Research Project: Molecular Identification and Characterization of Bacterial and Viral Pathogens Associated with Foods

Location: Produce Safety and Microbiology Research

2019 Annual Report

Objectives
The overall objective of this project is to develop novel typing methods to identify foodborne pathogens and characterize: bacterial foodborne pathogens through genomics, transcriptomics and proteomics; virulence factors and bacterial toxins; and antibiotic resistance in food production. Specifically, during the next five years we will focus on the following objectives: Objective 1: Develop improved identification technologies for human bacterial and viral pathogens to replace current testing methodologies. Sub-objective 1A: Develop a fast, simple and high throughput array-based method fortyping pathogens. Sub-objective 1B: Validate the array genotyping tool for the identification of viral and bacterial pathogens in samples from agricultural environments. Sub-objective 1C: Develop novel Campylobacteraceae species identification methods. Objective 2: Identify and characterize genetic factors associated with virulence and/or environmental adaptation of human bacterial pathogens using genomic, transcriptional and proteomic analyses. Sub-objective 2A: Identify the transcriptional network patterns of bacterial pathogens during environmental adaptation and modulation of their stress response. Sub-objective 2B: Identify genes involved in host/environmental adaptation and investigate variation in virulence potential through in-depth genome sequencing of selected taxa. Sub-objective 2C: Identify the genetic and epigenetic alterations or factors involved in the environmental adaptation of foodborne pathogens through genomic and methylome analyses. Sub-objective 2D: Quantitative proteomic and transcriptomic analysis of virulence factors of foodborne pathogens can be used to elucidate transcriptional vs. posttranscriptional control of virulence in foodborne pathogens. Sub-objective 2E: Top-down proteomic characterization of bacterial virulence proteins or toxins. Objective 3: Characterize molecular mechanisms contributing to the potency of bacterial toxins. Sub-objective 3A: Identification and characterization of Shiga toxin 2 (Stx2) subtypes in environmental STEC strains. Sub-objective 3B: Characterization of Stx2 expression levels and functional activities in environmental STEC strains. Sub-objective 3C: Characterization of pathogenic mechanisms associated with Stx2 subtypes produced by E. coli strains. Sub-objective 3D: Investigate toxin-inactivation mechanisms by natural plant compounds. Objective 4: Identify antimicrobial resistance gene reservoirs in the food production ecosystem and characterize the fitness and virulence of resistant pathogens. Sub-objective 4A: Complete genomic sequencing and functional metagenomic analyses of antibiotic-resistant Campylobacter. Sub-objective 4B: Characterization of the fitness and virulence of antimicrobial-resistant Campylobacter jejuni and Campylobacter coli.

Approach
Objective 1: A fast, simple and high throughput array-based method for typing pathogens will be developed. Capture probes will be designed to target norovirus and hepatitis A virus, clinically-important Salmonella serovars and Campylobacter spp. To evaluate probe specificity, viral RNA or bacterial DNA will be extracted from clinical samples or cultured strains. A Cooperative Research and Development Agreement (CRADA) has been established with Arrayit Corporation to develop a fast, simple, and cost-effective test, in conjunction with inexpensive instrumentation. The array genotyping method will be validated using samples from agricultural environments. Also, MALDI-TOF-MS will be assessed as a faster, more accurate and reliable identification of Campylobacteraceae taxa, when compared to current phenotype-based approaches. Objective 2: The transcriptomic patterns that correlate with environmental adaptation and stress modulation for Campylobacter, Salmonella and E. coli will be determined, using RNA-Seq under distinct and relevant environmental conditions. Gene content or alleles that tentatively correlate with niche preference, environmental adaptation or pathogenicity will be identified by sequencing Campylobacter and Arcobacter isolates from a more diverse strain set. Alleles or methylation patterns within a population that correlate with environmental adaptation or pathogenicity will be identified through next-generation genomic analysis. Also, proteomic and transcriptomic analysis will be used to investigate transcriptional/post-transcriptional control of virulence factors and to characterize bacterial toxins and virulence determinants. Objective 3: A genotypic and proteomic screen for identifying and classifying Shiga toxin subtypes, harbored by strains recovered from different sources and locations in a major agricultural region, will be conducted. Using enzyme-linked immunosorbent assay and cell-based assays, the amounts and functional activities of Shiga toxin 2a and 2c subtypes will be determined. Using surface plasmon resonance, the mechanisms contributing to the cytotoxicities associated with the Shiga toxin 2a and 2c subtypes will be characterized, by investigating their role in the inhibition of protein synthesis in mammalian cells, thus providing a better understanding of the toxin’s mode of action. Natural plant compounds, specifically polyphenolics, will be investigated as potential inactivators of bacterial toxins. Objective 4: Genome sequencing of antimicrobial-resistant Campylobacter, isolated from poultry farms, will be performed to identify (potentially novel) antibiotic resistance genes. Metagenomic analysis of bacteria isolated from samples (such as litter, insects and fecal droppings) from these same poultry farms will be performed to identify the pool of ‘available’ antibiotic resistance genes that could potentially be transferred into Campylobacter. The fitness and virulence of resistant Campylobacter will be measured, to determine if increased fitness explains the persistence of resistant strains. Fitness metrics will include survival in insects and on poultry, and fecal colonization.

Progress Report
Under Objective 1, progress was obtained on the development of typing methods for the simple and fast identification of foodborne pathogens. A collaboration with the technology sector and leafy-greens processors was continued to design a prototype for the real-time surveillance of pathogens. To address the request by leafy green processors in California, a probe-based detection method was validated for the detection of Listeria species from swabs collected from surfaces in processing facilities. The bacterial cells from these swab samples were captured by employing single-stranded DNA sequences (“aptamers”), and additional protocols were designed and optimized for sample concentration and mechanical lysis of the bacterial cells. To achieve a high level of sensitivity, the probes targeted high-copy sequences in Listeria species. Preliminary results demonstrated that the prototype method can detect less than 5-10 cells from sponge-swab samples collected at a processing facility. These findings established a precedent for developing an integrated system to rapidly detect Listeria species at low-cell concentrations from environmental samples. In support of Objective 2, advances were made in the identification of Arcobacter genes involved potentially in host/environmental adaptation or virulence. Arcobacter butzleri strains from Hurricane Florence floodwaters were typed by DNA analysis (multi-locus sequence typing (MLST)) and compared to A. butzleri MLST sequence types already present in the MLST database. Florence floodwater strains were present in two clusters that also contained previously-characterized sequence types representing strains recovered from both food animals and human clinical samples. The presence of food animal sequence types in these clusters is not surprising, since A. butzleri are recovered routinely from food animals and likely entered the floodwaters via run-off from this dense agricultural region. This initial analysis will be used to select strains for a more thorough chromosomal DNA sequencing to identify genes related to source, as well as genes involved in virulence and antimicrobial resistance. In another study, an Arcobacter cryaerophilus strain from New Zealand that was isolated from green mussels was characterized. This strain possesses an unusual extrachromosomal element that putatively codes for a secreted toxin. Under Objective 2, progress was also made in the characterization of emerging or novel Campylobacter associated with human illness or highly-related to pathogenic campylobacters. As part of an international collaboration, the chromosomal sequences of several Campylobacter showae strains associated with severe gastrointestinal illness (that is, ulcerative colitis, colon cancer) were determined, characterized and compared to other C. showae strains. Analysis indicated that the adherent/invasive strains possessed a gene set that was different from the non-adherent/non-invasive strains, suggesting that certain genes are important for the localization, colonization and virulence potential of C. showae. Additionally, the chromosomal sequences of three putative novel Campylobacter species were determined. Strains representing these species were isolated from: children under five years of age from several countries in sub-Saharan Africa and south Asia; human clinical samples in Brittany, France; and wild red foxes (this novel species is highly related to C. upsaliensis, a human pathogen recovered routinely from domestic cats and wild and domestic dogs) Within Sub-objective 2A, progress continued in the elucidation of Campylobacter jejuni transcriptional network patterns during environmental adaptation using RNA sequencing (RNAseq). In collaboration with researchers at Washington State University, C. jejuni strains were cultured in association with human macrophage that would be encountered during infection. C. jejuni global gene expression was characterized during a time course of bacteria with macrophage, thus identifying gene expression pathways important to infection. Within Sub-objective 2B, progress was made in the elucidation of Salmonella genes involved in environmental adaptation. Scientists in Albany, California, used genomic sequencing to identify rpoS gene variants among Salmonella strains involved in a food outbreak. The RpoS protein regulates a variety of genes involved in stress management within Salmonella. RNAseq analysis demonstrated that these variants altered how RpoS regulates many stress pathways. Within Sub-objective 2C, advances were made in the identification of genetic alterations involved in the environmental adaptation of foodborne pathogens through genomic analyses. Within the gut of chickens, C. jejuni is exposed to bile, a protective agent against bacterial colonization. C. jejuni adapts to survival in this environment by altering the expression of genes, causing bile resistance. In collaboration with Washington State University, we have continued to obtain isolates of C. jejuni able to grow in increasing levels of bile and their genomes were sequenced to identify changes that could be related to the bile resistance. Under Sub-objective 2D, Salmonella enterica enterica wild-type and mutant strains were analyzed for identification and quantification of expressed proteins using label-free proteomic data analysis. In addition, hypotheses were developed to explain detection of effector/invasion proteins (SipA-D) of Salmonella pathogenicity island 1 during proteolytic surface-shaving of Salmonella enterica enterica (SEE) Newport and Thompson (but not for Kentucky). Although SipB-D are surface-exposed as part of the injectosome needle complex of the Type III secretion system, detection of SipA was unusual because it is thought to be injected directly into eukaryotic cells. However, no eukaryotic cells were present in the experiment. The absence of SipA-D detection for SEE Kentucky is consistent with previous assessments of SEE Kentucky indicating that, although it is prevalent in poultry, it is not considered to be a robust human pathogen. Significant progress was made on Sub-objective 2E. A method was developed to enhance detection of bacterial proteins and toxins using high resolution mass spectrometry and top-down proteomic analysis. Using this new approach, the mature intact A-subunit of Shiga toxin 2 (approximately 33 kDa) was detected from expression by pathogenic E. coli O157:H7. Similarly, the C-subunit of cytolethal distending toxin (CdtC) was also detected from a non-motile pathogenic E. coli O157:H- strain. Significant progress was made on Sub-objective 3A. A suspected pathogenic E. coli O157:H7 strain isolated from a water source in an agricultural region in California that was negative for Shiga toxin (but positive for other virulence factors) by PCR was analyzed by whole genome sequencing (WGS). WGS revealed that this strain had 99.97 percent homology to a genomically sequenced strain isolated from cattle in Kansas. Both California and Kansas strains possess a prophage having a stx1 gene, however, this gene has multiple mutations that might explain the absence of Stx expression based on mass spectrometry analysis and the lack of detection of the stx1 gene by PCR. The high homology of these two strains may suggest a common bacterial lineage despite the geographical separation of the collection sites. Using routine surveillance methods, STEC O157:H7 cattle strains, recovered during a short time period and discrete sampling location, were initially identified based on a typical STEC colony morphology and positive PCR tests for the stx2c gene; however, subsequent results from cytotoxicity assays showed various levels of Shiga toxin activity. The use of WGS enabled the identification of the insertion sequence element (IS1203v) within the stx-carrying prophage in these highly similar O157:H7 cattle strains and provided an explanation for the observed attenuated pathogenic potential in some recovered strains. Under Sub-objectives 3C and 3D, significant progress was achieved on the characterization of pathogenic mechanisms associated with Shiga toxin 2 subtypes using cultured mammalian cells. The combinatorial use of cell sorting with a fluorescent cell-based assay enabled the identification of subpopulations of intoxicated mammalian cells. The results demonstrated that the toxin-resistant subpopulations were due to a weak binding affinity between the Shiga toxin subtype and the host mammalian cell, reducing the delivery of the Shiga toxin 2a and 2c subtypes when compared to Shiga toxin 1a subtype. In addition, this research quantified the amounts of bacterial toxins required for complete inhibition of protein synthesis and indicated that mammalian host cells can recover from intoxication under sub-lethal toxin concentrations. These findings have provided fundamental information on the amounts of Shiga toxins that result in reversible toxin effects, enabling the design of effective intervention strategies for toxin inactivation. Within Objective 4, progress was achieved on the characterization of antimicrobial resistance (AMR) in Campylobacter. In collaboration with North Carolina State University, AMR, poultry-associated C. jejuni and C. coli were characterized for several strains, and multiple AMR genes or mutations were identified. Progress was also achieved on the characterization of macrolide resistance in Campylobacter with the genome sequencing of Erythromycin (erm) sensitive strains of C. jejuni and C. coli that are being used to select for erm resistant mutants.

Accomplishments
1. Characterized relative potencies of Shiga toxin subtypes in the host cell. Shiga toxins produced by some Escherichia coli strains are responsible for the development of severe disease symptoms in humans, such as hemolytic-uremic syndrome. Distinct Shiga toxin subtypes have been identified with different potencies in humans, and the factors contributing to these differences among the subtypes are not fully understood. ARS scientists in Albany, California, collaborated on research with the University of Central Florida to demonstrate the use of cell sorting with an unstable fluorescent reporter for quantifying the relative potencies of Shiga toxin subtypes in individual mammalian cells. This research identified for the first time a subpopulation of toxin-resistant mammalian cells due to a relatively weak affinity between the Shiga toxin subtype and the host human cell and provided an explanation for the differential cellular toxicity between the subtypes. These findings have set a precedent for the development of effective strategies for the treatment of Shiga toxin-producing Escherichia coli infections as well as other toxin-mediated diseases.

2. Non-motile, Shiga toxin-producing E. coli (STEC) investigated by top-down proteomic analysis and whole genome sequencing. Shiga toxin-producing E. coli (STEC) endanger public health and represent an economic burden on agriculture, and methods are needed to rapidly identify bacterial contaminants that express Shiga toxin (Stx). ARS researchers in Albany, California, analyzed three non-motile STEC strains for expression of Stx using mass spectrometry (MS/MS) and top-down proteomic analysis. One STEC (E. coli O157:H-) linked to an outbreak of hemolytic uremic syndrome that expressed Stx2a and was typed by MS/MS and top-down analysis from the B-subunit of the Stx holotoxin. MS/MS was performed on both the disulfide bond-intact and disulfide bond-reduced B-subunit that confirmed the presence of this intramolecular bond critical for structure and function. Two other non-motile E. coli O145:H28 strains isolated from an agricultural region in California were also analyzed: one expressed Stx1a and the other Stx2a. This technique demonstrates a rapid method for detecting and identifying Shiga toxin expression from suspected STEC strains.

3. Identification of host-specific Campylobacter jejuni markers. The human pathogen, Campylobacter jejuni, is a commensal organism of the intestinal tracts of many wild and domestic avian and mammalian species and a leading cause of diarrheal disease in humans. Considerable genomic analysis of C. jejuni from domestic animals has demonstrated that certain genomic types of C. jejuni are associated with particular animal hosts, for example, poultry- and cattle-associated types. Moreover, poultry- and cattle-associated strains are among the types of strains that are recovered from human clinical samples. Recently, ARS scientists in Albany, California, determined the genome sequences of several C. jejuni strains isolated from wild birds and black bears, and although most C. jejuni strains from black bears were similar to cattle-associated strains, two black bear isolates and all C. jejuni isolates from wild birds were genomically distinct from the poultry- and cattle-associated strains. These two distinctive black bear isolates and C. jejuni isolates from blackbirds had genes that allow glucose utilization, while all the strains from wild birds had distinctive toxin genes. The identification of host-specific, genomic markers in C. jejuni will improve source attribution of clinical infections and inform intervention strategies to reduce campylobacteriosis.

Review Publications
Fagerquist, C.K., Zaragoza, W.J. 2018. Proteolytic surface-shaving and serotype-dependent expression of SPI-1 invasion proteins in Salmonella enterica subspecies enterica. Frontiers in Nutrition. 5:124. https://doi.org/10.3389/fnut.2018.00124.
Miller, W.G., Yee, E., Bono, J.L. 2018. Complete genome sequence of the Arcobacter halophilus type strain CCUG 53805. Microbiology Resource Announcements. 7(14):e01077-18. https://doi.org/10.1128/MRA.01077-18.
Miller, W.G., Yee, E. 2018. Complete genome sequence of the Arcobacter trophiarum type strain LMG 25534. Microbiology Resource Announcements. 7(13):e01110-18. https://doi.org/10.1128/MRA.01110-18.
Miller, W.G., Yee, E., Bono, J.L. 2018. Complete genome sequence of the Arcobacter ellisii type strain LMG 26155. Microbiology Resource Announcements. 7(16):e01268-18. https://doi.org/10.1128/MRA.01268-18.
Miller, W.G., Yee, E., Huynh, S., Parker, C. 2018. Complete genome sequence of the Arcobacter marinus type strain JCM 15502. Microbiology Resource Announcements. 7(16):e01269-18. https://doi.org/10.1128/MRA.01269-18.
Miller, W.G., Yee, E., Bono, J.L. 2018. Complete genome sequence of the Arcobacter molluscorum type strain LMG 25693. Microbiology Resource Announcements. 7(16):e01293-18. https://doi.org/10.1128/MRA.01293-18.
Miller, W.G., Yee, E. 2018. Complete genome sequence of the Arcobacter skirrowii type strain LMG 6621. Microbiology Resource Announcements. 7(17):e01308-18. https://doi.org/10.1128/MRA.01308-18.
Miller, W.G., Yee, E., Bono, J.L. 2018. Complete genome sequence of the Arcobacter suis type strain LMG 26152. Microbiology Resource Announcements. 7(17):e01307-18. https://doi.org/10.1128/MRA.01307-18.
Fagerquist, C.K., Zaragoza, W.J., Lee, B.G., Yambao, J.C., Quiñones, B. 2018. Clinically-relevant Shiga toxin 2 subtypes from environmental Shiga toxin-producing Escherichia coli identified by top-down/middle-down proteomics and DNA sequencing. Clinical Mass Spectrometry. 11:27-36. https://doi.org/10.1016/j.clinms.2018.12.001.
Fagerquist, C.K., Lee, B.G., Zaragoza, W.J., Yambao, J.C., Quiñones, B. 2018. Software for top-down proteomic identification of a plasmid-borne factor (and other proteins) from genomically sequenced pathogenic bacteria using MALDI-TOF-TOF-MS/MS and post-source decay. International Journal of Mass Spectrometry. 438:1-12. https://doi.org/10.1016/j.ijms.2018.12.006.
Miller, W.G., Yee, E., Bono, J.L. 2018. Complete genome sequences of the Arcobacter cryaerophilus strains ATCC 43158T and ATCC 49615. Microbiology Resource Announcements. 7(20):e01463-18. https://doi.org/10.1128/MRA.01463-18.
Duarte, A., Botteldoorn, N., Miller, W.G., Coucke, W., Martiny, D., Hallin, M., Seliwiorstow, T., De Zutter, L., Uyttendaele, M., Vandenberg, O., Dierick, K. 2018. Relation between broiler and human C. jejuni strains isolated in Belgium from 2011 to 2013. Journal of Applied Microbiology. 126:277-287. https://doi.org/10.1111/jam.14132.
Niedermeyer, J., Ring, L., Miller, W.G., Genger, S., Parr Lindsey, C., Osborne, J., Kathariou, S. 2018. Proximity to other commercial turkey farms affects colonization onset, genotypes and antimicrobial resistance profiles of Campylobacter in turkeys: suggestive evidence from a paired-farm model. Applied and Environmental Microbiology. 84(18):e01212-18. https://doi.org/10.1128/AEM.01212-18.
Bolinger, H.K., Zhang, Q., Miller, W.G., Kathariou, S. 2018. Lack of evidence for erm(B) infiltration into Erythromycin-resistant Campylobacter coli and Campylobacter jejuni from commercial turkey production in eastern North Carolina: a major turkey-growing region in the United States. Foodborne Pathogens and Disease. 15(11):698-700. https://doi.org/10.1089/fpd.2018.2477.
Crippen, C., Huynh, S., Miller, W.G., Parker, C., Szymanski, C.M. 2018. Complete genome sequence of Acinetobacter radioresistens strain LH6, a multidrug-resistant bacteriophage-propagating strain. Microbiology Resource Announcements. 7(5):e00929-18. https://doi.org/10.1128/MRA.00929-18.
Bian, X., Huynh, S., Chapman, M.H., Szymanski, C., Parker, C., Miller, W.G. 2018. Draft genome sequences of nine Campylobacter hyointestinalis subsp. lawsonii strains. Microbiology Resource Announcements. 7(10):e01016-18. https://doi.org/10.1128/MRA.01016-18.
Miller, W.G., Yee, E., Bono, J.L. 2018. Complete genome sequence of the Arcobacter bivalviorum type strain LMG 26154. Microbiology Resource Announcements. 7(12):e01076-18. https://doi.org/10.1128/MRA.01076-18.
Miller, W.G., Yee, E., Bono, J.L. 2018. Complete genome sequence of the Arcobacter mytili type strain LMG 24559. Microbiology Resource Announcements. 7(11):e01078-18. https://doi.org/10.1128/MRA.01078-18.
Elhadidy, M., Miller, W.G., Arguello, H., Álvarez-Ordóñez, A., Dierick, K., Botteldoorn, N. 2018. Molecular epidemiology and antimicrobial resistance mechanisms of Campylobacter coli from diarrheal patients and broiler carcasses in Belgium. Transboundary and Emerging Diseases. 66:463-475. https://doi.org/10.1111/tbed.13046.
Parker, C., Huynh, S., Bono, J.L., Miller, W.G., Cooley, M.B., Brandl, M. 2019. Complete genome sequences of three Shiga toxin-producing Escherichia coli O111:H8 strains exhibiting an aggregation phenotype. Microbiology Resource Announcements. 8(1):e01335-18. https://doi.org/10.1128/MRA.01335-18.
Hsu, T., Gemmell, M.R., Franzosa, E., Berry, S., Mukhopadhya, I., Hansen, R., Michaud, M., Nielsen, H., Miller, W.G., Nielsen, H., Bajaj-Elliott, M., Huttenhower, C., Garrett, W.S., Hold, G.L. 2019. Comparative genomics and genome biology of Campylobacter showae. Emerging Microbes & Infections. 8(1):827-840. https://doi.org/10.1080/22221751.2019.1622455.
Parker, C., Cooper, K.K., Huynh, S., Smith, T.P., Bono, J.L., Cooley, M.B. 2018. Genome sequences of eight Shiga toxin-producing Escherichia coli strains isolated from a produce-growing region in California. Genome Announcements. 7(1):1-3. https://doi.org/10.1128/MRA.00807-18.
Kennedy, C., Walsh, C., Karczmarczyk, M., O’Brien, S., Akasheh, N., Quirke, M., Farrell-Ward, S., Buckley, T., Fogherty, U., Kavanagh, K., Parker, C., Sweeney, T., Fanning, S. 2018. Multi-drug resistant Escherichia coli in diarrhoeagenic foals: pulsotyping, phylotyping, serotyping, antibiotic resistance and virulence profiling. Veterinary Microbiology. 223:144-152. https://doi.org/10.1016/j.vetmic.2018.08.009.
Tran, T.D., Huynh, S., Parker, C., Han, R., Hnasko, R.M., Gorski, L.A., McGarvey, J.A. 2018. Complete genome sequence of Lactococcus lactis subsp. lactis strain 14B4, which inhibits the growth of Salmonella enterica serotype Poona in vitro. Microbiology Resource Announcements. 7(19):e01364-18. https://doi.org/10.1128/MRA.01364-18.
Negretti, N.M., Clair, G., Gourley, C.R., Huynh, S., Adkins, J.N., Parker, C., Konkel, M.E. 2019. Campylobacter jejuni demonstrates conserved proteomic and transcriptomic responses when co-cultured with human INT 407 and Caco-2 epithelial cells. Frontiers in Microbiology. 10:755. https://doi.org/10.3389/fmicb.2019.00755.
Tran, T., Huynh, S., Parker, C., Hnasko, R.M., Gorski, L.A., McGarvey, J.A. 2018. Complete genome sequences of three Bacillus amyloliquefaciens strains that inhibit the growth of Listeria monocytogenes in vitro. Genome Announcements. 6(25):e00579-18. https://doi.org/10.1128/genomeA.00579-18.
Quiñones, B., Yambao, J.C., Silva, C.J., Lee, B.G. 2019. Draft genome sequences of Shiga toxin-producing Escherichia coli O157:H7 strains recovered from a major production region for leafy greens in California. Microbiology Resource Announcements. 8(27):e00644-19. https://doi.org/10.1128/MRA.00644-19.
Silva, C.J., Lee, B.G., Yambao, J.C., Erickson-Beltran, M.L., Quiñones, B. 2019. Using nanospray liquid chromatography and mass spectrometry to quantitate Shiga toxin production in environmental Escherichia coli recovered from a major produce production region in California. Journal of Agricultural and Food Chemistry. 67(5):1554-1562. https://doi.org/10.1021/acs.jafc.8b05324.