The role of lipopolysaccharide in the development of atopy in humans


  • A. Simpson,

    1. Manchester Academic Health Science Centre, NIHR Translational Research Facility in Respiratory Medicine, University Hospital of South Manchester NHS Foundation Trust, The University of Manchester, Manchester, UK
    Search for more papers by this author
  • F. D. Martinez

    1. Arizona Respiratory Center, University of Arizona, Tucson, AZ, USA
    Search for more papers by this author

Angela Simpson, Manchester Academic Health Science Centre, NIHR Translational Research Facility in Respiratory Medicine, University Hospital of South Manchester NHS Foundation Trust, The University of Manchester, Second Floor, Education and Research Centre, Manchester M23 9LT, UK.


Atopy is a highly prevalent condition and remains the single biggest risk factor for asthma. Although atopy has a heritable component, the time frame of the increase in the prevalence indicates that it is not due to genetic factors alone. The relationship between allergen exposure and sensitization is complex. Lipopolysaccharide (LPS) and its bioactive moiety endotoxin are common to all gram-negative bacteria, and have been used as a surrogate of microbial load. Endotoxin can be readily measured in dust collected from homes. Some studies have demonstrated a clear inverse dose–response relationship between exposure to endotoxin and the risk of atopy but this finding has not been reproduced in all studies. Our innate immune system recognizes LPS readily via the LPS signal transduction pathway, which has the trimolecular complex of CD14/TLR4/MD2 at the core. A common single-nucleotide polymorphism in the promoter region of CD14 rs2569190 C to T (CD14/−260 or CD14/−159) has been associated with elevated sCD14. Although early studies suggested that this variant was associated with more severe atopy, this finding was not uniformly replicated. It has now been demonstrated in four independent populations that high exposure to endotoxin in the domestic environment is protective against the development of atopy, but only among carriers of the C allele, that is, the environmental exposure is only relevant when taken in the context of the genotype. Furthermore, this interaction is biologically plausible. We propose that neither the environmental exposure nor the genotype in isolation is sufficient to cause complex diseases like asthma and atopy, but disease results from the one acting in the context of the other, of which CD14 and endotoxin is one example contributing to the risk for atopy.

Cite this as: A. Simpson and F. D. Martinez, Clinical & Experimental Allergy, 2010 (40) 209–223.


Atopy is a term ‘reserved to describe the common genetic predisposition to become IgE-sensitized to allergens commonly occurring in the environment and to which everyone is exposed but to which the majority do not produce a prolonged IgE antibody response’ [1]. Although atopy per se is not a disease and is neither necessary nor sufficient for the development of asthma, it remains the strongest risk factor for this condition and consequently merits investigation [2]. The prevalence of atopy has increased in over the later part of the last century [3], and currently occurs in approximately one in three of the population in the developed world [4]. The timeframe of this increase is too short for this to be due to genetic factors alone, although twin and family studies do suggest a heritable component [5]. Although few would argue that exposure to allergens is necessary for the development of sensitization, questions remain about the timing of exposure and the dose that presents the highest risk. In some countries where mites are prolific, dust from most domestic environments contains levels high enough to cause sensitization. Thus, one might then ask why does only a proportion of the population become sensitized? The timing of exposure and patterns of exposure (e.g. transient very high exposures vs. chronic moderate exposures) may be crucial but other factors such as co-exposures may be more important [e.g. endotoxin, (1→3)-β-d-glucan, other microbials, dietary factors, environmental tobacco smoke]. In this review, we will consider the roles of endotoxin and of genetic variation in molecules involved in the response to endotoxin in the development of atopy; sensitization in animal models and the effect of inhaled endotoxin in adults on airway inflammation will be reviewed by others as part of this series.


Using the 1958 UK birth cohort, Strachan first described number of siblings, in particular older siblings, as the strongest protective factor for current hayfever [6]. A possible explanation was ‘… if allergic diseases were prevented by infection in early childhood, transmitted by unhygienic contact with older siblings, …’. The epidemiological observations of protection against allergic disease afforded by exposure to microbials have been termed the Hygiene Hypothesis. In addition to the potential role of clinical bacterial and viral infections (and possibly the treatment thereof), it is recognized that environmental exposures to microbial compounds that do not result in clinical disease but act through innate immune response mechanisms may also influence the development of adaptive immunity and consequently allergy [7]. These include compounds such as endotoxin and (1→3)-β-d-glucan, which can be measured in domestic dust. Indeed, using more recent birth cohorts derived from general practice research databases, the finding of protection against hayfever with increasing numbers of older siblings was confirmed, but was not related to numbers of acute infectious illnesses presenting to primary care in infancy [8], providing further evidence that exposure to microbials other than those causing clinical disease may be important.

Dendritic cells linking innate and adaptive immunity in atopy

A major challenge for the immune system is to maintain tolerance of self while mounting an appropriate defence against ‘non-self’ molecules. The non-self molecules comprise both pathogens and other innocuous environmental antigens, which, in some individuals, result in an allergic reaction, whose defining feature is the production of allergen-specific IgE antibodies and the presence of an allergen-specific T helper type 2 (Th2)-polarized memory. Much exposure to foreign substances occurs through the respiratory tract as small particles are continuously inhaled into the conducting airways. Mucosal dendritic cells (DCs – the most powerful antigen-presenting cells) are located just below the epithelial cell layer (reviewed in [9]) and are able to extend their processes through the epithelial cell layer into the airway lumen, providing a continuous form of immune surveillance [10]. DCs take up antigens from the airway lumen by endocytosis and then migrate to mediastinal lymph nodes, where the antigen/pathogen is processed into peptides for presentation on major histocompatability complex class II molecules [11]. During this process, DCs mature into immunostimulatory effector cells, losing their endocytotic ability and up-regulating CD80 and CD86 co-stimulatory molecules. The mature DC becomes functionally distinct and develops a Th1-, Th2-, Th17- or regulatory-inducing phenotype when the antigen is presented to naïve CD4 Th cells. As such, DCs are key determinants of how the allergic immune response is initiated and is perpetuated. The local tissue microenvironment plays a role in determining the phenotypic expression of the maturing DC, and elucidation of factors controlling this microenvironment is fundamental to understanding this process.

The usual outcome following inhalation of innocuous protein is T cell division and a humoral immune response, which usually results in immune tolerance [12, 13]. The predominant subset of allergen-reactive T cells in blood is IL10-secreting CD4+ T regulatory 1 cells in healthy controls, whereas in sensitized individuals, the main subset is IL4-producing Th2 cells. The question is what causes DCs to induce a Th2 phenotype in naïve Th cells following antigen presentation? The nature of the effector phenotype is largely determined by the predominant cytokine environment [14], the T cell polarizing cytokines. Multiple factors are involved in determining this environment, which include host factors (genetic predisposition, T cell receptor signalling and co-factors, other innate immune cells) allergen factors (nature, e.g. presence of proteolytic activity, dose and timing of exposure) and exposure to environmental factors (which may be protective of or may be a risk for the development of allergy).

One mechanism by which environmental factors may influence the effector phenotype has been elegantly demonstrated in BALB/c mice during sensitization to ovalbumin (OVA – a system in which host factors and allergen factors can be standardized) [15]. No airway inflammation was seen in mice that were treated with intranasal lipopolysaccharide (LPS)-depleted OVA. Antigen-specific immune responses were seen when intranasal OVA was administered in the presence of both high- and low-dose LPS inducing Th1 and Th2 responses, respectively. The Th2 responses were Toll-like receptor 4 (TLR4) and MyD88 dependent [16]. Both TLRs and protease-activated receptors are expressed on the surface of airway epithelial cells, and it is also recognized that activation of these receptors by their ligands induces a cascade of cytokine production, which can result in the maturation of DCs.

It is therefore plausible that exposure to microbial compounds can influence the nature of the human immune response to antigen exposure, most likely by influencing the local tissue microenvironment.


LPS is a structure common to all gram-negative bacteria and as such is an archetypal pathogen-associated molecular pattern. It is made up of three distinct regions – the hydrophobic lipid A component (endotoxin, the bioactive moiety), a short core oligosaccharide and the O-antigen polysaccharide (whose presence or absence determines colony morphology – absence resulting in rough colonies, and presence resulting in smooth colonies). The lipid A/endotoxin domain anchors the LPS molecule within the outer membrane of gram-negative bacteria. In humans, systemic infection with gram-negative bacteria can result in septic shock with multi-organ failure, an extreme response to endotoxin exposure.

That endotoxin was present in house dust was first reported in 1964 [17]. The respiratory effects of endotoxin were first documented in the context of occupational exposures in cotton workers [18], where it was reported to cause chest tightness and fever, together with changes in lung function including a reduction in gas transfer [19], which led to the suggestion that exposures were important in the development of byssinosis [18].

In a controlled laboratory environment, inhalation of endotoxin can induce dose-dependent systemic inflammatory responses characterized by leucocytosis and increased C-reactive protein [20], as well as airway responses characterized by bronchial obstruction with cell activation and neutrophil invasion in the airways of humans [21]. However, in animal models, it has been demonstrated that the effects of exposure to endotoxin may be modified in the presence of exposure to other agents commonly inhaled in dust. For example in guinea-pigs, the massive neutrophil invasion of the airways seen after endotoxin exposure is muted in the presence of (1→3)-β-d-glucan [22]. Furthermore, in an allergen model, exposure to endotoxin alone exerted a stimulatory effect on the IgG response to inhaled OVA. When (1→3)-β-d-glucan was included with the endotoxin, there was a much more muted antibody response to OVA [23]. These experiments serve as a reminder that in vivo subjects are not exposed to inhaled contaminants in isolation and that response to one organic dust may influence response to another, resulting in complex interactions in the adaptive and innate immune responses.

Endotoxin exposure and measurements

Endotoxin can be readily measured in dust collected from homes using a biologically active kinetic assay – the Limulus amebocyte lysate assay [24, 25]. Differences in sample collection, storage and extraction as well as technical complexities of the assay have made it difficult to compare levels between studies.

Although it has been recognized as a component of house dust since the 1960s, there remains no consensus as to how best to measure exposure in the domestic environment. Samples are usually collected by vacuuming a 1 m2 area of flooring or mattress for a fixed time period (e.g. 1 or 2 min) using a specially designed vacuum sampler or a modification (e.g. a sock) that can be fitted to a domestic vacuum cleaner. Investigators have shown that sampling 1 m2 area is not materially different from sampling the whole carpet [26]. Although there is a high correlation between samples collected from the same dust reservoir using different sampling heads (e.g. sock vs. ALK sampling device) when used by the same fieldworker, the actual quantity of endotoxin recovered can vary several fold, making between-study comparisons difficult [27, 28]. Results are expressed either as the concentration of endotoxin in EU/mg of dust or as total endotoxin recovered (or per unit area) in EU/m2 that factors in the mass of dust collected in the sample. The correlation between the two is very high [26].

Samples collected in this way from the mattress and the living room floor showed a higher intraclass correlation than dust taken from vacuum cleaner bags or collected as settling dust or air, suggesting that these are more robust sampling techniques [28]. In addition, there is a good correlation between levels in the mattress and levels in the living room floor [28]. Some studies have shown little seasonal variation [28] within homes but others have suggested that levels are higher in the summer [29, 30]. Most investigators have concluded that the between-home coefficient of variation is much greater than that within homes [28, 29, 31]. Although coefficients of correlation are less high over a 6-year period than a 1-year period, they are still in the region of 0.5 if taken from the same item [32], indicating that a single measurement during early childhood is a reasonable index of exposure. Although they are probably a relatively poor measure of the true exposure, as it is not feasible to perform long-term and/or repeated airborne measurements in most studies, settled dust measurements are generally used. It is recognized that it would be valuable to develop better tools for exposure measurement.

Determinants of endotoxin levels within individual homes have been studied using questionnaires on household characteristics. This approach suggested that factors associated with higher endotoxin levels in settled dust are more occupants of the home, keeping cats and dogs, having carpets or rugs, having an older vacuum cleaner, steam cleaning the carpets and having a higher relative humidity within the carpet and not having insulation under the floor [33]. For mattress levels, mattress type was also important – foam rubber having the lowest levels [34]. However, the proportion of the variance explained by the models is generally low [26, 34], suggesting that this cannot be used as a substitute for objective measures.

Endotoxin can be measured in air in almost all homes [35] and is mostly associated with airborne particulate matter <1 μm diameter [36]. Levels were higher in homes containing dogs but not cats [35]. City homes generally contain less endotoxin than farm homes, which have lower levels than barns, with rural homes in developing countries having levels somewhere between farms and barns [37].

In summary, endotoxin levels need to be measured and cannot be estimated from answers to a self-administered questionnaire. It is adequate to sample only 1 m2 area, best collected from the mattress or the living room floor. Although the sample can be collected by a study subject using a sock or a fieldworker using a custom-built sampler, comparisons of absolute levels with other studies may not be possible when different sampling techniques have been used. There is generally a higher between- than within-home variability.

Endotoxin in the rural environment – the effects of farming

Despite the fact that the farming environment is a recognized cause of occupational asthma [38], many studies show that children of farmers have a lower prevalence of atopy [39, 40] and hayfever [39–41] compared with other children living in the same vicinity but not on farms. For asthma, there was a degree of heterogeneity between studies [39, 41, 42]. There are of course other differences between farming and non-farming children including more dogs, less smoking and more siblings as well as a lower family history of allergic disease, but associations with reduced atopy withstood adjustment for these parameters [39, 41]. Levels of endotoxin are higher within the homes of farmers compared with non-farmers living in the same villages [41, 43]. Studies investigating the association between endotoxin exposure and atopy in the rural environment are summarized in Table 1. Investigators for the Allergy and Endotoxin Study team demonstrated that among children living in rural areas of Germany, Austria and Switzerland, exposure to farm milk and stables in the first year of life and thereafter was associated with reduced atopy, hayfever and asthma [40]. They were also able to demonstrate an inverse association between levels of endotoxin in children's mattresses and atopy, hayfever and asthma [43]. Furthermore, in a multivariate model, these two protective effects were independent of each other [43]. The follow-up PARSIFAL study (prevention of allergy risk factors for sensitization in children related to farming and an anthroposophic lifestyle) aimed to identify the specific factors within the farming environment that were protective [44]. In the multivariate model for atopy, protective factors were exposure to agriculture (cultivation of grain), pigs, poultry and regular stays in a barn and endotoxin (but not farm milk) [44]. A more recent study of children of farmers and their non-farming neighbours in Shropshire failed to replicate this finding. It was demonstrated in this study that the only independent protective factor for atopy was the consumption of unpasteurized milk (and not farming status or endotoxin) [41]. A possible role for endotoxin in unpasteurized milk acting through the gut rather than the respiratory tract has largely been discounted by the finding that levels of endotoxin within milk are higher in non-farming than in farming families – due to the fact that non-farming families store their milk for longer before consumption [45].

Table 1.   Association between endotoxin and atopy – studies conducted in a rural environment
Author, year, country,
Design, numbers,
SettingDefinition of
exposure; units
FindingEndotoxin protective,
no effect or a risk
for atopy
Braun-Fahrlander [43], 2002, Austria, Germany, Switzerland, ALEXCross-sectional survey, N=814, 6–13 yearsRural, farming and non-farmingDust from child's mattress; EU/mg and EU/m2Atopy (sIgE)Endotoxin levels were inversely related to atopic sensitization, OR 0.76 (0.58–0.98)ProtectiveAlso protective for hayfever but high levels were associated with an increase in non-atopic wheeze. In the multivariate analysis, exposure to farming in 1st year of life and endotoxin were independent protective factors
Ege [44], 2007, Schram-Bijkerk [46], 2006, Europe, PARSIFALCross-sectional survey, N=8263, 5–13 yearsChildren of farmers and Rudolf Steiner schools and controlsDust from child's mattress; EU/mg and EU/m2, N=440 with endotoxinAtopy (sIgE), (N=2086)Endotoxin was inversely related to atopic sensitization, OR 0.38ProtectiveNo effect of endotoxin on asthma or current wheeze. Endotoxin effects were independent of farming exposures
Wickens [47], 2005, New ZealandCross-sectional, N=293, 7–10 yearsRural, farming and non-farmingDust from living room floor; EU/g and EU/m2Atopy (spt)No association with atopyNo effectIncreased prevalence of allergic disease on farms; endotoxin levels were lower on farms
Perkins [41], 2006, UKCross-sectional, N=4767, School-age childrenRural, farming and non-farmingDust from living room floor; EU/mg (N=879)Atopy (spt) (N=879)No association with atopy, OR 0.94 (0.59–1.49)No effectUnpasteurized milk was associated with less atopy, irrespective of farming status
Eduard [48], 2004, NorwayNested study of symptomatic farmers and controls, N=2253, mean age 46 yearsFarmersPersonal sampling of endotoxin exposure during farming tasks (N=1614)Atopy (sIgE)No association with atopy, OR 0.82 (0.58–1.2)No effectHigher endotoxin exposure was associated with non-atopic asthma and inversely associated with atopic asthma
Portengen [49], 2005, The NetherlandsCross-sectional, adults, N=162Pig farmersPersonal exposure; EU/m3Atopy (sIgE)Strong inverse association between airborne endotoxin and atopy, OR 0.03 (0–0.3)ProtectiveHigher endotoxin was associated with an increase in airway hyperresponsiveness

For specific sensitization to mite allergen, investigators in the PARSIFAL study demonstrated that the bell-shaped curve between exposure to mite allergen and sensitization was shifted down in the presence of higher levels of endotoxin [46].

In a small study from New Zealand, levels of endotoxin were lower in the homes of farmers than local non-farmers, and allergic diseases (although not atopy) were more prevalent among farmers' children than those of their neighbours [47].

Summarizing the epidemiological data collected in the farming environment, a range of factors have been associated with a reduced risk of atopy but none is common to all studies, and as often factors are shown to have independent effects in multivariate analyses, there may be more than one effect.

Endotoxin in the urban domestic environment

Studies of endotoxin in the domestic environment in relation to atopy and asthma are summarized in Table 2. The first suggestion that environmental endotoxin exposure may be protective again the development of the allergic phenotype came from a study of 61 high-risk infants [50]. Endotoxin levels were significantly lower in homes of children with allergies than those without. Furthermore, there was a correlation between household endotoxin levels and production of IFN-γ by CD4 and CD8 T cells in a subgroup of these children. However, the protective effect of domestic endotoxin exposure could not be reproduced in the much larger population-based birth cohort (LISA) aged 2 years in Germany [51]. The PIAMA study from the Netherlands examined the relationship between endotoxin exposure and atopy at age 4 years in children taking part in a mite avoidance study, but could find no association [52]. Analysis of the effect of domestic exposure to endotoxin in a subgroup of the MAS 90 study revealed no association between this exposure and atopy (or asthma) in schoolage children [53]. A study of two small populations from Estonia and Sweden identified that although the levels of endotoxin were generally higher in Estonia, a protective effect of high personal exposure on atopy was seen only in Swedish children [54]. A cross-sectional study of children in Germany (INGA) identified a weak protective effect for endotoxin, but this was only significant for sensitization to two or more allergens [55]. A case–control study of teenage children from Cyprus identified atopy to be associated with increasing exposure to endotoxin [56]. A small case–control study from Palestine found endotoxin to be lower in the homes of atopic children [57]. A nested case–control study of adults from the ECRHS identified exposure to endotoxin to be protective for allergy [58].

Table 2.   Association between endotoxin and atopy – studies conducted in an urban environment
Author, year,
country, acronym
Numbers, design, ageDefinition of exposure; unitsOutcome
FindingEndotoxin protective,
no effect or a risk for
Gereda [50], 2000, USAN=61, wheezing infants, cohort, age 6–24 monthsDomestic exposure (1 sample from living room floor, kitchen floor, sofa, bedroom floor and cot mattress); EU/mLAtopy (spt)Endotoxin levels were lower in the homes of children with allergies than those withoutProtectiveOnly 10 children were atopic
Bolte [51], 2003, Germany, LISAN=2000, population-based birth cohort, age 2 yearsDust from mothers mattress at age 3 months; EU/gAtopy (sIgE)No association with atopy when analysed in quartiles of exposure, e.g. OR 0.85 (0.54–1.35)No effectIncreased repeated wheeze at higher exposures
Douwes [52], 2006, the Netherlands, PIAMAN=696 (287 with IgE), intervention arm of PIAMA, high-risk children, age 1 and 4 yearsDust from infants mattress and living room floor at age 3 months; EU/m2Atopy (sIgE)Levels in living room floor were not associated with atopy, OR∼1No effectLevels in mattress were low and not associated with any outcome. High levels in living room floor were protective for wheeze
Lau [53], 2005, Germany, MAS90N=153, age 10 yearsDust from child's mattress age 10 years; EU/mgAtopy (sIgE)No association with atopyNo effectThis results is mentioned only in the discussion and is not the focus of the paper
Bottcher [54], 2003, Estonia and SwedenN=108 from Estonia, N=111 from Sweden, age 2 yearsDust from infants mattress and a carpet in first year of life; EU/mgAtopy (spt)High endotoxin was protective in Sweden, OR 0.48 (0.35–0.9), no effect in EstoniaProtective/no effectEndotoxin levels were generally higher in Estonia where prevalence of allergy is low compared with Sweden where prevalence is high
Gehring [55], 2002, Germany, INGAN=444 (nested atopy case–control study), age 5–10 yearsDust from living room floor; EU/m2Atopy (sIgE)Higher endotoxin was protective, OR 0.8 (0.67–0.97)ProtectiveNo effect seen on asthma, hayfever or eczema
Gehring [59], 2007, Europe, Airallerg (combining GINI, LISA, PIAMA, Bamse)Nested case control for atopy within each cohort, age 2–4 yearsDust from child's mattress and living room floor at age 3 months; EU/m2 and EU/g (exposure was measured 1–4 years after sensitization was measured)Atopy (sIgE)Higher endotoxin was protective when all populations were combinedProtectiveIncreasing mattress dust was also protective; this remained the only significant factor after mutual adjustment
Nicolaou [56], 2006, CyprusAsthma case control, N=128, age 15–16 yearsDust from child's mattress; EU/m2 and EU/gAtopy (spt)Higher endotoxin was associated with an increase in risk of atopy on skin test, OR 1.6 (1–2.4)Risk 
El-Sharif [57], 2006, PalestineN=109, nested case–control study within ISAAC, age 6–12 yearsDust from child's mattress and living room floor; EU/gAtopy (spt)Endotoxin concentrations were higher in the living room floor amongst non-atopic controls, OR 0.02 (0.002–0.3)Protective 
Gehring [30], 2004, GermanyN=350, nested case–control study within ECRHS, age 25–50 yearsDust from living room floor; EU/m2 and EU/gAtopy (sIgE)Higher endotoxin was protective against more severe allergy, OR 0.72 (0.56–0.92)ProtectiveAdults rather than children indicating that current exposure as well as early life exposure may be important

In summary, several studies have identified a weak effect of high levels of endotoxin within homes, reducing the risk of atopy, but several found no association and one identified an increased risk. The reasons for this failure of replication may include Type 1 error of the initial studies, measurement error of endotoxin, measurement of the wrong environmental exposure or failure to take account of the genetic context of the environmental exposure.

CD14, Toll-like receptor 4 and lipopolysaccharide-induced signal transduction

Because exposure to high concentrations of LPS in vivo can cause catastrophic circulatory collapse and death, much effort has gone into the deciphering of the LPS signal transduction pathway to identify targets for sepsis therapies. Although exposure to inhalant endotoxin in the domestic environment is low grade in comparison, it is likely that the mechanisms of host response are the same. In serum, LPS binds to the serum protein LPS-binding protein (LBP) [60]. The established endotoxin response pathway has the trimolecular complex of CD14/TLR4/MD2 at the core.

The human CD14 gene is located on chromosome 5q31.1, a region of linkage to asthma and associated allergic phenotypes. It consists of two exons and codes for a 375 amino acid protein with multiple leucine repeats. CD14 is expressed on the surface of myeloid cells (macrophages and monocytes) −mCD14 and is also present in the serum in microgram amounts −sCD14.

CD14-deficient mice that lack both soluble and membrane-bound CD14 are ∼100-fold less sensitive to shock induced by LPS [61]. Using anti-CD14 monoclonal antibodies in human macrophage models, it was demonstrated that CD14 is a receptor for the LPS–LBP complex and is necessary for LPS-induced activation [62]. However, CD14 is a glycosylphosphatidylinositol-anchored protein and as such lacks a transmembrane domain and an intracellular signalling phase [62]. Subsequent studies have indicated that TLR4 is the signalling receptor for LPS [63] and that the presence of MD2 with TLR4 is essential for this signalling [64], indicating a core tri-molecular complex at the heart of the innate immune response to LPS. More recent work indicates that different combinations of other receptors may be recruited to an ‘activation cluster’ (including hsp70, hsp90CXCR4 and CD55) within the lipid raft of the cell membrane, whose combination may be specific to the type of LPS present [65] and the cell type. An elegant description of events is that LBP catalyses the transfer of LPS to mCD14; signalling molecules are then recruited to the site of the CD14–LPS complex in the cell membrane (within the lipid raft or microdomain). LPS is then released onto the lipid bi-layer and binds to the TLR4/MD2-based complex of receptors [66], resulting in triggering of different intracellular signalling cascades (nuclear factor-κB, ERK and JNK) depending on the specifics of the cell type and dynamic receptor cluster. Enhanced responsiveness to bacterial pathogens in early life, mediated by CD14, could elicit increased IL-12/IL-18 expression by innate immune pathways that may favour Th1 rather than Th2-type responses and reduce the likelihood of mounting an IgE response to environmental allergens.

The crystal structure of the CD14 molecule (molecular weight 55 kDa) indicates that it is a dimer comprising a horseshoe-shaped structure with a hydrophobic pocket at the NH2-terminal side [67]. Mutagenesis studies indicate that lipid A binds to this pocket and the hydrophilic carbohydrate moiety of LPS binds to grooves in the neighbouring area [67].

A single-nucleotide polymorphism (SNP) in the promoter region of CD14 rs2569190, a C-to-T transition at −260 bp from the translation start site and at −159 bp from the transcription start site (known both as CD14/−260 and CD14/−159) has been associated with sCD14 levels, with levels being the highest among T allele homozygotes [68–70]. This is a common SNP with a minor allele frequency of ∼48%, resulting in heterozygosity being present in ∼50% of the population, the remainder being fairly evenly divided between CC and TT homozygotes. It is plausible that changes in CD14 expression may affect the host response to LPS and hence alter the balance in immune responses (Fig. 1).

Figure 1.

 Fitted predicted probability curves for allergic sensitization at age 5 years in relation to environmental endotoxin load in children with CC, CT and TT genotypes in the promoter region of the CD14 gene (CD14/−159 C to T) derived from the logistic regression analysis. Reprinted with permission from [54].

CD14 genotype and atopy

The first study investigating the relationship between the CD14/−260 genotype and allergy identified that among allergic children, those with the CC genotype had higher total IgE and more positive skin tests, i.e. among allergic children, CCs were more allergic [68]. A longitudinal cohort study from Australia identified the C allele to be a risk for allergies, but any effect was lost by adulthood [71]. A study from the Netherlands found the C allele to be a risk for higher total IgE and more positive skin tests among allergic subjects, but not for asthma [72]. However, three studies of German children have failed to replicate this finding [73–75]. Others have reported no association, and found no effect on meta-analysis [76]. A study of the Hutterite founder population identified the T allele to be the risk allele for allergy [77], as did one high-risk birth cohort for eczema and high total IgE [78]. A summary of published studies is provided in Table 3.

Table 3.   Summary of studies on relationship between CD14 genotype and allergic outcomes
StudyPopulationOutcome measureAssociationRisk alleleComment
Baldini [68]481 children from 4 ethnic groups in Tucson Children's respiratory studyAtopyNo association (data not shown)NoneTT children with allergies are ‘less allergic’ and have higher sCD14
 Total IgE in children with positive skin testsLower in TT group, TT 81 (52–128) vs. CC 168 (107–264), P=0.02C 
 Number of positive skin testsin atopic childrenFewer in TT group, TT 1.78 (1.4–2.2) vs. CC 2.77 (2.2–3.3), P=0.006C 
Sengler [75]German children form MAS90 study, N=558Asthma, rhinitis, eczema, polysensitization, IgENo association, e.g. number of positive skin tests in atopic children TT 2.93 vs. CC 2.44None 
Heinzmann [73]Asthma case–control study, children, N=352AsthmaNo association with CD14 genotype, e.g. T allele carried by 46% of controls and 48% of asthmaticsNoneControl were not tested for asthma
Leung [79]Chinese children asthma case control, N=350Total IgE in atopic childrenHigher in CC genotype, TT 2.58 vs. CC 2.82 (log scale), P=0.02C 
Kabesch [74]German children form ISAAC study N=2048Total IgE, number of positive skin tests, asthma, rhinitis, eczemaNo association, e.g. asthma present in 11% of CC and 11% TTNoneHigher sCD14 in TT group
O'Donnell [71]Cohort of Australian children, N=305, followed from age 8 years to age 25 yearsAtopyAssociated with CC genotype in childhood OR 2.0 (1.1–3.9), P=0.04CCC associated with atopy and AHR in childhood, with much of the effect lost by adulthood
 AHRAssociated with CC genotype at age 8 years, with trend at other time-points OR 2.6 (2.1–5.6) P=0.02C 
Wang [80]Taiwanese children with asthma, N=190High IgE in children with asthmaT allele associated with high IgE, OR 1.56 (1.03–2.36) P=0.03TIn haplotype analysis effect was only seen in those with a particular haplotype containing a microsatellite marker
Woo [81]Asthma and food allergy case control, N=175Non-atopic asthma, food allergyHigher frequency of T allele relative to controls, OR 2.7 (0.9–8.0) for non-atopic asthmaTLower rate of T allele in controls than in other published studies (39%)
Litonjua [78]High risk birth cohort, N=379EczemaMore common in carriers of T, OR 2.3 (1.4–3.8)T 
 Total IgELower in CCT 
Gao [82]British N=300 and Japanese N=200 asthma case controlAsthma, atopyNo associationNone 
 Total IgE in BritishHigher in CC, TT 21.9 vs. CC 79.4, P=0.02C 
Ober [77]Hutterites, N=693, founder populationPositive skin testsMore common in T allele carriers, P=0.009T 
Koppelman [72]Dutch adults with asthma and their spouses, N=317Total IgE in subjects with positive skin testsHigher in CC group, CC 163 (105–253) vs. TT 106 (67–168)C 
 Number of positive skin testsHigher in CC groupC 
 HayfeverMore common in CC group, OR 1.8 (1.1–3.0)C 
 MD asthmaNo associationNone 
Buckova [83]Case control Czech adults with asthma or rhinitis, N=882Asthma, rhinitis, total IgE, lung functionNo associationNone 
 Sensitization to mouldsAssociated with C alleleC 
Kedda [76]Australia, adults, asthma case control, N=1011Asthma, atopyNo associationNoneDid meta-analysis and found no association
Sharma [84]Case–control study of adults with atopic asthma from India, N=414Atopic asthma, higher IgEC allele present in 48% of cases and 38% of controlsC 
Lachheb [85]Case–control study of children with asthma, Tunisia, N=434AsthmaTT less likely to have asthma, OR 1.6 (1.2–2.2)COR for asthma in C allele carriers was 1.6

The results demonstrate a similar amount of heterogeneity in the relationship between CD14 and atopy as between endotoxin and atopy. The logical next step was to consider the relationship between CD14 genotype, endotoxin exposure and atopy.

Endotoxin, genotype and atopy

There are four published studies of the association between CD14 and allergic outcomes in the context of measured domestic endotoxin exposures, covering both adults and children and different ethnic groups, all with broadly similar findings (Table 4). Within the setting of a UK population-based birth cohort study, we have shown that exposure to high levels of endotoxin in the home is protective against the development of allergic sensitization and eczema but only in C allele homozygotes (comprising 28% of the population) [70]. At low levels of endotoxin, the same subjects were at an increased risk of allergic sensitization and eczema relative to the other two genotype groups. Therefore, level of endotoxin exposure was important in determining allergic sensitization but only in one genotype group, comprising a quarter of the population. Among adults in Barbados, high levels of endotoxin in the home were associated with a higher risk of asthma in the TT group (which also indicates that at high levels of endotoxin, carriers of the C alleles were at a decreased risk of asthma) [86]. In rural and farming children in Germany, children with the CC genotype and with high domestic exposure to endotoxin have less specific IgE (independent of the effect of animal exposure) [87]. Among pregnant women recruited to form a new birth cohort study in the United States, those with high levels of endotoxin in the home and with the CD14 genotype CC showed a strong trend towards lower total IgE levels [88]. In a subgroup of the offspring assessed at age 1 year, a similar finding was obtained, i.e. those with the CC genotype who were exposed to higher levels of endotoxin in the home had lower total IgE measurements [89].

Table 4.   Summary of studies reporting interaction between measured endotoxin exposure and CD14 genotype
Author, year,
acronym, country
Design, age,
Ethnic groupDefinition of
exposure; units
between CD14
genotype and
Relationship between Endotoxin
and outcome
Interaction between
CD14 genotype and
  1. MAAS, Manchester Asthma and Allergy study; BAGS, Barbados Asthma Genetics Study; WHEALS, Wayne County Health Environment, Atopy and Asthma Longitudinal Study; ALEX, Allergen and Endotoxin Study; AS, allergic sensitisation; A, Asthma; E, eczema; NA, no association; NR, not reported; GM, geometric mean; EU, endotoxin unit.

Simpson [70], 2006, MAAS, UKPopulation based birth cohort study, age 5 years, N=442Mixed European ancestry onlyLiving room floor, GM 2856 EU/m2 (16.1 EU/mg)AS: NA
AS: increasing endotoxin associated with decreased AS
AS: In CC only higher endotoxin associated with decreased risk of AS
A: In CC only higher endotoxin associated with increased risk of non-atopic wheeze
E: In CC only higher endotoxin associated with decreased risk of eczema
Zambelli-weiner [86], 2005, BAGS, BarbadosAsthmatic probands and their families, N=443 (adults and children)Of African descent/European admixtureLiving room composite endotoxin load – dichotomous (75th percentile), GM 23 144 EU/m2AS: NA
A: TT protective
A: more common in TT with high endotoxin (i.e. at high exposures C allele is protective)
Williams [88], 2006, WHEALS mothers, USAMothers of population based birth cohort (young adults >21 years), N=517All races analysed togetherDust from the home at age 1 month (child), GM 16.2 EU/mgAS: TT had lower total IgE
AS: In CC only, higher Endotoxin associated with lower total IgE
Williams [89], 2008, WHEALS, USAPopulation based birth cohort study, age 1 year
All races analysed together, stratifiedDust from the home at age 6 month (child), GM 18.2 EU/mgAS:
AS: In CC only, higher Endotoxin associated with lower total IgE
Eder [87], 2005, ALEX, Germany and AustriaChildren living on farms and local controls, N=624NRDust from child's mattress, GM in tertiles (2nd tertile 12 495–30 046 EU/m2)AS: NA
AS: high exposure protective
AS: C allele protective in the highest endotoxin tertile only

Others have looked for interactions between environmental exposures other than measured endotoxin and CD14 genotype (e.g. being brought up on a farm or owning pets in childhood). These studies are summarized in Table 5 and are much more heterogeneous, for example pet ownership among T allele homozygotes appears to be protective, whereas the effects of farming measured in different ways are varied. This highlights the need to objectively measure the environmental exposure, as lifestyles, e.g. living in Russia or being brought up on a farm, are associated with multiple different, possibly conflicting exposures.

Table 5.   Studies of environmental exposures (other than endotoxin) interacting with CD14 genotype
Author, year, acronym,
Design, age, numbersEthnic groupDefinition of
exposure; units
Relationship between CD14
genotype and outcome
Zhang [90], 2009, Karelian Finns and RussiansAdult females, N=604KareliaRussia vs. Finland (assumes Russians have higher exposure to microbial load)TT risk for eczema in Finnish and protective in Russians,Assuming Russians are exposed to higher endotoxin, this does not fit proposed hypothesis; the difference is seen in the TT group, with Russian TTs having lowest prevalence
Smit [91], 2007, Danish farmersFarming school students, case–control study for new onset asthma, N=188Mixed European ancestryAll farmersTT risk for atopy. NA with new onset asthmaPopulation was selected to investigate factors associated with new onset asthma. Rates of allergic sensitization were the same among cases and controls. Assuming all are exposed to high levels of endotoxin, these data fit the current hypothesis
Smit [92], 2009, France, EGEACase–control and family study of adults and children, N=825FrenchCountry livingCC protective against atopic wheeze in those who live in the countrySuggest CC is protective genotype at presumed higher endotoxin exposures, i.e. fits current hypothesis
Bottema [93], 2008, PIAMA, PREVASC, KOALA, The NetherlandsChildren for 3 birth cohort studies, N=3062DutchPet ownershipC allele without pets had lower total IgE levels, and less sensitisation to any allergenSimilar to results for pet exposure shown by Eder et al., which were independent of the endotoxin effect
Leynaert [94], 2006, ECRHS, FranceN=600, young adultsFrenchFarm exposure in childhoodTT protective against atopy in those exposed to farming environmentAssuming farmers are exposed to higher endotoxin does not fit proposed hypothesis; farming exposure was defined as at least 3 months on a farm during childhood. Farming was not protective against asthma but was against atopy and nasal allergies. Overall CC had higher total IgE, TT had less atopy
Gern [95], 2004, COASTN=285, high-risk birth cohort study, age 1 yearUSAPet exposureTT+dog was protective against eczemaEczema was most common in heterozygotes. Only 17 children were included as dog owners with TT genotype
Eder [87], 2005, ALEX, Germany/AustriaChildren living on farms and local controls, N=624Germany/AustriaPet exposureTT+pets was protective against specific sensitisation and was associated with lower total IgEThe effect of stable animals was different, whereby TT+stable animals conferred the highest risk, similarly TT plus high endotoxin

In all four populations where endotoxin was objectively measured, higher levels were protective, but only among carriers of the C allele. Use of luciferase reporter assays in human monocytes to assess the transcriptional activity of the two variants indicated that although the wild-type C allele demonstrated strong basal transcriptional activity, this activity was 32% higher in the variant T allele (in keeping with T allele homozygotes generally having higher circulating levels of sCD14) [96]. Following stimulation by LPS, CD14 expression was enhanced in a proportionate manner in both assays, i.e. production from transfected cells containing the T allele was still greater. The difference in the transcriptional activity between the two alleles was explained in further studies by their relative affinity for the transcription factors SP1 and SP3. A further study in humans confirmed that T allele homozygotes had higher circulating levels of sCD14 at baseline, but then performed an inhalational challenge with low-dose (20 mcg) endotoxin [69]. Levels of sCD14 increased more in the carriers of the C allele, such that there was no demonstrable difference in sCD14 between genotype groups after the challenge, suggesting that carriers of the C allele were more responsive to exposure to endotoxin than T allele homozygotes.

Toll-like receptor 4

There are two common co-segregating mis-sense mutations in exon 3 of this gene (Asp299Gly and Thr399Ile) that result in amino acid changes in the extracellular domain of the protein and have been widely studied. These variants have been associated with a blunted response to inhaled endotoxin on bronchial challenge testing [97] and a reduced systemic inflammatory response to low-dose inhaled endotoxin [98]. However, others have failed to confirm this [99, 100].

Variants in TLR4 have not been associated with asthma or outcomes associated with lung function in any study [91, 101, 102]. One study suggested that these variants are associated with more severe atopy but found no association with asthma [103]. A single study has been published on the interaction between endotoxin exposure in the home (in tertiles) and the Asp299Gly variant and found that higher endotoxin appeared to be protective for the variants against wheeze but not against allergy [102]. However, a significant effect was seen in the middle tertile for BHR and wheeze and the higher tertile for asthma, and so the results were not very consistent.


We have presented evidence to suggest that endotoxin is a plausible protective factor for atopy in the context of the hygiene hypothesis and that although it is difficult to compare levels between studies, within studies, this factor can be reliably measured in the domestic environment and is relatively stable over time. CD14 forms part of a pattern recognition receptor for endotoxin and a functional SNP in the promoter region of this gene may confer differential responsiveness to endotoxin (from in vivo studies). There is a similar degree of heterogeneity in results from studies relating domestic endotoxin exposure to atopy and CD14 genotype to atopy. However, once endotoxin exposure is considered in the context of the CD14 genotype, findings are much more consistent – higher levels of endotoxin were protective against the development of atopy, but only among carriers of the C allele of CD14/−260.

It is possible that endotoxin is a marker for hygiene rather than the functional molecule and that another exposure such as muramic acid, glucan or animal contact is the active factor. It is possible that another SNP in linkage disequilibrium with CD14/−260 is the functional SNP or that this is a haplotype or a pathway effect. Both these possibilities require further investigation. Further studies to identify such interactions need to be well designed with adequate numbers to ensure power while incorporating objective measures of relevant environmental exposures.

It is also likely that others have attempted to reproduce this finding and failed, but due to a negative publication bias, these results are not publically available. However, the fact remains that this interaction has been demonstrated in four geographically distinct populations and as such is one of the most replicated associations in the field of asthma and allergy epidemiology. Most importantly, the interaction is biologically plausible. We propose that neither the environmental exposure nor the genotype in isolation is sufficient to cause complex diseases like asthma and atopy, but disease results from the one acting in the context of the other, in which CD14 and endotoxin is merely one example contributing a fraction of the risk for atopy. The challenge for investigators now is to identify the others.