THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc

Vol. 269, No. 22, Issue of June 3, pp. 15808-15813, 1994 Printed in U.S.A.

Recombinant Pulmonary Surfactant Protein D POST-TRANSLATIONAL MODIFICATION AND MOLECULAR ASSEMBLY*

(Received for publication, January 31, 1994, and in revised form, March 25, 1994)

Edmond Crouch$§, Donald ChangS, Kevin Rust$, Anders Perssonn, and John Heusedl From the Departments of $.Pathology. IMedicine, and IlCell Biology and Physiology, Washington University School of Medicine, St. Louis, Missouri 63110

Pulmonary surfactant protein D (SP-D) is a member of a family of collagenous C-type lectins that includes the serum mannose binding proteins and surfactant protein A. Recent studies have shown that rat SP-D (rSP-D) mol- ecules are assembled as tetramers of trimeric subunits (12 mers) and that dodecamers can participate in higher orders of molecular assembly involving interactions of the amino-terminal peptide domains. In order to further study the assembly of SP-D in uitro, Chinese hamster ovary K1 cells were transfected with a full-length rat SP-D cDNA, and stable transfectants with high levels of SP-D production (approximately 6 x lo6 dodecamerd celll24 h) were obtained using a glutamine synthetase selection system. The secreted molecules (RrSP-D), which were purified by affinity chromatography on mal- tosyl-agarose, comigrated with rSP-D on SDS-polyacryl- amide gel electrophoresis in the presence and absence of reduction, and coeluted with rSP-D dodecamers from 4% agarose. The major bacterial collagenase-resistant peptide showed a decreased mobility on reduction con- sistent with the formation of intrachain disulfide bonds. A 17-kDa pepsin-resistant fragment was isolated follow- ing overnight digestion with pepsin at 27 “C, confirming the formation of a triple helical domain comparable in size and thermal stability to that of natural SP-D. The expressed protein contained sialylated endoglycosidase F-sensitive carbohydrate; amino acid analysis of acid and alkaline hydrolysates demonstrated essentially nor- mal levels of hydroxyproline, hydroxylysine, and hy- droxylysine-glycosides. Electron microscopic studies showed a molecular structure indistinguishable from lung SP-D, with a similar small subpopulation of mol- ecules showing higher orders of multimerization. Solid- phase neoglycoprotein binding assays gave the same saccharide inhibition profile as natural rat SP-D, and both proteins showed efficient saccharide-dependent agglutination of Escherichia coli. These studies demon- strate that a single genetically distinct chain type can account for the various and complex molecular assem- blies of SP-D, and further verify the potential physi- ologic significance of the disulfide-bonded multimers and higher aggregates isolated from rat, bovine, and hu- man lung lavage.

Surfactant protein D (SP-D)’ is a collagenous C-type lectin with structural similarities to pulmonary surfactant protein A,

* This work was supported by National Institutes of Health grants HL-44015 and GM-29647. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Jewish Hospital, 216 S. Kingshighway, St. Louis, MO 63110. Tel.: 314- 5 To whom correspondence should be addressed: Dept. of Pathology,

The abbreviations used are: SP-D, surfactant protein D; rSP-D, rat SP-D, RrSP-D, recombinant rSP-D; CHO, Chinese hamster ovary;

454-8462; Fax 314-454-5505.

serum mannose-binding protein, and bovine conglutinin (Crouch et al., 1993a; Sastry and Ezekowitz, 1993). SP-D is secreted into the air spaces by alveolar type I1 cells, and into the distal airways by nonciliated bronchiolar epithelial cells (Crouch et al., 1991b). Recent studies suggest that SP-D plays roles in pulmonary host defense (Kuan et al., 1992), and possi- bly, the extracellular metabolism of pulmonary surfactant (Persson et al., 1992).

Protein and cDNA sequencing studies have revealed that each SP-D chain consists of a short noncollagenous amino- terminal domain containing two cysteine residues, an uninter- rupted collagen-like sequence (59 Gly-X-Y triplets in rat and human SP-D), a short connecting sequence, and a noncollag- enous carboxyl-terminal domain that shows high sequence ho- mology with other members of the C-type lectin family (Crouch et al., 1993a; Lim et al., 1993; Lu et al., 1992; Rust et al., 1991; Shimizu et al., 1992). Among the known collagenous C-type lectins, the primary sequence and domain structure of SP-D is most similar to bovine conglutinin (Crouch et al., 1992; Lim et al., 1993).

Our previously published studies of rSP-D (Persson et al., 1988, 1989) and subsequent studies of the human and bovine proteins (Crouch et al., 1993a, 1993b; Lim et al., 1993; Lu et al., 1992, 1993; Rust et al., 1991) have shown that SP-D is as- sembled as high M, complexes of disulfide-bonded multimers of the 43-kDa subunits. Other recent studies have shown that rSP-D is predominantly assembled as a dodecamer with four trimeric subunits, each consisting of parallel arrangements of 43-kDa monomers (Crouch et al., 1992). Disulfide-mediated in- teractions involving the amino termini of the trimeric subunits yield a tetrameric structure with triple helical arms and pe- ripheral trimeric globular C-type lectin domains.2

Given the complexity of the molecular organization of SP-D it has not been possible to entirely exclude possible contribu- tions of a second chain type or other molecules in the assem- bly of SP-D multimers. Studies of SP-D synthesis and assem- bly have also been complicated by the rapid loss of SP-D expression during rat lung epithelial cell culture, and by the absence or low levels of SP-D production by the previously characterized human lung tumor cell lines (Crouch et al., 1 9 9 1 ~ ) . ~ For these reasons, we applied protein molecular bio- logic techniques to further examine the molecular structure and assembly of rat SP-D. Specifically, CHO K1 cells were transfected with rSP-D cDNA in an attempt to gain additional insights into the molecular requirements for the formation of the multimeric structures identified by electron microscopy. These studies provide the first demonstration of SP-D produc-

DMEM, Dulbecco’s modified Eagle’s medium; GMEM, Glasgow’s modi- fied Eaele’s medium: MSX. methionine sulfoximine; PAGE, polyacryl- amide gel electrophoresis.

. . ~

Crouch, E., Persson, A., Chang, D., and Heuser, J. (1994) J. B i d . Chem., in press.

A. Persson, unpublished data.

15808

Characterization of Recombinant SP-D 15809

tion and assembly in a mammalian transfection system. These studies also establish that a single chain type is sufii- cient for the formation of SP-D dodecamers and the higher or- der molecular assemblies identified in silicotic rat and human proteinosis pulmonary lavage.

MATERIALS AND METHODS Isolation of SP-D-Rat lung SP-D (rSP-D) was isolated from the

lavage supernatant of normal or silicotic adult rats using previously published methods (Persson et al., 1990; Crouch et al., 1991a). Briefly, bronchoalveolar lavage was clarified by centrifugation for 10 min at 150 x g prior to centrifugation for 30 min at 48,000 x g a t 4 “C. SP-D in the lavage supernatant was isolated by affinity chromatography on malto- syl-agarose and further purified by gel filtration chromatography on 4% agarose (Bio-Rad A-15M) in the presence of 10 mM EDTA.

Expression of Rat SP-D cDNA in CHO K1-For these studies we used a full-length ra t SP-D cDNA provided by Drs. J . H. Fisher and D. R. Voelker, Denver, CO (Shimizu et al., 1992). The cDNA was excised from pGEM-3Z with EcoRI and spliced into the corresponding site within the multiple cloning site of pEE14 (Ausubel, 1992; Cockett et al., 1990) distal to the powerful cytomegalovirus promoter/enhancer and proximal to the glutamine synthetase minigene driven by the SV40 late promoter (pEE14-rSPD). Subclones containing the appropriately oriented cDNA were identified by restriction mapping and confirmed by DNA sequenc- ing (data not shown).

Adherent CHO K1 cells (ATCC CCL-61) in Glascow’s modified Ea- gle’s medium (GMEM) were transfected with pEE14-rSPD or mock- transfected with pEE14 using Lipofectin (GIBCOBRL) according to the manufacturer’s instructions. Approximately 2 x lo5 cells in 60-mm plates were transfected with 8 pg of purified DNA in GMEM-10 in the presence of 33 1.11 of Lipofectin and in the absence of serum and antibi- otics. Incubations were performed for 6 2 2 h. Parallel plates containing transfected, mock-transfected, or untransfected CHO cells were then incubated in glutamine-free GMEM-10 containing 10% dialyzed fetal calf serum for 48 h. The cells were split at 1:8 and replated in the above medium supplemented with 25 PM methionine sulfoximine (MSX). After 2-3 weeks, colonies of vigorously proliferating and cohesive-appearing cells (approximately 1-5 colonies/lOO-mm dish) were isolated from the transfected and mock-transfected cultures using a cloning cylinder and separately replated in 35-mm tissue culture plates in the presence of MSX. Only rare groupings of loosely cohesive cells were identified in cultures of untransfected CHO cells exposed to MSX. After the cells reached confluence, fresh culture medium was applied and the cells were incubated for 16-24 h in the presence of fresh ascorbate (50 pg/ml). Although we have no evidence that ascorbate is required for synthesis or secretion of RrSP-D, ascorbate was included to help ensure complete hydroxylation of the collagen domain.

Aliquots of culture medium were then screened for secreted SP-D using a semi-quantitative solid-phase immunoassay, and by immuno- blotting with monospecific rabbit anti-rat SP-D antibody (Persson et al., 1989; Crouch et al., 1991a). Approximately 85% of all isolated colonies were positive, and the majority of the positive cultures showed compa- rable levels of immunoreactive material; no immunoreactivity was de- tected in media from the mock-transfected cells. In preliminary experi- ments, the size of the expressed mRNA was also assessed by Northern blotting of total RNA from one 60-mm plate using radiolabeled rSP-D cDNA as probe (Rust et al., 1991). Selected cell lines of transformants were further amplified following exposure to 100-1000 p MSX, and clones of interest were isolated by limiting dilution using microtiter plates. Prior to cloning, cells were continuously maintained in the pres- ence of MSX. For some preparative experiments, incubations were per- formed in the presence of fetal calf serum; this had no apparent effect on the yield of recombinant protein, but permitted multiple cycles of me- dium collection.

Metabolic Labeling of Recombinant SP-D-Nearly confluent cell cul- tures were briefly washed in fresh serum-free DMEM, and then incu- bated for 16 h in the same medium containing 20 pCi/ml~-[2,3-~H]pro- line (DuPont NEN). The harvested culture medium was centrifuged at 400 X g for 10 min, and the supernatant was dialyzed uersus TBS containing 10 mM EDTA and 0.2 mM phenylmethanesulfonyl fluoride. Following recalcification, the culture medium was applied to maltosyl- agarose, and bound proteins were eluted as described above. Nondia- lyzable radioactivity in starting material, and radioactivity in the un- bound and eluted fractions, was quantified by liquid scintillation spectrometry. The remaining material was temporarily stored a t 4 “C pending further characterization. For some studies, the cell layers were lysed in immunoprecipitation buffer (50 mM Tris-HC1, 150 mM NaCl,

0.5% (w/v) deoxycholate, 1% (v/v) Triton X-100, 0.1% (w/v) SDS, 2.5 mM EDTA, 0.1 mM phenylmethanesulfonyl fluoride, 2.5 mM N-ethylmaleim- ide, pH 7.5) and examined by SDS-PAGE (Persson et al., 1988).

Degradatiue Analysis-Peptic digestions were performed for 12-16 h at 27 “C in 0.5 M acetic acid in the presence of 100 pg/ml porcine pepsin A. Amino acid analysis of acid and alkaline hydrolysates, and digestions with purified bacterial collagenase, endoglycosidase F, and neur- aminidase were also performed as described for rat lung SP-D (Persson et al., 1988, 1989).

Gel Filtration Chromatography-Chromatography of natural or re- combinant SP-D was performed on a column of 4% agarose equilibrated with TBS containing 10 mM EDTA as previously described (Persson et al., 1990). The recovery of radiolabeled protein was monitored by liquid scintillation spectrometry. The column was routinely calibrated with internal standards of rSP-D, bovine thyroglobulin (670 kDa), rabbit IgG (150 kDa), and albumin (68 kDa).

SDS-PAGE-Proteins were resolved by SDS-PAGE on discontinuous 5%/10% polyacrylamide slab gels and visualized by silver staining, au- toradiography, or fluorography (Persson et al., 1988, 1990). Peptic and bacterial collagenase resistant fragments were resolved on 5%/12.5% or 5%/15% slab gels. Molecular weight were estimated relative to internal, unlabeled or 14C-labeled globular protein standards.

Solid-phase Saccharide Binding Assays-Solid-phase saccharide in- hibition assays were performed using maltosyl-BSA-coated microtiter wells as previously described (Persson et al., 1990). Briefly, SP-D neo- glycoprotein complexes that formed in the presence of serial dilutions of various competing saccharides were quantified using rabbit anti-rat SP-D and an indirect biotinylated goat anti-rabbit IgGhtreptavidin peroxidase detection system. Equivalent weights of natural and recom- binant rSP-D were found to give equivalent antibody binding in pre- liminary solid-phase immunoassays.

Electron Microscopy-Natural and recombinant SP-D were prepared for transmission electron microscopy by adsorption to mica flakes, freeze-drying, and platinum replication as described by Heuser (1983). Briefly, this involved adsorbing a 10 pg/ml solution of protein in the presence of 70 mM KCl, 30 m~ Hepes buffer, pH 7.4,5 mM MgCl,, and 3 mM EGTA to a slurry of wet-ground mica and then quick-freezing the mica with a homemade liquid helium-cooled copper block “Cryopress.” Thereafter, the mica pellet was freeze-fractured, deep-etched for 4 min at -100 “C, and rotary replicated with 2 nm of platinum deposited at 10 above the horizontal. The sample was thawed and mica was dis- solved out from under the platinum replica by flotation overnight on full-strength hydrofluoric acid. After extensive washing in water and retrieval onto 75-mesh Formvar-coated grids, the replica was viewed in a JEOL 200 CX electron microscope operated at 100 kV and photo- graphed at 70 ,000~ primary magnification in three dimensions. Final images for publication were contrast-reversed video thermal-prints of molecules selected for clarity and optimal morphology. Their contrast was enhanced against background by overlaying the panels of thermal- prints with a thin gray press-on mask (Lettraset no. 7) and brushing away the gray over the relevant parts of the molecules.

Bacterial Agglutination-Bacterial agglutination assays were per- formed usingEscherichia coli Y1088 and a spectrophotometric sedimen- tation assay essentially as described by Kuan et al., 1992. Following the addition of SP-D and calcium, the absorbance at 700 nm was monitored continuously for at least 2 h.

RESULTS

CHO K1 Cells Dansfected with rSP-D cDNA Express SP-D mRNA and Protein-Transfection of CHO K1 cells with rat SP-D cDNA using the PEE14 vector (see “Materials and Meth- ods”) (Fig. 1) resulted in the synthesis, assembly, and secretion of functional SP-D. The initial pools selected in the presence of 25 p~ MSX and incubated in the presence of ascorbate accu- mulated SP-D in the culture medium as detected by enzyme- linked immunosorbent assay. SDS-PAGE of the culture me- dium of cultures incubated overnight in the absence of serum and the presence of ascorbate and MSX showed a major protein component that comigrated with authentic rSP-D in the pres- ence and absence of reduction. This component bound to, and was specifically eluted from, maltosyl-agarose (Fig. 2). Identi- fication of the protein as SP-D was confirmed by immunoblot- ting using monospecific antibodies to rat SP-D (data not shown). Northern hybridization of total cellular RNA from the transfected cells with rSP-D cDNA showed a single hybridiz-

15810 Characterization of Recombinant SP-D 1 2 3 4 5

pEE6

SV40 Late Promoter - cDNA. A full-length rat SP-D cDNA (Shimizu et al., 1992) was excised

FIG. 1. Vector used for transfection of CHO K1 cells with rSP-D

from pGEM-3Z with EcoRI and spliced into the corresponding site within the multiple cloning site of pEE14, downstream to the cytomega- lovirus promoter/enhancer and upstream to the glutamine synthetase minigene driven by the SV40 late promoter (pEE14-rSPD).

1 2 3 4 5 6 7 a 9 x) n12 kDa ." ~ -" - " " - - - "

200- - 97"

68- 1

45-

29-

L T2

FIG. 3. Northern analysis of control and transfected CHO K1 " -' cells. RNA was extracted from untransfected and mock-transfected

cells, cells transfected with SP-D cDNA, rat type I1 cells ( 7 2 ) . and lung tissue ( L ) . Equivalent weights of total cellular RNA (IO pg) were re- solved on agarose gels, transferred to nitrocellulose, and probed with radiolabeled rSP-D cDNA. Comparable loadings were confirmed by ei- thidium bromide staining. Lane I, human lung; lane 2, rat type I1 cells; lane 3, CHO K1 parental cell line; lane 4, mock-transfected CHO cells; lune 5; CHO cells transfected with (pEE14-rSPD). The transfected cells show high levels of expression of a major mRNA that hybridizes with SP-D cDNA. The larger cross-hybridizing species are interpreted as

" incompletely processed nuclear transcripts.

clones were further characterized using techniques comparable to those employed for the characterization of rat SP-D synthe- sized by isolated type I1 cells (Persson et al., 1988). The use of radiolabeled proteins allowed convenient assessment of the re- covery from chromatographic columns and facilitated the un- ambiguous identification of cleavage products obtained follow- -

+ - DTT

FIG. 2. SDS-PAGE of recombinant rSP-D. CHO K1 cells stably expressing transfected rSP-D cDNA were washed and incubated for 12 h in serum-free DMEM supplemented with ascorbate. The culture me- dium was harvested, dialyzed uersus TBS, and applied to maltosyl- agarose. Bound proteins were eluted with 10 mM EDTA. Proteins in aliquots of the eluted fractions were dissolved in SDS sample buffer in the presence or absence of dithiothreitol (Dm), resolved by SDS-PAGE, and visualized by silver staining. Lane I, globular protein standards; lane 2, starting material; lunes 3-7, EDTA eluted fractions, reduced with D m , lanes 8-12, EDTA-eluted fractions, unreduced. The position of rSP-D disulfide-bonded trimers (2 ' ) and reduced monomers (M) is indicated a t right. The band above SP-D in lanes 2-7 represents re- sidual albumin.

able species that migrated near the position of authentic SP-D mRNA (Fig. 3).

The yield per confluent 100-mm plate after an overnight incubation of a selected cell line (BZA, and representative clones) was approximately 5 pg/ml with a calculated production rate of approximately 6 x 10" dodecamers/celV24 h. The pro- duction of SP-D was increased by at least %fold following iso- lation of colonies and re-expansion of cells that survived after a 72-h exposure to 1000 p~ MSX.

Characterization of Recombinant Cell Lines and Clones- The recombinant proteins expressed by selected cell lines or

ing proteolytic or glycosidase digestion procedures. Nearly confluent cultures were transferred to DMEM

supplemented with 50 pg/ml freshly prepared ascorbic acid and incubated for up to 24 h in absence of serum. The harvested medium was subjected to a brief centrifugation to remove any cellular debris (400 x g for 10 min), and applied to a column of maltosyl-agarose, and bound proteins were eluted with 100 mM maltose or 10 mM EDTA. The recombinant bound to maltosyl agarose and was specifically eluted with EDTA (Fig. 4A ). Chro- matography of metabolically labeled SP-D indicated that >70% of the total nondialyzable radioactivity in the culture medium was bound and specifically eluted; SDS-PAGE and fluorogra- phy of the wash showed no significant SP-D indicating that virtually all of the SP-D bound to the column. In addition, >80% of the total bound radioactivity eluted in the position of natural rSP-D dodecamers on gel filtration chromatography (Fig. 4B ). SDS-PAGE of aliquots of the radiolabeled cell layer lysate and equivalent volume proportions of the medium culture indicated that the majority of the newly synthesized recombinant protein was secreted and accumulated in the medium during the 16-h labeling period (not shown).

The major pepsin- and bacterial collagenase-resistant frag- ments of recombinant rSP-D migrated at the expected positions of the corresponding fragments of natural rSP-D or human SP-D in the absence and presence of sulfhydryl reduction (Fig.

Characterization of Recombinant SP-D

A 5000 -

4500 -- T

4000 ..

-c 3500 ..

v) 3000 .. 0

. 2500 .. ? E 2000 . .

8 1500 -. EDTA

1000 .-

.-obnr.ul-m~r.uJ~~~ .- - - - -

Fraction Number

B 7000 - !Jo SP-C

6000 .. 4 4 ; i

- 5000 .. P E

g 4000 ‘. II

? E

3000 ‘-

z kc :: z $951181??

Fraction Number

FK:. 4. Gel filtration chromatography of recombinant rSP-D. To facilitate quantitation of the recovery of rSP-D following various chromatographic procedures, recombinant rSP-D was metabolically with tritiatcd-proline, as described under “Materials and Methods,” and separated from minor radiolabeled contaminants by maltosyl-agarose chromatography. A, maltosyl-agarose chromatography of radiolabeled rSP-D. Greater than 80% of the nondialyzahle radioactivity in the me- dium bound to the column and was clutcd with EDTA. B, fractions eluted from maltosyl-agarose were examine by gel filtration chroma- tography on 4% agarose. Approximately 85% of the total radioactivity was recovered in a major peak that coeluted with authentic rat SP-D dodecamers. Only a small fraction of the total radioactivity eluted neal thyroglobulin (773) in the position of rSP-D trimers. or near the Vi. the expected position of rSP-D monomers.

5 1” (Crouch et al., 1993a; Persson et al., 1989). The major peptic fragment migrated at approximately 29 kDa relative to re- duced globular standards (lane 3, 17-kDa collagen standards) and showed the same apparent size in the absence of prior sulfhydryl reduction (not shown). The major collagenase-resist- ant fragment migrated at approximately 18 kDa (globular stan- dards, reduced), and migrated more rapidly without prior re- duction with dithiothreitol (compare lanes 1 and 2). The mobility was also increased following incubation with endogly- cosidase F (Fig. 5, right panel) or neuraminidase (not shown) indicating the presence of sialylated N-linked oligosaccharides. Amino acid analysis of acid hydrolysates of the unlabeled pro- tein showed a pattern of post-translational hydroxylation that was virtually indistinguishable from that of the natural protein with essentially equivalent ratios of Hypi(Pro + Hyp) and Hyl/ (Lys + Hyl (Table I). Analysis of alkaline hydrolysates showed that the most of the hylys consisted of acid labile species that coeluted with glucosyl-galactosyl-hydroxylysine in hydroly- sates of human placenta type IV collagen (data not shown).

Ultrastructural Studies of RrSP-D Demonstrate the Same Molecular Organization as Natuml SP-D-Electron micros- copy of maltosyl-agarose purified recombinant rSP-D spread on mica and shadowed with platinum showed a uniform popula- tion of molecules characterized by four rigid-appearing arms that were nonrandomly arrayed about a poorly defined central

Std 1 2 3

kDa

*

0 I I

+c +P

Std

15811

1 2 kDa

-6e

- +EndoF

FM:. 5. Degradative analysis of recombinant rSP-D. Maltosyl- agarosc purified radiolabled rSP-D was treated with purified bacterial collagenase i+C I. pepsin c+P I, or cndoglycosidase F i+E~ldo F J as dc- scribed under “Materials and Methods.“ The collagenase and peptic cleavage products were resolved by SDS-PAGE on 5%/12.5% slab gels rlc,fi pnr~~ls, /nrzr.s 13 J; intact rSP-D and the Endo-F cleavage products were resolved by SDS-PAGE on a 5%/10% slab gel (rQ$ll pcmel, lmes I and % I_ Both gels included internal “C-labeled globular protein stan- dards IS/C/I. The resolved proteins were visualized by fluorescence autoradiography. The positions apparent molecular mass of selected standards are indicated at /e/i and rig/z/.

TAIIIX I Amino acid composition fresicluesl 1000)

Amino acid analyses of acid hydrolysates. No attempt was made to correct for hydrolytic losses.

S-1) ISP-D rSP-n control recmlbinsnt silica , I) = 1 ,‘i 01 = 1,” il l = iSI’

cys 14 12 15 HYP 32 26 22 A% 71 83 78 Thl 3 5 33 31 SC1 66 85 55 GIX 142 136 149 Pro 48 65 62 GlY 212 221 214 Ala 113 99 112 Val 26 15 19 Met 8 8 10 Ile 19 21 18 Leu 60 60 61 ?Lr 6 ND” 6 Phc 20 22 29 HYI 24 16 21 His 4 16 7 Lvs 49 42 41 A;.g 47 41 40 RP ND ND ND

Hyp + Pro 80 91 84 Hyl + Lys 69 58 62 HypilPro + Hyp) 0.4 0.4 0.26 Hyl/(Lys + Hyl, 0.29 0.28 0.34

” Natural rat SP-D isolated from the adult 10,000 x g supernatant adult rat lung lavage (Persson et nl.. 1989).

” Recombinant rSP-D. representative of two analyses. ’ rSP-D from the 10,000 x g supernatant of silicotic rat lavage

(Crouch et (I[.. 1991a). ” ND, not determined.

15812 Characterization of Recombinant SP-D RrSP-D

combinant rSP-D. RrSP-D was isolated FIG. 6. Electron microscopy of re-

from the culture medium of transfected cells by maltosyl-agarose chromatogra- phy. Platinum shadowed molecules were visualized by transmission electron mi- croscopy as described under ”Materials and Methods.” RrSP-D (top two rows); rSP-D (bottom two rows). Note: prepara- tions contained SP-D dodecamers and a subpopulation of multimers morphologi- cally indistinquishable from those of natural rSP-D. The RrSP-D multimers appeared to account for <5% of the total molecular assemblies.

r SP-D

I .neeen

......................... ~ .......................... : .......................... i ......................................................

e.eeeee

1.eeeee

e.eeee nin 1 5 e . e ~

.......................... c .......................... i .......................... i ......................................................

- ................................................................................ (. .......................... ) ..........................

-. ........... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

e.oeooo e .enen n l n 150.00

FIG. 7. SP-D-mediated bacterial agglutination. The effects of equivalent weight amounts of natural and recombinant SP-D on the agglutination of E. coli Y1088 were examined as described under “Ma- terials and Methods.” The decrease in absorbance a t 700 nm that ac- companied sedimentation of agglutinated organisms was monitored continuously on a spectrophotometer. Equivalent weights of natural

hub (Fig. 6). Each arm measured approximately 46 nm in length and was terminated at the end opposite from the hub by an 8-9-nm diameter globule, identical to that of natural SP-D? These preparations also showed small numbers of higher order multimers consisting of SP-D molecules associated a t their amino termini. Assembly appeared complete, free arms (or ap- parent dimeric or trimeric arrangements of arms) were not identified in preparations of rSP-D obtained by maltosyl-agar- ose or sequential saccharide affinity and gel filtration chroma- tography.

Carbohydrate Binding ActivitySolid-phase binding assays to maltosyl-BSA showed the same saccharide inhibition profile as natural rSP-D (inositol > maltose > glucose > mannose) (Persson et al., 1990). The maltose inhibition curves for equiv- alent weights of the two proteins were superimposable with I,, values of approximately 3 mM. Similar results were obtained with glucose, except that both proteins required a %fold higher concentration of saccharide to achieve 50% inhibition.

Recombinant SP-D Is Functional as a Microbial Agglu- tinin-Recombinant rSP-D grossly agglutinated E. coli Y1088, in the absence but not the presence of maltose or EDTA (not shown). Using a spectrophotometric sedimentation assay (Kuan et al., 1992) equivalent weights of the natural and re- combinant proteins also gave identical kinetics (Fig. 7) and dose-dependence (not shown) of saccharide- and calcium-de- pendent bacterial agglutination.

DISCUSSION The expression of SP-D in CHO cells with the assembly of

dodecamers apparently identical to natural rSP-D establishes that the information required for the assembly of these struc- tures is specified in the primary structure of SP-D. These studies also establish that stable triple helical arms can form

(upper panel) and recombinant SP-D (middle panel) showed identical kinetics of agglutination. The agglutination of bacteria by RrSP-D was inhibited by maltose (bottom panel) or EDTA, as previously described for natural rat and human SP-D.

Characterization of Recombinant SP-D 15813 as homopolymers of a single chain type, and indicate that ad- ditional polypeptides are not required for the cross-linking of SP-D arms or for the assembly of the terminal globular domains.

There are previously published studies that describe the synthesis and secretion of full-length and fully assembled col- lagenous molecules in a mammalian expression system (Ala- Kokko et al., 1991; Bonaldo et al., 1990; Lee et al., 1988; Maz- zorana et al., 1993; Penman et al., 1991). However, to our knowledge, this is the first study to demonstrate a full complement of collagen post-translational modifications for the recombinant protein including: near physiologic levels of hydroxyproline, hydroxylysine, and hydroxylysine glycosides; the formation of sialylated N-linked oligosaccharides, the ac- quisition of a protease-resistant triple helical conformation, and the formation of appropriate inter- and intrachain disul- fide bonds. The observation that these cells have the capacity to hydroxylate and subsequently glycosylate lysine residues (i.e. express active lysyl hydroxylase and appropriate glycosyl transferases, further validates the use of these cells for the ex- pression of other collagenous proteins.

Structure-function studies of the carbohydrate recognition domains of mannose-binding protein and other C-type lectins effectively utilized carbohydrate recognition domains ex- pressed in bacteria (e.g. Quesenberry and Drickamer, 1992). However, there are considerably less data on the expression of full-length C-type lectins, including the collagenous C-type lec- tins. In general, the complexity of assembly and post-transla- tional modification of these proteins necessitate the utilization of eukaryotic expression systems. SP-A cDNAs and genomic sequences have previously been expressed in CHO (McCor- mack et al., 1990) and COS cells (Alcorn et al., 1992). These molecules show prolyl hydroxylation, N-linked glyclosylation, and trimer formation; and appropriate biologic activities in certain functional assays. On the other hand, there is now considerable evidence that these molecules are deficient in the assembly of hexameric complexes (Greenspan et al., 1989; Haagsman et al., 1989; Manz-Keinke et al., 1992; Voss et al., 1988, 1991). This is of concern given that multimerization has been shown influence certain biologic activities in vitro (Manz- Keinke et al., 1992). SP-A expressed in baculovirus has also been shown to be abnormally assembled, to lack hydroxypro- line, and to contain abnormal N-linked oligosaccharides (Kalina et al., 1993). Although recombinant forms of MBP have also been expressed and found to be biologically active (Schweinle et al., 1993), there is very little data on the struc- ture of the recombinant protein.

An interesting observation, relates to the identification of a subpopulation of higher order molecular assemblies in the preparations of recombinant SP-D. These assemblies, which consist of two or more dodecamers associated at or near the central hub of SP-D dodecamers, are indistinguishable from multimolecular assemblies identified in bronchoalveolar lavage from silica-treated rats. Similar structures also account for the majority of the SP-D eluted from the insoluble fraction of la-

vage from patients with alveolar protein~sis.~ In summary, functional SP-D can be successfully expressed

in CHO Kl cells. This system should provide a useful for model for studying SP-D synthesis and secretion and for future struc- ture-function studies employing molecular mutagenesis.

Acknowledgment-We thank Janet North for excellent secretarial assistance. We also thank Robyn Roth for technical assistance with the EM studies.

REFERENCES

Ala-Kokko, L., Hyland, J., Smith, C., Kivirikko, K. I., Jimenez, S. A,, and Prockop,

Alcorn, J. L., Chen, Q., Boggaram, V., and Mendelson, C. R. (1992)Am. J. Physiol. D. J. (1991) J. Biol. Chem. 266, 14175-14178

262, L437-L445 Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J.

A,, and Struhl, K. (1992) Current Protocols in Molecular Biology, Chap. 16, Sect. 14, John Wiley & Sons, New York, NY

Bonaldo, P., Mucignat M.-A,, and Colombatti, A. (1990) Matrix 10, 139-147 Cockett, M. I., Bebbington, C. R., and Yarranton, G. T. (1990) Biolnchnology 8,

Crouch, E., Persson, A,, Chang, D., and Parghi, D. (1991a) Am. J. Pathol. 139,

Crouch, E. C., Rust, K., Chang D., Persson, A,, Parghi, D., and Mariencheck, W.

Crouch, E. C., Rust, K., Persson, A,, Mariencheck, W., Moxley, M., and Longmore,

Crouch, E., Persson, A,, Heuser, J., and Chang, D. (1992) FASEB J. 6, A1160 Crouch, E., Persson, A,, and Chang, D. (1993a) Am. J. Pathol. 142,241-248 Crouch, E. C., K. Rust, R. Veile, H. Donis-Keller, and Grosso, L. (199313) J. Biol.

Greenspan, D. S., Hoffman, G. G., and Lee B-S. (1989) J. Biol. Chem. 264,20683-

Haagsman, H. P., White R. T., Schilling, J., Lau, K., Benson, B. J., Golden, J.,

Heuser, J. E. (1983) J. Mol. Bid. 169, 155-195 Kalina, M., McCormack, F. X., Crowley, H., Voelker, D. R., and Mason, R. J. (1993)

662-667

765-776

(1991b) Am.. J. Respir. Cell Mol. Biol. 5, 13-18

W. (1991c)Am. J. Physiol. 260, L247-L253

Chem. 268,2976-2983

20687

Hawgood, S., and Clements, J. A. (1989) Am. J. Physiol. 257, L421-L429

Kuan, S.-F., Rust, K., and Crouch, E. C. (1992) J. Clin. Invest. 90,97-106 Lee S.-T., Smith, B. D., and Greenspan, D. S. (1988) J. Biol. Chem. 263, 1341P

J. Histoehem. Cytochem. 41, 57-70

13418 Lim B. L., Lu, J., and Reid, K. B. M. (1993) Immunology 78, 159-165 Lu, J., A. C. Willis, and Reid, K. B. M. (1992) Biochem. J. 284, 795-802 Lu, J., Wiedemann, H., Holmskov, U., Thiel, S., Timpl, R, and Reid, K. B. M. (1993)

Manz-Keinke, H., Plattner, H., and Schlepper-Schafer, J. (1992) Eur. J. Cell Biol.

Mazzorana, M., Gruffat, H., Sergeant, A., and van der Rest, M. (1993) J. Biol.

McCormack F. X., Fisher, J. H., Suwabe, A,, Smith, D. L., Shannon, J. M., and

Penman, M., Lux, A,, Freedman, N. J., Rohrer, L., Ekkel, Y., McKinstry, H.,

Persson, A,, Rust, K., Chang, D., Moxley, M., Longmore, W., and Crouch, E. (1988)

Persson, A., Chang, D., Rust, K., Moxley, M., Longmore, W., and Crouch, E. (1989)

Persson, A., Chang, D., and Crouch, E. (1990) J. Bid. Chem. 265,5755-5760 Persson, A,, Gibbons, B., Schoemaker, J. D., Moxley, M., and Longmore, W. (1992)

Quesenbeny, M. S., and Drickamer, K. (1992) J. Biol. Chem. 267,10831-10841 Rust, K., Grosso, L., Zhang, V., Chang, D., Persson, A., Longmore, W., Cai, G.-Z.,

and Crouch, E. C. (1991) Arch. Biochem. Biophys. 290,11&126 Sastry, K., and Ezekowitz, R. A. (1993) Cum Opin. Immunol. 6, 59-66 Schweinle, J. E., Nichiyasu, M., Ding, T. Q., Sastry, K., Gillies, S. D., and Ezeko-

Shimizu, H., J. H. Fisher, P. Papst, B. Benson, K. Lau, R. J. Mason, and Voelker,

Voss, T., Eistetter, H., and Schafer, K. P. (1988) J. Mol. Biol. 201,219-227 Voss, T., Melchers, K., Scheirle, G., and SchSer, K. P. (1991) Am. J. Respir. Cell

Eur. J. Biochem. 215,793-799

57,95-100

Chem. 268,3029-3032

Voelker, D. R. (1990) Biochim. Biophys. Acta 1087, 190-198

Resnick, D., and Krieger, M. (1991) J. Biol. Chem. 266, 23985-23993

Biochemistry 27,8576-8584

Biochemistry 28,6361-6367

Biochemistry 31,12183-12189

witz, R. A. B. (1993) J. Bid. Chem. 268,364-370

D. R. (1992) J. Biol. Chem. 267,1853-1857

MOL. B~OZ. 4,88-94

E. Crouch, D. Chang, and J. Heuser, unpublished data.