Published ahead of print on February 13, 2003, doi:10.1164/rccm.200209-1064OC
© 2003 American Thoracic Society Characterization of a Single Nucleotide Polymorphism in the Lipopolysaccharide Binding Protein and Its Association with Sepsis
Department of Surgery, University of Texas Southwestern Medical Center, Dallas, Texas; and Department of Surgery Correspondence and requests for reprints should be addressed to Grant O'Keefe, MD, University of Washington, Department of Surgery, Box 359796, Harborview Medical Center, 325 Ninth Avenue, Seattle, WA 98104–2499. E-mail: gokeefe{at}u.washington.edu
We sought to characterize polymorphisms in the proximal coding region of the lipopolysaccharide binding protein gene and to determine whether a previously reported variant was associated with sepsis complicated by organ failure or shock after trauma. We used multiple analytical methods, including pyrosequencing, restriction fragment length polymorphism, and sequencing to characterize the proximal coding region. We also reexamined a prospective cohort of severely injured patients and healthy control individuals. The single nucleotide polymorphism at nucleotide 292 does not exist as previously reported. Instead, the adjacent nucleotide (291) was observed to be polymorphic. In 151 trauma patients, 37 (25%) developed severe sepsis, and 19 (13%) died. Thirteen of 50 (26%) C-allele carriers and 24 of 101 (24%) TT homozygotes developed severe sepsis. Unadjusted and adjusted analyses did not demonstrate any associations between genotype and severe sepsis, septic shock, or death. In conclusion, a single nucleotide polymorphism in the lipopolysaccharide binding protein coding region that was reported to exist at the 292 position and to result in an amino acid substitution actually exists at the adjacent 291 position and does not result in an amino acid substitution. Furthermore, this polymorphism does not appear to be associated with complicated sepsis after trauma.
Key Words: lipopolysaccharide • polymorphism • sepsis • gene
The lipopolysaccharide (LPS) binding protein (LBP) facilitates the transfer of bacterial LPS to the LPS receptor, CD14, and is important in the host response to LPS and gram-negative bacteria (1, 2). Given the variability in the inflammatory response between individuals, it is important to consider genetic differences in the inflammatory and innate immune responses (LBP, CD14, various cytokines) as possible determinants of outcomes after infectious and inflammatory conditions (3, 4). In our investigations of such associations with severe sepsis and septic shock, we have studied the proximal coding region of the LBP gene. Other investigators have observed an association between a T
After examination of the nucleotide sequences published in GenBank and EMBL and after careful examination of genetic material from a cohort of injury victims and uninjured control subjects, we observed that this previously reported T
Source of Genetic Material, DNA Extraction, and Amplification DNA was obtained from victims of traumatic injury who were admitted to the trauma center at Parkland Memorial Hospital in Dallas, Texas. We also obtained DNA from 133 uninjured control subjects. All study subjects or their surrogate provided informed consent for blood sampling and genotyping. The University of Texas Southwestern Medical Center Institutional Review Board approved all procedures. Genomic DNA was extracted from whole blood using the QIAamp DNA Blood Midi Kit (Qiagen, Valencia, CA) according to manufacturer's instructions, and DNA was stored at -20°C until amplification. The region spanning nucleotides 200 through 317 of the LBP gene was amplified from genomic DNA by polymerase chain reaction (PCR) using Taq DNA polymerase (Roche Diagnostics, Indianapolis, IN) and a PTC 200 thermal cycler (MJ Research, Watertown, MA). Published PCR primers and conditions were used in all phases of this study (5), with the exception of pyrosequence analysis, which required redesigned primer sequences and conditions (Table 1) .
Cloning and Sequence Analysis HpaII digestion of the PCR-amplified fragment was used to identify individuals representing each genotype (Figures 1 and 2) . Five individuals for each genotype were subsequently amplified, and the PCR products were inserted into the pCRII plasmid vector via TA cloning (Invitrogen, Carlsbad, CA). Briefly, this technology takes advantage of the terminal transferase activity of Taq DNA polymerase, which precipitates the addition of a single A residue to the 3' ends of PCR-amplified DNA fragments (6, 7). The plasmid vector was constructed by enzymatic linearization within a multiple cloning site that contained a copy of the LacZ-  complement and further modification to contain single nucleotide T overhangs at the 3' ends. These modifications allowed a single base pair (T/A) overlap during ligation that increased the efficiency of the cloning reaction. Insertion of the PCR product into the multiple cloning site interrupted the LacZ- gene and rendered it nonfunctional. The recircularized plasmids were then used to transform INVaF'-competent Escherichia coli cells, and the transformed bacteria were grown overnight on agar plates containing the lactose analogue X-Gal. The addition of this reagent to the bacterial growth medium allowed for blue-white screening of colonies that were likely to contain the fragment of interest. Colonies that contained a pCRII plasmid with a PCR-amplified insert were unable to cleave X-Gal, which gave the positive colonies a white, rather than a blue, color. Three positive (white) colonies were selected at random from each individual, and the plasmid DNA was isolated (8). The DNA sequence of the plasmids was then determined by fluorescent modification of the dideoxy chain termination sequencing method (9), and SNPs were identified by visual inspection of the sequence electropherograms (Figure 3)
.
Pyrosequencing Analysis All genotypes were determined by pyrosequencing analysis on a PSQ 96 Pyrosequencer (Pyrosequencing AB, Westborough, MA) according to manufacturer's specifications, and genotypes were resolved using PSQ 96 SNP Software, version 1.2 AQ. Each SNP was assayed with a specific primer sequence (Table 1), which enabled the unambiguous scoring of heterozygotes and alternate homozygotes (10). As a sequencing-based technique, pyrosequencing is a rapid method that has been used for SNP haplotyping and as a reliable alternative to microsatellite analysis for chimerism (11, 12).
Patient Recruitment, Data Presentation, and Analysis
In this report, our primary analysis was of the LPB genotype as a risk factor for severe sepsis or septic shock that are referred to together as complicated sepsis. We compared allele frequencies in patients with and without complicated sepsis by Fisher's exact test. The unadjusted relative risk for severe sepsis with its associated 95% confidence interval is also presented. We then conducted a logistic regression analysis, incorporating the LBP genotype in addition to the variables that we previously identified to be risk factors for complicated sepsis. These variables included age of 55 years or more, an arterial base deficit of 6 meq/L or more measured between 6–24 hours after injury, and carriage of the A-allele at the -308 position in the tumor necrosis factor-
Identification of a T  C SNP at the 291 Position of the LBP GeneDNA from 15 subjects, representing all three possible genotypes, was subjected to sequencing of the forward and reverse strands. Examination of the sequence data revealed that nucleotide 292 of the LBP cDNA was an invariant G. However, a T C transition SNP was identified at the adjacent nucleotide 291 (Figure 3). This synonymous SNP would result in alteration of an HpaII restriction endonuclease recognition site but would not precipitate an amino acid substitution. Pyrosequencing confirmed the results of sequencing in all 15 sequenced samples and was therefore used to genotype this SNP in the remaining trauma and control individuals. A sample of each genotype was subjected to pyrosequencing a second time, confirming the initial results in all cases. The allele frequencies for patients and control subjects according to ethnicity are shown in Table 2
. Genotype frequencies were in Hardy–Weinberg equilibrium in both groups. There were significant differences in allele frequencies across ethnic groups, but within ethnic groups, the allele frequencies were similar in the trauma patients and the uninjured control subjects.
Characteristics of the Trauma Patient Cohort The demographic, clinical, and outcome variables for this cohort were previously reported in detail and are summarized here (13). Most of the patients were men (112; 74%), and 27 (15%) were 55 years of age or more. The cohort was multiethnic, but the majority were white (86; 57%), followed by Hispanic (47; 31%) and African American (12; 8%) according to self-report. Shock, defined by a base deficit of 6 meq/L or more measured from 6 to 24 hours postinjury, was common, occurring in 40 (27%) patients. The majority of patients developed a nosocomial infection while in the intensive care unit (96; 64%). Sepsis was complicated by organ failure or shock in 37 (25%) patients, and 19 (13%) patients died.
Analysis of LBP 291 Genotype and Risk for Severe Sepsis
Logistic regression revealed that carriage of the A-allele at the LBP 291 SNP was not associated with an increased risk for severe sepsis after trauma (odds ratio = 1.157; 95% confidence interval, 0.463–2.891). The inclusion of the LBP genotype did not markedly alter the results of our previously reported analysis of this patient cohort. This original report did not include the LBP SNP but did include a SNP at nucleotide -308 of the tumor necrosis factor-
We conducted a series of additional, secondary analyses. First we examined the association between the LBP 291 C-allele and septic shock, the most severe manifestation of sepsis, and with death. Few patients (18; 12%) developed septic shock, and there was a similar risk across LBP genotypes. Of the C-allele carriers, 7 (14%) developed septic shock, and of the TT homozygotes, 11 (11%) developed septic shock. There was no association between the 291 C-allele and death. We then limited our analysis to the 96 patients who developed a nosocomial infection and similarly found no association between the 291 C-allele and either severe sepsis or septic shock. We finally limited the analyses to the 86 white patients and did not observe any associations between the 291 C-allele and severe sepsis, septic shock, or death (data not shown).
By direct sequencing of the cloned DNA fragment containing the proximal LPB gene coding region, we have determined that the HpaII RFLP site, thought to be due to a T G SNP at the 292 nucleotide, is in fact the result of a T C SNP at the adjacent 291 position. Either of these substitutions would result in the generation of the HpaII restriction site, allowing differentiation of the wild-type and variant alleles by RFLP analysis. However, this method of detection is unable to distinguish the sequence of the underlying polymorphism. This information is crucial, as the alternative SNPs would have substantially different effects on the resulting amino acid sequence. A T G transition at the 292 position would be nonsynonymous and would result in the replacement of cystine with glycine at amino acid residue 98. In contrast, the T C substitution at 291 is silent, with both variants encoding the amino acid proline at residue 97. We have used a third genotyping method, pyrosequencing, to confirm our findings. This technique directly sequences a short DNA fragment (5–10 bases), including the polymorphic region of interest, and provides a visual display of that sequence. Like RFLP, pyrosequencing requires prior knowledge of the nucleotide sequence adjacent to the SNP, but unlike RFLP, pyrosequencing reveals the exact sequence change or polymorphism. The HpaII enzyme recognizes the nucleotide sequence CCGG. Therefore, whereas either of the substitutions at nucleotide 291 or 292 would result in the generation of a restriction site that appears identical by RFLP analysis, the SNPs would easily be differentiated by sequencing or pyrosequencing. The results of pyrosequencing correspond exactly to the results of sequencing in our samples, identifying the SNP at the 291 position. Figure 3 illustrates the three variations on the nucleotide sequence of this region. The synonymous substitution at 291 would not directly alter the protein structure, whereas the previously reported SNP at 292 would precipitate the replacement of cystine with glycine. This amino acid substitution would have the potential to disrupt disulfide bonds and lead to altered function of the LBP protein (5). The association between this SNP and severe sepsis that was previously reported was observed only in a subgroup analysis of patients with sepsis (males but not females), was not strong, and was only one positive association of five polymorphisms that were tested (5). These previously reported findings of an association between the erroneously located nonsynonymous SNP may represent a false-positive association but may also represent a true association in which this polymorphism is part of an extended haplotype of genetically linked SNPs and may thus be a marker for other functional variations (17, 18). In the previously reported cohorts of patients with sepsis and healthy control subjects, this polymorphism was linked to another SNP in the LBP gene. Therefore, the possibility remains that the 291 SNP is a marker for, but not a functional contributor to, adverse outcomes. Thus, additional investigation into identifying other SNPs in this region and defining haplotypes may prove important. However, coupled with our determination that this coding region SNP is synonymous, a marginal association should not be considered definitive and excludes any functional importance of this particular variant. Our study had fewer subjects than the report referred to previously here but included a similar number of men. As they observed the association between LBP polymorphism and sepsis in only males, it is unlikely that our observation of no association is due a function of inadequate sample size. Of interest is our observation of a marked ethnic variation in allele frequencies. In both trauma victims and uninjured control subjects, the C-allele was significantly more common in the African American subjects. Whether this simply reflects the occurrence of a random mutation or whether this SNP is part of a biologically important haplotype that influences survival remains to be determined. It is thus important to identify and begin to explore ethnic differences such as observed here, rather than to avoid them through restricting studies such as this to individual ethnic groups (19, 20). These observations raise a number of important issues surrounding the study of genetic associations with complex disease, such as sepsis. First, our characterization of this subtle difference in published gene sequence reflects the importance of careful attention to genotyping methods in such studies, which are increasingly frequent in the medical literature. Inaccurate reporting of genotypes and associations can have a considerable negative impact, in part by directing time and resources along incorrect lines of investigation. Additionally, genotyping is likely to become increasingly important in clinical medicine, and maximizing the benefit to individual patients will require careful attention to genotype assignment (21). Although RFLP has often been replaced with newer and more rapid genotyping methods, it is an inexpensive and easily learned technique that remains in widespread use. Furthermore, errors are not specific to this technique. Numerous sequencing errors have been intermittently identified in the public sequence databases. For example, recent resequencing of the fibrinogen cluster gene on chromosome 4 has identified nine likely sequencing errors (22). It appears that a possible sequencing error may have in part contributed to the erroneous characterization of the LBP nucleotide. The second issue relates to the threshold significance level for genotype–phenotype associations. There is no consensus on how to deal with this issue, with a number of options that at least provide a framework in which to consider the issue (23, 24).
In summary, the HpaII RFLP site in the LPB gene, which was previously reported to be due to a T
Supported by NIGMS grant #5P50GM021681–370013 (G.E.O.). Received in original form September 17, 2002; accepted in final form February 8, 2003
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
