UspA1, UspA2, and UspA2H have been classified as members of the YadA family of nonfimbrial, nonpilus adhesins, which reach the bacterial surface by means of an autotransporter mechanism (9
). The UspA2H protein has been shown to function in attachment to Chang conjunctival epithelial cells in vitro (22
) and can also confer serum resistance on some strains of M. catarrhalis
). In the present study, it was shown that the UspA2H protein plays a role in biofilm formation by M. catarrhalis
strain ETSU-9 in vitro, as evaluated with the crystal violet-based assay. Studies of the Yersinia
YadA protein, the prototype of this autotransporter family, have indicated that this nonfimbrial adhesin appears to form a very dense layer on the surfaces of Yersinia
species and possesses an architecture that involves a “head-stalk-anchor” structure, with the “head” component binding various eukaryotic proteins (reviewed in reference 21
). Both the UspA1 and UspA2 proteins of M. catarrhalis
O35E were also shown to form structures with a similar “lollipop” appearance (19
), with the UspA2 protein being much more abundant (41
). UspA2H proteins are hybrid macromolecules, involving a leader peptide and an N-terminal region that resemble those of UspA1 proteins, together with a C-terminal half that most closely resembles that of a UspA2 protein (22
Mutant analysis revealed that lack of expression of the UspA1 protein by M. catarrhalis ETSU-9 had a relatively modest effect on the ability of the strain to form a biofilm in the crystal violet-based assay system (Fig. ). In contrast, a uspA2H mutant of ETSU-9 had a much reduced ability to form a biofilm in this system relative to both the uspA1 mutant and the ETSU-9 parent strain (Fig. ). This result is perhaps not surprising in view of the fact that the relative level of UspA2H expression by ETSU-9 is apparently much greater than that of UspA1 (Fig. ). What was unexpected was that the introduction of a 15-nt insertion in frame into numerous different sites within the uspA2H gene of ETSU-9 would have such a negative effect on the ability of the UspA2H protein to mediate biofilm formation by this strain (Fig. ). This insertion mutagenesis technique was intended to identify domains within the UspA2H protein that were directly (or indirectly) involved in the ability of the strain to form biofilms. Instead, it appeared that insertion of this 5-aa sequence at any one of many different sites in the UspA2H protein resulted in loss of function, at least with regard to biofilm formation in the crystal violet-based assay. Interestingly, nearly all of these mutant UspA2H proteins were still able to mediate attachment of the organism to Chang conjunctival epithelial cells (Fig. ), indicating that these 5-aa insertions were likely not causing a gross conformational change in the UspA2H protein or at least did not occur in the region(s) essential for bacterial attachment. These results also suggest that UspA2H-dependent epithelial cell attachment and biofilm formation mediated by the M. catarrhalis ETSU-9 UspA2H protein are separate and distinguishable activities.
Even though a number of different 5-aa insertions adversely affected biofilm formation ability, a mutant (ETSU-9.mut92) in which 418 aa was deleted from the UspA2H protein still formed a biofilm, albeit at a slightly reduced level relative to its parent strain (Fig. ). The same deletion mutant also attached at wild-type levels to Chang cells in vitro (Fig. ). These data indicate that the portion of the UspA2H protein deleted in this particular mutant is not essential for biofilm development or attachment to Chang cells. This 418-aa section is located within what would be predicted to be the “stalk” region of the “lollipop” structure proposed for UspA1 and UspA2 proteins by Hoiczyk and colleagues (19
). It is possible that this internal deletion resulted in the formation of a shorter “stalk” that still allowed proper orientation of the “head” region predicted to be involved in attachment of UspA1 to cell membranes (19
). This internal deletion does not include the predicted autotransporter domain (21
), which is consistent with the functional ability of this UspA2H mutant protein, which is apparently present on the surface of the outer membrane. This internal deletion does include the region which binds MAb 17C7 (2
) and therefore may also include a region with homology to the closely related UspA1 protein, which was recently shown to bind CEACAM (17
) and fibronectin (47
The “lollipop” model proposed by Hoiczyk et al. (19
) might explain why biofilm formation by ETSU-9 is so sensitive to perturbations in UspA2H. In this model, the N-terminal “head” region would be the domain directly involved in biofilm formation, so insertions in this region would prevent biofilms. The “stalk,” or coiled-coil, region of UspA2H is composed of six repeating amino acid sequences and would be expected to present the “head” domain in the correct orientation. Deletion of some or all of the repeating units of the “stalk” (mutants 81 and 92) resulted in only minor reductions in biofilm formation. However, interruption of the repeating units, either by deletion (mutant 103) or by insertion, could result in a kink in the “stalk” so that the “head” domain is no longer in the correct orientation to mediate biofilm formation.
Mutations in three other genes in addition to uspA2H
were shown to adversely affect the biofilm formation ability of M. catarrhalis
ETSU-9 in the crystal violet-based assay. These genes encoded a predicted aminoglycoside phosphotransferase, a predicted lytic murein transglycosylase D, and a predicted N
-alanine amidase (AmpD). The last two gene products likely have roles in bacterial cell wall metabolism in M. catarrhalis
, and mutations in ampD
may have an indirect effect on biofilm formation. More specifically, a transposon insertion in the ampD
gene of Pseudomonas aeruginosa
resulted in high-level expression of a chromosomally encoded β-lactamase (6
), and expression of some β-lactamases has been reported to inhibit biofilm formation by both E. coli
and P. aeruginosa
, perhaps by affecting peptidoglycan remodeling (13
). M. catarrhalis
also has a chromosomally encoded β-lactamase (7
), and we found that the level of β-lactamase activity in culture supernatant fluid from the ampD
mutant ETSU-9.988 was greater than that obtained with the parent strain (data not shown). As for the lytic murein transglycosylase, it is known that E. coli
recycles almost 50% of its murein in each generation (14
). Although the role of murein recycling has not yet been fully elucidated, Park (39
) has suggested that this function may be one method for sensing the condition of the cell wall. It is possible that a disruption of the normal murein-cycling pathway could have an adverse affect on biofilm formation.
One unexpected finding from this study was that the ability of the ETSU-9.222cat, ETSU-9.714cat, and ETSU-9.988cat mutants (described immediately above) to form biofilms in the crystal violet-based assay was affected by the type of inoculum (i.e., broth-grown cells versus agar-grown cells). Wells inoculated with mutants that had been grown on agar plates had more biofilm development than did wells that were inoculated with broth-grown cells (Fig. ). It has been suggested that inoculating these wells in the tissue culture plate with a cluster of cells scraped from the surface of an agar plate may be akin to seeding the wells with a preformed biofilm (J. W. Costerton, personal communication); this could account for these inoculum-dependent results. It can also be inferred from these data that future screens for biofilm-negative M. catarrhalis mutants should be carried out using broth-grown inocula to minimize the probability of missing mutants attenuated for biofilm development in this in vitro system.
It must also be noted that the crystal violet-based biofilm assay used in the present study, while very useful for efficiently screening large numbers of mutants, does have some limitations. First, many workers consider this assay appropriate for measuring early events in biofilm development on abiotic surfaces (37
). Second, data obtained on the biofilm formation abilities of wild-type strains and mutants in the crystal violet-based assay may be different from or similar to results obtained with the same strains in continuous-flow biofilm systems (11
). Third, there are obvious issues involving depletion of nutrients and potential lack of aeration in the crystal violet-based assay system (30
). In the present study, the biofilm-deficient M. catarrhalis
mutants were not tested in alternative biofilm systems.
Based on the data presented in this report, the UspA2H protein of M. catarrhalis ETSU-9 is necessary for biofilm formation by this strain in the crystal violet-based assay system. Moreover, the gain-of-function experiments involving recombinant E. coli expressing UspA2H reinforce the functional role of this M. catarrhalis gene product in biofilm development in this particular assay system. However, the existence of the other three biofilm-deficient mutants described above (i.e., ETSU-9.222, ETSU-9.714, and ETSU-9.988) indicate that UspA2H expression by itself is not sufficient for biofilm formation by ETSU-9 in this model system and that other gene products, including some that affect different aspects of cell wall synthesis, are also important for normal biofilm development in the system.