Pandoraea isolates were recovered from 18 sputum samples from a single CF patient attending the CF center of the University Hospital of Montpellier, South of France. All samples were analyzed according to national recommendations (Société Française de Microbiologie, 2015) and the initial isolate was identified by 16S rRNA gene sequencing. Isolates were named with numbers (1–18) according to chronological sampling order, and isolates with different colonial morphotypes in a sample were named with the same number followed by apostrophe(s). Selected co-colonizing P. aeruginosa (Pa13 isolated in September 2013) and methicillin-resistant Staphylococcus aureus (Sa14 isolated in January 2014) isolated at the time of or close to the most severe pulmonary exacerbation experienced by the patient during the study period, were also studied. Sputum culture results (bacterial identification, load), clinical data [forced expiratory volume in 1 s (FEV₁), body mass index, pulmonary exacerbations/stable status], and antibiotic courses were recorded.

Antimicrobial susceptibility testing was performed according to the 2016 recommendations of the Antibiogram Committee of the French Society for Microbiology¹. Disk diffusion method was performed for amikacin, amoxicillin +/- clavulanic acid, aztreonam, cefalotin, cefepime, cefoxitin, cefpirom, cefotaxime, ceftazidime, chloramphenicol, ciprofloxacin, colistin, cotrimoxazole, fosfomycin, gentamicin, imipenem, isepamicin, levofloxacin, moxalactam, nalidixic acid, netilmicin, ofloxacin, piperacillin, piperacillin/tazobactam, tetracycline, ticarcillin +/- clavulanic acid, and tobramycin (Bio-Rad^®). E-test (bioMérieux^®) was performed when inhibition zone diameters were observed around disks, and minimal inhibitory concentration (MIC) results that differed by more than two dilutions were considered as different. For ciprofloxacin, MICs were determined by the agar dilution reference method.

Motility of Pandoraea isolates was tested in semi-solid agar according to Kirov et al. (2004). Biofilm formation was evaluated after 72 h of growth for Pandoraea isolates and after 48 h of growth for P. aeruginosa and S. aureus at 37°C in microtitre plates NUNC^®-polystyrene (Abdouchakour et al., 2015). Isolates were categorized as non-adherent or weakly, moderately, or strongly adherent according to Stepanovic et al. (2000). Experiments were performed in triplicate.

Virulence assays were performed in Caenorhabditis elegans model as previously described (Mosser et al., 2015), except that nematode growth medium (NGM) agar plates were inoculated with bacterial culture grown in Trypticase Soy Broth (TSB) for 24 h and incubated at 37°C for 28 h. Each experiment, including 2–3 plates per strain, was conducted once, except for Escherichia coli OP50, avirulent in C. elegans model, and Pandoraea Pp14′ that were included in each of the four experiments conducted during the study. The effect of Pandoraea culture supernatant on virulence and biofilm formation of two co-colonizing isolates (S. aureus Sa14 and P. aeruginosa Pa13) was tested by replacing TSB used for the initial culture of Sa14 and Pa13 by a mix of TSB 2X and Pandoraea culture supernatant (0.22 μm filtration of 48 h culture in TSB at 37°C) (vol/vol).

16S rRNA and gyrB gene sequencing were performed as previously described (Carlier et al., 2002; Coenye and LiPuma, 2002). Pulsed-field gel electrophoresis (PFGE) after DNA macrorestriction by XbaI (Dupont et al., 2015), 16S rRNA gene PCR-temporal temperature gel electrophoresis (TTGE) (Dupont et al., 2015), and Enterobacterial repetitive intergenic consensus sequence-based PCR (ERIC-PCR) fingerprints (Atkinson et al., 2006) were performed as previously described. Pulsotype clustering analysis was based on Dice coefficient and Unweighted Pair Group Method with Arithmetic mean (UPGMA)². Simpson’s Index of Diversity and Adjusted Wallace Coefficient (AWC) were calculated by using Comparing Partitions³. AWC was used to evaluate the degree of congruence among all isolates’ features.

Data were analyzed using GraphPad Prism version 5.00 for Windows (GraphPad Software, San Diego, CA, United States). Comparisons of nematode survival curves in presence of Pandoraea or co-colonizing S. aureus and P. aeruginosa, as well as virulence assays in presence of Pandoraea supernatant were performed using Log-rank tests. A P-value ≤ 0.05 was considered significant.

Ethics approval was obtained through the Institutional Review Board of the University hospital of Nîmes (Interface Recherche Bioéthique IRB n°16.1101) for this observational study that fell within routine practice with non-additional diagnostic and monitoring procedures applied to the patient and retrospective analysis of primary data derived from routine clinical care.

Thirty-one Pandoraea isolates were recovered from 18 sputum samples (2–3 colonial morphotypes differing by their size, color and/or form and visually detected during the routine analysis of the sputum samples were analyzed for 10 samples) over a 7-year period of colonization (May 2008–February 2015) in a 22-year-old CF male patient. Partial 16S rDNA sequences of 1445 bp were identical for 29 isolates, the two remaining sequences (isolates Pp6 and Pp8) differing by one identical single nucleotide. The most related 16S rDNA sequences for type strains were those of P. pulmonicola CCUG 38759^T (>99.6% identity) and Pandoraea pnomenusa CCUG 38742^T (99.2%). Partial gyrB gene sequences (672 bp) were identical for 30 isolates, the sequence of isolate Pp14′ displaying one point mutation. The gyrB sequences presented 93.5% identity with that of the closest species P. pulmonicola strain DSM 16583^T, a value compatible with the intraspecific variability described in other Pandoraea species (Coenye and LiPuma, 2002). These results suggested that the 31 isolates belonged to P. pulmonicola.

Molecular fingerprints revealed the P. pulmonicola isolates to be clonally related thereby confirming the chronic colonization (Figure 1A). Main pulsotype P1 was observed for 25 isolates while 4 other pulsotypes differing by no more than 5 DNA fragments from pulsotype P1, were each observed for one or two isolates (Figure 1). ERIC-PCR appeared more discriminative (seven profiles, E1 to E7, and Simpson’s Index of Diversity, SID = 0.862) than PFGE (SID = 0.351) but non-congruent with PFGE clusterisation (AWC = 0.003) (Figure 1B). The grouping of closely related ERIC-PCR profiles in clusters named A to D did not enhance the congruence with PFGE profiles clusterisation (Figure 1B). The 16S rRNA gene PCR-TTGE showed that a majority of isolates (29/31) harbored identical V3 region in all rrs copies. Copies with sequences differing by a guanine base addition (position 389 in the V3 region, E. coli K12 numbering; Sundquist et al., 2007) were found in the isolates Pp6 and Pp8 (Figure 1B), isolates which displayed in consequence different 16S rDNA sequences.

All isolates were multidrug resistant. For most antibiotics, no inhibition zone diameters were observed around disks. MICs were determined for piperacillin-tazobactam, cefotaxime, imipenem, cotrimoxazole, and ciprofloxacin. Differences of two dilutions or more were observed for cefotaxime MIC for Pp8 and for piperacillin-tazobactam MIC for Pp6 compared with MIC values obtained for other isolates (Figure 1B). For the three other antibiotics, MIC values were either identical or did not differ by more than one dilution and were not considered as significant differences in antimicrobial susceptibility; particularly, MICs of ciprofloxacin were 32 mg/L (10 isolates) or 64 mg/L (21 isolates). Resistance traits (MICs) were congruent with 16S rRNA gene PCR-TTGE profiles (AWC = 1.00). Swimming and swarming motilities were observed for 58 and 29% of the isolates, respectively. Eight P. pulmonicola isolates (26%) were weak biofilm producers, the others being non-adherent. In most cases (except Pp14/14′ and Pp10′/10″), different colonial morphotypes in the same sample differed in their ability to swim, to swarm or to adhere (Figure 1B). Motility and adhesion were unrelated to other traits, such as resistance or genotype (AWC < 0.1).

Variability was observed for all the characteristics studied, either for isolates recovered from serial samples but also between all isolates showing distinct colonial morphotypes within a sample. When results were combined, 24 sub-types were identified for the 31 clonally related isolates, highlighting that studied Pandoraea population was highly diverse (Figure 1B). The only association observed between studied characteristics was that the two isolates Pp6 and Pp8 (displaying 16S rDNA differing by one identical single nucleotide and intragenomic heterogeneity) were the only isolates with distinct antibiotypes. No chronological trend was observed for phenotypic and genotypic feature variability (Figure 1).

For all P. pulmonicola isolates, the survival time of C. elegans was lower than that observed for the E. coli control strain with time required to kill 50% of the worm population or time to death 50 (TD50) in four experiments ranging from 11.2 to 12 days (mean time: 11.5 days, standard deviation: 0.4) (data not shown). Isolates exhibited variable virulence levels toward C. elegans since TD50 ranged from 8.4 to 11.1 days (average 9.8 days) (Figure 2). The relation between virulence expressed in nematode, isolation date, isolate sub-type and clinical status appeared unclear, except for the most virulent isolate Pp14′ (TD50 in four experiments ranging from 7.8 to 8.9 days, mean time: 8.4 days, standard deviation: 0.5) recovered in January 2014 at the time of the most severe pulmonary exacerbation (Figure 2).

Both co-colonizing isolates of S. aureus (Sa14) and P. aeruginosa (Pa13) exhibited weak virulence in the nematode model (TD50 = 11.9 and 11.1 days, respectively), lower than those of P. pulmonicola Pp14, Pp14′, and Pp13, respectively grown from the same sample (P-value < 0.001) (Figure 3). Growth in the presence of P. pulmonicola supernatant did not affect the virulence of the co-colonizing isolates (P-value > 0.05) (Figure 3). Regarding biofilm formation, P. aeruginosa Pa13 was weak biofilm producer, while S. aureus Sa14 was moderately biofilm producer. No effects on S. aureus or P. aeruginosa biofilm production were observed in assays with Pandoraea supernatant (data not shown).

The patient is the only patient colonized by Pandoraea in our regional center caring for about 250 CF patients. He was diagnosed with CF at birth and is homozygous for the δ-F508 mutation. He was chronically colonized by P. aeruginosa and S. aureus and first acquired Pandoraea in May 2008. During the 7-year study period, the patient experienced 11 episodes of pulmonary exacerbation and received 12 courses of antibiotics (Figure 2). During the overall study period, a 10%-decrease of FEV₁ was noted while body mass index increased from 14 to 20.5. The most severe pulmonary exacerbation occurred in January 2014 and was associated with significant pulmonary impairment (Figure 2).

Pandoraea pulmonicola was recovered from all the sputum sampled during the study period. The patient was also persistently colonized by P. aeruginosa, S. aureus, and Candida albicans but P. pulmonicola was the dominant bacterium or one of the dominant bacteria in 72% of the sputum samples. P. pulmonicola load ranged from 2.10⁴ to ≥10⁸ CFU/ml, being ≥10⁸ CFU/ml in 67% of the samples (Figure 2). Samples were analyzed at the time of 7 out of the 11 episodes of exacerbation. In five cases, Pandoraea was the dominant pathogen and bacterial load were ≥10⁸ CFU/ml in four cases. Isolates displaying minor pulsotypes were mostly recovered at the time of pulmonary exacerbation (4/6 isolates, 67%), including isolate Pp14′ which was non-motile, non-adherent, and more virulent in C. elegans than other Pandoraea isolates. Contrarily, swimming and swarming motilities were mainly observed for isolates not recovered during exacerbation (13/18 isolates, 72% and 5/9 stains, 55.5%, respectively). Weak biofilm formation, 16S rRNA gene sequence heterogeneity and distinct antibiotypes were equally observed for isolates recovered during exacerbation or stable clinical status and no relation between MIC values and antimicrobial courses received by the patient was noted. Particularly, no modification in MIC of ciprofloxacin was noted after the two documented antimicrobial courses including ciprofloxacin in September 2013 and October 2014. The five isolates that both swam and swarmed were isolated in the first year of patient colonization (Figure 1B) but adhesion, as well as other features, appeared unrelated to the date of isolation.