Transmission of Yersinia pestis from an Infectious Biofilm in the Flea...

Transmissible Y. pestis Biofilm in Fleas • JID 2004:190 (15 August) • 783

M A J O R A R T I C L E

Transmission of Yersinia pestis from an InfectiousBiofilm in the Flea Vector

Clayton O. Jarrett,1 Eszter Deak,1,a Karen E. Isherwood,3 Petra C. Oyston,3 Elizabeth R. Fischer,2 Adeline R. Whitney,1

Scott D. Kobayashi,1 Frank R. DeLeo,1 and B. Joseph Hinnebusch1

Laboratories of 1Human Bacterial Pathogenesis and 2Intracellular Parasites, Rocky Mountain Laboratories, National Institute of Allergyand Infectious Diseases, National Institutes of Health, Hamilton, Montana; 3Defence, Science, and Technical Laboratory, Porton Down,Salisbury, Wiltshire, United Kingdom

Transmission of plague by fleas depends on infection of the proventricular valve in the insect’s foregut by adense aggregate of Yersinia pestis. Proventricular infection requires the Y. pestis hemin storage (hms) genes;here, we show that the hms genes are also required to produce an extracellular matrix and a biofilm in vitro,supporting the hypothesis that a transmissible infection in the flea depends on the development of a biofilmon the hydrophobic, acellular surface of spines that line the interior of the proventriculus. The developmentof biofilm and proventricular infection did not depend on the 3 Y. pestis quorum-sensing systems. Theextracellular matrix enveloping the Y. pestis biofilm in the flea appeared to incorporate components from theflea’s blood meal, and bacteria released from the biofilm were more resistant to human polymorphonuclearleukocytes than were in vitro–grown Y. pestis. Enabling arthropod-borne transmission represents a novelfunction of a bacterial biofilm.

Yersinia pestis, the cause of bubonic and pneumonic

plague in humans, persists in populations of wild ro-

dents in many parts of the world and is transmitted

primarily by the bites of infected fleas [1]. Y. pestis

undergoes a characteristic development in the flea that

leads to transmission [2, 3]. During the first week after

being taken up by a flea in a blood meal, the bacteria

multiply in the lumen of the flea gut to form dense

aggregates. In some fleas, the infection also involves the

proventriculus, a valve that connects the midgut to the

esophagus. The interior wall of the proventriculus is

lined with rows of spines covered with cuticle, the same

hydrophobic, acellular material that composes the in-

sect exoskeleton. A Y. pestis aggregate can adhere to these

Received 12 December 2003; accepted 5 March 2004; electronically published12 July 2004.

Financial support: Ellison Medical Foundation (New Scholars in Global InfectiousDiseases award to B.J.H.).

a Present affiliation: Microbiology Graduate Program, Yale University School ofMedicine, New Haven, Connecticut.

Reprints or correspondence: Dr. B. Joseph Hinnebusch, National Institutes ofHealth, National Institute of Allergy and Infectious Diseases, Rocky MountainLaboratories, 903 S. 4th St., Hamilton, MT 59840 ([email protected]).

The Journal of Infectious Diseases 2004; 190:783–92� 2004 by the Infectious Diseases Society of America. All rights reserved.0022-1899/2004/19004-0017$15.00

spines, continue to grow and consolidate, and eventually

fill the proventriculus, impeding the passage of blood

into the midgut. Physical blockage and interference with

the action of the proventricular valve are required for

efficient transmission [4–6], which occurs when a com-

plete or partially blocked flea attempts to feed.

Two genetic factors of Y. pestis have been shown to

be specifically required for fleaborne transmission. Ini-

tial survival and growth after the infectious blood meal

requires a plasmid-encoded phospholipase D that pro-

tects the bacteria from lysis in the flea midgut [7]. A

transmissible infection, however, further depends on

the Y. pestis hemin storage (hms) gene products, which

are required for colonization and blockage of the pro-

ventriculus. Y. pestis hms mutants colonize the flea mid-

gut but are unable to infect the proventriculus [8]. The

hms genes constitute 2 unlinked operons on the Y. pestis

chromosome: hmsHFRS and hmsT [9–12]. HmsR and

F have amino acid sequence similarity to the glycosyl

transferase and deacetylase enzymes of Staphylococcus

epidermidis, which participate in the synthesis of an

extracellular polysaccharide that is required to produce

a biofilm [13–16]. The hms genes were so named be-

cause they are responsible for a pigmentation pheno-

at UC

SF Library and C

enter for Know

ledge Managem

ent on Decem

ber 16, 2014http://jid.oxfordjournals.org/

Dow

nloaded from

http://jid.oxfordjournals.org/

784 • JID 2004:190 (15 August) • Jarrett et al.

Table 1. Susceptibility pattern of the ex-tracellular matrix surrounding Hms-positiveYersinia pestis in the flea.

Treatment Resulta

Solvent0.1% SDS �

Acetone �

Chloroform plus methanol +Chloroform �

Methanol + (slight)3% H2O2 � (decolorizes)0.01N NaOH � (decolorizes)

EnzymeProteinase K, E �

Plasmin �

Lipase + (slight)Phospholipase C �

a +, Degrades matrix; �, no effect on matrix.

Figure 1. Development of temperature- and Hms-dependent biofilm in Yersinia pestis. A–C: Interior surface of glass flowcell inoculated with Hms-positive Y. pestis KIM6+ after 0 (A) and 48 (B) h of media flow at 21�C and after 48 h of media flow at 37�C (C). D–F: Flowcells after 48 h of mediaflow at 21�C inoculated with Hms-negative Y. pestis KIM6 (D), Hms-positive Y. pestis KIM6(pHMS1) (E), and Hms-positive, quorum-sensing–negativeY. pestis KIM6+ yepI::kan ypeI::kan DluxS (F). Bar, 20 mm.

type based on the ability of Y. pestis colonies to bind hemin or

the structurally analogous Congo red dye [17, 18].

Previous investigations have indicated that the dense aggre-

gates of Y. pestis that develop in the flea midgut and proven-

triculus are surrounded by an extracellular matrix, resembling

a biofilm [3, 19]. Y. pestis and Yersinia pseudotuberculosis also

produce biofilm-like growth on the external mouthparts of

Caenorhabditis elegans nematodes, and this phenotype also is

dependent on the hms genes [16, 20]. Many bacteria adhere to

and accumulate on a surface by forming a biofilm, a physio-

logically distinct community of cells embedded in an extra-

cellular matrix of bacterial origin [21]. The formation of biofilm

has been implicated in many persistent and chronic bacterial

infections, including dental plaque and peridontitis, infections

of implanted medical devices, and Pseudomonas aeruginosa in-

fection of the lungs of patients with cystic fibrosis [22]. In the

present study, we link Hms-dependent proventriculus infection

to formation of biofilm and examine the phenotype of the

bacteria that are released from the biofilm and transmitted

when a blocked flea attempts to feed.

MATERIALS AND METHODS

Bacteria. The Y. pestis strains KIM6+, KIM6 (the Hms-neg-

ative KIM6+ derivative that lacks the chromosomal locus con-

taining hmsHFRS), and KIM6 complemented with pHMS1,

which contains a wild-type (wt) copy of the hmsHFRS locus,

were used in the present study [23, 24]. A Y. pestis KIM6+ quo-

rum-sensing (QS)–defective mutant was constructed by use of

sequential allelic exchanges using the suicide vector pCVD442

[25] to replace the wt N-acyl-homoserine lactone (AHL)–type

autoinducer synthase genes yepI (YPO0984; identical to ytbI of

Y. pseudotuberculosis; GenBank accession number AF079136) and

ypeI (YPO2456; identical to ypsI of Y. pseudotuberculosis; Gen-

Bank accession number AF071401) [26] with insertion mutant

alleles. The suicide vectors used for mutagenesis contained the

ypeI or ypeI gene, into which a kanamycin (kan) resistance

marker was inserted. Integration of the suicide vector into the

wt gene on the chromosome was determined by screening for

pCVD442-encoded resistance to ampicillin. After growth of the

merodiploids in nonselective conditions, sucrose-resistant, ampi-

cillin-sensitive clones [25] were tested by use of polymerase chain

at UC

SF Library and C

enter for Know

ledge Managem

ent on Decem


Dow

nloaded from



Figure 2. Scanning electron microscopy of Hms-positive (A) and Hms-negative (B) Yersinia pestis grown on agar plates at 21�C. Bar, 0.5 mm.

Figure 3. Biofilm-like growth of Hms-positive Yersinia pestis in the proventriculus of blocked fleas. A, Digestive tract dissected from an uninfectedXenopsylla cheopis flea, indicating the location of the proventricular valve (PV), which connects the esophagus (E) to the midgut (MG). Bar, 0.25 mm.Scanning electron microscopy of the spines lining the interior of a proventriculus from an uninfected flea (B), a flea infected with Hms-negative Y.pestis (C), and a flea blocked with Hms-positive Y. pestis, showing bacteria embedded within an extracellular matrix (D). Asterisks indicate proventricularspines. Bar, 5 mm.

reaction (PCR) and nucleotide sequencing, to verify the desired

mutation. Loss of function of the Y. pestis AHL autoinducers was

also verified in a bioassay of culture supernatant using a Chro-

mobacterium violaceum reporter strain [27]. A third allelic ex-

change using pCVD44 generated a triple-QS mutant (Y. pestis

KIM6+ yepI::kan ypeI::kan DluxS), in which the luxS homolog

(YPO3300) was replaced with a mutant luxS allele with a 411-

bp in-frame deletion that truncated the predicted protein after

aa 8. The luxS deletion in this strain was confirmed by use of

PCR, sequence analysis, and Northern blot. The KIM6, KIM6+,

and triple-QS mutant strains were transformed with pGFP (Clon-

tech) by use of electroporation.

Infection of fleas. Xenopsylla cheopis fleas were infected by

allowing them to feed on fresh heparinized mouse blood con-

taining Y. pestis organisms/mL, using an artificial feeding85 � 10

system [3, 8]. Fleas (50 males and 50 females) that took an

infectious blood meal were kept at 21�C, fed twice weekly there-

after on normal mice, and monitored for 4 weeks for infection

and proventricular blockage rates, as described elsewhere [8].

Enzyme, solvent, and histological stain treatments of Y. pestis

biofilms. Bacterial aggregates were dissected from fleas’ di-

gestive tracts 1–3 weeks after infection, washed with PBS, and

placed on microscope slides. Slides were submerged in each of

the solvents listed in table 1 for 12 h at 21�C. Additional slides

at UC

SF Library and C

enter for Know

ledge Managem

ent on Decem


Dow

nloaded from



Figure 4. Comparison of flea infections produced by Hms-positive and Hms-negative Yersinia pestis. Digestive tracts dissected from a flea infectedand blocked with Hms-positive Y. pestis KIM6+ (A) and a flea infected with Hms-negative Y. pestis KIM6 (B). Bar, 0.25 mm. Higher-magnification viewsof bacterial masses from the midgut of fleas infected with Hms-positive Y. pestis KIM6+(pGFP) (C and E) or Hms-negative Y. pestis KIM6(pGFP) (Dand F) examined by brightfield and fluorescence microscopy. Arrows, border of the viscous extracellular matrix surrounding the Hms-positive Y. pestisaggregate, which was not seen for the Hms-negative aggregate. Bar, 0.05 mm.

were treated with proteinase K (200 mg/mL), proteinase E (250

mg/mL), fibrinolysin (plasmin, 2.5 U/mL), phospholipase C

(6.25 U/mL), and lipase (25,000 U/mL), for 12–24 h at 37�C

in buffers recommended by the supplier (Sigma). The ability

of these agents to dissolve the brown extracellular matrix was

evaluated by use of microscopy before and after treatment.

Dense suspensions of Y. pestis from 48-h, 21�C cultures on

heart-infusion agar (Difco) containing 0.2% galactose or liquid

TMH media [28] containing 1% (wt/vol) N-acetylglucosamine

were also prepared. Slides were stained with periodic acid–Schiff

(PAS), calcofluor white, ruthenium red, Sudan black, and OsO4

reagents, according to established protocols [29, 30].

Flowcell experiments. Y. pestis that had been incubated

for 48 h at 21�C in N-minimal medium [31] containing 0.1%

casamino acid, 38 mmol/L glycerol, and 8 mmol/L MgCl2 were

quantitated in a Petroff-Hausser counting chamber and diluted

to cells/mL with fresh medium; 1.5 mL of the bacterial71 � 10

suspension was injected into a 12-cm-long section of 0.5-mm-

thick, 3.5-mm2 glass tubing that was connected to a reservoir

of sterile medium via a peristaltic pump at the influent end

and to a discard reservoir at the effluent end. After a 20-min

period to allow bacteria to attach to the glass surface (designated

), sterile medium was pumped through the flowcell at 1t p 0

mL/min. The interior surface of the flow cell was photographed

48 h later using differential interference contrast microscopy.

Scanning electron microscopy. Samples were prepared as

described elsewhere [32], with the following modifications:

0.1% poly-l-lysine coated coverslips were applied to bacterial

at UC

SF Library and C

enter for Know

ledge Managem

ent on Decem


Dow

nloaded from



Table 2. Differential staining properties of Hms-positive (Hms+) and Hms-negative(Hms�) Yersinia pestis.

Source

Y. pestis KIM6+ (Hms+) Y. pestis KIM6 (Hms�)

In vivo (flea) In vitro culture In vivo (flea) In vitro culture

Sudan black + � � �

OsO4 + � � �

PAS + � � �

Ruthenium red � + � �

Calcofluor white � � � �

IFA (a-Y. pestis AS) �a + �a +

NOTE. AS, antiserum; IFA, indirect fluorescent antibody; PAS, periodic acid–Schiff; +, positive;�, negative.

a Surface of bacterial aggregates did not stain; external bacteria were positive (see figure 5).

Figure 5. A, In vivo biofilm dissected from a flea. B, In vitro biofilm produced on a glass surface by Yersinia pestis KIM6+(pGFP) stained by indirectfluorescent antibody with anti–Y. pestis KIM6+ serum and a rhodamine-labeled secondary antibody. Fluorescence micrographs taken with a combinationfluorescein isothiocyanate and rhodamine filter set are shown. Bar, 10 mm.

lawns after 48 h of growth at 21�C on heart-infusion agar

containing 0.2% galactose. Adherent bacteria were fixed over-

night at 4�C with 75 mmol/L lysine acetate, 0.075% (wt/vol)

alcian blue, 2% paraformaldehyde, and 2.5% glutaraldehyde

buffered with 100 mmol/L sodium cacodylate. Samples were

washed with 100 mmol/L sodium cacodylate, postfixed with

1% OsO4 for 1 h, and dehydrated in a graded ethanol series.

Samples were critical point–dried under CO2 with a Bal-Tec

cpd 030 drier (Balzers), mounted on aluminum studs, and

coated with 125 A of iridium in an IBS/TM200S ion beam

sputterer (South Bay Technologies) before viewing at 5 kV on

a S-4500 field emission scanning electron microscope (Hitachi).

The proventriculus was dissected from fleas, bisected with a

27-gauge needle, and fixed with 2.5% glutaraldehyde and 4%

paraformaldehyde in 100 mmol/L sodium cacodylate buffer

(pH 7.4). Samples were postfixed, mounted, and viewed as

described above.

Indirect fluorescent antibody (IFA) assays. Hyperimmune

serum was generated by injecting rabbits intravenously twice

weekly for 8 weeks with 0.2 mL of PBS containing 91 � 10

killed Y. pestis KIM6+ grown at 21�C in N-minimal medium

containing 2% glucose and 1% glucosamine. The serum was

used in IFA assays of bacterial aggregates dissected from the

midguts of infected fleas and of in vitro biofilms produced on

the walls of glass culture tubes by Y. pestis grown in TMH, 1%

N-acetylglucosamine, and 100 mmol/L hemin at 21�C. Washed,

unfixed aggregates were incubated with a 1:500 dilution of the

antiserum for 30 min at 37�C, washed, and reincubated with

a 1:50 dilution of goat anti–rabbit IgG labeled with rhodamine

(Kirkegaard & Perry). Slides were examined by fluorescence

microscopy.

Interaction with human polymorphonuclear leukocytes

(PMNLs). The midguts of 50 fleas were dissected 1–3 weeks

after infection and placed in 100 mL of RPMI 1640 medium

(Gibco) with 10 mmol/L HEPES. Twenty microliters of glass

sand was added, and the midguts were triturated and vortexed

thoroughly. After this procedure, both the Y. pestis KIM6+ and

KIM6 supernatants predominantly contained individual bac-

teria, with a few clusters of 2–5 bacteria but no large clumps.

In vitro control bacteria were grown in TMH and 1% N-ace-

at UC

SF Library and C

enter for Know

ledge Managem

ent on Decem


Dow

nloaded from



Figure 6. Fleas infected with a Yersinia pestis quorum-sensing mutant. A, Percentage of fleas infected and blocked during a 4-week period aftera single infectious blood meal containing Y. pestis KIM6+ (black bars) or KIM6+yepI::kan ypeI::kan DluxS (white bars). The average and range of 3independent flea infection experiments is shown. Shown are brightfield (B) and fluorescence microscopy (C) images of a digestive tract dissected froma flea blocked with Y. pestis KIM6+ yepI::kan ypeI::kan DluxS(pGFP). Bar, 0.1 mm.

tylglucosamine for 48 h at 21�C, volumes containing 61 � 10

bacteria were centrifuged, and the cells were resuspended in

100 mL of RPMI 1640 medium plus HEPES. The dissected

midguts of 50 uninfected fleas were added to each in vitro–

grown bacterial suspension, triturated, and vortexed. All bac-

terial supernatants were quantitated by direct count, adjusted

to Y. pestis organisms/mL in RPMI 1640 medium plus71 � 10

HEPES, and opsonized with 25% normal human serum for 30

min at 37�C. Phagocytosis and killing of Y. pestis by human

PMNLs was determined by adding 0.1 mL of each bacterial

suspension to triplicate wells of a 96-well tissue culture plate

containing fresh human PMNLs in 0.1 mL of RPMI61 � 10

1640 medium plus HEPES, prepared as described elsewhere

[33]. Matched control wells without PMNLs were also inoc-

ulated. The plates were then centrifuged at 350 g for 8 min at

4�C. After 10, 30, and 60 min of incubation at 37�C in 5%

CO2, the contents of individual wells were removed and treated

with saponin (final concentration, 0.1%) for 15 min at 4�C to

lyse PMNLs, and serial dilutions were plated on Luria-Bertani

agar. The percentage of killing was calculated by comparing the

average number of colony-forming units in the presence and

absence of PMNLs for each time point.

RESULTS

Temperature- and Hms-dependent production of Y. pestis bio-

film in vitro. We used a glass flowcell system to determine

whether Y. pestis can adhere to a hydrophobic surface and form

a biofilm in vitro. At room temperature, the Hms-positive Y.

pestis strain produced biofilm-like growth on the glass surface

(figure 1). An isogenic Y. pestis Hms-negative mutant did not

form a biofilm, but this ability was restored when the mutant

was complemented with the hms genes. Y. pestis did not form

a biofilm when the flowcell was incubated at 37�C, which cor-

relates with the lack of hms-dependent Congo red–binding

(pigmentation phenotype) at temperatures 128�C [17, 18, 34].

Growth-rate determinations at 21�C and 37�C showed no dif-

ferences among the Y. pestis strains in the medium. Thus, along

with the temperature-dependent ability of Y. pestis colonies to

bind hemin and Congo red, the ability to produce an in vitro

biofilm is also Hms dependent.

Given the similarity of Y. pestis hmsR and hmsF to the S.

epidermidis genes that are responsible for producing the extra-

cellular polysaccharide required for the formation of biofilm,

we examined bacterial lawns on agar plates by scanning electron

microscopy. Of note, extracellular material was produced by

Hms-positive, but not Hms-negative, Y. pestis grown at room

temperature (figure 2).

Biofilm-like growth of Y. pestis in the flea: effect of hms

gene products. Previous studies have shown that wt and

Hms-negative strains of Y. pestis colonize the flea equally after

an infectious blood meal and achieve equivalent bacterial loads

in the midgut [8]. The most striking difference between the 2

at UC

SF Library and C

enter for Know

ledge Managem

ent on Decem


Dow

nloaded from



Figure 7. Biofilm-like phenotype of Yersinia pestis at the flea-mammal interface. In the dissected, blocked flea digestive tract shown in panels Aand B, 2 small matrix-embedded clusters of Y. pestis (arrows) have broken free from the bacterial mass in the proventriculus (PV) and traveled downthe esophagus (E). In the blocked digestive tract shown in panels C and D, a contiguous mass of Y. pestis within a viscous extracellular matrix extendsfrom the proventriculus through the esophagus.

strains is that Hms-negative Y. pestis never proceed to infect

the proventriculus but remain confined to the midgut. As a

result, Y. pestis hms mutants are completely unable to block

fleas, whereas Hms-positive Y. pestis infect both the proven-

triculus and the midgut, leading to proventricular blockage in

25%–50% of X. cheopis fleas [8]. Examination of a blocked

proventriculus by scanning electron microscopy confirmed the

presence of a bacterial mass associated with extracellular ma-

terial fixed among the proventricular spines (figure 3).

Both Hms-positive and -negative Y. pestis grew in the form

of large aggregates in the flea midgut but were dissimilar. The

aggregates formed by Hms-positive Y. pestis in the flea were

darkly pigmented, had complex structures, and were enveloped

in a viscous extracellular matrix (figure 4A). In contrast, Hms-

negative Y. pestis aggregates were light colored, had little 3-

dimensional structure, and were not associated with an obvious

extracellular matrix, although individual bacilli in the aggregate

were difficult to resolve, even at high magnification (figure 4B

and 4D). Hms-positive Y. pestis aggregates dissected from fleas

were also more cohesive and less easily disrupted than were

Hms-negative Y. pestis aggregates.

In an initial attempt to characterize the amorphous brown

matrix that surrounds Y. pestis in the flea gut, bacterial aggre-

gates were dissected from infected fleas and exposed to different

solvents, degradative enzymes, and histological stains (tables 1

and 2). The flea gut–associated extracellular matrix was not

soluble in aqueous solvents but was degraded by polar lipid

solvents; it also was resistant to proteases and to plasmin, an

enzyme that hydrolyzes fibrin. Bacterial masses dissected from

the flea midgut could be decolorized by treatment with the

hemin solvent 0.01N NaOH and by oxidation with 3% H2O2,

suggesting that hemin derived from the flea’s blood meal was

responsible for the brown color of the extracellular matrix.

Because the extracellular matrix of bacterial biofilms is typically

composed of bacterially-derived polysaccharide, we analyzed Y.

pestis aggregates dissected from infected fleas with differential

stains. The flea gut–associated extracellular material surround-

ing Hms-positive Y. pestis stained with PAS, which reacts with

polysaccharides and other complex polymers. It did not stain

with ruthenium red or calcofluor white, which react with poly-

anionic and b 1,3-linked polysaccharides, respectively. Hms-

positive Y. pestis aggregates dissected from fleas also stained

positive with lipid-soluble dyes. In contrast, the Hms-depen-

dent extracellular matrix produced by Y. pestis on agar media

did stain with ruthenium red but not with PAS or the other

polysaccharide and lipid stains. Hms-negative Y. pestis aggre-

at UC

SF Library and C

enter for Know

ledge Managem

ent on Decem


Dow

nloaded from



Figure 8. Increased resistance of flea-derived Yersinia pestis to humanpolymorphonuclear leukocytes (PMNLs). Relative bactericidal activity ofPMNLs to Hms-positive Y. pestis KIM6+ from a 21�C in vitro culture (�)or from infected fleas (�) and to Hms-negative Y. pestis KIM6 from a21�C culture (�) or from infected fleas (�). The mean and SD of 3independent experiments are shown.

gates dissected from flea guts were negative with all the stains

used (table 2). In addition, the Hms-dependent amorphous

brown matrix surrounding the aggregates in the flea gut was

not recognized in IFA assays by anti–Y. pestis polyclonal anti-

body. Peripheral and external bacteria stained strongly positive,

but the matrix itself and the bacteria enclosed within it did not

(figure 5). Hms-negative Y. pestis aggregates dissected from the

flea gut also stained poorly by IFA assay, except for peripheral

bacteria. This result and their appearance at high magnification

(figure 4D) indicates that Hms-negative aggregates also adsorb

material from the flea midgut.

Y. pestis QS systems and biofilm-like growth in the flea.

Bacterial QS systems have been implicated in regulating biofilm

development [35, 36]. Three such systems have been identified

in Y. pestis [15, 26], 2 that use autoinducer-1 type AHL signals

and 1 that uses the autoinducer-2 LuxS. We tested a Y. pestis

strain with loss of function mutations in all 3 autoinducer

synthase genes for its ability to form a biofilm and block fleas.

This Y. pestis triple-QS–negative mutant was able to produce

a biofilm on a glass surface (figure 1F) and to infect and block

fleas as well as the wt parent strain did (figure 6).

Y. pestis phenotype at the flea-host transmission interface.

A transmissible infection in the flea fits the description of a

bacterial biofilm—surface colonization by a dense microcolony

embedded in an extracellular matrix (figure 3D). In a completely

blocked flea, the biofilm fills the lumen of the proventriculus.

Bacteria embedded in the viscous matrix, either contiguous with

the blocking mass in the proventriculus or detached pieces of it,

were exuded into the esophagus after application of a coverslip

during preparation of dissected digestive tracts for microscopy,

indicating that the matrix was cohesive but somewhat fluid (fig-

ure 7). The bacteria transmitted by fleas—either individual cells,

small clusters, or both—undoubtedly derive from this viscous

biofilm-like mass.

Bacteria embedded in a biofilm have been shown to be more

resistant to uptake or killing by phagocytes [37]. For some bac-

teria, antiphagocytic properties have been attributed to the ex-

tracellular matrix itself [38, 39]. To examine the effect of the flea-

specific phenotype on phagocytosis, aggregates of Hms-positive

and -negative Y. pestis were dissected from infected fleas and trit-

urated and vortexed to release individual cells into suspension,

which were then added to human PMNLs. Bacteria from fleas

were significantly more resistant to killing by PMNLs than were

the same bacteria grown in liquid media under conditions in

which the Hms-dependent extracellular matrix is synthesized

(figure 8). The increased resistance was attributable to reduced

uptake of flea-derived Y. pestis by the PMNLs, because the kinetics

of phagocytosis mirrored those of killing (data not shown).

DISCUSSION

Y. pestis exhibits a distinct life stage in the flea vector. Unlike

infection in the mammal, the bacteria are not invasive in the

flea. The bacteria do not adhere to or penetrate the flea gut

epithelium but remain confined to the lumen of the digestive

tract (figure 4). This presents 2 potential problems. First, be-

cause fleas feed and excrete frequently, the bacteria are at con-

stant risk of being eliminated in the feces. In fact, nearly half

of X. cheopis fleas clear themselves of infection in this manner

even after feeding on highly septicemic blood [4, 7, 8]. Second,

transmission requires that Y. pestis, which are nonmotile, move

against the direction of blood flow when the flea feeds. Y. pestis

overcomes both of these problems by adopting a biofilm mode

of growth in the flea gut. Successful infection of the flea depends

on the formation of large, dense bacterial aggregates in the gut

lumen that are too large to pass from the midgut to the hindgut.

The flea-blocking bacterial aggregates have long been as-

sumed to be enveloped in a fibrin clot matrix formed from the

blood meal by a procoagulant activity of the Y. pestis plasmin-

ogen activator (Pla) [40]. This model is often cited in infec-

tious-disease textbooks. Our results provide further evidence

against the Pla-generated fibrin clot model of flea blockage. Not

only are Y. pestis strains that lack the pla gene able to infect and

block fleas just as wt strains do [3], but the matrix in which the

bacteria are embedded in the flea gut is completely resistant to

the fibrinolytic enzyme plasmin and to other proteases (table 1).

Although it is not required in the flea, the Y. pestis Pla is believed

to be important after transmission into the dermis, where it

facilitates dissemination from the site of the flea bite [41].

at UC

SF Library and C

enter for Know

ledge Managem

ent on Decem


Dow

nloaded from



As an alternative to the fibrin matrix model, our results

support the biofilm model proposed by Darby et al. [16], who

showed that both Y. pestis and Y. pseudotuberculosis produce

an Hms-dependent biofilm that accumulates on the external

mouthparts and that blocks the feeding of C. elegans and who

suggested that the same phenomenon is responsible for pro-

ventricular blockage in the flea. Infection of the proventriculus

by Y. pestis (figures 3, 4, 5, and 6) fits the operational definition

of a bacterial biofilm, which is formed by many bacteria in

many different environments. Developing a matrix-enclosed

biofilm is a common means by which bacteria colonize a surface

in spite of currents and shear forces [37]. For example, many

aquatic bacteria can alternate between a communal, surface-

attached biofilm and a dispersed planktonic life stage. Certain

chronic bacterial infections, such as periodontitis, endocardi-

tis, and infections of implanted medical devices, result from

biofilm formation in vivo [22, 37]. Y. pestis appears to use a

similar means to infect the hydrophobic surface of the pro-

ventricular spines, thus overcoming the pulsating valvular ac-

tion of the proventriculus and the inward flow of blood during

feeding that would tend to wash the bacteria back into the

midgut. Transmission of Leishmania parasites also depends on

a foregut-blocking phenomenon in the sandfly [42], but Y. pestis

is unique among arthropod-borne bacteria in using biofilm

development to produce a transmissible infection in the vector.

Growth of Y. pestis in the biofilm phenotype, both in vitro

and in the flea, requires the temperature-dependent synthesis of

hms gene products. HmsH and HmsF are outer membrane pro-

teins [43]. However, the presence of these 2 surface proteins is

not sufficient to produce blockage of fleas or binding of Congo

red and biofilm-like growth in vitro, because hmsR and hmsT

mutations abolish these phenotypes [8, 11, 12] (authors’ un-

published data). Complementation analyses further indicate that

all 5 identified hms genes are required for binding of Congo red

in vitro [10, 43]. In contrast, none of the 3 known QS systems

of Y. pestis is required to form a biofilm in vitro or to cause

proventricular blockage in fleas, despite the high cell densities

achieved (figure 6).

The biochemical composition of the Hms-dependent extra-

cellular material associated with Y. pestis in vitro and in the

flea is not known. Given the similarity of hmsF and hmsR to

nucleotide sugar transferase genes involved in polysaccharide

biosynthesis and the differential staining results of Hms-posi-

tive Y. pestis colonies (table 2), it is likely that the in vitro

extracellular material is polysaccharide. However, the extracel-

lular matrix around the bacteria in the flea gut appeared to be

heterogeneous and more complex. It was dark brown, probably

as a result of adsorbed hemin derived from red blood cells in

the flea’s diet. In addition, the hydrophobic, viscous nature and

the staining and solubility characteristics of the flea-associated

matrix (tables 1 and 2) indicated that it contained lipid, perhaps

also derived from blood. Thus, unlike the extracellular matrix

produced in vitro, the flea gut–associated matrix may be com-

posed of both bacterially derived polysaccharide and exogenous

digestive products of blood present in the flea gut. The matrix

was not recognized by hyperimmune serum (figure 5), which

also suggests that it does not consist entirely of bacterial products.

Y. pestis from fleas were more resistant to human PMNLs

than were Y. pestis from in vitro cultures (figure 8). This phe-

nomenon was not Hms dependent, however, suggesting that

other, unknown Y. pestis factors specific to the flea environment

are responsible. Lillard et al. [14] also found that the Hms

phenotype did not affect the resistance to human PMNLs of

Y. pestis grown in vitro in the presence or absence of hemin.

Adsorbed material from the flea gut may be responsible for

increased resistance to PMNLs, and unknown Y. pestis anti-

phagocytic gene products whose expression is specifically in-

duced in the flea may also contribute. In any case, as bacteria

are released from the biofilm and enter the mammalian host,

the flea-specific phenotype may confer some protection against

the innate immune response. This may be important imme-

diately after transmission, because known Y. pestis antiphago-

cytic factors, such as the F1 capsule and the type III secretion

system, are not produced in the flea [34, 44]. In addition, if

the matrix does contain lipid or other generic material derived

from the flea’s blood meal, this may mask bacterial antigens.

Y. pestis is a clonal variant of Y. pseudotuberculosis that

emerged within the last 1500–20,000 years [45]. Presumably,

the change from the food- and waterborne transmission of the

Y. pseudotuberculosis ancestor to the fleaborne transmission of

Y. pestis occurred during this short evolutionary period. The

hms genes are not required in the mammal; thus, their primary

biological function is to enable fleaborne transmission [8, 14].

Of interest, the 2 chromosomal hms loci are present in both

Y. pestis and Y. pseudotuberculosis, but the hms phenotype of

the 2 species differs. Colonies of most laboratory strains of Y.

pseudotuberculosis strains do not bind hemin or Congo red,

which is the signature Hms-positive phenotype in Y. pestis [46,

47]. Y. pseudotuberculosis, even Congo red–binding strains, do

not block the proventriculus of fleas, although they are able to

establish a chronic midgut infection [7] (authors’ unpublished

data). Thus, the evolutionary transition of Y. pestis to fleaborne

transmission appears to have involved the recruitment of en-

dogenous chromosomal genes for a new function—the pro-

duction of a proventriculus-blocking biofilm.

Acknowledgments

We thank M. Otto, R. Rebeil, and F. Sebbane, for review of the man-uscript, and A. Mora, for help with preparing the figures.

at UC

SF Library and C

enter for Know

ledge Managem

ent on Decem


Dow

nloaded from



References

1. Poland JD, Quan, TJ, Barnes AM. Plague. In: Beran GW, ed. Handbookof zoonoses. Boca Raton, FL: CRC Press, 1994:93–112.

2. Bacot AW, Martin CJ. Observations on the mechanism of the trans-mission of plague by fleas. J Hygiene Plague 1914; 13(Suppl 3):423–39.

3. Hinnebusch BJ, Fischer ER, Schwan TG. Evaluation of the role of theYersinia pestis plasminogen activator and other plasmid-encoded factorsin temperature-dependent blockage of the flea. J Infect Dis 1998; 178:1406–15.

4. Pollitzer R. Plague. Monograph series 22. Geneva: World Health Or-ganization, 1954.

5. Burroughs AL. Sylvatic plague studies: the vector efficiency of ninespecies of fleas compared with Xenopsylla cheopis. J Hygiene 1947; 45:371–96.

6. Eskey CR, Haas VH. Plague in the western part of the United States.Public Health Bulletin 254. Washington, DC: US Public Health Service,1940.

7. Hinnebusch BJ, Rudolph AE, Cherepanov P, Dixon JE, Schwan TG,Forsberg A. Role of Yersinia murine toxin in survival of Yersinia pestisin the midgut of the flea vector. Science 2002; 296:733–5.

8. Hinnebusch BJ, Perry RD, Schwan TG. Role of the Yersinia pestis heminstorage (hms) locus in the transmission of plague by fleas. Science 1996;273:367–70.

9. Fetherston JD, Schuetze P, Perry RD. Loss of the pigmentation phe-notype in Yersinia pestis is due to the spontaneous deletion of 102 kbof chromosomal DNA which is flanked by a repetitive element. MolMicrobiol 1992; 6:2693–704.

10. Lillard JW, Fetherston JD, Pedersen L, Pendrak ML, Perry RD. Se-quence and genetic analysis of the hemin storage (hms) system ofYersinia pestis. Gene 1997; 193:13–21.

11. Hare JM, McDonough KA. High-frequency RecA-dependent and-independent mechanisms of Congo red binding mutations in Yersiniapestis. J Bacteriol 1999; 181:4896–904.

12. Jones HA, Lillard JW, Perry RD. HmsT, a protein essential for ex-pression of the haemin storage (Hms+) phenotype of Yersinia pestis.Microbiology 1999; 145:2117–28.

13. Heilmann C, Schweitzer O, Gerke C, Vanittanakom N, Mack D, GotzF. Molecular basis of intercellular adhesion in the biofilm-formingStaphylococcus epidermidis. Mol Microbiol 1996; 20:1083–91.

14. Lillard JW, Bearden SW, Fetherston JD, Perry RD. The haemin storage(Hms+) phenotype of Yersinia pestis is not essential for the pathogenesisof bubonic plague in mammals. Microbiology 1999; 145:197–209.

15. Parkhill J, Wren BW, Thomson NR, et al. Genome sequence of Yersiniapestis, the causative agent of plague. Nature 2001; 413:523–7.

16. Darby C, Hsu JW, Ghori N, Falkow S. Caenorhabditis elegans: plaguebacteria biofilm blocks food intake. Nature 2002; 417:243–4.

17. Burrows TW, Jackson S. The pigmentation of Pasteurella pestis on adefined medium containing haemin. Br J Exp Pathol 1956; 37:570–6.

18. Surgalla MJ, Beesley ED. Congo red agar plating medium for detectingpigmentation in Pasteurella pestis. Appl Microbiol 1969; 18:834–7.

19. Hinnebusch BJ, Rosso M-L, Schwan TG, Carniel E. High-frequencyconjugative transfer of antibiotic resistance genes to Yersinia pestis inthe flea midgut. Mol Microbiol 2002; 46:349–54.

20. Joshua GWP, Karlyshev AV, Smith MP, Isherwood KE, Titball RW,Wren BW. A Caenorhabditis elegans model of Yersinia infection: biofilmformation on a biotic surface. Microbiology 2003; 149:3221–9.

21. Costerton JW, Lewandowski Z, Caldwell DE, Korber DR, Lappin-ScottHM. Microbial biofilms. Annu Rev Microbiol 1995; 49:711–45.

22. Costerton JW, Stewart PS, Greenberg EP. Bacterial biofilms: a commoncause of persistent infections. Science 1999; 284:1318–22.

23. Perry RD, Pendrak ML, Schuetze P. Identification and cloning of ahemin storage locus involved in the pigmentation phenotype of Yersiniapestis. J Bacteriol 1990; 172:5929–37.

24. Fetherston JD, Perry RD. The pigmentation locus of Yersinia pestis

KIM6+ is flanked by an insertion sequence and includes the structuralgenes for pesticin sensitivity and HMWP2. Mol Microbiol 1994; 13:697–708.

25. Donnenberg MS, Kaper JB. Construction of an eae deletion mutantof enteropathogenic Escherichia coli by using a positive-selection suicidevector. Infect Immun 1991; 59:4310–7.

26. Atkinson S, Throup JP, Stewart GSAB, Williams P. A hierarchical quo-rum-sensing system in Yersinia pseudotuberculosis is involved in theregulation of motility and clumping. Mol Microbiol 1999; 33:1267–77.

27. Blosser RS, Gray KM. Extraction of violacein from Chromobacteriumviolaceum provides a new quantitative bioassay for N-acyl homoserinelactone autoinducers. J Microbiol Methods 2000; 40:47–55.

28. Straley SC, Bowmer WS. Virulence genes regulated at the transcrip-tional level by Ca2+ in Yersinia pestis include structural genes for outermembrane proteins. Infect Immun 1986; 51:445–54.

29. Thompson SW. Selected histochemical and histopathological methods.Springfield, IL: Charles C. Thomas, 1966.

30. Mills J, Pulliam L, Dall L, Marzouk J, Wilson W, Costerton JW. Exopoly-saccharide production by viridans streptococci in experimental endo-carditis. Infect Immun 1984; 43:359–67.

31. Garcia-Vescovi E, Soncini FC, Groisman AA. Mg2+ as an extracellularsignal: environmental regulation of Salmonella virulence. Cell 1996; 84:165–74.

32. Fassel TA, Edmiston CE. Bacterial biofilms: strategies for preparingglycocalyx for electron microscopy. Meth Enzymol 1999; 310:194–203.

33. Kobayashi SD, Voyich JM, Buhl CL, Stahl RM, DeLeo FR. Globalchanges in gene expression by human polymorphonuclear leukocytesduring receptor-mediated phagocytosis: cell fate is regulated at the levelof gene expression. Proc Natl Acad Sci USA 2002; 99:6901–5.

34. Straley SC, Perry RD. Environmental modulation of gene expressionand pathogenesis in Yersinia. Trends Microbiol 1995; 3:310–7.

35. Davies DG, Parsek MR, Pearson JP, Iglewski BH, Costerton JW, Green-berg EP. The involvement of cell-to-cell signals in the development ofa bacterial biofilm. Science 1998; 280:295–8.

36. Kjelleberg S, Molin S. Is there a role for quorum sensing signals inbacterial biofilms? Curr Opin Microbiol 2002; 5:254–8.

37. Donlan RM, Costerton JW. Biofilms: survival mechanisms of clinicallyrelevant microorganisms. Clin Microbiol Rev 2002; 15:167–93.

38. Johnson GM, Lee DA, Regelmann WE, Gray ED, Peters G, Quie PG.Interference with granulocyte function by Staphylococcus epidermidisslime. Infect Immun 1986; 54:13–20.

39. Meluleni GJ, Grout M, Evans DJ, Pier GB. Mucoid Pseudomonas aeru-ginosa growing in a biofilm in vitro are killed by opsonic antibodiesto the mucoid exopolysaccharide capsule but not by antibodies pro-duced during chronic lung infection in cystic fibrosis patients. J Im-munol 1995; 155:2029–38.

40. Cavanaugh DC. Specific effect of temperature upon transmission ofthe plague bacillus by the oriental rat flea, Xenopsylla cheopis. Am JTrop Med Hyg 1971; 20:264–73.

41. Sodeinde OA, Subrahmanyam YV, Stark K, Quan T, Bao Y, GoguenJD. A surface protease and the invasive character of plague. Science1992; 258:1004–7.

42. Sacks D, Kamhawi S. Molecular aspects of parasite-vector and vector-host interactions in leishmaniasis. Annu Rev Microbiol 2001; 55:453–83.

43. Pendrak ML, Perry RD. Proteins essential for expression of the Hms+phenotype of Yersinia pestis. Mol Microbiol 1993; 8:857–64.

44. Perry RD, Fetherston JD. Yersinia pestis—etiologic agent of plague.Clin Microbiol Rev 1997; 10:35–66.

45. Achtman M, Zurth K, Morelli G, Torrea G, Guiyoule A, Carniel E. Yersiniapestis, the cause of plague, is a recently emerged clone of Yersinia pseu-dotuberculosis. Proc Natl Acad Sci USA 1999; 96:14043–8.

46. Perry RD, Brubaker RR. Accumulation of iron by yersiniae. J Bacteriol1979; 137:1290–8.

47. Brubaker RR. Factors promoting acute and chronic diseases caused byyersiniae. Clin Microbiol Rev 1991; 4:309–24.

at UC

SF Library and C

enter for Know

ledge Managem

ent on Decem


Dow

nloaded from


Transmission of Yersinia pestis from an Infectious Biofilm in the Flea...

Documents

Transcript of Transmission of Yersinia pestis from an Infectious Biofilm in the Flea...