Polymorphism of ftsI gene in Haemophilus influenzae and emergence of cefotaxime resistance in two Tunisian hospitals.

The decreased affinity to β-lactams in Haemophilus influenzae is usually caused by specific alterations in penicillin-binding protein 3 due to varieties of substitutions in ftsI gene. This study aimed to characterize the polymorphism of ftsI gene in 19 H. influenzae strains, isolated between 2014 and 2016 (different resistance phenotypes to β-lactams (n = 9) and susceptible strains (n = 10) used for comparative purposes). All strains were characterized for capsular type by PCR and agglutination tests and for β-lactam resistance by amplification and sequencing of ftsI. Biotyping and clonality were performed by API-NH and pulsed-field gel electrophoresis, respectively. Four strains were β-lactamase-negative ampicillin-resistant and five were β-lactamase-positive clavulanic-acid-resistant. One strain from each group was resistant to cefotaxime. Our isolates belonged mainly to biotype IV and I and were non-typeable and genetically unrelated. According to mutation profiles of their ftsI, strains were classified as group I (n = 3), group II (n = 4), group-III-like (n = 1) and group III (n = 1). All group II strains were further classified as subgroup IIb, except for one strain, which harboured a new mutation (N422I). Ampicillin MICs of β-lactamase-negative ampicillin-resistant strains were 6 to 12 times the MICs of susceptible strains. Only bla TEM-1 was detected in β-lactamase-positive clavulanic-acid-resistant strains, and was responsible for high MICs for ampicillin (>256 mg/L), whatever the ftsI mutational resistance group. The emergence of cefotaxime-resistant isolates in our country is a matter of concern and requires strict surveillance and rationalization of antibiotic use to preserve these molecules.

Introduction

Haemophilus influenzae is a commensal bacterium of the human upper respiratory tract, oropharynx and nasopharynx. It is one of the most frequent pathogens responsible for bronchopulmonary, ear, nose and throat infections [1]. Also, it represents the principal aetiology of invasive infections such as purulent meningitis, bacteraemia and epiglottitis mainly in infants [1]. β-Lactams, mainly third-generation cephalosporins, are active against H. influenzae. However, the emergence and spread of resistant strains worldwide can severely affect their efficacy [2]. Resistance to β-lactams in H. influenzae is predominantly mediated by TEM-1 or ROB-1 β-lactamase production and is associated with resistance to aminopenicillins, of which the activity spectrum is limited to penicillins. Strains producing β-lactamases are termed β-lactamase-positive ampicillin-resistant. The second mechanism of resistance is non-enzymatic, due to decreased affinity of β-lactams for the altered transpeptidase domain of penicillin-binding protein 3 (PBP3), involved in septal peptidoglycan synthesis and encoded by the ftsI gene [1]. Strains expressing this mechanism are termed β-lactamase-negative ampicillin-resistant (BLNAR). Resistance by the latter mechanism can affect penicillins, penicillin and penicillinase inhibitor associations, cephalosporins and carbapenems, depending on the number and the type of mutations in the ftsI gene [1]. Strains that accumulate the two mechanisms are termed β-lactamase-positive clavulanic-acid-resistant (BLPCAR) [1]. In BLNAR strains, amino acid substitutions are usually surrounding the conserved motifs Lys512-Thr-Gly (KTG) and Ser379-Ser-Asn (SSN) of the PBP3 transpeptidase domain. More than 40 substitutions have been described in the literature and it has been found that a single BLNAR isolate could accumulate from one to 11 substitutions [1]. According to specific substitutions, BLNAR isolates have been classified into four major mutational groups (I, II, III and III-like). In group I, His-517 was substituted by Arg and in group II Lys-526 was substituted by Asn. Ampicillin MICs of these groups varied between 0.5 and 8 mg/L and they are considered low BLNAR. Group III and group III-like were defined by the second substitution S385T in addition to the first one N526K or R517H, respectively. They present full resistance to ampicillin (MIC range 1–32 mg/L) and cephalosporins (MIC range 0.12–2 mg/L) and considered high BLNAR [3,4]. Furthermore, isolates from group III and group–III–like with the additional substitution L389F showed generally higher MICs to extended-spectrum cephalosporins and meropenem. Hence, two new groups—III+ and group–III–like+—have been proposed by Skaare et al. for these strains [2]. In H. influenzae enzymatic resistance has historically predominated, with a prevalence >20% in many European countries, Australia and Canada. Recent studies showed that the prevalence of β-lactamase-positive ampicillin-resistant strains is stabilizing or decreasing. By contrast, a significant increase of BLNAR phenotype was observed in these same countries [1]. The situation in Japan is different and usually marked by high prevalence of BLNAR strains. In addition, high BLNAR strains have been rarely reported outside Asian countries, particularly Japan and Korea [5].

In Tunisia, multicentric studies on antimicrobial resistance of H. influenzae showed that β-lactamase production was the most common mechanism of resistance to β-lactams during all years of surveillance (1999–2017). The prevalence of β-lactamase-positive ampicillin-resistant strains varied from 17.3% in 1999 to 36.6% in 2017. BLNAR isolates emerged in 2006 at low frequencies with significant increase from 2.9% in 2007 to 8.2% in 2017 (www.infectiologie.org.tn). In Tunisia, BLNAR strains are routinely detected by phenotypic methods and few data are available on the genetic classification of their ftsI gene. Accordingly, we aimed to characterize the polymorphism of ftsI gene in a Tunisian collection of 19 H. influenzae strains, isolated between 2014 and 2016, and to assess the clonality among them.

Materials and methods

Strain collection

Haemophilus influenzae isolates included in the study were distributed as follows:

Seven strains recovered from the microbial Laboratory of Charles Nicolle Hospital, including three low BLNAR isolates (β-lactamase-negative, ampicillin MIC >1 mg/L) and four BLPCAR isolates (ampicillin/clavulanic acid MIC >2 mg/L).

Two cefotaxime-resistant H. influenzae strains (Hi16 and Hi19) isolated at the microbiology laboratories of the Charles Nicolle (Tunis city) and Abderrahman Mami (Ariana city) hospitals, respectively.

Ten control isolates, fully susceptible to β-lactams: β-lactamase-negative ampicillin-susceptible strains, randomly selected, used for comparative purposes.

Strain identification

Isolates were identified through Gram-staining that usually showed a pleomorphic Gram-negative bacilli, their requirements for β-NAD+ (V factor) and haemin (X factor) for growth and by API NH (bioMérieux, Marcy-l’Étoile, France). Chocolate agar plates (bioMérieux) were routinely used for subcultures of H. influenzae. Biotypes were determined using indole, urease and ornithine decarboxylase reactions revealed by API NH [6].

Antimicrobial susceptibility testing

Antimicrobial susceptibility was determined by disc diffusion method, according to the European Committee on Antimicrobial Susceptibility Testing recommendations (CA-SFM/EUCAST) [7]. The antibiotics tested were penicillin G (1 μg), ampicillin (2 μg), amoxicillin/clavulanic acid (2 μg/1 μg), cefotaxime (5 μg), nalidixic acid (30 μg), ciprofloxacin (5 μg), tetracycline (30 μg), chloramphenicol (30 μg), rifampicin (5 μg) and trimethoprim-sulfamethoxazole (1.25–23.75 μg). β-Lactamase production was determined by the chromogenic cephalosporin test (cefinase test) with nitrocefin as the substrate (bioMérieux). MICs of ampicillin, amoxicillin/clavulanic acid and cefotaxime were determined by E-test strips (bioMérieux) and were interpreted according to the EUCAST breakpoints.

Amplification and sequencing of ftsI gene

Mutations in ftsI gene encoding PBP3 were identified by PCR and sequencing as previously described [8].

β-lactamases gene detection

Haemophilus influenzae strains with BLPCAR phenotype were screened for bla and bla genes using multiplex PCR as previously described [9].

Capsular typing and genetic relationship

Capsular type was identified by slide agglutination using specific anti-sera (Difco, BD, Le Pont de Claix, France) and was confirmed by PCR [10]. The H. influenzae ATCC 10211 (strain with capsular type b was used as a positive control.

The genetic relationship between isolates was analysed by pulsed-field gel electrophoresis (PFGE). The PFGE Pulse Net protocol of Escherichia coli was adapted to H. influenzae strains using SmaI restriction enzyme (New England BioLabs, Ipswich, MA, USA; https://www.cdc.gov/pulsenet/pathogens/pfge.html). DNA profiles were examined with FP-Quest software (BioRad, Marnes la Coquette, France) and using the Dice coefficient and UPGMA (unweighted pair group method with arithmetic mean). Clusters were defined as DNA patterns sharing ≥70% similarity, which corresponds to the possibly related criteria of Tenover et al. [11].

Results

Clinical data

Demographic and clinical data of patients are summarized in Table 1. Most H. influenzae strains were isolated from sputum (n = 15). They were mainly recovered from pneumology (n = 5), otorhinolaryngology (n = 3) and paediatrics (n = 3) wards. Fifteen (78.9%) infections were classified as community-acquired (Table 1).

Table 1

Clinical characteristics and biotype of Haemophilus influenzae strains (n = 19)

	Reference strains	Date of isolation	Patient age/Gender	Specimens	Wards	Infection origin	Biotype
β-lactam-susceptible strains (Control group)	Hi 2	18/07/2014	79 years/F	Bronchial secretion	Intensive care unit	CA	I
	Hi 3	11/03/2015	–/F	Sputum	External consultation	CA	IV
	Hi 5	12/05/2015	–/M	Sputum	Pneumology	CA	IV
	Hi 6	01/06/2015	81 years/M	Bronchial secretion	Intensive care unit	HA	IV
	Hi 7	10/07/2015	–/F	Sputum	Pneumology	CA	I
	Hi 8	28/07/2015	–/M	Bronchial secretion	Pneumology	CA	I
	Hi 9	12/08/2015	75 years/M	Bronchial secretion	Surgery	HA	IV
	Hi 11	12/09/2015	14 years/M	Sputum	Paediatrics	CA	I
	Hi 12	30/10/2015	–/F	Sputum	Paediatrics	CA	I
	Hi 13	11/11/2015	14 years/F	Pus	Otorhinolaryngology	CA	I
β-lactam-resistant strains	Hi 1	17/07/2014	55 years/F	Sputum	Otorhinolaryngology	CA	I
	Hi 4	10/04/2015	–/F	Sputum	Pneumology	CA	II
	Hi 10	23/09/2015	9 years/F	Sputum	Paediatrics	CA	VIII
	Hi 14	30/10/2015	–/F	Sputum	Pneumology	CA	I
	Hi 15	30/12/2015	1 year/M	Pus	ForensicMedicine	CA	IV
	Hi 16	20/01/2016	62 years/M	Sputum	Gastroenterology	HA	IV
	Hi 17	05/03/2016	–/M	Pus	Otorhinolaryngology	CA	IV
	Hi 18	11/03/2016	–/F	Sputum	Paediatrics	CA	IV
	Hi 19a	30/08/2016	72 years/M	Sputum	Thoracic surgery	HA	IV

CA, community-acquired infections were defined as infections in which the onset of patient symptoms occurred before admission or within 48 h of admission to the hospital; HA, hospital-acquired infections were defined as infections in which the onset of symptoms occurred more than 48 h after admission.

Isolate from Abdurrahman Mami hospital.

For the two H. influenzae strains resistant to cefotaxime (Hi16 and Hi19) and given the importance of this novel resistance, detailed clinical histories of patients are given below.

Clinical observation no. 1

Hi16 was recovered from a 61-year-old man hospitalized in the gastroenterology ward for decompensated cirrhosis post-viral hepatitis C. Two weeks previously, he was treated with cefotaxime (4 g/day for 15 days) for bacteraemia caused by E. coli. Four days after having completed his course of antimicrobial therapy, his clinical state deteriorated and he developed stage II encephalopathy, with dyspnoea and recurrence of fever. An infectious investigation was conducted including chest X-ray, which revealed diffuse alveolar images. Sputum and urine cultures were positive for H. influenzae resistant to cefotaxime and Enterococcus faecium resistant to glycopeptides, respectively. The patient was treated with ofloxacin (800 mg/day) for 10 days, with clinical and biological improvement.

Clinical observation no. 2

Hi19 was isolated from a 72-year-old man hospitalized in thoracic surgery for acute coronary syndrome. He was a former smoker and was previously hospitalized for acute exacerbation of his chronic obstructive pulmonary disease in 2006 and in 2015. The patient did not receive antibiotics in the previous 6 months. For the current episode, an infectious investigation, including sputum and urine cultures was carried out. The two specimens were positive for H. influenzae resistant to cefotaxime and E. coli, respectively. The patient was treated with a high dose of cefotaxime (6 g/day) for 10 days and with ciprofloxacin (1500 mg/day) on discharge.

Strain characterization

The H. influenzae strains were classified into four biotypes. Biotypes IV and I were identified in nine and eight strains, respectively. Susceptible isolates belonged mainly to biotype I (60%), whereas 55.5% of resistant isolates belonged to biotype IV. Biotypes II and VIII were identified in two strains each (Table 1).

All strains were non-typeable by slide agglutination as well as by PCR amplification. PFGE analysis of H. influenzae isolates showed 19 unrelated pulsotypes (Fig. 1). They were susceptible to nalidixic acid, chloramphenicol and rifampicin, Eight and three strains were resistant to tetracycline and cotrimoxazole, respectively. All four β-lactamase-producing isolates harboured blaTEM-1 gene (Table 2).

Fig. 1

Dendrogram of pulsed-field gel electrophoresis DNA patterns of Haemophilus influenzae strains obtained after the UPGMA analysis of the Dice's coefficient. BLNAS, β-lactamase-negative ampicillin-susceptible; BLNAR, β-lactamase-negative ampicillin-resistant; BLPCAR, β-lactamase-positive clavulanic-acid-resistant).

Table 2

Antimicrobial resistance profiles and deduced amino acid substitution patterns in the transpeptidase region of penicillin-binding protein 3 (PBP3) of Haemophilus influenzae strains

Reference Strains	Resistance profile	Cefinase/bla genes	MICs			ftsI groupa		Mutations in ftsI gene
			AMP	AMC	CTX		D350	S357	M377	S385	L389	P392	N422	G490	A502	V511	R517	N526	S532	F535	V547	I549	Y557	N569
Hi 3	—	—	0.25	0.25	0.012	NA
Hi 7	—	—	0.25	0.38	0.016
Hi 9	TET	—	0.38	0.25	0.012
Hi 12	TET	—	0.19	0.125	0.023
Hi 13	—	—	0.25	0.125	0.016							P
Hi 11	TET	—	0.125	0.125	0.016															I
Hi 8	SXT	—	0.25	0.25	0.012															I
Hi 6	—	—	0.75	0.25	0.047																I
Hi 5	TET	—	0.5	0.38	0.08				s												I
Hi 2	—	—	0.25	0.25	0.016																	F
Hi 14	AMX-TET	—	2	0.125	0.064	Group I											H
Hi 1	AMX	—	4	0.125	0.023											A	H
Hi 10	AMX-TET	—	4	0.38	0.023											A	H
Hi 17a	AMX-AMC-SXT	+/blaTEM-1	>256	2	0.064	Group IIb	N		I						V			K		I
Hi 18 a	AMX-AMC	+/blaTEM-1	>256	4	0.047	Group Iib	N		I					E	V			K
Hi 4 a	AMX-AMC-TET-SXT	+/blaTEM-1	>256	2	0.047	Group Iib	N		I					E	V			K			I
Hi 15b	AMX-AMC	+/blaTEM-1	>256	2	0.032	Group II	N		I				I∗		V			K
Hi 19	AMX-AMC-CTX	+/blaTEM-1	>256	>256	1	Group–III–like+	N	N	I	T	F						H		T		I		H	S
Hi 16	AMX-AMC-CTX-TET	—	24	4	0.5	Group III+	N	N	I	T	F				T			K

AMC, ampicillin/clavulanic acid; AMP, ampicillin, CTX, cefotaxime.

a Classification according to Ubukata et al. [30] (Group I, II and III), Dabernat et al. [3] (Subgroup II: IIa, IIb, IIc and IId), García-Cobos et al. [4] (Group–III–like), Skaare et al. [23] (Group III+ and Group–III–like+): –, negative; NA, non-assigned resistance group, ∗, new mutation.

ftsI genotypes and correlation with β-lactam MICs

Sequence analysis of the transpeptidase region of PBP3 showed a total of 19 substitutions in 18 positions. The most frequent mutations were (D350N), (S357I), (R517H) and (N526K). In the Hi15 BLPCAR isolate, a new mutation was observed at position 422 (N422I). For this strain, MICs of ampicillin, amoxicillin/clavulanic acid and cefotaxime were >256, 2 and 0.032 mg/L, respectively.

Among strains with a β-lactamase-negative ampicillin-susceptible phenotype (control group), four did not harbour mutations in the ftsI gene and six displayed different point mutations (Table 2). Ampicillin MICs varied from 0.19 to 0.75 mg/L (MIC50 0.25 mg/L).

According to the mutation profile of their ftsI gene, strains with BLNAR phenotype (n = 4) were classified as group I (n = 3) and group III+ (n = 1) and those with BLPCAR phenotype (n = 5) were classified as group II (n = 4) and group–III–like (n = 1) (Table 2). Three strains (Hi4, Hi17 and Hi18) from group II were further assigned to subgroup II-b, and were also TEM-1 β-lactamase producers. The remaining strain (Hi15) that had a new mutation (N422I) could not be assigned to any of the previously described group II subgroups. Ampicillin MICs varied between 2 and 4 mg/L in group I and was >256 mg/L in group II.

Resistance to cefotaxime in strains Hi16 and Hi19 was associated with S385T and L389F mutations, respectively. Strain Hi19 was also a TEM-1 producer. Cefotaxime MICs were 0.5 mg/L and 1 mg/L for Hi16strain (group III+) and Hi19 strain (group III like+), respectively.

High resistance level of ampicillin (>256 mg/L) was associated with TEM-1 production whatever the ftsI mutational resistance group. ROB enzyme has not been found in our collection.

Discussion

In this study, molecular mechanisms of resistance to β-lactams in clinical H. influenzae isolates showing different β-lactam resistance phenotypes were investigated. All strains were non-typeable by slide agglutination as well as by PCR amplification. The predominance of non-typeable strains was also reported in other Tunisian studies, and in Korea, Spain and France [12,13]. However, previous reports demonstrated the limitation of slide agglutination for H. influenzae serotyping in comparison with the results provided by PCR [14,15]. This may be related to the individual characteristics of expression of capsule and/or other antigens on the bacterial surface [16,17].

In Tunisia, until 2002, all invasive infections in young children were caused by H. influenzae b strains [18], justifying the introduction of the anti-Hib conjugate vaccine. However, given its high cost, it was abandoned at the beginning of 2006. Then, based on extensive evidence demonstrating the economic impact of the anti-Hib conjugate vaccine through direct and indirect cost savings, as well as through contributions to the Tunisian economy in general, this vaccine was reintroduced into the vaccination schedule in 2011 [19].

In the present study, PFGE analysis of the H. influenzae isolates showed diverse pulsotypes, which is in agreement with the genetic heterogeneity of non-capsulated H. influenzae previously reported [1]. Also, according to resistance phenotype, many studies have shown that BLNAR and BLPCAR strains of H. influenzae are genetically diverse with a general absence of related PFGE DNA profiles [1]. Otherwise, clonal spread of BLNAR strains of serotype b was reported by a limited number of studies in Japan [20,21]. In comparison, some local outbreaks caused by BLNAR H. influenzae strains have been reported, in Norway, Spain and Canada. of BLNAR H. influenzae strains that were caused by closely related clones [4,8,22].

Among the β-lactamase-negative ampicillin-susceptible phenotypes (n = 10), four strains of our collection had the prototype amino acid sequence of the transpeptidase region of PBP3. In the remaining six strains, different point mutations were found. These observations were previously reported [2,8,23]. Many silent mutations have been reported in the DNA region encoding the transpeptidase domain of PBP3 in susceptible strains [23]. The results of García-Cobos et al. showed that β-lactams susceptible H. parainfluenzae strains presented various substitutions not previously assigned to any ftsI-resistant groups [4].

According to mutational profiles of ftsI genes, BLNAR and BLPCAR isolates of our collection were classified as group I, II, IIb, III-like+ and III+. Strain Hi15 belonging to group II, but not assigned to any group II subgroups, showed a new mutation (N422I) in its ftsI gene, suggesting a novel subgroup II. The β-lactam resistance in H. influenzae due to ftsI mutations is increasing worldwide [22]. Low-level resistance isolates, mainly with the N526K substitutions, predominated in most geographical regions, whereas high-level resistance isolates with additional L389F substitution (ampicillin MIC50: 128 mg/L and cephalosporins MIC50: 1 mg/L) are common in Japan and South Korea. Epidemiological data from European countries and Canada showed a gradual increase in resistant isolates from group II with sporadic cases of cefotaxime-resistant strains [1,13,22,24,25].

In Tunisia, resistance to cefotaxime has been recently reported among six H. influenzae isolates from Habib Bourguiba hospital. These strains belonged to group IIa, group IIb, group III and group–III–like [26]. The two cefotaxime-resistant strains described in our study belonged to group III+ and group–III–like+. According to clinical data, H. influenzae-resistant strains were isolated from two elderly individuals with severe underlying diseases, chronic cirrhosis in one and chronic obstructive pulmonary disease in the other. It has been demonstrated that people suffering from chronic obstructive pulmonary disease are frequently colonized by H. influenzae in their respiratory tract [12,27]. In addition, previous antimicrobial treatment by cefotaxime and iterative hospitalizations are contributing factors for selection of such resistant strains. According to Dabernat et al., the inappropriate use of oral antibiotics for the treatment of community-acquired bronchopulmonary and upper respiratory tract infections seems to be responsible for the selection of BLNAR strains [3].

In our series, the five BLPCAR isolates harboured the blaTEM-1 gene. This finding is in agreement with previous Tunisian studies [28,29]. Although the predominance of TEM-1 β-lactamase was largely reported worldwide, ROB-1 is rarely found outside North America [1].

Our study presents two major limitations. First, clinical data were collected retrospectively and detailed information for all patients, such as age, co-morbidities, severity of diseases, antimicrobial treatment history and clinical outcomes could not be obtained. Second, our results gave limited data on the distribution of the ftsI group in our country because of the low number of studied strains. Further large studies including other hospital centres will be necessary to assess the real clinical impact of this emerging resistance and to identify additional risk factors for selection of resistant strains.

In conclusion, this study revealed the diversity of mutation profiles in ftsI gene among BLNAR and BLPCAR H. influenzae strains. Our results further indicate that these strains were non-capsulated and were genetically unrelated. The emergence of strains resistant to extended spectrum β-lactams is alarming and requires strict epidemiological surveillance. Furthermore, rationalization and strict control of antibiotic use, mainly in the community, is needed to preserve the activity of these molecules.

Conflict of interest

The authors have no conflicts of interest to declare. No funding was received for the study.

Authors' contributions

FS and SM were responsible for the conception or design of the work; SI, ME and SL performed the data collection; and FS performed the data analysis and interpretation. FS drafter the article; BBBI critically revised the article and the final approval for publication was given by FS, SI, SM, ME, GA, SL, SA and BBBI.