Differential distribution of IgA-protease genotypes in mucosal and invasive isolates of Haemophilus influenzae in Sweden

The 177 isolates were collected from 177 unique patients. The descriptive features of patients from cohort 1 are presented in Table 1. The gender distribution was almost balanced, with a slight overrepresentation of women. The median age in the cohort was 32 years of age, 66 samples (37.7%) were from pre-school children (< 6 years of age). Approximately two thirds of isolates were collected in primary care facilities and one third in hospital care. Nineteen % of isolates were from cases of acute otitis media and 15% were from cases with lower respiratory tract infection. In 22% of patients, there was insufficient information in the referral text to identify the type of clinical infection.

The PCR screening results performed for all 177 isolates of cohort 1 is presented in Table 1. The igaA gene PCR was positive in all but one isolate (176/177; 99%). The igaA1 gene variant was found in 121/177 (68%) whereas the igaA2 gene variant was found in 55/177 (31%) of isolates. Isolates also carrying the igaB gene were common throughout the study period, as 81/177 (46%) of isolates tested positive in the PCR screening. Only one of the isolates was igaB2 positive. There was a year-to-year variation of the distribution, but no clear trend over time, with a lower proportion of igaB-positive isolates in 2011 compared with both 2010 and 2012, likely reflecting temporal variations in circulating strains.

The PCR screening results for all 53 isolates of cohort 2 is presented in Table 2. The igaA gene PCR was positive in all isolates. The capsule type-associated IgA-cleavage pattern from prior studies was mainly confirmed in this genotype analysis, as all included type b and d isolates had an igaA1 genotype, and type c and e isolates had an igaA2 genotype. Data on type f isolates have been conflicting as judged by prior studies, but in our study all four type f isolates included carried an igaA2 genotype. In our two capsule type a isolates tested, the PCR for igaB screening was positive. Among the tested isolates in cohort 2, igaB was found in 11 isolates (21%), including 3 igaB2. Among all invasive isolates and invasive NTHi isolates igaB was found in 8/42 (19%) and 7/33 (21%) isolates, respectively. Two of the three included urogenital isolates had an igaB genotype. Information regarding isolation site and capsule type for cohort 2 is outlined in Additional file 1: Table S1. Information on the clinical descriptive features was, however, only available in a subset of the cohort.

It is well established that a positive PCR screening does not necessarily correlate with protease activity. Therefore, we determined the IgA1 cleavage patterns by Western blot in a subset of 54 isolates. The isolates were chosen to represent all different genotypes, all years in cohort 1, in addition to all capsule types and isolation sites. Forty isolates were from cohort 1 (14 from 2010, 9 from 2011 and 17 from 2012), and 14 isolates from cohort 2 (representing all capsule types as well as isolation sites). Cleavage pattern A1 correlates with IgA-A1, cleavage pattern A2/B can be either IgA-A2 or IgA-B, and cleavage pattern A1 + A2/B indicates expression of two proteases simultaneously. Among 22 isolates tested with an igaA1 genotype without igaB, 20 had indeed an A1 cleavage pattern. Among the 14 isolates analysed with an igaA2 genotype without igaB, 13 had an A2/B cleavage pattern. In 11 out of 14 isolates tested with an igaA1 + igaB genotype, the igaB gene was expressed; 9 had an A2/B cleavage pattern, while two had an A1 + A2/B cleavage pattern. Two isolates with an igaB2 genotype were examined and only one of them had an A2/B cleavage pattern. Both analysed isolates with a combined igaA2 and igaB genotype had an A2/B cleavage pattern. All tested encapsulated isolates had cleavage patterns corresponding to the genotype except for capsule type a, which had a A1 cleavage pattern, suggesting that the igaB gene is not expressed in H. influenzae type a. Taken together, in the majority of isolates the cleavage pattern phenotype corresponded to the genotype.

Data of association and results from the regression analysis are presented in Tables 3 and 4, respectively. The igaB genotype was less common in invasive (bloodstream and CSF) isolates compared with isolates from other isolation sites. The difference was significant using Fischer’s test, 8/42 (19%) of invasive isolates compared with 84/188 (45%) of non-invasive isolates (p = 0.003), as well as for invasive non-typeable H. influenzae isolates alone, 7/33 (21%) compared with 83/182 (46%) of non-invasive isolates (p = 0.012). The negative association between igaB gene presence and bloodstream infection remained independently significant even when adjusting for covariates in a multinomial multivariate regression (Table 4). Younger age was associated with acute otitis media, whereas increasing age was associated with bloodstream infection in the same model. There was no significant association between iga genotype and type of respiratory tract infection (Table 3). A significant association between igaA2 genotype and young age was also observed (Table 3).

This study was performed in Malmö, Sweden, at a laboratory of clinical microbiology serving an area with a population of approximately 500,000 patients. It was the only microbiology laboratory in the region. Two cohorts were investigated. In the first cohort, all H. influenzae isolated from respiratory tract samples in January 2010, 2011 and 2012 (all tested as non-typeable) were selected for further testing. This strategy was chosen to allow for temporal variations in the distribution. The second cohort comprised of a collection of well-characterized invasive and encapsulated isolates from cases with different clinical infections. The cohort includes Swedish isolates from 1997 to 2010 and international isolates representing all capsule types. Details on isolation sites and origins of isolates in cohort 2 are given in Additional file 1: Table S1.

All isolates were grown on chocolate agar plates overnight (18 h) in a humid atmosphere at 37 °C and 5% CO₂ before any experiments were conducted. All isolates had been confirmed as H. influenzae using standard microbiology techniques or MALDI-TOF (Matrix Assisted Laser Desorption and Ionization - Time-of-flight mass spectrometry). DNA was prepared by adding a few colonies of bacteria to distilled water. After heating at 98 °C for 10 min, each sample was centrifuged (16,000 x g) followed by collection of the supernatant.

All isolates were screened using four different PCR primer pairs as described in prior studies [12], Y (identifying the igaA gene), W (identifying the igaA2 variant of the igaA gene) as well as T and U (identifying the igaB gene). Isolates positive in the igaB screening were subjected to a fifth PCR, identifying the igaB2 variant. The genotyping algorithm is visualized in Fig 1. Conditions and primer sequences are outlined in Additional file 2: Table S2. DNA was separated using a 1xTAE 1% agarose gel with 0.625 mg/mL ethidium bromide. Bands were visualized with UV light using a BioRad molecular Imager (Hercules, CA).

A subset of 54 isolates, representing all IgA protease gene variants, was chosen for confirmation of IgA1 cleavage by Western blot. The strains were grown in brain-heart-infusion (BHI) broth supplemented with 10 μg/ml NAD and hemin at 37 °C 5% CO₂ overnight. The bacteria were spun down and 20 μl of supernatant was incubated with 3 μl of 1 mg/ml human IgA1 (Calbiochem, Solna, Sweden) at 37 °C overnight. The proteins were separated by SDS-PAGE and transferred to nitrocellulose membranes. The membranes were incubated with polyclonal rabbit anti-human IgA antibodies at dilution 1:1000 overnight at 4 °C (A0262, Dako; Glostrup Denmark) followed by polyclonal HRP (horseradish peroxidase)-conjugated swine anti-rabbit IgG (dilution 1:1000) (P0217, Dako; Glostrup, Denmark) 2 h at room temperature and developed.

In the cohort of respiratory tract isolates, the information regarding type of clinical infection was gathered from the referral text. The disease types were lower respiratory tract infection, acute otitis media, sinusitis or unspecfic upper respiratory tract infection. We also collected data on patient age, gender and whether the culture had been taken in primary care or at a hospital facility. In the second cohort, information on capsule type and isolation site (blood, CSF or other) was available, and in a subset of cases also age and gender.

Data analysis was performed using STATA, version14 (College Station, TX). The presence or absence of igaB between invasive and non-invasive isolates was assessed using a two-sided Fischer’s exact test. Comparison of clinical traits among isolates with different genotypes was performed using the Kruskal-Wallis test for continuous variables and the chi-2 test for categorical variables. To assess the relevance of confounders in the associations between IgA1 protease genotypes and clinical diagnoses, multinomial multivariate logistic regression models using categorical clinical diagnosis as outcome were fitted using the purposeful selection algorithm [20]. Briefly, the main predictor (IgA protease genotype, igaA2 vs all and igaB vs all, respectively) and all covariates that in a univariate analysis had a p-value of < 0.2 were included in a crude model. The least significant covariate was then stepwise removed until only significant covariates or covariates that substantially affected the odds ratio of a remaining covariate were retained in the final model.