Correlation of Moraxella catarrhalis macrolide susceptibility with the ability to adhere and invade human respiratory epithelial cells

ABSTRACT Recently, the prevalence of macrolide-resistant Moraxella catarrhalis has been reported, especially among Chinese children. The fitness cost of resistance is reported to render the resistant bacteria less virulent. To investigate the correlation between macrolide susceptibility of M. catarrhalis and pathogenicity, the whole genome of 70 M. catarrhalis isolates belonging to four clonal complexes with different macrolide susceptibilities was sequenced. The gene products were annotated with the Gene Ontology terms. Based on 46 extracted essential virulence genes, 19 representative isolates were selected to infect type II alveolar cells (A549 cells). The ability of these isolates to adhere and invade human epithelial cells and to produce cytokines was comparatively analysed. Furthermore, mice were infected with a pair of M. catarrhalis isolates with different pathogenic behaviours and macrolide susceptibilities to examine pulmonary clearance, histological findings, and the production of cytokines. The percentages of annotations for binding, metabolic process, cellular process, and cell were non-significantly different between the macrolide-resistant and macrolide-susceptible groups. The presence of uspA2, uspA2H, pilO, lbpB, lex1, modM, mboIA, and mboIB significantly differed among the four clonal complexes and macrolide susceptibility groups. Furthermore, compared with those in macrolide-susceptible isolates, the adhesion ability was stronger (P = 0.0019) and the invasion ability was weaker (P < 0.0001) in the macrolide-resistant isolates. Mouse experiments revealed that pulmonary macrophages elicit immune responses against M. catarrhalis infection by significantly upregulating the Csf2, Il4, Il13, Il1b, Il6, Tnf, and Il18. Therefore, M. catarrhalis populations exhibited diverse pathogenicity in vitro and in vivo.


Introduction
Moraxella catarrhalis is a major etiological agent for acute otitis media and acute exacerbation of chronic obstructive pulmonary disease. Additionally, M. catarrhalis can cause other serious infections, such as meningoencephalitis [1] and communityacquired pneumonia [2]. Macrolides are commonly used to treat respiratory illnesses, including those caused by M. catarrhalis [3]. Recent studies have reported the prevalence of macrolide-resistant M. catarrhalis [4][5][6]. The fitness cost of antibiotic resistance is a phenomenon in which antibiotic-resistant mutants exhibit decreased fitness and virulence without an apparent reason [7,8]. Previously, we had demonstrated that macrolide-resistant isolates were highly concentrated in the CC449 (CCN10) and CC363 clonal complexes [9,10], while macrolide-susceptible isolates were highly concentrated in the CC224 and CC446 (CCN08) clonal complexes [9]. We hypothesized that populations of M. catarrhalis isolates with different macrolide susceptibilities exhibit differential pathogenic behaviours, such as the ability to adhere and invade human epithelial cells.
Some virulence factors, including UspA1, UspA2, UspA2H, OMPCD, OMP M. catarrhalis adherence protein (McaP), and Moraxella lipooligosaccharide (LOS), have been previously reported [11][12][13][14]. These virulence factors promote serum resistance to evade humoral immunity. Additionally, these virulence factors function as M. catarrhalis adhesins and bind to different human epithelial cell lines [15]. The interaction between M. catarrhalis and human-derived epithelial cell lines is multifactorial and involves the simultaneous activities of multiple adhesins [15]. We predict that various combinations of multiple adhesins co-ordinate and act differently in M. catarrhalis isolates with different backgrounds (such as the macrolide-resistant and macrolide-susceptible groups), conferring differential abilities to adhere and invade human epithelial cells.
To confirm this hypothesis, 70 M. catarrhalis isolates with different macrolide susceptibilities, including those belonging to CC449, CC363, CC446, and CC224, were sequenced. The gene products were annotated with Gene Ontology (GO) terms. Based on the 46 extracted essential virulence genes, 19 isolates with 17 unique combinations of virulence genes were selected to infect type II alveolar cells (A549 cells). The abilities of the isolates to adhere and invade human epithelial cells and produce cytokines were comparatively evaluated. Based on the results of this experiment, mice were infected with a pair of M. catarrhalis isolates with different pathogenic behaviours and macrolide susceptibilities to analyse pulmonary clearance, histological changes, and the production of cytokines. Macrolide-resistant and macrolide-susceptible M. catarrhalis isolates exhibited differential abilities to adhere and invade human epithelial cells with several important virulence genes facilitating this process. Additionally, mouse experiments suggested that pulmonary macrophages mediate the rapid immune response to M. catarrhalis infection. This immune response involved the upregulation of cytokines, such as Csf2, Il4, Il13, Il1b, Il6, Tnf, and Il18.

Bacterial strains
Seventy non-duplicate clinical M. catarrhalis isolates were collected from the sputum or broncho-alveolar lavage of adult patients (>18 years of age) diagnosed with lower respiratory tract infection and the sputum or ear purulent secretion of children diagnosed with lower or upper respiratory tract infections between 2010 and 2017. Of these, 22, 40, and 8 isolates were obtained from the Peking Union Medical College Hospital (PUMCH), the First Affiliated Hospital of Xiamen University, and other hospitals in China, respectively. The isolates were identified using the Bruker Biotyper matrix-assisted laser desorption ionisation-time of flight mass spectrometry system (Biotyper version 3.1 software, Bruker Daltonics, Billerica, Germany). The susceptibilities of 70 isolates to erythromycin and azithromycin were examined using the Kirby-Bauer method and E-test. Previously, we had performed 23S rRNA gene and multilocus sequence typing analyses [9]. In this study, 70 isolates belonged to the following four clonal complexes: CC449 (n = 14; 14 macrolide-resistant isolates with A2330T mutation), CC363 (n = 18; 13 macrolideresistant isolates with A2330T mutation and 5 macrolide-susceptible isolates without A2330T mutation), CC446 (n = 13; 13 macrolide-susceptible isolates without A2330T mutation), and CC224 (n = 25; 3 macrolide-resistant isolates with A2330T mutation, 22 macrolide-susceptible isolates without A2330T mutation). The isolates were stored at −80°C.
Genomic DNA extraction, sequencing, assembly, and annotation Genomic DNA was extracted using the TIANamp bacteria genomic DNA kit (TiangenBiotechCo. Ltd., Beijing, China) and subjected to shotgun sequencing using Illumina Genome Analyzer IIx technology. Paired-end libraries were prepared from 5 μg of isolated genomic DNA using the TruSeq DNA sample prep kit A (Illumina Inc., San Diego), following the manufacturer's instructions. Genomic paired-end libraries were sequenced with a read length of 2 × 150 nucleotides using an Illumina GAIIx instrument, following the manufacturer's instructions. Image analysis and base calling were performed using the standard Illumina pipeline. The raw Illumina sequencing reads were trimmed at a threshold of 0.01 (Phred score of 20) using Fastx-toolkit (http://hannonlab.cshl. edu/fastx_toolkit/commandline.html). SPAdes [16] was used for assembling the filtered reads with parameters "-k 33,45,55 -careful -cov-cutoff 5". All assembled contig sequences were annotated with Prokka (Tools used in Prokka: Prodigal for proteincoding features (CDS), RNAmmer for rRNA, Aragorn for tRNA) [17].

Virulence gene analysis
The sequences of 46 virulence genes, which were retrieved from literature, were downloaded from the Virulence Factor Database (VFDB, http://www.mgc. ac.cn/cgi-bin/VFs/v5/main.cgi, VFDB_setB_nt.fas.gz) and National Centre for Biotechnology Institute (NCBI). The detailed gene information is shown in Table S1. The virulence genes in the sequenced strains were predicted using SRST2 [18] and confirmed using Basic Local Alignment Search Tool (BLAST) with the following parameters: identity > 50% and coverage > 50%.

Gene homology analysis and functional enrichment
Gene homology analysis was performed based on the pan-genome results using Roary with the parameters "-e create a multiFASTA alignment of core genes using PRANK; -n fast core gene alignment with MAFFT" [19]. The gene distribution file was obtained from the "gene_presence_absence.csv" and "gene_-presence_absence.Rtab" files. Specific genes of a certain group were obtained using the χ 2 -test (P-value < 0.05) for categorical data transformed from the gene distribution obtained from Roary. The gene functions were further annotated using Blast2GO to obtain the enriched GO terms (https://www.blast2go.com/). GO annotation results were visualized and compared using WEGO [20,21]. Significant (P < 0.05) GO terms were determined using the Pearson χ 2 test.

Adhesion assay
To examine the effect of virulence genes on the ability of M. catarrhalis isolates to adhere and invade human respiratory epithelial cells, 19 isolates with 17 unique combinations of virulence genes were selected from the 70 isolates used in this study.
To determine the optimal ratio of bacterium to host cells for application in the adhesion and invasion assays, the following five multiplicity of infection (MOI) ratios were examined: 2:1, 4:1, 10:1, 20:1, and 200:1. The test MOI ratios did not significantly affect bacterial adhesion to A549 cells but markedly affected the invasion of M. catarrhalis isolates into A549 cells (data not shown). The optimal MOI with detectable bacterial numbers and low variability was 10. Therefore, an MOI of 10 was used in the subsequent adhesion and invasion assays.
The adhesion assay was performed as previously described [22] with some modifications. Briefly, A549 cells were plated in 24-well cell culture plates (1 × 10 5 cells/well) and incubated with an appropriate MOI (10) of M. catarrhalis for 4 h. Non-adherent bacteria were removed by washing with phosphatebuffered saline (PBS, PH 7.2-7.4). Cells were lysed with 200 µL of 0.25% trypsin (Sigma) at 37°C for 10 min and 300 µL of 0.1% saponin (Sigma) for 10 min. The lysates were transferred to 1.5 mL Eppendorf tubes and mixed vigorously for 30 s. Aliquots of the samples were serially diluted (1 × 10 −3 , 1 × 10 −4 , and 1 × 10 −5 ) and plated on blood agar plates (100 µL/plate) containing 5% sheep blood to determine the number of adherent bacteria. All assays were performed in six replicates. At least three independent experiments were performed.

Invasion assay
The susceptibility of 70 M. catarrhalis isolates to gentamycin was examined. All isolates were susceptible to gentamycin. The results of the in vitro time-kill assay revealed that gentamycin exhibited bactericidal activity (99.9% killing) at a concentration of 300 µg/ mL (treatment duration: 1 h). Therefore, 300 µg/mL gentamycin was used in the subsequent experiments in this study.
From the 70 isolates used in this study, 19 isolates with 17 unique combinations of virulence genes were selected to examine their abilities to invade A549 cells. The experimental procedure of the invasion assay was similar to that of the adhesion assay [22], except 300 µg/mL gentamycin was added to the infected monolayer cells before lysing with 0.25% trypsin and 0.1% saponin to kill any remaining adherent extracellular bacteria. The cells were infected with M. catarrhalis at density of 2 × 10 5 (MOI = 2), 4 × 10 5 (MOI = 4), 1 × 10 6 (MOI = 10), 2 × 10 6 (MOI = 20), or 2×10 7 CFU/well (MOI = 200) for 4 h. All assays were performed in five or six parallel wells. At least three independent experiments were performed. Results of representative experiments are presented as mean (colony-forming unit (CFU) per well) ± standard deviation.

Electron microscopy
A549 cells were plated in 24-well cell culture plates (1 × 10 5 cells/well) and incubated with M. catarrhalis [73-OR (CC446) or 35-OR (CC363)] at MOI 10 for 4 h. Non-adherent bacteria were removed by washing with PBS. Cells were lysed with 200 µL of 0.25% trypsin (Sigma) at 37°C for 10 min. The lysate was centrifuged and washed twice with PBS. The infected A549 cells were fixed in 2.5% glutaraldehyde in phosphate buffer (0.06 M) for up to 2 days (pH 7.3). Cells were prepared for electron microscopy as previously described [22]. The distribution of M. catarrhalis isolates outside and inside A549 cells after infection for 4 h was determined at a magnification of ×3000. Meanwhile, the lamellipodia enclosing M. catarrhalis isolates after 4 h of infection and different phases of macropinocytotic ingestion were observed at a magnification of ×15000.

Pulmonary clearance model
The mouse pulmonary clearance model was established as described previously [23][24][25][26]. The animal study was approved by the Institute of Microbiology, Chinese Academy of Sciences Animal Care and Use Committee (project number: APIMCAS2021144). C57/BL6J male mice (aged 8-10 weeks) were obtained from Jackson Laboratories (Bar Harbour, ME) and housed under standard pathogen-free conditions in an animal facility at the Institute of Microbiology, Chinese Academy of Sciences. Mice (n = 5-7 per group) were intraperitoneally anesthetised using isoflurane. After the trachea was surgically exposed, the mice in the control (PBS-treated), 73-OR (macrolide-susceptible isolate)-infected, and 35-OR (macrolide-resistant isolate)-infected groups were intratracheally injected with PBS, 10 8 CFU of 73-OR M. catarrhalis, and 10 8 of 35-OR M. catarrhalis, respectively. The bacterial suspension was prepared in 50 µL PBS. The surgical wound was closed using 5.0 silk sutures, and the animals were allowed to recover. At 3 h post-challenge, the animals were sacrificed via CO 2 exposure. The blood and lung tissues were collected. The number of white blood cells (WBCs) in the blood was measured using an automated haematological analyser Sysmex Xs-800i. The lungs were homogenized on ice in 1 mL of PBS (PH 7.2-7.4) using a tissue homogenizer. Next, the samples were serially diluted and plated on blood agar plates (100 µL/plate) containing 5% sheep blood (two replicates). After incubation overnight at 37°C and 5% CO 2 , the CFU was determined via manual counting.

Histological analysis
Mice (n = 5-7/group) were challenged with PBS or 10 8 CFU of M. catarrhalis (73-OR or 35-OR) in 50 µL PBS as described above. To examine pulmonary inflammation, both lungs were inflated with 10% buffered formaldehyde and perfuse-fixed as previously described [23]. The fixed lungs were embedded in paraffin, stained with haematoxylin and eosin, and examined under a microscope. The lung injuries, including alveolar structure damage and inflammatory cell infiltration, were scored in a blinded manner. Histological changes in the lung tissue sections of different groups were compared at magnifications of ×100 and ×400.

Statistical analysis
GO annotation results were visualized and compared using WEGO [20,21]. Significant (P < 0.05) GO terms were determined using Pearson's χ 2 -test (or Fisher's exact test, when appropriate). The prevalence of virulence genes among the four clonal complexes and the two macrolide susceptibility groups was compared using Pearson's χ 2 -test (or Fisher's exact test, when appropriate) for categorical data ( Table 1). The levels of cytokines and bacterial growth between the macrolide-susceptible and macrolide-resistant groups were compared using the two-tailed Mann-Whitney test, while those between the four clonal complexes were compared using one-way analysis of variance (ANOVA). Pulmonary clearance and WBC counts in mice belonging to the control, 73-OR-infected, and 35-OR-infected groups were compared using oneway ANOVA. Differences were considered significant at P < 0.05.

Ethical statement
Only de-identified clinically obtained bacterial isolates were used in this study. No human subjects participated in this study. The study was approved by the Human Research Ethics Committee of PUMCH (No. JS-2304).

Comparative analysis of functional annotations
In this study, 220 and 106 genes were extracted from the macrolide-resistant (n = 30) and macrolide-susceptible groups (n = 40), respectively, as they exhibited significant and differential prevalence between the two groups. Next, 91 of the 220 genes and 57 of the 106 genes were annotated to GO terms ( Figure 1). The GO term with the highest number of assigned unigenes in the molecular function (MF) category was catalytic activity, followed by binding, transporter activity, transcription regulator activity, structural molecule activity, and molecular function regulator. In the biological process (BP) category, the term with the most uni-genes was metabolic process, followed by cellular process, localization, response to stimulus, biological regulation, regulation of biological process, developmental process, cellular component organization or bio-genesis, biological adhesion, and negative regulation of biological process. Meanwhile, in the cellular component (CC) category, the term with the most uni-genes was cell, followed by membrane, membrane part, protein-containing complex, and organelle.

Comparative analysis of adhesion and invasion
The prevalence of the virulence genes varied depending on the clonal complex and macrolide susceptibility. Hence, 19 isolates with 17 unique combinations of virulence genes were selected from the four clonal complexes (CC224, n = 6; CC449, n = 3; CC446, n = 5; and CC363, n = 5). Compared with those in the macrolide-susceptible isolates, the adhesion ability was significantly stronger (P = 0.0019) and the invasion ability was significantly weaker (P < 0.0001) in the macrolide-resistant isolates ( Figure 3B and D). Additionally, the adhesion and invasion abilities were significantly different between the main clonal complexes. CC363, a major macrolide-resistant clonal complex, exhibited the highest host cell adhesion ability, followed by CC446 (a purely macrolide-susceptible clonal complex), CC449 (a purely macrolide-resistant clonal complex), and CC224 (a mostly macrolide-susceptible clonal complex) ( Figure 3C). In contrast, CC446, a pure macrolide-susceptible clonal complex, exhibited the highest host cell invasion ability, followed by CC224 (a majorly macrolide-susceptible clonal complex), CC363 (a majorly macrolide-resistant clonal complex), and CC449 (a purely macrolide-resistant clonal complex) ( Figure 3E). To further demonstrate the differential behaviours of the M. catarrhalis populations, one macrolideresistant isolate (35-OR) with strong host cell adhesion ability and weak host cell invasion ability (mean number of adhering cells = 2.6 × 10 7 CFU/ well; mean number of invading cells = 5 CFU/well; [(number of invading cells/number of adhering cells) × 100] = 0.000019%) and one macrolide-susceptible isolate (73-OR) with a weak host cell adhesion ability and a strong host cell invasion ability (mean number of adhering cells = 1.7 × 10 7 CFU/well; mean number of invading cells = 8.6 × 10 2 CFU/well; [(number of invading cells/number of adhering cells) × 100] = 0.005170%) were selected to infect A549 cells (P = 0.0286). The cells were fixed and subjected to transmission electron microscopy (TEM) (Figure 4). The adherent M. catarrhalis 35-OR cells formed multicellular grape-like aggregates on the A549 cell surface ( Figure 4A) but did not invade the cells. In contrast, some M. catarrhalis 73-OR cells adhered to the surface of A549 cells, while a visible number of isolates infiltrated the cells ( Figure 4B). These results are consistent with those of bacterial growth analysis. TEM analysis confirmed that the epithelial cell monolayer formed lamellipodia and filopodia, extending toward and enclosing individual adherent bacterial cells ( Figure 4C). Different phases of macropinocytotic ingestion are shown in Figure 4C.

Effects of M. catarrhalis populations on animal models
Histological analysis ( Figure 5) revealed that alveolar and bronchial structures were almost intact and visible in the three groups. Compared with that in the alveolar sputum of control mice ( Figure 5A and D), the number of inflammatory cells, especially pulmonary macrophages and neutrophils, was higher in the alveolar sputum of 73-OR-infected ( Figure 5B and E) or 35-OR-infected mice ( Figure 5C and F). Immune cell infiltration and fibrous tissue hyperplasia were also observed. Moreover, inflammation severity in the 35-OR-infected group was higher than that in the 73-OR-infected group as evidenced by increased numbers of inflammatory cells and the manifestation of fibrous tissue hyperplasia.
Next, bacterial counts in the total lung tissue and WBC counts in the blood after M. catarrhalis infection were determined (Figure 6). The bacterial counts in the 73-OR-infected and 35-OR-infected groups at between four clonal complexes. The levels of cytokines and bacterial growth between the macrolide-susceptible and macrolide-resistant groups were analysed using the two-tailed Mann-Whitney test, while those between the four clonal complexes were analysed using one-way analysis of variance. Differences were considered significant at P < 0.05.  3 h post-infection were significantly (P < 0.0001) higher than those at 0 h post-infection ( Figure 6A). The clearance of the 73-OR isolate was significantly rapid when compared with that of the 35-OR isolate (P = 0.0134). These results were consistent with the histological findings, which revealed severe inflammation in the 35-OR-infected group. Meanwhile, compared with those in the control group, the blood WBC counts were lower in the 73-OR-infected group and 35-OR-infected groups, especially in the 35-OR-infected group (P = 0.0393) ( Figure 6B).

Cytokine profile analysis
The plasma concentrations of cytokines were examined ( Figure 6C). The Il6, Il10, Il27, and Tnf levels in the infected groups were significantly (P < 0.05) upregulated when compared with those in the control group. However, the levels of the other 13 cytokines were not significantly different between the two groups. Additionally, the cytokine levels in the 73-OR-infected group were non-significantly higher than those in the 35-OR-infected group. The , and Tnf levels in the 73-OR-infected and 35-OR-infected groups were significantly (P < 0.05) higher than those in the control group, especially in the 73-OR-infected group. Pulmonary clearance and WBC counts in the control, macrolide-susceptible, and macrolide-resistant groups were compared using one-way analysis of variance. Differences were considered significant at P < 0.05. pulmonary levels of cytokines were also examined ( Figure 7). Compared with those in the control group, the pulmonary Il13, Il1b, Il4, Il6, Il17a, Il18, Il22, Tnf, and Csf2 levels were significantly upregulated (P < 0.05) in the infected group. However, the pulmonary levels of eight cytokines were not significantly different between the two groups. The pulmonary cytokine levels in the 73-OR-infected group were higher than those in the 35-OR-infected group, which was consistent with the results of mouse plasma analysis ( Figure 6C).

Discussion
Previously, we had demonstrated that macrolideresistant isolates mainly clustered in CC449 and CC363 although M. catarrhalis has a diverse genetic background [9]. Mycobacterium tuberculosis is reported to be associated with the fitness cost of antimicrobial resistance [8]. Hence, we hypothesize that the pathogenicity of macrolide-resistant isolates is higher than that of macrolide-susceptible isolates owing to efficient energy conservation. The role of macrolide susceptibility in the pathogenicity of M. catarrhalis to the host is unknown. This study compared the pathogenicity of 30 macrolide-resistant isolates and 40 macrolide-susceptible isolates. Based on previous studies, 46 essential virulence genes were extracted [15,27,28]. The distribution of these essential virulence genes in 70 M. catarrhalis isolates was analysed to examine its correlation with macrolide susceptibility. Additionally, the ability of bacterial clone isolates to adhere to A549 cells in vitro was comparatively analysed to establish its correlation with virulence genes. Based on in vitro studies, mice were infected with a pair of M. catarrhalis isolates with different pathogenic behaviours and macrolide susceptibilities. Pulmonary clearance, histological manifestations, and cytokine production were analysed in infected mice.
The prevalence of uspA2, uspA2H, pilO, lbpB, lex1, modM, mboIA, and mboIB significantly differed between the four clonal complexes and between the macrolide susceptibility groups. However, the roles of these genes in macrolide susceptibility or pathogenicity are unknown. To elucidate the potential correlation, this study aimed to examine these genes and their possible functions.
PilO and PilN are integral inner membrane proteins that play an important role in type IV pilus formation, which is essential for pathogenicity, DNA transformation, and motility [34,35]. The distribution of pilO varied among the four clonal complexes (Table  1 and Figure 2). In particular, pilO was absent in CC446. However, pilO prevalence was not correlated with bacterial adhesion or invasion.
Lactoferrin binding protein B (LbpB), a bi-lobed outer membrane-bound lipoprotein, plays an essential role in the iron acquisition process in some gramnegative pathogens [36,37]. In this study, the prevalence of lbpB varied between the four clonal complexes (Table 1 and Figure 2) and between the macrolide susceptibility groups. The distribution of lbpB, which was specifically absent in CC224, significantly varied between the macrolide-resistant and macrolide-susceptible groups (P = 0.0006).
In Haemophilus spp., Lex1 is involved in LOS biosynthesis, serum resistance, adherence, and invasion [38,39]. The function of the lex1 in M. catarrhalis, has not been elucidated. In this study, the distribution of lex1 varied between the four clonal complexes (Table 1 and Figure  2). In particular, lex1 was absent in CC224, which mostly comprises macrolide-susceptible isolates (P < 0.05). However, lex1 prevalence was not correlated with the ability of bacteria to adhere or invade A549 cells.
The modulation of the expression of ModM, a phase-variable type III DNA methyltransferase, affects the expression of various genes, known as a phase-variable regulon [40]. The three alleles of modM (modM1, modM2, and modM3) potentially regulate the expression of multiple genes associated with colonization, infection, and protection against host defences [41]. In this study, modM was detected in all isolates of CC446 and CC363 but not in those of CC224 and CC449. Additionally, modM2 and modM3 accounted for 96.8% (30/31) and 3.2% (1/ 31), respectively, of the detected modM. No modM1 was detected in the isolates, which was consistent with the results of a previous study [41].
The genes mboIA and mboIB are always simultaneously present or completely absent. Previously, we had reported that mboIA and mboIB were associated with macrolide resistance [42], which was further confirmed in this study. However, the functions of mboIA and mboIB in M. catarrhalis are unknown. In this study, mboIA and mboIB genes were specifically absent in CC449. However, the correlation of the absence of mboIA and mboIB with the weak invasion ability of CC449 is unclear.
In this study, M. catarrhalis populations exhibited diverse abilities to adhere and invade A549 cells. Macrolide-resistant isolates exhibited enhanced adhesion (P = 0.0019) and suppressed invasion (P < 0.0001) when compared with macrolide-susceptible isolates ( Figure 3B and D). Similarly, the adhesion and invasion significantly varied among the clonal complexes. CC363, a predominant macrolide-resistant clonal complex, exhibited the strongest host cell adhesion ability. Meanwhile, CC446, a purely macrolide-susceptible clonal complex, exhibited the strongest host cell invasion ability ( Figure 3C and E). Next, one macrolide-resistant isolate (35-OR) with strong host cell adhesion ability and a weak host cell invasive ability and a macrolide-susceptible isolate (73-OR) with weak host cell adhesion ability and a strong host cell invasive ability were selected to infect A549 cells. The differences in the host cell adhesion and invasion abilities of M. catarrhalis populations were examined using TEM (Figure 4). M. catarrhalis 35-OR aggregated on the A549 cell surface in a multicellular grape-like form ( Figure 4A) without cellular invasion. In contrast, some M. catarrhalis 73-OR isolates with a diplococcus form adhered to the surface of A549 cells and infiltrated the cells ( Figure 4B). These results were consistent with the bacterial growth analysis (adhesion and invasion assays) results.
Mouse experiments revealed differential changes in the lung tissues of the three groups. Compared with those in the control group, the number of inflammatory cells, especially infiltrated neutrophils and pulmonary macrophages, was higher and fibrous tissue hyperplasia was prominent in the infected groups. Meanwhile, inflammation severity in the 35-ORinfected group was higher than that in the 73-ORinfected group as evidenced by increased numbers of infiltrating inflammatory cells and fibrous tissue hyperplasia. The pulmonary clearance of the 73-OR isolate was rapid when compared with that of the 35-OR isolate ( Figure 6). This was consistent with the histological (which revealed severe inflammation in the 35-OR-infected group) and TEM analyses. The WBC counts in the blood of the 73-OR-infected and 35-OR-infected groups were downregulated. In particular, the WBC counts were significantly downregulated in the 35-OR-infected group (P = 0.0393) ( Figure 6B). We speculated that WBCs were rapidly recruited from the circulation to the injured lung tissues. Furthermore, the upregulated cytokines, such as Csf2, Il4, and Il13, which function in different phases of macrophage activation, in the lung tissues ( Figure  7) and blood ( Figure 6C), were associated with pulmonary macrophages [43]. The secretion of IL1B, IL6, TNF [44], and IL18 is triggered by activated macrophages, resulting in the recruitment of immune effector cells, such as neutrophils and monocytes and the death of pathogen-infected cells [45][46][47]. This explains the similar production levels of cytokines between the 35-ORinfected and 73-OR-infected groups in adhesion and invasion assays as no pulmonary macrophages were added. These findings indicate that pulmonary macrophages may play a critical role in the rapid immune response against M. catarrhalis infection. The immune response triggered by pulmonary macrophages may vary depending on M. catarrhalis populations. For example, the cytokine concentrations in the 73-ORinfected group were higher than those in the 35-ORinfected group. However, further studies are needed to examine the differential immune responses elicited by pulmonary macrophages against different M. catarrhalis populations.
M. catarrhalis populations exhibited diverse pathogenicity in vitro and in vivo. Adhesion and invasion are two consecutive steps in M. catarrhalis colonization and infection. Hence, the factors involved in the bacterial interaction with host cells, such as virulence genes, receptors, and cytokines must be determined.
One limitation of this study was that it focused only on four representative clonal complexes (CC449, CC363, CC224, and CC446) and determined their abilities to adhere and invade human epithelial cells and produce cytokines. Future studies must use isolates with diverse genetic backgrounds to verify the findings of this study. Additionally, the classical gene knockout and complementation experiments were not performed to examine the effect of virulence genes on bacterial ability to adhere and invade human respiratory epithelial cells. These experiments will be performed in the follow-up study.