OUP user menu

Assessing adhesion, biofilm formation and motility of Acidovorax citrulli using microfluidic flow chambers

Ofir Bahar, Leonardo De La Fuente, Saul Burdman
DOI: http://dx.doi.org/10.1111/j.1574-6968.2010.02094.x 33-39 First published online: 1 November 2010


Acidovorax citrulli is the causal agent of bacterial fruit blotch of cucurbits. We have shown previously that type IV pili (TFP) are required for wild-type levels of virulence of A. citrulli on melon and that this pathogen can colonize and move thorough the xylem vessels of host seedlings. Here, comparative studies between wild-type and TFP mutant strains using microfluidic flow chambers demonstrated that TFP play a critical role in both the surface attachment and the biofilm formation of A. citrulli under a medium flow. Additionally, TFP null mutants were unable to perform twitching movement against the direction of medium flow. Assays using a flagellin mutant showed that, in contrast to TFP, polar flagella do not contribute to the adhesion and biofilm formation of A. citrulli under tested conditions. Also, flagellum-mediated swimming motility of wild-type strains was not observed under medium flow. These results imply that TFP may play an important role in colonization and spread in the xylem vessels under sap flow conditions, while polar flagella could be more important for spread during periods of time when xylem flow is minimal.

  • Acidovorax citrulli
  • biofilm
  • motility
  • twitching
  • adhesion


Acidovorax avenae ssp. citrulli (Schaad et al., 1978; Willems et al., 1992) is a Gram-negative bacterium that causes bacterial fruit blotch (BFB) of cucurbit plants. The destructive potential of this bacterium was fully realized during the late 1980s, following severe BFB outbreaks in watermelon fields in the United States that led to high yield losses of up to 100% (Latin & Hopkins, 1995; Schaad et al., 2003). Since then, the disease has spread to different parts of the world, causing severe yield losses in watermelon and melon (Bahar & Burdman, 2010). Recently, Schaad (2008) suggested a new classification for subspecies of A. avenae, with A. avenae ssp. citrulli being reclassified as Acidovorax citrulli. Herein, we adopt this new nomenclature.

Molecular, biochemical and host-range characterization of A. citrulli isolates revealed the existence of two distinct groups: group I includes strains isolated mainly from nonwatermelon plants, while group II includes strains isolated mostly from watermelon (Walcott et al., 2000, 2004; Burdman et al., 2005; Feng et al., 2009). The genome of a group II strain (AAC00-1) has been sequenced recently by the Joint Genome Institute.

Few studies have shed light on the transmission mechanisms of A. citrulli inside the plant. It has been shown that this bacterium can penetrate the plant through blossoms and subsequently infect seeds (Walcott et al., 2003; Lessl et al., 2007). In a recent study, we showed that A. citrulli can systemically infect melon seedlings and can move basipetally and acropetally through the xylem vessels (Bahar et al., 2009).

The aforementioned studies suggest that motility contributes to both infection and translocation of the bacterium throughout the plant. Indeed, we demonstrated recently that type IV pili (TFP) and polar flagella, mediating twitching and swimming motilities, respectively, are required for the full virulence of A. citrulli on melon seedlings (Bahar et al., 2009; O. Bahar and S. Burdman, unpublished data). Nevertheless, the roles of TFP and polar flagella in xylem colonization and translocation inside the plant are not yet understood.

Microfluidic flow chambers (MFCs) mimic the xylem vessels of plant vascular systems (Meng et al., 2005) and have been used as a model system to investigate the behavior of bacteria under flow conditions. For instance, MFC studies with Xylella fastidiosa, a xylem-limited pathogen that lacks flagella and causes Pierce's disease of grapes (Meng et al., 2005; De La Fuente et al., 2007a, b), demonstrated the ability of X. fastidiosa to move against medium flow with the assistance of TFP and to strongly adhere to surfaces by means of type I pili (De La Fuente et al., 2007a, b).

We hypothesize that the observed reduced virulence of A. citrulli TFP and polar flagellum mutants on seedlings is at least in part due to their reduced abilities to adhere to and form biofilms on the vascular tissue, and to spread against xylem flow. Therefore, the objective of this study was to investigate A. citrulli behavior under xylem flow-mimicking conditions, with an emphasis on surface adhesion, biofilm formation and movement. In particular, we aim to define the role of TFP and flagella during the infection process of A. citrulli. Here, we used the MFC technology to compare the group I wild-type strain M6 with a TFP null mutant M6-M (M6 impaired in the TFP assembly gene pilM) and with a hyperpiliated mutant (M6-T, impaired in pilT that encodes an ATPase protein required for TFP retraction and twitching). An M6 mutant lacking polar flagella (M6-flg) was also assessed. To authenticate the role of TFP in A. citrulli in the MFC system, experiments using the group II wild-type strain W1 compared with its TFP null mutant W1-A (impaired in pilA, encoding pilin, the major TFP subunit) were also conducted.

Materials and methods

Bacterial strains and growth conditions

Acidovorax citrulli strains and their characteristics are described in Table 1. For MFC studies, strains were grown in Nutrient Broth (Difco) at 28 °C with shaking (200 r.p.m.) until the midlog phase. Cultures were then collected using a sterile 1-mL syringe and introduced into the MFCs. Assays were set at 25 °C according to De La Fuente (2007b) and lasted 3–8 days.

View this table:

Acidovorax citrulli strains used in this study and their characteristics

StrainCharacteristicsTFPPolar flagellaTwitching motilitySwimming motilityReference
M6Group I wild-type strain; ApR++++Burdman (2005)
M6-MM6 Tn5-mutant defective in pilM (TFP assembly protein); ApR, KmR++Bahar (2009)
M6-TM6 insertional mutant defective in pilT (ATPase protein); ApR, KmR++−/+Bahar (2009)
M6-flgM6 insertional mutant defective in the gene encoding flagellin (fliC); ApR, KmR++This study
W1Group II wild-type strain; ApR++Burdman (2005)
W1-AW1 insertional mutant defective in pilA (major TFP subunit pilin); ApR, KmRBahar (2009)
  • * KmR and ApR, kanamycin and ampicillin resistance, respectively.

  • Swimming motility of mutant M6-T is dependent on the experimental conditions.

Generation of the flagellin mutant M6-flg

A mutant impaired in flagellin was generated on the background of wild-type M6. Primers Flg-mut-F (5′-GCCGAATTCGCAGACCAAGACCGTCAACG-3′) and Flg-mut-R (5′-GCCGGATCCTTGATGTCCTTGCCCGACTCGTT-3′) were designed based on the Aave_4400 sequence (fliC) of strain AAC00-1 (http://genome.jgi-psf.org/aciav/aciav.info.html). The amplified fragment, which does not span the 3′- and 5′-ends of the gene, was digested with EcoRI and BamHI (the restriction sites are underlined in the above primer sequences) and cloned into the suicide vector pJP5603 (Penfold & Pemberton, 1992), conferring kanamycin (Km) resistance. Transformation into strain M6 was carried out as described previously (Bahar et al., 2009). Putative mutants were selected on NA with Km at 50 μg mL−1, and verified by Southern blot. Loss of swimming motility was confirmed in soft agar (0.3%) plates and under the microscope (not shown).

MFCs and microscopy

MFCs were fabricated as described previously (De la Fuente et al., 2007b). Briefly, the chamber body was constructed with polydimethylsilioxane and consisted of two parallel channels measuring 80 μm wide, 3.7 cm long and 50 μm high, separated by a 50 μm wide polydimethylsilioxane ridge. Chamber bodies were then sandwiched between a cover glass and a supporting glass microscope slide. Teflon tubes were attached to inlet and outlet channels, and media were introduced into the channels using syringes controlled by pumps (Pico Plus, Harvard Apparatus). The chambers were mounted on a Nikon Ti/U E20L80 microscope (Nikon Co.) using 40 × phase-contrast and differential interference contrast optics. Time-lapse images were recorded using a DS-Qi1Mc digital camera and analyzed using nis elements software (Nikon Co.).

Assessment of bacterial adhesion force

The adhesion abilities of bacterial cells were evaluated using a modification of a described procedure (De La Fuente et al., 2007b): (1) cells were introduced from side channels, while the flow in the main channels was stopped, allowing cells to attach; (2) introduction of cells from the side channels was stopped and medium flow in the main channels was resumed at a rate of 0.25 μL min−1 to remove unattached cells; and (3) the flow rate in the main channels was gradually increased from 0.25 to 0.5, 1, 2, 4, 8, 16, 32 and 64 μL min−1, each rate being maintained for one minute. Time-lapse movies were captured during the course of the assay and cells attached to the glass surface were quantified using nis elements software. Each repetition of steps 1–3 was considered a replicate. For each strain, at least three replicates in different locations along the channels were measured. For each flow rate, the amount of cells washed from the field of view was calculated as a function of the total number of cells present at the beginning of the assay. At the end of each flow rate, the number of attached cells was determined by averaging the amount of attached cells in the last three frames of that time period (corresponding to the last 15 s of the corresponding flow rate). Adhesion forces were determined according to De La Fuente (2007b).

Biofilm formation and cell motility

Biofilm formation was monitored inside the MFCs by maintaining a flow rate of 0.25 μL min−1 in the main channel and capturing images at 30-s intervals for a period of 6–24 h. Swimming and twitching were assessed for all strains inside the MFCs. Twitching motility rates were calculated for six bacterial cells according to De La Fuente (2007a).

Statistical analyses

All experiments were repeated at least three times and data were subjected to the Tukey–HSD test using jmp in v3.2.1 (SAS Institute Inc.). For comparison of adhesion forces, one-way anova were performed using statistix 8.0 (Analytical Software). Differences among means were determined using Fisher's protected LSD test at P=0.05.


TFP mutants possess reduced attachment ability

All the experiments described were based on the interaction between bacterial cells and the glass surface of the MFCs; however, similar observations were made on polydimethylsilioxane surfaces (not shown). Differences were evident among the wild types and TFP mutants in their ability to adhere to glass as soon as cells were introduced into the MFCs. While M6 and W1 cells attached immediately to the surface, the respective TFP mutants failed to do so (Fig. 1). Blocking medium flow at the main channel for up to 90 min resulted in the accumulation of TFP mutant cells in the field of view. However, when medium flow was resumed (0.25 μL min−1), all TFP mutants (all M6-M and the majority of M6-T and W1-A cells) were immediately displaced.


Differences in attachment abilities among wild-type and mutant strains of Acidovorax citrulli in microfluidic chambers. All pictures were taken during the first 40 min after cell introduction while the flow rate in the chambers was 0.25 μL min−1. Arrows indicate culture inlet points. Pictures are representative of results from three independent experiments.

In contrast to M6-M cells, which, under flow, were unable to adhere to the channel surface regardless of the incubation time, M6-T and W1-A cells showed sporadic attachment after 24 h of incubation. Interestingly, the hyperpiliated M6-T cells attached to the surface not only as solitary cells but also as small clusters of about 5–15 cells (Fig. 2). No apparent differences were observed between M6 and M6-flg, as both effectively attached to the surfaces (not shown).


Comparison of adhesion strength in microfluidic chambers during an increase of the medium flow rate and adhesion force measurements. (a) Sample pictures from one assay with M6 and M6-T strains. Numbers below pictures indicate the flow rate in μL min−1 and the time under the corresponding flow rate at which the picture was taken. (b) Quantitative estimation of cell attachment (left y-axis, dark gray columns) and adhesion force (right y-axis, light gray columns) of wild-type M6 and mutants M6-T and M6-flg. The left y-axis represents the percentage of cells that remained attached to the surface after 1 min under the maximum flow rate (64 μL min−1), relative to the beginning of the assay. The right y-axis represents the adhesion force of the cells. Averages and SEs from at least three replicates per strain are shown. Different letters indicate significant differences among strains (P=0.05 for cells attachment, upper case letters; P=0.01, for adhesion force, lower case letters).

TFP contribute to the strength of bacterial adhesion

Adhesion force evaluation assays were conducted to compare the strength of attachment among wild types and mutants. This assay was not performed with mutant M6-M, due to its inability to attach to the surface under tested conditions.

Gradually increasing the flow rate from 0.25 to 16 μL min−1 did not reveal substantial differences between strains M6 and M6-T in attachment ability. However, following the application of flow rates of 32 and 64 μL min−1, the majority (84%) of the M6-T cells were displaced from the surface (Fig. 2; Supporting Information, Movie S1). Under these conditions, only 37% of the M6 cells were displaced from the surface, and the differences between these strains were significant (P=0.05) (Fig. 2b). Accordingly, wild-type M6 showed a significantly (P=0.01) higher adhesion force (174.8 pN) than M6-T (104.4 pN) (Fig. 2b). For a qualitative assessment of the strength of attachment, the flow rate was increased to 100 μL min−1, equivalent to a drag force of 380 pN (De la Fuente et al., 2007b) for 1 min. Here too, the majority of M6 cells that withstood the previous rate of 64 μL min−1 remained attached to the surface. No significant differences were observed between M6 and M6-flg in adhesion assays (Fig. 2b).

Wild-type W1, which, in contrast to strain M6, does not produce polar flagella (Table 1), showed a behavior similar to that of M6, with an average of 49% of the initial cells being displaced from the surface at the end of the assays (not shown). On the other hand, the majority of W1-A cells were quickly removed from the surface following the application of a flow rate of 8 μL min−1. The few remaining W1-A cells were displaced from the surface when the flow rate was increased to 32 μL min−1 (not shown).

Polar flagella- and TFP-mediated motilities inside MFCs

During incubation inside the chambers, even at the minimum flow of 0.25 μL min−1, swimming motility was not observed for strains M6 and M6-M. However, when medium flow was stopped, random swimming was immediately observed for both strains. This implies that cells of these strains possessed functional flagella, and that the lack of swimming was likely due to the medium flow being too strong to allow swimming movement. As expected, swimming was not observed for strains W1 and M6-flg under the tested conditions (not shown).

Under the tested conditions in MFC, it was difficult to observe twitching of strain M6. The more common form of movement was characterized by cells moving 1–4 μm, up and down the channel, perpendicular to the direction of medium flow. Another typical form of movement for M6 was characterized by cells spinning around without moving to a certain direction. M6-flg showed movement patterns similar to M6. Twitching movement was not observed for either of the TFP mutants.

Twitching of W1, on the other hand, was frequently observed in the opposite direction of medium flow (0.25 μL min−1), immediately after cells attached to the surface. Cells moved for short distances, typically 10–20 μm against the flow, before being removed from the surface. An estimation of the twitching speed indicated that cells moved at approximately 9.9 ± 1.1 μm min−1.

Functional TFP contribute to biofilm formation

In all assays, whenever biofilms were formed, we observed a succession of characteristic events. First, a biofilm never formed sooner than 48 h after the beginning of the assay, and in some experiments, it occurred only after 72 h (shown in Fig. 3 for strain W1), regardless of the cell density. Second, after the biofilm was formed, and even before it had completely filled up the field of view, chunks of cells continuously disconnected from the biofilm, which immediately grew back to fill up the gaps formed by the disconnecting chunks (shown for W1 in Movie S2). Third, following biofilm disassembly, the time required for a biofilm to re-grow was considerably faster (6–8 h) than the time required for the initial biofilm to fill up the field of view (∼20–24 h). This pattern of biofilm disassembly and regrowth was described for other bacteria and is considered a form of cell redistribution (Dow et al., 2003).


Biofilm formation by Acidovorax citrulli W1 inside the microfluidic chambers. The first frame (top) was taken 72 h after assay initiation. The capture time of subsequent pictures is indicated on each picture in parentheses. The pictures are from a replicate of one experiment of three with similar results. A complete movie of this replicate can be seen in Movie S2.

Biofilm formation as described above was typical of wild types M6 and W1, as well as mutant M6-flg. Strain M6-T was able to form a biofilm, but was slower in filling up the field of view (not shown). It appeared that the M6-T biofilm grew mainly due to cell division rather than both movement and cell division as observed for the wild types. Because mutant M6-T possesses TFP, but is impaired in twitching motility, this is understandable. The TFP-null mutants M6-M and W1-A did not form biofilms at any stage (not shown).


While most biological assays are based on a final outcome, the MFC device enables continuous monitoring of cells, allowing a fascinating glance into the microorganism's world in ‘real time.’ Here, we used MFCs to assess several behaviors of wild types and TFP/polar flagellum mutants of A. citrulli.

TFP and polar flagella are involved in motility, attachment and biofilm formation in different bacterial species (Josenhans & Suerbaum, 2002; Mattick, 2002; Craig et al., 2004). We have demonstrated previously that TFP and polar flagella are involved in the pathogenicity of A. citrulli (Bahar et al., 2009; O. Bahar and S. Burdman, unpublished results). We also showed that functional TFP are required for biofilm formation of this bacterium on glass and polystyrene surfaces (Bahar et al., 2009).

Acidovorax citrulli has the ability to colonize the xylem vessels of melon seedlings (Bahar et al., 2009). Here, studies with xylem-mimicking MFCs revealed an even more drastic effect of TFP on surface attachment and biofilm formation. Under flow conditions, cells of the TFP-null mutant M6-M were unable to attach to the surface. This result was in contrast to findings from conventional assays, where cell attachment and biofilm formation by this mutant were observed to some extent (Bahar et al., 2009). These results were corroborated by the use of an additional TFP-null mutant in a different A. citrulli strain, W1-A, which is impaired in pilA (major TFP subunit pilin), and showed a behavior similar to that of M6-M in MFCs. The W1-A mutant, generated in the background of wild-type M6, was used in these assays because numerous attempts to generate a pilA mutant in the background of strain M6 were unsuccessful (Bahar et al., 2009). It is important to mention that strain W1 is not a typical A. citrulli strain as it lacks a polar flagellum and possesses reduced virulence in comparison with other group II strains of this bacterium (Bahar et al., 2009). Nevertheless, in this specific study, utilization of the W1-A mutant served as an additional means to assess the role of A. citrulli TFP in the MFC system.

An interesting phenotype was seen with the hyperpiliated pilT mutant M6-T. In contrast to M6-M, M6-T cells were able to attach to the surface; however, the strength of attachment was significantly weaker than M6, supporting the fact that functional TFP is crucial for surface attachment under flow. Our findings also demonstrate that under flow, functional TFP play an important role in biofilm growth by A. citrulli. In contrast, under the conditions tested, polar flagella appear to be less important for adhesion and biofilm formation of A. citrulli. This statement is supported by the fact that the flagellin mutant M6-flg and wild-type W1 (both lacking flagella) were able to attach to the surface and form a biofilm in a manner similar to that of M6.

TFP are well-established virulence determinants of animal pathogenic bacteria, and were recently shown to contribute to the virulence of several phytopathogenic bacteria, including Ralstonia solanacearum, Xanthomonas oryzae pv. oryzicola, Xylella fastidiosa and A. citrulli (Kang et al., 2002; Meng et al., 2005; Wang et al., 2007; Bahar et al., 2009). While the contribution of TFP to the virulence of animal pathogens has been investigated, the mechanisms by which TFP contribute to the virulence of phytopathogenic bacteria are poorly understood. The findings from this study may provide a possible explanation for the reduced virulence of A. citrulli TFP mutants (Bahar et al., 2009). It is well known that xylem sap in plant vessels does not flow at a constant rate, and at nights, may even be reduced to a minimum. However, under average rates, sap flow may minimize cell adhesion and subsequent biofilm formation on xylem walls, thus affecting virulence, particularly in the case of TFP mutants.

Biofilms are thought to contribute to the virulence of phytopathogenic bacteria through several mechanisms, including blockage of xylem sap, increased resistance to plant antimicrobial substances and/or enhanced colonization of specific niches (Danhorn & Fuqua, 2007). Nevertheless, the picture can often be more complex than expected. For instance, Guilhabert & Kirkpatrick (2005) showed that a hemagglutinin mutant of X. fastidiosa, which is impaired in cell aggregation and biofilm maturation, was hypervirulent on grapevines. The authors hypothesized that the formation of an immature monolayered-biofilm structure by this mutant was sufficient to induce severe disease symptoms, while the lack of cell aggregation promoted a faster distribution of the pathogen in the plant, yielding a phenotype more severe than that of the wild type. In A. citrulli, the hyperpiliated M6-T mutant was shown to form cell aggregates in MFC to a much greater extent than wild-type M6. Interestingly, previously reported virulence assays revealed that not only is the M6-T mutant less virulent than the wild type, it is also less virulent than the TFP-null mutant M6-M (Bahar et al., 2009), suggesting that cell aggregation could negatively affect virulence, probably by hampering the distribution of the pathogen inside the plant.

In addition to the effect of TFP on virulence through biofilm formation, TFP-mediated twitching may also contribute to bacterial spread along the plant, especially against the flow direction, as observed here and in studies with X. fastidiosa (Meng et al., 2005). Indeed, stem inoculation experiments demonstrated that both A. citrulli and X. fastidiosa possess the ability to spread against the sap flow in xylem vessels (Meng et al., 2005; Bahar et al., 2009). In our study, the twitching speed of A. citrulli was approximately 9.9 ± 1.1 μm min−1. Similar twitching assays showed that wild-type cells of X. fastidiosa moved at 0.86 ± 0.04 μm min−1; however, an X. fastidiosa mutant lacking type I pili (which slows down twitching) moved at 4.85 ± 0.27 μm min−1 (De La Fuente et al., 2007a). Thus, the twitching speed of A. citrulli is roughly comparable to that of the X. fastidiosa mutant lacking type I pili. This is not surprising as, according to the annotation of the sequenced AAC00-1 strain, A. citrulli lacks type I pili.

Our findings, however, do not explain the impaired virulence of strains W1 and M6-flg. Although these strains lack polar flagella, they do possess adhesion and biofilm formation abilities similar to those of strain M6 in the MFCs. It is possible that, in contrast to our observations in the present studies, polar flagellum does play a role in attachment to, colonization of and biofilm formation on xylem vessels. Moreover, the role of polar flagella in virulence may not be limited to these features. We speculate that under conditions of minimal xylem sap flow, swimming contributes to long spread of the pathogen thorough the xylem, thus allowing further colonization of parts distant from the infection site.

An obvious limitation of MFCs is that they mimic the xylem vessels only to a certain extent: not only are the surface and the medium different, the chambers lack the complex dynamics of a plant–pathogen interaction system. Nevertheless, this technology provided powerful insights into several behaviors of A. citrulli under flow conditions and raised new questions that can now be addressed and examined in a full-biological system, using the host plant and suitable experiments.

Supporting Information

Additional supporting information may be found in the online version of this article:

Movie S1. Adhesion assay with increasing flow rate with strains M6 (upper channel) and M6-T (lower channel).

Movie S2. Biofilm formation of wild-type strain W1.

Please note: Wiley-Blackwell is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.


We thank Ms Jennifer Parker and Dr Yael Helman for critically reading the manuscript. The research of Ofir Bahar at Auburn University was supported by a graduate student fellowship from the United States–Israel Binational Agricultural Research and Development (BARD) Fund.


  • Editor: Stephen Smith


View Abstract