A Common Polymorphism in the SFTPD Gene Influences Assembly, Function, and Concentration of Surfactant Protein D
http://www.100md.com
免疫学杂志 2005年第3期
Abstract
Surfactant protein D (SP-D) plays important roles in the host defense against infectious microorganisms and in regulating the innate immune response to a variety of pathogen-associated molecular pattern. SP-D is mainly expressed by type II cells of the lung, but SP-D is generally found on epithelial surfaces and in serum. Genotyping for three single-nucleotide variations altering amino acids in the mature protein in codon 11 (Met11Thr), 160 (Ala160Thr), and 270 (Ser270Thr) of the SP-D gene was performed and related to the SP-D levels in serum. Individuals with the Thr/Thr11-encoding genotype had significantly lower SP-D serum levels than individuals with the Met/Met11 genotype. Gel filtration chromatography revealed two distinct m.w. peaks with SP-D immunoreactivity in serum from Met/Met11-encoding genotypes. In contrast, Thr/Thr11 genotypes lacked the highest m.w. form. A similar SP-D size distribution was found for recombinant Met11 and Thr11 expressed in human embryonic kidney cells. Atomic force microscopy of purified SP-D showed that components eluting in the position of the high m.w. peak consist of multimers, dodecamers, and monomers of subunits, whereas the second peak exclusively contains monomers. SP-D from both peaks bound to mannan-coated ELISA plates. SP-D from the high m.w. peak bound preferentially to intact influenza A virus and Gram-positive and Gram-negative bacteria, whereas the monomeric species preferentially bound to isolated LPS. Our data strongly suggest that polymorphic variation in the N-terminal domain of the SP-D molecule influences oligomerization, function, and the concentration of the molecule in serum.
Introduction
The innate immune system uses pattern recognition receptors (PRRs)3 to mark microorganisms for destruction either by direct lysis, by complement-induced lysis, or by phagocytosis. In response to binding of conserved pathogen-associated molecular patterns, various PRRs can either induce or inhibit the immune response (1). Surfactant protein D (SP-D) is a member of the protein family of collectins that also includes mannose-binding lectin and surfactant protein A (2, 3, 4). SP-D serves as a PRR by binding selectively to the surfaces of bacteria, viruses, and fungi, thereby enhancing phagocytosis and intracellular killing (2, 4, 5). In addition, SP-D was recently shown to directly inhibit growth of Gram-negative bacteria and fungi by increasing membrane permeability (6, 7). Mice deficient in SP-D accumulate alveolar surfactant phospholipids, exhibit enhanced acute inflammatory responses in the lung to a variety of stimuli (8), and develop spontaneous chronic inflammation, emphysema, and fibrosis (9). However, in vitro data have indicated additional proinflammatory effects of SP-D, and recent data suggested that SP-D, through binding of different receptors, acts in a dual manner to suppress or enhance inflammation depending on binding orientation of the SP-D molecule (10).
SP-D is composed of homotrimeric subunits with each polypeptide in the subunit consisting of an N-terminal domain with two conserved cysteines critical in cross-linking within and between subunits, a collagen-like region, an helical neck region, and a C-type lectin carbohydrate recognition domain (CRD) (11). The SP-D molecule is found as single subunits or as four subunits in a cruciform structure, but higher oligomers are also observed (12, 13). SP-D binds to carbohydrate and lipid ligands by the CRDs, but high affinity interactions require oligomeric assembly mediated by N-terminal cross-linking of trimeric arms (14).
SP-D is mainly synthesized in alveolar type II cells of the lung, but SP-D also localizes to epithelial cells of mucosa-associated tissue in the genitourinary and gastrointestinal tracts, and SP-D is present in amniotic fluid as well as in serum (15, 16, 17, 18). The origin of SP-D in the vascular compartment is not known, but it is generally assumed that SP-D is released from the lung into the bloodstream (19). This is consistent with the observation that the levels of SP-D in the blood increase in association with certain types of lung injury (18).
Three polymorphisms have been identified in the coding sequence of human SP-D: codons corresponding to amino acid residue 11 (Met11Thr), residue 160 (Ala160Thr), and residue 270 (Ser270Thr) (20, 21, 22) in the mature protein. Two clinical studies have associated the SP-D variants of amino acid 11 with disease. The SP-D allele coding for Met11 was associated with severe respiratory syncytial virus infection in infants, whereas Thr11 was suggested to increase susceptibility to tuberculosis (21, 23).
Polymorphisms of PRRs like mannose-binding lectin, TLR2, and TLR4 are associated with altered risk of a variety of disorders like autoimmune diseases, asthma, and atherosclerosis and to infections (3, 24, 25). The present work describes how known allelic variation in the coding region of SP-D influences the structure, the function, and the serum level of the protein, changes that may imply altered susceptibility to various infectious or immune-related diseases.
Materials and Methods
SP-D genotyping by PCR sequence-specific priming
SP-D single-nucleotide polymorphisms in the form of the structural variants named SP-D11/Met, SP-D11/Thr, SP-D160/Thr, SP-D160/Ala, SP-D270/Ser, and SP-D270/Thr were typed by PCR sequence-specific priming using the following six reactions: SP-D11/Met (T): 5'-primer, ACCTACTCCCACAGAACAAT; 3'-primer, GGAAGAAACACGTCTCCAGA; SP-D11/Thr (C): 5'-primer, ACCTACTCCCACAGAACAAC; 3'-primer, AGGAAGAAACACGTCTCCAGA; SP-D160/Thr (A): 5'-primer, TCTCTCTGACCCTAAAGTTGC; 3'-primer, CTCACCTGCTGCCCCTGT; SP-D160/Ala (G): 5'-primer, TCTCTCTGACCCTAAAGTTGC; 3'-primer, CTCACCTGCTGCCCCTGC; SP-D270/Ser (T): 5'-primer, CTGGTGGACAGTTGGCCT; 3'-primer, TCTGACCCGCCATCATCGTT; SP-D270/Thr (A): 5'-primer, CTGGTGGACAGTTGGCCA; 3'-primer, TCTGACCCGCCATCATCGTT. As internal positive control, we included a PCR covering exon 4 of the human mannose-binding lectin (mbl2) gene: 5'-primer: GAGTTTCACCCACTTTTTCACA; 3'-primer: GCCTGAGTGATATGACCCTTC. The PCR was performed essentially as described by Aldener-Cannava and Olerup (26).
Abs and ELISA
The rabbit anti-human SP-D Abs K477 (18) and P13 (27) were used in the SP-D quantification assay and in the solid-phase binding assays quantifying the amount of SP-D bound to bacteria, influenza A virus (IAV), LPS, and mannan, respectively. A mAb Hyb 246-4 was used for detection in the SP-D quantification assay (18). SP-D was quantified in serum and amniotic fluid by ELISA (18).
Purification of SP-D from amniotic fluid
A pool of amniotic fluid (n = 6) was centrifuged at 4000 rpm and 4°C for 30 min and purified by maltosyl agarose affinity chromatography (28). Purified SP-D was concentrated with a Vivaspin 6 (10,000 m.w. cut-off) concentrator (Vivascience), and the structurally different forms of SP-D were separated by gel filtration chromatography.
Gel filtration chromatography
Gel filtration chromatography was performed on 200-μl samples of normal human serum from genotyped individuals’ pooled amniotic fluid (n = 6), individual amniotic fluid, and purified amniotic fluid SP-D. The samples were applied to an analytical Superose 6 column connected to a fast-performance liquid chromatography system (Amersham Biosciences) using TBS (pH 7.4) containing 10 mM EDTA and 0.05% emulphogene as eluant at a flow rate of 24 ml/h. Fractions of 0.8 ml were collected and quantified by the SP-D ELISA. SP-D eluted as two structurally different forms and was collected in fractions 3 and 4 ("SP-D high") and 6 and 7 ("SP-D low"), respectively.
SDS-PAGE/Western blotting
Individual amniotic fluid containing both the SP-D high and the SP-D low form were separated by gel filtration chromatography, and the fractions were quantified for SP-D by ELISA. SP-D (500 ng) from fractions 4 and 7 were separated by SDS-PAGE gel electrophoresis in 4–20% (w/v) polyacrylamide gradient gels with a discontinuous buffer system and blotted onto polyvinylidene difluoride membranes (Immobilon P; Millipore) as described previously (18). K477 was used for detection.
Atomic force microscopy (AFM)
SP-D molecules were analyzed by AFM. For AFM imaging, 20-μl droplets of fractions of purified SP-D (10 μg/ml) were incubated on freshly cleaved mica (1 cm in diameter) for 10–15 min at room temperature. They were rinsed with deionized water (1 min) and dried with a flow of N2. The samples were imaged in Tapping Mode on Nanoscope IIIa (Digital Instruments) in air under ambient conditions (FM tip: Nanosencors; interaction force, 1 nN/m; scan size, 2 nμ; scan speed, 1 μm/s).
Expression of recombinant forms of SP-D in human embryonic kidney (HEK) 293T-rex cells using the "Flp-In" system
All reagents were purchased from Invitrogen Life Technologies unless otherwise stated. The Met11 and Thr11 forms of SP-D were amplified by RT-PCR and TA-cloned into the "Flp-In" pcDNA5.1/FRT/V5-His TOPO TA expression vector according to the manufacturer’s protocol. Shortly, first-strand synthesis was performed with the Thermoscript kit and oligo(dT) priming from 1 μg of a pool of human total RNA from the trachea (Clontech). Primers used in the following PCR had the sequence: HuSP-D leader f1(flp-in), 5'-GGG GGA TCC GCC ATG CTG CTC TTC CTC CTC-3', HuSP-D(stop) r1(flp-in), 5'-GGG TCT AGA TCA GAA CTC GCA GAC CAC AAG ACG A-3'. PCR was performed with the PFU polymerase (Stratagene) according to the manufacturer’s protocol, and A'-overhangs were added post-PCR. The fragments were TA cloned into the expression vector and constructs were verified by sequencing. An endotoxin-free preparation (12 μg) of the plasmid encoding the recombinase and 1.5 μg of the specific construct were transfected into 50% confluent Flp-In HEK 293T-rex cells in 80-cm2 culture flasks using the transfection reagent jetPEI according to manufacturer’s protocol (Qbiogene). The cells were grown in complete medium (DMEM containing 10% FCS, penicillin/streptomycin, and L-glutamine). After 2 days, the cells were transferred to medium containing 150 μg of hygromycin/ml of medium, and the medium was changed every third or fourth day until hygromycin-resistant cells became confluent. The cells were trypsinized and split into 150-cm2 culture flasks and allowed to attach in complete medium containing hygromycin. After 6 h, the medium was changed to DMEM only containing L-glutamine. The culture supernatant was harvested after 72 h at 37°C. The medium was then clarified by centrifugation at 10,000 x g for 10 min and stored at 4°C until further analysis.
Bacterial preparation
Clinical isolates of three Gram-negative (Escherichia coli, Klebsiella oxytoca, Haemophilus influenzae) and two Gram-positive (Staphylococcus aureus and Streptococcus mutans) bacterial strains were grown overnight at 37°C in serum broth followed by centrifugation (4000 rpm for 5 min). The bacteria were resuspended in formaldehyde buffer (0.9% NaCl, 0.5% formaldehyde) for 1 h with shaking, washed three times in TBS at pH 7.4, and resuspended in 0.05 M sodium carbonate buffer (pH 9.6) to the desired density determined spectrophotometrically at 660 nm.
Binding of SP-D to bacteria
All bacterial strains were suspended at a density of OD660 0.7 and adhered to Polysorp microtiter plates (Nunc) by overnight incubation at 4°C. The plates were blocked with TBS and 0.05% Tween 20 containing 10 mM CaCl2 for 30 min. The same buffer was used as washing buffer. All incubations were performed at room temperature with shaking unless otherwise stated. Two-fold dilutions of the two variant forms of SP-D (SP-D high and SP-D low) were added to the plate diluted in TBS, 0.05% Tween 20 containing 10 mM CaCl2 or 10 mM EDTA, and incubated overnight at 4°C. The binding between bacteria and SP-D was measured using SP-D polyclonal P13 Ab (1:5000, 2 h) followed by incubation with alkaline phosphatase goat anti-rabbit Ig conjugate (Sigma-Aldrich) diluted 1/2000 for 1 h, and para-nitrophenyl phosphate (1 mg/ml). Binding was determined spectrophotometrically at 405 nm. The experiments were performed in duplicate and repeated three times.
Binding of SP-D to IAV
Phillipine 82/H3N2 strain IAV was grown in the chorioallantoic fluid of 10-day-old chicken eggs, purified on a discontinuous sucrose gradient, stored, and characterized as previously described (29). Binding of purified SP-D variants to IAV was tested using an ELISA as previously described using polyclonal P13 Ab for detection of SP-D binding (12, 30, 31).
Binding of SP-D to mannan and LPS
Maxisorp microtiter plates (Nunc) were coated with 10 μg/ml mannan purified from Saccharomyces cerevisiae (32), rough LPS (E. coli O26:B6; Sigma-Aldrich) or smooth LPS (E. coli O55:B5; Sigma-Aldrich) in sodium carbonate buffer overnight at 4°C. After wash in washing buffer, the plates were incubated overnight at 4°C with fractions from a gel filtration chromatography of pooled amniotic fluid (n = 6). Before addition to the plates, the fractions were diluted 1/4 in 1) TBS, 0.05% Tween 20, 10 mM CaCl2; 2) TBS, 0.05% Tween 20, 10 mM CaCl2, 25 mM maltose; 3) TBS, 0.05% Tween 20, 10 mM CaCl2, 30 μg/ml rough LPS; or 4) TBS, 0.05% Tween 20, and 10 mM EDTA and incubated overnight at 4°C. The binding of SP-D was measured with polyclonal P13 Ab, alkaline phosphatase conjugate, and para-nitrophenyl phosphate as described above in the bacteria binding assay.
Human samples
Blood samples for SP-D genotyping were obtained from 206 healthy Danish Caucasian blood donors. From 143 individuals, a corresponding serum sample was available for SP-D serum measurements. Selected serum samples (n = 28) were used for structural studies of the SP-D variants. Blood samples were obtained with informed consent, and the use of the blood samples was approved by the local ethics committee in Copenhagen and Frederiksberg (ref. no. 01-286/99). SP-D was purified from amniotic fluid obtained from Caesarean sections performed at 38–42 wk of gestation and made available from the Department of Obstetrics and Gynaecology, Odense University Hospital (Odense, Denmark).
Statistical analysis
Statistical comparisons were made with Student’s unpaired t test of logarithmic transformed values. Linkage disequilibrium between the different alleles was calculated using contingency tables analyses (2). Statistical significance was defined as p < 0.05. Two-sided tests were used throughout.
Results
Genotyping of allelic variants of SP-D
The distribution of the genotypes, allele frequencies, and the means and confidence intervals of SP-D serum concentrations are shown in Table I. Seventy-three (35.4%) of 206 Danes carried the genotype encoding Met/Met11, 97 (47.1%) had the genotype Met/Thr11, and 36 (17.5%) had the genotype Thr/Thr11. The alleles were distributed as 59% of the Met-encoding allele and 41% of the Thr-encoding allele. The logarithmic transformed average serum SP-D concentrations of the homozygous allelic variants of codon 11 (Met/Met and Thr/Thr) differed significantly (p = 0.015), demonstrating that individuals with the Met/Met11 variant had higher mean serum SP-D levels than individuals with the Thr/Thr11 variant. There were no significant differences between SP-D serum levels in the population with the variant genotypes of 160 and 270. The allele variants of amino acid residue 160 were distributed almost equally with frequencies of 44 and 56% for Thr and Ala, respectively. The Thr270 allele was relatively infrequent in this material with a frequency of only 4% compared with 96% for Ser270. The population homozygous for allele variants of aa 11 and 160 represented 38% of the total group. In this selected group, 40 (90%) of the individuals with the Met/Met11 variant were linked to the Thr/Thr160 variant, whereas 34 (100%) of the Thr/Thr11 variant were linked to the Ala/Ala160 variant. Likewise, all individuals containing the Ser/Thr270 variant also carried the Met/Met11 or Met/Thr11 variants and the Thr/Ala160 or Ala/Ala160 variants. The alleles at locus determining aa 11 and 160 were in strong linkage disequilibrium (p < 0.0001), as were the alleles at locus determining aa 11 and aa 270 (p = 0.0076), whereas a weaker association was found between the alleles at locus determining aa 160 and 270 (p = 0.07). This strongly indicates linkage between Met11, Thr160, and Ser270, and between Thr11, Ala160, and Thr270 alleles, respectively.
Table I. Distribution of genotypes, allele frequencies, and corresponding SP-D serum levelsa
Structural differences of SP-D allelic variants
Serum samples were separated by gel filtration chromatography, and SP-D was quantified in every fraction (Fig. 1). Serum from Met/Met11 individuals showed two structurally different forms of SP-D (SP-D high and SP-D low), whereas serum from Thr/Thr11 individuals displayed a marked predominance of one form that co-eluted with the lower m.w. form of Met/Met11 individuals. Integration of the area under the curves showed an average distribution of SP-D high to SP-D low of 1:1.6 for the Met/Met11 individuals and 1:5.1 for the Thr/Thr11 individuals. Recovery of the two SP-D forms from serum was too low in this experiment to allow for analysis by AFM. SP-D was therefore purified from amniotic fluid by maltose-agarose affinity chromatography followed by gel filtration chromatography, and the coeluting peaks were analyzed by AFM. The SP-D low form was predominantly composed of single subunits, whereas the SP-D high form displayed a mixture of single subunits, dodecamers, and multimers (Fig. 2). The single subunit formation observed in the SP-D high form preparation might be due to degradation taking place after the initial gel filtration chromatography. Chromatography of SP-D high after a storage period led to increasing amounts of SP-D low in the preparation (data not shown).
FIGURE 1. Gel filtration chromatography of serum followed by ELISA quantification of SP-D in all fractions. , Met/Met11 (average of n = 10); , Met/Thr11 (average of n = 10); , Thr/Thr11 (average of n = 8). Blue dextran (BD), 2000 kDa; thyroglobulin, 670 kDa; and IgG, 158 kDa.
FIGURE 2. AFM of the SP-D high and SP-D low forms isolated from amniotic fluid. The SP-D high form shows predominantly dodecamers and multimers (a and b), whereas the SP-D low form is characterized by single trimeric subunit structures (c). Image within a and c, 500-nm scans; b, 200-nm scan.
SDS-PAGE followed by Western blotting of the two forms of SP-D showed a band corresponding to the mass of 150 kDa under nonreduced conditions in both forms (Fig. 3). The SP-D high form further expressed higher oligomerized forms of SP-D of >250 kDa, whereas the SP-D low form expressed protein bands of 90, 43, and 40 kDa, indicating several structural differences of the two forms. In both cases, the reduced form of SP-D high and SP-D low showed protein bands of 46 and 43 kDa.
FIGURE 3. Western blotting of SP-D high (fraction 4) and SP-D low (fraction 7) forms from two different amniotic fluids (AF1 and AF2) separated by gel filtration chromatography and recognized by polyclonal anti SP-D Ab 477. Reduced and nonreduced samples are marked + and –, respectively.
SP-D binding to bacteria
The two variant forms of purified SP-D were tested for binding to clinical isolates of three Gram-negative and two Gram-positive bacterial strains in a solid-phase assay in the presence of calcium or EDTA (Fig. 4). Both forms of SP-D bound in a concentration-dependent manner to all the bacteria strains tested. At all concentrations, the SP-D high form showed a higher degree of binding to the bacteria than the SP-D low form, and the binding was dependent on the presence of calcium.
FIGURE 4. Solid-phase binding assay with formalin-fixed bacterial strains adhered to microtiter plates. Purified SP-D (?, SP-D high; , SP-D low) was 2-fold diluted from 2.5 μg/ml in the presence of calcium. The highest SP-D concentration of both forms was also diluted in washing buffer containing 10 mM EDTA. Note the different y-axes. The data presented are representative of four experiments.
SP-D binding to IAV
The two forms of SP-D were tested for binding to solid-phase IAV (Fig. 5). The SP-D high form bound significantly more to the virus than the SP-D low form at all concentrations tested.
FIGURE 5. Solid-phase binding assay with IAV adhered to microtiter plates. Purified SP-D (?, SP-D high; , SP-D low) was 2-fold diluted and bound to IAV. The data presented are representative of four experiments.
SP-D binding to mannan and LPS
To further study the binding properties of the different forms of SP-D, we examined their ability to bind to solid-phase microbial ligands, yeast mannan, and Gram-negative LPS (Fig. 6). A pool of amniotic fluid from six individuals was separated by gel filtration chromatography. The fractions were quantified for SP-D, and equal levels of SP-D high and SP-D low were present in the pool used for binding studies (Fig. 6a). The same fractions were analyzed for binding of SP-D to mannan (Fig. 6b). Both forms of SP-D bound to mannan. However, a preferential binding ofthe SP-D high form compared with the SP-D low form was seen. Thebinding of both SP-D forms could be inhibited by maltose tothe EDTA background, but residual binding not influenced by maltose or EDTA was observed in both fractions, most pronounced for the SP-D low form. The SP-D fractions were further analyzed for direct binding to solid-phase rough LPS from E. coli 026:B6. The SP-D low form showed significant binding to LPS, with the binding being influenced by calcium but not by maltose (Fig. 6c). No difference in the binding between SP-D high and LPS was observed in the presence of calcium, calcium and maltose, or EDTA. A similar binding pattern was observed when analyzing binding of SP-D oligomers to smooth LPS from E. coli 055:B5 (data not shown). Binding between SP-D high and solid-phase mannan was partly inhibited by LPS in a concentration-dependent manner, whereas the binding of SP-D low was inhibited to the EDTA background level (Fig. 6d). These data indicate that the binding between either of the two forms of SP-D and mannan or LPS is not exclusively dependent on calcium-mediated lectin activity of the CRD, and that the SP-D high form preferentially binds mannan, whereas the SP-D low form preferentially binds LPS.
FIGURE 6. Binding of SP-D to mannan and LPS. a, Quantification of fractions from a pool of amniotic fluid separated by gel filtration chromatography. The fractions were used as SP-D contributor (diluted 1/4) in the binding assays b–d; b, mannan coat/SP-D binding in the presence of calcium (), EDTA (–), or calcium with 75 mM maltose (?); c, LPS coat/SP-D binding in the presence of calcium (), EDTA (–), or with 75 mM maltose (?) in the presence of calcium; d, mannan coat/SP-D binding in the presence of calcium (), EDTA (–), or with preincubation with 30 μg/ml LPS (?) in the presence of calcium. The data presented are representative of four or more experiments.
Recombinant full-length SP-D
The Met11 and Thr11 variants were expressed in mammalian cells to determine whether differences in the proportions of oligomerized SP-D variants identified in serum reflect differences in the primary structure of the alleles. The variants were expressed and secreted as recombinant proteins by HEK cells grown in serum-free medium. The cell medium was separated by gel filtration chromatography, and SP-D was quantified in each fraction (Fig. 7). The recombinant forms of SP-D showed similar elution profiles as SP-D found in serum. Integration of the area under the curves showed an average distribution of SP-D high to SP-D low of 1:3.1 for the Met11 form and 1:10 for the Thr11 form. The ratios between the SP-D low form in the Met11 and Thr11 variants were 10:3.1 = 3.2 and were thus comparable to the ratios between the SP-D low form found in Met/Met and Thr/Thr variants in serum, which were 5.1:1.6 = 3.2. Western blotting of the two structural forms of recombinant SP-D showed identical electrophoretic mobility as the two forms purified from amniotic fluid, and the binding properties of recombinant SP-D to mannan were identical with binding properties with purified SP-D from amniotic fluid (data not shown).
FIGURE 7. Gel filtration chromatography of culture supernatant of HEK 293T-rex cells stably transfected with human cDNAs encoding the Met11 and Thr11 allelic variants. The secreted proteins in the individual fractions were quantified by ELISA. , Met11 recombinant SP-D; , Thr11 recombinant SP-D.
Discussion
Three single-nucleotide polymorphisms are identified in the coding region of the SP-D gene resulting in amino acid variations (20, 21). We have shown that the variation located at codon 11 within the sequence encoding the mature protein resulted in significantly different SP-D serum levels. We have previously estimated the relative genetic influence on serum SP-D levels in a twin study including children at the age of 6–9 years. The heritability factor was calculated to be 0.91, indicating that the SP-D serum levels in children predominantly is determined by genes, with a relatively small influence from environmental factors (33). The data presented in this study demonstrate that the variation of amino acid residue 11 partially accounts for the genetic influence on the serum SP-D level.
The Met11Thr variation also significantly influenced the oligomerization of serum SP-D. The Met/Met11 variant was composed of a low and a high m.w. variant when analyzed by gel filtration chromatography, whereas the Thr/Thr11 variant mainly was composed of the low m.w. variant. Recombinant full-length SP-D of the two allelic variants expressed in HEK cells showed a similar size distribution when analyzed by gel filtration chromatography. This observation emphasizes the importance of residue 11 on the oligomerization of human SP-D. AFM demonstrated that the SP-D low species isolated from amniotic fluid consisted of trimeric subunits, whereas trimeric, dodecameric, and multimeric forms were present in the SP-D high form. Notably, the latter forms were only represented in the blood of individuals with Met11 alleles. The individual polymorphisms of aa 160 or 270 had no detectable influence on the oligomeric state of SP-D observed on gel filtration chromatography, but based on linkage disequilibrium calculations, we cannot exclude an influence of these polymorphisms and perhaps promoter polymorphisms on the serum levels of SP-D. Methionine and threonine have very different hydrophobic properties, which might lead to differences in the molecular structures of the N-terminal peptide domain. Previous studies have further indicated the existence of various structural variants both with trimers, dodecamers, and high multimers, when SP-D was purified from proteinosis lung washings (34). As demonstrated in this study, SP-D from amniotic fluid shows similar complexity. Both SP-D high and SP-D low forms show a dominant SDS-PAGE protein band with the expected mobility of SP-D trimers. Although SP-D high included higher orders of disulfide-bonded aggregates, SP-D low exclusively contained components migrating with the expected mobility of disulfide-cross-linked dimers and monomers. We observed two monomers of 46 and 43 kDa. Mason et al. (35) identified a 50-kDa variant of SP-D, which could be identical with the 46-kDa band, and showed O-linked glycosylation of Thr11 herein. The 50-kDa variant was recovered as trimeric subunits, raising the possibility that differences in the glycosylation of residue 11, which is immediately N-terminal to Cys15, could influence the oligomerization of the subunits.
To characterize the potential functional consequences of variable oligomerization of SP-D, the corresponding oligomeric forms of SP-D were isolated from human amniotic fluid and high and low m.w. forms were resolved by gel filtration chromatography. SP-D high and SP-D low were used to study the interaction with intact microorganisms or with components of the microbial cell wall. One monoclonal and two polyclonal anti-SP-D Abs were tested for use in the functional studies mentioned below with identical outcome (data not shown). Each Ab bound equally well to the two SP-D forms in simple ligand-Ab binding studies. SP-D high clearly bound to adsorbed formaldehyde-fixed Gram-negative and Gram-positive bacterial strains and to IAV. There was a marked variation in the binding for the bacterial strains, but SP-D high showed calcium-dependent binding to the selected strains in a dose-dependent manner. For all strains analyzed, the same concentration of SP-D low resulted in significantly lower binding to the microorganisms. It is well known that LPS core saccharides of Gram-negative bacteria as well as mannan purified from S. cerevisiae are recognized by SP-D via the CRD (36, 37). Our data indicate that the SP-D high and SP-D low forms bind differentially to these components. The SP-D high form binds significantly better to mannan compared with the SP-D low form. The binding between SP-D high and mannan was inhibited with maltose to the EDTA level but only partially inhibited by competing LPS. In contrast, the SP-D low form bound significantly better to rough and smooth LPS compared with the SP-D high form. The binding between the SP-D low form and LPS was calcium dependent but could not be inhibited by maltose. Previous studies have shown that binding between rat SP-D and solid-phase rough LPS only partly were inhibited by mannose (38). These results suggest that LPS and mannan use different, but closely related, binding sites on the SP-D CRD.
Several studies indicate that the aggregating properties of SP-D are determined by valency, carbohydrate affinity, and the spatial arrangement of the CRDs. Recombinant trimeric single subunit of SP-D is a functional lectin with the same saccharide selectivity as higher ordered oligomers but appears to show a more restricted range of biological activities (14, 30, 39, 40, 41, 42, 43). Highly multimerized preparations of SP-D are significantly more potent than dodecamers in their ability to agglutinate different strains of IAV and inhibit IAV hemagglutination and infectivity, whereas trimeric CRDs induce minimal agglutination (12, 30, 31). In line with these results, we demonstrate that SP-D high binds more efficiently to IAV compared with SP-D low.
The functional role of the oligomeric forms of SP-D has further been studied in vivo. Transgenic mice that express a single arm mutant (RrSP-DSer15,20) were expressed in the respiratory epithelium of SP-D gene-targeted (SP-D–/–) mice (44, 45). The expression of RrSP-DSer15,20 partially restored antiviral activity butotherwise failed to rescue the deficient phenotype. This indicated that disulfide cross-linked SP-D oligomers are required for the regulation of surfactant phospholipid homeostasis and the prevention of emphysema and foamy macrophages in vivo.
The differential binding of SP-D high and SP-D low to intact microorganisms and LPS is intriguing and implies that SP-D high is important in the binding, aggregation, and clearance of microorganisms, whereas SP-D low by its preferential binding to simpler ligands like LPS may have alternative physiological functions. Previous studies have shown that SP-D binds to CD14 and modulates the LPS-CD14 interaction (38), and recently, it was shown that SP-D act in a dual manner to enhance or suppress inflammatory mediator production depending on binding orientation. SP-D, which is not complexed to ligands, binds SIRP through their globular heads to initiate a signaling pathway that blocks proinflammatory mediator production. In contrast SP-D binding to pathogen-associated molecular patterns via the CRDs, presents the aggregated collagenous tails to calreticulin/CD91, and stimulates phagocytosis and proinflammatory responses (10). We speculate that genetic differences in the proportions of trimeric subunits, and higher order multimers are associated with differential activation of anti-inflammatory and proinflammatory signaling pathways, thereby influencing the efficiency of clearance and inflammatory response to inhaled microorganisms and other Ags.
The high frequency of both allotypes in the Met11Thr polymorphism indicates that heterozygosity in general may be advantageous. Nevertheless, the different genotypes may predispose individuals to different infection or immune-related diseases.
Acknowledgments
We thank Vivi M?ller, Bente Frederiksen, and Vibeke Weirup for skilled technical assistance.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by grants from the Danish Medical Research Council and Novo Nordisk Research Foundation (to P.G. and H.O.M.), Marie Curie Fellowship, European Union (to H.J.), and National Heart, Lung, and Blood Institute Grant HL-44015 (to E.C.). U.H., J.M., and R.L.-L. were supported by the Danish Medical Research Council (Grant 9902278), The Novo Nordic Foundation, The Fifth (EC) Framework Program (contract QLK2000-00325), "Fonden til L?gevidenskabens Fremme," and the Benzon Foundation.
2 Address correspondence and reprint requests to Dr. Uffe Holmskov, Medical Biotechnology Center, Institute for Medical Biology, University of Southern Denmark, Winsloewparken 253, DK-5000 Odense, Denmark. E-mail address: uholmskov@health.sdu.dk
3 Abbreviations used in this paper: PRR, pattern recognition receptor; AFM, atomic force microscopy; CRD, carbohydrate recognition domain; SP-D, surfactant protein D; IAV, influenza A virus; HEK, human embryonic kidney.
Received for publication June 22, 2004. Accepted for publication September 29, 2004.
References
Janeway, C. A., Jr, R. Medzhitov. 2002. Innate immune recognition. Annu. Rev. Immunol. 20:197.
Holmskov, U., S. Thiel, J. C. Jensenius. 2003. Collectins and ficolins: humoral lectins of the innate immune defense. Annu. Rev. Immunol. 21:547.
Turner, M. W.. 2003. The role of mannose-binding lectin in health and disease. Mol. Immunol. 40:423.
McCormack, F. X., J. A. Whitsett. 2002. The pulmonary collectins, SP-A and SP-D, orchestrate innate immunity in the lung. J. Clin. Invest. 109:707.
Crouch, E., J. R. Wright. 2001. Surfactant proteins a and d and pulmonary host defense. Annu. Rev. Physiol. 63:521.
McCormack, F. X., R. Gibbons, S. R. Ward, A. Kuzmenko, H. Wu, G. S. Deepe, Jr. 2003. Macrophage-independent fungicidal action of the pulmonary collectins. J. Biol. Chem. 278:36250.
Wu, H., A. Kuzmenko, S. Wan, L. Schaffer, A. Weiss, J. H. Fisher, K. S. Kim, F. X. McCormack. 2003. Surfactant proteins A and D inhibit the growth of Gram-negative bacteria by increasing membrane permeability. J. Clin. Invest. 111:1589
LeVine, A. M., J. A. Whitsett, J. A. Gwozdz, T. R. Richardson, J. H. Fisher, M. S. Burhans, T. R. Korfhagen. 2000. Distinct effects of surfactant protein A or D deficiency during bacterial infection on the lung. J. Immunol. 165:3934
Wert, S. E., M. Yoshida, A. M. LeVine, M. Ikegami, T. Jones, G. F. Ross, J. H. Fisher, T. R. Korfhagen, J. A. Whitsett. 2000. Increased metalloproteinase activity, oxidant production, and emphysema in surfactant protein D gene-inactivated mice. Proc. Natl. Acad. Sci. USA 97:5972.
Gardai, S. J., Y. Q. Xiao, M. Dickinson, J. A. Nick, D. R. Voelker, K. E. Greene, P. M. Henson. 2003. By binding SIRP or calreticulin/CD91, lung collectins act as dual function surveillance molecules to suppress or enhance inflammation. Cell 115:13.
Hakansson, K., N. K. Lim, H. J. Hoppe, K. B. Reid. 1999. Crystal structure of the trimeric -helical coiled-coil and the three lectin domains of human lung surfactant protein D. Structure Fold Des. 7:255.
Hartshorn, K., D. Chang, K. Rust, M. White, J. Heuser, E. Crouch. 1996. Interactions of recombinant human pulmonary surfactant protein D and SP-D multimers with influenza A. Am. J. Physiol. 271:L753.
Crouch, E., D. Chang, K. Rust, A. Persson, J. Heuser. 1994. Recombinant pulmonary surfactant protein D: post-translational modification and molecular assembly. J. Biol. Chem. 269:15808.
Palaniyar, N., L. Zhang, A. Kuzmenko, M. Ikegami, S. Wan, H. Wu, T. R. Korfhagen, J. A. Whitsett, F. X. McCormack. 2002. The role of pulmonary collectin N-terminal domains in surfactant structure, function, and homeostasis in vivo. J. Biol. Chem. 277:26971.
Madsen, J., A. Kliem, I. Tornoe, K. Skjodt, C. Koch, U. Holmskov. 2000. Localization of lung surfactant protein D on mucosal surfaces in human tissues. J. Immunol. 164:5866.
Akiyama, J., A. Hoffman, C. Brown, L. Allen, J. Edmondson, F. Poulain, S. Hawgood. 2002. Tissue distribution of surfactant proteins A and D in the mouse. J. Histochem. Cytochem. 50:993.
Stahlman, M. T., M. E. Gray, W. M. Hull, J. A. Whitsett. 2002. Immunolocalization of surfactant protein-D (SP-D) in human fetal, newborn, and adult tissues. J. Histochem. Cytochem. 50:651.
Leth-Larsen, R., C. Nordenbaek, I. Tornoe, V. Moeller, A. Schlosser, C. Koch, B. Teisner, P. Junker, U. Holmskov. 2003. Surfactant protein D (SP-D) serum levels in patients with community-acquired pneumonia small star, filled. Clin. Immunol. 108:29.
Hermans, C., A. Bernard. 1999. Lung epithelium-specific proteins: characteristics and potential applications as markers. Am. J. Respir. Crit. Care Med. 159:646.
DiAngelo, S., Z. Lin, G. Wang, S. Phillips, M. Ramet, J. Luo, J. Floros. 1999. Novel, non-radioactive, simple and multiplex PCR-cRFLP methods for genotyping human SP-A and SP-D marker alleles. Dis. Markers 15:269.
Lahti, M., J. Lofgren, R. Marttila, M. Renko, T. Klaavuniemi, R. Haataja, M. Ramet, M. Hallman. 2002. Surfactant protein D gene polymorphism associated with severe respiratory syncytial virus infection. Pediatr. Res. 51:696.
Crouch, E., K. Rust, R. Veile, H. Donis-Keller, L. Grosso. 1993. Genomic organization of human surfactant protein D (SP-D): SP-D is encoded on chromosome 10q22.2–23.1. J. Biol. Chem. 268:2976.
Floros, J., H. M. Lin, A. Garcia, M. A. Salazar, X. Guo, S. DiAngelo, M. Montano, J. Luo, A. Pardo, M. Selman. 2000. Surfactant protein genetic marker alleles identify a subgroup of tuberculosis in a Mexican population. J. Infect. Dis. 182:1473.(Rikke Leth-Larsen, Peter )
Surfactant protein D (SP-D) plays important roles in the host defense against infectious microorganisms and in regulating the innate immune response to a variety of pathogen-associated molecular pattern. SP-D is mainly expressed by type II cells of the lung, but SP-D is generally found on epithelial surfaces and in serum. Genotyping for three single-nucleotide variations altering amino acids in the mature protein in codon 11 (Met11Thr), 160 (Ala160Thr), and 270 (Ser270Thr) of the SP-D gene was performed and related to the SP-D levels in serum. Individuals with the Thr/Thr11-encoding genotype had significantly lower SP-D serum levels than individuals with the Met/Met11 genotype. Gel filtration chromatography revealed two distinct m.w. peaks with SP-D immunoreactivity in serum from Met/Met11-encoding genotypes. In contrast, Thr/Thr11 genotypes lacked the highest m.w. form. A similar SP-D size distribution was found for recombinant Met11 and Thr11 expressed in human embryonic kidney cells. Atomic force microscopy of purified SP-D showed that components eluting in the position of the high m.w. peak consist of multimers, dodecamers, and monomers of subunits, whereas the second peak exclusively contains monomers. SP-D from both peaks bound to mannan-coated ELISA plates. SP-D from the high m.w. peak bound preferentially to intact influenza A virus and Gram-positive and Gram-negative bacteria, whereas the monomeric species preferentially bound to isolated LPS. Our data strongly suggest that polymorphic variation in the N-terminal domain of the SP-D molecule influences oligomerization, function, and the concentration of the molecule in serum.
Introduction
The innate immune system uses pattern recognition receptors (PRRs)3 to mark microorganisms for destruction either by direct lysis, by complement-induced lysis, or by phagocytosis. In response to binding of conserved pathogen-associated molecular patterns, various PRRs can either induce or inhibit the immune response (1). Surfactant protein D (SP-D) is a member of the protein family of collectins that also includes mannose-binding lectin and surfactant protein A (2, 3, 4). SP-D serves as a PRR by binding selectively to the surfaces of bacteria, viruses, and fungi, thereby enhancing phagocytosis and intracellular killing (2, 4, 5). In addition, SP-D was recently shown to directly inhibit growth of Gram-negative bacteria and fungi by increasing membrane permeability (6, 7). Mice deficient in SP-D accumulate alveolar surfactant phospholipids, exhibit enhanced acute inflammatory responses in the lung to a variety of stimuli (8), and develop spontaneous chronic inflammation, emphysema, and fibrosis (9). However, in vitro data have indicated additional proinflammatory effects of SP-D, and recent data suggested that SP-D, through binding of different receptors, acts in a dual manner to suppress or enhance inflammation depending on binding orientation of the SP-D molecule (10).
SP-D is composed of homotrimeric subunits with each polypeptide in the subunit consisting of an N-terminal domain with two conserved cysteines critical in cross-linking within and between subunits, a collagen-like region, an helical neck region, and a C-type lectin carbohydrate recognition domain (CRD) (11). The SP-D molecule is found as single subunits or as four subunits in a cruciform structure, but higher oligomers are also observed (12, 13). SP-D binds to carbohydrate and lipid ligands by the CRDs, but high affinity interactions require oligomeric assembly mediated by N-terminal cross-linking of trimeric arms (14).
SP-D is mainly synthesized in alveolar type II cells of the lung, but SP-D also localizes to epithelial cells of mucosa-associated tissue in the genitourinary and gastrointestinal tracts, and SP-D is present in amniotic fluid as well as in serum (15, 16, 17, 18). The origin of SP-D in the vascular compartment is not known, but it is generally assumed that SP-D is released from the lung into the bloodstream (19). This is consistent with the observation that the levels of SP-D in the blood increase in association with certain types of lung injury (18).
Three polymorphisms have been identified in the coding sequence of human SP-D: codons corresponding to amino acid residue 11 (Met11Thr), residue 160 (Ala160Thr), and residue 270 (Ser270Thr) (20, 21, 22) in the mature protein. Two clinical studies have associated the SP-D variants of amino acid 11 with disease. The SP-D allele coding for Met11 was associated with severe respiratory syncytial virus infection in infants, whereas Thr11 was suggested to increase susceptibility to tuberculosis (21, 23).
Polymorphisms of PRRs like mannose-binding lectin, TLR2, and TLR4 are associated with altered risk of a variety of disorders like autoimmune diseases, asthma, and atherosclerosis and to infections (3, 24, 25). The present work describes how known allelic variation in the coding region of SP-D influences the structure, the function, and the serum level of the protein, changes that may imply altered susceptibility to various infectious or immune-related diseases.
Materials and Methods
SP-D genotyping by PCR sequence-specific priming
SP-D single-nucleotide polymorphisms in the form of the structural variants named SP-D11/Met, SP-D11/Thr, SP-D160/Thr, SP-D160/Ala, SP-D270/Ser, and SP-D270/Thr were typed by PCR sequence-specific priming using the following six reactions: SP-D11/Met (T): 5'-primer, ACCTACTCCCACAGAACAAT; 3'-primer, GGAAGAAACACGTCTCCAGA; SP-D11/Thr (C): 5'-primer, ACCTACTCCCACAGAACAAC; 3'-primer, AGGAAGAAACACGTCTCCAGA; SP-D160/Thr (A): 5'-primer, TCTCTCTGACCCTAAAGTTGC; 3'-primer, CTCACCTGCTGCCCCTGT; SP-D160/Ala (G): 5'-primer, TCTCTCTGACCCTAAAGTTGC; 3'-primer, CTCACCTGCTGCCCCTGC; SP-D270/Ser (T): 5'-primer, CTGGTGGACAGTTGGCCT; 3'-primer, TCTGACCCGCCATCATCGTT; SP-D270/Thr (A): 5'-primer, CTGGTGGACAGTTGGCCA; 3'-primer, TCTGACCCGCCATCATCGTT. As internal positive control, we included a PCR covering exon 4 of the human mannose-binding lectin (mbl2) gene: 5'-primer: GAGTTTCACCCACTTTTTCACA; 3'-primer: GCCTGAGTGATATGACCCTTC. The PCR was performed essentially as described by Aldener-Cannava and Olerup (26).
Abs and ELISA
The rabbit anti-human SP-D Abs K477 (18) and P13 (27) were used in the SP-D quantification assay and in the solid-phase binding assays quantifying the amount of SP-D bound to bacteria, influenza A virus (IAV), LPS, and mannan, respectively. A mAb Hyb 246-4 was used for detection in the SP-D quantification assay (18). SP-D was quantified in serum and amniotic fluid by ELISA (18).
Purification of SP-D from amniotic fluid
A pool of amniotic fluid (n = 6) was centrifuged at 4000 rpm and 4°C for 30 min and purified by maltosyl agarose affinity chromatography (28). Purified SP-D was concentrated with a Vivaspin 6 (10,000 m.w. cut-off) concentrator (Vivascience), and the structurally different forms of SP-D were separated by gel filtration chromatography.
Gel filtration chromatography
Gel filtration chromatography was performed on 200-μl samples of normal human serum from genotyped individuals’ pooled amniotic fluid (n = 6), individual amniotic fluid, and purified amniotic fluid SP-D. The samples were applied to an analytical Superose 6 column connected to a fast-performance liquid chromatography system (Amersham Biosciences) using TBS (pH 7.4) containing 10 mM EDTA and 0.05% emulphogene as eluant at a flow rate of 24 ml/h. Fractions of 0.8 ml were collected and quantified by the SP-D ELISA. SP-D eluted as two structurally different forms and was collected in fractions 3 and 4 ("SP-D high") and 6 and 7 ("SP-D low"), respectively.
SDS-PAGE/Western blotting
Individual amniotic fluid containing both the SP-D high and the SP-D low form were separated by gel filtration chromatography, and the fractions were quantified for SP-D by ELISA. SP-D (500 ng) from fractions 4 and 7 were separated by SDS-PAGE gel electrophoresis in 4–20% (w/v) polyacrylamide gradient gels with a discontinuous buffer system and blotted onto polyvinylidene difluoride membranes (Immobilon P; Millipore) as described previously (18). K477 was used for detection.
Atomic force microscopy (AFM)
SP-D molecules were analyzed by AFM. For AFM imaging, 20-μl droplets of fractions of purified SP-D (10 μg/ml) were incubated on freshly cleaved mica (1 cm in diameter) for 10–15 min at room temperature. They were rinsed with deionized water (1 min) and dried with a flow of N2. The samples were imaged in Tapping Mode on Nanoscope IIIa (Digital Instruments) in air under ambient conditions (FM tip: Nanosencors; interaction force, 1 nN/m; scan size, 2 nμ; scan speed, 1 μm/s).
Expression of recombinant forms of SP-D in human embryonic kidney (HEK) 293T-rex cells using the "Flp-In" system
All reagents were purchased from Invitrogen Life Technologies unless otherwise stated. The Met11 and Thr11 forms of SP-D were amplified by RT-PCR and TA-cloned into the "Flp-In" pcDNA5.1/FRT/V5-His TOPO TA expression vector according to the manufacturer’s protocol. Shortly, first-strand synthesis was performed with the Thermoscript kit and oligo(dT) priming from 1 μg of a pool of human total RNA from the trachea (Clontech). Primers used in the following PCR had the sequence: HuSP-D leader f1(flp-in), 5'-GGG GGA TCC GCC ATG CTG CTC TTC CTC CTC-3', HuSP-D(stop) r1(flp-in), 5'-GGG TCT AGA TCA GAA CTC GCA GAC CAC AAG ACG A-3'. PCR was performed with the PFU polymerase (Stratagene) according to the manufacturer’s protocol, and A'-overhangs were added post-PCR. The fragments were TA cloned into the expression vector and constructs were verified by sequencing. An endotoxin-free preparation (12 μg) of the plasmid encoding the recombinase and 1.5 μg of the specific construct were transfected into 50% confluent Flp-In HEK 293T-rex cells in 80-cm2 culture flasks using the transfection reagent jetPEI according to manufacturer’s protocol (Qbiogene). The cells were grown in complete medium (DMEM containing 10% FCS, penicillin/streptomycin, and L-glutamine). After 2 days, the cells were transferred to medium containing 150 μg of hygromycin/ml of medium, and the medium was changed every third or fourth day until hygromycin-resistant cells became confluent. The cells were trypsinized and split into 150-cm2 culture flasks and allowed to attach in complete medium containing hygromycin. After 6 h, the medium was changed to DMEM only containing L-glutamine. The culture supernatant was harvested after 72 h at 37°C. The medium was then clarified by centrifugation at 10,000 x g for 10 min and stored at 4°C until further analysis.
Bacterial preparation
Clinical isolates of three Gram-negative (Escherichia coli, Klebsiella oxytoca, Haemophilus influenzae) and two Gram-positive (Staphylococcus aureus and Streptococcus mutans) bacterial strains were grown overnight at 37°C in serum broth followed by centrifugation (4000 rpm for 5 min). The bacteria were resuspended in formaldehyde buffer (0.9% NaCl, 0.5% formaldehyde) for 1 h with shaking, washed three times in TBS at pH 7.4, and resuspended in 0.05 M sodium carbonate buffer (pH 9.6) to the desired density determined spectrophotometrically at 660 nm.
Binding of SP-D to bacteria
All bacterial strains were suspended at a density of OD660 0.7 and adhered to Polysorp microtiter plates (Nunc) by overnight incubation at 4°C. The plates were blocked with TBS and 0.05% Tween 20 containing 10 mM CaCl2 for 30 min. The same buffer was used as washing buffer. All incubations were performed at room temperature with shaking unless otherwise stated. Two-fold dilutions of the two variant forms of SP-D (SP-D high and SP-D low) were added to the plate diluted in TBS, 0.05% Tween 20 containing 10 mM CaCl2 or 10 mM EDTA, and incubated overnight at 4°C. The binding between bacteria and SP-D was measured using SP-D polyclonal P13 Ab (1:5000, 2 h) followed by incubation with alkaline phosphatase goat anti-rabbit Ig conjugate (Sigma-Aldrich) diluted 1/2000 for 1 h, and para-nitrophenyl phosphate (1 mg/ml). Binding was determined spectrophotometrically at 405 nm. The experiments were performed in duplicate and repeated three times.
Binding of SP-D to IAV
Phillipine 82/H3N2 strain IAV was grown in the chorioallantoic fluid of 10-day-old chicken eggs, purified on a discontinuous sucrose gradient, stored, and characterized as previously described (29). Binding of purified SP-D variants to IAV was tested using an ELISA as previously described using polyclonal P13 Ab for detection of SP-D binding (12, 30, 31).
Binding of SP-D to mannan and LPS
Maxisorp microtiter plates (Nunc) were coated with 10 μg/ml mannan purified from Saccharomyces cerevisiae (32), rough LPS (E. coli O26:B6; Sigma-Aldrich) or smooth LPS (E. coli O55:B5; Sigma-Aldrich) in sodium carbonate buffer overnight at 4°C. After wash in washing buffer, the plates were incubated overnight at 4°C with fractions from a gel filtration chromatography of pooled amniotic fluid (n = 6). Before addition to the plates, the fractions were diluted 1/4 in 1) TBS, 0.05% Tween 20, 10 mM CaCl2; 2) TBS, 0.05% Tween 20, 10 mM CaCl2, 25 mM maltose; 3) TBS, 0.05% Tween 20, 10 mM CaCl2, 30 μg/ml rough LPS; or 4) TBS, 0.05% Tween 20, and 10 mM EDTA and incubated overnight at 4°C. The binding of SP-D was measured with polyclonal P13 Ab, alkaline phosphatase conjugate, and para-nitrophenyl phosphate as described above in the bacteria binding assay.
Human samples
Blood samples for SP-D genotyping were obtained from 206 healthy Danish Caucasian blood donors. From 143 individuals, a corresponding serum sample was available for SP-D serum measurements. Selected serum samples (n = 28) were used for structural studies of the SP-D variants. Blood samples were obtained with informed consent, and the use of the blood samples was approved by the local ethics committee in Copenhagen and Frederiksberg (ref. no. 01-286/99). SP-D was purified from amniotic fluid obtained from Caesarean sections performed at 38–42 wk of gestation and made available from the Department of Obstetrics and Gynaecology, Odense University Hospital (Odense, Denmark).
Statistical analysis
Statistical comparisons were made with Student’s unpaired t test of logarithmic transformed values. Linkage disequilibrium between the different alleles was calculated using contingency tables analyses (2). Statistical significance was defined as p < 0.05. Two-sided tests were used throughout.
Results
Genotyping of allelic variants of SP-D
The distribution of the genotypes, allele frequencies, and the means and confidence intervals of SP-D serum concentrations are shown in Table I. Seventy-three (35.4%) of 206 Danes carried the genotype encoding Met/Met11, 97 (47.1%) had the genotype Met/Thr11, and 36 (17.5%) had the genotype Thr/Thr11. The alleles were distributed as 59% of the Met-encoding allele and 41% of the Thr-encoding allele. The logarithmic transformed average serum SP-D concentrations of the homozygous allelic variants of codon 11 (Met/Met and Thr/Thr) differed significantly (p = 0.015), demonstrating that individuals with the Met/Met11 variant had higher mean serum SP-D levels than individuals with the Thr/Thr11 variant. There were no significant differences between SP-D serum levels in the population with the variant genotypes of 160 and 270. The allele variants of amino acid residue 160 were distributed almost equally with frequencies of 44 and 56% for Thr and Ala, respectively. The Thr270 allele was relatively infrequent in this material with a frequency of only 4% compared with 96% for Ser270. The population homozygous for allele variants of aa 11 and 160 represented 38% of the total group. In this selected group, 40 (90%) of the individuals with the Met/Met11 variant were linked to the Thr/Thr160 variant, whereas 34 (100%) of the Thr/Thr11 variant were linked to the Ala/Ala160 variant. Likewise, all individuals containing the Ser/Thr270 variant also carried the Met/Met11 or Met/Thr11 variants and the Thr/Ala160 or Ala/Ala160 variants. The alleles at locus determining aa 11 and 160 were in strong linkage disequilibrium (p < 0.0001), as were the alleles at locus determining aa 11 and aa 270 (p = 0.0076), whereas a weaker association was found between the alleles at locus determining aa 160 and 270 (p = 0.07). This strongly indicates linkage between Met11, Thr160, and Ser270, and between Thr11, Ala160, and Thr270 alleles, respectively.
Table I. Distribution of genotypes, allele frequencies, and corresponding SP-D serum levelsa
Structural differences of SP-D allelic variants
Serum samples were separated by gel filtration chromatography, and SP-D was quantified in every fraction (Fig. 1). Serum from Met/Met11 individuals showed two structurally different forms of SP-D (SP-D high and SP-D low), whereas serum from Thr/Thr11 individuals displayed a marked predominance of one form that co-eluted with the lower m.w. form of Met/Met11 individuals. Integration of the area under the curves showed an average distribution of SP-D high to SP-D low of 1:1.6 for the Met/Met11 individuals and 1:5.1 for the Thr/Thr11 individuals. Recovery of the two SP-D forms from serum was too low in this experiment to allow for analysis by AFM. SP-D was therefore purified from amniotic fluid by maltose-agarose affinity chromatography followed by gel filtration chromatography, and the coeluting peaks were analyzed by AFM. The SP-D low form was predominantly composed of single subunits, whereas the SP-D high form displayed a mixture of single subunits, dodecamers, and multimers (Fig. 2). The single subunit formation observed in the SP-D high form preparation might be due to degradation taking place after the initial gel filtration chromatography. Chromatography of SP-D high after a storage period led to increasing amounts of SP-D low in the preparation (data not shown).
FIGURE 1. Gel filtration chromatography of serum followed by ELISA quantification of SP-D in all fractions. , Met/Met11 (average of n = 10); , Met/Thr11 (average of n = 10); , Thr/Thr11 (average of n = 8). Blue dextran (BD), 2000 kDa; thyroglobulin, 670 kDa; and IgG, 158 kDa.
FIGURE 2. AFM of the SP-D high and SP-D low forms isolated from amniotic fluid. The SP-D high form shows predominantly dodecamers and multimers (a and b), whereas the SP-D low form is characterized by single trimeric subunit structures (c). Image within a and c, 500-nm scans; b, 200-nm scan.
SDS-PAGE followed by Western blotting of the two forms of SP-D showed a band corresponding to the mass of 150 kDa under nonreduced conditions in both forms (Fig. 3). The SP-D high form further expressed higher oligomerized forms of SP-D of >250 kDa, whereas the SP-D low form expressed protein bands of 90, 43, and 40 kDa, indicating several structural differences of the two forms. In both cases, the reduced form of SP-D high and SP-D low showed protein bands of 46 and 43 kDa.
FIGURE 3. Western blotting of SP-D high (fraction 4) and SP-D low (fraction 7) forms from two different amniotic fluids (AF1 and AF2) separated by gel filtration chromatography and recognized by polyclonal anti SP-D Ab 477. Reduced and nonreduced samples are marked + and –, respectively.
SP-D binding to bacteria
The two variant forms of purified SP-D were tested for binding to clinical isolates of three Gram-negative and two Gram-positive bacterial strains in a solid-phase assay in the presence of calcium or EDTA (Fig. 4). Both forms of SP-D bound in a concentration-dependent manner to all the bacteria strains tested. At all concentrations, the SP-D high form showed a higher degree of binding to the bacteria than the SP-D low form, and the binding was dependent on the presence of calcium.
FIGURE 4. Solid-phase binding assay with formalin-fixed bacterial strains adhered to microtiter plates. Purified SP-D (?, SP-D high; , SP-D low) was 2-fold diluted from 2.5 μg/ml in the presence of calcium. The highest SP-D concentration of both forms was also diluted in washing buffer containing 10 mM EDTA. Note the different y-axes. The data presented are representative of four experiments.
SP-D binding to IAV
The two forms of SP-D were tested for binding to solid-phase IAV (Fig. 5). The SP-D high form bound significantly more to the virus than the SP-D low form at all concentrations tested.
FIGURE 5. Solid-phase binding assay with IAV adhered to microtiter plates. Purified SP-D (?, SP-D high; , SP-D low) was 2-fold diluted and bound to IAV. The data presented are representative of four experiments.
SP-D binding to mannan and LPS
To further study the binding properties of the different forms of SP-D, we examined their ability to bind to solid-phase microbial ligands, yeast mannan, and Gram-negative LPS (Fig. 6). A pool of amniotic fluid from six individuals was separated by gel filtration chromatography. The fractions were quantified for SP-D, and equal levels of SP-D high and SP-D low were present in the pool used for binding studies (Fig. 6a). The same fractions were analyzed for binding of SP-D to mannan (Fig. 6b). Both forms of SP-D bound to mannan. However, a preferential binding ofthe SP-D high form compared with the SP-D low form was seen. Thebinding of both SP-D forms could be inhibited by maltose tothe EDTA background, but residual binding not influenced by maltose or EDTA was observed in both fractions, most pronounced for the SP-D low form. The SP-D fractions were further analyzed for direct binding to solid-phase rough LPS from E. coli 026:B6. The SP-D low form showed significant binding to LPS, with the binding being influenced by calcium but not by maltose (Fig. 6c). No difference in the binding between SP-D high and LPS was observed in the presence of calcium, calcium and maltose, or EDTA. A similar binding pattern was observed when analyzing binding of SP-D oligomers to smooth LPS from E. coli 055:B5 (data not shown). Binding between SP-D high and solid-phase mannan was partly inhibited by LPS in a concentration-dependent manner, whereas the binding of SP-D low was inhibited to the EDTA background level (Fig. 6d). These data indicate that the binding between either of the two forms of SP-D and mannan or LPS is not exclusively dependent on calcium-mediated lectin activity of the CRD, and that the SP-D high form preferentially binds mannan, whereas the SP-D low form preferentially binds LPS.
FIGURE 6. Binding of SP-D to mannan and LPS. a, Quantification of fractions from a pool of amniotic fluid separated by gel filtration chromatography. The fractions were used as SP-D contributor (diluted 1/4) in the binding assays b–d; b, mannan coat/SP-D binding in the presence of calcium (), EDTA (–), or calcium with 75 mM maltose (?); c, LPS coat/SP-D binding in the presence of calcium (), EDTA (–), or with 75 mM maltose (?) in the presence of calcium; d, mannan coat/SP-D binding in the presence of calcium (), EDTA (–), or with preincubation with 30 μg/ml LPS (?) in the presence of calcium. The data presented are representative of four or more experiments.
Recombinant full-length SP-D
The Met11 and Thr11 variants were expressed in mammalian cells to determine whether differences in the proportions of oligomerized SP-D variants identified in serum reflect differences in the primary structure of the alleles. The variants were expressed and secreted as recombinant proteins by HEK cells grown in serum-free medium. The cell medium was separated by gel filtration chromatography, and SP-D was quantified in each fraction (Fig. 7). The recombinant forms of SP-D showed similar elution profiles as SP-D found in serum. Integration of the area under the curves showed an average distribution of SP-D high to SP-D low of 1:3.1 for the Met11 form and 1:10 for the Thr11 form. The ratios between the SP-D low form in the Met11 and Thr11 variants were 10:3.1 = 3.2 and were thus comparable to the ratios between the SP-D low form found in Met/Met and Thr/Thr variants in serum, which were 5.1:1.6 = 3.2. Western blotting of the two structural forms of recombinant SP-D showed identical electrophoretic mobility as the two forms purified from amniotic fluid, and the binding properties of recombinant SP-D to mannan were identical with binding properties with purified SP-D from amniotic fluid (data not shown).
FIGURE 7. Gel filtration chromatography of culture supernatant of HEK 293T-rex cells stably transfected with human cDNAs encoding the Met11 and Thr11 allelic variants. The secreted proteins in the individual fractions were quantified by ELISA. , Met11 recombinant SP-D; , Thr11 recombinant SP-D.
Discussion
Three single-nucleotide polymorphisms are identified in the coding region of the SP-D gene resulting in amino acid variations (20, 21). We have shown that the variation located at codon 11 within the sequence encoding the mature protein resulted in significantly different SP-D serum levels. We have previously estimated the relative genetic influence on serum SP-D levels in a twin study including children at the age of 6–9 years. The heritability factor was calculated to be 0.91, indicating that the SP-D serum levels in children predominantly is determined by genes, with a relatively small influence from environmental factors (33). The data presented in this study demonstrate that the variation of amino acid residue 11 partially accounts for the genetic influence on the serum SP-D level.
The Met11Thr variation also significantly influenced the oligomerization of serum SP-D. The Met/Met11 variant was composed of a low and a high m.w. variant when analyzed by gel filtration chromatography, whereas the Thr/Thr11 variant mainly was composed of the low m.w. variant. Recombinant full-length SP-D of the two allelic variants expressed in HEK cells showed a similar size distribution when analyzed by gel filtration chromatography. This observation emphasizes the importance of residue 11 on the oligomerization of human SP-D. AFM demonstrated that the SP-D low species isolated from amniotic fluid consisted of trimeric subunits, whereas trimeric, dodecameric, and multimeric forms were present in the SP-D high form. Notably, the latter forms were only represented in the blood of individuals with Met11 alleles. The individual polymorphisms of aa 160 or 270 had no detectable influence on the oligomeric state of SP-D observed on gel filtration chromatography, but based on linkage disequilibrium calculations, we cannot exclude an influence of these polymorphisms and perhaps promoter polymorphisms on the serum levels of SP-D. Methionine and threonine have very different hydrophobic properties, which might lead to differences in the molecular structures of the N-terminal peptide domain. Previous studies have further indicated the existence of various structural variants both with trimers, dodecamers, and high multimers, when SP-D was purified from proteinosis lung washings (34). As demonstrated in this study, SP-D from amniotic fluid shows similar complexity. Both SP-D high and SP-D low forms show a dominant SDS-PAGE protein band with the expected mobility of SP-D trimers. Although SP-D high included higher orders of disulfide-bonded aggregates, SP-D low exclusively contained components migrating with the expected mobility of disulfide-cross-linked dimers and monomers. We observed two monomers of 46 and 43 kDa. Mason et al. (35) identified a 50-kDa variant of SP-D, which could be identical with the 46-kDa band, and showed O-linked glycosylation of Thr11 herein. The 50-kDa variant was recovered as trimeric subunits, raising the possibility that differences in the glycosylation of residue 11, which is immediately N-terminal to Cys15, could influence the oligomerization of the subunits.
To characterize the potential functional consequences of variable oligomerization of SP-D, the corresponding oligomeric forms of SP-D were isolated from human amniotic fluid and high and low m.w. forms were resolved by gel filtration chromatography. SP-D high and SP-D low were used to study the interaction with intact microorganisms or with components of the microbial cell wall. One monoclonal and two polyclonal anti-SP-D Abs were tested for use in the functional studies mentioned below with identical outcome (data not shown). Each Ab bound equally well to the two SP-D forms in simple ligand-Ab binding studies. SP-D high clearly bound to adsorbed formaldehyde-fixed Gram-negative and Gram-positive bacterial strains and to IAV. There was a marked variation in the binding for the bacterial strains, but SP-D high showed calcium-dependent binding to the selected strains in a dose-dependent manner. For all strains analyzed, the same concentration of SP-D low resulted in significantly lower binding to the microorganisms. It is well known that LPS core saccharides of Gram-negative bacteria as well as mannan purified from S. cerevisiae are recognized by SP-D via the CRD (36, 37). Our data indicate that the SP-D high and SP-D low forms bind differentially to these components. The SP-D high form binds significantly better to mannan compared with the SP-D low form. The binding between SP-D high and mannan was inhibited with maltose to the EDTA level but only partially inhibited by competing LPS. In contrast, the SP-D low form bound significantly better to rough and smooth LPS compared with the SP-D high form. The binding between the SP-D low form and LPS was calcium dependent but could not be inhibited by maltose. Previous studies have shown that binding between rat SP-D and solid-phase rough LPS only partly were inhibited by mannose (38). These results suggest that LPS and mannan use different, but closely related, binding sites on the SP-D CRD.
Several studies indicate that the aggregating properties of SP-D are determined by valency, carbohydrate affinity, and the spatial arrangement of the CRDs. Recombinant trimeric single subunit of SP-D is a functional lectin with the same saccharide selectivity as higher ordered oligomers but appears to show a more restricted range of biological activities (14, 30, 39, 40, 41, 42, 43). Highly multimerized preparations of SP-D are significantly more potent than dodecamers in their ability to agglutinate different strains of IAV and inhibit IAV hemagglutination and infectivity, whereas trimeric CRDs induce minimal agglutination (12, 30, 31). In line with these results, we demonstrate that SP-D high binds more efficiently to IAV compared with SP-D low.
The functional role of the oligomeric forms of SP-D has further been studied in vivo. Transgenic mice that express a single arm mutant (RrSP-DSer15,20) were expressed in the respiratory epithelium of SP-D gene-targeted (SP-D–/–) mice (44, 45). The expression of RrSP-DSer15,20 partially restored antiviral activity butotherwise failed to rescue the deficient phenotype. This indicated that disulfide cross-linked SP-D oligomers are required for the regulation of surfactant phospholipid homeostasis and the prevention of emphysema and foamy macrophages in vivo.
The differential binding of SP-D high and SP-D low to intact microorganisms and LPS is intriguing and implies that SP-D high is important in the binding, aggregation, and clearance of microorganisms, whereas SP-D low by its preferential binding to simpler ligands like LPS may have alternative physiological functions. Previous studies have shown that SP-D binds to CD14 and modulates the LPS-CD14 interaction (38), and recently, it was shown that SP-D act in a dual manner to enhance or suppress inflammatory mediator production depending on binding orientation. SP-D, which is not complexed to ligands, binds SIRP through their globular heads to initiate a signaling pathway that blocks proinflammatory mediator production. In contrast SP-D binding to pathogen-associated molecular patterns via the CRDs, presents the aggregated collagenous tails to calreticulin/CD91, and stimulates phagocytosis and proinflammatory responses (10). We speculate that genetic differences in the proportions of trimeric subunits, and higher order multimers are associated with differential activation of anti-inflammatory and proinflammatory signaling pathways, thereby influencing the efficiency of clearance and inflammatory response to inhaled microorganisms and other Ags.
The high frequency of both allotypes in the Met11Thr polymorphism indicates that heterozygosity in general may be advantageous. Nevertheless, the different genotypes may predispose individuals to different infection or immune-related diseases.
Acknowledgments
We thank Vivi M?ller, Bente Frederiksen, and Vibeke Weirup for skilled technical assistance.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by grants from the Danish Medical Research Council and Novo Nordisk Research Foundation (to P.G. and H.O.M.), Marie Curie Fellowship, European Union (to H.J.), and National Heart, Lung, and Blood Institute Grant HL-44015 (to E.C.). U.H., J.M., and R.L.-L. were supported by the Danish Medical Research Council (Grant 9902278), The Novo Nordic Foundation, The Fifth (EC) Framework Program (contract QLK2000-00325), "Fonden til L?gevidenskabens Fremme," and the Benzon Foundation.
2 Address correspondence and reprint requests to Dr. Uffe Holmskov, Medical Biotechnology Center, Institute for Medical Biology, University of Southern Denmark, Winsloewparken 253, DK-5000 Odense, Denmark. E-mail address: uholmskov@health.sdu.dk
3 Abbreviations used in this paper: PRR, pattern recognition receptor; AFM, atomic force microscopy; CRD, carbohydrate recognition domain; SP-D, surfactant protein D; IAV, influenza A virus; HEK, human embryonic kidney.
Received for publication June 22, 2004. Accepted for publication September 29, 2004.
References
Janeway, C. A., Jr, R. Medzhitov. 2002. Innate immune recognition. Annu. Rev. Immunol. 20:197.
Holmskov, U., S. Thiel, J. C. Jensenius. 2003. Collectins and ficolins: humoral lectins of the innate immune defense. Annu. Rev. Immunol. 21:547.
Turner, M. W.. 2003. The role of mannose-binding lectin in health and disease. Mol. Immunol. 40:423.
McCormack, F. X., J. A. Whitsett. 2002. The pulmonary collectins, SP-A and SP-D, orchestrate innate immunity in the lung. J. Clin. Invest. 109:707.
Crouch, E., J. R. Wright. 2001. Surfactant proteins a and d and pulmonary host defense. Annu. Rev. Physiol. 63:521.
McCormack, F. X., R. Gibbons, S. R. Ward, A. Kuzmenko, H. Wu, G. S. Deepe, Jr. 2003. Macrophage-independent fungicidal action of the pulmonary collectins. J. Biol. Chem. 278:36250.
Wu, H., A. Kuzmenko, S. Wan, L. Schaffer, A. Weiss, J. H. Fisher, K. S. Kim, F. X. McCormack. 2003. Surfactant proteins A and D inhibit the growth of Gram-negative bacteria by increasing membrane permeability. J. Clin. Invest. 111:1589
LeVine, A. M., J. A. Whitsett, J. A. Gwozdz, T. R. Richardson, J. H. Fisher, M. S. Burhans, T. R. Korfhagen. 2000. Distinct effects of surfactant protein A or D deficiency during bacterial infection on the lung. J. Immunol. 165:3934
Wert, S. E., M. Yoshida, A. M. LeVine, M. Ikegami, T. Jones, G. F. Ross, J. H. Fisher, T. R. Korfhagen, J. A. Whitsett. 2000. Increased metalloproteinase activity, oxidant production, and emphysema in surfactant protein D gene-inactivated mice. Proc. Natl. Acad. Sci. USA 97:5972.
Gardai, S. J., Y. Q. Xiao, M. Dickinson, J. A. Nick, D. R. Voelker, K. E. Greene, P. M. Henson. 2003. By binding SIRP or calreticulin/CD91, lung collectins act as dual function surveillance molecules to suppress or enhance inflammation. Cell 115:13.
Hakansson, K., N. K. Lim, H. J. Hoppe, K. B. Reid. 1999. Crystal structure of the trimeric -helical coiled-coil and the three lectin domains of human lung surfactant protein D. Structure Fold Des. 7:255.
Hartshorn, K., D. Chang, K. Rust, M. White, J. Heuser, E. Crouch. 1996. Interactions of recombinant human pulmonary surfactant protein D and SP-D multimers with influenza A. Am. J. Physiol. 271:L753.
Crouch, E., D. Chang, K. Rust, A. Persson, J. Heuser. 1994. Recombinant pulmonary surfactant protein D: post-translational modification and molecular assembly. J. Biol. Chem. 269:15808.
Palaniyar, N., L. Zhang, A. Kuzmenko, M. Ikegami, S. Wan, H. Wu, T. R. Korfhagen, J. A. Whitsett, F. X. McCormack. 2002. The role of pulmonary collectin N-terminal domains in surfactant structure, function, and homeostasis in vivo. J. Biol. Chem. 277:26971.
Madsen, J., A. Kliem, I. Tornoe, K. Skjodt, C. Koch, U. Holmskov. 2000. Localization of lung surfactant protein D on mucosal surfaces in human tissues. J. Immunol. 164:5866.
Akiyama, J., A. Hoffman, C. Brown, L. Allen, J. Edmondson, F. Poulain, S. Hawgood. 2002. Tissue distribution of surfactant proteins A and D in the mouse. J. Histochem. Cytochem. 50:993.
Stahlman, M. T., M. E. Gray, W. M. Hull, J. A. Whitsett. 2002. Immunolocalization of surfactant protein-D (SP-D) in human fetal, newborn, and adult tissues. J. Histochem. Cytochem. 50:651.
Leth-Larsen, R., C. Nordenbaek, I. Tornoe, V. Moeller, A. Schlosser, C. Koch, B. Teisner, P. Junker, U. Holmskov. 2003. Surfactant protein D (SP-D) serum levels in patients with community-acquired pneumonia small star, filled. Clin. Immunol. 108:29.
Hermans, C., A. Bernard. 1999. Lung epithelium-specific proteins: characteristics and potential applications as markers. Am. J. Respir. Crit. Care Med. 159:646.
DiAngelo, S., Z. Lin, G. Wang, S. Phillips, M. Ramet, J. Luo, J. Floros. 1999. Novel, non-radioactive, simple and multiplex PCR-cRFLP methods for genotyping human SP-A and SP-D marker alleles. Dis. Markers 15:269.
Lahti, M., J. Lofgren, R. Marttila, M. Renko, T. Klaavuniemi, R. Haataja, M. Ramet, M. Hallman. 2002. Surfactant protein D gene polymorphism associated with severe respiratory syncytial virus infection. Pediatr. Res. 51:696.
Crouch, E., K. Rust, R. Veile, H. Donis-Keller, L. Grosso. 1993. Genomic organization of human surfactant protein D (SP-D): SP-D is encoded on chromosome 10q22.2–23.1. J. Biol. Chem. 268:2976.
Floros, J., H. M. Lin, A. Garcia, M. A. Salazar, X. Guo, S. DiAngelo, M. Montano, J. Luo, A. Pardo, M. Selman. 2000. Surfactant protein genetic marker alleles identify a subgroup of tuberculosis in a Mexican population. J. Infect. Dis. 182:1473.(Rikke Leth-Larsen, Peter )