A Common Single Nucleotide Polymorphism in the CD14 Promoter Decreases the Affinity of Sp Protein Binding and Enhances Transcriptional Activity¹

The CD14 pattern recognition receptor has a unique ability to discriminate nonself lipoglycans of infectious pathogens—first and foremost, LPS—from noninfectious self (1, 2, 3). At the molecular level, CD14 acts by transferring LPS and other bacterial ligands from circulating LPS-binding protein to the Toll-like receptor 4/MD-2 signaling complex (4). Engagement of this complex results in the activation of innate host defense mechanisms such as release of inflammatory cytokines, and in up-regulation of costimulatory molecules, thus providing cues that are essential to directing adaptive immune responses (3, 5).

Although CD14 exists as a single-copy gene, CD14 protein is found in two distinct forms: a 50- to 55-kDa glycosylphosphatidylinositol-anchored membrane molecule, membrane CD14 (mCD14),³ expressed primarily on the surface of monocyte/macrophages and neutrophils (6), and a soluble form lacking the glycosylphosphatidylinositol anchor (7). Soluble CD14 (sCD14) appears to derive from monocytes (8) as well as the liver (9, 10) and is found in normal serum at microgram concentrations (7). Both mCD14 and sCD14 are critical for LPS-dependent signal transduction. LPS-responsive cells that lack mCD14, such as endothelial cells, epithelial cells, and astrocytes, become sensitive to low concentrations of LPS in the presence of sCD14 (11).

Regulation of CD14 gene expression appears to be important in several disease states. Increased serum levels of sCD14 are associated with high mortality in Gram-negative septic shock (12). Consistent with these findings, CD14 knockout mice were found to be at least 10-fold more resistant to LPS-induced shock than wild-type controls (13). In addition, inflammatory infectious diseases such as atopic dermatitis (14), HIV infection (15), and malaria (16), as well as extensive tissue damage in polytrauma and severe burns (17), are marked by elevated sCD14 in the circulation.

This compelling epidemiologic evidence led us to investigate whether CD14 expression is influenced by genetic variation. Recently, our group has identified a single nucleotide polymorphism (SNP) in the proximal CD14 promoter at position −159 (18). CD14/−159C→T is very frequent among both Hispanic and non-Hispanic white populations, with approximately one-half of all chromosomes carrying the C allele and one-half carrying the T allele. Interestingly, homozygous carriers of the T allele have a significant increase in serum levels of sCD14 and a concomitant decrease in total serum IgE (18), suggesting that CD14 may play a role in the regulation of IgE synthesis and IgE-mediated diseases such as allergy and asthma. In contrast, the association between CD14/−159C→T and risk for myocardial infarction recently described in at least three different populations (19, 20, 21) eloquently highlights the far-reaching effects that genetic variation in CD14 may have on the pathogenesis of cardiovascular diseases. In this context, the role of CD14 in inflammation and/or its ability to bind and transport lipids, including cholesterol (22), may be particularly relevant. Altogether, these findings suggest that CD14/−159 may affect the regulation of CD14 gene expression, thus modulating the impact that CD14-mediated events have on innate and adaptive immune responses (23).

These findings warranted studies aimed at providing a molecular basis for the effects of CD14/−159C→T on the regulation of CD14 gene expression. We developed a reporter system to assess the transcriptional activity of the two allelic variants of the CD14 proximal promoter. We found a monocyte-specific increase in transcription from the T allele that was associated with a decreased affinity in DNA/Sp protein interactions at a GC box that includes the polymorphic nucleotide.

The human monocytic Mono Mac 6 cell line was generously provided by Dr. H. W. Ziegler-Heitbrock (University of Munich, Munich, Germany) and was cultured as described (24). Human hepatocytic HepG2 cells (HB8065) were obtained from the American Type Culture Collection (Manassas, VA). Primary human monocytes were isolated by countercurrent elutriation from normal volunteers as previously described (25). All cell culture reagents were endotoxin free as determined by the Limulus amebocyte lysate assay (limits of detection 3–6 pg/ml; BioWhittaker, Walkersville, MD).

Fig. 1,A shows the sequence of the CD14 5′-flanking region that contains the proximal promoter (26, 27) and CD14/−159C→T (Fig 1 A, bold and underlined). The −159C/luc and −159T/luc reporter constructs contain a 304-bp insert (position −232/+72; GenBank accession no. U00699) and were created by PCR amplification of genomic DNA from individuals homozygous for either C or T at position −159 using as primers 5′-GTGCCAACAGATGAGGTTCA (position −513/−493) and 5′-CGCAGCGGAAATCTTCATC (position +274/+292). The PCR products were digested with NarI/SacI, shuttled through pBluescript KS⁺ (Stratagene, La Jolla, CA) via ClaI/SacI, and then directionally cloned into the KpnI and SacI sites of the promoterless, enhancerless luciferase reporter plasmid, pGL3-Basic (Promega, Madison, WI). Constructs were verified by sequence analysis and were prepared for transfection using the EndoFree Plasmid Mega kit (Qiagen, Valencia, CA).

Mono Mac 6 cells (5 × 10⁶ cells/500 μl of serum-free RPMI 1640) were transiently transfected with −159C/luc, −159T/luc, or pGL3-Basic (20 μg) and a control Renilla luciferase reporter plasmid (pRL-CMV, 25 ng; Promega) using electroporation. Transfection was performed using a T820 Electro Square Porator (BTX, San Diego, CA) at 2.0 kV with three 99-μs pulses. Cells were then incubated at room temperature for 15 min before the addition of growth medium with 10% human serum (Sigma-Aldrich, St. Louis, MO). Following a 6-h incubation, luciferase activity was measured using the Dual-Luciferase Reporter Assay system (Promega) on a TD 20/20 Luminometer (Promega). Protein concentration of cell extracts was determined by the bicinchoninic acid method (Pierce, Rockford, IL). Results were normalized for Renilla activity and protein concentration, and were expressed as relative luciferase activity (RLA). In selected experiments, transfected Mono Mac 6 cells were stimulated with LPS from Salmonella minnesota (100 ng/ml; Sigma-Aldrich) for 6 h before measuring luciferase activity.

HepG2 cells (8 × 10⁵/ml in DMEM) were transiently transfected with −159C/luc or −159T/luc (1 μg) and pRL-CMV (1.25 ng) using Lipofectamine 2000 (6 μl; Life Technologies, Rockville, MD). Following a 24-h incubation in growth medium, luciferase activity was measured as described above.

EMSA analysis was performed using 22-bp oligonucleotides (22C or 22T) corresponding to the −167/−146 region in the CD14 promoter that encompasses CD14/−159 (Fig. 1,B). Mutant oligonucleotides were generated by introducing three nonoverlapping blocks of 5-bp transversions into wild-type 22C (Fig. 1 C). In addition, we used oligonucleotides containing binding sites for Sp1 (5′-ATTCGATCGGGGCGGGGCGAGC) and STAT6 (5′-CTGTTCTTGGGAAGTC; consensus sequences are underlined). Single-stranded oligonucleotides were obtained from Life Technologies. Complementary oligonucleotide pairs were annealed and isolated as previously described (28). When used as EMSA probes, double-stranded oligonucleotides (50 ng) were end-labeled using [γ-³²P]ATP (3000 Ci/mM; ICN Pharmaceuticals, Costa Mesa, CA) and T4 polynucleotide kinase (Life Technologies). Free radioactivity was removed by gravity chromatography over Sephadex G25 columns (Boehringer Mannheim, Indianapolis, IN).

Mono Mac 6 cells (10⁷ cells) were washed with ice-cold PBS, resuspended in 1 ml of ice-cold hypotonic buffer (29) with protease inhibitors (1 mM PMSF, and 3 μg/ml each of leupeptin, aprotinin, antipain, and pepstatin) and phosphatase inhibitors (5 mM β-glycerol phosphate, 1 mM benzamidine, 1 mM orthovanadate, and 1 mM sodium fluoride), and were then incubated for 15 min on ice. Tubes were subsequently centrifuged for 30 s at 16,000 × g at 4°C. The nuclear pellets were washed twice with ice-cold hypotonic buffer and were then treated with 100 μl of extraction buffer (29) with protease and phosphatase inhibitors as above for 20 min on ice. Nuclear debris was pelleted by centrifugation for 10 min at 16,000 × g at 4°C. The protein concentration of the nuclear extracts was determined using the bicinchoninic acid assay (Pierce). Aliquots of the isolated nuclear extracts were snap frozen in liquid nitrogen and stored at −70°C until used. Nuclear extracts from HeLa cells were obtained from Geneka (Montreal, Canada). Nuclear extracts from HepG2 were prepared as previously described (28) with the addition of 0.1% Nonidet P-40 to the lysis buffer. Nuclear extracts from freshly isolated monocytes were prepared following the Andrews and Faller modification (30) of the procedure described by Dignam (31).

After preliminary data indicated that Sp family proteins bind differentially to the 22C and 22T oligonucleotides, EMSA conditions were optimized for Sp protein/DNA interactions as follows: nuclear extracts (5–10 μg) were incubated for 20 min on ice with binding buffer (20 mM HEPES-HCl (pH 7.6), 100 mM NaCl, 12.5% glycerol, 14 mM 2-ME, 1 mM EDTA, 0.5% Nonidet P-40, and poly(dI-dC) × poly(dI-dC) (1 μg; Amersham Pharmacia Biotech, Piscataway, NJ) in the presence or absence of oligonucleotide competitors. A ³²P-labeled probe (30 fM, 100,000–150,000 cpm/lane) was then added to the binding reaction (final volume, 25 μl) for an additional 20-min incubation on ice. The binding reactions were run on a 3.2% polyacrylamide gel (acrylamide:bisacrylamide ratio of 37.5:1; Bio-Rad, Hercules, CA) in 0.25× Tris-borate-EDTA buffer (25 mM Tris, 25 mM boric acid, and 0.5 mM EDTA, pH 8) at 200 V for 2–3 h at 4°C. Gels were dried before exposure to autoradiographic film. Autoradiographs were quantitated by densitometric analysis using Molecular Analyst software (Bio-Rad).

For Ab supershift assays, polyclonal antisera (4 μg) were added to nuclear extracts for 20 min on ice before incubation with the probe. Polyclonal antisera specific for human Sp2 (rabbit K-20), Sp3 (goat d-20), Sp4 (rabbit V-20), and Fra1 (rabbit R-20) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit polyclonal antiserum to Sp1 was purchased from Geneka. Recombinant Sp1 protein was obtained from Promega. An AP2 EMSA kit (including AP2 oligonucleotide (5′-CCACAAACGACCGCCCGCGGGCGGT), AP2 mut oligonucleotide (5′-CCACAAACGACCGATTGCGGGCGGT), JEG-3 nuclear extracts, binding buffer, and anti-AP2-α Ab) was obtained from Geneka and was used for the detection of AP2/DNA interactions.

The transcriptional activity of the CD14/−159C and T variants was compared by transiently transfecting the −159C/luc and −159T/luc reporter constructs into monocytic Mono Mac 6 cells. Luciferase activity in cell extracts was assessed 6 h post-transfection and was expressed as fold increase in the activity of the CD14 reporter constructs compared with the empty vector, pGL3-Basic. The data presented in Fig. 2 indicate that, consistent with previous reports (26), the wild-type CD14 promoter (−159C/luc) had strong basal transcriptional activity. Importantly, the C→T SNP at −159 (−159T/luc) resulted in a 32% increase in constitutive gene expression. This difference in activity was significant (p < 0.0001) and consistent in 14 independent transfections.

LPS stimulation enhances CD14 expression in Mono Mac 6 cells, as assessed by quantitation of CD14 mRNA (Ref. 32 and T. D. LeVan, unpublished observations). Therefore, we compared the activity of −159C/luc and −159T/luc in Mono Mac 6 cells stimulated with LPS (100 ng/ml) for 6 h. LPS induced a 49% increase in the activity of −159C/luc and −159T/luc, which was similar to the increase observed for CD14 mRNA under the same conditions (n = 8, data not shown). Most importantly, a 20% difference (p < 0.003) between the expression of −159C/luc and −159T/luc was still apparent after LPS stimulation. These data show that the CD14/−159T allele is transcribed more efficiently in both unstimulated and LPS-stimulated monocytic cells.

To understand the mechanism(s) responsible for the increased transcriptional activity of the CD14/−159T allelic variant, the sequence of the CD14 promoter encompassing the SNP was analyzed for the presence of potential transcription factor binding sites (33). A putative Sp protein-binding GC box and an AP2 motif were predicted to span CD14/−159 (Fig. 1 B). Therefore, EMSA analysis was performed to assess the ability of AP2 and Sp proteins to bind the C and/or T allelic variants using 22-bp oligonucleotides (22C and 22T, position −167/−146) that include both the GC box and the AP2 site. AP2 complexes were undetectable using nuclear extracts from AP2-expressing JEG-3 cells (Ref. 34 and data not shown), thus ruling out an interaction between the polymorphic region of the CD14 promoter and AP2 proteins.

Transcription factors of the Sp family activate a wide subset of mammalian genes containing GC box promoter elements (reviewed in Ref. 35). Currently, this family consists of four proteins designated Sp1–4. The GC-rich region encompassing the −159 polymorphic site (Fig. 1 B) represents a potential Sp protein binding motif (36, 37, 38). Of note, the presence of a T at −159 decreases the homology between the CD14 promoter GC box and the Sp consensus sequence.

Sp protein/DNA interactions were investigated by EMSA. Fig. 3 (top) shows that incubation of a 22-bp probe containing a high-affinity binding site for Sp1 with nuclear extracts from HeLa cells that express high levels of Sp proteins (39) resulted in the formation of three specific complexes, all of which were disrupted by the addition of a 100-fold molar excess of unlabeled Sp1 oligonucleotide, but not by an unrelated STAT6 oligonucleotide. Complex I appeared to contain Sp1 and Sp2 because it was supershifted by Sp1- or Sp2-specific antisera. Complexes II and III were identified by Ab supershifts as containing two Sp3 species of different mobility (40). None of the complexes contained Sp4 as assessed by supershifting using Sp4-specific antiserum (data not shown). A control antiserum to Fra1 had no effect on the migration of the complexes. Notably, all three specific complexes generated by the Sp1 probe were competed by unlabeled 22C and 22T oligonucleotides, suggesting that Sp proteins may bind at, or in the vicinity of, CD14/−159.

To demonstrate more directly the interaction between Sp proteins and the CD14 promoter, EMSAs were conducted using 22C or 22T oligonucleotide probes and HeLa cell nuclear extracts. Three specific complexes, virtually identical in mobility to those detected with the Sp1 probe, were observed (Fig. 3, center and bottom). Indeed, all three complexes were competed specifically by an Sp1 oligonucleotide and contained Sp1 (Fig. 3, complex I), Sp2 (Fig. 3, complex I), and Sp3 (Fig. 3, complexes II and III), as identified by supershift assays. In addition, recombinant Sp1 protein was found to interact with the 22C and 22T as well as Sp1 probes (data not shown). Thus, our results demonstrate that the transcription factors Sp1, Sp2, and Sp3 can bind the CD14 promoter region that contains CD14/−159C→T. Importantly, the competition curves with 22C and 22T oligonucleotides provided the first evidence to suggest a difference in affinity of the two allelic variants for Sp family members.

To further dissect the impact of −159C→T on CD14 promoter/protein interactions, we performed mutational analysis to assess how different regions of the GC box contribute to Sp protein binding. Three mutants, each containing a block of five transversions, were generated on the 22C backbone (Fig. 1,C) and were used as competitors in EMSA analysis. Fig. 4 shows that Mut2 (in which the inner core of the GC box had been disrupted) was severely impaired in its capacity to compete 22C probe binding to HeLa nuclear proteins. By contrast, Mut1 (which includes CD14/−159) and Mut3 retained substantial protein binding ability. We conclude that the core of the GC box in the CD14 proximal promoter is essential for Sp protein binding, whereas the polymorphic region immediately upstream may contribute to the overall affinity of Sp protein/DNA interactions.

Nuclear extracts from HeLa cells represented a useful tool to characterize Sp protein/CD14 promoter interactions because they contain high levels of Sp1–3. However, to correlate the functional data from transient transfections with the results provided by EMSA analysis, we needed to test nuclear extracts from the monocytic cell line used in reporter assays, Mono Mac 6. Fig. 5 (left) shows that incubation of Mono Mac 6 nuclear extracts with an Sp1 probe resulted in the formation of three specific complexes of mobility similar to that observed using HeLa nuclear extracts (see Fig. 3), even though complex III was barely detectable. All complexes were disrupted by the addition of a 100-fold excess of unlabeled Sp1 oligonucleotide, but not by an unrelated STAT6 oligonucleotide. Ab-mediated supershifting indicated that complex I contained Sp1 and Sp2 proteins. Complex II and the faint complex III were completely abrogated by anti-Sp3 Abs. When EMSA analysis was performed using Mono Mac 6 nuclear extracts and a 22C probe, a similar pattern of protein/DNA interactions was detected (Fig. 5, center).

Finally, nuclear extracts from primary peripheral blood monocytes isolated by elutriation were tested for their ability to bind the region of interest in the CD14 promoter. Fig. 5 (right) shows that the 22C probe was able to bind Sp1, Sp2, and Sp3, resulting in the formation of complexes I, II, and III. All three complexes were specifically abrogated by Sp1 as well as by 22C and 22T oligonucleotides, with competition and Ab supershift patterns similar to those obtained with HeLa and Mono Mac 6 nuclear proteins. Overall, our results confirm a similar pattern of Sp1–3/CD14 promoter interactions for HeLa cells, Mono Mac 6 cells, and freshly isolated human monocytes, except for the low expression of the high mobility form of Sp3 in Mono Mac 6 cells.

The finding that Sp1, Sp2, and Sp3 bound both the 22C and 22T oligonucleotides indicated that there was no qualitative difference in the ability of Sp proteins to interact with the two allelic variants of the CD14 promoter. In contrast, the 22C and 22T oligonucleotides appeared to differ in their capacity to compete for Sp protein binding (see competition lanes in Figs. 3 and 5). Therefore, we sought to assess whether the presence of a C or a T at −159 resulted in a difference in the affinity of Sp proteins for the CD14 promoter. To this purpose, nuclear extracts from HeLa cells were incubated with an Sp1 probe and an increasing molar excess of Sp1, 22C, or 22T oligonucleotide competitor. The intensity of the resulting bands was quantitated by scanning densitometry, and the percentage of binding relative to uncompeted probe was plotted against the molar excess of competitor added to each sample (Fig. 6, top). The mean ± SEM of three independent experiments is shown for each competitor in each panel. As expected, an unlabeled Sp1 motif competed Sp protein binding quite effectively even when added at a 25-fold molar excess. By contrast, 22C and 22T bound less avidly and, furthermore, markedly different affinities were consistently observed for the two oligonucleotides. Indeed, a 100-fold molar excess of 22T was required to obtain the same degree of competition detected using 22C at 50-fold molar excess. These results indicate a preference for Sp protein binding to the CD14/−159C allelic variant. Of note, a similar pattern of competition was observed using nuclear extracts from freshly isolated human monocytes (Fig. 6, center).

An apparent paradox emerged from our results described above: higher transcriptional activity of the CD14/−159T allele was associated with DNA/protein interactions of lower affinity. However, our results also demonstrated that the nuclear factors binding to CD14/−159C→T are members of the Sp family, which includes proteins with different functional properties. Indeed, Sp1 has been shown to function as a transcriptional activator, whereas Sp3 is characterized usually as a repressor (reviewed in Ref. 35). Therefore, we reasoned that the higher transcriptional activity of the T allele in monocytes may be due to its decreased affinity for Sp3-containing complexes with inhibitory potential. As a corollary, monocytes would be expected to express low and possibly limiting levels of Sp3 relative to Sp1 and Sp2. By the same token, an excess of Sp3 should overcome affinity-related transcriptional differences between the two CD14 alleles.

To test our hypothesis, we took advantage of the existence of another major cell type that expresses CD14 in vivo. The hepatocytic cell line HepG2, as well as primary hepatocytes, shows moderate but readily detectable constitutive levels of CD14 transcription (10, 41). In preliminary experiments, we used EMSA analysis to determine the relative expression of Sp family members in HepG2 cells. Fig. 7,A shows that HepG2 nuclear extracts formed three major complexes with an Sp1 or 22C probe. Complex I appeared to contain Sp1 and Sp2 because it was supershifted by Sp1- and Sp2-specific antisera. Complexes II and III contained Sp3 because the bands disappeared upon addition of an Sp3-specific Ab. As expected, all complexes were specifically competed by 22C and 22T oligonucleotides, but not by an unrelated oligonucleotide (STAT6). Of note, binding studies using the two allelic variants of the CD14 promoter and HepG2 nuclear extracts again showed a marked decrease in the affinity of the T allele (Fig. 6, bottom). Even more importantly, upon densitometric analysis of the complexes formed by Sp factors, we noted that the relative intensity of the Sp3-containing bands was dramatically increased in HepG2 cell EMSAs relative to monocytic cells. Indeed, the ratio of Sp3 (complexes II and III) to Sp1 and Sp2 (complex I) was 2:1 in HepG2 cells, whereas an inverted 1:2 ratio was observed in both Mono Mac 6 cells and freshly isolated monocytes (Fig. 7 B).

Because HepG2 and Mono Mac 6 cells exhibit opposite patterns of Sp family member expression, these two cell lines represent an ideal in vivo model to test whether increased transcriptional activity of the T allele correlates with a decreased sensitivity to Sp3-dependent inhibition. We therefore compared the activities of the −159T/luc and −159C/luc reporter vectors in Mono Mac 6 and HepG2 cells. The results of these experiments are shown in Fig. 7 C, and the experiments provided information that supports our hypothesis. On the one hand, the increased activity of the T allele was lost in Sp3-rich HepG2 cells when compared with Mono Mac 6 cells. Thus, an excess of Sp3 overrode the difference in affinity created by the polymorphism. In addition, the overall transcriptional activity of the CD14 reporter vectors was diminished in HepG2 cells, probably because of Sp3-mediated repression through other Sp binding motifs known to exist in the proximal CD14 promoter (26). Taken together, these findings suggest that the interplay between CD14 promoter affinity and the [Sp3]:[Sp1 + Sp2] ratio plays a critical mechanistic role in regulating CD14 transcription, and more specifically, in determining the differential activity of the two allelic variants of the CD14 promoter.

CD14 is involved in LPS signal transduction and therefore is essential both for interfacing the innate immune system with microbial pathogens and for directing adaptive immune responses along specific T helper differentiation pathways (5). The association between naturally occurring CD14 promoter genotypes and levels of serum sCD14 recently observed by our group in vivo (18), and the concomitant association with total serum IgE detected in the same population, reiterates the potential role of CD14 as a candidate gene for asthma and allergy. Indeed, our preliminary analysis indicates that subjects homozygous for CD14/−159T are more responsive to LPS inhalation and at the same time are less prone to become atopic (T. D. LeVan, R. Rylander, and O. Michel, manuscript in preparation). In contrast, the association between the same SNP and risk for coronary artery disease (19, 20, 21) underscores the importance of CD14 in inflammatory processes.

Because of the central role of CD14 in innate immunity, the regulation of CD14 gene expression has been extensively investigated. A 227-bp region upstream of the major transcription initiation site was found to be sufficient to drive maximal reporter activity in transient transfection assays (26). This region contains binding sites for Sp1 and members of the CCAAT/enhancer-binding protein (C/EBP) family of transcription factors (26, 42). Basal promoter activity was strictly dependent on the integrity of Sp1 and C/EBP motifs located at position −110 and −135, respectively. The same C/EBP site was also involved in the TGF-β/Vit D3-dependent activation of CD14 expression (42). More recently, in vivo studies with transgenic mice have shown that an 80-kb human genomic fragment that includes the CD14 coding region contains critical regulatory elements necessary to direct tissue-specific CD14 expression in monocytes and hepatocytes (9, 41). Within this fragment, a 0.7-kb enhancer located ∼6 kb 5′ of the CD14 transcription initiation sites appears to be essential for CD14 expression in hepatocytes, while additional upstream elements are required to direct transcription in monocytes (41).

Our present data show that a common SNP in the proximal CD14 promoter results in increased transcriptional activity. A reporter vector containing CD14/−159T was more active than the C allelic variant in transient transfection assays using CD14-expressing monocytic cells. This increase in activity was paralleled by a decreased affinity of the interactions between the Sp1, 2, and 3 proteins and the polymorphic GC box in the CD14 promoter. All Sp family members contain a highly conserved DNA binding domain and bind the same consensus sequence, but show promoter context-related differences in their transactivating properties (35). It is generally believed that the function of Sp protein-dependent promoters is regulated by the relative ratio between activating and repressing members of the Sp family. In this scenario, the −159/C→T polymorphism increases transcription by lowering the affinity of the CD14 regulatory region for Sp3, a factor known to inhibit the activity of a number of promoters (40, 43, 44). This line of reasoning is strongly supported by the comparison of the transcriptional activities of the C and T allele in two cell lines that exhibit opposite patterns of Sp1/Sp2 and Sp3 expression. Indeed, transcription of the T allele was increased in the monocytic cell line Mono Mac 6 in which Sp3 expression is low. By contrast, in Sp3-rich hepatocytic HepG2 cells, the CD14/−159C→T transcriptional difference was lost. Thus, the affinity of the promoter GC box and the ratio between activating and repressing Sp family members play a complex but critical role in regulating the transcription of the two allelic variants of the CD14 promoter.

In our population studies, levels of serum sCD14 were higher in the CD14/−159T homozygotes (18), and as described in this study, the CD14/−159T reporter construct was transcriptionally more active than the C allelic variant in monocytic cells but not in hepatocytes. Yet sCD14 is known to be secreted by both monocytes (8) and hepatocytes (9, 10). Thus, it is possible that genetic variation in the CD14 proximal promoter may affect its ability to engage in long-range interactions with liver-specific, promoter-distal elements such as the enhancer situated ∼6 kb upstream of the CD14 transcription start site (41). These elements are not included in the reporter constructs analyzed in this study.

More and more reports are becoming available that describe novel polymorphisms in disease-related genes, and associations between these SNPs and specific phenotypes. However, few mechanistic studies have so far addressed the functional impact that SNPs in regulatory regions have on the control of gene expression (45, 46, 47, 48). A common theme of these studies, as well as in ours, is that, unlike conventional vectors containing artificial mutations, most reporter constructs that model naturally occurring polymorphic variants exhibit relatively modest differences in transcriptional activity when compared with their wild-type counterpart in transient transfection assays. However puzzling, this paradigm is consistent with the basic notion that these instances of genetic variation may be insufficient to cause disease when taken individually. However, they may induce significant modulation of gene expression when they become involved in gene-by-gene and/or gene-by-environment interactions. Thus, complex phenotypes and/or polygenic diseases would not be expected to emerge as all-or-none events, but rather to result from a combination of small quantitative effects. This may be especially true of innate immunity, the genes for which are highly conserved across species and therefore are not likely to tolerate dramatic shifts in their functional patterns. Therefore, we envision a scenario in which a constellation of variations in critical genes fine-tunes innate and adaptive immune responses, and the way they interface with the environment, resulting in a wide spectrum of related phenotypes.