Advertisement

Analytical and Bioanalytical Chemistry

, Volume 410, Issue 22, pp 5445–5454 | Cite as

Rapid detection and differentiation of Staphylococcus colonies using an optical scattering technology

  • Tawfiq S. Alsulami
  • Xingyue Zhu
  • Maha Usama Abdelhaseib
  • Atul K. Singh
  • Arun K. Bhunia
Communication
Part of the following topical collections:
  1. Food Safety Analysis

Abstract

Staphylococcus species are a major pathogen responsible for nosocomial infections and foodborne illnesses. We applied a laser-based BARDOT (bacterial rapid detection using optical scattering technology) for rapid colony screening and detection of Staphylococcus on an agar plate and differentiate these from non-Staphylococcus spp. Among the six growth media tested, phenol red mannitol agar (PRMA) was found most suitable for building the Staphylococcus species scatter image libraries. Scatter image library for Staphylococcus species gave a high positive predictive value (PPV 87.5–100%) when tested against known laboratory strains of Staphylococcus spp., while the PPV against non-Staphylococcus spp. was 0–38%. A total of nine naturally contaminated bovine raw milk and ready-to-eat chicken salad samples were tested, and BARDOT detected Staphylococcus including Staphylococcus aureus with 80–100% PPV. Forty-five BARDOT-identified bacterial isolates from naturally contaminated foods were further confirmed by tuf and nuc gene-specific PCR and 16S rRNA gene sequence. This label-free, non-invasive on-plate colony screening technology can be adopted by the food industries, biotechnology companies, and public health laboratories for Staphylococcus species detection including S. aureus from various samples for food safety and public health management.

Graphical abstract

Keywords

Optical sensor BARDOT Staphylococcus Food sample 16S RNA gene 

Introduction

Staphylococcus species are a major human and veterinary pathogen responsible for nosocomial infections and foodborne illnesses. Antibiotic-resistant Staphylococcus also presents a serious public health concern. Staphylococcus genus includes over 30 species and subspecies, which are Gram-positive, non-motile, and non-spore-forming bacteria. Staphylococcus species are differentiated based on their ability to produce coagulase, which converts fibrinogen (soluble) to fibrin (insoluble) in the plasma. Coagulase-positive species are Staphylococcus aureus and Staphylococcus hyicus, and the coagulase-negative species are Staphylococcus lentus, Staphylococcus epidermidis, and Staphylococcus xylosus. The most pathogenic ones are S. aureus, S. epidermidis, and Staphylococcus saprophyticus. S. aureus causes an estimated 241,000 foodborne illnesses annually in the USA, an estimate that probably does not account for sporadic cases. Therefore, advanced methods should be exploited for rapid detection of Staphylococcus species including S. aureus in food samples to avoid the risk of food contamination, foodborne outbreaks, and financial crisis to food industries.

Culturing methods are invaluable for detecting and identifying staphylococci and micrococci that involve biochemical and physiological characteristics analysis for differentiation and classification [1]. However, such methods are arduous and may lead to inconclusive differentiation at the genus level due to high similarity in housekeeping and metabolic enzymes. On the other hand, PCR method amplifies a single copy of a target gene in an hour but fails to differentiate live and dead cells, and its performance is susceptible to inhibitors. Thus, a culture enrichment step is employed for detecting live pathogens while diluting away the inhibitors. Advanced screening technologies and methods should be explored that are operated independent of probes or primers, are resistant to inhibitors, and can yield results relatively quickly [2].

In this study, we employed a laser-based light scattering sensor, BARDOT (bacterial rapid detection using optical scattering technology) that was developed in our laboratory at Purdue University, to detect and identify bacterial colonies on an agar plate [3, 4]. BARDOT is able to detect and identify a single colony among a mixture of bacterial colonies based on the unique scatter signature, which is produced when a red-diode laser beam (635 nm) passes through the center of the colony of about 1 mm diameter [5]. The accuracy of bacterial identification depends on a robust scatter image library and image analysis [6] which account for colony texture, also referred to as surface roughness, a common feature in morphological description [7]. Texture-based analysis has been used for classification of the bacterial colony [8]. BARDOT has been used for detection of multiple bacterial pathogens including Listeria spp. [3, 9], Salmonella enterica [5, 10], Shiga-toxin producing Escherichia coli [11], Bacillus spp. [12], and many others. Additionally, BARDOT has been used to study streptomycin-induced antibiotic stress response in Salmonella enterica serovar Typhimurium [13] and in the differentiation of virulence gene-associated mutant colonies of Listeria monocytogenes [14].

BARDOT was used as a potential rapid colony screening tool to test the bovine raw milk supplies and ready-to-eat chicken salad for the presence of Staphylococcus including S. aureus colonies on phenol red mannitol agar (PRMA) plate. The BARDOT-identified colonies were further confirmed by PCR and 16S rRNA gene sequencing. BARDOT results were further validated in parallel by analyzing samples using the US Food and Drug Administration (FDA) method for raw milk testing [15]. Overall, results from this study support a potential use of BARDOT in detecting Staphylococcus species in food.

Materials and methods

Bacterial strains and growth media

Bacterial strains (Table 1) were grown either by inoculating a loop full of frozen culture (− 80 °C) or inoculating a pure colony into 5 ml of tryptic soy broth containing 0.6% yeast extract (TSB-YE, Becton Dickinson, Franklin Lakes, NJ). The cultures were incubated at 37 °C at 190 rpm in an orbital shaker for 16–18 h, diluted in 100 mM phosphate-buffered saline (PBS; 1.38 g/L of sodium phosphate monobasic, 2.68 g/L of sodium phosphate dibasic, 0.2 g/L of potassium chloride, and 8.5 g/L sodium chloride), and plated on different agar media to obtain 50–100 colonies per plate to capture the scatter patterns using BARDOT [5].
Table 1

List of bacterial strains used for generation of BARDOT library

Bacteriaa

Strains tested (n)

Source (Ref)

Percent positive predictive value (PPV) ± SD

Staphylococcus

Non-Staphylococcus

Staphylococcus species

S. aureus

12

ATCC, NRRL, PRI,

87.8 ± 12.8

12.2 ± 12.8

S. epidermidis

1

ATCC

91

9

S. hyicus

1

BLCC

98

2

S. lentus

1

BLCC

100

0

S. xylosus

1

BLCC

100

0

Non-Staphylococcus species

Bacillus cereus

2

ATCC

0

100

Escherichia coli O157:H7

2

ATCC, EDL

0

100

Streptococcus faecalis

1

PRI

38

62

Acinetobacter calcoaceticus

1

NRRL

33

67

Enterococcus faecium

1

NRRL

38

62

Klebsiella pneumoniae

1

NRRL

4

96

Listeria monocytogenes F4244

1

CDC

0

100

Pseudomonas aeruginosa

1

ATCC

0

100

Shigella flexneri

1

ATCC

0

100

Salmonella enterica serovar Typhimurium

1

ISDH

Yersinia enterocolitica

1

ATCC

Total

29

   

ATCC American Type Culture Collection, Manassas, VA; NRRL Northern Regional Research Laboratory, Peoria, IL; PRI Presque Isle Cultures; ISDH Indiana State Department of Health, Indianapolis, IN, USA; BLCC Bhunia Lab Culture Collection, Purdue University, West Lafayette, IN; CDC Centers for Disease Control and Prevention; SD standard deviation

a16 strains of Staphylococcus species and 13 strains of non-Staphylococcus species were used in this study and tested on PRMA

Optimal media selection for BARDOT-based analysis

Six different agar media were tested in this study to find a suitable medium, which can provide a high classification accuracy in terms of positive predictive value (PPV) [6] based on the colony scatter signature of Staphylococcus species. To build the scatter image library for the differentiation of Staphylococcus from non-Staphylococcus bacteria, or S. aureus from non-S. aureus, appropriate dilutions of bacterial cultures were plated on PRMA, mannitol salt agar (MSA), tellurite glycine agar (TGA), and Vogel and Johnson agar (VJA). All media were procured from Becton Dickinson (Franklin Lakes, NJ). TGA and VJA contain 1% tellurite solution required for differential growth of Staphylococcus species forming gray or black colonies, leading to the opacity of colonies that restricted the propagation of incident laser and hindered the formation of colony scatter pattern. Therefore, we prepared modified TGA (mTGA) and modified VJA (mVJA) by eliminating tellurite solution supplement. Appropriate dilutions of bacterial cultures in PBS were plated on these media and incubated at 37 °C, and colony scatter patterns were captured when the colony size reached 0.9 ± 0.2 mm in diameter, which was measured using a Leica DMLB microscope (Leica Microsystems USA, Bannockburn, IL). Baird-Parker (BP) agar was also used to follow the FDA procedure for food testing and to validate physiological characteristics of natural isolates. BARDOT was applied to acquire colony scatter patterns from BP but was unsuccessful due to the opacity of the media preventing laser penetration.

Acquisition and analysis of bacterial colony scatter pattern

Detailed BARDOT system has been explained [9, 11]. In brief, when a Petri dish containing bacterial colonies is imaged, a map on the BARDOT’s screen shows the location of each colony. The cutoff diameter of colonies to be acquired was set at 0.7–1.1 mm. The device moves the Petri dish in the x-y coordinate and passes the 635 nm red-diode laser beam through the center of each colony, collecting scatter images [11]. For building a scatter image library, at least images from 50 colonies were acquired from each Petri dish from two independent set of experiments. For initial solid agar media selection, bacteria were grown on PRMA, MSA, mTGA, and mVJA and a total of 1694 colony scatter images were collected to build a scatter image library. A built-in image analysis software was used to train the library [12]. The image analysis process has been previously described [6]. In short, Zernike (circular/ring) and Haralick (granular/roughness) features were extracted from every single colony scatter pattern. As a result, all cross-validation rounds were summarized in a matrix computing positive predictive value (PPV). The PPV shows the probability that the identified colonies belong to a specific genus of interest within the image library [11].

Building scatter image library of Staphylococcus on PRMA

PRMA was found to be the most suitable medium for building the scatter image library since it produced differential scatter signature of Staphylococcus spp. and non-Staphylococcus spp. and S. aureus from non-S. aureus species tested only after 14 h of incubation (see Fig. 1). Staphylococcus spp. including 12 strains of S. aureus and a single strain of each S. epidermidis, S. xylosus, S. hyicus, and S. lentus were used for library construction. Non-Staphylococcus cultures included two strains of each Bacillus cereus and Escherichia coli O157:H7 and a single strain of each Acinetobacter calcoaceticus, Enterococcus faecium, Listeria monocytogenes, Pseudomonas aeruginosa, Shigella flexneri, Salmonella enterica serovar Typhimurium, Yersinia enterocolitica, Klebsiella pneumoniae, and Streptococcus faecalis. All cultures were grown in TSB-YE at 37 °C overnight and serially diluted and appropriate dilutions were plated on PRMA to obtain about 50–100 colonies per plate. The Staphylococcus spp. library contained 852 images, while the non-Staphylococcus spp. contained 842 images, and scatter images were analyzed as described in the previous section.
Fig. 1

Selection of a suitable growth medium to build scatter image library for the screening and detection of Staphylococcus spp. a Colony profiles and scatter patterns of Staphylococcus species were acquired when colony diameter reached 0.9 ± 0.2 mm. Incubation time to reach the appropriate colony size is mentioned in colony profile as an inset. b Average positive predictive value (PPV) represents the average classification accuracy of the BARDOT acquired scatter pattern of Staphylococcus species on different media. MSA, mannitol salt agar; PRMA, phenol red mannitol agar; mTGA, modified tellurite glycine agar; mVJA, modified Vogel and Johnson agar. c Scatter pattern and colony profile of representative natural isolates from chicken salad and bovine raw milk on phenol red mannitol agar (PRMA). Colony 1 and colony 2 represent scatter images from two different PRMA plates of the same experiment. These scatter patterns were analyzed using an image classifier trained with Staphylococcus and non-Staphylococcus libraries (see also Table 2). Alfa numeric starting with KY are the NCBI GenBank accession numbers for the 16S rRNA gene sequences deposited in the GenBank

Application of BARDOT for detection of Staphylococcus from naturally contaminated food

Bovine raw milk (n = 6, 1 L each per visit) and ready-to-eat chicken salad (n = 3, one sample per visit) samples were procured from local dairy farms and local stores, respectively, and were processed according to the FDA method [15]. Appropriate dilution of food samples was also plated on tryptic soy agar with 0.6% yeast extract (TSA-YE) or brain-heart infusion (BHI) agar to assess total aerobic plate counts (TAPC). Chicken salad (50 g) and milk (50 mL) samples were diluted in 450 mL Butterfield’s buffer (1.25 mL/L of the stock solution, 34 g/L of KH2PO4, pH 7.2) in a stomacher bag. The samples were serially diluted in PBS and plated on PRMA and incubated at 37 °C for 12–14 h or until the colony diameter reached 0.9 ± 0.2 mm for the acquisition of the scattering patterns by BARDOT. A total of 2481 scatter patterns from PRMA were captured and images were matched with the scatter image libraries of Staphylococcus spp. and non-Staphylococcus spp. for bacterial identification. Two representative isolates from each batch of food sample showing > 95% PPV with the Staphylococcus library were further matched with S. aureus and non-S. aureus library and confirmed by PCR (see below). Isolates that did not show > 80% PPV with Staphylococcus library were not confirmed with BARDOT or PCR for S. aureus identification. Simultaneously, diluted food samples were also plated on BP and incubated at 37 °C for 24 h or until the colony with black (tellurite reduction) center and surrounding halo zone (phospholipase activity) appeared, characteristics of S. aureus. Colonies on BP prevented laser penetration and thus did not yield any scatter signature. Colonies from BP (FDA procedure) and PRMA (BARDOT procedure) were further verified by PCR.

DNA extraction and PCR confirmation

BARDOT-positive isolates were analyzed by PCR using tuf [16] or nuc [17] gene-specific primers and 16S rRNA gene sequencing [18]. DNA templates were extracted using either DNeasy blood and tissue kit (Qiagen, Hilden, Germany) or by boiling. The tuf gene-specific Tstag forward and Tstag reverse primers include TstaG422: 5′-GGCCGTGTTGAACGTGGTCAAATCA-3′ and Tstag765: 5′-TIACCATTTCAGTACCTTCTGGTAA-3′) [16]. The nuc gene-specific primers include 5′-GTGCTGGCATATGTATGGCAATTGT-3′ (For) and 5′-TCTTTGACCTTTGTCAAACTCGA-3′ (Rev) [17]. Universal eubacterial primers specific for 16S rRNA gene (U1F: AGAGTTTGATCCTGGCTCAG and U1R: GGTTACCTTGTTACGACTT) were used to identify isolates as Eubacteria and for ribosomal RNA gene homology [18]. The amplification reaction mixture (25 μL) contained 30 ng of DNA template, 1× GoTaq Flexi buffer, 1 U of GoTaq, 2.5 mM MgCl2, 200 μM (each) deoxynucleoside triphosphates (dNTP), and 4 μM each of forward and reverse primers. DNA template from standard S. aureus was used as a positive control, and water was used as a negative control. For 16S rRNA gene sequencing, clear GoTaq Flexi buffer was used to avoid interference in sequencing from the green dye in GoTaq Flexi buffer. Each amplicon was sequenced at the Purdue University Genomics Core Facility and the 16S rRNA gene sequence were submitted to NCBI GenBank with following accession numbers KY385269, KY385270, KY385271, KY385272, KY385273, KY385274, KY385275, KY385276, KY385277, KY385278, KY385279, KY385280, KY385281, KY385282, KY385283, KY385284, KY385285, KY385286, KY385287, KY385288, KY385289, and KY385290.

Statistical analysis

The GraphPad Prism (version 6) software was used to perform ANOVA analysis followed by Tukey’s multiple comparison tests to analyze any significant difference in colony scatter patterns obtained from PRMA, MSA, mTGA, and mVJA media. Cohen’s kappa value was calculated for some food isolates to assess the inter-rater reliability of BARDOT identification and 16S rRNA gene sequencing-based identification using the online QuickCalcs program, GraphPad Prism.

Results and discussion

Optimal media for BARDOT-based analysis and bacterial growth profile

To find a suitable plating medium for generating differential scatter patterns of Staphylococcus spp. identification, PRMA, MSA, TGA, VJA, mTGA, and mVJA were used (Fig. 1a). Among these, PRMA showed the highest classification accuracy (PPV = 97.56%) followed by MSA (PPV = 94.71%). On TGA and VJA, Staphylococcus spp. colonies appeared black forming dark precipitates and the colony scatter patterns were undefined, possibly due to the opacity of colonies. Thus, these two media formulations were modified (mTGA and mVGA) by eliminating tellurite supplement and the calculated PPV was 90.5 and 95.1%, respectively. Furthermore, the PPVs for the scatter images of Staphylococcus species captured on PRMA and mVJA were significantly different (P < 0.05) from those of MSA and mTGA (Fig. 1b). Though MSA, TGA, and VJA are commonly used selective/differential media for isolation of Staphylococcus species, we chose PRMA for obvious reasons—for producing highest distinguishable scatter features and for supporting faster growth (Fig. 1). The colony growth of all tested Staphylococcus species on PRMA varied from 14 to 27 h to reach the colony diameter of 0.9 ± 0.2 mm, while the colonies took 14–48 h on mVGA, 14–64 h on MSA, and 14–46 h on mTGA to reach the same colony diameter (Fig. 1a). We also observed that colony scatter patterns on PRMA were symmetric and colonies produced distinguishable defined scatter features compared to other media. Moreover, visually, PRMA was also able to distinguish mannitol-fermenting colonies (yellow) from the non-fermenting ones (reddish pink) due to the presence of pH indicator dye, phenol red. For example, S. epidermidis (ATCC 49134), being a non-mannitol fermenter, produced reddish pink colonies on MSA and PRMA. Taken together, the PRMA was selected as a suitable agar medium to build the Staphylococcus and non-Staphylococcus spp. scatter image libraries for Staphylococcus species including S. aureus screening and for detection from food samples.

The scatter patterns of bacterial colonies are affected by the type of media used, since the colony morphotype, accumulation of metabolic by-products, extracellular compounds, and three-dimensional arrangement and distribution of single cells within a colony affect the scatter patterns [3, 9, 19, 20]. Previous studies have demonstrated that agar concentration, thickness, medium formulation, humidity, and oxygenation can also affect bacterial growth and colony phenotype [21, 22]. Keeping all parameters constant, PRMA supported faster growth, provided distinguishable scatter patterns and the highest classification accuracy; thus, it was the primary choice for building scatter image library for detection and identification of Staphylococcus spp.

Light scattering event at the bacterial colony surface is a complex biophysical phenomenon. The laser beam illuminating the bacterial colony generates a signal referred to as scatter pattern [3]. Bacterial colony focuses the incoming optical field depending on the colony profile, which affects the focusing properties. Bacterial colony profile changes over time due to changes in colony size and metabolism [19]. Therefore, colony illuminated with the laser beam functions as an optical element with adaptive light focusing properties. As a result, conditions of light diffraction are affected by the phase modulation of bacterial colony due to its convex shape and the refractive index [4, 20, 23].

In a previous study [24], it was established that intensity pattern for Staphylococcus intermedius colonies shows the central low-intensity zone, and scatter pattern is generally affected by the light diffraction at the edges of the colony, whereas, for the E. coli colonies, central intensity ring occurs on the boundaries of the bacterial colony center, which is mostly populated with the oldest bacterial cells and highest concentration of the extracellular substance. The second intensity ring in E. coli results from the light diffraction on the bacterial colony edge. Moreover, the presence of additional radial spokes between the two intensity rings in both patterns results from the internal heterogeneity of the colony. This heterogeneity resulted from the migration of young cells from the center of the colony, leading to local changes in refractive index [24].

Based on available data from our previous work and others on light scattering at bacterial colony surface, we interpret that scatter pattern is measured as the function of colony profile, which includes colony height, diameter (shape and size), and refractive index [19, 20]. These factors together define the colony texture, which is referred to as surface or roughness [4, 6]. However, due to the dearth of literature on the correlation between measured light scattering signals with the morphological substructure of the bacteria in a colony opens future scope on the effect of gene manipulation in bacterial substructure on light scattering patterns.

Scatter image library

Previously, we observed that for accurate screening and detection of a target (inclusive) pathogen, it is important to match the scatter pattern of the suspect colony not only against the target pathogen-specific library but also against the non-target (exclusive) bacterial library comprising of microorganisms that likely to grow on the same agar plate during food testing [5]. Here, we built scatter image libraries for both Staphylococcus species and non-Staphylococcus species on PRMA (Fig. 1a and Table 1) and our libraries contained scatter patterns from 12 strains of S. aureus and one strain of each S. epidermidis, S. hyicus, S. lentus, and S. xylosus. The library for non-Staphylococcus species consisted of 842 scatter images representing 13 strains from 11 species representing 11 different genera. To validate and establish the robustness of the libraries, we used the image classifier to test the classification accuracy (PPV) for strains of standard Staphylococcus species that were not included in the library. All test strains gave a high PPV (87–100%) match with the Staphylococcus library and a low PPV (0–38%) match with the non-Staphylococcus library (Table 1). These results reiterate that the Staphylococcus scatter pattern library could be used for the screening and detection of natural Staphylococcus isolates, similar to our previous pathogen-specific libraries [5, 11, 12].

BARDOT-based detection of staphylococci from naturally contaminated food

A total of nine bovine raw milk and ready-to-eat chicken salad samples were analyzed for the presence of Staphylococcus species. Averages of total aerobic plate count (TAPC) on TSA-YE/BHI for six raw milk samples were 3.2 × 105 ± 1.5 × 105 CFU/mL, which were slightly above the US standard established by NRCNA [25], while for chicken salad sample 1.16 × 102 ± 0.52 × 102 CFU/mL, and this level is below the limit established by NACMCF [26]. A total of 18 isolates from chicken salad and 27 isolates from bovine raw milk were analyzed with a polyphasic approach (culture-dependent method, BARDOT, PCR, biochemical test, and 16S rRNA gene) (Table 2). BARDOT and other molecular methods confirmed the presence of Staphylococcus spp. in all nine chicken salad and bovine raw milk samples. Our finding corroborates with a previous study [27], where Staphylococcus was also detected in more than 60% of bovine raw milk samples. Since all test samples were positive for Staphylococcus species, and BARDOT was able to detect Staphylococcus from these samples, we did not feel it was necessary to test additional samples for validation of BARDOT.
Table 2

Application of BARDOT for the identification of Staphylococcus species from chicken salad and bovine raw milk

Isolatesa

Growth on BPAb

PCR analysis (tuf gene)

16S rRNA gene-based identificationc (accession number)

BARDOT library matching (%PPV)d

PCR analysis (nuc gene)e

Staph

Non-Staph

S. aureus

Non-S. aureus

Chicken salad-I

 CSBS1

+

+

Staphylococcus sp. (KY385269)

86

14

NA

NA

NT

 CSBS2

+

+

Staphylococcus sp. (KY385270)

80

20

NA

NA

NT

 CSBS3

+

+

Staphylococcus sp. (KY385271)

96

4

NA

NA

NT

 CSBS4

+

+

Staphylococcus sp. (KY385272)

96

4

77

23

 CSBS5

+

+

Staphylococcus sp. (KY385273)

100

0

73

27

+

 CSBS6

+

+

Staphylococcus sp. (KY385274)

94

6

NA

NA

NT

 CSBS7

+

+

Staphylococcus sp. (KY385275)

97

3

NA

NA

NT

 CSBS8

+

+

Staphylococcus sp. (KY385276)

84

16

NA

NA

NT

 CSBS9

+

+

Leuconostoc sp. (KY385277)

100

0

NA

NA

NT

 CSBS10

+

+

Staphylococcus sp. (KY385278)

73

27

NA

NA

NT

Chicken salad-II

 CSBA11

Serratia sp. (KY385279)

41

59

NA

NA

NT

 CSBA12

N

Serratia sp. (KY385280)

62

38

NA

NA

NT

 CSBA13

N

Pseudomonas sp. (KY385281)

42

58

NA

NA

NT

 CSBA14

N

Pantoea sp. (KY385282)

34

66

NA

NA

NT

Chicken salad-III

 CSBA15

Pantoea sp. (KY385283)

43

57

NA

NA

NT

 CSBA16

Pantoea sp. (KY385284)

50

50

NA

NA

NT

 CSBA17

+

Pseudomonas sp. (KY385285)

30

70

NA

NA

NT

 CSBA18

Pseudomonas sp. (KY385286)

88

12

NA

NA

NT

Raw milk-I

 RMBA19

N

Streptococcus sp. (KY385287)

80

20

NA

NA

NT

 RMBA20

N

Streptococcus sp. (KY385288)

67

33

NA

NA

NT

 RMBA23

+

+

Staphylococcus sp. (KY385289)

100

0

6

94

 RMBA24

+

+

Staphylococcus sp. (KY385290)

98

02

68

32

Raw milk-II

 RMBS25f

+

NS / +

0

100

NA

NA

NT

 RMBS26

NG

+

NS / +

0

100

NA

NA

NT

 RMBS27

+

NS / +

96

4

12

88

+

 RMBS28

+

+

NS / +

97.4

2.6

88

12

+

 RMBS30

NG

+

NS / +

0

100

NA

NA

NT

 RMBS32

NG

+

NS / +

0

100

NA

NA

NT

Raw milk-III

 RMBS33

+

+

NS / +

98

2

NA

NA

NT

 RMBS35

+

NS / +

100

0

NA

NA

NT

 RMBS36

+

+

NS / +

98

2

NA

NA

NT

 RMBS38

+

NS / +

100

0

65

35

 RMBS39

+

+

NS / +

90

10

NA

NA

NT

 RMBS40

NG

N

NS / +

75

25

NA

NA

NT

Raw milk-IV

 RMBS41

+

+

NS / +

100

0

96

4

+

 RMBS44

+

+

NS / +

98

2

NA

NA

NT

 RMBS45

+

+

NS / +

89

11

NA

NA

NT

 RMBS48

+

+

NS / +

100

0

NA

NA

NT

 RMBS51

+

NS / +

94.3

5.7

NA

NA

NT

 RMBS52

+

NS / +

100

0

NA

NA

NT

 RMBS53

+

+

NS / +

100

0

NA

NA

NT

 RMBS54

+

+

NS / +

98

2

94

6

+

 RMBS56

+

+

NS / +

98

2

NA

NA

NT

 RMBS57

+

NS / +

100

0

NA

NA

NT

 RMBS58

+

NS / +

100

0

NA

NA

NT

NG no growth, N non-specific amplification, NS not sequenced, NT not tested, NA not analyzed, CSBS Chicken Salad Bhunia Singh, CSBA Chicken Salad Bhunia Alsulami, RMBS Raw Milk Bhunia Singh, RMBA Raw Milk Bhunia Alsulami, PPV positive predictive value (classification accuracy)

aThe total of 45 isolates from 9 food samples was tested with BARDOT on PRMA. Cohen’s kappa values were calculated for isolates CSBS1-RMBA24 (22 isolates)

b+, a colony with the characteristic black center and/or halo zone indicate Staphylococcus on Baird-Parker agar (BPA); −, no growth of colony on BPA

cEubacterial 16S rRNA gene amplification

dStaph and non-Staph represent a class of scatter image library to differentiate isolates based on scatter pattern at the genus level, whereas S. aureus and non-S. aureus class of library was used to classify at the species level

eThese isolates were verified by S. aureus-specific PCR targeting nuc genes. +, PCR positive; −, PCR negative. Two representative isolates from each batch showing > 95% PPV with the Staphylococcus library were further matched with S. aureus and non-S. aureus library and confirmed with nuc gene-specific primers. Isolates that did not yield > 80% PPV with Staphylococcus library were not confirmed with BARDOT or PCR for S. aureus identification

fBovine raw milk isolates, RMBS25–RMBS58, were identified with double blind experiment, where BARDOT identification was performed by one author and tuf gene-based PCR was performed by another author

Based on the classification accuracy (%PPV) obtained for S. aureus (Table 1), a PPV of ≥ 80% was set as a threshold to differentiate true-positive and false-positive natural isolates using the BARDOT method. Out of 45 natural isolates, 28 (62%) isolates (chicken salad isolates, CSBS1-CSBS8; raw milk isolates RMBA23, RMBA24, RMBS27, RMBS28, RMBS33, RMBS35, RMBS36, RMBS38, RMBS39, RMBS41, RMBS44, RMBS45, RMBS48, RMBS51-RMBS54, and RMBS56–58) were identified as Staphylococcus spp. with BARDOT (≥ 80% PPV) and tuf gene-specific PCR/16S rRNA gene sequencing (Table 2). Of these isolates, about 24% were presumptively identified as S. aureus since they gave > 80% PPV when matched with S. aureus library (Table 2). Of these, 80% isolates were positively confirmed as S. aureus by nuc gene-specific PCR (Table 2). Further, 8 isolates (chicken salad isolates, CSBA11-CSBA17; raw milk isolate RMBA20) identified as non-Staphylococcus species with BARDOT (< 80% PPV) were also confirmed as non-Staphylococcus with PCR/16S rRNA gene sequencing (Table 2).

Furthermore, for reliable application of BARDOT as a screening tool, we also calculated Cohen’s kappa values to measure inter-rater agreement for the BARDOT method and 16S rRNA gene sequence-based homology for the characterization of 22 natural isolates from chicken salad and bovine raw milk. BARDOT-based identification of chicken salad and bovine raw milk isolates resulted in 6.6% false positives (Leuconostoc sp. CSBS9, Pseudomonas sp. CSBS18, and Streptococcus sp. RMBA19) (Table 2). Some Staphylococcus isolates (CSBS10, RMBS25, RMBS26, RMBS30, and RMBS32) showed < 80% match with the Staphylococcus library and classified as non-Staphylococcus. Interestingly, these isolates neither grew on BP nor produced S. aureus-like properties such as tellurite reduction (black colony) and phospholipase activity (halo zone) on BP (Table 2); thus, these isolates are considered atypical. Nevertheless, these data demonstrate that BARDOT can be used as a rapid colony screening tool for Staphylococcus on PRMA agar for the analysis of food samples, which do not rely on the chromogenic attributes but produces features that are related to bacterial genotype and physiology [5].

Scatter patterns of all Staphylococcus isolates were qualitatively different from the non-Staphylococcus isolates (Fig. 1a). Most of the Staphylococcus isolates obtained from chicken salad and bovine raw milk samples produced concentric ring-like scatter features (Fig. 1c), which show similarity with the scatter features of the standard Staphylococcus strains included in the library (Fig. 1c). Scatter patterns of Staphylococcus isolates possess characteristic scatter feature of concentric rings (Fig. 1a). All Staphylococcus isolates reached a desirable diameter (0.9 ± 0.2 mm) in 14 h to 27 h, whereas all non-Staphylococcus isolates took 14 to 64 h of incubation to reach the same size on PRMA. These differences in the scatter patterns of Staphylococcus and non-Staphylococcus isolates underscore the application potential of BARDOT for the screening and detection of Staphylococcus species from naturally contaminated samples.

Conclusions

The results show that the laser optical sensor, BARDOT, could be used for real-time detection and identification of Staphylococcus colonies on PRMA agar plate with 87–100% classification accuracy. Staphylococcus and non-Staphylococcus colony scatter patterns were profoundly different on the PRMA. Staphylococcus scatter pattern libraries were successfully used for the screening and detection of natural Staphylococcus isolates including S. aureus from chicken salad and bovine raw milk providing an opportunity to screen for unknown Staphylococcus isolates with similar scatter patterns from a variety of samples. Therefore, this novel label-free on-plate colony screening technology has the potential for adaptation by the food industry, biotechnology companies, and public health laboratories in the future for detection of Staphylococcus from various samples to aid in food safety and public health management.

Notes

Acknowledgments

T.S.A acknowledges King Saud University, the Kingdom of Saudi Arabia, for a graduate student fellowship. X.Z., a visiting scholar at Purdue University from College of Science, Nanjing University of Aeronautics and Astronautics, Nanjing, China, acknowledges the scholarship support from China Scholarship Council, Beijing, China. We also acknowledge the help of Bruce M. Applegate and Marcela M. Chavez in procuring raw milk samples.

Funding information

This research was supported through a cooperative agreement with the Agricultural Research Service of the US Department of Agriculture project number 8072-42000-077, USDA National Institute of Food and Agriculture (Hatch) project, and the Center for Food Safety Engineering at Purdue University.

Compliance with ethical standards

The authors declare that there were no ethical implications or research involving human or animals.

Conflict of interest

The authors declare that they have no conflict of interest.

References

  1. 1.
    Baird-Parker A. An improved diagnostic and selective medium for isolating coagulase positive staphylococci. J Appl Microbiol. 1962;25(1):12–9.Google Scholar
  2. 2.
    Bhunia AK. One day to one hour: how quickly can foodborne pathogens be detected? Future Microbiol. 2014;9(8):935–46.CrossRefPubMedGoogle Scholar
  3. 3.
    Banada PP, Guo S, Bayraktar B, Bae E, Rajwa B, Robinson JP, et al. Optical forward-scattering for detection of Listeria monocytogenes and other Listeria species. Biosens Bioelectron. 2007;22(8):1664–71.CrossRefPubMedGoogle Scholar
  4. 4.
    Bae E, Banada PP, Huff K, Bhunia AK, Robinson JP, Hirleman ED. Biophysical modeling of forward scattering from bacterial colonies using scalar diffraction theory. Appl Opt. 2007;46(17):3639–48.CrossRefPubMedGoogle Scholar
  5. 5.
    Singh AK, Bettasso AM, Bae E, Rajwa B, Dundar MM, Forster MD, et al. Laser optical sensor, a label-free on-plate Salmonella enterica colony detection tool. MBio. 2014;5(1):e01019–3.CrossRefPubMedPubMedCentralGoogle Scholar
  6. 6.
    Ahmed WM, Bayraktar B, Bhunia AK, Hirleman ED, Robinson JP, Rajwa B. Classification of bacterial contamination using image processing and distributed computing. IEEE J Biomed Health Inform. 2013;17(1):232–9.CrossRefPubMedGoogle Scholar
  7. 7.
    Sousa AM, Pereira MO, Lourenço A. MorphoCol: an ontology-based knowledgebase for the characterisation of clinically significant bacterial colony morphologies. J Biomed Inform. 2015;55:55–63.CrossRefPubMedGoogle Scholar
  8. 8.
    Zieliński B, Plichta A, Misztal K, Spurek P, Brzychczy-Włoch M, Ochońska D. Deep learning approach to bacterial colony classification. PLoS One. 2017;12(9):e0184554.CrossRefPubMedPubMedCentralGoogle Scholar
  9. 9.
    Banada PP, Huff K, Bae E, Rajwa B, Aroonnual A, Bayraktar B, et al. Label-free detection of multiple bacterial pathogens using light-scattering sensor. Biosens Bioelectron. 2009;24:1685–92.CrossRefPubMedGoogle Scholar
  10. 10.
    Abdelhaseib MU, Singh AK, Bailey M, Singh M, El-Khateib T, Bhunia AK. Fiber optic and light scattering sensors: complimentary approaches to rapid detection of Salmonella enterica in food samples. Food Control. 2016;61:135–45.CrossRefGoogle Scholar
  11. 11.
    Tang Y, Kim H, Singh AK, Aroonnual A, Bae E, Rajwa B, et al. Light scattering sensor for direct identification of colonies of Escherichia coli serogroups O26, O45, O103, O111, O121, O145 and O157. PLoS One. 2014;9(8):e105272.CrossRefPubMedPubMedCentralGoogle Scholar
  12. 12.
    Singh AK, Sun XL, Bai XJ, Kim H, Abdalhaseib MU, Bae E, et al. Label-free, non-invasive light scattering sensor for rapid screening of Bacillus colonies. J Microbiol Methods. 2015;109:56–66.CrossRefPubMedGoogle Scholar
  13. 13.
    Singh AK, Drolia R, Bai X, Bhunia AK. Streptomycin induced stress response in Salmonella enterica serovar Typhimurium shows distinct colony scatter signature. PLoS One. 2015;10(8):e0135035.CrossRefPubMedPubMedCentralGoogle Scholar
  14. 14.
    Singh AK, Leprun L, Drolia R, Bai X, Kim H, Aroonnual A, et al. Virulence gene-associated mutant bacterial colonies generate differentiating two-dimensional laser scatter fingerprints. Appl Environ Microbiol. 2016;82(11):3256–68.CrossRefPubMedPubMedCentralGoogle Scholar
  15. 15.
    Andrews WH, Hammack TS. Food sampling and preparation of sample homogenate. In: Bacteriological analytical manual online vol 3. 8th ed. Silver Spring: Center for Food Safety & Applied Nutrition, US Food and Drug Administration; 2003. p. 2004.Google Scholar
  16. 16.
    Morot-Bizot SC, Talon R, Leroy S. Development of a multiplex PCR for the identification of Staphylococcus genus and four staphylococcal species isolated from food. J Appl Microbiol. 2004;97(5):1087–94.CrossRefPubMedGoogle Scholar
  17. 17.
    Sunagar R, Deore S, Deshpande P, Rizwan A, Sannejal A, Sundareshan S, et al. Differentiation of Staphylococcus aureus and Staphylococcus epidermidis by PCR for the fibrinogen binding protein gene. J Dairy Sci. 2013;96(5):2857–65.CrossRefPubMedGoogle Scholar
  18. 18.
    Singh AK, Ramesh A. Evaluation of a facile method of template DNA preparation for PCR-based detection and typing of lactic acid bacteria. Food Microbiol. 2009;26(5):504–13.CrossRefPubMedGoogle Scholar
  19. 19.
    Bae E, Bai N, Aroonnual A, Robinson JP, Bhunia AK, Hirleman ED. Modeling light propagation through bacterial colonies and its correlation with forward scattering patterns. J Biomed Opt. 2010;15(4):045001.CrossRefPubMedGoogle Scholar
  20. 20.
    Kim H, Doh I-J, Sturgis J, Bhunia AK, Robinson JP, Bae E. Reflected scatterometry for noninvasive interrogation of bacterial colonies. J Biomed Opt. 2016;21(10):107004.CrossRefPubMedGoogle Scholar
  21. 21.
    Cooper A, Dean A, Hinshelwood C. Factors affecting the growth of bacterial colonies on agar plates. Proc R Soc Lond B Biol Sci. 1968;171(1023):175–99.CrossRefPubMedGoogle Scholar
  22. 22.
    Mialon M, Tang Y, Singh AK, Bae E, Bhunia AK. Effects of preparation and storage of agar media on the sensitivity of bacterial forward scattering patterns. Open J Appl Biosens. 2012;1(03):26.CrossRefGoogle Scholar
  23. 23.
    Suchwałko A, Buzalewicz I, Podbielska H. Bacteria identification in an optical system with optimized diffraction pattern registration condition supported by enhanced statistical analysis. Opt Express. 2014;22(21):26312–27.CrossRefPubMedGoogle Scholar
  24. 24.
    Buzalewicz I, Kujawińska M, Krauze W, Podbielska H. Novel perspectives on the characterization of species-dependent optical signatures of bacterial colonies by digital holography. PLoS One. 2016;11(3):e0150449.CrossRefPubMedPubMedCentralGoogle Scholar
  25. 25.
    Skovgaard N. Scientific criteria to ensure safe food: Institute of Medicine, Natural Research Council of the National Academies. The National Academy Press, Elsevier, Washington DC. 2004.Google Scholar
  26. 26.
    NACMCF. Response to questions posed by the Department of Defense Regarding Microbiological Criteria as indicators of process control or insanitary conditions, National Advisory Committee on Microbiological Criteria for Foods. USDA-FSIS, Washington, DC. 2015.Google Scholar
  27. 27.
    Chye FY, Abdullah A, Ayob MK. Bacteriological quality and safety of raw milk in Malaysia. Food Microbiol. 2004;21(5):535–41.CrossRefGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2018

Authors and Affiliations

  • Tawfiq S. Alsulami
    • 1
    • 2
  • Xingyue Zhu
    • 1
  • Maha Usama Abdelhaseib
    • 1
    • 3
  • Atul K. Singh
    • 1
    • 4
  • Arun K. Bhunia
    • 1
    • 5
  1. 1.Molecular Food Microbiology Laboratory, Department of Food SciencePurdue UniversityWest LafayetteUSA
  2. 2.Department of BacteriologyUniversity of WisconsinMadisonUSA
  3. 3.Department of Food HygieneAssiut UniversityAssiutEgypt
  4. 4.Clear LabsMenlo ParkUSA
  5. 5.Department of Comparative PathobiologyPurdue UniversityWest LafayetteUSA

Personalised recommendations