APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Oct. 2011, p. 7031–7039 Vol. 77, No. 190099-2240/11/$12.00 doi:10.1128/AEM.05138-11Copyright © 2011, American Society for Microbiology. All Rights Reserved.

Cell Surface Attachment Structures Contribute to Biofilm Formationand Xylem Colonization by Erwinia amylovora�

Jessica M. Koczan,1,2 Bryan R. Lenneman,2 Molly J. McGrath,2 and George W. Sundin2,3*Department of Plant Biology,1 Department of Plant Pathology,2 and Center for Microbial Pathogenesis,3

Michigan State University, East Lansing, Michigan 48824

Received 14 April 2011/Accepted 30 July 2011

Biofilm formation plays a critical role in the pathogenesis of Erwinia amylovora and the systemic invasion ofplant hosts. The functional role of the exopolysaccharides amylovoran and levan in pathogenesis and biofilmformation has been evaluated. However, the role of biofilm formation, independent of exopolysaccharideproduction, in pathogenesis and movement within plants has not been studied previously. Evaluation of therole of attachment in E. amylovora biofilm formation and virulence was examined through the analysis ofdeletion mutants lacking genes encoding structures postulated to function in attachment to surfaces or incellular aggregation. The genes and gene clusters studied were selected based on in silico analyses. Microscopicanalyses and quantitative assays demonstrated that attachment structures such as fimbriae and pili areinvolved in the attachment of E. amylovora to surfaces and are necessary for the production of mature biofilms.A time course assay indicated that type I fimbriae function earlier in attachment, while type IV pilus structuresappear to function later in attachment. Our results indicate that multiple attachment structures are needed formature biofilm formation and full virulence and that biofilm formation facilitates entry and is necessary forthe buildup of large populations of E. amylovora cells in xylem tissue.

Biofilm development is often utilized by bacterial pathogensto aid in host establishment, in population expansion, andultimately in disease proliferation (7, 27). The biofilm matrixprotects cells from stressful environmental conditions andenables increased nutrient acquisition. The formation ofbiofilms is a coordinated and highly regulated process thatexhibits distinct transitions between phases. These develop-mental phases include planktonic (free swimming), attach-ment (reversible and irreversible), mature biofilm, and detach-ment (30). The specific regulatory triggers governing thetransition between phases are largely unknown; however, it hasbeen shown that mechanical signals, nutritional and metabolicsignals, quorum-sensing signals, and host-derived signals canshift biofilm development through the different phases (7, 17).By understanding the functional mechanisms of distinct bio-film phases in pathogenesis, potential novel targets for diseasecontrol can be discovered.

Bacteria produce numerous proteinaceous structures thatcan be used in cell adhesion and attachment to surfaces. Thesestructures range from monomeric proteins to protein com-plexes (24) and include the pili and fimbriae, which consist ofmultiple different appendages, and other structures such ascurli, adhesins, intimins, and invasins (16, 17, 24). Our under-standing of the roles these structures play in attachment, over-all biofilm formation, and pathogenesis is still in its earlystages; however, it has been shown that Escherichia coli andPseudomonas aeruginosa both utilize pili and fimbriae in bio-film formation (7). Recently, the roles of afimbrial and fimbrialadhesins of Xylella fastidiosa and pili of Ralstonia solanacearum

and Acidovorax avenae have been explored, further demon-strating that biofilm formation within vascular plant pathogensis an important factor in virulence (1, 9, 14). However, thoughcell surface structures have been implicated in biofilm forma-tion and attachment, the roles of the structures can vary greatlyamong different species (7).

The Gram-negative plant pathogen Erwinia amylovora is thecausal agent of fire blight. This organism is highly virulent andcapable of rapid systemic movement within plant hosts and ofrapid dissemination among rosaceous species, including appleand pear trees, when environmental conditions are favorable.The internal movement of the pathogen through the vascularsystem of plants and the ability of the pathogen to infect flowers,actively growing shoots, and rootstocks makes the managementof fire blight difficult (21). Previous work has demonstratedthat the exopolysaccharides amylovoran and levan are impor-tant elements in the biofilm formation of E. amylovora (18).Amylovoran is a pathogenicity factor that is also thought tofunction as a shield that protects cells from host-elicited anti-microbial responses from plants (3). Levan is a known viru-lence factor (10), though its specific role in pathogenesis isunknown.

We are interested in determining the specific role of biofilmformation in E. amylovora pathogenesis. We hypothesized thatwe could use a genetic approach to uncouple biofilm formationfrom exopolysaccharide biosynthesis, thus enabling an eval-uation of the importance of biofilm formation in virulencewithout compromising the pathogen due to a defect in exo-polysaccharide biosynthesis. We further hypothesized thatE. amylovora encodes for surface structures that are necessaryfor the first critical step of biofilm formation, attachment. E.amylovora is known to produce peritrichous flagella and a typeIII secretion apparatus (13, 27); however, the role of these andother surface appendages in attachment is not known.

* Corresponding author. Mailing address: Michigan State Univer-sity, Department of Plant Pathology, 103 Center for Integrated PlantSystems, East Lansing, MI 48824. Phone: (517) 355-4573. Fax: (517)353-5598. E-mail: sundin@msu.edu.

� Published ahead of print on 5 August 2011.

7031

on Decem

ber 1, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

In this study, we utilized a bioinformatic approach and therecently sequenced genome of E. amylovora (32, 33) to identifygenes encoding putative cell surface attachment structures.Individual genes and gene clusters were deleted, and we useda combination of in vitro attachment assays and plant virulenceassays to demonstrate that multiple attachment structures arepresent in E. amylovora and play a role in biofilm formation,which is critical to pathogenesis and systemic movement in thehost.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions. The bacterial strains,plasmids, and oligonucleotide primers used in this study are listed in Table 1. Allstrains were grown in Luria broth (LB) at 28°C. For biofilm formation assays,strains were grown in 0.5� LB. Growth media were supplemented with theantibiotics ampicillin (50 �g ml�1), chloramphenicol (20 �g ml�1), gentamicin(15 �g ml�1), and tetracycline (12 �g ml�1) as necessary. For PCR amplification,Invitrogen reagents, protocol, and settings were used (Invitrogen Corp., Carls-bad, CA), unless otherwise noted.

Gene identification, deletion mutagenesis, and complementation. Regions ofinterest containing genes targeted for deletion were identified in the unanno-tated Sanger sequence of E. amylovora ATCC 49946 (available at http://www.sanger.ac.uk/resources/downloads/bacteria/erwinia-amylovora.html) throughthe use of tblastn against known sequences from GenBank. Sequence-specificprimers were designed, and deletion mutants lacking individual genes or gene

clusters were generated using the � phage recombinase as previously de-scribed (41). The names, putative functions, annotated ID numbers, andmaps of the E. amylovora genes and gene clusters mutated in this studyare illustrated in Fig. 1 and listed in Table 2. The following deletions wereconstructed: �flhCD, �fliADST, �EAM_2544; �flgABCDEFGHIJKLMN togenerate the mutant designated �flg-3; �fliEFGHIJKLMNOPQ to generatethe mutant designated �flg-4; �hofBC ppdD to generate the designated mu-tant �hof; �hofC; �fimAD clpEF fimAC faeHI to generate the mutant desig-nated �fim; �fimD; �crl; and �eae. The �hof, �hofC, �fim, �fimD, �crl, and�eae deletion mutants were complemented with plasmid clones containingthe corresponding genes amplified from E. amylovora Ea1189 along with theirnative promoters, digested at restriction sites KpnI and SacI, and then ligatedinto the cloning vector pBBR1MCS-3 (Table 1). For complementation of the�fim mutant, gene products were amplified using Roche Expand Long RangedNTPack according to the manufacturer’s recommendations (Roche Diag-nostics GmbH, Mannheim, Germany). The �flg-3 (deletion of 18 kb) and�flg-4 (deletion of 10 kb) mutants were not complemented due to the largersize of the deletions.

Pathogenicity assays. Strains were tested for virulence using an immature pearfruit assay and an apple shoot assay as previously described (18, 40). For theimmature pear fruit assay, we used a cell dose of �1 � 104 CFU/ml andmeasured the surface area of lesion size on fruit at 0, 2, 4, 6, and 8 dayspostinoculation. Bacterial populations within immature pear fruit were quanti-fied at 0, 1, 2, and 3 days postinoculation. Apple shoots were inoculated bycutting the youngest leaves using scissors dipped in a suspension of E. amylovora,and tissue was bagged overnight to maintain high humidity conditions. For shootassays, we used a cell dose of �2 � 108 CFU/ml and measured symptom

TABLE 1. Deletion mutants constructed in this study, sequences of the respective genes, and putative structures targeted for deletion

Strain Relevant characteristic(s) Source orreference

Ea1189 Wild type 5Ea1189�ams Amylovoran operon deletion mutant 41Ea1189�hofC Deletion of EAM_0729; putative type IV pilus structure This studyEa1189�hof Deletion of EAM_0729-0731; genes ppdD, hofB, and hofC of a putative type IV pilus structure This studyEa1189�fimD Deletion of EAM_0230; putative type I fimbria structure This studyEa1189�fim Deletion of EAM_0230-0237; genes fimAD, clpEF, fimAC, and faeHI of a putative type I

fimbria structureThis study

Ea1189�flg-3 Deletion of EAM_2541-2562; genes flhCD, fliADST, and flgABCDEFGHIJKLMN of a putativeflagellum structure

This study

Ea1189�flg-4 Deletion of EAM_2569-2581; genes fliEFGHIJKLMNOPQ of a putative flagellum structure This studyEa1189�crl Deletion of EAM_0898; putative curlin regulator This studyEa1189�eae Deletion of EAM_3759; putative invasin structure This study

FIG. 1. Genetic maps of the individual genes and sets of genes deleted in this study. Individual genes are shown as block arrows, arrow directioncorresponds to the direction of transcription, and gene length is drawn to scale in base pairs. The EAM gene designations are from the genomesequence of E. amylovora ATCC 49946 (EMBL accession no. FN666575). If multiple genes are shown in a set (e.g., ppdD, hofB, and hofC), theentire set of genes was deleted. In the �fim, �flg-3, and �flg-4 mutants, 7, 21, and 13 genes, respectively, were deleted as shown.

7032 KOCZAN ET AL. APPL. ENVIRON. MICROBIOL.

on Decem

ber 1, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

TABLE 2. Plasmids and oligonucleotide primers used in mutagenesis and complementation

Plasmid orprimer Relevant characteristics or sequencea Source or

reference

Plasmids:pKD3 Plasmid utilized in � phage recombinase mutagenesis method; Cmr 41pKD46 Expresses recombinases red, �, �, and exo for construction of deletion mutants; Ampr 41pMP2444 pBBR1MCS-5 backbone; gfp expressed from lac promoter; Gmr 41pJMK2 pBBR1MCS-3 backbone; hofC gene inserted as KpnI-SacI fragment; Tcr This studypJMK3 pBBR1MCS-3 backbone; hof gene cluster inserted as KpnI-SacI fragment; Tcr This studypJMK4 pBBR1MCS-3 backbone; fimD gene inserted as KpnI-SacI fragment; Tcr This studypJMK5 pBBR1MCS-3 backbone; fim gene cluster inserted as KpnI-SacI fragment; Tcr This studypJMK6 pBBR1MCS-3 backbone; crl gene inserted as KpnI-SacI fragment; Tcr This studypJMK7 pBBR1MCS-3 backbone; eae gene inserted as KpnI-SacI fragment; Tcr This study

PrimersEa1189�hofC F: 5�-ATGGGTGAACGCTTACTTTTCCGCTGGCAGGCTATTGACGATAGTGG

GCAGTGTAGGCTGGAGCTGCTTC-3�This study

R: 5�-TTACCCAAGCGCATCTCCCAGCCTGAATACGGGCAAATACATGGCCACCACATATGAATATCCTCCTTA-3�

Ea1189�hof F: 5�-ATGGGTGAACGCTTACTTTTCCGCTGGCAGGCTATTGACGATAGTGGGCAGTGTAGGCTGGAGCTGCTTC-3�

This study

R: 5�-TCATGGCAACTGCTCCTCATCAAAACGGAACATATCCAGGCAAGCGTCCTCATATGAATATCCTCCTTA-3�

Ea1189�fimD F: 5�-ATGACGCCTAAGGTGAAGAGGTATGTGCTATTTGACGAAGCTTTCTGCCGGTGTAGGCTGGAGCTGCTTC-3�

This study

R: 5�-AATAAGTGTGCAARAGAGGACATTGAAAAGACAGAGAACGTCGAAGCGACCATATGAATATCCTCCTTA-3�

Ea1189�fim F: 5�-ATGAAAAAAGTAATCAACTTTATTTTCCTGCTGCTGGCAGGTGCGGGTGAGTGTAGGCTGGAGCTGCTTC-3�

This study

R: 5�-GATGACTATGAGTTATGACTGGCCGTCACTGTGCACGGTCGCGGCCCCTTCATATGAATATCCTCCTTA-3�

Ea1189�flg 3 F: 5�-ATGATTATGAAAATAAAGCTTTTCCTGTTGGGCTGTGGACTTGGACTCGGGTGTAGGCTGGAGCTGCTTC-3�

This study

R: 5�-TCAGAAAGACAAATCGACGGTCACCGTATCGCTATAGGACCCCGCCACCGCATATGAATATCCTCCTTA-3�

Ea1189�flg 4 F: 5�-GGCAGCGGGGTGAAGTGATACAACAGATCGGCTTAGAATCAGGCTTTACGGTGTAGGCTGGAGCTGCTTC-3�

This study

R: 5�-TTACGAATTGAGACTAAACAGTGACAGCTTCGACATCTGCTGGAATACGGCATATGAATATCCTCCTTA-3�

Ea1189�crl F: 5�-ATGACGTTACCGAGTGGACATCCTAAGAGTCGAATAATTAAGCGCTTTCAGTGTAGGCTGGAGCTGCTTC-3�

This study

R: 5�-TCAGGCGGTCAGCTTCACCGGCTGGTCAGCGAAATCCGTTGCCGGTATCACATATGAATATCCTCCTTA-3�

Ea1189�eae F: 5�-ATGCAGGGGGGTAAAGCGGCTCCTCCTGCAATAATCTGGGACAAAGATGAGTGTAGGCTGGAGCTGCTTC-3�

This study

R: 5�-TTATTTTATAATATCAACATAAGGCTTGTTTATTGCGCTGCCGCTGCCAACATATGAATATCCTCCTTA-3�

�hofC comp F: 5�-GGTACCATGGGTGAACGCTTACTTTTCC-3� This studyR: 5�-GAGCTCTTACCCAAGCGCATCTCC-3�

�hof comp F: 5�-GGTACCATGGGTGAACGCTTACTTTTCC-3� This studyR: 5�-GAGCTCTCATGGCAACTGCTCCTCAT-3�

�fimD comp F: 5�-GGTACCATGACGCCTAAGGTGAAGAGG-3� This studyR: 5�-GAGCTCTTATTCACACGTTATCTCCTGTAACTT-3�

�fim comp F: 5�-GGTACCTGAAATTATTAACGAATGCCTGC-3� This studyR: 5�-GAGCTCATGATAAAACTTGCGGATAAATATCG-3�

�crl comp F: 5�-GGTACCATGACGTTACCGAGTGGACAT-3� This studyR: 5�-GAGCTCTCAGGCGGTCAGCTTCAC-3�

�eae comp F: 5�-GGTACCGGTAAAGCGGCTCCTCCT-3� This studyR: 5�-GAGCTCAATATCAACATAAGGCTTGTTTATTGC-3�

a KpnI (GGTACC) and SacI (GAGCTC) restriction sites are underlined. Abbreviations: F, forward; R, reverse; Amp, ampicillin; Cm, chloramphenicol; Gm,gentamicin; Tc, tetracycline.

VOL. 77, 2011 MULTIPLE ATTACHMENT STRUCTURES AND BIOFILM FORMATION 7033

on Decem

ber 1, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

progression at 3, 7, and 14 days postinoculation. Trees were maintained ingrowth chambers at 25°C with a light duration of 16 h. All assays were repeatedin triplicate, with at least four independent representatives in each experiment.Statistical analyses of treatment means (lesion size of Ea1189 and various deletionmutants) was done by one-way analysis of variance, and mean separation (P 0.05)was accomplished using Fisher’s protected least significant difference test. Meanswithin a column followed by the same letter are significantly different.

In vitro crystal violet assays for biofilm formation. We used an in vitro biofilmformation assay that was modified from an established crystal violet stainingassay (22) as previously described (18). For bright-field imaging, cellular aggre-gation on glass coverslips was observed after 48 h of growth in static culturefollowing staining with 10% crystal violet and examination using an OlympusIX71 inverted microscope (Olympus America Inc., New York, NY). Quantifica-tion measured absorbance values (A600) of resolubilized crystal violet stain after48 h as previously described (18).

In vitro attachment time course assay. We modified a standard in vitro biofilmassay (22) to develop an attachment assay to measure the timing of attachmentof bacterial cells to glass coverslips. Strains were each grown overnight in LBbroth to a concentration of 2 � 108 CFU/ml. Twenty microliters of this overnightculture was then added to 2 ml of sterile LB in individual wells of a 24-wellculture plate (Corning, Corning, NY), and glass coverslips (Thermo Fisher Sci-entific Inc., Waltham, MA) were angled within the wells. The coverslips wereremoved following incubation for 2, 4, 6, 8, 16, and 24 h; stained with 10% crystalviolet for 1 h; rinsed; and air dried. Bacterial cells attached to coverslips withinthe microscope field of view were enumerated. Ten random field-of-view imagesof each sample from each time point were taken at �100 using an Olympus IX71inverted microscope (Olympus America Inc., New York, NY). Each experimentwas repeated at least four times.

Confocal laser scanning microscopic visualization of E. amylovora biofilms.Three-dimensional aspects of biofilm formation were examined using a flow cellapparatus (Stovall Life Sciences, Greensboro, NC) as described previously (18).Strains were labeled with green fluorescent protein through the introduction ofthe plasmid pMP2444 via electroporation. Flow cell chambers were examinedusing the Zeiss 510 Meta ConfoCor3 LSM confocal laser scanning microscope(Carl Zeiss Microimaging GmbH), and images were captured at �10 using LSMimage browsing software (Carl Zeiss Microimaging GmbH). z stacks were com-piled, and a three-dimensional image that measured intensity was produced byusing the “2.5 dimensions” with the LSM browser.

Visualization of E. amylovora and deletion mutants in shoot tissue usingscanning electron microscopy. The two youngest leaves of three independentapple shoots (cv. Gala) were inoculated with bacterial strains by cutting leavesperpendicular to the major vein with scissors dipped in an E. amylovora suspen-sion (1 � 1010 CFU/ml). Leaves were collected at 3, 7, and 14 days postinocu-lation; sectioned into 1-cm sections; and fixed in paraformaldehyde-glutaralde-hyde (2.5% of each compound in 0.1 M sodium cacodylate buffer; ElectronMicroscopy Sciences, Hatfield, PA). The tissue was dehydrated successively andthen critical point dried using a critical-point drier (Balzers CPD, Liechtenstein).Dried petiole tissue was sectioned from slices after critical-point drying to reducepotential artifacts from the fixation process, mounted on aluminum mountingstubs (Electron Microscopy Sciences), and coated with gold using a gold sputtercoater (Emscope SC500; Emitech Ltd., Ashford, Kent, Great Britain). Imageswere captured on a JEOL 6400V scanning electron microscope (SEM; JapanElectron Optics Laboratories, Tokyo, Japan) with a LaB6 emitter (Noran EDS;Thermo Fisher Scientific Inc., Waltham, MA) using analySIS software (SoftImaging System GmbH).

Visualization of E. amylovora in vitro and in planta using transmission electronmicroscopy. For in vitro examination of Ea1189, cells were grown in LB overnightand diluted to �1 � 104 CFU/ml in sterile water. The cell suspension wasnegatively stained with 0.25% uranyl acetate, and a 2-�l sample was placed on atransmission electron microscope (TEM) grid. Samples were examined on aJEOL 100CX II microscope (Japan Electron Optics Laboratories). For in plantasamples, the two youngest leaves of three independent apple shoots (cv. Gala)were inoculated with Ea1189 by cutting leaves perpendicular to the major veinwith scissors dipped in a cell suspension (1 � 1010 CFU/ml). Samples were fixedovernight in paraformaldehyde-glutaraldehyde. Petiole tissue was embedded in2% agarose, thin sectioned with a razor blade, and stored in 0.1 M sodiumcacodylate buffer (Electron Microscopy Sciences). Postfixing was done with 2%osmium tetroxide in 0.1 M cacodylate buffer. Samples were dehydrated succes-sively in graded acetone series and then infiltrated and embedded in Poly/Bed812 (Polysciences, Warrington, PA). Sections were obtained with a PowerTomeXL ultramicrotome (Boeckeler Instruments, Tucson, AZ). Thin sections wereplaced on TEM grids, and samples were examined on the JEOL 100CX IImicroscope (Japan Electron Optics Laboratories).

RESULTS

Visualization of attachment structures of E. amylovora.TEM images of individual bacterial cells grown in broth culture,after negative staining, revealed the presence of peritrichous fla-gella but no other obvious surface appendages (Fig. 2A). TEMimages of xylem tube cross sections from petioles of leavesinoculated with wild-type E. amylovora Ea1189 revealed addi-

FIG. 2. Images of putative attachment structures of E. amylovora.(A) TEM imaging of a planktonic E. amylovora cell grown in brothculture and negatively stained. Peritrichous flagella are indicated byarrows. (B) TEM image of E. amylovora in planta. Putative attachmentstructures connect bacterial cells to host cells. (C) SEM image of E.amylovora cells attached to vascular tissue within a Gala apple tree.Imaged E. amylovora cells were found within a biofilm, with multipleappendages that protrude from the bacterial cell and attach to the hostsurface, as indicated by the arrows.

7034 KOCZAN ET AL. APPL. ENVIRON. MICROBIOL.

on Decem

ber 1, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

tional appendages anchoring bacterial cells to the host cell wall(Fig. 2B). SEM imaging also revealed apparently similar sur-face structures of Ea1189 that appeared to attach cells to theinterior of xylem tubes of Gala apple tissue (Fig. 2C), appar-ently anchoring the cells to the inner xylem wall. SEM imagesalso revealed cellular aggregation, a process that is also pre-sumably mediated by surface attachment structures in initiat-ing biofilms that stretched into the interior space of a xylemvessel (Fig. 2C).

Identification and generation of deletion mutants in puta-tive attachment structures. Regions of interest containinggenes encoding putative attachment structures were identifiedusing tblastx against the unannotated Sanger sequence ofE. amylovora ATCC 49946 using gene sequences of knownstructures from GenBank. Deletion mutations were generatedin genes or operons of interest as previously described (41),resulting in deletions of putative type IV pili, fimbriae, flagella,curli, and an invasin (Table 1).

Biofilm formation in vitro visualization and quantification.Using a previously published, modified in vitro crystal violetstaining method (18), we determined that, in general, mutantswith deletions of cell surface attachment structures exhibitedsignificant reductions in biofilm formation under these staticgrowth conditions. Absorbance values of resolubilized stainindicated a significant reduction in biofilm formation on glasscoverslips after 48 h in the �hofC, �hof, �fim, �flg-4, �crl, and�eae mutants (Fig. 3A). In only a few cases was biofilm for-mation by the mutants not significantly different from that ofEa1189 (Fig. 3A). Bright-field images of bacterial cells on glasscoverslips stained with crystal violet and captured at 48 hdemonstrated a lack of large aggregates, indicative of a lack ofbiofilm formation in almost all of the mutants (Fig. 3C). Com-plementation of deletion mutations restored biofilm formationto wild-type, if not higher, levels (Fig. 3B).

Examination of biofilm formation after 48 h within a flowcell system yielded similar results, with the �hofC, �hof, �fim,�flg-4, �crl, and �eae mutants all showing little to no aggrega-tion within the chamber, compared to the large aggregatesformed by wild-type Ea1189 (Fig. 4A and B and data notshown). Although the �fimD mutant formed large aggregateson glass coverslips under static conditions, we observed muchsmaller aggregates within the flow cell chamber that did nottransition to a mature biofilm (Fig. 4C).

In vitro biofilm attachment time course assay. The attach-ment of wild-type Ea1189 and mutant bacterial cells to glasscoverslips under static conditions was assessed at 2, 4, 6, 8, 16,and 24 h. All of the attachment mutants examined showedsignificant differences from wild-type Ea1189 throughout theexperiment (Table 3). At 2 h, only the �fim mutant exhibitedany significant reduction in attachment compared to that ofEa1189. However, at 4 h, all except the �fim mutant weresignificantly reduced in attachment. Attachment at 6 h wassignificantly reduced in all of the mutants except the �fimDand �crl mutants. An overall numerical increase in attachmentwas noted for most of the strains at 8 h, and only the �hof and�flg-3 mutants had significant reductions compared to the wildtype. At 16 h, all of the mutants, except the �flg-3, �crl, and�eae mutants, exhibited significant decreases, and at 24 h, the�fim, �flg-4, �crl, and �eae mutants all showed significantlyreduced attachment, while the �fimD mutant exhibited a sig-

nificant increase in attachment compared to that of Ea1189(Table 3).

Virulence assays in immature pear fruit and apple shoottissue after inoculation with E. amylovora and deletion mu-tants. All of the mutants exhibited reductions in lesion size atday 4 and day 6 postinoculation in immature pear fruit (Table4). However, by 8 days postinoculation, only the �hof and�fimD mutants showed a significantly reduced lesion size com-pared to that of Ea1189. The �ams mutant was nonpathogenicin immature pear fruit, as previously noted (3, 39). Comple-mentation of the deletion mutations restored necrosis to levelsthat were not significantly different from those produced by thewild type on day 8 in all cases (Table 4).

Significant differences in bacterial population size in imma-ture pear fruit were observed over the time course of 3 daysfollowing inoculation, including shifts in the timing of popula-tion growth and decreases in population size (Fig. 5). Signifi-cant decreases in population size compared to that of the wildtype at day 3 were observed for the �fim, �flg-3, and �flg-4mutants. The �hofC and �hof type IV pilus mutants both hada significant increase in population size compared to the wild

FIG. 3. Biofilm formation in vitro by E. amylovora. (A) Quantifica-tion of biofilm formation on glass coverslips of deletion mutants of E.amylovora and complemented strains. All of the mutants, except the�fimD mutant, demonstrated significant deficiencies in biofilm for-mation compared to Ea1189 (P 0.05, Student t test). Comple-mentation restores the phenotype to one similar to that of the wildtype. (B) Bright-field imaging of biofilm formation on glass cover-slips. All of the deletion mutants tested exhibit a significant visualreduction in attachment to the glass surface after 48 h, except the�fimD mutant.

VOL. 77, 2011 MULTIPLE ATTACHMENT STRUCTURES AND BIOFILM FORMATION 7035

on Decem

ber 1, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

type at days 2 and 3 postinoculation, suggesting a shift inpopulation growth.

Virulence of E. amylovora in apple shoots was evaluated bymeasuring the length of wilt symptoms as disease progressedinto shoot tissue. We observed a significant reduction in dis-ease progression in shoots inoculated with the �hofC, �fimD,�fim, and �flg-4 mutants compared to those inoculated withEa1189, while the length of wilt symptoms in shoots inoculatedwith the �hof and �flg-3 mutants was not significantly differentfrom that of symptoms caused by the wild type (Fig. 6). The�ams mutant was nonpathogenic in apple shoot tissue. In allvirulence assays, complementation restored virulence to levelsthat were not significantly different from that of the wild type(Table 4).

Visualization of E. amylovora and deletion mutants in shoottissue using SEM. Ea1189 cells initially attach to the internalsurfaces of the host xylem cell wall and produce biofilms thatextend into the vessel (Fig. 2C). Longitudinal sections ofshoots inoculated with Ea1189 enabled us to visualize bio-

FIG. 4. Flow cell imaging after 48 h of growth measuring aggre-gates as brightness of fluorescence, where higher intensity equalslarger aggregates. The image shows aggregation and biofilm develop-ment over area (x and y) and intensity (z). Note that intensity mappinghad the maximum value set at 250 and maps were developed with fivecolor layers. Additional layers of color indicate greater signal intensity.(A) Ea1189 exhibiting a typical biofilm consisting of aggregates.(B) The �fim mutant, a representative of biofilm-negative behavior,including very few to no aggregates but showing bacterial growth.(C) The �fimD mutant showing few aggregates throughout but not atlevels as great as those of the wild type.

TABLE 3. Time course assay measuring attachment of cells to glass coverslips at 2, 4, 6, 8, 16, and 24 ha

StrainAvg no. of cells attached at:

2 h 4 h 6 h 8 h 16 h 24 h

Ea1189 30.6 (abc) 117.6 (a) 123.0 (a) 252.0 (a) 210.8 (a) 281.4 (bc)�hofC mutant 21.1 (bcd) 72.2 (b) 28.9 (d) 179.2 (abc) 74.6 (d) 163.5 (cd)�hof mutant 33.3 (ab) 40.3 (cd) 40.0 (d) 100.7 (c) 86.5 (cd) 142.7 (cd)�fimD mutant 18.9 (cd) 65.6 (bc) 92.4 (abc) 206.7 (ab) 78.2 (d) 451.9 (a)�fim mutant 15.1 (d) 101.0 (a) 63.1 (bcd) 271.3 (a) 80.7 (cd) 132.4 (d)�flg-3 mutant 42.7 (a) 37.6 (d) 51.2 (cd) 128.3 (bc) 177.7 (ab) 365.6 (ab)�flg-4 mutant 42.9 (a) 43.2 (cd) 52.8 (cd) 216.5 (ab) 120.6 (bcd) 116.5 (d)�crl mutant 25.9 (bcd) 25.9 (d) 100.2 (ab) 225.1 (ab) 167.6 (abc) 98.1 (d)�eae mutant 29.0 (bc) 21.5 (d) 62.1 (bcd) 217.7 (ab) 145.7 (abcd) 86.5 (d)

a Each value represents the average number of cells attached within a field of view at �100 magnification. Means within a column followed by the same letter arenot significantly different according to Fisher’s protected least significant difference test (P � 0.05).

TABLE 4. Percent lesion size on immature pear fruit inoculatedwith E. amylovora Ea1189 or various attachment mutants at

4, 6, and 8 days postinoculationa

Strain% lesion size at:

Day 4 Day 6 Day 8

Ea1189 8 51 (a) 87 (abc)�ams mutant 0 0 (b) 0 (f)�hofC mutant 6 46 (a) 77 (bcde)�hofC/pJMK2 8 25 (b) 79 (bcd)�hof mutant 4 29 (b) 72 (de)�hof/pJMK3 3 28 (b) 92 (abc)�fimD mutant 3 18 (b) 57 (e)�fimD/pJMK4 2 34 (b) 83 (abc)�fim mutant 4 41 (a) 84 (abc)�fim/pJMK5 3 45 (a) 93 (abc)�flg-3 mutant 1 25 (b) 93 (abc)�flg-4 mutant 2 23 (b) 85 (abc)�crl mutant 1 21 (b) 76 (cde)�crl/pJMK6 9 57 (a) 99 (a)�eae mutant 2 19 (b) 80 (bcd)�eae/pJMK7 9 58 (a) 96 (ab)

a Statistical analyses of treatment means (sizes of lesions caused by Ea1189and various deletion mutants) was done by one-way analysis of variance, andmean separation (P 0.05) was accomplished with Fisher’s protected leastsignificant difference test. Means within a column followed by the same letter arenot significantly different. No significant differences were observed on day 4.

7036 KOCZAN ET AL. APPL. ENVIRON. MICROBIOL.

on Decem

ber 1, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

films progressing within xylem tissue (Fig. 7A). In addition,we discovered discontinuous aggregates, suggesting that dis-persal and initiation of new biofilms were occurring as amechanism of systemic movement within individual xylemvessels.

All of the bacterial mutants except the �flg-3 mutant werefound in high concentrations in the mesophyll tissue (Fig. 7D).Infrequently, mutants were found inside vascular tissue, spe-cifically, the young helical xylem, in smaller populations andwith little to no aggregates (Fig. 7C) compared to the wild type,which was found more frequently and in larger populationsof aggregates in both helical and developed xylem tissue(Fig. 7A). The �flg-3 mutant was phenotypically similar tothe wild type. Overall, the attachment mutants tended to beunable to localize or develop large populations within thevascular system.

DISCUSSION

We used a bioinformatic approach to identify multiple genesof E. amylovora encoding putative surface appendages anddemonstrated their role in attachment, biofilm formation, andpathogenesis. These are the first reported results to indicatethat type I fimbriae, flagella, type IV pili, and curli of E.amylovora may contribute to biofilm formation in static andflowing environments and that defects in many of the genesencoding these appendages result in decreased virulence inplanta. Our previous results demonstrated that E. amylovoraforms a biofilm in vitro and in planta (18). Pathogenesis andbiofilm formation appear to be linked, but without identifyinggenes encoding traits independent of amylovoran production,the mechanistic role in virulence of biofilm formation in E.amylovora could not be studied. Interestingly, mutants withreduced biofilm formation ability appear unable to successfullyestablish large populations in apple xylem. Colonization ofxylem is critical to the systemic movement of the pathogenthrough plants (5); thus, biofilm-deficient mutants remain lo-calized within an inoculated leaf and are strongly impaired inthe ability to invade the rest of the plant.

This study of multiple attachment structures shows both theimportance and the complexity of the attachment phase inbiofilm establishment. This was evident when examining therole of type IV pili using a cluster of genes identified in thisstudy, �hof, including the genes ppdD, hofB, and hofC. Thesegenes are known to encode functional type IV pili in P. aerugi-nosa (31). Type IV pili have been shown to mediate the tran-sition from reversible to irreversible attachment in P. aerugi-nosa and contribute to the virulence of another vascularinvading plant pathogen, R. solanacearum (8, 14, 34). Thoughthe timing of the transition between reversible and irreversibleattachment has not been determined, it can be assumed that

FIG. 5. Bacterial populations in immature pear fruit over 3 days.A shift in the population growth of the �hofC and �hof mutants andreductions in the �fim, �flg-3, and �flg-4 mutants are evident.Significant alterations in population are indicated by asterisks, andsignificant differences (P 0.05) were determined using the Stu-dent t test.

FIG. 6. Measurement of necrosis progression in Gala apple tissueat 3 and 7 days postinoculation. Scissor cuts perpendicular to themid-vein of a leaf were used for inoculations. The �hofC, �fimD, �fim,and �flg-4 mutants were significantly reduced in virulence compared tothe wild-type Ea1189 strain, and the �ams mutant was nonpathogenic(significant differences [P 0.05, t test] are indicated by asterisks) at7 days postinoculation.

FIG. 7. SEM images of E. amylovora Ea1189 within xylem tissue(A) and mesophyll tissue (B) of apple leaves. Distinct biofilms arepresent within the vascular tissue (black arrow), as indicated by theaggregates filling the vascular space, and smaller populations are pres-ent within the mesophyll tissue (white arrows). The biofilm formation-deficient �fim mutant was visible within vascular tissue (C) and meso-phyll tissue (D), which was typical of all of the deletion mutantsstudied, and exhibited larger populations within mesophyll tissue. Rel-atively few cells were able to gain entry into xylem tissue (black arrow).

VOL. 77, 2011 MULTIPLE ATTACHMENT STRUCTURES AND BIOFILM FORMATION 7037

on Decem

ber 1, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

the transition would occur after reversible attachment (the firstfew hours of contact with a surface) and before or at thebeginning of the expansion of the microcolony in the biofilm orthe initial growth phase of the bacteria. Under ideal growthconditions, in E. amylovora, this would occur between 3 and5 h, at the end of the lag time and the beginning of the growthphase (19). During the time course assay, the �hof mutant wasable to initiate attachment; however, from 4 to 6 h, the mutantwas significantly reduced in surface attachment compared tothe wild type. This implies that a disruption of the type IV pilialters irreversible attachment in E. amylovora. The disruptionof irreversible attachment also seems to stop any further sig-nificant attachment or expansion into micro- and macrocolo-nies.

The role of flagella in the attachment process was also ex-amined. It is well known that flagellum-driven motility allowsseveral bacterial species, including E. coli and Shewanella one-idensis MR-1, to escape unfavorable environments and swimtoward more favorable ones (23, 36, 37). Flagella have alsobeen shown to play significant roles in biofilm formation inseveral pathogens (7, 11, 15, 25, 38). For example, E. coliflagella exhibit multiple functions in biofilm formation, includ-ing motility to a surface, mediating surface contact, and expan-sion of the biofilm (36). In this study, we deleted two of thefour gene clusters encoding the production and regulation offlagella in E. amylovora (32) and demonstrated various effectson biofilm formation and virulence. One cluster, flagellar genecluster three, appears to aid in irreversible attachment but doesnot have a direct role in biofilm formation by or virulence ofthe pathogen. Similarly, the second gene cluster does not seemto have a specific role in reversible or irreversible attachment;however, when it is deleted, there is a significant reduction inbiofilm formation and virulence. These results indicate thatflagellum production appears to be controlled by multiple geneclusters that may function independently. As a result, the fla-gella of E. amylovora seem to have multifaceted functions inthe biofilm formation process.

Additional putative attachment structures were identifiedthrough our bioinformatic approach, including a regulator ofcurli genes, crl, and an invasin island, eae. Curli, or amyloidfibers, of E. coli are known to mediate the attachment offlagella and fimbriae to surfaces, ultimately aiding in biofilmformation and virulence (2, 26, 29). Curli-like functions, in-cluding a role in pellicle formation, have been associated withthe type III secretion system of Dickeya dadantii, an entericplant pathogen (12, 39). Invasins of Yersinia spp. or intimins ofE. coli are outer membrane proteins known to mediate theattachment of bacterial cells to their eukaryotic hosts but haveno known function in biofilm formation (20). Our deletions incrl and eae in E. amylovora demonstrate that curli and invasinsfunction in biofilm formation; however, no specific functionduring attachment was found. This implies that either curli andinvasins aid in the attachment of other structures or their rolein biofilm formation does not occur during the attachmentphase. The curli of E. coli have been shown to function incell-to-cell contact (26). Similarly curli of E. amylovora couldfunction in cellular aggregation and the building of maturebiofilms. Reductions in virulence due to deletion of the curliregulator indicate that not only functional attachment but alsomature biofilm formation is needed for full virulence in the

host. SEM imaging of the mutants in xylem tissue demon-strated that �crl mutant cells were not able to develop a ma-ture biofilm within the xylem, supporting this hypothesis.

The final putative surface structures identified were type Ifimbriae. These were of interest because these rod-shapedsurface organelles, found in most enterobacteria, have beenwell studied for their role in biofilm formation (36). In E. coli,type I fimbriae play a critical role in initial cell-to-surfacecontact and are important for adhesion and biofilm formationin both static and flowing systems (4, 25). To determine if thesestructures play a similar role in E. amylovora, a time courseassay of bacterial attachment was conducted. Based on thelength of the lag time of E. amylovora under optimal cultureconditions, we thought that initial attachment should occurwithin 2 h of exposure to a surface (19). Our results indicatedthat mutants with a deletion in type I fimbriae exhibited sig-nificant reductions of attachment after 2 h of exposure, indi-cating that there is a defect in the initial attachment essentialfor biofilm formation.

In addition to the role type I fimbriae play in attachment,these structures are also important virulence factors in severalenteric pathogens (6, 35). Even in cases where type I fimbriaewere not originally present in E. coli, the introduction of fim-briae genes increased the severity of the virulence of the patho-gen (6). Similarly, in E. amylovora, type I fimbria deficiencieslead to an overall reduction in the virulence of the pathogen.Significant reductions in lesion size in immature pear fruit anda reduction of disease progression within shoot tissue furtherindicate that initial attachment during biofilm formation con-tributes to virulence.

SEM imaging revealed that all of the mutant strains of E.amylovora studied (with the exception of the �flg-3 mutant)grew to large population sizes within the apple mesophyll.Thus, a fully functional biofilm does not seem to be requiredfor survival and growth in planta. However, we also found thatthese mutants were unable to enter and grow to large popula-tion sizes within vascular tissue, unlike the wild-type strain,indicating that biofilm formation plays a role in xylem coloni-zation by E. amylovora. This was consistent with our previousfinding that a levansucrase (lsc) mutant deficient in biofilmformation also remained confined to the mesophyll (18). Thesefindings were somewhat surprising, in that we expected that afunctional biofilm would be necessary to attain a large popu-lation size in planta and that large populations would beneeded to initiate entry into the xylem. Instead, it appears thatbiofilms play a crucial role in the actual entry into vasculartissue. Further studies are being conducted to determine howthe biofilm aids in entry into xylem and growth. In addition, wepreviously found that the �ams mutant was incapable ofgrowth in planta. This seems to indicate that amylovoran, in-dependent of a fully functional biofilm, is needed to protect E.amylovora from antimicrobial responses from the plant, assuggested by others (3).

In conclusion, we used a genetic approach to identify mul-tiple structures directly or indirectly involved in the attachmentphase of biofilm formation and in the virulence of E. amylo-vora. Structures that appear to play a direct role in the attach-ment of E. amylovora function in the virulence of the pathogen;structures that have indirect roles in attachment did not alwaysimpact virulence. We do not yet fully understand the linkage

7038 KOCZAN ET AL. APPL. ENVIRON. MICROBIOL.

on Decem

ber 1, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

between in vitro results demonstrating alterations in biofilmformation or in the timing of attachment to glass and biofilmformation and disease progression in planta. However, theseresults ultimately demonstrated the importance of biofilm for-mation in the virulence and xylem entry of E. amylovora. Func-tional characterization of additional genes involved in attach-ment and other steps in biofilm formation may provideadditional insight into the role of biofilm formation in E. amy-lovora pathogenesis.

ACKNOWLEDGMENTS

This work was supported by a special grant from the United StatesDepartment of Agriculture CSREES; Project GREEEN, a Michiganplant agriculture initiative at Michigan State University; and the Mich-igan Agricultural Experiment Station.

We thank Gail Ehret for statistical assistance and Alicia Pastor forwonderful TEM work.

REFERENCES

1. Bahar, O., T. Goffer, and S. Burdman. 2009. Type IV pili are required forvirulence, twitching motility, and biofilm formation of Acidovorax avenaesubsp. citrulli. Mol. Plant Microbe Interact. 22:909–920.

2. Barnhart, M. M., and M. R. Chapman. 2006. Curli biogenesis and function.Annu. Rev. Microbiol. 60:131–147.

3. Bellemann, P., S. Bereswill, S. Berger, and K. Geider. 1994. Visualization ofcapsule formation by Erwinia amylovora and assays to determine amylovoransynthesis. Int. J. Biol. Macromol. 16:290–296.

4. Beloin, C., et al. 2004. Global impact of mature biofilm lifestyle on Esche-richia coli K-12 gene expression. Mol. Microbiol. 51:659–674.

5. Burse, A., H. Weingart, and M. S. Ullrich. 2004. NorM, an Erwinia amylovoramultidrug efflux pump involved in in vitro competition with other epiphyticbacteria. Appl. Environ. Microbiol. 70:693–703.

6. Connell, H., et al. 1996. Type I fimbrial expression enhances Escherichia colivirulence for the urinary tract. Proc. Natl. Acad. Sci. U. S. A. 93:9827–9832.

7. Davey, M. E., and G. A. O’Toole. 2000. Microbial biofilms: from ecology tomolecular genetics. Microbiol. Mol. Biol. Rev. 64:847–867.

8. Delaquis, P. J., D. E. Caldwell, J. R. Lawrence, and A. R. McCurdy. 1989.Detachment of Pseudomonas fluorescens from biofilms on glass surfaces inresponse to nutrient stress. Microb. Ecol. 18:199–210.

9. Feil, H., W. S. Feil, and S. E. Lindow. 2007. Contribution of fimbrial andafimbrial adhesions of Xylella fastidiosa to attachment to surfaces and viru-lence in grape. Phytopathology 97:318–324.

10. Gross, M., G. Geier, K. Rudolph, and K. Geider. 1992. Levan and levansu-crase synthesized by the fire blight pathogen Erwinia amylovora. Physiol.Mol. Plant Pathol. 40:371–381.

11. Hossain, M. M., and S. Tsuyumu. 2006. Flagella-mediated motility is re-quired for biofilm formation by Erwinia carotovora subsp. carotovora. J. Gen.Plant Pathol. 72:34–39.

12. Jahn, C. E., D. K. Willis, and A. O. Charkowski. 2008. The flagellar sigmafactor FliA is required for Dickeya dadantii virulence. Mol. Plant MicrobeInteract. 21:1431–1442.

13. Jin, Q., et al. 2001. Visualization of secreted Hrp and Avr proteins along theHrp pilus during type III secretion in Erwinia amylovora and Pseudomonassyringae. Mol. Microbiol. 40:1129–1139.

14. Kang, Y., H. Liu, S. Genin, M. A. Schell, and T. P. Denny. 2002. Ralstoniasolanacearum requires type 4 pili to adhere to multiple surfaces and fornatural transformation and virulence. Mol. Microbiol. 46:427–437.

15. Karatan, E., and P. Watnick. 2009. Signals, regulatory networks, and mate-rials that build and break biofilm formation. Microbiol. Mol. Biol. Rev.73:310–347.

16. Kim, T., B. M. Young, and G. M. Young. 2008. Effect of flagellar mutationson Yersinia enterocolitica biofilm formation. Appl. Environ. Microbiol. 74:5466–5474.

17. Kline, K. A., S. Falker, S. Dahlberg, S. Normark, and B. Henriques-Nor-mark. 2009. Bacterial adhesins in host-microbe interactions. Cell Host Mi-crobe 5:580–592.

18. Koczan, J. M., M. J. McGrath, Y. Zhao, and G. W. Sundin. 2009. Contri-bution of Erwinia amylovora exopolysaccharides amylovoran and levan tobiofilm formation: implications in pathogenicity. Phytopathology 99:1237–1244.

19. McGhee, G. C., and G. W. Sundin. 2011. Evaluation of kasugamycin for fireblight management, effect on nontarget bacteria, and assessment of kasuga-mycin resistance potential in Erwinia amylovora. Phytopathology 101:192–204.

20. McGraw, E. A., J. Li, R. K. Selander, and T. S. Whittam. 1999. Molecularevolution and mosaic structure of , �, and � intimins of pathogenic Esch-erichia coli. Mol. Biol. Evol. 16:12–22.

21. Norelli, J., A. L. Jones, and H. S. Aldwinkle. 2003. Fire blight managementin the twenty-first century: using new technologies that enhance host resis-tance in apple. Plant Dis. 87:756–765.

22. O’Toole, G., et al. 1999. Genetic approaches to study of biofilms. MethodsEnzymol. 310:91–107.

23. Paulick, A., et al. 2009. Two different stator systems drive a single polarflagellum in Shewanella oneidensis MR-1. Mol. Microbiol. 71:836–850.

24. Pizarro-Cerda, J., and P. Cossart. 2006. Bacterial adhesion and entry intothe host cell. Cell 124:715–727.

25. Pratt, L. A., and R. Kolter. 1998. Genetic analysis of Escherichia coli biofilmformation: roles of flagella, motility, chemotaxis, and type I pili. Mol. Mi-crobiol. 30:285–293.

26. Prigent-Combaret, C., et al. 2000. Developmental pathway for biofilm for-mation in curli producing Escherichia coli strains: role of flagella, curli, andcolanic acid. Environ. Microbiol. 2:450–464.

27. Ramey, B. E., M. Koutsoudis, S. B. von Bodman, and C. Fuqua. 2004.Biofilm formation in plant-microbe associations. Curr. Opin. Microbiol.7:602–609.

28. Reference deleted.29. Saldana, Z., et al. 2009. The Escherichia coli common pilus and the bundle-

forming pilus act in concert during the formation of localized adherence byenteropathogenic E. coli. J. Bacteriol. 191:3451–3461.

30. Sauer, K., A. K. Camper, G. D. Ehrlich, J. W. Costerton, and D. G. Davies.2002. Pseudomonas aeruginosa displays multiple phenotypes during develop-ment as a biofilm. J. Bacteriol. 184:1140–1154.

31. Sauvonnet, N., P. Gounon, and A. Pugsley. 2000. PpdD type IV pilin ofEscherichia coli K-12 can be assembled into pili in Pseudomonas aeruginosa.J. Bacteriol. 182:848–854.

32. Sebaihia, M., et al. 2010. Complete genome sequence of the plant pathogenErwinia amylovora strain ATCC 49946. J. Bacteriol. 192:2020–2021.

33. Smits, T. H., et al. 2010. Complete genome sequence of the fire blightpathogen Erwinia amylovora CFBP 1430 and comparison to other Erwiniaspp. Mol. Plant Microbe Interact. 23:384–393.

34. Stoodley, P. Sauer, K. D. G. Davies, and J. W. Costerton. 2002. Biofilms ascomplex differentiated communities. Annu. Rev. Microbiol. 56:187–209.

35. van der Velden, A. W. M., A. J. Baumler, R. M. Tsolis, and F. Heffron. 1998.Multiple fimbrial adhesins are required for full virulence of Salmonellatyphimurium in mice. Infect. Immun. 66:2803–2808.

36. Van Houdt, R., and C. W. Michiels. 2005. Role of bacterial cell surfacestructures in Escherichia coli biofilm formation. Res. Microbiol. 156:626–633.

37. Verstraeten, N., et al. 2008. Living on a surface: swarming and biofilmformation. Trends Microbiol. 16:496–506.

38. Wood, T. K. 2009. Insights on Escherichia coli biofilm formation and inhibi-tion from whole-transcriptome profiling. Environ. Microbiol. 11:1–15.

39. Yap, M. N., C. H. Yang, J. D. Barak, C. E. Jahn, and A. O. Charkowski. 2005.The Erwinia chrysanthemi type III secretion system is required for multicel-lular behavior. J. Bacteriol. 187:639–648.

40. Zhao, Y., S. E. Blumer, and G. W. Sundin. 2005. Identification of Erwiniaamylovora genes induced during infection of immature pear tissue. J. Bac-teriol. 187:8088–8103.

41. Zhao, Y., G. W. Sundin, and D. Wang. 2009. Construction and analysis ofpathogenicity island deletion mutants of Erwinia amylovora. Can. J. Micro-biol. 55:457–464.

VOL. 77, 2011 MULTIPLE ATTACHMENT STRUCTURES AND BIOFILM FORMATION 7039

on Decem

ber 1, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from