|Home | About | Journals | Submit | Contact Us | Français|
In addition to genetics, there appear to be a number of environmental variables that impact on allergic risk. Meta-analyses of epidemiological studies presented in this paper demonstrate a correlation between specific ambient exposures (i.e. livestock, pets, endotoxin, and unpasteurized milk ingestion) and a reduction in allergic risk during childhood. Additional laboratory investigations discussed in this review characterized the intrinsic immunostimulatory activities of living environments. Considered together, results of these investigations suggest a novel paradigm by which early life home exposures to microbial products and other allergen non-specific immunostimulants modify allergic risk.
Over the last half century, allergic diseases have become far more common in industrialized countries, while atopy rates remain low in most of the third world[1–3]. Although reasons for these trends remain speculative, the rapidity with which allergic disease prevalence rates have increased in affected countries strongly suggests environmental factors have had a dominant role. Therefore, there is a great deal of interest in determining which ambient exposures are responsible for the low and high allergic disease prevalence rates of poor and affluent countries, respectively.
Allergen exposures are clearly required for the development of Th2 biased hypersensitivities. For some allergens (i.e. cockroach and house dust mite), the risk of developing hypersensitivities has been found to increase considerably when the home allergen burden increases above quantifiable threshold levels[4–6]. However, for other allergens (i.e. dogs, cats), increased levels of home exposure appear linked to a decreased risk of sensitization, both to the allergen of interest and to other unrelated allergens[6,7]. These and other lines of investigation suggest that aside from allergens themselves, living environments contain additional molecules that influence the immunological balance between allergen specific tolerance and hypersensitivity.
Epidemiological studies have found that environmental variables linked to life style, i.e. urban versus rural living), diet, exposures to diesel exhaust and other man-made pollutants, and infectious and noninfectious exposures to microbes, influence allergic risk. One consistent finding derived from these studies is that children raised on farms are less likely to develop allergic diseases than children raised in cities [8,11,12]. In a previously published meta-analysis of 9 pertinent studies, we found an odds ratio (OR) of 0.74 with 95% confidence intervals (CIs) of 0.61–0.91 for allergic stigmata in children raised on farms compared to children raised in non-farming environments . However, the reason(s) why farm living reduces allergic risk remain speculative. As houses located on farms, particularly those with livestock, are rich in their microbial content [8,14–16], it has been suggested that microbial stimulation educates host immunity in a manner that prevents dysregulated immune responses to ambient allergens. This theory, “The Hygiene Hypothesis”, is also supported by investigations in which other variables linked to microbial exposure, including pet ownership, family size, day care attendance (community acquired infections), vaccination status, antibiotic use, animal exposure, and infectious disease history were found to influence allergic disease risk[1,8,17].
Ever since children raised in farming communities were shown to enjoy reduced atopy rates [8,11,12], researchers have tried to identify specific farm associated exposures that might protect against the genesis of allergic diseases, both in rural and urban settings. Although results have not always been consistent, a large number of epidemiological studies suggest that regular exposures to livestock and pets, unpasteurized milk consumption during childhood, and elevated home endotoxin levels, protect against the allergic march. To better assess the real impact of these exposures on pediatric allergic risk, we conducted meta-analyses of all relevant studies published between 1966 and 2008.
Initially, MEDLINE searches were conducted to identify pertinent articles, using the following search commands: “(atopy, allergy, asthma, eczema, wheeze, or rhinitis) and (endotoxin, dog, cat, livestock, or unpasteurized milk)”. This search identified 6758 papers of potential relevance for these meta-analyses. All abstracts were reviewed independently by two investigators. Abstracts obviously unrelated to the topic at hand were discarded. For the rest, copies of full articles were retrieved and reviewed. To be included in these meta-analyses, investigations were required to meet the following criteria: (1) they had to have assessed for associations between relevant environmental exposures during childhood and atopic risk, (2) they were considered of compatible design with other studies included in the analyses, (3) results were reported as ORs to facilitate the execution of meta-analyses, and (4) their design and quality of their data sets were deemed adequate and appropriate, based on descriptions provided in the text of the paper.
Of the initial 6758 papers identified in the MEDLINE data base with our search commands, 6712 were excluded because they were considered irrelevant, incompatible for comparative analyses with other selected studies, and/or did not report results as ORs. Forty-six reports were considered appropriate for the conduct of meta-analyses to determine if correlations exist between childhood incidence/prevalence rates for allergic manifestations and pet ownership (Figure 1A), regular contact with livestock (Figure 1B), unpasteurized milk consumption (Figure 1C), and home endotoxin levels (Figure 1D).
Using ORs and CIs reported within individual investigations, we calculated pooled effects estimates (ORs and 95% CIs) using both fixed and random-effects models. Heterogeneity, a value which uses a chi-square test to determine goodness-of-fit, was calculated for each fixed-effects model. Significant heterogeneity was found between studies (P<0.1) with fixed-effects modeling. Therefore, random-effects models were selected to calculate summary ORs and CIs presented within this paper, as these estimates tend to be more conservative, taking into account between-study and within-study sampling variability. Fixed-effects and random-effects models were run with the R statistical software package (Vienna, Austria) using the “rmeta” command.
Twenty-seven studies were considered appropriate for inclusion in a meta-analysis of associations between pet ownership and the development of allergic stigmata (Figure 1A). In these investigations, atopic wheeze, eczema, allergic rhinitis and/or conjunctivitis symptoms were used as criteria for defining atopy. Several studies found no effect or a small positive effect of regular pet exposures on allergic risk, and 2 studies found a marked increase in the incidence of allergic stigmata for subjects raised in homes with cats. Nonetheless, the combined OR for all 27 studies was 0.86 (95% CI: 0.79 – 0.93), suggesting that pet ownership during childhood leads to an approximate 14% decrease in allergic risk. While this clinical effect appears small, as the CIs for the meta-analysis did not cross 1, it is considered statistically significant. Similar results were obtained when evidence of allergen specific IgE was used to define atopy (n= 20 studies; OR: 0.84, CI: 0.73–0.96). Interestingly, when meta-analyses for dog (OR: 0.76, CI: 0.65–0.89) and cat (OR: 0.97, CI: 0.87–1.1) were conducted separately, dog ownership appeared more protective against the genesis of allergic diseases than cat ownership. Potential explanations for the discordant OR values found for dog and cat ownership include variability in experimental design and/or other idiosyncrasies unique to each study. Alternatively, dog and cat exposures may have distinct immunological effects on allergic risk.
We identified 8 studies that compared the prevalence of allergic manifestations in children living on farms with livestock and children without regular livestock exposures. Atopy criteria used to assess for this association were the same as those used for the Figure 1A meta-analysis. The OR for clinical manifestations of atopy 0.58 (CI: 0.39–0.87) was significantly reduced for children with regular livestock exposures, compared to control children, representing a 42% reduction in allergic manifestations (Figure 1B). While not statistically different, the OR for developing atopic diseases was lower for children raised with livestock than for children raised with pets, suggesting that farm animals or other associated factors are more protective.
Many children raised on farms with livestock have the opportunity to drink unpasteurized milk on a regular or occasional basis. While we could identify only 7 studies that considered the influence of unpasteurized milk consumption during childhood on allergic risk, all of them found a protective effect (Figure 1C). Children drinking unpasteurized milk during the first few years of life had an OR of 0.68 (CI: 0.61–0.76) for allergic stigmata or a 32% reduction in allergic risk, compared to children who never drank unpasteurized milk. While not proven, considered in conjunction with previous meta-analyses, this finding suggests that unpasteurized milk consumption might contribute to the protective influence of being raised on a farm with livestock.
In previous studies in which pet ownership was found to protect against the allergic march (Figure 1A), it was also shown that household pets increased home endotoxin levels. Given the important role toll-like receptors (TLRs) play in immune regulation and the reported TLR4 dependence of endotoxin induced immune responses , this observation has received a great deal of attention from epidemiologists and laboratory based scientists interested in the origins of allergic diseases. However, in our meta-analysis of 13 pertinent studies, the OR for allergic stigmata was only reduced to 0.90 (CI: 0.78–1.0) or 10% for children living in homes with high rather than low endotoxin levels (Figure 1D). This relatively weak association suggests that either endotoxin is not an important environmental variable with respect to its influence on allergic risk, or that additional microbial products ubiquitous in living environments have an equally import and potentially confounding influence on the genesis of allergic diseases.
Previously discussed epidemiological studies suggest that environmental exposures during the first years of life play an important role in immune homeostasis and in determining allergic risk throughout childhood. A majority of time is spent at home during early life and accumulating experimental evidence suggests that living environments have a major educational influence on developing immune systems. Nonetheless, the molecular basis for immunomodulation by ambient exposures and understanding of their downstream influence on allergic risk remain highly speculative. By design, a majority of investigations aimed at characterizing how living environments affect host immunity have made a priori assumptions about which exposures to pay attention to and which to ignore. As an alternative, we reasoned that the immunological “ether” associated with homes might be better understood by investigating clinically relevant, sterile, but unpurified environmental samples. Logic suggests that gravity should concentrate immunostimulatory particulates into settled dust and endotoxin levels have previously been found to be predictive surrogate markers of allergic risk (Figure 1D). Therefore, our laboratory has begun to characterize the immunostimulatory activities of sterile house dust extracts (HDEs). Studies conducted to date have yielded provocative and reproducible results, which will be the focus of the following sections in this paper [61,62].
Dust samples were first collected from the bedrooms of 15 suburban homes in San Diego, California and then processed by standardized techniques, which included suspension in PBS, physical agitation, and sterile filtration. These HDEs were found to be sterile and non-toxic. In initial experiments, HDEs were shown to activate bone marrow derived dendritic cells (BMDDCs) in a concentration dependant manner [62,63]. Moreover, higher concentrations of most HDEs and optimized concentrations of TLR ligands elicited similar levels of IL-6 production. In contrast, LPS (TLR4) and immunostimulatory sequence oligodeoxynucleotide (ISS; TLR9) induced stronger IL-12p40 responses than any of the HDEs investigated. In a subsequent study, we determined whether a sampling of HDEs elicited the production of bioactive IL-12 (IL-12p70), a heterodimer of IL-12p40 and IL-12p35. However, HDE induced BMDDC secretion of IL-12p70 was weak compared to responses induced by LPS, ISS, and R848 (TLR7) and similar to the response elicited by Pam-3-Cys (TLR2). Although relatively ineffective at stimulating IL-12p70 production, in unpublished experiments we recently observed that HDEs potently induce IL-23p19 mRNA synthesis, suggesting that HDEs may preferentially promote the synthesis of bio-active IL-23, a heterodimer of IL-12p40 and IL-23p19, rather than IL-12p70. Purified TLR ligands and HDEs also elicit low levels of BMDDC IL-10 production, while IL-4, IL-13 and TNF-α were not detected in any culture supernatants[62,63].
In additional studies, HDE regulation of BMDDC co-stimulatory molecule expression was assessed. BMDDCs stimulated with HDEs displayed increased expression of CD40, CD80, CD86 and MHC Class II compared to unstimulated BMDDCs. Moreover, co-stimulatory molecule expression levels were similar on BMDDCs activated with HDEs and purified TLR ligands. In unpublished investigations we further established that like BMDDCs, murine splenocytes and human PBMCs were highly responsive to HDEs. Taken together, these observations demonstrate that HDEs can be prepared with standardized methods and that their bioactivities can be readily investigated with traditional laboratory techniques.
Consistent with other studies, we found the mean endotoxin content of house dust samples obtained from homes with pets (n=7) was more than twice that of house dust samples obtained from homes without pets (N=8) . In addition, while mean IL-6 responses were similar, HDEs from homes with pets elicited IL-12p40 responses that were 60% stronger on average than those of HDEs from pet free homes. In further analyses, correlations between HDE endotoxin levels and BMDDC cytokine inducing capacities were assessed. Considered separately, HDEs from homes with and without pet exposures had correlation coefficients (r values) above 0.5, but they were not statistically significant by Z testing. However, while r values were not strengthened, correlations between endotoxin levels and IL-6 (r=0.523; P=0.044) and IL-12p40 (r=0.573; P=0.024) inducing activities did reach statistical significance when all HDEs were considered together. Although the number of HDEs compared was small, these experimental findings support 3 major assertions: 1) compared to pet free homes, HDEs derived from pet exposure homes have increased levels of endotoxin, 2) HDE bioactivities correlate loosely but significantly with their endotoxin content, and 3) endotoxin is unlikely to be the only immunostimulatory molecule contained within HDEs.
To further evaluate the contribution of TLR4 in mediating responsiveness to HDEs, wild type (WT) and TLR4 knockout (ko) BMDDC responses were compared. TLR4 ko BMDDCs demonstrated a marked reduction in HDE (n=10) induced cytokine production and co-stimulatory molecule expression but residual responsiveness remained. In additional experiments, WT, TLR2 ko and TLR9 ko BMDDCs responses to HDEs were compared. While HDE stimulated TLR2 ko BMDDCs produced less IL-6 than WT BMDDCs, IL-12p40 production and co-stimulatory molecule expression were preserved. In contrast, HDE stimulated TLR9 ko BMDDCs were found to produce less IL-6 and IL-12p40 than WT BMDDCs. Furthermore, while TLR4 ko BMDDCs displayed a greater deficit, HDE activated TLR9 ko BMDDCs expressed lower levels of co-stimulatory molecules than WT BMDDCs. These observations support the view that in addition to TLR4, both TLR2 and TLR9 contribute to HDE mediated BMDDC responses.
Experimental findings presented thus far suggested that TLR signaling pathways play an important role in mediating HDE induced BMDDC responses. Nonetheless, these results did not exclude the possibility that HDEs might also activate BMDDCs by TLR independent pathways. Therefore, as MyD88 plays a critical role in signaling through all TLRs except TLR3[64,65], a final series of experiments compared cytokine production and co-stimulatory molecule up regulation by HDE activated WT and MyD88 ko BMDDCs. In these studies, HDE stimulated MyD88 ko BMDDCs produced only small amounts of IL-6 and IL-12p40 and increased co-stimulatory molecule expression only slightly. These results establish that TLR signaling pathways play a central role in BMDDC activation by HDEs.
In order to assess the adjuvant activities of HDEs, mice were intranasally (i.n.) immunized with ovalbumin (OVA) alone or with 21μ-l of HDE (100mg/ml; concentration prior to filtration) on 3 occasions, at weekly intervals. Additional groups of control mice were i.n. immunized with OVA and Pam-3-Cys, LPS, or ISS, according to the same vaccination schedule. While adjuvant potential varied, mice i.n. immunized with OVA and HDE had far stronger adaptive responses, than mice i.n. immunized with OVA alone, establishing that HDEs have adjuvant activities in the airways. Furthermore, HDEs (n=10) were consistently found to act as Th2 biasing adjuvants, as they induced strong allergen specific IgE and Th2 polarized cytokine responses but weak IgG2a and IFNγ responses. If fact, most HDEs studied were more potent Th2 adjuvants than Pam-3-Cys or low dose LPS, both of which have previously been described as Th2 adjuvants. Moreover, the adjuvant activities of HDEs were dependent on MyD88, further suggesting their dependence on signaling through TLRs. In addition to developing Th2 biased adaptive responses, mice immunized with OVA and HDE developed Th2 biased airway hypersensitivities, as reflected in their eosinophil rich airway inflammatory response and increased bronchial responsiveness to methacholine after i.n. OVA challenge. These results challenge the commonly held belief that microbial products in general, and TLR ligands in particular, protect against the allergic march by inherently favoring development of Th1 biased immune profiles.
Experiments just discussed might be construed to suggest that many, if not all, living environments intrinsically promote the development of allergic asthma. However, in these studies mice were airway exposed to the immunostimulatory contents of HDEs at weekly intervals and at levels likely to be in great excess of daily physiological exposures. In contrast, individuals are thought to inhale air laced with low concentrations of immunostimulatory elements, on a semi-continuous basis . Therefore, additional experiments were designed to better model real world exposures. In these investigations, mice received 3 weekly i.n. OVA immunizations, as in previously described experiments, while low dose HDE (1/7th weekly dose; 3μl) was i.n. delivered daily, beginning 1 week before the first and ending with the last dose of OVA, weekly with OVA (as in the previous experiments), or both.
Daily i.n. HDE delivery had little adjuvant effect on OVA specific responses. More importantly, daily airway HDE exposures prevented mice concurrently receiving weekly i.n. OVA and HDE (adjuvant dose) from developing both Th2 biased adaptive responses and experimental asthma. Additional studies demonstrated that both the Th2 adjuvant and tolerogenic activities of HDEs could be replicated with purified LPS. Further unpublished studies determined whether i.n. daily HDE/weekly OVA delivery induced long lasting allergen tolerance. In these studies mice received a series of 3 weekly i.n. OVA vaccinations either alone or with weekly adjuvant doses (21μl) or daily low doses (3μl) of HDE, as just described. One month after the last of the primary OVA immunizations, all mice were OVA sensitized by weekly i.n. OVA/adjuvant dose HDE delivery (3 doses). Mice receiving i.n. OVA and daily HDE during primary immunization were found to be highly resistant to Th2 sensitization, while mice in other primary immunization groups (OVA alone or weekly OVA with HDE) were not.
Recognizing that immunostimulatory molecules are ubiquitous in inspired air but that levels vary widely, these experimental results suggest a new paradigm by which ambient exposures might modulate airway immunity and allergic risk during the first years of life. According to this model, basal levels of daily exposure to endotoxin and other immunostimulatory materials present in ambient air are generally not sufficient to provide airway adjuvant activity but rather serve to attenuate innate responsiveness to these molecules. However, episodic exposures to ambient air laced with high concentrations of immunostimulatory molecules can provide sufficient adjuvant activity to induce a breakdown in allergen tolerance if prior immunologic dampening by basal exposures is inadequate. Although far from proven, this model provides an alternative view of how ambient environmental exposures to materials with Th2 adjuvant activities can paradoxically, also promote allergen tolerance.
Both epidemiological and laboratory investigations reviewed in this paper strongly suggest that ambient exposures to allergen non-specific immunostimulants have the potential to impact significantly on allergic risk. Nonetheless, understanding of the molecular variables and mechanisms responsible is far from complete. Studies discussed in the first half of this review demonstrate a correlation between pet, farm, animal, unpasteurized milk, and endotoxin exposures during childhood and a reduced incidence of allergic manifestations. However, as discussed, these epidemiological trends have been inconsistently reported, and in select studies associations were relatively weak, non-existent, or reversed. Moreover, these investigations provide little insight as to the mechanisms by which living environments influence allergic risk.
Laboratory investigations presented in the second half of the paper offer an alternative approach to characterizing how living environments modify host immunity in general and allergic risk in particular. In these studies, TLRs were found to play a central role in sensing and responding to allergen non-specific immunostimulatory molecules contained within HDEs and ubiquitous in living environments. Additional i.n. vaccination experiments revealed that weekly airway exposures to adjuvant doses of HDEs induced Th2 biased airway hypersensitivities to co-administered allergen, while daily HDE exposures promoted the development of long lived allergen tolerance. The implication of these observations is that the primary immunological consequence of airway exposures to allergen non-specific immunostimulants present in living environments is either to promote the development of Th2 biased hypersensitivities or allergen tolerance, rather than to drive the development of “protective” Th1 biased responses to allergens.
In additional experiments, we found that even the innate airway response to bolus HDE exposure (neutrophilic inflammation and cytokine release) is inhibited by pretreatment of mice with a week of daily i.n. low dose HDE delivery. The phenomenon of reduced responsiveness with repetitive exposure has previously been described with LPS tolerance and can be induced by other TLR ligands as well [67–69]. Moreover, in unpublished studies, we observed that daily i.n. HDE delivery increases local expression of mRNAs for molecules thought to mediate LPS tolerance (IL-10, STAT3, IRAKM, SHIP) [67,69–71]. These observations may explain why human lungs remain uninflamed despite continuous inhalation of pro-inflammatory molecules contained in HDEs. Furthermore, they suggest that mechanisms associated with LPS tolerance (innate immunity) may also play an important role in the physiological development of allergen specific tolerance by non-atopic infants and toddlers, a focus of ongoing investigations in our laboratory.
If regular and adequate TLR stimulation drives the development of immune and clinical tolerance to ambient allergens by mechanisms associated with LPS tolerance, then exposure levels for individual molecules could prove far less important than the net exposure level for all ambient immunostimulatory molecules in determining a child's allergic risk. This consideration may help to explain why epidemiological studies have yet to identify a specific molecule for which ambient exposure levels strongly and consistently correlate with relative allergic risk. Another implication of this view is that bioassays of HDE immunostimulatory activity could prove highly predictive of the allergic risk associated with living environments. We are currently testing this hypothesis in ongoing investigations.
This work was supported by grant AI61772 and T32RR023254 from the National Institutes of Health