Differences and fingerprints of ESBL-producing E. coli from chicken faeces

Abstract

E. coli is an important source of β-lactam resistance. Extended Spectrum β-Lactamase (ESBL) producing E. coli and their phylogenetic transmission are an important problem for health. On the other hand, MALDI-TOF MS allows sensitive and specific applications in food safety and many clinical studies, and it is also a method approved by the Food and Drug Administration (FDA) for microbial identification. We thought that the protein phenotypic character of the bacteria might provide data on resistance. This study aims to investigate β-lactamases in E. coli and to examine the differences in protein properties of ESBL-producing bacteria. In the present study that is the first report for Duzce (Türkiye), 28.6% of 122 isolates were identified as Chicken Faeces E. coli (CFEC)-ESBL. bla_CTX−M, bla_CTX−M−1, bla_CTX−M−15, bla_SHV, bla_TEM, bla_OXA−10, bla_CIT and bla_MOX genes were determined with PCR and bla_CTX−M gene was detected with the highest rate (88.5%). At least one of the resistance genes was detected in the phenotype screening tests, except one of the isolates (CFEC-ESBL-90). On the other hand, CFEC-ESBL-38 contained bla_CTX−M−15 and the fact that this isolate was the only atypical ESBL strain with indole (−) and lac (−) characteristics among all isolates explains the highest variance (41%). It was also different from other Principle Component Analysis (PCA) components. Moreover, this isolate had a high degree of similarity (87%; CCI) with the other isolate (CFEC-ESBL-90), which had low similarity to CFEC-ESBLs. The study determined the differences and similarities between E. coli isolates with all PCA analyses. Our findings indicate that the ESBL group generally differed from susceptible strains, and the isolates had some heterogeneities and homogeneities. As a result, phyloproteomic analyses with MALDI-TOF MS are considered beneficial in characterising phenotypic bacterial behaviour. Despite that, there were some limitations in the study that need to be solved with further research.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-14114-9.

Introduction

The opportunistic pathogen Escherichia coli (E. coli) is a common member of the gut microbiota in humans and animals. It is is an important source of many antibiotic-resistance genes in the ecosystem¹.

Antimicrobial resistance (AMR) is a very important growing crisis worldwide. It is predicted by some experts that there will be 10 million deaths related to AMR every year in the world by 2050^2,3. Because E. coli is zoonotic and spreads easily in food-environment-human interaction, it leads to a potential change in the microbiome at the global level. It was reported that E. coli is important, especially in the spread of extended-spectrum β-lactamases (ESBL) and acquired AmpC β-lactamases^4,5. E. coli can hydrolyze almost all penicillin and cephalosporins via its ESBL encoding genes. The uninterrupted transmission occurs and these resistance genes can pass from one bacterium to another.In other words, the distribution of resistant bacteria is an ecological evolutionary process⁵. Understanding this can only be possible by examining the phylogenetic and phyloproteomic relationships between bacteria with various analytical methods. While genotypical methods like sequencing (whole genome or multilocus etc.) are expensive and laborious, analyses of proteins are relatively inexpensive and effortless and have significantly contributed to this field in recent years⁶. In this case Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry (MALDI-TOF MS) method is useful and has become widespread in the past ten years.

MALDI-TOF MS analyzes the proteins of bacteria with the help of validated databases created with reference spectra of standard strains. It allows sensitive and specific applications in food safety and many clinical studies, and it is also a method approved by the Food and Drug Administration for microbial identification^7,8. A comparative study conducted in 2010 reported that MALDI-TOF MS had a very high rate of identifying the bacteria in cases (99.1%) compared to routinely used biochemical methods⁹. Another study reported that Gram-negative bacteria were directly detected as a species in 99.2% of blood and urine samples¹⁰. MALDI-TOF MS is identified after comparing highly conserved ribosomal proteins of microorganisms with reference proteomic profiles of standard strains abundantly available in the database. At the same time, it is possible to perform phyloproteomic analyses when working with multiple isolates. In this respect, it is possible to compare fingerprints with unique mass spectra created for each bacterium by MALDI-TOF MS and then compare them. Thus, this series of analyses that can be performed with MALDI-TOF MS, besides microbial identification, contributes to an idea of similarities or differences in some processes operating in metabolism^8,11. For example, Suen et al. (2019) identified pathogenic Staphylococcus species from indoor samples with MALDI-TOF MS and analyzed the multi-drug resistance profiles of isolates with MALDI biotyper software¹². In another study, it was reported that Mycobacterium spp. identification takes 7–21 days after colony formation with traditional biochemical methods, but it can be accurately identified in as little as 1 h with MALDI-TOF MS¹³. Alharbi et al. (2021) identified S. aureus and coagulase-negative staphylococci at 100% from 400 samples, comparing MALDI-TOF MS with conventional methods¹⁴. There are other studies conducted on MALDI-TOF MS in which high rates of bacteria are identified from different sources (wastewater, rural areas, etc.)^15–17. Identifying Gram-negative bacteria of faecal origin from food and farm animals is increasingly preferred^18,19.

Previous studies have provided information that there are differences in the protein analysis of resistant and susceptible strains. In general, there are microorganisms with very different characteristics in poultry, and many studies have shown that MALDI-TOF MS can reliably identify these bacteria²⁰. The general concept of the current study is to investigate β-lactamases especially ESBL production in bacteria and characterize some of its properties. Therefore, this study aimed (i) to investigate the presence of some β-lactamases in E. coli obtained from the gut contents of broiler chickens and (ii) to characterize the isolates by phenotypic-proteomic analyses.

Materials and methods

Chicken feaces (CF) samples and chemicals

One hundred thirty chicken faeces samples were fresh from 4 different slaughterhouses in the city centre of Duzce (Türkiye) (in April 2022). Only stool samples from inside the gut were taken because the study aimed to obtain E. coli. Mac Conkey Agar (MAC) (Merck, Germany), Tryptic Soy Agar (TSA) (Merck, Germany) and Tryptic Soy Broth (TSB) (Condalab, Spain) were used for isolation and culture, and α-Cyano-4-hydroxycinnamic acid (CHCA; Bruker, Germany) was used as a MALDI-TOF MS matrix. Acetonitrile (ACN, HPLC grade; Sigma-Aldrich, Missouri, USA), trifluoroacetic acid (TFA; Sigma-Aldrich, Missouri, USA), 0.1 µm filtered ultrapure water (Sigma-Aldrich, Missouri, USA) free of DNAse and RNAse, and a Bruker Bacterial Test Solution (BTS) containing E. coli, RNAase and myoglobin protein profiles were also used.

Isolation of Escherichia coli

The faeces were inoculated directly into the medium with a sterile swab and incubated at 37 °C for 24 h. The identification was made by conventional methods (Gram stain, catalase, oxidase, oxidation/fermentation (OF), indole, methyl red (MR), Voges Proskauer (VP), citrate, urease, triple sugar iron, and H₂S)²¹. MAC medium was preferred in the initial isolation, while TSA was used in the secondary passages. The isolates’ ability to ferment simple sugars (inositol, lactose, xylose, and mannose) and hemolysis on an agar medium containing 5–10% defibrinated sheep blood were also recorded.

Phenotypic determination of β-lactamases

Β-lactam antibiotics were used for phenotypic determination of resistance. Aztreonam is a monobactam; while the others belong to 3rd generation of cephalosporins. All isolates were passaged into MAC mediums containing 1 mg/L cefotaxime, and growth was recorded²². Then, all isolates were subjected to double disk phenotype screening and confirmation test. The antibiotic susceptibility tests were performed according to the Kirby Bauer method recommended by the Clinical Laboratory Standards Institute²³. CLSI recommends that these disks screen β-lactam resistance in animal gram-negative bacteria²³.

Screening assays were used to determine the sensitivity of isolates to ceftazidime, cefpodoxime, cefotaxime, aztreonam and ceftriaxone²³. The presence of AmpC β-lactamases were determined by investigating the cefoxitin and cefepime activities on isolates. Cefoxitin-resistant (≤ 14 mm) and cefepime-susceptible (≥ 18 mm) isolates were considered phenotypically positive for AmpC. Ceftazidime (30 µg), cefotaxime (30 µg), aztrenoam (30 µg), ceftriaxone (30 µg), cefpodoxime (10 µg), ceftazidime-clavulanic acid (40 µg), cefotaxime-clavulanic acid (40 µg), cefoxitin (30 µg), cefepime (30 µg) discs were used. The evaluation of zone diameters was made according to CLSI directives, and the threshold values were determined according to CSLI 2018 and CTX: R ≤ 27, ATM: R ≤ 27, CPD: R ≤ 27, CAZ: R ≤ 22, CRO: R ≤ 25were accepted. Also, cefotaxime and ceftazidime zone diameters progressing more than 5 mm in zone diameters with clavulanic acid were evaluated as positive²³.

Investigation of β-lactamase genes

Genomic DNA was extracted using boiling method (boiling in distilled water at 95 °C for 10 min). After boiling, it was centrifuged at 12,000 rpm for 5 min and used as supernatant DNA. For β-lactamase, bla_CTX-M, bla_CTX-M-1, bla_CTX-M-15, bla_SHV, bla_TEM, bla_OXA-10, with conventional PCR, and bla_CIT, bla_{MOX genes} for AmpC β-lactamase were investigated with multiplex PCR. The primers used in PCR reactions are given in Table 1. Each PCR reaction was run in 35 cycles with a total volume of 25 µl. For each gene, PCR mix (K0171 Thermo Scientific) 10 pmol reverse and forward primers and PCR water were used. In each reaction, pre-denaturation at 95 °C for 5 min, denaturation at 95 °C for 30 s. bla _CTX-M at 57 °C, bla CTX- M-15 at 48 °C, bla_CTX-M-1 and bla _OXA-10 at 45 °C, bla_TEM at 44 °C, bla_SHV primer bonding temperature at 42 °C, 45 s synthesis at 72 °C and 7 min final synthesis at 72 °C. The primers for bla _CIT and bla_MOX were added in half and, the same PCR cycle was run with an annealing temperature of 53 °C. E. coli NCTC 13,461-NCTC 13,462-NCTC 13,463, E. coli ATCC 35,218, and Klebsiella pneumoniae ATCC 700603 strains were used as positive controls.

Table 1.

Oligonucleotides used in the study for PCR.

Gene	5′–3′	References
blaCTX-M	F-SCSATGTGCAGYACCAGTAA R-CCGCRATATGRTTGGTGGTG	24
blaCTX-M-1	F-AAAAATCACTGCGCCAGTTC R-AGCTTATTCATCGCCACGTT	25
blaCTX-M-15	F-TGG GGG ATA AAA CCG GCA G R-GCG ATA TCG TTG GTG GTG C	26
blaSHV	F-CTTTACTCGCTTTATCG R-TCCCGCAGATAAATCACCA	27
blaOXA-10	F-GTCTTTCGAGTACGGCATTA R-ATTTTCTTAGCGGCAACTTAC	28
blaTEM	F-ATGAGTATTCAACATTTCCG R-CCAATGCTTAATCAGTGAGC	29
blaCIT	F-TGGCCAGAACTGACAGGCAAA R-TTTCTCCTGAACGTGGCTGGC	30
blaMOX	F-GCTGCTCAAGGAGCACAGGAT R-CACATTGACATAGGTGTGGTGC	30

Identification of bacterial colonies from chicken faeces by MALDI-TOF MS

The updated IVD database containing 10,694 MSPs (Bruker Daltonics, respectively) was applied to identify the peptide and protein spectra. For microbial biomass analysis using the MALDI-TOF MS method, a single colony was placed onto a special steel 96 micro scout plate (MSP) (Bruker Daltonics) and spread onto the wells in the plate as a thin film. After drying, 1 µL CHCA matrix solution (12.5 mg/mL CHCA in a 50% ACN and 2.5% TFA mixture) was added and allowed to dry completely at room temperature. The MALDI 96 MSP was placed in the MALDI-TOF MS Device, and the system was operated using the optimised method for identifying micro-organisms in linear positive ion mode at a 2.000–20.000 Dalton (Da) mass range. As the ion source, a 60 Hz nitrogen laser was used at 337 nm. The laser pulses consisting of 40 packets of 240 were applied in the measurement of each colony to obtain the spectra. Each sample was studied in triplicate, and the highest readings were included in the analysis. The internal quality control for MALDI-TOF MS in general bacteriology is partly achieved by using a Bruker BTS, consisting of an extract of E. coli proteins for mass calibration of the instrument³¹. Mass spectrum calibration was completed with seven peaks in the present study (m/z, 5095.39141 Da; 5381.28948 Da; 6265.88537 Da; 7254.94790 Da; 10,289.99287 Da; 13,692.32900 Da and 16,962.67711 Da) assigned with a standard deviation of 58.52 ppm and maximum peak error of 78.19 ppm.

The use of principle component analysis (PCA) in MALDI-TOF MS

The spectra were analysed using Bruker Daltonics MALDI Biotyper Flex Analysis version 3.4 automation-controlled Biotyper Compass Explorer 1.4 software and the MALDI Biotyper 3.1 database. The identification score criteria were applied following the manufacturer’s recommendations (Bruker). MALDI-TOF MS biotyping analysis elicits the sample’s characteristic mass and peak density distribution of ribosomal 16S proteins. Since this mass spectrum is species-specific for many micro-organisms, it represents a “molecular fingerprint”³². The spectra were massed using the PCA method supported by external MATLAB software integrated into the MALDI Biotyper.

Based on the unique peptide and protein peaks within each spectrum, PCA helped create clustered spectra groups with similar variational properties and visualise their differences. With phyloproteomic-PCA, the data were given on a three-dimensional (3D) coordinate system, and the dimensionality of the data set was reduced, preserving the original information. Optimised preliminary procedures (correction method: Savitski-Golay; subtraction method: multi-polygon; normalisation method) were applied for each spectrum to increase the speed of the analysis and reduce the size of the data body³³. The variance among the bacteria was automatically calculated with software support. In addition, virtual gel images (VGI) containing the projection of the peaks within the bacteria spectra were created. Vertical traces of VGI ranging from red to light blue corresponded to each peak within the spectrum and were given by a colour scale ranging from low relative abundance (light green) to high relative abundance (red). PCA dendrograms and 3D or 2D scatter plots representing each spectrum’s relationship and closeness were created for cluster analyses³⁴. Finally, the similarity (proximity) and difference (distance) relationships of each bacteria to the others, whose composite correlation index (CCI) was calculated statistically using the software, were determined.

Results

The identification of bacteria from chicken faeces samples by MALDI-TOF MS

A total of 124 bacteria were obtained, and 122 of them were identified as E. coli. They were coded and numbered as CFEC (Chicken Faeces E.coli), and these numbers were adhered to during the entire data evaluation process (Supplementary Table S1). The other two isolates were chicken faeces (CF), abbreviated with initials (Klebsiella oxytoca: CFKO; and Enterobacter kobei: CFEBC). CFKO and CFEBC isolates were not included in the phyloproteomic analysis.

The presence of ESBL and AmpC in CFEC bacteria

As a result of the phenotypic survey, 28.6% (n = 35) of 122 E. coli was detected and identified as a CFEC-ESBL E. coli Except for one isolate, a positive genotypic result was found. Also, AmpC was not detected in any of the isolates (0%). Genotypic β-lactamases were investigated via PCR for phenotypically ESBL-positive isolates (Table 2). In this respect, the bla_SHV gene was not detected in any of the 35 isolates, and the bla_CTX-M gene was detected at the highest rate (88.5%) (n = 31). It was found that 3 of the ESBL-producing isolates contained 4 of the screened resistance genes. In another four isolates, three genes were screened, and two different genes were detected in 21 isolates. While only one of the genes was detected in 6 isolates, none of the tested genes were detected in one isolate.

Table 2.

Rate of resistance genes for β-lactamases type of E. coli.

ESBL n = 35
blaCTX-M	blaCTX-M-1	blaCTX-M-15	blaSHV	blaTEM	blaOXA-10
31 (88.5%)	23 (65.7%)	1 (2.8%)	0	9 (25.7%)	14 (40%)

AmpC n = 35 (%)
blaCIT			blaMOX
0			0

bla_OXA-10 is not directly an ESBL gene but still it was investigated for all isolates because carbapenemases are also a type of β-lactamases. One or more of the other genes were detected in all isolates that contained the bla_OXA-10. Similarly, no isolates containing only one bla_TEM were detected. In contrast, all isolates containing the bla_TEM gene and the bla_TEM were detected as one or more of the CTX-Ms or bla_OXA-10 genes. However, phenotypic negative results, bla_CIT and bla_MOX genes were also investigated in ESBL-positive isolates and were not detected in any isolates. All results are presented in Table 3. Table 3 gives the genotypic results of ESBL-producing strains that phenotypically. Some biochemical features (such as urease, methyl red, H₂S, etc.) were not presented because they had typical E.coli results. The coding of ESBL detected E.coli is the same as the numbering in CFEC, and CFEC-ESBL is abbreviated by adding ESBL in front of this abbreviation.

Table 3.

Biochemical and genetic features of ESBL- E.coli isolates.

Sample code for ESBL- E.coli	Biochemical characteristics				β-lactam resistance genes
	Indole	Lactose	Catalase	Hemolysis	blaCTX-M	blaCTX-M-1	blaCTX-M-15	blaTEM	blaOXA-10
CFEC-ESBL-1	+	+	W	α	+	+	−	−	−
CFEC-ESBL-7	+	+	W	α	+	+	−	−	−
CFEC-ESBL-9	+	+	S	α	+	+	−	−	−
CFEC-ESB-14	+	+	S	α	+	+	−	−	−
CFEC-ESBL-16	+	+	W	α	+	+	−	−	−
CFEC-ESBL-18	+	+	W	a	+	+	−	−	−
CFEC-ESBL-22	+	+	S	a	−	−	−	−	−
CFEC-ESBL-27	+	+	S	a	+	+	−	−	−
CFEC-ESBL-28	+	+	W	b	+	+	−	−	−
CFEC-ESBL-38	−	-	St	α /−*	+	−	−	−	−
CFEC-ESBL-40	+	+	St	a	+	−	−	−	−
CFEC-ESBL-41	+	+	St	a	+	+	−	−	−
CFEC-ESBL-43	+	+	S	a	+	+	−	−	+
CFEC-ESBL-45	+	+	S	a	+	+	−	−	+
CFEC-ESBL-47	+	+	S	a	+	+	−	+	+
CFEC-ESBL-51	+	+	S	a	+	−	−	+	−
CFEC-ESBL-55	+	+	St	a	+	−	−	+	−
CFEC-ESBL-62	+	+	St	−	+	+	−	−	−
CFEC-ESBL-67	+	+	St	a	+	+	−	−	−
CFEC-ESBL-68	+	+	S	a	+	+	−	−	−
CFEC-ESBL-69	−	+	W	−	+	+	−	−	−
CFEC-ESBL-70	+	+	S	a	+	+	−	−	−
CFEC-ESBL-73	+	+	S	a	+	+	−	−	−
CFEC-ESBL-77	+	+	S	β	+	+	−	+	+
CFEC-ESBL-79	+	+	W	α	+	−	−	+	−
CFEC-ESBL-80	+	+	S	α	−	−	−	+	+
CFEC-ESBL-85	+	+	S	α	+	+	−	+	+
CFEC-ESBL-89	+	+	S	α	+	+	−	−	−
CFEC-ESBL-90	–	+	W	-	−	−	−	−	−
CFEC-ESBL-93	+	+	W	β	+	−	−	−	+
CFEC-ESBL-105	+	+	St	β	+	−	−	−	−
CFEC-ESBL-106	+	+	St	β	+	−	−	−	−
CFEC-ESBL-107	+	+	St	β	+	−	−	−	+
CFEC-ESBL-108	+	–	S	α	−	−	−	−	+
CFEC-ESBL-110	+	+	S	α	+	−	−	−	+

Other biochemical features (such as urease, methyl red, H₂S, etc.) were not presented in the table because they are the same for each bacterium.

Also, since bla_SHV, bla_CIT, and bla_MOX were not detected in any isolate, they were not given in the table.

*Bacterium has shown variable hemolysis. W: Weak; S: Strong.

The use of PCA in MALDI-TOF MS for chicken faeces E. coli

A general PCA analysis (I. Phyloproteomic study) of 122 CFEC isolates was performed, and a dendrogram profile and a 2D scattering profile were formed (Fig. 1). Also, the variance values were automatically calculated and shown on the dendrogram in Fig. 1A. With these three analyses (dendrogram, scattering profile, and variance), rough preliminary information was obtained according to the cluster formation of the most distant and the farthest and closest among 35 isolates among 122 CFECs and the scattering of these isolates. These two profiles and variance values were evaluated together. As seen in Figs. 1A and B, seven CFEC isolates far from the main cluster, as they have some differences, are separated from other CFECs in the dendrogram profile and settled into separate clusters. For example, CFEC-38 (CFEC-ESBL-38), CFEC-90 (CFEC-ESBL-90), and CFEC-19 isolates with the highest variance (PC1) of 31% in this analysis were placed on the far right in a separate cluster. Also, three E.coli CFEC-108, CFEC-69, and CFEC-35, are separated by a separate line, although they appear to be part of the larger cluster by the second variance value (PC2: 14%) (Fig. 1A and B). Consistent with the results in the dendrogram, the scattering profile in Fig. 1C also shows that these seven isolates scatter far from the large cluster. In summary, ESBL-producing strains are separated from others. However, those with some atypical features in this group show a distant clonal relationship according to the protein grouping made.

Fig. 1.

Dendrogram profile of 122 CFEC isolates (A and B) and 2D scattering profile (C).

Inspection of biochemical results based on PCA results

In this part, the biochemical test results of some of the CFECs that were separated from the large cluster according to the cluster commonality, scattering profile, and variance values according to the MALDI-TOF MS-based PCA results were examined, and it was seen that there were data compatible with the PCA analysis results. In this respect, for example, the indole test of two E.coli (CFEC-38, CFEC-90 far right) separated by 31% variance and CFEC-108 (PC2; 14% variance) were negative, and the remaining 119 E. coli were positive. On the other hand, one of the isolates separated from the large cluster with 14% variance (PC2), CFEC-108, was alpha-hemolytic, although there was no hemolysis in the remaining strains CFEC-71, CFEC-69, and CFEC-35.

MALDI-TOF MS spectral analysis of CFEC- ESBL bacteria

Although general information was obtained first with PCA analyses of all 122 CFEC bacteria, the study focused on comparing the results of MALDI-TOF MS-based PCA analysis of 35 E. coli bacteria with ESBL detected with other phenotypic methods. A representative mass spectrum (CFEC-ESBL-68) showing ESBL characteristics and gel images (virtual gel images; VGI) of 35 CFEC-ESBLs are given in Fig. 2. In this respect, when the mass spectrum of the CFEC-ESBL-68 isolate in Fig. 2A, and the VGIs of all CFEC-ESBLs are examined, especially the high abundance (intensity) peptide (m/z; 5107 Da, 6270 Da, 7290 Da, 9085 Da) and 9762 Da) protein (m/z; 10,489 Da) projections appear to be nearly identical for each. In this context, there is no major difference when looked at roughly. However, when detailed analyses of all components of PCA analyses were made (dendrogram, 2D/3D scatter plotting, variance, composite correlation index, etc.), it was determined that there were significant differences between these E. coli with ESBL characteristics.

Fig. 2.

(A) Representative mass spectrometry of CFEC-ESBL-68 and (B) Virtual gel profile of 35 CFEC-ESBL isolates.

Clusters, scatters, and composite correlation index analysis of ESBL-producing E. coli

At this step, with the support of the MATLAB program, the unique spectra, which are the fingerprints of each bacterium, were compared and analyzed (II. Phyloproteomic study), and the data are given separately in Fig. 3. First, the dendrogram (Fig. 3A) and 2D scattering profiles (Fig. 3B) generated for 35 CFEC-ESBLs together with a total of 10 variance values are shown (Fig. 3A).

Fig. 3.

The principal component analysis of the total 35 CFEC-ESBL isolates and comparison of both genetic and biochemical analysis results of them. (A) Dendrogram and variance analysis (B) 2D Scatter plotting (C) The total CCI % value corresponds to each isolate ( The -axis is the total CCI% value which includes itself and is calculated for each isolate and has a projection of each isolate of those values on the x-axis) (D) The color matrix of the total CFEC-ESBL isolates, (E) The color matrix and CCI % value of three isolates (CFEC-ESBL-38, CFEC-ESBL-68, and CFEC-ESBL-90.

As a result of the dendrogram analysis of isolates showing only ESBL characteristics, only CFEC-ESBL-38 (CFEC-38) took its place on a separate line on the far right (Fig. 3A), and it was located farthest (yellow dot) in the 2D scattering profile. It was found that this isolate differed from 34 CFEC-ESBLs with a variance of 41%. On the other hand, the second isolate, which is separated from the large cluster (n = 33) with a 15% variance value, was also CFEC-ESBL-90.

The composites for calculating the distance and proximity indices of CFEC-ESBL CCI were also calculated automatically with the MATLAB program (data not shown). In the graphic in Fig. 3C, the total value of CCI % of 34 isolates with itself, calculated for 35 CFEC-ESBL and each isolate, is on the y-axis, and each isolate itself is on the x-axis. As the closeness to the isolates within the group increases, the total %CCI value on the y-axis increases for each isolate. On the other hand, as the proximity value decreases, the total %CCI value also decreases. In this respect, when Fig. 3C is examined, it is seen that the lowest total CCI percentage value belongs to CFEC-ESBL-38. Also, it was found that the total %CCI value of CFEC-ESBL-90 was lower than the other group members. It is a result that is highly compatible with dendrogram, variance, and 2D results. According to 33 isolates of CFEC-ESBL, CFEC-ESBL38 is very different, and CFEC-ESBL-90 is different.

Another isolate that drew attention in 3C in the figure is CFEC-ESBL-93. It can be seen that the total per cent CCI value of this isolate is low. However, the variance value is 1% and is nested in the main cluster with 32 other CFEC-ESBLs in the dendrogram and scattering profile.

The CCI color matrix of all ESBL-CFEC isolates is presented in Fig. 3D. In this colour matrix, the dark red colour represents the highest CCI% value (each isolate has 100% similarity to itself, and this colour is dark red), and the dark blue colour corresponds to the lowest (0–1%) CCI value. If we take an overview of this, green tiles in the CCI colour matrix correspond to 40% to 59%, and yellow to dark red tiles correspond to a CCI index of 60% to 100%. It indicates that the light blue to dark blue boxes also have a CCI index from 39 to 0%.

The colour matrices of two isolates with low affinity (CFEC-ESBL-38, CFEC-ESBL-90) and an isolate with the highest affinity (CFEC-ESBL-68) among 35 CFEC-ESBL are given in Fig. 3E by matching their CCI% values. In this respect, it was found that the boxes corresponding to CFEC-ESBL-38 in the colour matrix were mostly in blue tones, and the %CCI values were low in proportion to this. In contrast, the box corresponding to CFEC-ESBL-90 appears light red (87% CCI). Also, the closeness of this isolate to CFEC-ESBL-41 and CFEC-ESBL-69 isolates, which are members of the large cluster, was determined to be 60% and 54%, respectively, in the dendrogram. On the other hand, it was reported to have the lowest affinity value (1%) against CFEC-ESBL-9, CFEC-ESBL-106, and CFEC-ESBL-110 isolates, among other isolates (Fig. 3E). It is seen that CFEC-ESBL-38 has the lowest overall CCI % value since it has the lowest affinity values to other CFEC-ESBLs except for a few isolates.

It was found that the second different isolate, CFEC-ESBL-90, has the highest affinity (87%) against CFEC-ESBL-38 isolate. Like CFEC-ESBL-38, the affinity index to CFEC-ESBL-41 and CFEC-ESBL-69 isolates is 64% and 62%, respectively. Also, the affinity index values for some isolates (n = 6) in the large cluster are over 50%, while the affinity indexes for 14 group members are between 30%-40%. This explains the lower variance value compared to CFEC-ESBL-38 (Fig. 3 E).

As expected, mostly yellow- to dark-red boxes in the CCI colour matrix of CFEC-ESBL-68, which has the highest percentage of closeness in total, draw attention. When we look at the colour matrix in general, most of the CFEC-ESBL isolates that have a low variance value (≤ 5%) in the large cluster in the dendrogram (Fig. 3A) and are very close to each other in the scattering profile (Fig. 3B) have plenty of coloured boxes from yellow to red on the colour scale (Fig. 3D).

Compatibility of biochemical and gene analyses and PCA results of some CFEC-ESBL

These results from MALDI-TOF MS-based PCA analyses pointed to important clues that some of the 35 ESBL-CFEC isolates had significant differences. In this respect, the biochemical characteristics of the isolates with the highest affinity (CFEC-ESBL-68) and the lowest (CFEC-ESBL-38) and low affinity (CFEC-ESBL-90) to all isolates and findings for β-lactam resistance genes were detected. Unlike the other 34 isolates, bla_CTX-M-15 is one of the β-lactam resistance genes in only CFEC-ESBL-38 isolates. It is considered that the presence of 41% variance causes the scattering profile to be located farthest in the scattering profile and with a different line in the dendrogram. Besides, in both (CFEC-ESBL-38 and CFEC-ESBL-90) isolates, the absence of bla_CTX-M-1, bla_TEM, bla_OXA-10 genes may be a reason for the 87% affinity between them, as well as leading to their separation from the main cluster (n = 33).

Discussion

The most suitable bacterium for monitoring ESBL production is E. coli. In the present study, 122 isolates were identified as E. coli by traditional (biochemical tests) methods and MALDI-TOF MS. Within the scope of phyloproteomic analysis, firstly, PCA analyses of all E. coli isolates were carried out, and it was determined that there are isolates with different characteristics. Then, more detailed analyses were conducted on a smaller group of E. coli with ESBL characteristics. With the support of all the analyses performed, it was concluded that there were significant differences in the light of the data on variance, CCI index values, dendrogram, and scattering profile placements, even though hints on bacteria belonging to the same species are provided.

The present study determined the differences and similarities between E. coli isolates with all PCA analyses (Dendrogram, scatter plotting, variance, and CCI). It was described that the ESBL group generally differed from susceptible strains, and the isolates had some heterogeneities and homogeneities. In conclusion, phyloproteomic analyses with MALDI-TOF MS may be useful for characterising phenotypic behaviours. The research run mostly by MALDI-TOF MS. Some conventional tests were performed to understand the differences in proteomic groups. The aim was to compare between them and it was seen that the two methods were parallel.

As a result of the investigation of some genes and the browsing of ESBL disk tests suggested by CLSI²³, β-lactamase was detected in almost one of the three isolates. It is a very high rate, especially according to the scale of industrial production facilities. E. coli is a very important strain regarding food pathogens and antibiotic resistance profile.For example, Yang et al.³⁵ reported that 22.9% of the bacteria were ESBL- E. coli. In other studies, Badr et al.³⁶ determined the rate of ESBL-producing E. coli isolated from chickens is 46.7%, while Gazal et al.³⁷ detected 66%, and Fournier et al.³⁸ detected ESBL 84%. Golpasand et al.³⁹ was found 30 strains from 108 isolates (27.8%) were detected as ESBL producing E. coli. Cormier et al.¹ reported that this rate reached 90%. The present study found that predominantly bla_CTX-M-1 resistance genes were detected. In previous studies, the most frequently reported resistance genes were CTX-Ms and their derivatives⁴⁰. These genes, which are reported to be common on a global scale, may indicate that bacteria are in contact with each other because the same resistance genes are often found in bacteria from very different sources. It is also known that bacteria transfer resistance genes by coming into contact with each other^2,36.

As mentioned in the first figure in the current study, ESBL-producing strains were included in a different cluster from non-ESBL-producing strains. In addition, it is observed that strains showing atypical biochemical properties are excluded in the ESBL-producing group. For example CFEC-ESBL-38 is a strain that came to the fore with its difference in the study as the only atypical ESBL strain that shows only indole (-) and lac (-) among 122 isolates. E. coli is lac ( +) commonly, and lactose permease enzyme (LacY protein) is a very important protein that enables the use of lactose in E. coli. However, in some bacteria, lac (-) variants occur due to deficiencies in the level of this enzyme encoded by this LacY gene^41,42. Stępień-Pyśniak et al.⁸ showed that Enterococcus faecalis and E. mundtii isolates were separated in the dendrogram with phyloproteomic analysis using the isolates’ spectral profiles. In the same study, it was also noted that there was clustering of very similar strains in terms of phenotype and genotype according to galactosidase and mellobiose characteristics, and it was stated in the study that a single gelatinase negative isolate gave a different peak. Unlike the other 34 CFEC-ESBL isolates, it is thought that the presence of blaCTX-M-15 only in CFEC-ESBL-38 causes it to settle furthest in the scattering profile and with a different line in the dendogram with a variance of 41%. The expression of a peptide/protein directly or indirectly related to phenotypic resistance might cause a difference in this strain^43–45.

In the I. In the phyloproteomic study, the CFEC-ESBL-90 was completely excluded from the large cluster with a 14% variance in the full dendrogram (n = 122, for CFEC) profile. This isolate had positive results phenotypically unlike genotypically in the case of β-lactamases. In comparison, it was located on a separate line in the dendrogram connected to the large cluster with a 15% variance in the II. Phyloproteomic study (n = 35, for CFEC-ESBL). Additionally, it was determined that the closeness ratios (CCI) to large cluster members, except for a few, were at a low level. Biochemically, this strain had weak catalase ability. The presence of catalase enzyme is very characteristic of E. coli. E. coli harbours two catalase genes: katG encodes hydroperoxidase I (HPI), and katE encodes HPII.

The activity of both catalases increases when both are present together, and the expression level of genes increases⁴⁶. In the dendrogram profile, the difference in common with the cluster members, including CFEC-ESBL-90 (CFEC-19 and CFEC-ESBL-38), is that it does not have hemolysis activity. The presence of the hemolysis enzyme and the observed hemolysis ability are mostly variable within the species in E. coli^47,48. On the other hand, another possible feature that makes this strain different from others is that the indole test is negative. Because the indole test is an indicator of the tryptophanase and tryptophan permease enzymes of the tryptophanase operon (TNA Operon), and the indole test of E. coli (90–95%) is mostly positive⁴⁹. Deficiency of these proteins was also determined as a possibility affecting all PCA analysis results. Torres-Corral and Santos⁵⁰ pointed out that Lactococcus garvieae isolates gave three characteristic peaks and as the reason for grouping of isolates, enzymes such as epimerase, methyltransferases, and acetyl phosphatases possessed by the isolates. Some studies suggest that changes in enzyme structures may affect results because protein analyses of biological structures are performed with MALDI-TOF MS^51,52. On the other hand, according to the present study, catalase, tryptophanase, and tryptophan permease enzymes of E. coli may be the reasons for the differences, but further studies must support them because bacteria produce many specific and non-specific proteins⁵³.

Besides, other genes (which were not tested in the study) are considered responsible for ESBL in CFEC-ESBL-90. Most genes responsible for ESBL appear as CTX-M, SHV, and TEM variants. However, other genes are also responsible for resistance^36,54–56. In the study of Laudy et al.⁵⁷, similar to the present study, although ESBL-producing Pseudomonas aeruginosa strains were obtained from phenotypic test results, they could not detect all the genes it screened at the same rate. They also reported three new ESBL-producing genes with their further studies. Because many genes responsible for β-lactamases were reported and continue to be reported^2,36,57, further studies can be carried out to detect ESBL genes in CFEC-ESBL-90.

In the present study, as well as the detection of differences between E.coli and ESBL-E.coli, it was observed that similar ones were numerically higher. For example, the greatest affinity was detected in the CFEC-ESBL-68 isolate. Although the ratio of CFEC-ESBL-68 to 13 cluster members was 49–69%, the closeness ratio to the remaining 21 cluster members was 70–99%. CFEC-ESBL-68 has typical biochemical features for E.coli, like most isolates in the study (strong catalase property, alpha hemolysis ability, indole and lactose positive, etc.). However, further tests (other simple sugar fermentation, gelatinase, nitrate, arginine, biofilm, multidrug resistance profile, etc.) are needed to understand.

The present study determined the differences and similarities between E. coli isolates with all PCA analyses (Dendrogram, scatter plotting, variance, and CCI). It was also found that the ESBL group generally differed from susceptible strains, and the isolates had some heterogeneities and homogeneities. Alharbi et al.¹⁴ showed that dendrogram cluster analysis can separate MSSA and MRSA. Similarly, in another study, it was shown that peaks of different masses distinguished MSSA and MRSA, and it was emphasized that MALDI TOF MS saves time according to molecular studies⁵⁸. Alegria et al.⁵⁹ indicated that ESBL-producing strain produced peaks at 370 and 414 m/z, while non-ESBL producer gave peaks at 456 and 458 m/z. All these data suggest that ESBL-producing E. coli are phylogenetically separated and may differ greatly in natural ecosystems. Previous studies have shown the feasibility of MALDI-TOF–MS for the clonal identification of bacteria. Our study is also an example of that. Moreover, this reseach accentuates the results of Matsumura and Ikegaya⁶⁰ who stated that MALDI-TOF assay is an excellent tool for quickly classifying ESBL-producing strains of foodborne Enterobacteriaceae. This method is considered as a fast and safe method in biological analysis⁶¹.

Conclusion

Phyloproteomic analyses with MALDI-TOF MS may be useful for characterising phenotypic behaviours. As it is known, a Whole Genome Analysis is a major challenge for analytical studies but at the same time it is a high-cost method. Thus, the most important result found in the present study is that performing advanced analyses and identification with the less costly MALDI-TOF MS contributes significantly to the validation of traditional analysis results. This is an important output of the study. By further developing the existing methodology, important problems can be solved in a shorter time. However MALDI-TOF MS is not enough for describing the isolates that are completely excluded from the large cluster. Future investigations are needed to solve this matter. This is also the limiting aspect of the study. In addition, the fact that a single sample was made in the study is also a limiting factor for the findings. On the other hand, this study represents a first in terms of ESBL screening and characterisation in broiler chickens for the region (Duzce, Türkiye). No previous study has been conducted or published in the region. In recent years, epidemiological studies have focused on the spreading of resistant strains, which are extremely important in clinical and food safety. In this sense, it is anticipated that this study will contribute to monitoring the data in the region. Although important clues were obtained, further analyses are planned to make sense of the effect of biochemical characteristics on variance values.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

CFEC

Chicken faeces E. coli

ESBL

Extended spectrum β-lactamase

PCA

Principal component analysis

CCI

Composit corelation index

CFKO

Chicken faeces Klebsiella oxytoca

CHCA

Cyano-4-hydroxycinnamic acid

ACN

Acetonitrile

TFA

Trifluoroacetic acid

MSP

Micro scout plate (MSP

CFEBC

Chicken faeces Enterobacter kobei

Author contributions

N.S. Conceptualization, Investigation, Methodology, Data curation, Visualization and Writing – original draft and editing.Y.N.C. Investigation, Methodology, Data curation, Visualization and Writing – original draft.

Funding

This research received no specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability

All data generated or analysed during this study are included in this published article.

Declarations

Competing interests

The authors declare no competing interests.