In Vivo Secretion of β-Lactamase-Carrying Outer Membrane Vesicles as a Mechanism of β-Lactam Therapy Failure

Outer membrane vesicles carrying β-lactamase (βLOMVs) protect bacteria against β-lactam antibiotics under experimental conditions, but their protective role during a patient’s treatment leading to the therapy failure is unknown. We investigated the role of βLOMVs in amoxicillin therapy failure in a patient with group A Streptococcus pyogenes (GAS) pharyngotonsillitis. The patient’s throat culture was examined by standard microbiological procedures. Bacterial vesicles were analyzed for β-lactamase by immunoblot and the nitrocefin assay, and in vivo secretion of βLOMVs was detected by electron microscopy. These analyses demonstrated that the patient’s throat culture grew, besides amoxicillin-susceptible GAS, an amoxicillin-resistant nontypeable Haemophilus influenzae (NTHi), which secreted βLOMVs. Secretion and β-lactamase activity of NTHi βLOMVs were induced by amoxicillin concentrations reached in the tonsils during therapy. The presence of NTHi βLOMVs significantly increased the minimal inhibitory concentration of amoxicillin for GAS and thereby protected GAS against bactericidal concentrations of amoxicillin. NTHi βLOMVs were identified in the patient’s pharyngotonsillar swabs and saliva, demonstrating their secretion in vivo at the site of infection. We conclude that the pathogen protection via βLOMVs secreted by the flora colonizing the infection site represents a yet underestimated mechanism of β-lactam therapy failure that warrants attention in clinical studies.


Introduction
Group A Streptococcus pyogenes (GAS) is the most common cause of acute bacterial pharyngotonsillitis, accounting for 20-30% of cases in children and 5-15% of cases in adults [1]. GAS is highly susceptible to β-lactam antibiotics, so penicillin and amoxicillin are the treatments of choice [1]. However, the inabilities of these antibiotics to eradicate GAS from patients with pharyngotonsillitis have been increasingly reported [2]. One cause of the therapy failure is colonization of the pharynx and tonsils by β-lactamase-producing bacteria such as Moraxella catarrhalis, Haemophilus influenzae, and Staphylococcus aureus that protect GAS against β-lactam antibiotics [2]. Yet, the mechanisms of this protection are incompletely understood. Here, we investigated the involvement of bacterial outer membrane vesicles (OMVs) in amoxicillin therapy failure.

Patient and Microbiological Examinations
A 39-year-old man presented with acute pharyngotonsillitis accompanied by a fever up to 39.8 • C. His throat swabs were cultured on blood agar and chocolate agar (Thermo Fisher Scientific, Prague, Czech Republic) and the isolates were identified by standard bacteriological procedures [27] and MALDI-TOF mass spectrometry (Microflex LT, Bruker Daltonics, Bremen, Germany). Antimicrobial susceptibilities were determined by the disc diffusion method (discs from Oxoid, Brno, Czech Republic) and the broth microdilution method according to the Clinical and Laboratory Standards Institute (CLSI) procedures and criteria [28]. The bla TEM-1 gene was detected by PCR followed by digestion of 600 bp amplicon with MboI (New England Biolabs, Frankfurt am Main, Germany) as described previously [29].

Ethical Approval
The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of the National Institute of Public Health, Prague (protocol code EK-SZU/08076/2021), on 1 June 2021. Written informed consent was obtained from the patient.

Statistical Analysis
Data were analyzed with one-way ANOVA (analysis of variance); p < 0.05 was considered significant.

Amoxicillin Treatment Failure in a Patient with GAS Pharyngotonsillitis
The throat culture of the patient with pharyngotonsillitis grew GAS, which was susceptible to penicillin, ampicillin, and amoxicillin ( Table 2). The patient was treated with amoxicillin (750 mg three times daily, p.o.) for 10 days, but no improvement was observed. A repeated throat culture performed after completing the amoxicillin treatment continued growing amoxicillin-susceptible GAS. Moreover, this second culture revealed NTHi, which was resistant to amoxicillin (MIC of 16 µg/mL), carried the bla TEM-1 gene, and was likely selected by the amoxicillin therapy. It was susceptible to amoxicillin/clavulanate ( Table 2). The patient was treated with amoxicillin/clavulanate (1 g twice daily, p.o.) and fully recovered after 4 days; the therapy was continued for up to 10 days. Control throat cultures performed 2 days and 10 days after termination of the amoxicillin/clavulanate therapy were negative for GAS and NTHi.  1 The data refer to the GAS isolates obtained before amoxicillin treatment and after amoxicillin treatment.

NTHi Patient's Isolate Secretes β-Lactamase-Carrying OMVs That Are Induced by Amoxicillin
To gain insight into the role of NTHi OMVs in the amoxicillin therapy failure, we isolated OMVs from the NTHi patient's isolate (Figure 1a) and analyzed them for the presence of β-lactamase and β-lactamase activity. The OMVs contained β-lactamase (Figure 1b), which was located inside OMVs, as demonstrated by its protection against proteinase K (PK) in the PK assay ( Figure 1b). The β-lactamase was enzymatically active, as evidenced by the ability of OMVs to cleave the β-lactamase substrate nitrocefin (Figure 1c). Notably, the β-lactamase activity (Figure 1c) and the amount (Figure 1d) of OMVs produced by NTHi at the time of its isolation from the patient's tonsils (reflecting the situation in vivo during therapy) significantly decreased when the isolate was passaged in vitro in a medium without amoxicillin, and significantly increased when amoxicillin in concentrations reported in the tonsillar tissue during amoxicillin treatment (0.17 µg/mL to 3.9 µg/mL) [30][31][32] was added to the NTHi culture (Figure 1c,d). Thus, amoxicillin in concentrations reached in the tonsils during therapy significantly increased secretion and enzymatic activity of NTHi βlactamase-containing OMVs. This led us to hypothesize that these β-lactamase-containing, amoxicillin-induced OMVs (hereafter termed NTHi OMVs βL+AMX+ ) were involved in the amoxicillin failure to eradicate GAS from the patient by protecting GAS against amoxicillin. OMVs βL+AMX+ subjected to proteinase K (PK) assay, which demonstrates intravesicular localization of β-lactamase. (c,d) β-lactamase activities (c) and the amounts (d) of OMVs produced by NTHi freshly isolated from the patient's throat culture (fresh isolate), by NTHi passaged twice in BHI broth without amoxicillin (P1 and P2, AMX-), and by NTHi from passage 2 grown in BHI broth with amoxicillin concentrations reported in the tonsils during therapy (0.17 µg/mL, 1.1 µg/mL, or 3.9 µg/mL). Data are presented as means ± standard deviations from three independent experiments; ** p < 0.01 compared to fresh isolate; xx p < 0.01 compared to P2 AMX-; xxx p < 0.001 compared to P2 AMX-(statistical analysis was performed with one-way ANOVA).

NTHi OMVs βL+AMX+ Protect GAS against Bactericidal Concentrations of Amoxicillin
To test the hypothesis that NTHi OMVs βL+AMX+ are involved in the amoxicillin failure to eradicate GAS from the patient, we determined whether these OMVs protect GAS against the reported amoxicillin tonsillar concentrations of 0.17 µg/mL and 3.9 µg/mL [30,32], which represent~10-fold and 244-fold MICs, respectively, for this isolate (MIC of 0.016 µg/mL) ( Table 2). To this end, GAS growth was monitored for 24 h in the presence of each amoxicillin concentration and NTHi OMVs βL+AMX+ in the doses of 724 µg/mL or 1.2 mg/mL, which were induced by the respective amoxicillin concentrations (Table 1). Indeed, each NTHi OMVs βL+AMX+ dose protected GAS against amoxicillin, with the protection being slightly delayed with 724 µg/mL of NTHi OMVs βL+AMX+ against 3.9 µg/mL of amoxicillin (Figure 2a,b). This demonstrated that NTHi OMVs βL+AMX+ in the amounts induced by amoxicillin concentrations reached in the tonsils during therapy protected GAS against bactericidal effect of amoxicillin. No GAS protection was conferred by β-lactamase-negative OMVs from a control amoxicillin-susceptible NTHi (Figure 2a,b), indicating that the β-lactamase associated with NTHi OMVs βL+AMX+ was the GAS-protecting component.

NTHi OMVs βL+AMX+ Increase Amoxicillin MIC for GAS
To elucidate the basis for the NTHi OMVs βL+AMX+ -mediated GAS protection against amoxicillin, we determined the effect of these OMVs on amoxicillin MIC for GAS. We found that in the presence of 724 µg/mL and 1.2 mg/mL of NTHi OMVs βL+AMX+ , the amoxicillin MIC for GAS increased from 0.016 µg/mL to 4 µg/mL (250-fold) and to 16 µg/mL (1000-fold), respectively (Table 3), making GAS resistant to amoxicillin. Importantly, both the amoxicillin MIC increase for GAS and the GAS protection against amoxicillin via NTHi OMVs βL+AMX+ were inhibited by the β-lactamase inhibitor clavu-lanate (Figure 2a,b, Table 3), confirming that these effects were mediated by the NTHi OMVs BL+AMX+ -associated β-lactamase. Moreover, GAS failed to grow on a medium with amoxicillin in the absence of NTHi OMVs βL+AMX+ (Figure 2c), demonstrating that these OMVs, not acquisition of the bla TEM-1 gene from NTHi, accounted for its amoxicillin resistance. The absence of bla TEM-1 in GAS was confirmed by PCR (Table 2). Table 3. The influence of NTHi OMVs βL+AMX+ on amoxicillin MIC for GAS isolated from the patient's throat culture.

NTHi β-Lactamase-Carrying OMVs Are Secreted In Vivo at the Infection Site
To provide a final piece of evidence for the involvement of NTHi OMVs βL+AMX+ in the amoxicillin therapy failure, we searched for their secretion in vivo at the site of infection. We identified NTHi bacteria secreting β-lactamase-carrying OMVs as well as released, free β-lactamase-carrying OMVs in the patient's pharyngotonsillar swabs ( Figure  3a), tonsillar crypt exudate (Figure 3b), and saliva (Figure 3c). Taken together, our findings demonstrate that amoxicillin-resistant NTHi colonizing the pharyngotonsillar mucosa of the GAS-infected patient secreted in situ β-lactamase-carrying OMVs, which were inducible by amoxicillin and protected GAS against the antibiotic, thereby accounting for the therapy failure.

Discussion
This study brings a new insight into the mechanisms of amoxicillin therapy failure in patients with GAS pharyngotonsillitis. The involvement of β-lactamase-carrying OMVs secreted by NTHi colonizing the patient's pharynx and tonsils in this failure is supported: (i) by the induction of NTHi OMV secretion and OMV-associated β-lactamase activity by amoxicillin concentrations reached in the tonsils during therapy; (ii) by the ability of NTHi OMVs βL+AMX+ to significantly increase amoxicillin MIC for GAS and to protect GAS against bactericidal concentrations of amoxicillin; (iii) by the inhibition of each of these NTHi OMVs βL+AMX+ -mediated effects by the β-lactamase inhibitor clavulanate; (iv) by the inability of GAS to resist amoxicillin in the absence of NTHi OMVs βL+AMX+ , which is in accordance with its excellent susceptibility to amoxicillin [1]; and (v) by the secretion of NTHi OMVs βL+AMX+ in vivo at the site of infection. This is, to the best of our knowledge, the first evidence that OMV-mediated protection of bacteria against β-lactam antibiotics previously observed under in vitro conditions [18,[22][23][24][25][26] has a clinical parallel in the ability of β-lactamase-carrying OMVs secreted in vivo to protect pathogens against β-lactams during patients' treatment, thus leading to therapy failure. For patients with GAS pharyngotonsillitis, this mechanism may be of a particular importance, since more than one fourth of children with GAS pharyngotonsillitis have their tonsils colonized with H. influenzae or M. catarrhalis [37], both of which can produce β-lactamase-carrying OMVs ( [22,23], this study). Mechanistically, the structural identity between the bacterial outer membrane and OMV membrane allows β-lactams to enter, through the porin channels, the OMV lumen, where β-lactamase, originating from the bacterial periplasm, is located and hydrolyzes the antibiotics [22][23][24]38]. Through this process, OMVs secreted outside bacterial cells serve as a first line of protection against β-lactams before the antibiotics reach the target bacterial population. Since OMVs secreted by Gram-negative β-lactam-resistant bacteria in vitro carry a broad spectrum of β-lactamases [18,[22][23][24][25][26], it is likely that a similar mechanism that we have described for amoxicillin plays a role in therapeutic failures of other β-lactam antibiotics including carbapenems, which are the "last resort" β-lactams used to combat multidrug resistant pathogens [26]. Moreover, experimental data suggest that OMVs may also be involved in antimicrobial resistance in other ways, including the sequestration of membrane active antibiotics (polymyxin B, colistin) [8,20,21,39] and dissemination of antibiotic resistance genes [33,40,41]. Thus, OMVs may serve as universal bacterial tools contributing, by different mechanisms, to antibiotic resistance. A broad involvement of OMVs in antimicrobial resistance is strongly supported by the observations that the increase in OMV secretion (e.g., by hypervesiculating mutans) increased the resistance, and the reduction in or inhibition of OMV secretion increased the susceptibility of various bacteria to a range of antibiotics [21,39,42].
The mechanism by which amoxicillin (and other β-lactam antibiotics such as imipenem [17] and meropenem [16]) increases OMV production is presently not known. Based on the models of OMV biogenesis [6] and the mechanism of action of β-lactam antibiotics [43], we hypothesize that the inhibition of peptidoglycan polymerization due to the β-lactam binding to the penicillin-binding proteins [43] plays a key role. This hypothesis is supported by the occurrence of OMV budding at the sites of locally decreased crosslinking between the peptidoglycan and the outer membrane [6]; it is in accordance with the peptidoglycan being a central structure that accounts, via its crosslinks with various membrane proteins, for the stability of the bacterial envelope [6].
The therapeutic success of amoxicillin/clavulanate in our patient demonstrated that amoxicillin's failure to eradicate GAS was not due to a poor tonsillar penetration of the drug. However, the amoxicillin concentration in the patient's tonsils could not be determined, as he did not undergo a tonsillectomy. This is the reason why we used the concentrations reported in the tonsils of amoxicillin-treated patients who did undergo a tonsillectomy [30][31][32] to determine the effects of amoxicillin on the amount of OMVs produced by NTHi and the OMV β-lactamase activity. Since these concentrations encompass a broad range (0.17 µg/mL-3.9 µg/mL), it is likely that the amoxicillin concentration in the tonsils of our patient was within this range. To further evaluate the mechanism of β-lactam therapy failure reported in this study in other clinically relevant situations, we will continue our investigations in additional patients with pharyngotonsillitis, in whom a coinfection with β-lactam-susceptible (GAS and others) and beta-lactam-resistant bacteria is detected, and the β-lactam antibiotic therapy failure occurs.

Conclusions
The pathogen protection via β-lactamase-carrying OMVs secreted in situ by the flora coinhabiting the infection site represents a yet underestimated mechanism of β-lactam therapy failure. The extent of the involvement of this mechanism in β-lactam therapy failure in clinical praxis needs to be evaluated in further clinical-microbiological studies. The emerging role of OMVs in antibiotic resistance should be taken into account in strategies directed at combating this serious medical and public health problem.