Graphene Quantum Dots as Nanozymes for Electrochemical Sensing of Yersinia enterocolitica in Milk and Human Serum

The genus Yersinia contains three well-recognized human pathogens, including Y. enterocolitica, Y. pestis, and Y. pseudotuberculosis. Various domesticated and wild animals carry Yersinia in their intestines. Spread to individuals arises from eating food or water contaminated by infected human or animal faeces. Interaction with infected pets and domestic stock may also lead to infection. Yersinia is able to multiply at temperatures found in normal refrigerators; hence, a large number of the bacteria may be present if meat is kept without freezing. Yersinia is also rarely transmitted by blood transfusion, because it is able to multiply in stored blood products. Infection with Yersinia can cause yersiniosis, a serious bacterial infection associated with fever, abdominal pain and cramps, diarrhea, joint pain, and symptoms similar to appendicitis in older children and adults. This paper describes a novel immunosensor approach using graphene quantum dots (GQDs) as enzyme mimics in an electrochemical sensor set up to provide an efficient diagnostic method for Y. enterecolitica. The optimum assay conditions were initially determined and the developed immunosensor was subsequently used for the detection of the bacterium in milk and human serum. The GQD-immunosensor enabled the quantification of Y. enterocolitica in a wide concentration range with a high sensitivity (LODmilk = 5 cfu mL−1 and LODserum = 30 cfu mL−1) and specificity. The developed method can be used for any pathogenic bacteria detection for clinical and food samples without pre-sample treatment. Offering a very rapid, specific and sensitive detection with a label-free system, the GQD-based immunosensor can be coupled with many electrochemical biosensors.


Introduction
Yersinia enterocolitica is a gram-negative bacillus shaped bacterium that leads to a zootonic disease called yersiniosis. The infection is demonstrated as mesenteric adenitis, acute diarrhea, terminal ileitis, and pseudoappendicitis. Rarely, it can even result in sepsis [1]. According to the 2017 report of the European Food Safety Authority (EFSA) and European Centre for Disease Prevention and Control (ECDC), Y. enterocolitica has been realized as the third most common foodborne-zoonotic disease after campylobacteriosis and salmonellosis in the European Union [2].
Nutrients are stored in refrigerators for a long time at 4 • C and Y. enterocolitica has the ability to quickly and easily grow and proliferate at this temperature. This situation increases the risk of infection via contaminated food and dairy products [1]. The bacterium is transmitted to humans by contaminated water and food. Raw and pasteurized milk, as well as various vegetables, may be increasing signals at mA range based on the increasing GQD concentration used. The enhanced electrocatalytic property of the GQD-modified electrode could be attributed to the intimate electronic interactions between Au and GQDs, which improves electron transfer. The detection of bacteria was based on the degree of inhibited electron transfer of the laminated GQDs on the Au surface, which was hampered by the formation of antigen-antibody complex on the GQD-modified Au electrode. Thus, the signal response of the immunosensor decreased along with the increased Y. enterecolitica concentration in the sample solution. To best of our knowledge, we could achieve the highest sensitivity and specificity for Y. enterocolitica by using the label-free, real time GQDs-based electrochemical sensor. This may help in the identification of possible infectious diseases and any pathogenic bacteria which exist in food products.
Materials 2018, 11, x FOR PEER REVIEW 3 of 15 electronic interactions between Au and GQDs, which improves electron transfer. The detection of bacteria was based on the degree of inhibited electron transfer of the laminated GQDs on the Au surface, which was hampered by the formation of antigen-antibody complex on the GQD-modified Au electrode. Thus, the signal response of the immunosensor decreased along with the increased Y. enterecolitica concentration in the sample solution. To best of our knowledge, we could achieve the highest sensitivity and specificity for Y. enterocolitica by using the label-free, real time GQDs-based electrochemical sensor. This may help in the identification of possible infectious diseases and any pathogenic bacteria which exist in food products. Scheme 1. Principle of the graphene quantum dots (GQDs)-based immunosensor for Y. enterocolitica detection.

Fabrication, Cleaning and Coating of Working Electrodes
The electrodes were laid out on the glass slide using a fine metal mask, which was made of a laser-cut patterned stainless steel. Gold was deposited on the wafer using an electron beam evaporator (Ebeam system Nanovak NVEB-600, Nanovak, Ankara, Turkey). Prior to the application of Au (200 nm), a 40 nm Ti layer was implemented onto the wafer as an intermediary adhesive layer to enhance the adhesion between the glass slide and the Au [24]. The electrodes were then washed with double distilled water and dried thoroughly under a stream of nitrogen gas. Electrochemical measurements were performed using a Keithley-4200 semi-conductor parameter analyzer KTE I version (V9.1). The illustration of electrodes and the measurement set up are presented in Figure 1.
The electrodes were cleaned by employing nitrogen plasma prior to laminating the EDC-NHS activated GQDs on the sensing surfaces [25]. The GQDs, diluted in double distilled water, were mixed with EDC (0.4 M)-NHS (0.1 M) solution in a tube, which was stirred on a shaker for 8 h at 4 °C. A 200 µL of this solution was then injected onto the gold working electrodes and incubated overnight to laminate the GQDs on the surface.

Fabrication, Cleaning and Coating of Working Electrodes
The electrodes were laid out on the glass slide using a fine metal mask, which was made of a laser-cut patterned stainless steel. Gold was deposited on the wafer using an electron beam evaporator (Ebeam system Nanovak NVEB-600, Nanovak, Ankara, Turkey). Prior to the application of Au (200 nm), a 40 nm Ti layer was implemented onto the wafer as an intermediary adhesive layer to enhance the adhesion between the glass slide and the Au [24]. The electrodes were then washed with double distilled water and dried thoroughly under a stream of nitrogen gas. Electrochemical measurements were performed using a Keithley-4200 semi-conductor parameter analyzer KTE I version (V9.1). The illustration of electrodes and the measurement set up are presented in Figure 1.
Six different concentrations of GQDs (0.5, 5, 50, 500, 5000 and 50,000 ppm) were prepared in double distilled water and then measured on the bare electrode surface to determine the optimum GQD concentration based on the signal/concentration relationship towards the reduction of 5 mM H2O2. The experiments for each concentration were repeated three times and the optimum GQD concentration was determined for bioassays according to average electrochemical sensor signals.

Antibody Immobilization and Characterization
The GQD-laminated sensor surface was washed with PBS buffer (pH: 7.4) prior to the covalent immobilization of the anti-Y. enterocolitica antibody (50 μg mL −1 , prepared in NaAc buffer, pH 4.5) during 1 h incubation. To inactivate the remaining carboxyl groups on the surface and avoid possible non-specific interactions during the analyte detection assays, the electrodes were incubated with BSA (100 μg mL −1 , prepared in PBS buffer) and ethanolamine (1 M, prepared in double distilled water, pH 8.5) for 4 min each. Each step of antibody immobilization was performed at room temperature in a Petri dish to protect it from dust or evaporation. All reagents (antibody, BSA, ethanolamine) were applied to the sensor surface at 200 µL volume. Between the subsequent steps the sensor surface was carefully rinsed with PBS buffer (pH 7.4). The bare gold, GQD-laminated, and antibody-immobilized sensor surfaces were characterized by an atomic force microscope (AFM) (Naio, Nanosurf AG, The electrodes were cleaned by employing nitrogen plasma prior to laminating the EDC-NHS activated GQDs on the sensing surfaces [25]. The GQDs, diluted in double distilled water, were mixed with EDC (0.4 M)-NHS (0.1 M) solution in a tube, which was stirred on a shaker for 8 h at 4 • C. A 200 µL of this solution was then injected onto the gold working electrodes and incubated overnight to laminate the GQDs on the surface.

Determination of Optimal GQDs Concentration for Bioassays
The morphology and uniformity of GQDs were initially characterized using a transmission electron microscope (TEM, JEM-2100, JEOL Ltd. Tokyo, Japan). The GQD sample was prepared by filtering 10 µL through a 1.2 µm glass fiber syringe filter and depositing the filtrate on a silicon chip attached to a TEM holder, and leaving them to dry overnight in a fume hood prior to measurement. Six different concentrations of GQDs (0.5, 5, 50, 500, 5000 and 50,000 ppm) were prepared in double distilled water and then measured on the bare electrode surface to determine the optimum GQD concentration based on the signal/concentration relationship towards the reduction of 5 mM H 2 O 2 . The experiments for each concentration were repeated three times and the optimum GQD concentration was determined for bioassays according to average electrochemical sensor signals.

Antibody Immobilization and Characterization
The GQD-laminated sensor surface was washed with PBS buffer (pH: 7.4) prior to the covalent immobilization of the anti-Y. enterocolitica antibody (50 µg mL −1 , prepared in NaAc buffer, pH 4.5) during 1 h incubation. To inactivate the remaining carboxyl groups on the surface and avoid possible non-specific interactions during the analyte detection assays, the electrodes were incubated with BSA (100 µg mL −1 , prepared in PBS buffer) and ethanolamine (1 M, prepared in double distilled water, pH 8.5) for 4 min each. Each step of antibody immobilization was performed at room temperature in a Petri dish to protect it from dust or evaporation. All reagents (antibody, BSA, ethanolamine) were applied to the sensor surface at 200 µL volume. Between the subsequent steps the sensor surface was carefully rinsed with PBS buffer (pH 7.4). The bare gold, GQD-laminated, and antibody-immobilized sensor surfaces were characterized by an atomic force microscope (AFM) (Naio, Nanosurf AG, Liestal, Switzerland). Commercially available AFM probes (NCLR) from NanoWorld (NanoWorld AG, Liestal, Switzerland) were used for the measurements at a spring constant of 23.191 N/m, resonant frequency of 170,481 Hz, scan size of 3.027 µm, and scan speed of 2 s. All AFM measurements were carried out at room temperature using the intermittent air mode.

Development of the Bioassay for Milk Samples
The initial assays were carried out in PBS buffer in a concentration range of 1-6.23 × 10 8 cfu mL −1 using 5000 ppm of GQDs. For the assays in milk (50%, diluted with buffer), two sets of samples were prepared to determine the most convenient GQD concentration that is necessary for the quantification of Y. enterocolitica at lowest possible level. One set of milk samples contained 6 cfu mL −1 of the bacterium, whereas the second set comprised of 6.23 × 10 8 cfu mL −1 . Each sample set was measured on two different electrode surfaces, which were prepared with GQDs at a concentration of 5000 or 50,000 ppm. The bacteria samples (200 µL) were incubated on the anti-Y. enterocolitica antibody immobilized electrode surfaces for 8 min, washed with PBS prior to the amperometric measurements at −0.2 V in the presence of H 2 O 2 (5 mM in PBS) [23]. Accordingly, the detection of bacterium in a wide concentration range was investigated using the electrodes prepared with 5000 ppm GQDs.

Development of the Bioassay for Human Serum Samples
The optimum concentration of GQDs for serum assays was investigated as described in Section 2.5. As the lower amount of GQD (5000 ppm) did not allow the quantification of the bacterium at 6 cfu mL −1 , the higher GQD concentration (50,000 ppm) was selected. The quantification of Y. enterocolitica from 1 to 6.23 × 10 8 cfu mL −1 was then studied using the GQD immunosensor. Each sample was injected onto the antibody immobilized surface at a volume of 200 µL, incubated for 8 min, and measured electrochemically in the presence of H 2 O 2 (5 mM in PBS). All bioassays were performed using 50% of human serum diluted with buffer.

Determination of Optimal GQDs Concentration for Bioassays
In this study, GQDs (<5 nm) were used as nanozymes replacing enzymatic systems. The intimate electronic interactions between the Au electrode and GQDs supplied high electrical conductivity and catalytic surface area, which turned them into the exceedingly active electrocatalysts for H 2 O 2 reduction [22,23]. In this study, we first determined the convenient amount of GQDs for use in sensor development, based on the signal-concentration relationship. The GQDs resulted in sensor signals at mA range on the bare gold electrodes surfaces. Among the six different concentrations tested, 5000 ppm was initially selected for use in bioassays. This concentration led to 23.8 ± 2.5 mA response, which is significantly higher than normal bioassay systems, where the measured concentration of an analyte generally results in a response ranging from µA to nA. Although the higher GQDs concentration (50,000 ppm) led to a 38.76 ± 2.6 mA response, it was not chosen because this would have increased the cost of the developed system ( Figure 2).
The uniform size and morphology of the GQDs were confirmed by employing TEM (Figure 3). The uniform size and morphology of the GQDs were confirmed by employing TEM (Figure 3).

Label Free Direct Assay for Y. enterocolitica Determination in Milk and Human Blood
The initial investigations for Y. enterocolitica detection were carried out in buffer to establish the bioassays. The bacterium specific antibody was covalently immobilized to the GQD-laminated Au electrode surface. The bare, GQD-laminated, and antibody-immobilized sensor surfaces were characterized by AFM (Figure 4). The 3D surface topology images in 1 × 1 µm 2 scanning area resulted in the heights of 12 nm, 18.5 nm, and 23.4 nm for bare ( Figure 4A), GQD-laminated ( Figure 4B), and antibody-immobilized ( Figure 4C) surfaces, respectively. The gradual increase of the surface height confirmed the successful preparation of the sensing surface for bacteria detection assays [26].

Label Free Direct Assay for Y. enterocolitica Determination in Milk and Human Blood
The initial investigations for Y. enterocolitica detection were carried out in buffer to establish the bioassays. The bacterium specific antibody was covalently immobilized to the GQD-laminated Au electrode surface. The bare, GQD-laminated, and antibody-immobilized sensor surfaces were characterized by AFM (Figure 4). The 3D surface topology images in 1 × 1 µm 2 scanning area resulted in the heights of 12 nm, 18.5 nm, and 23.4 nm for bare ( Figure 4A), GQD-laminated ( Figure 4B), and antibody-immobilized ( Figure 4C) surfaces, respectively. The gradual increase of the surface height confirmed the successful preparation of the sensing surface for bacteria detection assays [26]. Eight Y. enterocolitica samples (1-6.23 × 10 8 cfu mL −1 ) were prepared in PBS buffer (pH 7.4). These samples were then injected onto the sensor surface and the measurements were carried out using a three electrode system connected to a Keithley-4200 semi-conductor parameter analyzer. A 5000 ppm concentration of GQDs was sufficient to quantify 1 cfu mL −1 of Y. enterocolitica in 200 µL (the actual LOD was 5 cfu mL −1 ). However, the lower GQDs concentration (i.e., 500 ppm) could not provide this sensitivity level. The individual real time measurement results of bacteria detection are given in Figure 5A. The overall results of the assays in PBS buffer were subjected to logarithmic regression analysis, revealing R 2 values of 0.98 and 0.92 in the concentration range of 6.23 × 10 2 -6.23 × 10 8 cfu mL −1 and 1-6.23×10 8 cfu mL −1 , respectively ( Figure 5B). The lowest concentration of the bacterium generated a 2.8 ± 0.1 mA signal in buffer, which is significantly higher than those of earlier studies (~20 nA for 50 cfu mL −1 of E. coli), where HRP was used as the enzyme label in AuNP amplified sandwich assay in buffer [27]. This confirms the excellent properties of GQDs as enzyme mimics for biosensing applications. Moreover, the current bioassay strategy does not require the conjugation of nanomaterials with a secondary antibody, which decreases the cost and provides easy-to-apply sensor arrays [24,27]. Eight Y. enterocolitica samples (1-6.23 × 10 8 cfu mL −1 ) were prepared in PBS buffer (pH 7.4). These samples were then injected onto the sensor surface and the measurements were carried out using a three electrode system connected to a Keithley-4200 semi-conductor parameter analyzer. A 5000 ppm concentration of GQDs was sufficient to quantify 1 cfu mL −1 of Y. enterocolitica in 200 µL (the actual LOD was 5 cfu mL −1 ). However, the lower GQDs concentration (i.e., 500 ppm) could not provide this sensitivity level. The individual real time measurement results of bacteria detection are given in Figure 5A. The overall results of the assays in PBS buffer were subjected to logarithmic regression analysis, revealing R 2 values of 0.98 and 0.92 in the concentration range of 6.23 × 10 2 -6.23 × 10 8 cfu mL −1 and 1-6.23×10 8 cfu mL −1 , respectively ( Figure 5B). The lowest concentration of the bacterium generated a 2.8 ± 0.1 mA signal in buffer, which is significantly higher than those of earlier studies (~20 nA for 50 cfu mL −1 of E. coli), where HRP was used as the enzyme label in AuNP amplified sandwich assay in buffer [27]. This confirms the excellent properties of GQDs as enzyme mimics for biosensing applications. Moreover, the current bioassay strategy does not require the conjugation of nanomaterials with a secondary antibody, which decreases the cost and provides easy-to-apply sensor arrays [24,27].
To ensure the optimum GQD amount for milk and serum assays, we first tested two different concentrations of Y. enterecolitica (6 and 6.23 × 10 8 cfu mL −1 ) by using the working electrodes prepared with 5000 or 50,000 ppm GQD. The lower GQD concentration (5000 ppm) was selected for milk assays, as it allowed the quantification of 6 cfu mL −1 of the bacterium with a sufficient sensor signal (17.2 ± 2.1 nA). On the other hand, this GQD concentration did not generate a signal in human serum for 6 cfu mL −1 of Y. enterecolitica and led to only 120 ± 20 pA signal for the highest concentration of the bacterium (6.23 × 10 8 cfu mL −1 ). This is attributed to the presence of various interfering compounds in human serum, such as hormones, minerals, and proteins. Based on the preliminary tests, the assays in milk and serum were performed using the GQD concentrations of 5000 and 50,000 ppm, respectively ( Figure 6). 1 cfu mL -1 6.23  10 2 cfu mL -1 6.23  10 3 cfu mL -1 6.23  10 4 cfu mL -1 6.23  10 5 cfu mL -1 6.23  10 6 cfu mL -1 6.23  10 7 cfu mL -1 6.23  10 8 cfu mL -1 To ensure the optimum GQD amount for milk and serum assays, we first tested two different concentrations of Y. enterecolitica (6 and 6.23 × 10 8 cfu mL −1 ) by using the working electrodes prepared with 5000 or 50,000 ppm GQD. The lower GQD concentration (5000 ppm) was selected for milk assays, as it allowed the quantification of 6 cfu mL -1 of the bacterium with a sufficient sensor signal (17.2 ± 2.1 nA). On the other hand, this GQD concentration did not generate a signal in human serum for 6 cfu mL -1 of Y. enterecolitica and led to only 120 ± 20 pA signal for the highest concentration of the bacterium (6.23 × 10 8 cfu mL −1 ). This is attributed to the presence of various interfering compounds in human serum, such as hormones, minerals, and proteins. Based on the preliminary tests, the assays in milk and serum were performed using the GQD concentrations of 5000 and 50,000 ppm, respectively ( Figure 6). Ten different concentrations of the bacterium, from 1 to 6.23 × 10 8 cfu·mL −1 , were spiked in milk and serum. Each sample was injected to the antibody immobilized sensor for amperometric measurements at −0.2V. The lowest amounts of Y. enterocolitica measured in milk and serum were 1 and 6 cfu mL −1 , respectively. The sensor generated 0.6 and 7.73 nA signals for only milk and 1 cfu mL −1 bacterium in milk, respectively. The negative control (serum) and 6 cfu mL −1 of Y. enterocolitica in serum resulted in 0.22 and 0.329 nA sensor signals, respectively. It is worth noting that all samples were prepared as 1 mL solutions and only 200 µL of the samples were injected to the electrode surface for measurements. Therefore, the actual LODs of milk and serum assays were determined to be 5 and 30 cfu mL −1 , respectively. The overall results of bioassays for milk and serum samples are given in Figure 7. The correlation coefficient of both assays was found to be 0.98 when the data were subjected to the logarithmic regression analysis (Figure 7). were prepared as 1 mL solutions and only 200 µL of the samples were injected to the electrode surface for measurements. Therefore, the actual LODs of milk and serum assays were determined to be 5 and 30 cfu mL −1 , respectively. The overall results of bioassays for milk and serum samples are given in Figure 7. The correlation coefficient of both assays was found to be 0.98 when the data were subjected to the logarithmic regression analysis (Figure 7).  The current methods for Y. enterecolitica detection include cultivation [7], ELISA [9], PCR [10,11], and SPR-based tests [12,13]. These techniques are either labor intensive or time consuming, and lack the desired sensitivity. The limit of detection of these methods varies between 10 2 -10 6 cfu mL −1 (Table 1). A few electrochemical sensors have been reported for the detection of Yersinia species, targeting bacteria specific genes [15,16,28]. The direct detection of bacteria using electrochemical transducers has not been studied so far. Additionally, although the nanomaterials integrated sensors have dominated in pathogenic bacteria detection in recent years [22,25,29,30], to best of our knowledge, such a system has not been introduced for Yersinia determination. Compared to our earlier works [24,27], reporting on the quantification of E. coli in water (LOD: 50 cfu·mL −1 ) and Salmonella typhimirium in human stool (LOD: 12 cfu mL −1 ) using gold nanoparticles amplified electrochemical sensors, we achieved higher sensitivities (LOD milk = 5 cfu mL −1 , LOD serum = 30 cfu mL −1 ) in the current work, despite working with more complex media. It is worth noting that we performed a static assay procedure in the current work with a limited sample volume. In case of applying the developed strategy for fully automated microfluidic systems, we expect the sensitivity of the assays to increase, which may lead to quantifying one single bacterium. The current work allows label-free detection of Y. enterocolitica with a direct assay methodology, not requiring enzyme labels and detector antibody. The superior features of GQD-immunosensor supply a much shorter assay time than those of previous studies [24,27]. 1 cfu mL -1 6.23  10 2 cfu mL -1 6.23  10 3 cfu mL -1 6.23  10 4 cfu mL -1 6.23  10 5 cfu mL -1 6.23  10 6 cfu mL -1 6.23  10 7 cfu mL -1 6.23  10 8 cfu mL -1  The current methods for Y. enterecolitica detection include cultivation [7], ELISA [9], PCR [10,11], and SPR-based tests [12,13]. These techniques are either labor intensive or time consuming, and lack the desired sensitivity. The limit of detection of these methods varies between 10 2 -10 6 cfu mL −1 ( Table  1). A few electrochemical sensors have been reported for the detection of Yersinia species, targeting bacteria specific genes [15,16,28]. The direct detection of bacteria using electrochemical transducers has not been studied so far. Additionally, although the nanomaterials integrated sensors have dominated in pathogenic bacteria detection in recent years [22,25,29,30], to best of our knowledge, such a system has not been introduced for Yersinia determination. Compared to our earlier works [24,27], reporting on the quantification of E. coli in water (LOD: 50 cfu·mL −1 ) and Salmonella typhimirium in human stool (LOD: 12 cfu mL −1 ) using gold nanoparticles amplified electrochemical sensors, we achieved higher sensitivities (LODmilk= 5 cfu mL −1 , LODserum= 30 cfu mL −1 ) in the current work, despite working with more complex media. It is worth noting that we performed a static assay procedure in the current work with a limited sample volume. In case of applying the developed strategy for fully automated microfluidic systems, we expect the sensitivity of the assays to increase, which may lead to quantifying one single bacterium. The current work allows label-free detection of Y. enterocolitica with a direct assay methodology, not requiring enzyme labels and detector antibody. The superior features of GQD-immunosensor supply a much shorter assay time than those of previous studies [24,27].

Cross-Reactivity Studies for Y. enterocolitica
To determine the specificity of the developed assay for Y. enterocolitica, the interaction of non-specific bacteria with the bacterium-specific antibody-immobilized sensor was studied. For this, the reference samples of S. enteritidis, B. anthracis, E. coli, and Y. peptis were separately prepared in milk and human serum at a fix concentration of 10 7 cfu mL −1 . Each sample was injected to the antibody immobilized surfaces and measured during 200 s. An extremely low cross-reactivity was recorded for all reference pathogens in milk, suggesting a very high specificity of the bioassay for the target bacterium ( Table 2). The relative cross-reactivities of the non-specific pathogenic bacteria were found to be higher in human serum (Table 3) compared to milk. However, the maximum cross-reaction was still at an acceptable level (22.8%), which was recorded for another Yersinia species. It is also worth noting that the selected concentration of bacteria in cross-reactivity tests was quite high. Table 2. Current response profile for Y. enterocolitica binding in milk compared with other pathogenic bacteria, and the percentage of cross reactivity of non-specific bacteria (n = 3). To realize the applicability of the GQD-immunosensor in the co-existence of various bacteria species, we mixed the target bacterium and non-specific bacteria at equal concentrations (10 7 cfu·mL −1 ), and subsequently measured the sensor signal to determine the relative activity for Y. enterecolitica. Considering that Y. pestis led to higher cross-reactivity for individual tests as being another Yersinia genus, we prepared two different mixtures of non-specific bacteria: 1) A mixture of S. enteritidis, E. coli, and B. anthracis; and 2) a mixture of S. enteritidis, E. coli, B. anthracis, and Y. pestis. The relative activity of the sensor for milk assays decreased only 2.2% and 5.5% in mixtures 1 and 2, respectively ( Table 2). The suppression of the sensor signal in serum was determined to be higher, with 12.5% and 24.2% decrease in relative activity in mixtures 1 and 2, respectively (Table 3). Nevertheless, these results suggest that GQD-immunosensor is capable of quantifying the target bacterium at high specificity even in the co-existence of several non-specific bacteria.

Conclusions
In this work, a novel GQD-based immunosensor was developed using an electrochemical transducer. Among all methods reported for the detection of Y. enterecolitica, the GQDs-enriched sensor supplied the highest selectivity and sensitivity in complex media, such as milk (LOD: 5 cfu·mL −1 ) and human serum (LOD: 30 cfu·mL −1 ). The developed method can be used for any pathogenic bacteria detection for clinical and food samples without pre-sample treatment. Offering a very rapid, specific and sensitive detection with a label-free system, the GQD-based immunosensor can be coupled with various electrochemical biosensors.