Penicillin G, Piperacillin, Ceftriaxone, Cefepime, Cefazolin, Meropenem were purchased from Solarbio Co., Ltd. (Beijing, China). Imipenem was bought from Fresenius Kabi (Bad Homburg, Germany). HEPES was obtained from Sangon Biotech Co., Ltd. (Shanghai, China). All other reagents were purchased from Sigma-Aldrich Co (St. Louis, USA). The controlled pore glass (CPG) beads (125–140 μm particle diameter and 50 nm pore size) were originally purchased from VEB Trisola, Steinach, Germany. The construction of plasmid encoding NDM-1, expression and purification of recombinant NDM-1(rNDM-1) were performed as previously described (Meng et al., 2021a; Meng et al., 2021b). The purity of the purified rNDM-1 is > 95%, The unit activity of rNDM-1 to hydrolyze meropenem is 108 IU/mg.

Phosphate buffer (PB) and HEPES buffer in different ionic strengths (0.01, 0.05 0.1 M) and pHs (pH 6.8, 7.2, 7.6, 8.0) were prepared as running buffers and for sample preparation. All water used for preparing the buffers and samples was from a Milli-Q system (Millipore, Bedford, MA, USA).

Stock solutions of each of the antibiotics: Penicillin G, Piperacillin, Ceftriaxone, Cefepime, Cefazolin, Meropenem, and Imipenem, were prepared in PB and HEPES at the various ionic strengths and pHs, which were then serially diluted: 10 mM, 5 mM, 2.5 mM, 1.25 mM, and 0.625 mM.

The Enzyme Thermal biosensor (ET) is a flow-injected analytical device (Omik Bioscience AB, Sweden) has been employed in this study. The enzyme, e.g. β-lactamase, was immobilized on spherical porous glass beads (CPG), loaded in an enzyme column and substrate samples were injected (e.g. antibiotics) with running buffers at various concentrations. Meanwhile the simultaneously generated heat can be precisely and continuously detected with high resolution thermistor sensors (Ramanathan and Danielsson, 2001; Zhou et al., 2013). The thermistor sensor converts the heat generated by the reaction into a temperature signal (Xie et al., 2000; Qie et al., 2013). In an adiabatic system, if all the heat released is used to increase the temperature of the reaction system, the following formula can be used:

Where n is the number of moles of product; ΔH is the enthalpy change of the biochemical reaction; ΔT is the temperature change of the reaction system; C_S is the heat capacity of the reaction system. This formula shows that there is a linear relationship between the temperature change and the substrate concentration. Thus, the peak height or area can be used for determination of an analyte concentration (Xie et al., 2000; Yakovleva et al., 2013). Thermometric biosensors have numerous advantages including: the universality of the sensing mechanism, label-free analysis and high specificity. In addition, thermal sensors are less sensitive to sample interference than optical and electrical biosensors. The technology has been widely used to detect sample concentration and enzyme activity in clinical analysis, fermentation process control, food safety and environmental monitoring (Xie et al., 1994; Yakovleva et al., 2012; Chen et al., 2013; Yao et al., 2015; Andersson et al., 2017; Xu et al., 2017; Adlerberth et al., 2020).

Here, we establish a novel method for fast identification and classification of AMR β-lactamases by thermometric biosensor profiling of the enzyme catalytic activity for a panel of antibiotics ( Figure 1 ). The device is simple to operate and automatically monitors the degradation reactions of various antibiotics. Considering the clinical interesting, 7 mostly used antibiotics including penicillin G, piperacillin, cefepime, ceftriaxone, cefazolin, meropenem, and imipenem were selected for the NDM-1 and penicillinase profiling studies.

The Enzyme Thermal biosensor used in this study was composed of a pump, a sample valve, split two enzyme columns (one as measurement and the other as reference) and high resolution thermistors in coupled with signal amplification.

The enzyme column was loaded with ca. 50 IU immobilized β-lactamase (e.g. penicillinase or NDM-1) onto CPG. The working temperature was set at 25°C. A flow rate of 0.6 mL min^-1 and sample volumes of 30 and 75 μL were used. However, for the comparative study, the response signals were adjusted to 10 μL/analyte. Each sample measurement was performed in triplicate.

We first evaluated effects of different buffers on degradation of 7 β-lactam antibiotics: penicillin G, piperacillin, ceftriaxone, cefepime, cefazolin, meropenem and imipenem. Among these 7 antibiotics, penicillin G and piperacillin belong to penicillin class; ceftriaxone, cefepime, cefazolin belong to cephalosporin class; and meropenem and imipenem belong to carbapenem class. Cefazolin, ceftriaxone and cefepime are representative of the 1^st, 3^rd and 4^th generation cephalosporins, respectively.

Studies were performed to determine the effect that buffer had on assay sensitivity ( Figures 2 , 3 ). In this study, thermometric biosensor was employed for instant detection of the catalytic enthalpy (e.g. NDM-1 or penicillinase catalysis of antibiotics) that was converted to a corresponding temperature changes. The intensity of the thermal signal was proportional to the concentration of the substrates, e.g. antibiotics. In this study, a panel of 7 different antibiotics at concentrations of 0.625, 1.25, 2.5, 5.0 and 10 mM were determined sequentially. Therefore, a complete profile for one AMR enzyme could be generated in 35 min based on a cycle time of 5 min.

As penicillinase can only efficiently catalyze penicillin class, only penicillin G and piperacillin response curves were presented in buffers of PB ( Figures 2A, B ) and HEPES ( Figures 2C, D ) at a concentration of 10, 50 and 100 mM. However, the 3D profiles of the catalytic activities for the 7 antibiotics were generated using a concentration of 10 mM using PB and HEPES buffers ( Figures 2E, F ).

The two buffer systems presented similar sensitivities for the individual antibiotics. However, when comparing the three buffer concentrations, the sensitivity was the highest at 100 mM and lowest at 10 mM. Meanwhile the highest variation in sensitivity was seen at 10 mM buffer concentration for both PB and HEPES ( Figure 2 ). The lowest sensitivity variation was observed for Penicillin G in PB, which means penicillinase catalyzing penicillin G was relatively stable in PB buffer.

In order to verify the catalytic efficiency of NDM-1 enzyme, three classes of antibiotics including: Penicillins (penicillin G, piperacillin), cephalosporins (cefepime, ceftriaxone, cefazolin) and Carbapenems (meropenem, imipenem), were investigated using PB ( Figure 4 ) and HEPES ( Figure 3 ) buffers at concentrations of 10, 50 and 100 mM.

In contrast to Penicillinase in Figure 2 , the highest sensitivity was obtained with 10 mM buffer and the lowest with 100 mM buffer for both PB ( Figure 4 ) and HEPES ( Figure 3 ) for all 7 antibiotics with the exception of penicillin G in HEPES buffer ( Figure 3C ). In addition, effect of the buffer concentration on profile variation was the highest in PB ( Figures 4E-G ), but the lowest in HEPES buffer ( Figures 3E-G ) for the Cephalosprins antibiotic class (cefepime, ceftriaxone, cefazolin). This finding could be possibly used for differentiating cephalosprins from other classes of antibiotics.

Figure 5 showing the signal response comparison between PB and HEPES buffers was similar to those seen in Figure 3 and 4 , i.e. that the highest sensitivity was obtained at 10 mM buffer for almost all antibiotic reactions. Cefazolin showed the lowest response indicating that it generated the least catalytic enthalpy as compared with the other antibiotics. In general, higher sensitivity was observed in PB than HEPES buffer.

The pH effects on the catalytic reactions were evaluated in PB and HEPES buffers at pHs 6.8, 7.2, 7.6 and 8.0 ( Figure 6 ). Using the HEPES buffer, an apparent pH dependent effect on catalytic activity of NDM-1 was observed in both carbapenem antibiotics ( Figures 6A, B ). In addition, an apparent pH dependent effect on catalytic activity of NDM-1 was observed in both carbapenem and penicillin antibiotics using PB buffer ( Figures 6A-D ). In addition, the results show that higher sensitivity was observed using PB than HEPES, and the sensitivity variation was also higher for PB than HEPES. Meanwhile, the cefazolin showed the smallest response in HEPES buffer and was least effected by pH. Overall, the differential effect of pH on the catalytic activity of NDM-1 and penicillinase might facilitate differentiation of these two β-lactamases.