- Email: [email protected]
Discrimination among Listeria monocytogenes isolates using a mixed genome DNA microarray
Veterinary Microbiology 92 (2003) 351±362
Discrimination among Listeria monocytogenes isolates using a mixed genome DNA microarray Monica K. Boruckia,b,*, Melissa J. Krugb, Wayne T. Muraokab, Douglas R. Callb a
Animal Disease Research Unit, Agricultural Research Service, United States Department of Agriculture, Pullman, WA 99164-6630, USA b Department of Veterinary Microbiology and Pathology, College of Veterinary Medicine, Washington State University, Pullman, WA 99164-7040, USA Received 24 June 2002; received in revised form 22 October 2002; accepted 18 November 2002
Abstract Listeria monocytogenes can cause serious illness in humans, usually following the ingestion of contaminated food. Epidemiologic investigation requires identi®cation of speci®c isolates, usually done by a combination of serotyping and subtyping using pulsed-®eld gel electrophoresis (PFGE). DNA microarrays provide a new format to resolve genetic differences among isolates and, unlike PFGE, to identify speci®c genes associated with the infecting pathogen. A 585 probe, mixed genome microarray was constructed and 24 strains of L. monocytogenes were hybridized to the array. Microarray analysis allowed discrimination among L. monocytogenes isolates within a serotype and obtained from similar geographic and epidemiologic sources. Importantly, the microarray results preserved previously described phylogenetic relationships between major serogroups and, in a limited comparison, agreed with PFGE subtypes. The association of individual probes with isolates allowed identi®cation of speci®c genes. Sequencing of 10 polymorphic probes identi®ed nine matches with previously described bacterial genes including several suspected virulence factors. These results demonstrate that mixed genomic microarrays are useful for differentiating among closely related L. monocytogenes isolates and identifying genetic markers that can be used in epidemiologic and possibly pathogenesis studies. # 2002 Elsevier Science B.V. All rights reserved. Keywords: Listeria monocytogenes; Subtyping; Microarray; Virulence
* Corresponding author. Tel.: 1-509-335-7407; fax: 1-509-335-8328. E-mail address: [email protected] (M.K. Borucki).
0378-1135/02/$ ± see front matter # 2002 Elsevier Science B.V. All rights reserved. doi:10.1016/S0378-1135(02)00423-6
352
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
1. Introduction Listeria monocytogenes is a Gram positive, facultative intracellular bacterial pathogen that is ubiquitous in nature and capable of causing serious disease in humans and animals. L. monocytogenes infection of cattle and sheep can lead to abortion, central nervous system disease and death. Human listeriosis is a potentially fatal food-borne disease often associated with the consumption of contaminated dairy products. Although more than 14 serotypes of L. monocytogenes have been designated (Graves et al., 1999), only three serotypes (1/2a, 1/2b, and 4b) cause the vast majority of veterinary and human clinical cases (Tappero et al., 1995). Consequently, knowledge of the serotype alone is of limited value in identifying source of infection. Pulsed-®eld gel electrophoresis (PFGE) is the most widely used subtyping technique and a standardized protocol has been established, thereby allowing PFGE patterns to be compared via PulseNet (Graves and Swaminathan, 2001). Although PFGE is reproducible and relatively discriminatory, PFGE provides only limited genetic characterization of isolates and does not identify the presence or absence of speci®c genes. To address the need for a genotyping assay that can identify speci®c genetic differences among L. monocytogenes isolates, we developed a 585 probe DNA microarray. The array was constructed using a genomic library constructed from 10 different strains of L. monocytogenes. We hypothesized that a ``mixed genome'' array would allow discrimination among isolates of the same serotype and among isolates collected from epidemiologically and geographically similar sources. In the present study, we report the testing of this hypothesis and identify speci®c gene probes that are polymorphic and thus important for discriminating L. monocytogenes isolates. 2. Materials and methods 2.1. Bacterial strains A genomic library was constructed from four serotypes of L. monocytogenes. Isolates included four serotype 4b strains (CDC F5070, CDC F4565, ATCC 19115, and V8807 (isolated from ovine brain tissue)), four 1/2a strains (milk tank isolates M12716A, M35568A, M37952A, M32920A; each isolated from a different source and representing four different pulsovars), one 1/2c strain (CDC G3321) and one 3a strain (CDC G1127). The bacterial isolates characterized with the array are listed in Table 1. 2.2. Serotyping Listeria antiserum was obtained from Accurate Scienti®c (Westbury, NY, USA). Serotyping was performed according to the manufacturer's recommendations with the following modi®cations. Isolates were cultured in EB motility media (0.3% beef extract, 0.1% peptone, 0.5% NaCl, 0.4% agar; pH 7.4) prior to H-antigen determination (Food and Drug Administration, 1992). Bacteria from the outer edge of the motility media was used to inoculate 5 ml Luria Bertani medium and incubated at 308 C for 16 h. After harvesting cells by centrifugation, the pellet was resuspended in 400 ml 0.2% NaCl. Antigen (100 ml)
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
353
Table 1 Listeria monocytogenes strains used as genomic targets for microarray analysis Strain
Sourcea
Type
Serotype
Pulsovarb
M10867C M10867D M32490G M35568A M32771C M37952A M36046A M36582B M36509A H1445 H842 H1164 H9900104 G3321 H9333 H7973 H9900096 F2365 F5070 G1092 V8807 V013668A M36467A M35584A O51742
USDA USDA USDA USDA USDA USDA USDA USDA USDA WADOH WADOH WADOH WADOH CDC CDC CDC WADOH CDC CDC CDC USDA USDA USDA USDA ATCC
Bulk milk Bulk milk Bulk milk Bulk milk Bulk milk Bulk milk Bulk milk Bulk milk Bulk milk Human Human Human Human Human Human Human Environmental Human epidemic Human epidemic Human epidemic Ovine brain Bovine brain Bulk milk Bulk milk L. innocua
1/2a 1/2a 1/2a 1/2a 1/2a 1/2a 1/2a 1/2a 1/2a 1/2a 1/2b 1/2b 1/2b 1/2c 1/2c 1/2c 1/2c 4b 4b 4b 4b 4b 4c 4c
A A B B C D D D D
a Sample sources are as follows: United States Department of Agriculture (USDA), Pullman, WA; Washington State Department of Health (WASDOH), Shoreline, WA; Centers for Disease Control and Prevention (CDC), Atlanta, GA. b Isolates with identical PFGE ApaI endonuclease digestion profiles were considered to be the same pulsovar.
was combined with 40 ml antisera in a 6 50 mm culture tube and agglutination was observed after incubation for 1 h at 518 C. A slide agglutination test for O-antigen was performed according to the manufacture's recommendations with the exception that 20 ml of antigen was combined with 20 ml of antibody. 2.3. Development of array probes Genomic DNA was extracted from the 10 L. monocytogenes strains using an Easy DNA kit (Invitrogen, Carlsbad, CA, USA). DNA was quanti®ed by UV spectrophotometry and equal amounts of genomic DNA from each strain were mixed. This pooled genomic DNA was used to construct a random shotgun library (Amplicon Express, Pullman, WA, USA). Brie¯y, 10 mg DNA was cut with the restriction enzyme CviJI (Chimerx, Milwaukee, WI, USA) and fragments of approximately 600 bp were gel isolated, extracted and ligated into pUC18. Ligation products were transformed into E. coli and 6000 positive recombinant
354
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
clones were picked and arrayed into 96-well plates. Clone inserts were ampli®ed by polymerase chain reaction (PCR) using M13 primers (55 pmol each), 1.5 ml bacterial culture (template DNA), 4 units Taq with 1 reaction buffer (Fisher, Pittsburgh, PA, USA), 0.2 mM each dNTP (Eppendorf, Westbury, NY, USA), and 2.5 mM MgCl2 in 100 ml reaction volume. PCR cycle conditions were 95 8C for 5 min followed by 35 cycles of 95 8C for 30 s, 52 8C for 30 s, and 72 8C for 1 min, followed by 72 8C for 10 min after cycling was completed. Insert size was determined using gel electrophoresis (1% agarose). PCR products of correct size (500±1000 bp) were puri®ed using a Montage PCR96 Cleanup kit (Millipore Corp, Bedford, MA, USA) and stored at 20 8C until ready for printing. 2.4. Array construction Twelve-well, Te¯on masked slides (Erie Scienti®c, Portsmouth, NH, USA) were acid washed and derivatised with 3-glycidoxypropyltrimethoxysilane (Sigma±Aldrich, Milwaukee, WI, USA) (Call et al., 2001). Puri®ed PCR products were resuspended in print buffer (0.05% SDS, 50 mM NaOH, pH 12.0) at approximately 350 ng/ml followed by heat denaturing (95 8C for 5 min) in 96-well trays. PCR products were allowed to cool and were then spotted onto prepared glass slides (two arrays per slide) using a robotic arrayer (Model 417, Affymetrix, Santa Clara, CA, USA). The size and layout of the Te¯on masking necessitated spotting unique subarrays into each of four masked wells. Two probes (biotinylated 25-mer oligonucleotides, 5 mM) were added to each well to serve as positive controls for detection chemistry. Slides were then baked 1 h at 130 8C in a vacuum oven followed by long-term storage at room temperature. 2.5. Genomic target preparation and hybridization Genomic DNA was extracted from target strains using a DNeasy Tissue kit (Qiagen, Valencia, CA, USA) and quanti®ed using UV spectrophotometry. Genomic DNA (1 mg) was nick-translated in the presence of biotin-dATP following the manufacturer's instructions (BioNick Labeling System, Invitrogen). Labeled DNA was then ethanol precipitated, resuspended in 210 ml hybridization buffer consisting of 4X SSC (60 mM NaCl, 0.6 mM Na-citrate, pH 7.0) and 5 Denhardt's solution (0.1% Ficoll, 0.1% polyvinylpyrrolidone, 0.1% bovine serum albumin). Slides were pre-blocked at 23 8C for 30 min with TNB buffer (100 mM Tris±HCl, pH 7.5, 150 mM NaCl, 0.5% blocking reagent [TSA Biotin System, Perkin-Elmer, Boston, MA, USA]). For each isolate being tested, labeled DNA was heat denatured (95 8C for 3 min), divided between eight wells (two replicates of each subarray) and hybridized overnight at 55 8C in humidi®ed chambers. DNA was removed by aspiration and slides were sequentially washed at 55 8C for 4 min (1 SSC, 0.2% SDS), 23 8C for 4 min (0.1 SSC, 0.2% SDS), 23 8C for 4 min (0.1 SSC) (Hegde et al., 2000), and three times at 23 8C for 1 min (TNT buffer; 100 mM Tris±HCl, pH 7.5, 150 mM NaCl, 0.05% Tween 20). All washes included agitation. Subsequent wash steps used three washes (1 min) in TNT and all subsequent manipulations occurred at approximately 23 8C. Streptavidin conjugated to horseradish peroxidase (1:100 in TNB, TSA Biotin System) was incubated 30 min on each slide followed by washing and incubation with 10% equine serum (Sigma±Aldrich) in 2 SSC (30 min). Biotinyl Tyramide (1:50 in
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
355
ampli®cation buffer, TSA Biotin System) was then incubated on each slide (10 min) followed by washing and a 1 h incubation with 2 mg/ml streptavidin conjugated to Alexa Fluor 546 (Molecular Probes, Eugene, OR, USA) in 1 SSC and 5 Denhardt's solution. Slides were given a ®nal wash followed by drying and imaging using an Applied Precision ArrayWoRx scanner (Issaquah, WA, USA). The scanner was equipped with 548 nm excitation and 595 nm emission ®lters (0.5±1.0 s exposures). Probes with hybridized targets appeared as spots with pixel values ranging from 100 to 12,000 optical density units. 2.6. Microarray image analysis Spot software (CSIRO Mathematical and Information Sciences, North Ryde, Australia) was used to quantify signal intensity. The ®nal output included median pixel values that were used in this analysis. All slides that produced total probe intensities that deviated from a normal probability distribution were rejected from further analysis. Because the entire array was subdivided into four separate wells, each probe was normalized by dividing signal intensity by the average signal intensity for all probes found in that well. In most cases, this simple normalization permitted direct comparisons between microarray experiments. Normalized data was averaged for replicate probes and all data was managed using a relational database (MS Access, Microsoft Corp., Redmond, WA, USA). The frequency distribution of hybridization signals was examined for each probe (n 585) across all samples being tested (n 27). Histograms for normalized intensities were constructed (Fig. 1) using NCSS 2001 software (Number Cruncher Statistical Systems, Kaysville, UT, USA). Each histogram presented one of four possible distributions. Two distributions represented the extremes where either minimal signal was generated for all samples, or all samples generated signal near maximum (Fig. 1A and B). In other cases, signal intensity was normally or uniformly distributed between minimum and maximum signal intensities (Fig. 1C). Finally, signal intensity was distributed in a bi-modal fashion for a subset of probes (Fig. 1D). The analysis was limited to the latter distribution because clear threshold values could be assigned for each probe having a bi-modal signal distribution. All histograms having a bi-modal distribution were examined individually and intensity thresholds were selected that clearly delineated between the two peaks in the distribution. This was done without knowledge of how speci®c samples were placed in the histogram. After assigning probe speci®c thresholds, array data was converted into a binary matrix where hybridization intensity was considered high (1) or low (0) based on the assigned threshold. 2.7. Data analysis The binary matrix was processed with PAUP (version 4.0b8a, Sinauer Associates Inc., Sunderland, MA, USA) and the unweighted pair group method using arithmetic averages (UPGMA) was used to construct a dendrogram that summarized genetic relationships between samples. Bootstrap con®dence values (n 1000 iterations) were calculated for the nodes in the dendrogram and TreeView (Page, 1996) was used to produce the ®nal dendrogram ®gure. Data were also examined using a spreadsheet (Microsoft Excel) to identify probes that consistently discriminated between various dendrogram clusters.
356 M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362 Fig. 1. Example of four frequency distributions for normalized hybridization data. The x-axis shows normalized intensity values and the y-axis is the number of samples. Spots can have either low (A) or high (B) signal intensity for all samples or a spot may have a uniform or normally distributed spot intensities ranging between the high and low values (C). Finally, spots may have a clear bi-modal distribution (D) for which thresholds can be clearly assigned to distinguish between low and high signal intensities.
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
357
Clones for these probes were retrieved from the original library plates and sequenced using two-pass automated sequencing (Amplicon Express). 2.8. Pulsed-field gel electrophoresis (PFGE) Strain typing using a 30-h PFGE protocol was performed as previously described (Graves and Swaminathan, 2001) except that genomic DNA was digested with 20 U ApaI instead of 160 U. 3. Results and discussion A mixed genome microarray was constructed using a genomic library based on 10 different strains of L. monocytogenes. Twenty-four isolates of L. monocytogenes were then hybridized to replicate arrays (Table 1). Although some spot-to-spot variation was evident from the printing and hybridization procedures, most probes were easily detected and quanti®ed using Spot software (Fig. 2). Each array was composed of 585 probes (not including control probes), of which 92 probes (15.7%) were identi®ed as having bi-modal distributions and being polymorphic for at least two L. monocytogenes isolates. This set of 92 probes was used to distinguish genetic differences among isolates using binary scoring. The binary matrix was used to generate a UPGMA dendogram (Fig. 3). Serotypes 1/2a and 1/2c formed cluster I while serotypes 1/2b, 4b, and 4c formed cluster II. These two primary ``sero-clusters'' have been previously described based on other molecular techniques (Piffaretti et al., 1989; Bibb et al., 1990; Norrung and Skovgaard, 1993; Brosch et al., 1994; Rasmussen et al., 1995), and thus, these results are consistent with previous observations. The phylogenetic subclusters de®ned by the microarray analysis were not necessarily limited to a single serotype (subcluster Ib, Fig. 3). This is not surprising because serotypes are determined by somatic and ¯agellar antigenic variation whereas microarray analysis samples a greater diversity of the genome. To verify that assay replicates yielded similar results, genomic DNA from isolate V013368A was extracted, puri®ed, labeled, and hybridized in three separate experiments. Different laboratory technicians prepared each replicate DNA sample on different dates. A total of six individual arrays were processed with this target (three sets of duplicate arrays). As expected, the replicates of V013368 all clustered within a single group (Fig. 3). To test the hypothesis that the microarray could discriminate among L. monocytogenes isolates of the same serotype and similar source, nine serotype 1/2a isolates obtained from Paci®c Northwest bulk milk samples were characterized by microarray analysis (Table 1, Fig. 3). Microarray subtyping grouped the milk isolates according to serotype (Fig. 3) and divided the serotype 1/2a milk isolates into two subclusters (Ia and Ic) based on 10 probe differences. Additionally, genetic subtypes present within a subcluster were resolved by microarray analysis. For example, two milk isolate groups were identi®ed within subcluster Ia with two probe differences and a robust bootstrap value differentiating isolates M32490G and M35568A from the other Ia milk isolates (Fig. 3). The major clusters and subclusters of L. monocytogenes isolates were clearly de®ned by strongly polymorphic probes. Thus, while the microarray grouping generally agreed with
358
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
Fig. 2. Example of spot patterns after image segmentation with Spot software. No probes were present in the upper right quadrant of these arrays and only a portion of the full array is shown in these images. Spots with little or no hybridization signal (e.g. top panel, row 8, column 13) yielded small or irregular shapes whereas spots with clear hybridization signal were fully encircled (white border) regardless of spot shape or size. Hybridization of another genome to a replicate array (lower panel) produced a different hybridization pattern. For example, two probes in row 1 (columns 2 and 6) were clearly present in the upper panel, but absent in the corresponding locations for the lower panel.
the PFGE analysis and was of comparable resolution, the microarray analysis identi®ed speci®c polymorphic probe sequences. Four unique sequences (probes 317, 369, 791, 955) were present in all members of cluster I but no members of cluster II, while one unique sequence (probe 302) was present in all members of cluster II but no members of cluster I
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
359
Fig. 3. UPGMA representation (PAUP version 4.0b8a) of genetic relationships between 24 isolates of Listeria monocytogenes based on hybridization patterns derived from a 585 probe DNA microarray. Bootstrap values (nrep 1000) are indicated for major nodes. Sample V013668A was tested for processing and analysis reproducibility in three separate experiments (V013668Ar1±3).
(Table 2). Additionally, three subclusters (Ia, Ib, Ic) were present within cluster I, one of which was composed almost entirely of 1/2c isolates (Ib). Subclusters Ib and Ic also shared six unique sequences (probes 103, 134, 150, 367, 1072, 1092) with cluster II (Table 2). Eleven cluster-associated probes and four other polymorphic probes were sequenced to identify potential genetic markers and to assess the level of probe redundancy for the array. Thirteen of the probes had less than 87% nucleotide sequence similarity. Probes 134 and 1072 shared 99.7% nucleotide sequence identity and were considered redundant. A BLAST
360
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
Table 2 Microarray probes associated with L. monocytogenes clusters Ia, Ib, Ic, and II Strain (cluster)
Microarray probe identifier 103
M37952A (Ia) M36046A (Ia) M36582B (Ia) M32490G (Ia) M35568A (Ia) M32771C (Ia) M36509A (Ia) H1445 (Ib) G3321 (Ib) H9333 (Ib) H7973 (Ib) H9900096 (Ib) M10867C (Ic) M10867D (Ic) H842 (II) H1164 (II) H9900104 (II) F2365 (II) F5070 (II) G1092 (II) V8807 (II) V013668A (II) M36467A (II) M35584A (II) O51742b a b
a
0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0
134
150
367
1072
1092
317
369
791
955
302
0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0
0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0
0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0
0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0
0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1
0 low signal intensity; 1 high signal intensity. L. innocua.
search based on putative protein translations matched six cluster speci®c sequences with putative genes including several suspected virulence factors (Table 3). Four of the probe sequences were most similar to internalin proteins. In particular, sequences from probes 103, 134/1072, and 367 were similar to InlE, whereas probe 1069 coded for a LPTXG motif characteristic of many surface-associated internalins (Vazquez-Boland et al., 2001). A BLASTn search revealed that three probe sequences (103, 134/1072, and 367) share identity with the same region of the L. monocytogenes genome and are most likely sequence variants of the same ORF. Probe 302, unique to cluster II, had sequence similarity to L. monocytogenes ferrous iron transport protein B. Probe 955 had greatest similarity to PTS system beta-glucoside-speci®c enzyme II that has been implicated in virulence (Brehm et al., 1999; Glaser et al., 2001). One sequence (probe 192) was only found in three strains and it may be related to phage mediated, horizontal gene transfer (Table 3). Although microarray analysis of L. monocytogenes isolates did identify several polymorphic probes with sequence similarity to virulence factors, the signi®cance of this ®nding is unclear. In particular, the L. monocytogenes (strain EGD-e) genome contains 41 open reading frames encoding an LPXTG motif and 39 genes for PTS system enzyme II
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
361
Table 3 Summary of the select probe sequences from the L. monocytogenes microarray Probe no. (accession no.)
Tally of positive samples by cluster Ia
Ib/Ic
II
103 134 150 367 1072 1092 369 955
(BH170514) (BH170515) (BH170516) (BH170520) (BH170526) (BH170528) (BH175021) (BH170524)
0 0 0 0 0 0 11 11
8 8 8 8 8 8 8 8
12 12 12 12 12 12 0 0
791 1069 317 192 302 950 1085
(BH170522) (BH170525) (BH170519) (BH170517) (BH170518) (BH170523) (BH170527)
11 11 11 0 0 4 2
8 8 8 3 0 5 5
0 0 1 0 12 11 12
a
Similar proteinsa (E score)
Membrane associated lipoprotein, L. monocytogenes (3e-94) Membrane associated lipoprotein, L. monocytogenes (e-108)a Transcription regulator, L. monocytogenes (e-101) Membrane associated lipoprotein, L. monocytogenes (2e-91) Membrane associated lipoprotein, L. monocytogenes (e-108)a Hypothetical protein lmo0461, L. monocytogenes (5e-70) 3-Isopropylmalate dehydrogenase, L. monocytogenes (e-101) PTS system, beta-glucoside-specific enzyme IIABC component, L. monocytogenes (1e-66) Hypothetical protein lmo0904, L. monocytogenes (e-115) Internalin proteins, L. monocytogenes (2e-82 and below) ABC transporter, L. monocytogenes (2e-49) Integrase (phage-related protein), Bacillus halodurans (3e-43) Ferrous iron transport protein B, L. monocytogenes (1e-96) Polypeptide deformylase, L. monocytogenes (1e-53) Gamma-glutamyl kinase, L. monocytogenes (2e-27)
Redundant probes.
permeases (Glaser et al., 2001). Therefore, a more thorough analysis is needed including the typing of many more strains of known phenotype (virulent versus attenuated) on a microarray constructed from a greater number and diversity of polymorphic probes. 4. Conclusion A 585 probe mixed genome array was used to differentiate 24 L. monocytogenes isolates into two sero-clusters that agree with previously described phylogenetic lineages. Subclusters were also identi®ed within serotypic groups and among isolates from similar epidemiologic and geographic sources. Ten cluster-associated sequences were identi®ed, including ®ve that had similarity to suspected virulence factors. Increasing the number of representative strains used to construct the library while increasing the size of the array will likely increase both subtyping resolution and the number of cluster-associated sequences identi®ed. Cluster-associated sequences can then be used to design speci®c PCR primers for rapid L. monocytogenes strain detection and identi®cation. Identi®cation of speci®c genetic differences among isolates with de®ned phenotypes will provide insight into the epidemiology and potentially the pathogenesis of L. monocytogenes. Acknowledgements We thank Peggy Hayes at CDC for providing bacterial strains and serotyping assistance, Jinxin Hu at Washington State Department of Health for providing bacterial strains,
362
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
Myrvie Fuentes and Josiel Lopez for DNA extraction, James Reynolds for assistance with PFGE analysis, Glen Scoles for assistance with phylogenetic analysis, and Donald Knowles and Thomas Besser for helpful discussions. We thank Guy Palmer for critical review of the manuscript. Funding was provided by USDA Agricultural Research Service CWU 5348-32000-017-00D and by the Agricultural Animal Health Program (College of Veterinary Medicine, Washington State University). References Bibb, W.F., Gellin, B.G., Schwartz, B., Plikaytis, B.D., Reeves, M.W., Pinner, R.W., Broome, C.V., 1990. Analysis of clinical and food-borne isolates of Listeria monocytogenes in the United States by multilocus enzyme electrophoresis and application of the method to epidemiologic investigations. Appl. Environ. Microbiol. 56, 2133±2141. Brehm, K., Ripio, M.T., Kreft, J., Vazquez-Boland, J.A., 1999. The bvr locus of Listeria monocytogenes mediates virulence gene repression by beta-glucosides. J. Bacteriol. 181, 5024±5032. Brosch, R., Chen, J., Luchansky, J.B., 1994. Pulsed-field fingerprinting of listeriae: identification of genomic divisions for Listeria monocytogenes and their correlation with serovar. Appl. Environ. Microbiol. 60, 2584± 2592. Call, D.R., Chandler, D.P., Brockman, F.J., 2001. Fabrication of DNA microarrays using unmodified oligonucleotide probes. BioTechniques 30, 368±379. FDA, 1992. Bacteriological Analytical Manual, 7th ed. Glaser, P., Frangeul, L., Buchrieser, C., Rusniok, C., Amend, A., Baquero, F., Berche, P., Bloecker, H., Brandt, P., Chakraborty, T., Charbit, A., Chetouani, F., Couve, E., de Daruvar, A., Dehoux, P., Domann, E., Dominguez-Bernal, G., Duchaud, E., Durant, L., Dussurget, O., Entian, K.D., Fsihi, H., Portillo, F.G., Garrido, P., Gautier, L., Goebel, W., Gomez-Lopez, N., Hain, T., Hauf, J., Jackson, D., Jones, L.M., Kaerst, U., Kreft, J., Kuhn, M., Kunst, F., Kurapkat, G., Madueno, E., Maitournam, A., Vicente, J.M., Ng, E., Nedjari, H., Nordsiek, G., Novella, S., de Pablos, B., Perez-Diaz, J.C., Purcell, R., Remmel, B., Rose, M., Schlueter, T., Simoes, N., Tierrez, A., Vazquez-Boland, J.A., Voss, H., Wehland, J., Cossart, P., 2001. Comparative genomics of Listeria species. Science 294, 849±852. Graves, L.M., Swaminathan, B., 2001. PulseNet standardized protocol for subtyping Listeria monocytogenes by macrorestriction and pulsed-field gel electrophoresis. Int. J. Food Microbiol. 65, 55±62. Graves, L.M., Swaminathan, B., Hunter, S., 1999. Subtyping Listeria monocytogenes. In: Ryser, E.T., Marth, E.H. (Eds.), Listeria, Listeriosis, and Food Safety. Marcel Dekker, New York, pp. 279±298. Hegde, P., Qi, R., Abernathy, K., Gay, C., Dharap, S., Gaspard, R., Hughes, J., Snesrud, E.N., Less, N., Quackenbush, J., 2000. A concise guide to cDNA microarray analysis. BioTechniques 29, 548±562. Norrung, B., Skovgaard, N., 1993. Application of multilocus enzyme electrophoresis in studies of the epidemiology of Listeria monocytogenes in Denmark. Appl. Environ. Microbiol. 59, 2817±2822. Page, R.D.M., 1996. TreeView: an application to display phylogenetic trees on personal computers. Comput. Applic. Biosci. 12, 357±358. Piffaretti, J.C., Kressebuch, H., Aeschbacher, M., Bille, J., Bannerman, E., Musser, J.M., Selander, R.K., Rocourt, J., 1989. Genetic characterization of clones of the bacterium Listeria monocytogenes causing epidemic disease. Proc. Natl. Acad. Sci. U.S.A. 86, 3818±3822. Rasmussen, O.F., Skouboe, P., Dons, L., Rossen, L., Olsen, J.E., 1995. Listeria monocytogenes exists in at least three evolutionary lines: evidence from flagellin, invasive associated protein and listeriolysin O genes. Microbiology 141, 2053±2061. Tappero, J.W., Schuchat, A., Deaver, K.A., Mascola, L., Wenger, J.D., 1995. Reduction in the incidence of human listeriosis in the United States. Effectiveness of prevention efforts? The Listeriosis Study Group. J. Am. Med. Assoc. 273, 1118±1122. Vazquez-Boland, J.A., Kuhn, M., Berche, P., Chakraborty, T., Dominguez-Bernal, G., Goebel, W., GonzalezZorn, B., Wehland, J., Kreft, J., 2001. Listeria pathogenesis and molecular virulence determinants. Clin. Microbiol. Rev. 14, 584±640.
Discrimination among Listeria monocytogenes isolates using a mixed genome DNA microarray Monica K. Boruckia,b,*, Melissa J. Krugb, Wayne T. Muraokab, Douglas R. Callb a
Animal Disease Research Unit, Agricultural Research Service, United States Department of Agriculture, Pullman, WA 99164-6630, USA b Department of Veterinary Microbiology and Pathology, College of Veterinary Medicine, Washington State University, Pullman, WA 99164-7040, USA Received 24 June 2002; received in revised form 22 October 2002; accepted 18 November 2002
Abstract Listeria monocytogenes can cause serious illness in humans, usually following the ingestion of contaminated food. Epidemiologic investigation requires identi®cation of speci®c isolates, usually done by a combination of serotyping and subtyping using pulsed-®eld gel electrophoresis (PFGE). DNA microarrays provide a new format to resolve genetic differences among isolates and, unlike PFGE, to identify speci®c genes associated with the infecting pathogen. A 585 probe, mixed genome microarray was constructed and 24 strains of L. monocytogenes were hybridized to the array. Microarray analysis allowed discrimination among L. monocytogenes isolates within a serotype and obtained from similar geographic and epidemiologic sources. Importantly, the microarray results preserved previously described phylogenetic relationships between major serogroups and, in a limited comparison, agreed with PFGE subtypes. The association of individual probes with isolates allowed identi®cation of speci®c genes. Sequencing of 10 polymorphic probes identi®ed nine matches with previously described bacterial genes including several suspected virulence factors. These results demonstrate that mixed genomic microarrays are useful for differentiating among closely related L. monocytogenes isolates and identifying genetic markers that can be used in epidemiologic and possibly pathogenesis studies. # 2002 Elsevier Science B.V. All rights reserved. Keywords: Listeria monocytogenes; Subtyping; Microarray; Virulence
* Corresponding author. Tel.: 1-509-335-7407; fax: 1-509-335-8328. E-mail address: [email protected] (M.K. Borucki).
0378-1135/02/$ ± see front matter # 2002 Elsevier Science B.V. All rights reserved. doi:10.1016/S0378-1135(02)00423-6
352
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
1. Introduction Listeria monocytogenes is a Gram positive, facultative intracellular bacterial pathogen that is ubiquitous in nature and capable of causing serious disease in humans and animals. L. monocytogenes infection of cattle and sheep can lead to abortion, central nervous system disease and death. Human listeriosis is a potentially fatal food-borne disease often associated with the consumption of contaminated dairy products. Although more than 14 serotypes of L. monocytogenes have been designated (Graves et al., 1999), only three serotypes (1/2a, 1/2b, and 4b) cause the vast majority of veterinary and human clinical cases (Tappero et al., 1995). Consequently, knowledge of the serotype alone is of limited value in identifying source of infection. Pulsed-®eld gel electrophoresis (PFGE) is the most widely used subtyping technique and a standardized protocol has been established, thereby allowing PFGE patterns to be compared via PulseNet (Graves and Swaminathan, 2001). Although PFGE is reproducible and relatively discriminatory, PFGE provides only limited genetic characterization of isolates and does not identify the presence or absence of speci®c genes. To address the need for a genotyping assay that can identify speci®c genetic differences among L. monocytogenes isolates, we developed a 585 probe DNA microarray. The array was constructed using a genomic library constructed from 10 different strains of L. monocytogenes. We hypothesized that a ``mixed genome'' array would allow discrimination among isolates of the same serotype and among isolates collected from epidemiologically and geographically similar sources. In the present study, we report the testing of this hypothesis and identify speci®c gene probes that are polymorphic and thus important for discriminating L. monocytogenes isolates. 2. Materials and methods 2.1. Bacterial strains A genomic library was constructed from four serotypes of L. monocytogenes. Isolates included four serotype 4b strains (CDC F5070, CDC F4565, ATCC 19115, and V8807 (isolated from ovine brain tissue)), four 1/2a strains (milk tank isolates M12716A, M35568A, M37952A, M32920A; each isolated from a different source and representing four different pulsovars), one 1/2c strain (CDC G3321) and one 3a strain (CDC G1127). The bacterial isolates characterized with the array are listed in Table 1. 2.2. Serotyping Listeria antiserum was obtained from Accurate Scienti®c (Westbury, NY, USA). Serotyping was performed according to the manufacturer's recommendations with the following modi®cations. Isolates were cultured in EB motility media (0.3% beef extract, 0.1% peptone, 0.5% NaCl, 0.4% agar; pH 7.4) prior to H-antigen determination (Food and Drug Administration, 1992). Bacteria from the outer edge of the motility media was used to inoculate 5 ml Luria Bertani medium and incubated at 308 C for 16 h. After harvesting cells by centrifugation, the pellet was resuspended in 400 ml 0.2% NaCl. Antigen (100 ml)
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
353
Table 1 Listeria monocytogenes strains used as genomic targets for microarray analysis Strain
Sourcea
Type
Serotype
Pulsovarb
M10867C M10867D M32490G M35568A M32771C M37952A M36046A M36582B M36509A H1445 H842 H1164 H9900104 G3321 H9333 H7973 H9900096 F2365 F5070 G1092 V8807 V013668A M36467A M35584A O51742
USDA USDA USDA USDA USDA USDA USDA USDA USDA WADOH WADOH WADOH WADOH CDC CDC CDC WADOH CDC CDC CDC USDA USDA USDA USDA ATCC
Bulk milk Bulk milk Bulk milk Bulk milk Bulk milk Bulk milk Bulk milk Bulk milk Bulk milk Human Human Human Human Human Human Human Environmental Human epidemic Human epidemic Human epidemic Ovine brain Bovine brain Bulk milk Bulk milk L. innocua
1/2a 1/2a 1/2a 1/2a 1/2a 1/2a 1/2a 1/2a 1/2a 1/2a 1/2b 1/2b 1/2b 1/2c 1/2c 1/2c 1/2c 4b 4b 4b 4b 4b 4c 4c
A A B B C D D D D
a Sample sources are as follows: United States Department of Agriculture (USDA), Pullman, WA; Washington State Department of Health (WASDOH), Shoreline, WA; Centers for Disease Control and Prevention (CDC), Atlanta, GA. b Isolates with identical PFGE ApaI endonuclease digestion profiles were considered to be the same pulsovar.
was combined with 40 ml antisera in a 6 50 mm culture tube and agglutination was observed after incubation for 1 h at 518 C. A slide agglutination test for O-antigen was performed according to the manufacture's recommendations with the exception that 20 ml of antigen was combined with 20 ml of antibody. 2.3. Development of array probes Genomic DNA was extracted from the 10 L. monocytogenes strains using an Easy DNA kit (Invitrogen, Carlsbad, CA, USA). DNA was quanti®ed by UV spectrophotometry and equal amounts of genomic DNA from each strain were mixed. This pooled genomic DNA was used to construct a random shotgun library (Amplicon Express, Pullman, WA, USA). Brie¯y, 10 mg DNA was cut with the restriction enzyme CviJI (Chimerx, Milwaukee, WI, USA) and fragments of approximately 600 bp were gel isolated, extracted and ligated into pUC18. Ligation products were transformed into E. coli and 6000 positive recombinant
354
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
clones were picked and arrayed into 96-well plates. Clone inserts were ampli®ed by polymerase chain reaction (PCR) using M13 primers (55 pmol each), 1.5 ml bacterial culture (template DNA), 4 units Taq with 1 reaction buffer (Fisher, Pittsburgh, PA, USA), 0.2 mM each dNTP (Eppendorf, Westbury, NY, USA), and 2.5 mM MgCl2 in 100 ml reaction volume. PCR cycle conditions were 95 8C for 5 min followed by 35 cycles of 95 8C for 30 s, 52 8C for 30 s, and 72 8C for 1 min, followed by 72 8C for 10 min after cycling was completed. Insert size was determined using gel electrophoresis (1% agarose). PCR products of correct size (500±1000 bp) were puri®ed using a Montage PCR96 Cleanup kit (Millipore Corp, Bedford, MA, USA) and stored at 20 8C until ready for printing. 2.4. Array construction Twelve-well, Te¯on masked slides (Erie Scienti®c, Portsmouth, NH, USA) were acid washed and derivatised with 3-glycidoxypropyltrimethoxysilane (Sigma±Aldrich, Milwaukee, WI, USA) (Call et al., 2001). Puri®ed PCR products were resuspended in print buffer (0.05% SDS, 50 mM NaOH, pH 12.0) at approximately 350 ng/ml followed by heat denaturing (95 8C for 5 min) in 96-well trays. PCR products were allowed to cool and were then spotted onto prepared glass slides (two arrays per slide) using a robotic arrayer (Model 417, Affymetrix, Santa Clara, CA, USA). The size and layout of the Te¯on masking necessitated spotting unique subarrays into each of four masked wells. Two probes (biotinylated 25-mer oligonucleotides, 5 mM) were added to each well to serve as positive controls for detection chemistry. Slides were then baked 1 h at 130 8C in a vacuum oven followed by long-term storage at room temperature. 2.5. Genomic target preparation and hybridization Genomic DNA was extracted from target strains using a DNeasy Tissue kit (Qiagen, Valencia, CA, USA) and quanti®ed using UV spectrophotometry. Genomic DNA (1 mg) was nick-translated in the presence of biotin-dATP following the manufacturer's instructions (BioNick Labeling System, Invitrogen). Labeled DNA was then ethanol precipitated, resuspended in 210 ml hybridization buffer consisting of 4X SSC (60 mM NaCl, 0.6 mM Na-citrate, pH 7.0) and 5 Denhardt's solution (0.1% Ficoll, 0.1% polyvinylpyrrolidone, 0.1% bovine serum albumin). Slides were pre-blocked at 23 8C for 30 min with TNB buffer (100 mM Tris±HCl, pH 7.5, 150 mM NaCl, 0.5% blocking reagent [TSA Biotin System, Perkin-Elmer, Boston, MA, USA]). For each isolate being tested, labeled DNA was heat denatured (95 8C for 3 min), divided between eight wells (two replicates of each subarray) and hybridized overnight at 55 8C in humidi®ed chambers. DNA was removed by aspiration and slides were sequentially washed at 55 8C for 4 min (1 SSC, 0.2% SDS), 23 8C for 4 min (0.1 SSC, 0.2% SDS), 23 8C for 4 min (0.1 SSC) (Hegde et al., 2000), and three times at 23 8C for 1 min (TNT buffer; 100 mM Tris±HCl, pH 7.5, 150 mM NaCl, 0.05% Tween 20). All washes included agitation. Subsequent wash steps used three washes (1 min) in TNT and all subsequent manipulations occurred at approximately 23 8C. Streptavidin conjugated to horseradish peroxidase (1:100 in TNB, TSA Biotin System) was incubated 30 min on each slide followed by washing and incubation with 10% equine serum (Sigma±Aldrich) in 2 SSC (30 min). Biotinyl Tyramide (1:50 in
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
355
ampli®cation buffer, TSA Biotin System) was then incubated on each slide (10 min) followed by washing and a 1 h incubation with 2 mg/ml streptavidin conjugated to Alexa Fluor 546 (Molecular Probes, Eugene, OR, USA) in 1 SSC and 5 Denhardt's solution. Slides were given a ®nal wash followed by drying and imaging using an Applied Precision ArrayWoRx scanner (Issaquah, WA, USA). The scanner was equipped with 548 nm excitation and 595 nm emission ®lters (0.5±1.0 s exposures). Probes with hybridized targets appeared as spots with pixel values ranging from 100 to 12,000 optical density units. 2.6. Microarray image analysis Spot software (CSIRO Mathematical and Information Sciences, North Ryde, Australia) was used to quantify signal intensity. The ®nal output included median pixel values that were used in this analysis. All slides that produced total probe intensities that deviated from a normal probability distribution were rejected from further analysis. Because the entire array was subdivided into four separate wells, each probe was normalized by dividing signal intensity by the average signal intensity for all probes found in that well. In most cases, this simple normalization permitted direct comparisons between microarray experiments. Normalized data was averaged for replicate probes and all data was managed using a relational database (MS Access, Microsoft Corp., Redmond, WA, USA). The frequency distribution of hybridization signals was examined for each probe (n 585) across all samples being tested (n 27). Histograms for normalized intensities were constructed (Fig. 1) using NCSS 2001 software (Number Cruncher Statistical Systems, Kaysville, UT, USA). Each histogram presented one of four possible distributions. Two distributions represented the extremes where either minimal signal was generated for all samples, or all samples generated signal near maximum (Fig. 1A and B). In other cases, signal intensity was normally or uniformly distributed between minimum and maximum signal intensities (Fig. 1C). Finally, signal intensity was distributed in a bi-modal fashion for a subset of probes (Fig. 1D). The analysis was limited to the latter distribution because clear threshold values could be assigned for each probe having a bi-modal signal distribution. All histograms having a bi-modal distribution were examined individually and intensity thresholds were selected that clearly delineated between the two peaks in the distribution. This was done without knowledge of how speci®c samples were placed in the histogram. After assigning probe speci®c thresholds, array data was converted into a binary matrix where hybridization intensity was considered high (1) or low (0) based on the assigned threshold. 2.7. Data analysis The binary matrix was processed with PAUP (version 4.0b8a, Sinauer Associates Inc., Sunderland, MA, USA) and the unweighted pair group method using arithmetic averages (UPGMA) was used to construct a dendrogram that summarized genetic relationships between samples. Bootstrap con®dence values (n 1000 iterations) were calculated for the nodes in the dendrogram and TreeView (Page, 1996) was used to produce the ®nal dendrogram ®gure. Data were also examined using a spreadsheet (Microsoft Excel) to identify probes that consistently discriminated between various dendrogram clusters.
356 M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362 Fig. 1. Example of four frequency distributions for normalized hybridization data. The x-axis shows normalized intensity values and the y-axis is the number of samples. Spots can have either low (A) or high (B) signal intensity for all samples or a spot may have a uniform or normally distributed spot intensities ranging between the high and low values (C). Finally, spots may have a clear bi-modal distribution (D) for which thresholds can be clearly assigned to distinguish between low and high signal intensities.
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
357
Clones for these probes were retrieved from the original library plates and sequenced using two-pass automated sequencing (Amplicon Express). 2.8. Pulsed-field gel electrophoresis (PFGE) Strain typing using a 30-h PFGE protocol was performed as previously described (Graves and Swaminathan, 2001) except that genomic DNA was digested with 20 U ApaI instead of 160 U. 3. Results and discussion A mixed genome microarray was constructed using a genomic library based on 10 different strains of L. monocytogenes. Twenty-four isolates of L. monocytogenes were then hybridized to replicate arrays (Table 1). Although some spot-to-spot variation was evident from the printing and hybridization procedures, most probes were easily detected and quanti®ed using Spot software (Fig. 2). Each array was composed of 585 probes (not including control probes), of which 92 probes (15.7%) were identi®ed as having bi-modal distributions and being polymorphic for at least two L. monocytogenes isolates. This set of 92 probes was used to distinguish genetic differences among isolates using binary scoring. The binary matrix was used to generate a UPGMA dendogram (Fig. 3). Serotypes 1/2a and 1/2c formed cluster I while serotypes 1/2b, 4b, and 4c formed cluster II. These two primary ``sero-clusters'' have been previously described based on other molecular techniques (Piffaretti et al., 1989; Bibb et al., 1990; Norrung and Skovgaard, 1993; Brosch et al., 1994; Rasmussen et al., 1995), and thus, these results are consistent with previous observations. The phylogenetic subclusters de®ned by the microarray analysis were not necessarily limited to a single serotype (subcluster Ib, Fig. 3). This is not surprising because serotypes are determined by somatic and ¯agellar antigenic variation whereas microarray analysis samples a greater diversity of the genome. To verify that assay replicates yielded similar results, genomic DNA from isolate V013368A was extracted, puri®ed, labeled, and hybridized in three separate experiments. Different laboratory technicians prepared each replicate DNA sample on different dates. A total of six individual arrays were processed with this target (three sets of duplicate arrays). As expected, the replicates of V013368 all clustered within a single group (Fig. 3). To test the hypothesis that the microarray could discriminate among L. monocytogenes isolates of the same serotype and similar source, nine serotype 1/2a isolates obtained from Paci®c Northwest bulk milk samples were characterized by microarray analysis (Table 1, Fig. 3). Microarray subtyping grouped the milk isolates according to serotype (Fig. 3) and divided the serotype 1/2a milk isolates into two subclusters (Ia and Ic) based on 10 probe differences. Additionally, genetic subtypes present within a subcluster were resolved by microarray analysis. For example, two milk isolate groups were identi®ed within subcluster Ia with two probe differences and a robust bootstrap value differentiating isolates M32490G and M35568A from the other Ia milk isolates (Fig. 3). The major clusters and subclusters of L. monocytogenes isolates were clearly de®ned by strongly polymorphic probes. Thus, while the microarray grouping generally agreed with
358
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
Fig. 2. Example of spot patterns after image segmentation with Spot software. No probes were present in the upper right quadrant of these arrays and only a portion of the full array is shown in these images. Spots with little or no hybridization signal (e.g. top panel, row 8, column 13) yielded small or irregular shapes whereas spots with clear hybridization signal were fully encircled (white border) regardless of spot shape or size. Hybridization of another genome to a replicate array (lower panel) produced a different hybridization pattern. For example, two probes in row 1 (columns 2 and 6) were clearly present in the upper panel, but absent in the corresponding locations for the lower panel.
the PFGE analysis and was of comparable resolution, the microarray analysis identi®ed speci®c polymorphic probe sequences. Four unique sequences (probes 317, 369, 791, 955) were present in all members of cluster I but no members of cluster II, while one unique sequence (probe 302) was present in all members of cluster II but no members of cluster I
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
359
Fig. 3. UPGMA representation (PAUP version 4.0b8a) of genetic relationships between 24 isolates of Listeria monocytogenes based on hybridization patterns derived from a 585 probe DNA microarray. Bootstrap values (nrep 1000) are indicated for major nodes. Sample V013668A was tested for processing and analysis reproducibility in three separate experiments (V013668Ar1±3).
(Table 2). Additionally, three subclusters (Ia, Ib, Ic) were present within cluster I, one of which was composed almost entirely of 1/2c isolates (Ib). Subclusters Ib and Ic also shared six unique sequences (probes 103, 134, 150, 367, 1072, 1092) with cluster II (Table 2). Eleven cluster-associated probes and four other polymorphic probes were sequenced to identify potential genetic markers and to assess the level of probe redundancy for the array. Thirteen of the probes had less than 87% nucleotide sequence similarity. Probes 134 and 1072 shared 99.7% nucleotide sequence identity and were considered redundant. A BLAST
360
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
Table 2 Microarray probes associated with L. monocytogenes clusters Ia, Ib, Ic, and II Strain (cluster)
Microarray probe identifier 103
M37952A (Ia) M36046A (Ia) M36582B (Ia) M32490G (Ia) M35568A (Ia) M32771C (Ia) M36509A (Ia) H1445 (Ib) G3321 (Ib) H9333 (Ib) H7973 (Ib) H9900096 (Ib) M10867C (Ic) M10867D (Ic) H842 (II) H1164 (II) H9900104 (II) F2365 (II) F5070 (II) G1092 (II) V8807 (II) V013668A (II) M36467A (II) M35584A (II) O51742b a b
a
0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0
134
150
367
1072
1092
317
369
791
955
302
0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0
0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0
0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0
0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0
0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1
0 low signal intensity; 1 high signal intensity. L. innocua.
search based on putative protein translations matched six cluster speci®c sequences with putative genes including several suspected virulence factors (Table 3). Four of the probe sequences were most similar to internalin proteins. In particular, sequences from probes 103, 134/1072, and 367 were similar to InlE, whereas probe 1069 coded for a LPTXG motif characteristic of many surface-associated internalins (Vazquez-Boland et al., 2001). A BLASTn search revealed that three probe sequences (103, 134/1072, and 367) share identity with the same region of the L. monocytogenes genome and are most likely sequence variants of the same ORF. Probe 302, unique to cluster II, had sequence similarity to L. monocytogenes ferrous iron transport protein B. Probe 955 had greatest similarity to PTS system beta-glucoside-speci®c enzyme II that has been implicated in virulence (Brehm et al., 1999; Glaser et al., 2001). One sequence (probe 192) was only found in three strains and it may be related to phage mediated, horizontal gene transfer (Table 3). Although microarray analysis of L. monocytogenes isolates did identify several polymorphic probes with sequence similarity to virulence factors, the signi®cance of this ®nding is unclear. In particular, the L. monocytogenes (strain EGD-e) genome contains 41 open reading frames encoding an LPXTG motif and 39 genes for PTS system enzyme II
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
361
Table 3 Summary of the select probe sequences from the L. monocytogenes microarray Probe no. (accession no.)
Tally of positive samples by cluster Ia
Ib/Ic
II
103 134 150 367 1072 1092 369 955
(BH170514) (BH170515) (BH170516) (BH170520) (BH170526) (BH170528) (BH175021) (BH170524)
0 0 0 0 0 0 11 11
8 8 8 8 8 8 8 8
12 12 12 12 12 12 0 0
791 1069 317 192 302 950 1085
(BH170522) (BH170525) (BH170519) (BH170517) (BH170518) (BH170523) (BH170527)
11 11 11 0 0 4 2
8 8 8 3 0 5 5
0 0 1 0 12 11 12
a
Similar proteinsa (E score)
Membrane associated lipoprotein, L. monocytogenes (3e-94) Membrane associated lipoprotein, L. monocytogenes (e-108)a Transcription regulator, L. monocytogenes (e-101) Membrane associated lipoprotein, L. monocytogenes (2e-91) Membrane associated lipoprotein, L. monocytogenes (e-108)a Hypothetical protein lmo0461, L. monocytogenes (5e-70) 3-Isopropylmalate dehydrogenase, L. monocytogenes (e-101) PTS system, beta-glucoside-specific enzyme IIABC component, L. monocytogenes (1e-66) Hypothetical protein lmo0904, L. monocytogenes (e-115) Internalin proteins, L. monocytogenes (2e-82 and below) ABC transporter, L. monocytogenes (2e-49) Integrase (phage-related protein), Bacillus halodurans (3e-43) Ferrous iron transport protein B, L. monocytogenes (1e-96) Polypeptide deformylase, L. monocytogenes (1e-53) Gamma-glutamyl kinase, L. monocytogenes (2e-27)
Redundant probes.
permeases (Glaser et al., 2001). Therefore, a more thorough analysis is needed including the typing of many more strains of known phenotype (virulent versus attenuated) on a microarray constructed from a greater number and diversity of polymorphic probes. 4. Conclusion A 585 probe mixed genome array was used to differentiate 24 L. monocytogenes isolates into two sero-clusters that agree with previously described phylogenetic lineages. Subclusters were also identi®ed within serotypic groups and among isolates from similar epidemiologic and geographic sources. Ten cluster-associated sequences were identi®ed, including ®ve that had similarity to suspected virulence factors. Increasing the number of representative strains used to construct the library while increasing the size of the array will likely increase both subtyping resolution and the number of cluster-associated sequences identi®ed. Cluster-associated sequences can then be used to design speci®c PCR primers for rapid L. monocytogenes strain detection and identi®cation. Identi®cation of speci®c genetic differences among isolates with de®ned phenotypes will provide insight into the epidemiology and potentially the pathogenesis of L. monocytogenes. Acknowledgements We thank Peggy Hayes at CDC for providing bacterial strains and serotyping assistance, Jinxin Hu at Washington State Department of Health for providing bacterial strains,
362
M.K. Borucki et al. / Veterinary Microbiology 92 (2003) 351±362
Myrvie Fuentes and Josiel Lopez for DNA extraction, James Reynolds for assistance with PFGE analysis, Glen Scoles for assistance with phylogenetic analysis, and Donald Knowles and Thomas Besser for helpful discussions. We thank Guy Palmer for critical review of the manuscript. Funding was provided by USDA Agricultural Research Service CWU 5348-32000-017-00D and by the Agricultural Animal Health Program (College of Veterinary Medicine, Washington State University). References Bibb, W.F., Gellin, B.G., Schwartz, B., Plikaytis, B.D., Reeves, M.W., Pinner, R.W., Broome, C.V., 1990. Analysis of clinical and food-borne isolates of Listeria monocytogenes in the United States by multilocus enzyme electrophoresis and application of the method to epidemiologic investigations. Appl. Environ. Microbiol. 56, 2133±2141. Brehm, K., Ripio, M.T., Kreft, J., Vazquez-Boland, J.A., 1999. The bvr locus of Listeria monocytogenes mediates virulence gene repression by beta-glucosides. J. Bacteriol. 181, 5024±5032. Brosch, R., Chen, J., Luchansky, J.B., 1994. Pulsed-field fingerprinting of listeriae: identification of genomic divisions for Listeria monocytogenes and their correlation with serovar. Appl. Environ. Microbiol. 60, 2584± 2592. Call, D.R., Chandler, D.P., Brockman, F.J., 2001. Fabrication of DNA microarrays using unmodified oligonucleotide probes. BioTechniques 30, 368±379. FDA, 1992. Bacteriological Analytical Manual, 7th ed. Glaser, P., Frangeul, L., Buchrieser, C., Rusniok, C., Amend, A., Baquero, F., Berche, P., Bloecker, H., Brandt, P., Chakraborty, T., Charbit, A., Chetouani, F., Couve, E., de Daruvar, A., Dehoux, P., Domann, E., Dominguez-Bernal, G., Duchaud, E., Durant, L., Dussurget, O., Entian, K.D., Fsihi, H., Portillo, F.G., Garrido, P., Gautier, L., Goebel, W., Gomez-Lopez, N., Hain, T., Hauf, J., Jackson, D., Jones, L.M., Kaerst, U., Kreft, J., Kuhn, M., Kunst, F., Kurapkat, G., Madueno, E., Maitournam, A., Vicente, J.M., Ng, E., Nedjari, H., Nordsiek, G., Novella, S., de Pablos, B., Perez-Diaz, J.C., Purcell, R., Remmel, B., Rose, M., Schlueter, T., Simoes, N., Tierrez, A., Vazquez-Boland, J.A., Voss, H., Wehland, J., Cossart, P., 2001. Comparative genomics of Listeria species. Science 294, 849±852. Graves, L.M., Swaminathan, B., 2001. PulseNet standardized protocol for subtyping Listeria monocytogenes by macrorestriction and pulsed-field gel electrophoresis. Int. J. Food Microbiol. 65, 55±62. Graves, L.M., Swaminathan, B., Hunter, S., 1999. Subtyping Listeria monocytogenes. In: Ryser, E.T., Marth, E.H. (Eds.), Listeria, Listeriosis, and Food Safety. Marcel Dekker, New York, pp. 279±298. Hegde, P., Qi, R., Abernathy, K., Gay, C., Dharap, S., Gaspard, R., Hughes, J., Snesrud, E.N., Less, N., Quackenbush, J., 2000. A concise guide to cDNA microarray analysis. BioTechniques 29, 548±562. Norrung, B., Skovgaard, N., 1993. Application of multilocus enzyme electrophoresis in studies of the epidemiology of Listeria monocytogenes in Denmark. Appl. Environ. Microbiol. 59, 2817±2822. Page, R.D.M., 1996. TreeView: an application to display phylogenetic trees on personal computers. Comput. Applic. Biosci. 12, 357±358. Piffaretti, J.C., Kressebuch, H., Aeschbacher, M., Bille, J., Bannerman, E., Musser, J.M., Selander, R.K., Rocourt, J., 1989. Genetic characterization of clones of the bacterium Listeria monocytogenes causing epidemic disease. Proc. Natl. Acad. Sci. U.S.A. 86, 3818±3822. Rasmussen, O.F., Skouboe, P., Dons, L., Rossen, L., Olsen, J.E., 1995. Listeria monocytogenes exists in at least three evolutionary lines: evidence from flagellin, invasive associated protein and listeriolysin O genes. Microbiology 141, 2053±2061. Tappero, J.W., Schuchat, A., Deaver, K.A., Mascola, L., Wenger, J.D., 1995. Reduction in the incidence of human listeriosis in the United States. Effectiveness of prevention efforts? The Listeriosis Study Group. J. Am. Med. Assoc. 273, 1118±1122. Vazquez-Boland, J.A., Kuhn, M., Berche, P., Chakraborty, T., Dominguez-Bernal, G., Goebel, W., GonzalezZorn, B., Wehland, J., Kreft, J., 2001. Listeria pathogenesis and molecular virulence determinants. Clin. Microbiol. Rev. 14, 584±640.