SEARCH

SEARCH BY CITATION

Keywords:

  • 16S rRNA;
  • atopy;
  • bacteria;
  • childhood asthma;
  • environment

Abstract

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Conflict of interest
  8. Authors contributions
  9. References

Background

We have previously found an inverse association of bacterial diversity with childhood asthma. It remains unclear whether certain bacteria account for the protective effect.

Methods

The high variability of the bacterial 16S rRNA gene allows assessing diversity and specificity of bacterial communities by single-strand configuration polymorphism (SSCP). DNA was extracted from mattress dust samples of 489 school-age children from rural and suburban regions in Germany. A fragment of the bacteria-specific 16S rRNA gene was amplified by PCR, digested to single-strand DNA, and subjected to electrophoresis. The resulting band patterns reflect the underlying DNA sequences. The individual bands were tested for associations with asthma, hay fever, and atopy in quantitative and qualitative multivariable analyses. Significantly associated bands were isolated and sequenced. The sequences were compared to a database, and distinct bacteria were identified.

Results

Seven of 76 independent bands were found to be inversely associated with asthma, atopic sensitization, and hay fever with odds ratios ranging from 0.17 to 0.73. The bands contained the sequences of Acinetobacter sp., Lactobacillus spp., Neisseria spp., Staphylococcus sciuri, Jeotgalicoccus sp., Corynebacterium spp., and others.

Conclusions

In a diverse microbial environment, certain bacteria may account for the protective effect on the development of asthma and atopy.

After decades of successfully combating infections with vaccinations and antibiotics, beneficial health effects of microorganisms are gaining attention as the decline of infectious diseases is paralleled by an increase of allergic and autoimmune diseases [1]. For atopic diseases such as asthma and hay fever, there is conclusive evidence that environmental exposure to certain bacteria is inversely related to disease manifestation. This evidence is based on ecologic studies [2] and surveys using questionnaire data and individualized measurements of generic markers of bacterial exposure [3]. However, these approaches are limited by a potential ecologic fallacy, imprecise questionnaire data, and too broad categories of bacterial exposure such as Gram-negative vs Gram-positive bacteria. Classical culture-based methods fail to identify far more than 90% of all bacteria; hence, there is a need for detecting bacteria by molecular methods such as single-strand conformation polymorphism (SSCP) [4].

Recently, we have applied this method to mattress dust samples of children from rural regions in Bavaria, Germany, and found an inverse association of the diversity of bacterial exposure and childhood asthma [5]. Remarkably, this finding was paralleled by an inverse association of fungal exposure and asthma. To reduce the dimensions of the bacterial SSCP markers, we performed a factor analysis and identified two factors inversely related to asthma.

Here, we go beyond the previously published factor analysis and assess directly the genetic fingerprints of specific environmental bacteria in an exploratory approach. Furthermore, we extend the disease spectrum from asthma to hay fever.

Material and methods

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Conflict of interest
  8. Authors contributions
  9. References

Population

The PARSIFAL study was a cross-sectional survey on children of farmers, children attending Rudolf Steiner schools, and their respective reference groups [6]. In the German arm of the study, 9240 children were invited and 6963 children (75%) returned their questionnaires. For blood and dust analyses, subsamples of all children whose parents consented to the respective study modules were selected (Fig. 1). For a previous analysis of biomarkers, 182 dust samples enriched for children with atopic and nonatopic wheeze during the last 12 months had been extracted [7] and could therefore not be used for the SSCP analysis. As the prevalence of wheeze was higher in the reference children, the previous usage of samples resulted in an enrichment of farm children in the present analysis. Of the remaining 619 mattress dust samples, the amount of dust was sufficient for SSCP analysis in 489 children, of whom 255 were farm children. The study was approved by the ethical committee of the physicians’ board of Bavaria, and written informed consent was obtained from the children's parents or guardians.

image

Figure 1. Selection of the study population. RS, random selection; SRS, stratified random selection. The samples marked with asterisks are compared in Table 1.

Download figure to PowerPoint

Questionnaire and IgE measurements

The questionnaire included questions on sociodemographic background, family history of asthma and atopy, exposure to farm environment, and on child's health. Questions on health outcomes were derived from the ISAAC II questionnaire [8]. Children with reported physician-diagnosed asthma once or obstructive bronchitis more than once in their lifetime were defined as having asthma ever. Hay fever was defined as a reported lifetime diagnosis of hay fever or symptoms of itchy eyes or a runny nose without a cold during the last 12 months. Atopic sensitization was defined as any allergen-specific IgE value of 3.5 kU/L or greater in serum for common inhalant (Phadiatop) and food allergens (fx5; Pharmacia CAP System; Pharmacia Diagnostics AB, Uppsala, Sweden). A child who lived on a farm and whose family ran the farm was coded as being a farm child, whereas all other children were considered reference children.

Single-strand configuration polymorphism

DNA extraction [9] and SSCP [4] were performed as previously published. In brief, a fragment of the variable regions 4 and 5 of the 16S ribosomal DNA gene was amplified. PCR products were digested to single-stranded DNA and analyzed on SSCP gels, which were subsequently silver stained. The SSCP profiles were analyzed with the GelCompar II software (version 4.6; Applied Maths, Sint-Martens-Latem, Belgium). The SSCP gels were normalized using a standard present on every gel in two or three tracks. Bands of interest were excised from at least three individual tracks of the gels, amplified, and sequenced [4]. The DNA sequences were analyzed for similarities of at least 98% using the Ribosomal Database Project II for phylogenetic analysis (http://rdp.cme.msu.edu/seqmatch/seqmatch_intro.jsp).

Statistical analysis

Bands with maximum density units below 5 were considered noise and were not included in further analyses. Because of a skewed distribution and several zero values, the band density values were log-transformed after adding 1. For the quantitative analysis, robust logistic regression was applied. Additionally, a qualitative approach was explored by dichotomizing band densities at a low (≥5 density units) or high (≥10 density units) cut-off. Each band was tested for the respective health outcomes with adjustment for farming and potential confounders. All bands with P-values <0.05 were entered in a stepwise logistic regression. Because of the exploratory character of the analysis, corrections for multiple testing were not performed.

Results

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Conflict of interest
  8. Authors contributions
  9. References

Table 1 demonstrates that farm children were enriched in the sample selected for SSCP analysis. Consequently, characteristics related to farming such as maternal smoking or family history of atopy tended to differ between the samples included or not included in the SSCP analysis. Within the SSCP sample, farm and nonfarm children differed in the prevalence of asthma (OR = 0.56 [0.29–1.09]), atopic sensitization, diagnosis or current symptoms of hay fever, number of older siblings, parental education, and family history of atopy, as previously reported [10, 11]. Of these variables, only farming and family history of atopy were significantly associated with a physician's diagnosis of asthma, atopic sensitization, and diagnosis or current symptoms of hay fever. The most relevant farm-related exposures for asthma and atopy were regular contact to farm animals, frequent animal shed and hay loft visits, and consumption of unprocessed farm milk [11].

Table 1. Children living on a family-run farm are enriched in the SSCP analysis dataset
CharacteristicsGerman PARSIFAL population (N = 6843a)
Included in SSCP analysis (n = 489)Not included in SSCP analysis (n = 6354)P-value
  1. SSCP, single-strand configuration polymorphism.

  2. a

    Specific IgE was only measured in 1119 children.

Living on a family-run farm52%11%<0.001
Age in years (mean ± standard error of mean)8.7 ± 0.18.9 ± 0.00.004
Female gender46%49%0.276
At least one older sibling58%58%0.772
Low parental education30%29%0.725
Maternal smoking during pregnancy4%9%<0.001
Physician's diagnosis of asthma8%9%0.534
Atopic sensitization (specific IgE ≥3.5 kU/l)17%18%0.719
Atopic sensitization (specific IgE ≥0.35 kU/l)37%32%0.108
Diagnosis or current symptoms of hay fever10%11%0.186
Family history of asthma10%11%0.337
Family history of atopy31%35%0.066

A family history of atopy confounded the associations of farming with all health outcomes. In addition, high parental education confounded the association of farming with diagnosis or current symptoms of hay fever. Therefore, the subsequent models were adjusted for farming, parental education, and a family history of atopy.

The presence or absence of microorganisms and also their quantity might be related to the health outcomes. In the quantitative analysis, farm and nonfarm children differed in 51 of the 76 bands (Fig. 2). In contrast, only few bands were different with respect to the health outcomes (Fig. 2). Table 2 gives the numbers of bands significantly associated with asthma, hay fever, and atopy in the quantitative and the qualitative analyses. The majority of associations were inverse corresponding to protective effects, whereas the few positive associations represent potential risk bands.

image

Figure 2. Quantitative analysis of single-strand configuration polymorphism gel densities and bands with childhood asthma, hay fever, and atopy. The density of silver-stained gels is plotted against the position in the running direction of the gel. The uppermost panel shows mean values of log-transformed density units for farm (green) vs reference children (brown). The lower three panels give the mean values for children with (blue) or without (black) the indicated health condition. Bands with P-values <0.05 in adjusted robust logistic regression models are marked in red.

Download figure to PowerPoint

Table 2. Number of bands positively and negatively associated with farming, childhood asthma, hay fever, and atopy
CharacteristicsQuantitative analysisQualitative analysis
Positive associationInverse associationCut-off = 5 unitsCut-off = 10 units
Positive associationInverse associationPositive associationInverse association
  1. Of all 76 bands, only bands with significant associations (P < 0.05) are listed. The associations with the health conditions were adjusted for ‘living on a farm’.

Living on a farm4383282511
Physician's diagnosis of asthma0110321
Hay fever030011
Atopic sensitization (≥3.5 kU/l)031217

Microorganisms live in communities; therefore, the association between one bacterial species and disease may not be specific but merely indicative of another exposure. We therefore sought to identify specific associations between microorganisms and the health outcomes by mutually adjusting the particular bands in multivariable models. In these models, four bands were inversely associated with asthma, two with sensitization, and one with hay fever (Table 3). When lowering the cut-off of atopic sensitization to ≥0.35 kU/l, the association of band 539 with atopic sensitization remained for the quantitative and the qualitative approach (data not shown).

Table 3. Bacterial species significantly related to childhood asthma, hay fever, and atopy in mutually adjusted models
Health outcomeBandQuantitative analysisQualitative analysis
Cut-off = 5 unitsCut-off = 10 units
PositionBacteriumOROROR
  1. The models are adjusted for farming, family history of atopy, parental education, and mutually for all associated bands. OR = odds ratio with 95% confidence intervals.

Physician's diagnosis of asthma248Corynebacterium mycetoides; Zoogloea sp.; Duganella sp.; Aurantimonas ureolytica, sp.; uncultured bacterium

0.46 [0.24–0.89]

P = 0.021

  
300Serratia marcescens, S. nematodiphila, Serratia sp.; Pseudomonas fluorescens; uncultured bacterium  

2.42 [1.18–4.95]

P = 0.016

318 Corynebacterium tuberculostearicum   

2.32 [1.07–5.06]

P = 0.034

394 Gardnerella vaginalis

0.56 [0.32–0.97]

P = 0.037

0.17 [0.04–0.72]

P = 0.017

 
427Lactobacillus curvatus, Lactobacillus sakei; Streptococcus sp.; Moraxella sp.; uncultured bacterium  

0.45 [0.23–0.89]

P = 0.022

506Staphylococcus sciuri, sp.; Jeotgalicoccus sp.; Salinicoccus sp.; Macrococcus brunensis; Bacillus sp.

0.58 [0.38–0.88]

P = 0.010

0.41 [0.20–0.83]

P = 0.013

 
Hay fever261 Aerobacter ureolytica; P. fluorescens   

3.40 [1.11–10.41]

P = 0.032

352Corynebacterium mucifaciens, C. freiburgense, C. variabile, C. sp. Triatoma infestans; Neisseria meningitidis, N. mucosa, N. subflava

0.61 [0.43–0.87]

P = 0.007

 

0.28 [0.09–0.85]

P = 0.024

Atopic sensitization (≥3.5 kU/l)160Enterobacter cloacae, E. aerogenes, E. cancerogenus, E. ludwigii, E. sp.; Pantoea sp.; Kluyvera cryocrescens; Erwinia persicina; uncultured bacterium 

2.24 [1.02–4.88]

P = 0.043

 
265Aurantimonas altamirensis; A. ureolytica, A. sp.; Mesorhizobium sp.; uncultured bacterium  

8.36 [2.00–34.98]

P = 0.004

333 Lactobacillus iners

0.73 [0.53–1.00]

P = 0.050

 

0.37 [0.16–0.88]

P = 0.025

539Acinetobacter lwoffii, A. sp.; uncultured bacterium

0.65 [0.47–0.89]

P = 0.009

0.53 [0.29–0.98]

P = 0.042

 

All bands listed in Table 3 were excised from the SSCP gels, and the amplified fragments of the 16S rRNA gene contained in the bands were sequenced. The sequences were compared to a database of reference sequences of the bacterial 16S rRNA gene. The identified bacteria with an agreement of at least 98% between detected and reference sequences are listed in Table 3.

Discussion

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Conflict of interest
  8. Authors contributions
  9. References

Significant inverse associations of asthma, hay fever, and atopy with distinct bands reflecting certain types of bacterial exposures were detected. These associations were independent of living on a farm. Various bacteria were identified in the bands inversely associated with childhood asthma, hay fever, and atopy. Both qualitative and quantitative models revealed significant associations indicating a combination of dose-related and threshold-related effects.

In a previous analysis of the same dataset, we focused on bacterial diversity and combinations of SSCP bands [5]. The major advantage of that strategy was a low false-positive rate; however, major disadvantages were an increased false-negative rate and the restriction to bacterial clusters. In other words, we could not identify single SSCP bands representing bacterial species or genera. In addition, we might have missed important candidate bacteria.

In the present analysis, we extend the outcome spectrum to hay fever as a manifest allergic disease and circumvent the disadvantages of the previous analysis accepting a possible increase of the false-positive rate. The latter issue can be overcome by experimental confirmation of candidate bacteria, which is exemplified in a separate paper (Hagner et al., submitted to Allergy). In addition, our study sample is rather large for a DNA-based method, and we are not aware of any other population-based study with such comprehensive information on bacterial exposure.

A minor limitation consisted in the missing informative observations owing to previous usage of samples, which might have led to an overestimation of the variances and consequently to a loss of power. The false-positive rate, however, was not affected.

Sequencing of the bands mainly revealed the presence of environmental bacteria. For Acinetobacter lwoffii (band 539), the biological plausibility of the association with atopic sensitization is obvious as its atopy-protective potential has already been demonstrated in vitro and in vivo [12]. Similar effects were found in mice exposed to Lactococcus lactis strains [12] or Lactobacillus rhamnosus GG [13]. These bacteria are close relatives of Lactobacillus iners (band 333), Lactobacillus curvatus and Lactobacillus sakei (band 427).

The other bacteria, which emerged in our analysis, had so far not been implicated in immunomodulation or asthma protection. With the exception of Bacillus sp., all bacteria detected in band 506 belong to the family of Staphylococcaceae, which has been reported as the predominant bacteria family in house dust samples from the Karelian region with low prevalence of atopic diseases [14]. Staphylococcus sciuri colonizes the skin of a variety of farm animals [15] and is often isolated from food products of animal origin such as raw meat [16], raw milk [17], and cheese [18]. S. sciuri has also been reported to form biofilms in food-processing sites [19], thereby inhibiting adhesion and growth of competing bacteria [20]. Also, Macrococcus spp. usually colonize the skin and other body surfaces of cattle and horses and are found in food-processing factories [21]. Jeotgalicoccus spp. and Salinicoccus spp. are halophilic bacteria and have been detected in raw goat milk [22].

Another interesting group of bacteria belong to the genus Corynebacterium (bands 248 and 352) and the closely related Bifidobacterium Gardnerella vaginalis (band 394). Corynebacteria very commonly colonize straw and are found in air samples of flax-processing farms [23]. Like Mycobacterium tuberculosis, they contain mycolic acid, a cell wall component with the ability to prevent allergic airway inflammation in a mouse model [24].

The three Neisseria species found in band 352 are known causes of meningitis and endocarditis, but frequently colonize the oropharyngeal mucosa without leading to manifest infections. The cell wall of Neisseria meningitidis contains porins, which engage Toll-like receptor (TLR) 2 and induce maturation of dendritic cells [25].

Furthermore, sequences of ‘uncultured bacteria’ were detected. These represent most likely environmental bacteria so far only characterized by 16S rRNA analyses. Also, the other bacteria listed in Table 3 are environmental bacteria, whose immunologic properties are essentially unknown. Moreover, they are not accessible to PCR analysis, as their genomic sequences are currently unknown with the exception of fragments of the 16S rRNA gene, which are not suitable for specific PCR.

Intriguingly, many of the detected bacteria species naturally occur in raw milk [19, 22]. Moreover, band 506 containing predominantly species of the family Staphylococcaceae was significantly associated with current consumption of unprocessed farm milk independently of farming (adjusted OR = 2.18 [1.35–3.54]) and explained 33% of the inverse association of farm milk consumption and asthma. Beyond the PARSIFAL population [26], a strong and consistent protective effect of consumption of unprocessed farm milk on asthma or atopy has repeatedly been found [27].

The underlying immunologic mechanisms are largely unknown [28]. Two different routes of exposure are conceivable: ingestion and inhalation, whereas the latter one might include swallowing contaminated dust particles. On both routes, the incorporated microorganisms initially encounter mucosal barriers and cells of the innate immune system, which might be activated by ligation of pattern recognition receptors such as the TLRs. In turn, the innate immune system may shape the adaptive immune system to induce tolerance [29].

Staphylococcaceae might colonize not only the human skin but also other body surfaces such as the gut [30] or the respiratory mucosa. There, they might counterbalance the unfavorable effects of pathogenic bacteria by preventing their adhesion and growth. A recent study in neonates has shown that colonization with Streptococcus pneumoniae, Haemophilus influencae, or Moraxella catharralis at birth was associated with a higher risk of subsequent wheezing and asthma [31]. It is conceivable that colonization with certain staphylococci and related bacteria might limit colonization with those pathogens and thereby prevent asthma. In contrast to Staphylococcus aureus, S. sciuri does not produce staphylococcal enterotoxin B, which has been associated with decreased T-regulatory cytokines, increased Th2 cytokines, and induction of immunoglobulin E production in humans [32].

The parents of the asthmatic children included in the present analysis reported a physician's diagnosis of asthma ever. In a previous German study, cold air challenge had a negative predictive value of 93.7% for a reported diagnosis of asthma ever [33]. This indicates that a reported diagnosis of asthma hardly misses children with bronchial hyperresponsiveness. Only very mild cases of asthma without a physician's diagnosis but current wheezing might have been missed in the present analysis. Current wheezing, however, is much less reliable than a physician's diagnosis of asthma as the German language does not use a colloquial word for ‘wheeze’.

As expected, farm and nonfarm children differed in their bacterial exposure. The majority of bands were more common in farm children, thereby reflecting the broader spectrum of bacterial exposure in this group. The mattress is a long-term reservoir for individual microbial exposure as bacteria from animal sheds have been shown to be transferred to the child's mattress [34]. Therefore, mattress dust is a valid source for determining children's exposure to environmental bacteria.

Obviously, the simultaneous assessment of environmental exposure and disease status complicates inference of causality. However, the assumption that the detected bacteria protect from asthma and atopy seems more plausible than the hypothetical probability that asthma and atopy facilitate colonization with these bacteria, as most are environmental bacteria. Future confirmatory studies, however, should assess bacterial exposure before the onset of disease.

The identification of distinct bacterial genera and species does not contradict our earlier findings with respect to microbial diversity [5]. Rather, it is in line with the argumentation that high microbial diversity is more likely to cover the relevant microbial exposures for protection from asthma [5]. Whether specific bacteria or specific products of several related bacteria carry the effect remains open as well as the question whether microbial exposures provide true primary prevention or whether they exert therapeutic effects on latent disease.

In summary, the present analysis suggests that in a rich and diverse microbial environment, certain bacterial exposures play an important role for the protective effect on asthma and atopy. This complements earlier findings without contradiction [5]. Among the identified bacteria, Gram-positive (staphylococci, corynebacteria, lactic acid fermenters) and Gram-negative bacteria (neisseriae, Acinetobacter) were found. Transfer of these epidemiologic findings to animal models [12, 13, 34] may help understanding the immunologic mechanisms by which these bacteria may influence the development of asthma and atopy, and finally translate into guidelines of asthma prevention and treatment [35].

Acknowledgments

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Conflict of interest
  8. Authors contributions
  9. References

This work was supported by the Deutsche Forschungsgemeinschaft (SFB Transregio 22 Pulmonary Allergies; Project A1) and a European Union research grant (QLRT 1999-01391). We gratefully thank Barbara Fritz, Cornelia Öhme, and Andrea Klaus for excellent technical assistance. Furthermore we thank the participating children, their families, and the field workers of the PARSIFAL study.

Conflict of interest

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Conflict of interest
  8. Authors contributions
  9. References

Institutional funding by Deutsche Forschungsgemeinschaft: MJE, MM, KS, JM, JB, EvM; pending patent by employer: MJE, MM, KS, JM, JB, EvM; institutional funding by European Commission: AS, GP, EvM; EvM has received consulting fees from Protectimmun, Novartis, and GlaxoSmithKline and reimbursement of travel expenses by InfectoPharm; MvH has received lecture fees from Phadia.

Authors contributions

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Conflict of interest
  8. Authors contributions
  9. References

Conception and design of the study: MJE, MM, GP, JB, EvM; acquisition, analysis and interpretation of data: MJE, MM, KS, JM, MvH, AS, JB, EvM; drafting the article: MJE; revising manuscript: MJE, MM, KS, GP, MvH, AS, JB, EvM.

References

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Conflict of interest
  8. Authors contributions
  9. References