino acid composition, 21,817 Da, suggesting that the marked minimum at Ç190 nm and a clear positiveellipticity at 220–230 nm (Fig. 5). Overall the spectrummolecule is indeed a monomer. If a prolate ellipsoid is

assumed for the shape of the molecule, an axial ratio of suggests little classical secondary structure.SCR(1–3) inhibited complement-mediated lysis of1:5.2 and effective dimensions of 21 1 110 A may be

determined. These latter two terms must be considered sheep red blood cells in a concentration-dependentmanner with an IH50% of approximately 15 nMas upper limits as the model assumes a smooth ellipsoid

of revolution and no consideration has been given to the (Fig. 6).intrinsic roughness of the protein surface or boundaries

DISCUSSIONbetween subdomains (23).Circular dichroism studies were carried out to inves- A T7 RNA polymerase promoter expression system

in E. coli was chosen to express an SCR oligomer com-tigate secondary structure. The spectrum shows a

/ m3971$0527 10-30-95 09:01:05 pepa AP-PEP

DODD ET AL.734

(the oligomer potentially has six) proteins isolated frominclusion bodies is problematic.

First, in order to minimize intermolecular disulfidebridge formation, the oligomer was partially purifiedby batch adsorption using the ion-exchange matrixSP–Sepharose. This process yielded 60 to 70% pureprotein in a concentrated, fully denatured state, readyfor folding.

Second, we wished to develop a method for foldingthat was simple and capable of scale-up and that wouldminimize incorrect intramolecular disulfide bridges.This indicated a dilution method rather than the moretraditional dialysis system and as high a protein con-centration as could be tolerated. Although a wide vari-ety of diluents for folding proteins have been described,it is generally acknowledged that identification of themost suitable one for any particular protein is empiri-cal. We used sequential dilutions in a variety of buffers

FIG. 5. CD spectrum of SCR(1–3). The spectrum was obtained at with turbidity measurements to rapidly screen a large257C in 50 mM phosphate, pH 7.4. The protein concentration was 2.3 number of diluents for their ability to maintain solubil-1 1005 M. ity of the protein under conditions where denaturant

concentrations were believed to be low enough to allowfolding. Ethanolamine at 20 mM (pH 10.0) was identi-fied as the best buffer of those screened. This buffer hasprising residues Q1 to K196 of human CR1. It was

constructed using synthetic oligonucleotides designed been used previously for the folding of porcine growthhormone (25). Additional studies suggested that addi-on the basis of the known cDNA sequence for CR1 (17).

Certain modifications were made to the cDNA se- tion of a glutathione redox couple 1 h after the initialdilution was beneficial to recovery.quence. Unique restriction sites were incorporated into

the DNA sequence (without altering the coding se- Third, despite the buffer screening procedure, sometarget protein appeared to misfold and aggregate. Thequence) to facilitate subsequent manipulation of the

cDNA. Additionally, the nonrandom use of synonymous precipitate could be removed easily using ammoniumsulfate treatment prior to final purification of SCR(1–codons has been demonstrated in E. coli and there is

some evidence to support the belief that protein produc- 3) to homogeneity using Toyopearl Butyl.tion from genes containing nonoptimal or minor codons(particularly at the 5* end of the gene) is less efficientthan that from genes with no such codons (e.g., 24). Allof the first 30 codons of all constructs (where compati-ble with restriction enzyme sites) were optimized forhigh level expression. The codons for the seven aminoacids Arg, Gly, Ile, Leu, Pro, Ser, Ala were optimized(where compatible with restriction enzyme sites)throughout the coding sequence. Finally, to enable ex-pression of protein, an ATG codon was added to the 5*end of the gene immediately preceding the codon forthe first amino acid of mature CR1. The codon ATG ispart of the NdeI restriction site which can be used forcloning into vectors such as pBROC413.

The cDNA encoding the SCR oligomer was insertedinto the modified pT7-7 vector pBROC413 to yieldpDB1013 and expressed in E. coli as inclusion bodies.The oligomer was recovered from the whole-cell pelletusing a standard inclusion body isolation method and

FIG. 6. Anti-hemolytic activity of purified SCR(1–3) The effect ofsolubilized in a fully denaturing buffer.various concentrations of purified SCR(1–3) on the inhibition of com-The strategy for obtaining correctly folded target pro- plement-mediated lysis of sensitized sheep red blood cells was exam-

tein from the denatured solution was multifaceted as ined; full details are given under Materials and Methods. IH values of0 and 1 represent no inhibition and complete inhibition, respectively.it is generally accepted that folding of multidisulfide

/ m3971$0527 10-30-95 09:01:05 pepa AP-PEP

PROKARYOTE EXPRESSION OF SCR(1–3) OF HUMAN CR-1 735

Fourth, the protein needed to be formulated into a The anti-hemolytic potency (classical pathway IH50%)of SCR(1–3) from several preparations was reproduci-product suitable for evaluation. Structural studies, in

particular, need concentrated product. Sephadex G25 bly determined asÇ15 nM, confirming reported activityin this region of CR1 (e.g., ref. 7). As all the disulfideusing the volatile buffer formic acid and subsequent

lyophilization were used to obtain the final product. bridges were believed to be correctly formed and it ap-pears that SCR(1–3) is correctly folded, we concludeCharacterization showed that the material was es-

sentially pure as judged by two techniques and was that this potency is a valid basis for comparison withother constructs in this assay. The IH50% determinedeasily soluble at concentrations suitable for structural

studies, that all six proposed disulfides were present, for sCR1 in the same assay was 40 pM (results notshown). A proper comparison of CR1 constructs re-although it was not possible to quantify the extent of

disulfide bridging, and that the product was function- quires the use of a variety of activation stimuli includ-ing those using undiluted serum but our data are con-ally active in vitro. These results suggest that the use

of nonphysiological pH (pH 10 during the refolding, pH sistent with proposals that additional SCR modules arerequired for full activity (7) or that the multiple ligand2.5 during the final buffer exchange) can be tolerated

during the isolation process, at least for the SCR con- binding sites of sCR1 (5) may enhance activity.The methods described here have been used subse-struct described here.

The final product yield of 6 to 15 mg liter01 of culture quently and successfully for a number of other CR1SCR constructs and their comparative biological evalu-obtained under unoptimized fermentation conditions,

compares favorably to the yields of 0.1 to 0.2 mg liter01 ation will be reported. These studies, together with theongoing structural analysis of the three-domain mod-described for a single SCR from Factor H produced in

a soluble form from a yeast expression system (9). ule, should yield interesting insights not only into thestructure of SCR modules but also into structure–ac-A prolate ellipsoid, as opposed to an oblate, is as-

sumed for the shape of SCR(1–3) as electron micros- tivity relationships within the CR1 molecule itself, sol-uble forms of which are attracting increasing researchcopy studies on sCR1 (26) show the individual SCR

domains to be connected in extended linear fashion. interest as potential therapeutic inhibitors of comple-ment activation with applications in a number of dis-The dimensions for the ellipsoid of 21 1 110 A are

reasonable, as the NMR structure of the 16th repeat ease states (11,26,33,34).of Factor H has approximate dimension of 15 1 40 A(9). This is thought to be a reasonable model of one ACKNOWLEDGMENTSsubdomain of SCR(1–3). Three of these Factor H do-

We thank W. Neville for valuable support with the reverse-phasemains placed end to end would have dimensions ofÇ15HPLC studies, A. Fosberry for synthesis and purification of the oligo-1 120 A. nucleotides, C. Slater for supporting CD work, J. M. Dewdney and

The preliminary investigation into secondary struc- A. H. Nishikawa for encouragement, and A. Herrington for help inpreparing the manuscript.ture of SCR(1–3) using CD indicated little classical

secondary structure. Similar studies on Factor Hshowed a remarkably similar CD spectrum (27). In con- REFERENCEStrast, NMR analyses of the Factor H SCR domains indi-

1. Muller-Eberhard, H. J. (1988) Molecular organisation and func-cated significant amounts of b-sheet (9,10). The appar- tion of the complement system. Annu. Rev. Biochem. 57, 321–ently contradictory results are most likely explained 347.by the fact that the b-strands in (the Factor H) SCRs 2. Ross, G. D., and Medof, M. E. (1985) Membrane receptors specificare very short, the longest only five residues long, and for bound fragments of C3. Adv. Immunol. 37, 217–267.this may obscure the typical b-sheet spectrum, which 3. Ahearn, J. M., and Fearon, D. T. (1989) The structure and func-

tion of the complement receptors CR1 (CD35) and CR2 (CD21).shows a minimum at approximately 218 nm, althoughAdv. Immunol. 46, 183–219.this may be shifted to the 205- to 210-nm region if short

4. Fearon, D. T. (1980) Identification of the membrane glycoproteinlengths of b-sheet separated by loops of random coilthat is the C3b receptor of the human erythrocyte, polymorpho-are present (28) or to approximately 200 nm in proteins nuclear leukocyte, B lymphocyte, and monocyte. J. Exp. Med.

rich in proline. For example, proline-rich glycoprotein 152, 20–30.(29,30) and histidine-rich glycoprotein (31) both have 5. Klickstein, L. B., Wong, W. W., Smith, J. A., Weis, J. H., Wilson,a minimum at 200 nm. As SCR(1–3) has 18 proline J. G., and Fearon, D. T. (1987) Human C3b/C4b receptor (CR1):

Demonstration of long homologous repeating domains that areresidues we cannot rule out the possibility that prolinecomposed of the short consensus repeats characteristic of theellipticity contributes to the observed effect. AnotherC3/C4 binding proteins. J. Exp. Med. 165, 1095–1112.unusual feature of the spectrum is the positive ellip-

6. Day, A. J., Ripoche, J., Willis, A. C., and Sim, R. B. (1987) Struc-ticity at 220–230 nm. Although such a maximum is ture and polymorphism of human Factor H. Complement 4, 147–typical of tryptophan (32), SCR(1–3) contains only four 148.tryptophan residues. Definition of secondary structure 7. Makrides, S. C., Scesney, S. M., Ford, P. J., Evans, K. S., Carson,

G. R., and Marsh, H. C. (1992) Cell surface expression of theawaits the outcome of ongoing NMR structural studies.

/ m3971$0527 10-30-95 09:01:05 pepa AP-PEP

DODD ET AL.736

C3b/C4b receptor (CR1) protects Chinese hamster ovary cells Vogel, C. W. (1986) The effects of metal ions and temperatureon the interaction of cobra venom factor and human complementfrom lysis by human complement. J. Biol. Chem. 267, 24754–

24761. factor B. J. Biol. Chem. 261, 11038–11044.22. Brooks, I., Wetzel, R., Chan, W., Lee, G., Watts, D. G., Soneson,8. Krych, M., Hourcade, D., and Atkinson, J. P. (1991) Sites within

K. K., and Hensley, P. (1994) Association of REI light chain VLthe complement C3b/C4b receptor important for the specificitydomains: The functional linearity of parameters in equilibriumof ligand binding. Proc. Natl. Acad. Sci. USA 88, 4353–4357.analytical ultracentrifuge data for self associating systems, in9. Barlow, P. N., Baron, M., Norman, D. G., Day, A. J., Willis,‘‘Modern Analytical Ultracentrifugation: Acquisition and Inter-A. C., Sim, R. B., and Campbell, I. D. (1991) Secondary structurepretation of Data for Biological and Synthetic Polymer Systems’’of a complement control protein module by two-dimensional 1H(Schuster, T. M., and Laue, T. M., Eds.), pp. 15–36, Birkhauser,NMR. Biochemistry 30, 997–1004.Boston.

10. Barlow, P. N., Norman, D. G., Steinkasserer, A., Horne, T. J., 23. Teller, D. C., Swanson, E., and de Haen, C. (1979) The transla-Pierce, J., Driscoll, P. C., Sim, R. B., and Campbell, I. D. (1992) tional friction coefficient of proteins. Methods Enzymol. 61, 103–Solution structure of the fifth repeat of Factor H: A second exam- 124.ple of the complement control protein module. Biochemistry 31,

24. Chen, G-F, and Inouye, M. (1990) Suppression of the negative3626–3634.effect of minor arginine codons on gene expression; preferential

11. Dupe, R. J., Goddard, M. E., Freeman, A. M., Hibbs, M. J., Lifter, usage of minor codons within the first 25 codons of the Esche-J., Mossakowska, D., Wilkinson, M. J., and Smith, R. A. G. (1991) richia coli genes. Nucleic Acids Res. 18, 1465–1473.Utility of complement inhibition during myocardial reperfusion: 25. Puri, N. K. (1991) Refolding of recombinant porcine growth hor-Pharmacology of soluble complement inhibition during myocar- mone in a reducing environment limits in vitro aggregate forma-dial reperfusion: Pharmacology of soluble complement receptor tion. FEBS Lett. 292, 187–190.1. Thromb. Haemostasis 65, 695.

26. Weisman, H. F., Bartow, T., Leppo, M. K., Marsh, H. C., Carson,12. Smith, R. A. G., Dodd, I., Freeman, A., and Mossakowska, D. E. G. R., Concino, M. F., Boyle, M. P., Roux, K. H., Weisfeldt,

(1993) UK Patent Application No. WO 94/00571. M. L., and Fearon, D. T. (1990) Soluble human complement re-13. Gibb, A. L., Freeman, A. M., Smith, R. A. G., Edmonds, S., and ceptor Type 1: In vivo inhibitor of complement suppressing post-

Sim, E. (1993) The interaction of soluble human complement ischemic myocardial inflammation and necrosis. Science 249,receptor type 1 (sCR1, BRL55730) with human complement com- 146–151.ponent C4. Biochim. Biophys. Acta 1180, 313–320. 27. Discipio, R. G., and Hugli, T. E. (1982) Circular dichroism studies

of human Factor H. A regulatory component of the complement14. Studier, F. W., and Moffat, B. A. (1986) Use of bacteriophage T7system. Biochim. Biophys. Acta 709, 58–64.RNA polymerase to direct selective high-level expression of

cloned genes. Use of bacteriophage T7 RNA polymerase to direct 28. Garcia, K. G., de Sauvage, F. J., Strubel, M., Henzel, W., Reilly,selective high-level expression of cloned genes. J. Mol. Biol 189, D., and Goeddel, D. V. (1993) Processing and characterization of113–130. human proguanylin expressed in Escherichia coli. J. Biol. Chem.

268, 22397–22401.15. Tabor, S. (1990). In ‘‘Current Protocols in Molecular Biology’’29. Aubert, J. P., Porchet, N., Boersma, A., Loucheux, L. M. H., and(Ausubel, F. A., Brent, R., Kingston, R. E., Moore, D. D., Seid-

Degand, P. (1982) Circulation dichroism studies on the proline-man, J. G., Smith, J. A., and Struhl, K., Eds.), pp. 16.2.1–16.2.11,rich glycoprotein of human parotid saliva. Biochim. Biophys.Greene and Wiley–Interscience, New York.Acta 704, 361–365.16. Lawlor, E. J., Elson, S. W., Holland, S., Cassels, R., and Hodgson,

30. Loomis, R. E., Bergey, E. J., Levine, M. J., and Tabak, L. A (1985)J. E. (1994) Expression in Escherichia coli of a clavaminic acidCircular dichroism and fluorescence spectroscopic analyses ofsynthase isozyme: a trifunctional oxygenase involved in clavu-a proline-rich glycoprotein from human parotid saliva. Int. J.lanic acid biosynthesis. Tetrahedron 50, 8737–8748.Peptide Protein Res. 26, 621–629.17. Klickstein, L. B., Bartow, T. J., Miletic, V., Rabson, L. D., Smith,

31. Morgan, W. T. (1985) The histidine-rich glycoprotein of serumJ. A., and Fearon, D. T. (1988) Identification of distinct C3bhas a domain rich in histidine, proline, and glycine that bindsand C4b recognition sites in the human C3b/C4b receptor (CR1,heme and metals. Biochemistry 24, 1496–501.CD35) by deletion mutagenesis. J. Exp. Med. 168, 1699–1717.

32. Wu, M-L., and Morgan, W. T. (1993) Characterization of hemo-18. Dacie, J. V., and Lewis, S. M., Eds. (1975) ‘‘Practical Haematol-pexin and its interaction with heme by differential scanning calo-ogy,’’ fifth ed., pp. 3–4, Churchill, Edinburgh/New York.rimetry and circular dichroism. Biochemistry 32, 7216–7222.

19. Brooks, I., Watts, D. G., Soneson, K. K., and Hensley, P. (1994) 33. Mulligan, M. S., Yeh, C-C.G., Rudolph, A. R., and Ward, P. A.Determining confidence intervals for parameters derived from (1992) Protective effects of soluble CR1 in complement and neu-analysis of equilibrium analytical ultracentrifuge data. Methods trophil-mediated tissue injury. J. Immunol. 148, 1479–1485.Enzymol. 240, 459–478.

34. Hill, J., Lindsay, T. F., Ortiz, F., Yeh, C-C.G., Hechtmann,20. Johnson, M. L., and Frasier, S. G. (1985) Nonlinear least squares H. B., and Moore, F. D. (1992) Soluble complement receptor type

analysis. Methods Enzymol. 117, 301–346. 1 ameliorates the local and remote organ injury after intestinalischemia-reperfusion in the rat. J. Immunol. 149, 1723–1728.21. Hensley, P., O’Keefe, M. C., Spangler, C. J., Osborne, J., and

/ m3971$0527 10-30-95 09:01:05 pepa AP-PEP