Hfq Regulates Biofilm Gut Blockage That Facilitates Flea-Borne Transmission of Yersinia pestis

Abstract

The plague bacillus Yersinia pestis can achieve transmission by biofilm blockage of the foregut proventriculus of its flea vector. Hfq is revealed to be essential for biofilm blockage formation and acquisition and fitness of Y. pestis during flea gut infection, consistent with posttranscriptional regulatory mechanisms in plague transmission.

INTRODUCTION

Yersinia pestis, the etiological agent of bubonic plague, is transmitted to humans by fleabite. Colonization and biofilm formation by Y. pestis in the flea gut are essential steps for proventricular foregut blockage and facilitate subsequent transmission of Y. pestis during “frustrated feeding” by an infected flea (16, 20). The known transmission factors hmsHFRS, gmhA, and ymt are required for flea-borne transmission but are dispensable for infection within the mammalian host (8, 16–19). Whole-genome comparative transcriptional profiling revealed, however, that these three genes are not differentially transcribed during Y. pestis biofilm formation and gut blockage of the flea relative to temperature-matched in vitro conditions (40). Clearly, the transcriptome provides only a first order of gene regulation, while a second order of posttranscriptional regulation may shape the unique flea-associated biofilm that allows transmission. This is in keeping with the emerging paradigm for extensive posttranscriptionally regulated virulence in numerous bacterial pathogens (6, 25, 29, 32–34) mediated by the ubiquitous RNA chaperone protein Hfq. The characterization of Hfq function by the use of rodent infections with several bacterial pathogens, including Salmonella spp. (34), Francisella tularensis (28), uropathogenic Escherichia coli (24), and Y. pestis (13), has shown significant virulence attenuation in hfq mutants.

The hfq gene is abundantly transcribed during Y. pestis biofilm proventricular blockage (40), predicting a second order of posttranscriptional regulation directed by Hfq during Y. pestis flea infection. To investigate the role of Hfq in mediating flea gut blockage, an hfq deletion mutant in the Y. pestis KIM6+ strain was constructed using the lambda red recombinase system (9). This strain contains the hmsHFRS gene locus required for the synthesis of extracellular matrix polysaccharide (EPS) essential for biofilm formation and flea foregut blockage (18, 31, 39). We hypothesized that the Δhfq mutant growing in vitro at the 21°C insect host temperature or within the flea gut would be impaired in its growth and biofilm gut blockage capability essential for plague survival and transmission.

Growth of a Y. pestis KIM6+ hfq deletion mutant.

Functional mutations of hfq frequently result in compromised bacterial growth fitness (6, 13), including in Y. pestis growing at the mammalian host temperature of 37°C versus laboratory growth at 28°C (3, 13). Here, at 21°C, the insect host temperature, the Δhfq mutant, relative to its isogenic wild type, exhibited a significantly impaired specific growth rate and a decrease in stationary-phase cell density (Fig. 1A) similar to those seen with growth at 37°C (3, 13) and at 28°C (Fig. 1B) in brain heart infusion broth (BHI). The Δhfq mutant, complemented with the native gene and promoter region of hfq on a high-copy-number plasmid, pCR4-TOPO (Invitrogen), sufficiently rescued this growth fitness defect in BHI broth. However, the Δhfq mutant showed no significant growth fitness defect compared to the wild-type strain when grown in biofilm-supporting minimal TMH medium (supplemented with 0.2% galactose as the carbon source instead of gluconate and 1 μg/ml hemin) (37).

Fig 1.

Growth fitness comparison of Y. pestis KIM6+ wild-type (●), Δhfq mutant (■), and complemented mutant Δhfq (pCR4::hfq) (▲) strains at 21°C (A) and 28°C (B) in brain heart infusion (BHI) broth and wild-type (●) and Δhfq mutant (■) strains at 21°C in TMH medium (0.2% galactose, 1 μg/ml hemin) (C). The specific growth rate, μ, represents the linear regression slope calculated from exponential-phase time spans 4 to 10 h and 17 to 28 h for growth in BHI broth and TMH broth, respectively. Means and standard deviations of the results of 3 independent experiments are indicated. Repeated analysis of variance (ANOVA) measurements were used to determine significant differences in slopes (μ). * indicates P < 0.001.

The Δhfq mutant displays a differential ability to form biofilm in vitro.

Biofilm formation is a dynamic process that is initiated by bacterial adherence and development of multicellular aggregates on a favorable surface (22, 27). To examine biofilm formation on different surfaces, Y. pestis strains were grown in borosilicate glass tubes and on polystyrene plates at 21°C with shaking in biofilm-supporting minimal TMH medium (supplemented with 0.2% galactose and 1 μg/ml hemin), and the biofilm was quantified by safranin staining (7, 39). Relative to the wild-type strain, the Δhfq mutant was unable to grow a comparable biofilm on polystyrene (Fig. 2A) but retained a wild-type ability to form biofilm on borosilicate glass (Fig. 2B) over time. The slightly lower growth rate of the Δhfq mutant may account for the reduced ability to form biofilm up to 48 h. However, there was no difference in the stationary-phase cell density between the wild-type and Δhfq mutant strains under these minimal culture conditions (Fig. 1C), indicating disparate abilities of these strains to form biofilm on different surfaces.

Fig 2.

Differential ability of the Δhfq mutant to form biofilm in vitro. (A and B) Growth of Y. pestis KIM6+ wild-type (black bars), Δhfq mutant (unfilled bars), and complemented mutant Δhfq (pCR4::hfq) (dark gray bars) strains in TMH medium (0.2% galactose, 1 μg/ml hemin) seeded at 10³ bacteria/ml on polystyrene microtiter plates (A) and in borosilicate glass tubes (B). (C) The Δhfq mutant does not form rugose (wrinkled) colonies. Means and standard deviations of the results of 3 independent experiments are indicated for panels A and B. Data were analyzed using two-way ANOVA and a Bonferroni posttest. * indicates P < 0.05; **** indicates P < 0.001.

A Y. pestis strain with the hmsHFRS gene locus can be assessed for biofilm formation at <28°C by streaking bacteria on agar medium containing Congo red dye on which pigmented red colonies form when the bacterial cells bind Congo red, which correlates with the ability to form a biofilm that is largely dependent on production of EPS (39). The wild-type and complemented Δhfq mutant strains formed large red pigmented colonies with rugose centers, while the Δhfq mutant colonies appeared to have smooth centers (Fig. 2C) on Congo red agar. Rugose colony morphology represents enhanced EPS production and biofilm (2, 35, 38, 42).

The Δhfq mutant is unable to form proventricular biofilm blockage of the flea gut.

To determine whether the Δhfq mutant infects, survives, and forms biofilm blockage of the proventriculus, 150 Xenopsylla cheopis fleas were infected with wild-type, Δhfq mutant, or complemented mutant Δhfq (pCR4::hfq) strains using a previously described artificial feeding system (18, 20) and a blood meal containing a mean of 9.61 × 10⁹ (±1.52 × 10¹⁰) bacteria/ml. Fleas that took a blood meal were maintained at 21°C and 75% relative humidity, fed twice weekly on uninfected mice, and monitored for proventricular blockage (18). To quantify bacterial survival, 20 fleas were processed at days 0, 7, and 28 postinfection (18, 20, 39). In the flea gut, the Δhfq mutant was able to grow at the same rate as the wild-type strain (Fig. 3A), achieving similar bacterial loads at 28 days postinfection. However, significantly fewer fleas remaining infected with the Δhfq mutant by 28 days (Fig. 3B). Notably, the Δhfq mutant was unable to form a biofilm blockage of the flea gut (Fig. 3C).

Fig 3.

The Δhfq mutant is deficient in its ability to form biofilm blockage of the flea proventriculus. (A) The Δhfq mutant grows at a rate similar to that of the wild-type strain during flea infection. ■, wild-type strain; ○, Δhfq mutant strain; ▲, complemented mutant Δhfq (pCR4::hfq) strain; Δ, Δhfq mutant strain pregrown at 21°C. (B) The Δhfq mutant strain had reduced infection rates in fleas 28 days postinfection. Black bars, wild-type strain; unfilled bars, Δhfq mutant strain; dark gray bars, complemented mutant Δhfq (pCR4::hfq) strain; hatched bars, Δhfq mutant strain pregrown at 21°C. (C) The Δhfq mutant has a reduced ability to form biofilm blockage of the flea proventriculus. The means and standard deviations of the results of 3 independent experiments are represented for all infections except for the infections of the Δhfq mutant strain pregrown at 21°C, for which the data represent 2 independent experiments. A one-way ANOVA and a Newman-Keuls posttest (* indicates P < 0.05) were used to statistically analyze wild-type, Δhfq mutant, and Δhfq (pCR4::hfq) strains grown at 37°C.

The ability to form biofilm blockage may be limited by infectious dose (10, 20) or defective acquisition of the pathogen; fleas infected with the Δhfq mutant showed lower infectious doses (≤50% wild-type CFU/flea at T = 0) immediately following feeding compared to the wild type (Fig. 3A). The lower infectious dose may be correlated with the poor growth, physiological effects, and elongated cell morphology of the Δhfq mutant grown at 37°C in BHI broth (3). Therefore, the Δhfq mutant was grown at 21°C, to adapt cells to the flea growth temperature, following which fleas were infected as described above using 1.95 × 10⁸ (±7.07 × 10⁶) bacteria/ml. The fleas acquired a higher infectious dose of the Δhfq mutant pregrown at 21°C; however, there was no biofilm blockage. This confirmed that the Δhfq mutant was defective in forming biofilm blockage in a manner independent of the infectious dose.

To further evaluate the liability of the hfq mutation, fleas were coinfected with the wild type carrying the green fluorescent protein (GFP)-expressing pACGFP1 plasmid (Clontech) and the Δhfq mutant carrying plasmid pmCherry (Clontech) expressing red fluorescent protein by the use of a 1:1 ratio of 3.5 × 10⁸ (±1.2 × 10⁸) bacteria/ml from each strain cultured at 37°C. The competitive index (CI) was calculated as previously reported (26) 28 days postinfection to determine the fitness of the Δhfq mutant versus that of the wild type following flea infection. The CI was determined by calculating the ratio of mean wild-type Y. pestis CFU/flea (X) and the mean Δhfq mutant CFU/flea (Y) at 28 days postinfection divided by this same ratio at 0 days postinfection [(X/Y)_{t = 28 days}/(X/Y)_{t = 0 days}]. A CI of <1 would denote the wild type as the dominant strain, whereas a CI of >1 would denote dominance of the Δhfq mutant. In two independent coinfection experiments, the CI was 0.22 (±0.06), which indicated a lower mean bacterial load per flea and reduced fitness of the Δhfq mutant relative to the wild type in competition. Fluorescence microscopy of the coinfected flea guts demonstrated that the Δhfq mutant formed smaller aggregates that developed separately from the wild-type strain (Fig. 4).

Fig 4.

Coinfection of fleas with green fluorescent protein (GFP)-expressing wild-type (light-colored aggregates—green arrow) and mCherry red fluorescent protein-expressing Δhfq mutant (dark-colored aggregates—red arrow) strains exhibited that the Δhfq mutant strains form smaller aggregations that develop separately from the wild-type strains during flea infection, as visualized by fluorescence microscopy. (A) The left panel illustrates a whole uninfected flea midgut, and the right panel illustrates a coinfected flea gut containing GFP-expressing wild-type (green arrow) and mCherry-expressing Δhfq mutant (red arrow) strains. (B) Coinfected flea gut: panel 1, bright-field image showing GFP-expressing wild-type (green arrow) and mCherry-expressing Δhfq mutant (red arrow) strains; panel 2, separately developing aggregations of GFP-expressing wild-type and mCherry-expressing Δhfq mutant strains; panel 3, GFP-expressing wild-type bacteria only; panel 4, mCherry-expressing Δhfq mutant aggregations only. Images represent 8 fleas per experiment dissected between 21 and 28 days after infection. PV, proventriculus (yellow arrows).

The Δhfq mutant's inability to form a biofilm blockage of the foregut while surviving at high numbers in the flea is reminiscent of the phenotype of the Y. pestis KIM6 pgm mutant strain, which lacks the hmsHFRS EPS locus (18). However, 28 days after infection, 50 of 50 colonies grown from fleas (n = 16) infected with the Δhfq mutant were pigmented on Congo red agar and were PCR positive for the hmsF gene. This ruled out spontaneous loss of the pgm locus (12), a known phenomenon which could have been enhanced in the Δhfq mutant background.

We accept our hypothesis that Hfq is required for biofilm-mediated gut blockage in the flea. Development and assembly of the biofilm are affected by the coordinated synthesis of components, including surface appendages, signaling molecules promoting intercellular communication, and molecules essential for nutrient acquisition, stress adaptation, and biofilm regulation (1, 22, 27). In Y. pestis, the hmsHFRS-encoded EPS is a key component of the biofilm structure and its composition and is posttranslationally regulated by the chemical second messenger molecule, cyclic-di-GMP (5, 39). Although other outer surface factors impacting biofilm formation have not yet been defined, the observed autoaggregative phenotype of a Y. pestis mutant impaired in EPS synthesis (39) may imply a role for such factors.

It is advantageous for Y. pestis to form large, dense multicellular aggregates to establish infection in the flea gut lumen, as these cannot be easily expelled in the feces (20). Importantly, biofilm blockage of the foregut proventriculus of the flea vector is the classical biological mechanism for transmission of the plague. Fewer fleas may remain infected with the Δhfq mutant because it forms smaller bacterial aggregates, and altered intercellular signaling may contribute to the separated development of its bacterial aggregates. The lower numbers of the mutant during coinfection, however, may also reflect its inefficiency at acquiring available nutrients essential for survival and growth when in competition with the wild type. This could be influenced by an altered spatial structure of the Δhfq mutant biofilm (1, 27). Stress adaptation signatures associated with nutrient starvation and membrane stresses are common in bacterial biofilms (1), including that of Y. pestis (40), and may be reduced in the Δhfq mutant. This deficiency of the Δhfq mutant may prove a competitive fitness disadvantage in sylvatic plague transmission, when the flea gut may be prone to increased colonization by other bacteria (11, 21). The difference in biofilm production between the wild type and the Δhfq mutant may therefore be a consequence of alterations in several factors required for biofilm formation, development, and structure. This suggests a role for Hfq in modulating the dynamics of biofilm formation, optimal acquisition, and fitness during Y. pestis flea infection.

How Hfq functions in biofilm formation in the flea is currently unknown. However, a known function of Hfq is to stabilize and facilitate interaction of small regulatory RNA (sRNAs) with their cognate mRNA targets. The Hfq-regulated sRNAs function by base pairing to their cognate mRNAs and hindering their translation and/or increasing their degradation or sometimes enhancing translation (33, 36, 41). Hfq and its associated sRNAs are implicated in biofilm development in uropathogenic E. coli (24) and Vibrio cholera (4) and modulation of the outer surface composition and envelope stress responses in other Gram-negative enterobacteria (14, 15, 30).

A Y. pestis Δhfq mutant has been shown to be attenuated for virulence in a mouse infection model (13) and was observed here to be deficient in transmission by biofilm blockage of the flea proventricuclus. This implicates Hfq-mediated mechanisms in the complete infective life cycle of the plague. The influence of a single regulatory protein on Y. pestis infection in two physiologically distinct host environments with discrete gene expression profiles (40) is interesting and not a previously described phenotype for Hfq. This suggests that Hfq may have separate targets in the two different host environments. Hfq-dependent sRNAs have recently been reported to contribute to virulence of Y. pestis (23). Therefore, uncovering the mechanisms and targets by which Hfq mediates transmission is the next step in better understanding of plague and in improving disease control.