Abstract
The timely and accurate detection of carbapenemase-producing Enterobacterales (CPE) is imperative to manage this worldwide problem in an effective fashion. Herein we addressed the question of whether the protons produced during imipenem hydrolysis could be detected using an ion sensitive field effect transistor (ISFET). Application of the methodology on enzyme preparations showed that the sensor is able to detect carbapenemases of the NDM, IMP, KPC and NMC-A types at low nanomolar concentrations while VIM and OXA-48 responded at levels above 100 nM. Similar results were obtained when CPE cell suspensions were tested; NDM, IMP, NMC-A and KPC producers caused fast reductions of the output potential. Reduction rates with VIM-type and especially OXA-48 producing strains were significantly lower. Based on results with selected CPEs and carbapenemase-negative enterobacteria, a threshold of 10 mV drop at 30 min was set. Applying this threshold, the method exhibited 100% sensitivity for NDM, IMP and KPC and 77.3% for VIM producers. The OXA-48-positive strains failed to pass the detection threshold. A wide variety of carbapenemase-negative control strains were all classified as negative (100% specificity). In conclusion, an ISFET-based approach may have the potential to be routinely used for non OXA-48-like CPE detection in the clinical laboratory.
Similar content being viewed by others
Introduction
The ongoing spread of multidrug and carbapenemase-producing Enterobacterales (CPE), primarily Klebsiella pneumoniae, has been recognized by WHO and other international institutions as a major public health problem. CPE infect hospitalized patients worldwide and, given the limited therapeutic options, may result in unacceptably high mortality rates especially among debilitated patients treated in ICUs. Carbapenemases encountered in clinical enterobacteria comprise a variety of evolutionary distinct β-lactamase types of unknown origin the most common being the serine reactive KPC variants (class A), the zinc-dependent metallo-β-lactamases (MβLs) of the NDM, VIM, and IMP types (class B), and the serine reactive OXA-48 enzyme (class D). The respective genes are invariably associated with promiscuous genetic structures that facilitate further dissemination1,2,3,4,5,6.
Successful containment of CPEs through implementation of appropriate infection control measures seems, at least currently, as the main way to limit the serious burden posed by these pathogens on health systems. The task of containment falls mainly with the microbiology laboratory that should be able to detect these strains in an accurately and timely fashion. Carbapenemase production is not always associated with increased carbapenem MICs7. Therefore several such strains may evade a laboratory that solely relies in antimicrobial susceptibility testing (AST) for detection of carbapenemase producers. Molecular methods and immunochromatography can provide the required high sensitivity but again would fail to detect the gene or production of a novel carbapenemase8,9,10. Phenotypic techniques based on synergy tests using carbapenem AST discs that contain inhibitory molecules can provide reliable results not only on the production of even novel enzymes but regarding their type as well (e.g. serine reactive or MβL;11). Carbapenemase production may as well be revealed indirectly by carbapenem inactivation assays using a sensitive indicator strain (e.g. modified Hodge test;12). A definite answer on whether a bacterium produces a carbapenemase or not may be provided by monitoring hydrolysis of a carbapenem through UV spectrophotometry using crude protein extracts13. Enzymatic assays though, are not preferred in a clinical laboratory, due to the additional step of protein extraction and the required specialized equipment. Mass spectrometric profiling of the carbapenem hydrolysis products has also being employed utilizing the MALDI-TOF instruments used for bacterial species identification14,15.
Detection of the acidic hydrolysis product (Fig. 1b) of carbapenems by pH indicators is a promising alternative as there is no need for specialized equipment and laborious protein extraction16. The acidimetric determination of β-lactam hydrolysis has been a method introduced more than 60 years ago in enzymology studies of penicillinases17. Nordman et al. developed a modification of the technique that uses imipenem as a substrate and therefore could detect production of an enzyme with carbapenemase activity18. The technique was subsequently commercialized (CARBA NP) while a similar derivative that uses a different pH indicator and modifications that do not require cell disruptive reagents has been developed (Blue-Carba;19). Electrochemical detection of the protons produced during imipenem hydrolysis by CPE has been also employed using a polyaniline coated pH-redox sensor with its performance being comparable to that of the pH indicator techniques (BYE Carba;20,21).
Theoretically, potentiometric detection of the protons produced during imipenem hydrolysis by carbapenemases could also be carried out using an Ion Sensitive Field Effect Transistor (ISFET) sensor. ISFETs were the first chemical sensors to be developed 50 years ago22 and are actually Metal Oxide Semiconductor Field Effect Transistors (MOSFETs) where the metal gate has been substituted with a reference electrode submerged in the aqueous solution interfacing the metal oxide23. When the gate insulator material can buffer specific ions and the drain current is stable, changes in the potential across the capacitor (ΔΨ) would be analogous to the changes of the ions’ concentration23. In pH sensing ISFETs the gate insulator is made of amphoteric materials such as silicon nitride (Si3N4). There, in the solution/transistor interface, the following equilibria are being established24:
With the pH changes in the solution controlling the changes in the potential according to the following equation23:
where R is the universal gas constant, F the Faraday constant, T the temperature and α the sensitivity factor of the gate insulator material (usually < 1) that when approaching unity the response of the transistor equals the sensitivity predicted by the Nernst equation (58.2 mV per pH unit change at 20 °C).
Due to their ease in construction, low cost, reusability and low volume, ISFET sensors have been exploited in a variety of applications not only as sensitive and fast response pH meters but also in more advanced methodologies requiring detection of the protons produced during biochemical reactions23,25. Replacement of the conventional Ag/AgCl reference electrode by coupling the ISFET with another field effect transistor (REFET;26,27) permitted further volume reductions and the development of microchips such as those used in the IonTorrent semiconductor DNA sequencing technology28. Manufacturing of circuits that would amplify the signal to well above the Nernstian response enabled the detection of pH shifts during growth of E. coli at densities as low as 102 cfu/ml24. Herein we explore the applicability of such a sensor to detect imipenem hydrolysis by free carbapenemases and suspensions of CPE cells using a low cost commercially available flow cell ISFET with a conventional reference electrode.
Results
Application of the ISFET method on β-lactamase preparations
The ability of the ISFET method in detecting imipenem hydrolysis was assessed using β-lactamase preparations obtained from recombinant E. coli clones. Initial experiments were performed using protein extracts from the E. coli pZ-NDM-1 clone and from E. coli carrying the empty vector pZE21-MCS. Addition of the NDM-1 containing preparation in the imipenem solution caused a significant drop in the output voltage of the apparatus, in contrast to equal quantities of the preparation obtained from the pZE21-MCS which yielded rates that were similar with the imipenem control (Fig. 1c). In the NDM-1/imipenem reaction mix the voltage changed linearly with time until a fall of approximately 90–100 mV where the rate started to slow reaching a plateau at 120 mV (corresponded to approx. 2.7 pH units’ fall i.e. the pH fell from 7.2 to 4.5) probably due to the lower enzyme activity and stability at the acidic pH. The rate of voltage reduction, estimated from control corrected curves, was analogous to the quantity of NDM-1 added in the reactions (Fig. 2a). Similar patterns were observed for the VIM-1 and IMP-1 MβLs (Fig. 2a) as well as for the KPC-2, NMC-A and OXA-48 serine reactive carbapenemases (Fig. 2b). The CTX-M-15 and CMY-2 β-lactamases, not hydrolyzing imipenem, exhibited a relative voltage reduction near zero as compared to control.
The responses of the sensor to the various types of carbapenemases differed with some enzymes requiring higher quantities in order to observe measurable pH changes (Fig. 2a,b). The efficiency of the ISFET sensor to detect imipenem hydrolysis by the various carbapenemase types was estimated through comparisons of the rate of voltage reduction per one microliter of crude extract (VISFET) with the rate of absorbance units fall at 300 nm (VUV) caused by the same preparation during spectrophotometric monitoring of 80 μM imipenem hydrolysis in the ZnSO4 solution in parallel experiments (Table 1). The VISFET/VUV ratios indicated that from the tested β-lactamases NDM-1 could be detected with higher efficiency followed, in descending order, by NMC-A, IMP-1, KPC-2, VIM-1 and OXA-48 (Table 1). Moreover, comparisons of the lowest concentrations of the enzymes that could be detected with the ISFET (using a threshold of 10 mV fall at 30 min) showed that detection limits were at the sub-nanomolar level for NMC-A and NDM-1, nanomolar for IMP-1 and KPC-2, and sub-micromolar level for VIM-1 and OXA-48 (Table 1; Fig. 2). The ISFET detection efficiency for the various carbapenemases correlated with the respective steady-state catalysis constants (kcat) with enzymes exhibiting rapid imipenem turn-over causing higher pH shifts (Table 1).
Whole cell assays using laboratory E. coli clones
Preliminary experiments showed that the apparatus was able to detect imipenem hydrolysis caused by dense suspensions of intact MβL producing recombinant E. coli clones (e.g. E. coli pZE21-VIM-1; Fig. 3a). The voltage fall observed with the mixture of the bacterial suspensions and imipenem was considerably higher compared to either the bacterial suspension alone or imipenem (Fig. 3a). Yet, in the measurements of bacterial suspensions (Fig. 3; control 2 curves) the pH was not stable and after an initial increase phase it dropped at rates higher than the imipenem measurement (Fig. 3; control 1 curves). This was rather anticipated, as experiments were performed in un-buffered conditions and intact bacterial cells could affect the ion composition of their environment. The above observations suggested that in experiments of whole bacterial cells an additional control measurement of the bacterial suspension alone had to be included and that the bacterium/imipenem curve had to be corrected by subtracting both control curves (Fig. 3a; lower graph). Application of the ISFET method on recombinant clones over-producing non-carbapenemases (e.g. E. coli pBCMY-2; Fig. 3b) revealed that the voltage fall observed with the bacterium/imipenem mixtures was lower compared with that of the bacterium alone with the latter measurements yielding even higher rates of pH changes (probably due to a buffering effect of the imipenem solution), thus suggesting that any minor imipenem hydrolysis by non-carbapenemase producing bacteria could not be detected by the apparatus.
The above results provided strong evidence that the apparatus could differentiate bacteria producing carbapenemases from those that did not. To assert this, additional recombinant E. coli clones overproducing various β-lactamases (i.e. NDM-1, IMP-1, VIM-27, KPC-2, NMC-A, OXA-48, GES-5 and CTX-M-15) as well as, laboratory generated E. coli clones harboring wild type carbapenemase encoding plasmids, were analyzed. E. coli hosts as well as clones carrying empty plasmid vectors were also included in the experiments. The corrected voltage—time curves of bacteria/imipenem reactions showed that the total of MβL producers induced significant reductions of the apparatus readings, while from the strains producing serine reactive carbapenemases only NMC-A, KPC-2 and OXA-48 exhibited a voltage drop (Fig. 3c). The GES-5 producing strain yielded a positive corrected voltage change indicating that the low hydrolysis rates conferred by this enzyme could not be detected by the method (Fig. 3c; medium graph). Similarly to the GES-5 clone, the strains not producing a carbapenemase yielded a voltage fall that was lower compared to that of the controls (Fig. 3c; lower graph).
Among the carbapenemase overproducing clones the pZ-OXA-48 induced the lowest responses averaging a drop of 16 mV at 30 min while the rest exhibited a voltage reduction above 50 mV (Table 2). Expression of VIM-1, KPC-2 and NDM-1 by natural plasmids in E. coli was detectable by the ISFET sensor in all cases (Table 2). Here, the VIM-1 producing clones gave the weakest responses (as low as 22 mV; Table 2). The majority of efficient carbapenemase producers passed the threshold of − 10 mV in less than 5 min with the exceptions of pZ-OXA-48 clone and the VIM-1 expressing E. coli 1R716/trc-TZ116 transconjugant (14 and 18 min respectively; Table 2). The magnitude of voltage changes in carbapenemase producing E. coli laboratory strains did not correlate with the respective carbapenem resistance levels as indicated by imipenem and meropenem MICs (Table 2).
Application on clinical isolates
The ISFET method was subsequently challenged against a collection of carbapenemase positive or negative clinical enterobacterial isolates (Tables 3, 4). Production of an MβL or KPC-2 was detectable by the sensor in all of the K. pneumoniae strains tested (Table 3). In contrast none of the OXA-48 producing strains and the GES-6 isolate induced a voltage reduction within the time course of monitoring. Positive corrected voltage changes were also obtained in the experiments of the carbapenemase negative strains. The strains producing NDM, IMP and KPC type carbapenemases gave the strongest signals while for the VIM-producing isolates a range of responses was observed with some strains yielding a voltage fall near the threshold of 10 mV.
Similar results regarding VIM detection were obtained when analyzing non K. pneumoniae enterobacterial strains (Table 4). In E. coli VIM production was detectable in both isolates tested with one strain inducing a weak (E. coli TZ116) and the other a strong response (E. coli TZ541). As the same difference was observed for the E. coli transconjugants of the above strains (Table 2) it was more likely due to the lower expression of the MβL gene by the TZ116 VIM carrying plasmid. A weak acidification was also evident in the two E. cloacae VIM-1 producers analyzed with one of them not exceeding the threshold of 10 mV at 30 min (Table 4). Detection of this MβL was moreover problematic in the five P. mirabilis isolates tested with only one of them reaching the threshold and two of them exhibiting a positive corrected voltage change. In contrast, VIM-1 production in P. stuartii and S. liquefaciens induced significant pH reduction. As was observed with K. pneumoniae isolates, KPC-2 producing E. coli caused rapid voltage changes in all isolates while the strain expressing the less efficient class A carbapenemase GES-5 did not yield a measurable, by the ISFET sensor, imipenem hydrolysis. In carbapenemase negative E. coli, E. cloacae, E. aerogenes and P. mirabilis isolates a greater than zero corrected voltage change was observed in all but two cases where a minor decrease, below 5 mV, was documented (Table 4).
Based on the above data, a voltage reduction of 10 mV in 30 min in the control corrected curve would indicate production of an efficient carbapenemase (in terms of kcat/Km) by the analyzed strain. Application of this threshold to the dataset of clinical isolates examined herein would identify 42 out of the 53 enterobacterial carbapenemase producers (79.2% sensitivity) while it would exclude all the 30 strains not producing such an enzyme as negative (100% specificity). Regarding specific carbapenemase types the ISFET could detect 100% of the NDM-1, KPC-2 and IMP-1 producers and 17 out of 22 strains expressing solely a carbapenemase of the VIM type (77.3% sensitivity) while failing to identify any of the OXA-48 and GES type carbapenemase strains. The relatively low sensitivity in detecting VIM producers was confined to E. cloacae (with 1 out of 2 strains yielding a false negative result) and P. mirabilis (4/5 false negatives).
In a number of OXA-48 K. pneumoniae measurements it was apparent that the control corrected relative voltage, though positive, had a downward trend indicating that the 30 min time window may not be adequate to reveal imipenem hydrolysis by this enzyme when expressed by natural promoters. In order to assess this, additional measurements were performed where the voltage was monitored for 80 min. In three out of the five OXA-48 strains the − 10 mV detection limit was reached between 45 and 76 min (Table 4). Voltage monitoring for extended periods may also enable the detection of VIM producing P. mirabilis, as indicated by prolonged measurements for one such isolate (Table 4). The fact that from the five VIM producing P. mirabilis the apparatus could detect imipenem hydrolysis only in one isolate (i.e. opposite to detection rates observed for this enzyme in the other enterobacterial species) suggested that there may be expression issues of the enzyme in these species. This was indicated by the fact that the VIM producing E. coli trc-EUG91 clone obtained through conjugation with P .mirabilis EUG91 exhibited stronger ISFET responses compared to the donor. In order to assess if this was indeed caused by lower VIM production and not due to other species specific factors (e.g. cell wall endurance), protein extracts of P. mirabilis EUG91 and its E. coli transconjugant were assayed for imipenem hydrolysis using spectrophotometry. The measured imipenem hydrolysis units (33 nmol∙min−1∙mg−1 for P. mirabilis versus 136 nmol∙min−1∙mg−1 for E. coli) indicated that the enzyme was produced at lower levels in the periplasm of P. mirabilis thus causing difficult to resolve detection issues.
Additional indications that the ISFET false negative results for OXA-48 and VIM producers were due to low enzyme levels in the measurement solution were obtained through comparisons with the commercial colorimetric method RAPIDEC CARBA NP. The latter test uses denser bacterial suspensions than those used in ISFET (due to lower reaction volume) with the measurement being preceded by a 30 min bacterial cell lysis step. CARBA NP enabled the detection of three out of five OXA-48 producers while it failed to identify two isolates with results being in accordance with prolonged ISFET measurements (Table 4). Regarding VIM producing P. mirabilis CARBA NP identified all the three isolates tested (Table 4). With the exception of one P .mirabilis strain that yielded a yellow color, all the ISFET false negative strains tested with CARBA NP did not cause significant pH shifts as revealed by the orange color of the indicator (Table 4). Though the colorimetric method was more sensitive as compared to the ISFET it suffered from specificity issues as two out of the four carbapenemase negative Enterobacterales strains tested, which were overproducing CTX-M-15 and CMY-2 type beta-lactamases, yielded a positive result (Table 4). In contrast, the ISFET method, even in prolonged measurements, did not yield any false positive result (Table 4).
Discussion
Herein we assessed the ability of a commercially available ion sensitive field effect transistor sensor to detect the protons produced when imipenem is hydrolyzed by clinically relevant carbapenemases. Application of the method on β-lactamase preparations and whole cells of various enterobacterial species indicated that the ISFET could detect most of the clinically relevant carbapenemases with high specificity and relatively high sensitivity. Enzymes exhibiting low turn-over rates during imipenem hydrolysis (e.g. OXA-48 and VIM-1) required higher quantities in order to be detected under the experimental conditions tested (i.e. at an imipenem concentration well above the respective Kms) compared to the faster carbapenemases (NDM-1, NMC-A, IMP-1 and KPC-2). This difference was reflected in the responses observed during the assays of laboratory and clinical enterobacteria where the latter enzymes gave output potential reductions at 30 min above 50 mV while VIM type producers exhibited on average lower signals with high variation (− 29 ± 31 mV; Fig. 4). In strains producing OXA-48, GES type carbapenemases or non-carbapenemases a voltage fall that was lower than that of the combined controls was evident (Fig. 4). The high variation observed in the ISFET responses when VIM producers were assayed was most probably due to different levels of the enzyme in the periplasm of the respective strains. In species such as E. cloacae and P. mirabilis it seems that VIM production is too low to be detected in 100% of the cases, probably due to low copy number of the gene (likely caused by plasmid stability issues or chromosome incorporation) or inefficient localization in the periplasm.
With the experimental setup used in this study sensitivity is considerably lower for VIM and OXA-48 carbapenemases compared to the pH indicator techniques. Yet, the results obtained permitted a quantitative assessment of the phenomena taking place during application of the latter methodologies while they could serve as a proof of concept for an investment in the development of more sensitive carbapenemase detecting ISFETs. Sensitivity could be improved either by introducing modifications that would increase the concentration of the enzymes in solution (e.g. reaction volume reduction or use of cell wall disruption agents with no buffering effects) or through signal amplification. A considerable volume reduction even below 10 μl could be achieved through the use of a REFET instead of a conventional reference electrode while signal amplification may be enabled either electronically or by coupling the ISFET with an actuator electrode of a coulometric sensor that would permit titration of the acidic hydrolysis product23,29. The successful implementation and commercialization of the ISFET technology in microchips used for the detection of the protons produced during DNA polymerization reactions (e.g. IonTorrent, DNA Electronics-Genalysis;28,30) suggest that similar setups may be used for the massive, parallel and sensitive detection of imipenemase activity in bacteria producing even unknown carbapenemases directly in clinical samples.
Materials and methods
Laboratory strains and bla-carrying plasmids
Natural plasmids and recombinant vectors expressing various β-lactamases were hosted by the following E. coli laboratory strains: E. coli DH5α [F–, endA1, glnV44, thi-1, recA1, relA1 gyrA96, deoR, nupG, purB20, φ80dlacZΔM15, Δ(lacZYA-argF)U169, hsdR17(rK–mK+), λ–), E. coli C600Z1 (spr, laciq, PN25-tetR, lacY1, leuB6, mcrB + , supE44, thi-1, thr-1, tonA21), E. coli JM109 [endA1, glnV44, thi-1, relA1 gyrA96, recA1, mcrB+, Δ(lac-proAB), e14-, [F' traD36 proAB+ lacIq lacZΔM15], hsdR17(rK-mK+)] and E. coli 1R716 (lac-, rifr). High level expression of NMC-A, VIM-27, CMY-2, CTX-M-15 and, GES-5 was achieved using the recombinant plasmids pNTN3-blaNMC-A (E. coli JM109), pBC-blaVIM-27, pBC-blaKPC-2, pBC-blaCMY-2, pBC-blaCTX-M-15 and, pBC-blaGES-5 (E. coli DH5α)31,32,33,34,35,36. In the latter clones, the bla genes were expressed under the control of the respective natural promoters. The open reading frames of VIM-1, NDM-1, IMP-1 and, OXA-48 were cloned into the BamHI and KpnI sites of the pZE21-MCS vector and the resulting plasmids (pZ-bla) were used to transform the TetR producing E. coli C600Z1 strain37. Integrity of the cloned regions and of the expression module was confirmed by Sanger sequencing using the primers pZE21-Fs (5′-TGGCAATTCCGACGTCTAAG-3′) and pZE21-Rs (5′-GTCTAGGGCGGCGGATTTG-3′). In these clones expression of β-lactamases was controlled by the pTetOL1 promoter/operator and was inducible by tetracycline analogues38.
Clinical isolates
A total of 83 non-repetitive enterobacterial clinical strains isolated in Greek hospitals were analyzed. The collection comprised of 35 Klebsiella pneumoniae isolates producing at least one carbapenemase type (10 VIM, 5 NDM, 1 IMP, 7 KPC, 5 OXA-48, 1 GES-6, 3 co-producing VIM and KPC and 3 co-producing VIM and NDM β-lactamases) and 12 isolates producing various types of serine reactive β-lactamases not hydrolyzing carbapenems. Additionally 18 E. coli isolates were assayed from which eight were producing a carbapenemase (two VIM, five KPC and one GES-5) and 10 expressing for β-lactamases with no such functionality. The collection included also 10 VIM producing enterobacterial isolates belonging to Enterobacter cloacae (n = 2), Proteus mirabilis (n = 5), Providencia stuartii (n = 2) and Serratia liquefaciens (n = 1) as well as carbapenemase negative E. cloacae (n = 3), Klebsiella aerogenes (n = 4) and P. mirabilis (n = 1). Species identification, antimicrobial susceptibility and β-lactamase content characterization were performed as described previously7,39. The genetic diversity of the isolates expressing the same β-lactamases has been previously asserted using Multi Locus Sequence Typing (MLST) and Restriction Fragment Length Polymorphism/Pulse Field Gel Electrophoresis (RFLP-PFGE).
Carbapenem MIC measurements
Imipenem and meropenem MICs were determined using the broth micro-dilution in Mueller–Hinton in accordance with EUCAST guidelines. During the testing of the strains carrying the pZE21 recombinant plasmids, anhydrotetracycline was included in the broth at a concentration of 200 ng/ml, for induction of the cloned β-lactamases.
Preparation of crude protein extracts
β-Lactamase preparations from the recombinant E. coli clones as well as from the P. mirabilis EUG91 strain and its transformant E. coli trfEUG91 were obtained after ultra-sonication of cells grown in Luria–Bertani (LB) broth as described previously39. Cultures of the pZE21 strains were allowed to reach the late logarithmic phase (OD600nm ≈ 0.5) and then β-lactamase production was induced for 18 h by addition of 200 ng/ml anhydrotetracycline. MβL producers were grown in the presence of 300 μM ZnSO4 and β-lactamases were released in 0.05 Μ HEPES buffer pH 7. Sonication of the remaining clones was performed in 0.1 M potassium phosphate buffer pH 7. Total protein content of the extracts was determined by the Bradford method. Imipenem hydrolysis was monitored using UV spectrophotometry at 300 nm at a substrate concentration of 80 μM and pH 7. Carbapenemase concentrations in crude extracts of the recombinant clones were estimated using the measured initial velocities and the published steady state kinetic constants (kcat and Km) for imipenem hydrolysis through the Michaelis–Menten equation39,40,41,42,43,44,45,46.
Equipment and application of the ISFET method
The ISFET sensor used in the study is commercially available from WINSENSE Ltd., Thailand (Fig. 1a) and consisted of three terminals (source, drain and a reference electrode substituting the metal gate). Silicon nitride (Si3N4) made up the gate insulator and the reference electrode was of the Ag/AgCl type (Fig. 1b). A circuit provided by WINSENSE Ltd. was used to keep a stable drain current (Ids) for the ISFET operation (Fig. 1b). Output voltage was monitored using a VA18B digital multimeter (Mastech, China) connected to a PC. The ISFET was placed into the bottom of a custom made flow cell provided by Winsense Ltd. with the reference electrode adapted in the top of the compartment that had a volume capacity of approximately 0.5 ml (Fig. 1a). Solutions were applied to the apparatus using a 5 ml syringe through a T tubing connector (Bio-Rad, USA). The output voltage was linearly depended on the pH of the solution (one pH unit fall corresponded to approximately 45 mV of voltage reduction). Performance of the device was routinely assessed using standard buffers of pH values 4, 7 and 10 (ThermoFisher Scientific, USA).
The ability of the sensor to detect carbapenem hydrolysis (Fig. 1b) was initially assessed using β-lactamase preparations obtained from the recombinant E. coli clones. For this, 10 mg/ml imipenem stock solutions of the commercially available imipenem/cilastatin formulation (50 mg imipenem/50 mg cilastatin) were prepared in water, 10 mM Tris/HCl pH 7 or 0.3 mM ZnSO4 with pH adjusted at 7.2 using drops of 10 N NaOH. Each experiment was performed by diluting a quantity of the enzyme in 0.5 ml of solution which was then mixed with 1 ml of the imipenem stock yielding a final substrate concentration of 6.67 mg/ml. The total 1.5 ml of the reaction was aspired with a 5 ml syringe and transferred to the ISFET flow cell. Voltage was monitored continuously for 20–30 min and the resulting curve was compared with control measurements of imipenem. The flow cell was washed with 20 ml of deionized water between each measurement. Preparations containing VIM-1, NDM-1 and IMP-1 metallo-β-lactamases, KPC-2, NMC-A and OXA-48 serine reactive carbapenemases as well as CMY-2 and CTX-M-15 β-lactamases-not exhibiting meaningful imipenem hydrolysis-were assayed. Significant voltage reduction was observed only for the efficient enzymes indicating that the method is able to specifically detect the presence of a carbapenemase in solution. The ZnSO4 solution yielded the best results and it was used in subsequent experiments. In order to estimate the lowest detection limits for each carbapenemase various quantities of protein extracts were assayed and the rate of voltage reduction was measured for the stable phase of the control corrected 30 min curves. A voltage reduction of 10 mV in 30 min (rate ≈ − 0.33 mV∙min−1) was set as the hydrolysis detection threshold (discussed in the last paragraph of the section). Detection limits were determined through mV∙min−1 versus [E] plots and were expressed as the enzyme concentration contained in the reaction in nM resulting in a voltage reduction rate of − 0.33 mV∙min−1. Imipenem hydrolysis rates by the carbapenemase preparations were also measured through UV spectrophotometry at 300 nm in parallel experiments using 80 μΜ substrate in the ISFET measurement solution.
Whole cell assays were performed using the same setup as above. Bacteria were grown in Tryptic Soy Agar (TSA) containing 300 μM ZnSO4 at 37 °C for 16 h. For the pZE21 clones 200 ng/ml of anhydrotetracycline was added in the medium for induction of the cloned β-lactamase genes. Heavy bacterial suspensions were prepared in 0.5 ml of the ZnSO4 solution (5 full 10 μl loops for all but the pZ-NDM-1 and pE-NMC-A clones for which suspensions having a density of 1 and 2 in the MacFarland scale were used due to rapid imipenem hydrolysis when using denser preparations). 1 ml of the imipenem stock solution was added and the mixture was thoroughly vortexed and transferred into the ISFET apparatus using a 5 ml syringe. The output voltage was monitored for 30 min. Control measurements of bacterial suspensions without imipenem and of the imipenem solution were used to correct the bacterium/imipenem measurement through subtraction of normalized curves according to the following equations:
where ΔVrel(t) is the relative voltage change at each time point that was obtained by subtracting the maximum voltage reading observed during the course of each measurement (Vmax) from the voltage reading at each time point (V(t)) and ΔVcor(t)bacterium/imipenem is the corrected relative voltage change of the bacterium/imipenem mix.
Between each measurement the flow cell was washed with 50 ml deionized water, 20 ml of 70% v/v ethanol and again with 50 ml of deionized water and finally it was equilibrated with the ZnSO4 solution.
After experiments, the ISFET was washed as above and dried while the reference electrode was stored in a saturated KCl solution (Orion electrode storage solution, ThermoFisher Scientific).
The corrected voltage drop at 30 min that would yield the optimal results in terms of sensitivity and specificity when the method is applied in whole cell assays was estimated using the Receiver Operating Characteristic (ROC) curve analysis20 in Prism 6 software (GraphPad, USA). The signals obtained from all measurements of carbapenemase negative strains were grouped in the “Control” column while those of carbapenemase producers in the “Patient” column and the data were analyzed using the software’s default parameters. A signal of < − 9.5 mV at 30 min yielded a sensitivity of 80.82% [95% Confidence Interval (CI): 69.92–89.10%] and 100% specificity (95% CI 91.96–100%). Hence the detection threshold was set at a voltage drop of − 10 mV at 30 min.
Comparisons with a colorimetric technique
Clinical strains which yielded false negative results with the ISFET method were also assayed with the commercial pH indicator colorimetric technique RAPIDEC CARBA NP (BioMérieux, France). Strains found true positive and true negative, regarding carbapenemase production, with the ISFET method were also tested for control purposes. Bacteria were grown on TSA containing 300 μM ZnSO4 at 37 °C for 16 h and the test was performed and interpreted according to the manufacturer’s instructions.
References
Nordmann, P., Naas, T. & Poirel, L. Global spread of carbapenemase-producing enterobacteriaceae. Emerg. Infect. Dis. 17, 1791–1798 (2011).
Kostyanev, T. et al. Detection and prevalence of carbapenem-resistant Gram-negative bacteria among European laboratories in the COMBACTE network: A COMBACTE LAB-Net survey. Int. J. Antimicrob. Agents 53, 268–274 (2019).
Martinez-Martinez, L. & Gonzalez-Lopez, J. J. Carbapenemases in Enterobacteriaceae: Types and molecular epidemiology. Enferm. Infecc. Microbiol. Clin. 32(Suppl 4), 4–9 (2014).
Tzouvelekis, L. S., Markogiannakis, A., Psichogiou, M., Tassios, P. T. & Daikos, G. L. Carbapenemases in Klebsiella pneumoniae and other Enterobacteriaceae: An evolving crisis of global dimensions. Clin. Microbiol. Rev. 25, 682-+ (2012).
Queenan, A.M., Bush, K. Carbapenemases: The versatile beta-lactamases. Clin. Microbiol. Rev. 20, 440–458 (2007).
Bush, K. & Bradford, P. A. Interplay between beta-lactamases and new beta-lactamase inhibitors. Nat. Rev. Microbiol. 17, 295–306 (2019).
Kotsakis, S. D. et al. Laboratory evaluation of Brilliance CRE Agar for screening carbapenem-resistant Enterobacteriaceae: Performance on a collection of characterised clinical isolates from Greece. J. Glob. Antimicrob. Resist. 1, 85–90 (2013).
Bilozor, A. et al. Application of molecular methods for carbapenemase detection. Front. Microbiol. 10, 1755 (2019).
Glupczynski, Y. et al. Evaluation of two new commercial immunochromatographic assays for the rapid detection of OXA-48 and KPC carbapenemases from cultured bacteria. J. Antimicrob. Chemother 71, 1217–1222 (2016).
Aguirre-Quinonero, A. & Martinez-Martinez, L. Non-molecular detection of carbapenemases in Enterobacteriaceae clinical isolates. J. Infect. Chemother 23, 1–11 (2017).
Miriagou, V. et al. Combined disc methods for the detection of KPC- and/or VIM-positive Klebsiella pneumoniae: Improving reliability for the double carbapenemase producers. Clin. Microbiol. Infect. 19, E412-415 (2013).
Girlich, D., Poirel, L. & Nordmann, P. Value of the modified Hodge test for detection of emerging carbapenemases in Enterobacteriaceae. J. Clin. Microbiol. 50, 477–479 (2012).
Bernabeu, S., Poirel, L. & Nordmann, P. Spectrophotometry-based detection of carbapenemase producers among Enterobacteriaceae. Diagn. Microbiol. Infect. Dis. 74, 88–90 (2012).
Oviano, M., Bou, G. Matrix-assisted laser desorption ionization-time of flight mass spectrometry for the rapid detection of antimicrobial resistance mechanisms and beyond. Clin. Microbiol. Rev. 32, e00037–18 (2019).
Papagiannitsis, C. C. et al. Matrix-assisted laser desorption ionization-time of flight mass spectrometry meropenem hydrolysis assay with NH4HCO3, a reliable tool for direct detection of carbapenemase activity. J. Clin. Microbiol. 53, 1731–1735 (2015).
Noel, A. et al. Comparative Evaluation of Four Phenotypic Tests for Detection of Carbapenemase-Producing Gram-Negative Bacteria. J. Clin. Microbiol. 55, 510–518 (2017).
Saz, A. K., Lowery, D. L. & Jackson, L. J. Staphylococcal penicillinase. I. Inhibition and stimulation of activity. J. Bacteriol. 82, 298–304 (1961).
Nordmann, P., Poirel, L. & Dortet, L. Rapid detection of carbapenemase-producing Enterobacteriaceae. Emerg. Infect. Dis. 18, 1503–1507 (2012).
Pires, J., Novais, A. & Peixe, L. Blue-carba, an easy biochemical test for detection of diverse carbapenemase producers directly from bacterial cultures. J. Clin. Microbiol. 51, 4281–4283 (2013).
Bogaerts, P. et al. Multicentre evaluation of the BYG Carba v2.0 test, a simplified electrochemical assay for the rapid laboratory detection of carbapenemase-producing Enterobacteriaceae. Sci. Rep 7, 9937 (2017).
Bogaerts, P., Yunus, S., Massart, M., Huang, T. D. & Glupczynski, Y. Evaluation of the BYG carba test, a new electrochemical assay for rapid laboratory detection of carbapenemase-producing Enterobacteriaceae. J. Clin. Microbiol. 54, 349–358 (2016).
Bergveld, P. Development of an ion-sensitive solid-state device for neurophysiological measurements. IEEE Trans. Biomed. Eng Bm17, 70- (1970).
Bergveld, P. Thirty years of ISFETOLOGY—What happened in the past 30 years and what may happen in the next 30 years. Sens. Actuators B Chem. 88, 1–20 (2003).
Jiang, Y. et al. A high-sensitivity potentiometric 65-nm CMOS ISFET sensor for rapid E. coli screening. IEEE Trans. Biomed. Circuits Syst. 12, 402–415 (2018).
Moser, N., Keeble, L., Rodriguez-Manzano, J., Georgiou, P. ISFET arrays for lab-on-chip technology: A review. https://doi.org/10.1109/icecs46596.2019.8965034:57-60 (2019)
Matsuo, T. & Nakajima, H. Characteristics of reference electrodes using a polymer gate ISFET. Sens. Actuators 5, 293–305 (1984).
Chudy, M., Wróblewski, W. & Brzózka, Z. Towards REFET. Sens. Actuators B Chem. 57, 47–50 (1999).
Rothberg, J. M. et al. An integrated semiconductor device enabling non-optical genome sequencing. Nature 475, 348–352 (2011).
Vanderschoot, B. & Bergveld, P. An isfet-based microliter titrator—integration of a chemical sensor actuator system. Sens. Actuators 8, 11–22 (1985).
Toumazou, C. et al. Simultaneous DNA amplification and detection using a pH-sensing semiconductor system. Nat. Methods 10, 641-646 (2013).
Papagiannitsis, C. C. et al. Characterization of metallo-beta-lactamase VIM-27, an A57S mutant of VIM-1 associated with Klebsiella pneumoniae ST147. Antimicrob. Agents Chemother 55, 3570–3572 (2011).
Nordmann, P., Mariotte, S., Naas, T., Labia, R. & Nicolas, M. H. Biochemical properties of a carbapenem-hydrolyzing beta-lactamase from Enterobacter cloacae and cloning of the gene into Escherichia coli. Antimicrob. Agents Chemother 37, 939–946 (1993).
Miriagou, V. et al. Imipenem resistance in a salmonella clinical strain due to plasmid-mediated Class A carbapenemase KPC-2. Antimicrob. Agents Chemother. 47, 1297–1300 (2003).
Kotsakis, S. D., Papagiannitsis, C. C., Tzelepi, E., Tzouvelekis, L. S. & Miriagou, V. Extended-spectrum properties of CMY-30, a Val211Gly mutant of CMY-2 cephalosporinase. Antimicrob. Agents Chemother 53, 3520–3523 (2009).
Hachler, H. et al. Characterisation of CTX-M-117, a Pro174Gln variant of CTX-M-15 extended-spectrum beta-lactamase, from a bovine Escherichia coli isolate. Int. J. Antimicrob. Agents 41, 94–95 (2013).
Kotsakis, S. D., Miriagou, V., Tzelepi, E. & Tzouvelekis, L. S. Comparative biochemical and computational study of the role of naturally occurring mutations at Ambler positions 104 and 170 in GES beta-lactamases. Antimicrob. Agents Chemother 54, 4864–4871 (2010).
Frohlich, C. et al. Structural and biochemical characterization of the environmental MBLs MYO-1, ECV-1 and SHD-1. J. Antimicrob. Chemother 75, 2554–2563 (2020).
Lutz, R. & Bujard, H. Independent and tight regulation of transcriptional units in Escherichia coli via the LacR/O, the TetR/O and AraC/I1-I2 regulatory elements. Nucleic Acids Res. 25, 1203–1210 (1997).
Voulgari, E. et al. Evaluation of the performance of Acuitas(R) Resistome Test and the Acuitas Lighthouse(R) software for the detection of beta-lactamase-producing microorganisms. J. Glob. Antimicrob. Resist. 22, 184–189 (2020).
Marathe, N. P. et al. Functional metagenomics reveals a novel carbapenem-hydrolyzing mobile beta-lactamase from Indian river sediments contaminated with antibiotic production waste. Environ. Int. 112, 279–286 (2018).
Franceschini, N. et al. Purification and biochemical characterization of the VIM-1 metallo-beta-lactamase. Antimicrob. Agents Chemother 44, 3003–3007 (2000).
Marcoccia, F. et al. Kinetic study of laboratory mutants of NDM-1 metallo-beta-lactamase and the importance of an isoleucine at position 35. Antimicrob. Agents Chemother 60, 2366–2372 (2016).
Laraki, N. et al. Biochemical characterization of the Pseudomonas aeruginosa 101/1477 metallo-beta-lactamase IMP-1 produced by Escherichia coli. Antimicrob. Agents Chemother 43, 902–906 (1999).
Yigit, H. et al. Carbapenem-resistant strain of Klebsiella oxytoca harboring carbapenem-hydrolyzing beta-lactamase KPC-2. Antimicrob. Agents Chemother 47, 3881–3889 (2003).
Mariotte-Boyer, S., Nicolas-Chanoine, M. H. & Labia, R. A kinetic study of NMC-A beta-lactamase, an Ambler class A carbapenemase also hydrolyzing cephamycins. FEMS Microbiol. Lett 143, 29–33 (1996).
Docquier, J. D. et al. Crystal structure of the OXA-48 beta-lactamase reveals mechanistic diversity among class D carbapenemases. Chem. Biol. 16, 540–547 (2009).
Acknowledgements
This work was financially supported by the Hellenic Pasteur Institute. George Miliotis was supported by a doctoral fellowship in the frame of the CycloN-Hit program (FP7-PEOPLE-ITN-2013, Project No 608407). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. We would like to thank Prof. Joakim Larsson and Assoc. Prof. Carl-Fedrick Flach (Centre for Antibiotic Resistance Research-CARe, University of Gothenburg) for their kind donation of the pZE21-MβL recombinant plasmids, the pZE21-MCS vector and the C600Z1 E. coli host.
Author information
Authors and Affiliations
Contributions
SDK conceptualized the work, developed the methodology, performed the experiments, supervised the work and wrote the initial draft of the manuscript. GM performed the experiments. ET characterized the clinical strains. LST analyzed the data. VM characterized the clinical strains and supervised the work. All authors reviewed and edited the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Kotsakis, S.D., Miliotis, G., Tzelepi, E. et al. Detection of carbapenemase producing enterobacteria using an ion sensitive field effect transistor sensor. Sci Rep 11, 12061 (2021). https://doi.org/10.1038/s41598-021-91202-6
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-021-91202-6
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.