Skip to main content
NCBI home page
Food Chemistry: X logoLink to Food Chemistry: X
. 2024 Mar 15;22:101281. doi: 10.1016/j.fochx.2024.101281

Colorimetric sensor array for the rapid distinction and detection of various antibiotic-resistant psychrophilic bacteria in raw milk based-on machine learning

Yanan Qin 1,1,⁎, Jingshuai Sun 1,1, Wanting Huang 1, Haitao Yue 1, Fanxing Meng 1, Minwei Zhang 1
PMCID: PMC10966183  PMID: 38544935

Highlights

  • •

    d-amino acid-modified Au NPs were developed for detecting psychrophilic bacteria.

  • •

    The limit of detection was 102 CFU mL−1.

  • •

    The linear discriminant analysis was applied to the distinction between bacteria.

  • •

    Nanosensors were available for antibiotic susceptibility testing rapidly.

Keywords: Psychrophilic bacteria, Colorimetric detection, Gold nanoparticles, d-amino acid, Antibiotic susceptibility assay

Abstract

In this study, a rapid, inexpensive, and accurate colorimetric sensor for detecting psychrophilic bacteria was designed, comprising gold (Au) nanoparticles (NPs) modified by d-amino acid (D-AA) as color-metric probes. Based on the aggregation of Au NPs induced by psychrophilic bacteria, a noticeable color shift occurred within 6 h. Depending on the various metabolic behaviors of bacteria to different D-AA, four primary psychrophilic bacteria in raw milk were successfully distinguished by learning the response patterns. Furthermore, the quantification of single bacteria and the practical application in milk samples could be realized. Notably, a rapid colorimetric method was constructed by combining Au/D-AA with antibiotics for the minimum inhibitory concentration of psychrophilic bacteria, which relied on differences in bacteria metabolic activity in response to diverse antibiotic treatments. Therefore, the method enables the rapid detection and susceptibility evaluation of psychrophilic bacteria, promoting clinical practicability and antibiotic management.

Introduction

Psychrotrophic bacteria are defined as a heterogeneous group of microorganisms able to grow below 7 °C, regardless of their higher optimal growth temperature (Ribeiro Júnior et al., 2018). These bacteria account for > 75 % of the microflora in raw milk, especially in situations of poor hygiene (Salgado et al., 2020, Vithanage et al., 2016). Pseudomonas spp. and Acinetobacter spp. are the most frequently isolated from refrigerated raw milk, which can produce heat-stable protease and lipases that spoil dairy products (Fusco et al., 2020, Machado et al., 2017, Von Neubeck et al., 2015). Furthermore, the misuse of antibiotics in recent years has led to psychrophilic bacteria becoming antibiotic-resistant (Böger et al., 2021, Dietvorst et al., 2020). Pseudomonas spp. can spread antibiotic-resistant genes and cause severe clinical mastitis, posing a severe hazard to the dairy industry (Meng et al., 2020, Quintieri and Caputo. , 2019). Therefore, detecting psychrotrophic bacteria and their antibiotic resistance in the dairy processing industry is critical.

Traditional strategies for psychrophilic bacteria detection primarily include the plate-counting method (Hameed et al., 2018), polymerase chain reaction (Machado et al., 2013), and flow cytometry (FCM) (M. Wang et al., 2023, Wang et al., 2023). However, these methods have limitations restricting their general implementation, such as their time-consuming nature, complexity, and associated costs (Karo et al., 2008, Martín et al., 2008, Ziyaina et al., 2020). Therefore, a practical, easily accessible, and affordable approach for detecting psychrophilic bacteria is urgently required. Colorimetric detection approaches have recently been found to be potentially more effective than traditional methods in the field due to their simplicity of use, rapid reaction, cheap cost, and naked-eye visibility (Bu et al., 2019, Geleta, 2022; X. Wang et al., 2023, Wang et al., 2023). For the development of colorimetric biosensors, gold (Au) nanoparticles (NPs) offer an excellent platform due to their unique optical properties and their simple synthesis and functionalization (Liu and Dong, 2023, Luo et al., 2019, Peng and Chen, 2019, Wu et al., 2022). Using aptamers to decorate Au NPs for bacterial targeting assays is a common strategy at present (Chen et al., 2020, Chen et al., 2021, Qi et al., 2023). Unfortunately, the technology depends on specialized expertise for aptamer generation, and the success rate for producing targeted aptamers is comparatively low (Banu et al., 2022).

d-amino acid (D-AA) is a peptidoglycan component, the primary part of the bacterial cell wall surrounding the bacteria (Caldwell et al., 2023, Sasabe et al., 2016, Typas et al., 2012). Due to the enzymatic promiscuity of bacterial transpeptidases, researchers have shown that various bacteria can easily incorporate exogenous D-AA from the surrounding media into their peptidoglycan by displacing the terminal D-AA of the oligopeptide strand (Gao et al., 2022, Hsu et al., 2019, Pundir et al., 2018). D-AA-coated Au NPs have been developed to detect Staphylococcus aureus (S. aureus) within ascite samples from patients (Yang et al., 2018). The approach based on bacterial metabolism was also used for methicillin-resistant Staphylococcus aureus fingerprinting and antibiotics profiling (Gao et al., 2022). Psychrophilic bacteria, a type of microorganism, exhibit analogous patterns in cell membrane metabolism akin to other bacterial strains. Therefore, detecting psychrophilic bacteria could be considered with D-AA-modified Au NPs.

Herein, a rapid and economical colorimetric sensor array was established to identify psychrophilic bacteria based on machine learning. Due to their specific metabolic behavior, the sensor array can also conduct antibiotic susceptibility testing (AST) of psychrophilic bacteria. First, D-AA was used to specifically metabolize into the bacterial cell wall, whereas fungus or mammalian cells cannot. The Au NPs became unstable and aggregated on the bacteria surface after exogenous D-AA unbound with Au NPs. A good distinction between four psychrophilic bacteria was realized in tryptic soy broth (TSB) medium and milk samples. Finally, a rapid approach for AST was constructed based on the distinctive metabolic behavior of D-AA, where Au/D-AA and psychrophilic bacteria were incubated with various antibiotic concentrations.

Experimental section

Materials and chemicals

d-alanine (d-Ala), d-glutamic acid (d-Glu), ciprofloxacin (CIP), and tetracycline (TET) were obtained from Yuanye Biotechnology Co., Ltd. (Shanghai, China). l-alanine (l-Ala) and l-glutamic acid (l-Glu) were obtained from Sigma-Aldrich (Shanghai, China). Chloroauric acid trihydrate (HAuCl4·3H2O) and sodium borohydride (NaBH4) were purchased from Solarbio Science & Technology Co., Ltd. (Beijing, China) and Aladdin Industrial Company (Shanghai, China), respectively. Ceftriaxone (CRO) and gentamicin (GM) were purchased from Macklin Biochemical Co., Ltd. (Shanghai, China). All compounds were analytical grade without additional purification, using ultrapure water throughout the process.

Apparatus

The ultraviolet–visible (UV–vis) spectra were obtained on a UV-3100PC spectrometer (MAPADA, Shanghai, China). The morphology of the nanoprobes was measured using transmission electron microscopy (TEM; FEI Tecnai G2 F20, Hillsboro, OR, USA). The composition of sensor probes was characterized using a Fourier-transform infrared (FT-IR; Thermo Scientific Nicolet iS20, Madison, WI, USA) spectrometer. All pH values were measured on a PHS-3C instrument (INESA, Shanghai, China). All data were processed in Originpro 2021 (version 9.8.0.200, Northampton, MA, USA) and Systat (version 13.2, San Jose, CA, USA).

Synthesis of Au/D-AA

Using Au/d-Ala for example, 200 μL of HAuCl4 (48.6 mM), 40 μL of d-Ala (24.3 mM), and 20 mL of deionized water were mixed vigorously at room temperature. Then, 500 μL of freshly prepared NaBH4 (0.1 M) was added incrementally to the mixture, and the mixture progressively transformed from light yellow to purple and then to orange–red. The mixed solution was stirred for 15 min at room temperature, dialyzed (5 kDa, Biosharp) with ultrapure water for 24 h, and changed every 4 h. Sterilization was performed using a 0.22-μm filter and then kept at 4 °C until usage. The same procedure as for Au/d-Ala was used to prepare Au/d-Glu. Furthermore, probes comprising d-Ala and d-Glu mixed in different molar ratios (1:1, 1:2, and 2:1) were prepared using the above method.

Isolation and identification of psychrophilic bacteria

The raw milk samples were obtained from Fuhai (45°05′N, 88°04′E) and Jeminai (47°68′N, 86°20′E), Altay Area, Xinjiang, in September 2021, which were fully suspended by vortexing 9 mL of sterile saline (0.85 %) and subsequently deposited onto the solid modified plate count agar plates at 7 °C for 10 days. Hundreds of strains of psychrophilic bacteria were isolated, and bacterial genome deoxyribonucleic acid (DNA) was extracted with the FastDNA Spin Kit for Soil (MP Biomedicals, USA). The 16 s ribosomal ribonucleic acid (rRNA) gene primers were 338F (5́-ACTCCTACGGGAGGCAGCAG-3́) and 806R (5́-GGACTACHVGGGTWTCTA-AT-3́) and 1.5 % agarose gel electrophoresis was used to assess the quality of the 16 s rRNA gene. The 16 s rRNA gene was sequenced and analyzed using the BLAST method by Bioengineering Co., Ltd. (Shanghai, China). Moreover, the phylogenetic tree of psychrophilic bacteria was constructed with MEGA 7.0.

Detection of psychrophilic bacteria

First, 20 μL of psychrophilic bacteria were combined with 40 μL of TSB medium and 100 μL of distilled water. Thereafter, 40 μL of probes were added to the mixtures at different concentrations separately. Spectra UV–vis of the mixed solutions were obtained after incubating at 25 °C for 6 h, and the samples underwent five parallel tests. Finally, linear discriminant analysis (LDA) was applied to identify and categorize psychrophilic bacteria using the ratio of the absorption spectra values at 650 nm and 520 nm (A650/A520) for each sample. An initial concentration of psychrophilic bacteria (104 CFU mL−1, 105 CFU mL−1, and 106 CFU mL−1) was added to the pasteurized milk, respectively. The procedures were similar for pasteurized milk samples outlined for the TSB medium, except that the additional bacteria were spiked to a 10 × dilution of milk.

Antibiotic susceptibility testing

Briefly, to determine the MIC, 20 μL of psychrophilic bacteria were added to a mixture including the TSB medium (40 μL), water (100 μL), a single probe (40 μL), and various antibiotic concentrations (20 μL) and incubated for 6 h at 25 °C. Four antibiotics (ceftriaxone (CRO), gentamicin (GM), tetracycline (TET), and ciprofloxacin (CIP)) were used to show the AST ability of the probes. Furthermore, the disk diffusion susceptibility testing was conducted according to the Clinical and Laboratory Standards Institute (CLSI) criteria by culturing the bacteria on nutrient agar plates at 25 °C for 24 h, measuring the diameter of a clear region to indicate the bacterial inhibition zone.

Results and discussion

Isolation and identification of psychrophilic bacteria

The 16 s rRNA sequences of the screened psychrophilic bacteria were compared with the sequences in the National Center for Biotechnology Information database using BLAST, and the similarities between the strains allowed us to determine their respective genera. Pseudomonas spp. (23.68 %), Klebsiella spp. (5.26 %), Acinetobacter spp. (26.32 %), and Chryseobacterium spp. (5.79 %) accounted for most of the psychrophilic bacteria identified. The phylogenetic tree of the four strains (marked as A, B, C, and D, respectively) was generated using the neighbor-joining method in MEGA 7.0 software (Fig. S1). Strain A revealed a 99.09 % similarity to Pseudomonas endophytica (P. endophytica, BSTT44), whereas strain B displayed a 99.65 % similarity to Klebsiella oxytoca (K. oxytoca, ATCC 13182). Strains C and D were both 99.72 % similar to Acinetobacter johnsonii (A. johnsonii, XAAS) and Chryseobacterium timonianum (C. timonianum, G972), respectively. These four psychrophilic bacteria were used in subsequent detection experiments with Au/D-AA.

Synthesis and characterization of D-AA-functionalized Au NPs

D-AA-functionalized Au NPs were prepared in a one-pot reaction without other complex modifications (Fig. 1A). Using the absorbance ratio (A650/A520) and the color of Au/D-AA as indicators, an appropriate synthetic technique for Au/D-AA was proposed by optimizing the amount of NaHB4 and the molar ratio of HAuCl4 and D-AA. The Au/D-AA grew from light red to brilliant red with increasing amounts of NaHB4 (35 μM–80 μM) and molar ratios of HAuCl4 and D-AA (from 1:1 to 10:1). A650/A520 progressively dropped, which could be attributable to NaHB4, speeding up the reduction of HAuCl4 and increasing the rate at which Au/D-AA was produced (Fig. S2) (Daruich De Souza et al., 2019). However, HAuCl4 and D-AA molar ratios more than 10:1 or NaHB4 concentrations greater than 80 μM caused the Au/D-AA to aggregate (Fig. S2). Therefore, 80 μM and 10:1 were determined as the optimal amounts of NaHB4 and the proper molar ratio of HAuCl4 and D-AA, respectively.

Fig. 1.

Fig. 1

Open in a new tab

(A) Schematic diagram of the synthesis of D-AA-modified Au NPs. (B) Absorption spectra of d-Ala-modified Au NPs and pure Au NPs. (C) TEM image of Au/d-Ala. (D) FTIR spectra of Au/d-Ala and d-Ala. (E) Absorption spectra of d-Glu-modified Au NPs and pure Au NPs. (F) TEM image of Au/d-Glu. (G) FTIR spectra of Au/d-Glu and d-Glu.

With the modification of D-AA, the plasmon absorption peaks of Au/d-Ala and Au/d-Glu were almost unchanged compared to pure Au NPs (Fig. 1B and E). The synthesized NPs showed typical spherical morphologies (Fig. 1C and F), approximately 3 nm in diameter by TEM images (Fig. S3). Furthermore, the nanoprobes were characterized using FT-IR spectra (Fig. 1D and G). Several distinctive pure D-AA peaks were discovered in the nanoprobes. Regarding Au/d-Ala, d-Ala exhibited characteristic bands at 1626 and 3382 cm−1, corresponding to the bending modes and stretching vibrations of N—H, respectively. Additionally, the band at 1361 cm−1 was the stretching mode of the C—N bond, indicating the existence of a carboxyl group (Yang et al., 2018). The Au/d-Glu combination yielded findings similar to those obtained with the matching d-Glu, with distinctive N—H (3229 and 1632 cm−1), carboxyl group (1407 cm−1), and C—N (1354 cm−1) bands being observed. The above results indicated that the Au NPs had been effectively coated with D-AA.

The stability of Au/D-AA significantly influences its detection ability. The stability of Au/D-AA in gradient pH ranges (4–12), at a sodium chloride concentration of 2.5–40 mM, and stored at 4 °C for 30 days was investigated. The solution color was the same, and the UV–vis spectra of Au/D-AA were nearly unchanged (Figs. S4–S6). These results demonstrated that the probes were stable and suitable for application in further research.

Colorimetric detection of psychrophilic bacteria via Au/D-AA

Principle of detecting psychrophilic bacteria with Au/D-AA

While Au/D-AA was incubated with the bacteria, the D-AA detached from the Au NP surfaces and was metabolized by the bacteria into peptidoglycan (Fig. S7A). Then, the stability of the Au NPs was disturbed, accumulating on the surface of bacteria, and the solution changed from red to blue. Correspondingly, the absorption spectra peak of Au NPs shifted from 520 nm to 650 nm (Fig. S7B).

To validate the detecting principle of the colorimetric sensor, Au/d-Ala and Au/d-Glu were cultivated with the bacteria (107 CFU mL−1), respectively. The probes in the pure TSB medium in the absence of psychrophilic bacteria remained red after incubation (Fig. 3A and B), proving the good stability of these probes in the TSB medium. After psychrophilic bacteria were incubated, all samples changed in color and shifted in the absorption spectrum, which decreased in absorbance at 520 nm and increased in absorbance at 650 nm (Fig. 3A and B), showing that the aggregation of Au NPs occurred. Moreover, taking A. johnsonii as an example, the TEM pictures showed the noticeable aggregation of Au NPs on the surface of the bacteria after incubation, demonstrating that the aggregation of Au NPs was related to D-AA (Fig. S8). Furthermore, Au/l-Ala and Au/l-Glu, which were from l-Ala and l-Glu modified Au NPs, respectively, were used as control nanoprobes. As anticipated, no noticeable change occurred in either of their absorption spectra after incubation (Figs. S9A and S9B), indicating the unique metabolic properties of psychrophilic bacteria toward D-AA.

Fig. 3.

Fig. 3

Open in a new tab

(A) Radar plots obtained with three colorimetric sensors against A. johnsonii at different concentration levels (103-107 CFU mL−1). (B) Absorption spectra of the 2:1 sensor after being incubated with A. johnsonii at different concentration levels (103-107 CFU mL−1). (C) The linear relationship between the A 650 nm/A 520 nm value and the concentration of A. johnsonii with the 2:1 sensor.

Rapid quantitative analysis of psychrophilic bacteria in the TSB medium

To investigate the colorimetric response value among Au/D-AA synthesized by mixing different molar ratios of d-Ala and d-Glu (1:1, 1:2, and 2:1), the probes (1:1, 1:2, and 2:1) were cultured with the bacteria under the same conditions. Among the five colorimetric sensors, the highest response values were obtained for the mixing probe (2:1) (Figs. S9C–F). Therefore, three sensors (Au/d-Ala, Au/d-Glu, and 2:1) were selected for the subsequent study.

The LDA based on the A650/A520 values was used to show the discriminating abilities of the colorimetric sensor array. Notably, four strains (P. endophytica, K. oxytoca, A. johnsonii, and C. timonianum) were tightly grouped without overlap and separated, revealing the strong bacterial discriminating abilities of the sensor probes (Fig. 2C). The color key from the heat map can reflect the ability of the bacteria to metabolize D-AA. In this case, a higher color key indicates a larger metabolic capacity for the bacteria. Here, A. johnsonii demonstrated the highest D-AA metabolic capacity compared with the other three bacteria (Fig. 2D). Furthermore, among the three sensor probes (Au/d-Ala, Au/d-Glu, and the 2:1 sensor), the response value of the 2:1 sensor is the most pronounced (Fig. 2D).

Fig. 2.

Fig. 2

Open in a new tab

Absorption spectra of Au/d-Ala (A) and Au/d-Glu (B) after being incubated with 4 different types of psychrophilic bacteria. LDA plots (C) and heat maps (D) of Au/d-Ala, Au/d-Glu, and the 2:1 sensor against 4 different types of psychrophilic bacteria. Insets in A − B: the color of the probe solutions without or with the bacteria.

The quantitative analytical efficacy of the designed sensors was evaluated using A. johnsonii as a model. The response values of three colorimetric sensors induced by A. johnsonii at different concentrations (103–107 CFU mL−1) were learned from the radar plot (Fig. 3A). The data suggested that the response values increased with the addition of bacteria concentrations, indicating an increased aggregation of Au NPs. A greater discrepancy in the metabolism of the 2:1 probe was displayed compared with Au/d-Ala and Au/d-Glu. Additionally, similar results were obtained from Fig. S10, indicating that the psychrophilic bacteria were more preferred to metabolize d-Ala than d-Glu.

Different amounts of A. johnsonii were added to the Au/D-AA solutions to evaluate the sensitivity of the method. The absorption spectra peak of the 2:1 sensor gradually shifted from 520 nm to 650 nm with increasing A. johnsonii concentrations, causing the solution to change from red to blue (Fig. 3B). The linear relationship between the A650 nm/A520 nm value and the concentration of A. johnsonii with the 2:1 sensor displayed a strong linear correlation within 103–107 CFU mL−1, and the limit of detection (3σ/S) was 102 CFU mL−1 (Fig. 3C). Similar results were observed using three colorimetric sensors for detecting four bacteria (Figs. S11–S13). By recording the A650/A520 values within 12 h, it was observed that the response values reached a peak when cultivated for 6 h (Fig. S14). Moreover, the detection time of the colorimetric sensor was noticeably reduced compared to the conventional incubation method, which required 24 h to find clear colony morphology (Fig. S15).

Rapid antibiotic susceptibility testing

Due to the influence of antibiotics on bacterial metabolism behavior, which resulted in changes in the degree of Au NP aggregation, the probe can be used to assess antibiotic resistance. Before proceeding, the effect of various antibiotics on the nanoprobes was investigated. No obvious difference occurred in the absorption spectra, demonstrating that the pure antibiotics did not cause the agglomeration of Au NPs (Fig. S16). The AST was conducted at designated concentrations specified by the CLSI to ascertain whether the strain was susceptible (S) or resistant (R). Au/d-Ala was incubated with P. endophytica and K. oxytoca at different TET concentrations (0–16 μg mL−1), respectively. The study showed that the samples (≤2 μg mL−1 TET) had an A650/A520 value exceeding 0.84, similar to the control group without antibiotics (Fig. 4A and B), indicating the regular metabolic function of P. endophytica and K. oxytoca toward d-Ala. However, samples with ≥ 4 μg mL−1 TET showed a noticeable decrease in the A650/A520 value (<0.84), corresponding to the inhibited metabolism of P. endophytica and K. oxytoca (Fig. 4A and B). Therefore, P. endophytica and K. oxytoca were both S toward TET, and the MIC was 4 μg mL−1. However, the A650/A520 values did not change much, even at the highest concentration (16 μg mL−1) of TET (Fig. 4C and D), demonstrating that A. johnsonii and C. timonianum were both TET-resistant bacteria. The heat maps further confirmed the results, where a noticeable color key change appeared with P. endophytica and K. oxytoca at the MIC concentration, whereas no obvious change was observed with A. johnsonii and C. timonianum (Fig. 4E–H). Subsequently, similar results were observed using Au/d-Glu and the 2:1 probe (Fig. S17), underscoring the AST ability of Au/D-AA.

Fig. 4.

Fig. 4

Open in a new tab

A650 nm /A520 nm values (A-D) of Au/d-Ala and the heat maps (E-H) of Au/d-Ala after incubation with four kinds of psychrophilic bacteria against various concentrations of TET.

Fig. S18 shows the AST results for the other three antibiotics (CRO, GM, and CIP) with Au/d-Ala. P. endophytica and K. oxytoca were classified as S for the three antibiotics (Figs. S18A and S18B). Furthermore, A. johnsonii was S (CRO and GM) and I (CIP) (Fig. S18C). The corresponding C. timonianum was R (CRO and GM) and S (CIP) (Fig. S18D). The results from Au/d-Glu and the 2:1 probe to perform the AST were similar (Fig. S19), confirming the good capability of the colorimetric sensor for AST.

For comparison with the results of Au/D-AA, the broth microdilution method was used to evaluate the antibiotic susceptibility of four psychrophilic bacteria (Fig. S20). The bacteria with inhibition zones larger than the criterion for interpreting the CLSI zone diameter were classified as S or R. For instance, the P. endophytica inhibition diameters for CRO, GM, TET, and CIP were 20, 20, 22.5, and 23 mm, respectively (Figs. S20A and S20B), displaying that P. endophytica was S for the four antibiotics according to CLSI zone diameter criteria. Then, the antibiotic susceptibilities of K. oxytoca, A. johnsonii, and C. timonianum were evaluated under the same procedure (Figs. S20C–S20H). The results were consistent with the AST using the colorimetric sensor basis of Au/D-AA. Furthermore, identifying AST via the Au/D-AA sensor requires only 6 h, unlike the broth microdilution method (at least 24 h), significantly enhancing efficiency.

A machine learning-based colorimetric sensor array for psychrophilic bacteria identification in milk

To investigate the practicality of colorimetric sensor array, pasteurized milk was selected as a real sample. The impact of pasteurized milk samples without psychrophilic bacteria on Au/D-AA was analyzed. No noticeable variation appeared in the absorption spectra, demonstrating that the stability of Au/D-AA was unaffected by pasteurized milk (Fig. S21). The practicality of Au/D-AA was demonstrated by analyzing authentic samples obtained by inoculating pasteurized milk samples with four psychrophilic bacteria. The results outlined excellent recovery rates within 94.5 %–106.0 % (Fig. S22), with a relative standard deviation of < 5 %. These results indicated that the Au/D-AA sensor exhibited an accurate detection ability, which can be used to assay the bacteria in raw milk.

Four psychrophilic bacteria in pasteurized milk were identified using the sensor array to explore the possible application of Au/D-AA in complicated samples. The LDA plot displayed twelve distinct groups with no overlap, indicating the sensor array discriminated effectively, even when spiked with psychrophilic bacteria at different concentrations in milk samples (Fig. S23). Since different types of psychrophilic bacteria can exist in one sample, the discriminating capacity to the combination of psychrophilic bacteria was further evaluated. The mixtures were classified, including four psychrophilic bacteria in varying ratios and concentrations. The LDA result confirmed that the psychrophilic bacteria mixtures in pasteurized milk can be well separated (Fig. S24). Furthermore, the ternary (Fig. S25) and qua-ternary (Fig. 5) psychrophilic bacteria mixtures with different concentrations were distinguished, indicating that the method can discriminate complex mixtures. A detection model was constructed and trained using a database comprising A650/A520 values measured from different sample batches with psychrophilic bacteria. Furthermore, the blind testing on a randomly selected group of psychrophilic bacteria in our training set was studied, where 80 unidentified bacteria samples were chosen randomly from the four psychrophilic bacteria added to the pasteurized milk. Unknown psychrophilic bacterial samples were identified by matching the colorimetric response value to the appropriate LDA group determined by the training set. The blind testing in pasteurized milk yielded a 93.75 % accuracy (Table S1), demonstrating the high potential of the sensory array for practical applications.

Fig. 5.

Fig. 5

Open in a new tab

LDA plots of sensor array against four psychrophilic bacteria with different ratios (1:1:1:1, 1:2:1:1, 1:1:2:1, 1:1:1:2, and 2:1:1:1) at 103 CFU mL−1 (A1-E1), 104 CFU mL−1 (A2-E2), and 105 CFU mL−1 (A3-E3) in pasteurized milk, respectively (P, K, A, and C represent P. endophytica, K. oxytoca, A. johnsonii, and C. timonianum, respectively.).

Conclusion

In summary, an inexpensive, accurate, and rapid colorimetric sensor comprising Au NPs and two D-AA ratios was constructed to detect psychrophilic bacteria and for AST. Distinct colorimetric response patterns resulted from the aggregation behavior of Au NPs, which, in turn, was directly connected to the metabolic aptitude of bacteria toward D-AA. Four psychrophilic bacteria were successfully discriminated against and detected within 6 h, significantly accelerating the progress of bacteria testing compared to the traditional method (24 h). In addition, the colorimetric response patterns realized the practical application in milk samples. Notably, a rapid method of AST combined with Au/D-AA was developed for the MIC, which satisfied an extensive spectrum of clinical needs ranging from prompt diagnosis to accurate drug treatment. However, the technology is still immature due to limited research on the behavioral responses of other psychrophilic bacteria, such as Listeria spp. and Bacillus spp., toward Au/D-AA. Nevertheless, it is effective for rapidly detecting psychrophilic bacteria and antibiotic susceptibility, significantly reducing bacterial infections and the harm caused by the misuse of antibiotics.

CRediT authorship contribution statement

Yanan Qin: Writing – review & editing, Writing – original draft, Supervision, Resources, Project administration, Funding acquisition. Jingshuai Sun: Writing – review & editing, Methodology, Formal analysis. Wanting Huang: Data curation. Haitao Yue: Investigation. Fanxing Meng: Writing – review & editing. Minwei Zhang: Writing – review & editing, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgement

This work was supported by the Natural Science Foundation of Xinjiang Uygurutonomous Region (No.2022D01C404, 2023B02034-1, and No.2019Q002), Tianshan Youth Foundation of Xinjiang (No. 2020Q061), the science and technology young top-notch talent project of Autonomous Region (2022TSYCCX0064), the research start-up fund of Xinjiang University (No. 2018-660010).

Footnotes

Appendix A

Supplementary data to this article can be found online at https://doi.org/10.1016/j.fochx.2024.101281.

Appendix A. Supplementary material

The following are the Supplementary data to this article:

Supplementary data 1
mmc1.docx (10.6MB, docx)

Data availability

The authors do not have permission to share data.

References

  1. Banu K., Mondal B., Rai B., Monica N., Hanumegowda R. Prospects for the application of aptamer based assay platforms in pathogen detection. Biocybernetics and Biomedical Engineering. 2022;42(3):934–949. doi: 10.1016/j.bbe.2022.07.005. [DOI] [Google Scholar]
  2. Böger B., Surek M., Vilhena R.D.O., Fachi M.M., Junkert A.M., Santos J.M., Domingos E.L., Cobre A.D.F., Momade D.R., Pontarolo R. Occurrence of antibiotics and antibiotic resistant bacteria in subtropical urban rivers in Brazil. Journal of Hazardous Materials. 2021;402 doi: 10.1016/j.jhazmat.2020.123448. [DOI] [PubMed] [Google Scholar]
  3. Bu T., Jia P., Liu J., Liu Y., Sun X., Zhang M., Tian Y., Zhang D., Wang J., Wang L. Diversely positive-charged gold nanoparticles based biosensor: A label-free and sensitive tool for foodborne pathogen detection. Food Chemistry: X. 2019;3 doi: 10.1016/j.fochx.2019.100052. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Caldwell M., Hughes M., Wei F., Ngo C., Pascua R., Pugazhendhi A.S., Coathup M.J. Promising applications of D-amino acids in periprosthetic joint infection. Bone Research. 2023;11(1):14. doi: 10.1038/s41413-023-00254-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Chen S., Yang X., Fu S., Qin X., Yang T., Man C., Jiang Y. A novel AuNPs colorimetric sensor for sensitively detecting viable Salmonella typhimurium based on dual aptamers. Food Control. 2020;115 doi: 10.1016/j.foodcont.2020.107281. [DOI] [Google Scholar]
  6. Chen S., Zong X., Zheng J., Zhang J., Zhou M., Chen Q., Man C., Jiang Y. A colorimetric strategy based on aptamer-catalyzed hairpin assembly for the on-site detection of Salmonella typhimurium in milk. Foods. 2021;10(11):2539. doi: 10.3390/foods10112539. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Daruich De Souza C., Ribeiro Nogueira B., Rostelato M.E.C.M. Review of the methodologies used in the synthesis gold nanoparticles by chemical reduction. Journal of Alloys and Compounds. 2019;798:714–740. doi: 10.1016/j.jallcom.2019.05.153. [DOI] [Google Scholar]
  8. Dietvorst J., Vilaplana L., Uria N., Marco M.P., Muñoz-Berbel X. Current and near-future technologies for antibiotic susceptibility testing and resistant bacteria detection. Trends in Analytical Chemistry. 2020;127 doi: 10.1016/j.trac.2020.115891. [DOI] [Google Scholar]
  9. Fusco V., Chieffi D., Fanelli F., Logrieco A.F., Cho G., Kabisch J., Böhnlein C., Franz C.M.A.P. Microbial quality and safety of milk and milk products in the 21st century. Comprehensive Reviews in Food Science and Food Safety. 2020;19(4):2013–2049. doi: 10.1111/1541-4337.12568. [DOI] [PubMed] [Google Scholar]
  10. Gao X., Li M., Zhao M., Wang X., Wang S., Liu Y. Metabolism-triggered colorimetric sensor array for fingerprinting and antibiotic susceptibility testing of bacteria. Analytical Chemistry. 2022;94(19):6957–6966. doi: 10.1021/acs.analchem.1c05006. [DOI] [PubMed] [Google Scholar]
  11. Geleta G.S. A colorimetric aptasensor based on gold nanoparticles for detection of microbial toxins: An alternative approach to conventional methods. Analytical and Bioanalytical Chemistry. 2022;414(24):7103–7122. doi: 10.1007/s00216-022-04227-9. [DOI] [PubMed] [Google Scholar]
  12. Hameed S., Xie L., Ying Y. Conventional and emerging detection techniques for pathogenic bacteria in food science: A review. Trends in Food Science & Technology. 2018;81:61–73. doi: 10.1016/j.tifs.2018.05.020. [DOI] [Google Scholar]
  13. Hsu Y.P., Hall E., Booher G., Murphy B., Radkov A.D., Yablonowski J., Mulcahey C., Alvarez L., Cava F., Brun Y.V., Kuru E., VanNieuwenhze M.S. Fluorogenic d-amino acids enable real-time monitoring of peptidoglycan biosynthesis and high-throughput transpeptidation assays. Nature Chemistry. 2019;11(4):335–341. doi: 10.1038/s41557-019-0217-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Karo O., Wahl A., Nicol S.B., Brachert J., Lambrecht B., Spengler H.P., Nauwelaers F., Schmidt M., Schneider C.K., Müller T.H., Montag T. Bacteria detection by flow cytometry. Clinical Chemistry and Laboratory Medicine. 2008;46(7) doi: 10.1515/CCLM.2008.156. [DOI] [PubMed] [Google Scholar]
  15. Liu D.M., Dong C. Gold nanoparticles as colorimetric probes in food analysis: Progress and challenges. Food Chemistry. 2023;429 doi: 10.1016/j.foodchem.2023.136887. [DOI] [PubMed] [Google Scholar]
  16. Luo K., Ryu J., Seol I.H., Jeong K.B., You S.M., Kim Y.R. Paper-based radial chromatographic immunoassay for the detection of pathogenic bacteria in milk. ACS Applied Materials & Interfaces. 2019;11(50):46472–46478. doi: 10.1021/acsami.9b16075. [DOI] [PubMed] [Google Scholar]
  17. Machado S.G., Baglinière F., Marchand S., Van Coillie E., Vanetti M.C.D., De Block J., Heyndrickx M. The biodiversity of the microbiota producing heat-resistant enzymes responsible for spoilage in processed bovine milk and dairy products. Frontiers in Microbiology. 2017;8 doi: 10.3389/fmicb.2017.00302. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Machado S.G., Bazzolli D.M.S., Vanetti M.C.D. Development of a PCR method for detecting proteolytic psychrotrophic bacteria in raw milk. International Dairy Journal. 2013;29(1):8–14. doi: 10.1016/j.idairyj.2012.09.007. [DOI] [Google Scholar]
  19. Martín M.C., Del Rio B., Martínez N., Magadán A.H., Alvarez M.A. Fast real-time polymerase chain reaction for quantitative detection of Lactobacillus delbrueckii bacteriophages in milk. Food Microbiology. 2008;25(8):978–982. doi: 10.1016/j.fm.2008.07.004. [DOI] [PubMed] [Google Scholar]
  20. Meng L., Liu H., Lan T., Dong L., Hu H., Zhao S., Zhang Y., Zheng N., Wang J. Antibiotic resistance patterns of Pseudomonas spp. isolated from raw milk revealed by whole genome sequencing. Frontiers in Microbiology. 2020;11:1005. doi: 10.3389/fmicb.2020.01005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Peng H., Chen I.A. Rapid colorimetric detection of bacterial species through the capture of gold nanoparticles by chimeric phages. ACS Nano. 2019;acsnano.8b06395 doi: 10.1021/acsnano.8b06395. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Pundir C.S., Lata S., Narwal V. Biosensors for determination of D and L- amino acids: A review. Biosensors and Bioelectronics. 2018;117:373–384. doi: 10.1016/j.bios.2018.06.033. [DOI] [PubMed] [Google Scholar]
  23. Qi H., Feng L., Zhao S., Li H., Li F. Aptamer recognition-promoted specific intercalation of iridium complexes in G-quadruplex DNA for label-free and enzyme-free phosphorescence analysis of kanamycin. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy. 2023;284 doi: 10.1016/j.saa.2022.121758. [DOI] [PubMed] [Google Scholar]
  24. Quintieri F., Caputo. Antibiotic resistant Pseudomonas spp. spoilers in fresh dairy products: An underestimated risk and the control strategies. Foods. 2019;8(9):372. doi: 10.3390/foods8090372. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Ribeiro Júnior J.C., De Oliveira A.M., Silva F.D.G., Tamanini R., De Oliveira A.L.M., Beloti V. The main spoilage-related psychrotrophic bacteria in refrigerated raw milk. Journal of Dairy Science. 2018;101(1):75–83. doi: 10.3168/jds.2017-13069. [DOI] [PubMed] [Google Scholar]
  26. Salgado C.A., Baglinière F., Vanetti M.C.D. Spoilage potential of a heat-stable lipase produced by Serratia liquefaciens isolated from cold raw milk. LWT-Food Science and Technology. 2020;126 doi: 10.1016/j.lwt.2020.109289. [DOI] [Google Scholar]
  27. Sasabe J., Miyoshi Y., Rakoff-Nahoum S., Zhang T., Mita M., Davis B.M., Hamase K., Waldor M.K. Interplay between microbial d-amino acids and host d-amino acid oxidase modifies murine mucosal defence and gut microbiota. Nature Microbiology. 2016;1(10):16125. doi: 10.1038/nmicrobiol.2016.125. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Typas A., Banzhaf M., Gross C.A., Vollmer W. From the regulation of peptidoglycan synthesis to bacterial growth and morphology. Nature Reviews Microbiology. 2012;10(2):123–136. doi: 10.1038/nrmicro2677. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Vithanage N.R., Dissanayake M., Bolge G., Palombo E.A., Yeager T.R., Datta N. Biodiversity of culturable psychrotrophic microbiota in raw milk attributable to refrigeration conditions, seasonality and their spoilage potential. International Dairy Journal. 2016;57:80–90. doi: 10.1016/j.idairyj.2016.02.042. [DOI] [Google Scholar]
  30. Von Neubeck M., Baur C., Krewinkel M., Stoeckel M., Kranz B., Stressler T., Fischer L., Hinrichs J., Scherer S., Wenning M. Biodiversity of refrigerated raw milk microbiota and their enzymatic spoilage potential. International Journal of Food Microbiology. 2015;211:57–65. doi: 10.1016/j.ijfoodmicro.2015.07.001. [DOI] [PubMed] [Google Scholar]
  31. Wang M., Bai Z., Liu S., Liu Y., Wang Z., Zhou G., Gong X., Jiang Y., Sui Z. Accurate quantification of total bacteria in raw milk by flow cytometry using membrane potential as a key viability parameter. LWT-Food Science and Technology. 2023;173 doi: 10.1016/j.lwt.2022.114315. [DOI] [Google Scholar]
  32. Wang X., Yuan W., Sun Z., Liu F., Wang D. Ultrasensitive multicolor electrochromic sensor built on closed bipolar electrode: Application in the visual detection of Pseudomonas aeruginosa. Food Chemistry. 2023;403 doi: 10.1016/j.foodchem.2022.134240. [DOI] [PubMed] [Google Scholar]
  33. Wu Z., Huang C., Dong Y., Zhao B., Chen Y. Gold core @ platinum shell nanozyme-mediated magnetic relaxation switching DNA sensor for the detection of listeria monocytogenes in chicken samples. Food Control. 2022;137 doi: 10.1016/j.foodcont.2022.108916. [DOI] [Google Scholar]
  34. Yang X., Dang Y., Lou J., Shao H., Jiang X. D-alanyl-d-alanine-modified gold nanoparticles form a broad-spectrum sensor for bacteria. Theranostics. 2018;8(5):1449–1457. doi: 10.7150/thno.22540. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Ziyaina M., Rasco B., Sablani S.S. Rapid methods of microbial detection in dairy products. Food Control. 2020;110 doi: 10.1016/j.foodcont.2019.107008. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary data 1
mmc1.docx (10.6MB, docx)

Data Availability Statement

The authors do not have permission to share data.

Articles from Food Chemistry: X are provided here courtesy of Elsevier

RESOURCES