Intranasal and intratracheal challenge of mice with disrupted mature biofilms allowed us to directly test whether pneumococci within biofilms were in a virulent state and modeled the aspiration of bacterial aggregates that might be present in mucosal secretions from the nasopharynx or that which might be introduced during intubation 
. Our results indicate that pneumococci within biofilms are highly suited for attachment to mucosal epithelial cells, but as a result are avirulent. This was unexpected as the formation of biofilms has been suggested to be a pivotal event to numerous infectious diseases 
. One important limitation of this study is that this approach does not examine the pathogenic potential of biofilms that form in vivo
and de novo
, such as within the nasopharynx during normal colonization. Thus there is the possibility that in vivo
biofilms might act differently.
Based on our experimental results, the hyper-adhesive phenotype of biofilm pneumococci could be attributed to: i) reduced capsule which exposes bacterial surface proteins 
, ii) selection for the transparent phenotype which carries greater amounts of ChoP that binds to the host-protein PAFr 
, iii) enhanced production of CbpA which binds to Laminin receptor 
, as well as PsrP, which binds to Keratin 10 
. Unconfirmed by protein analysis, but possibly also contributing to the hyper-adhesive phenotype, we observed increased expression of the gene encoding Neuraminidase A by qRT-PCR, which has been shown to enhance bacterial adhesion by cleaving sialic acid moieties and thereby exposing cryptic ligands on the host-cell surface 
. The attenuated phenotype of biofilm pneumococci could be attributed to: i) a reduced metabolic rate that would delay its ability to respond to stressors in a novel host-environment, ii) enhanced ChoP, which would enhance opsonization by C-reactive protein 
, iii) reduced capsule, which would also facilitate phagocytosis 
, iv) a reduction in pneumolysin production 
, v) reduced PcpA and possibly type I pilus 
. While a reduction in capsule and pneumolysin expression along with enhanced neuraminidase has been shown for S. pneumoniae
, ours is the first study to suggest they act collectively to dramatically impact the ability of biofilm pneumococci to progress from the nasopharynx and cause invasive disease; in particular bloodstream infections. Thus, implying that pneumococci within biofilms do not directly contribute to the development of invasive disease.
Using TEM to examine pneumococci within a mature biofilm structure we were surprised to determine that only a small percentage of the mature biofilm was composed of electron dense and presumably viable pneumococci. A finding that suggests robust pneumococcal biofilm formation occurred through the accumulation of dead pneumococci. Most recently, Trappetti et al.
have shown that it is the opaque phase variant of S. pneumoniae
that is responsible for formation of the EPM and not the transparent. As our biofilms contained both transparent and opaque S. pneumoniae
, the opaque variant most likely accounts for the EPM we detected by electron microscopy. Furthermore, and in contrast to our in vivo
findings, Trappetti et al
. observed that opaque but not transparent biofilm-derived pneumococci, were able to translocate from the nasopharynx to the lungs and brain of mice 
. One possible explanation for this discrepancy in results is that the transparent pneumococci present in our biofilms facilitated the opsonophagocytosis of the attached opaque bacteria. This would suggest that naturally occurring mixed biofilms are avirulent. Alternatively, is our use of a continuous flow reactor for mature biofilm development; Trappetti et al
. used a static biofilm model. In separate studies both Trappetti et al.
and ourselves found that that the use of different biofilm models resulted in variable phenotypes 
. Finally, is our use of the TIGR4 strain of S. pneumoniae
whereas Trappetti et al.
used a 19F clinical isolate.
Despite our observation of considerable EPM surrounding the electron dense and presumably viable bacteria, biofilm pneumococci were determined to be hyper-adhesive, suggesting that in addition to a loss in capsule and increased ChoP, CbpA, and PsrP protein levels by pneumococci, the EPM may also contain adhesive elements. This possibility is also supported by findings by Trappetti et al.
, which determined that opaque sessile (i.e biofilm) pneumococci adhere to A549 and Detroit 562 cells better than transparent sessile pneumococci. The latter was unexpected as the transparent phenotype is associated with increased expression of CbpA and ChoP and transparent planktonic pneumococci have been shown to adhere to cells in an enhanced manner 
. It is for this latter reason that we believe mice challenged with biofilm-derived planktonic pneumococci, which would be mostly transparent, developed pneumonia but were unable to cause bloodstream infection. Importantly, our observation of enhanced biofilm adhesion by numerous strains and enhanced PsrP, ChoP, and CbpA production indicates that the hyper-adhesive phenotype is a pan-pneumococcal biofilm property that is multi-factorial, involving numerous components along with the production of EPM.
In an effort to develop a working model that coalesces the published data with our own, we propose that the selective death of opaque pneumococci might be occurring during biofilm formation. Opaque cell death would provide a mechanism for the release of DNA and other components that are known make up the EPM. It would also provide an explanation for the high numbers of dead bacteria observed in our biofilms as well as our recovery of predominantly transparent pneumococci from mature biofilms. In context of in vivo
transmission our model implies that the opaque variant would be responsible for formation of the EPM in the nasopharynx and thereby confer in vivo
persistence, whereas the transparent variant, which is better able to attach to cells and colonize naïve animals 
, would remain available in greater numbers for spread to the next host and gain from the enhanced adhesive capacity of the surrounding EPM. In the next host, partial reversion to the opaque variant would t also be necessary to reform a biofilm. Thus, studies are warranted to ascertain if differential cell death dependent on phase-variation occurs within biofilms and to test its impact on transmission.
The observed tolerance of biofilm pneumococci to antimicrobials was in agreement with previously published studies 
, moreover, was indicative that we were in fact examining mature biofilms. Importantly, biofilm pneumococci remained susceptible to cell wall acting antimicrobials suggesting that maintenance of the cell wall remained a critical function during the quiescent state. Ours is the most comprehensive analysis of pneumococcal gene expression during biofilm growth to date. However, our gene expression data reflects the biases of our biofilm model which includes biofilm-related changes in the ratio of opaque and transparent pneumococci as well as increasing amounts of remnant mRNA from dead bacteria. This most likely explains why our microarray results do not exactly match previous studies that explore differences between opaque and transparent pneumococci 
We determined that biofilm bacteria down-regulated >50 genes involved in protein translation, the ATP synthase machinery, fatty acid metabolism, phospholipid synthesis, and replication. A strong reduction in capsule operon cassette expression was observed consistent with a previous study by Moscoso et al.
, which showed that cps3A
, the first gene in the capsule operon cassette was down-regulated during biofilm production 
. Previously, for serotype 3 strains, a non-phase variable deletion within the capsule operon cassette resulting in a rough mutant has been shown to occur and contribute towards biofilm formation 
. A reduction in capsule would serve to expose surface components such as adhesins and facilitate attachment. This notion is supported by our previous findings with PsrP, where a version of the protein unable to extend past the capsule layer failed to mediate adhesion, as well studies as completed by Munoz-Elias et al.
, that found use of an unencapsulated strain facilitated the identification of genes involved in biofilm formation in vitro 
. Concomitantly, a reduction in capsule would reduce the virulence potential of individual pneumococci.
As indicated, the observed reduction in the physiological state of bacteria may also contribute to their attenuated phenotype. Metabolically inert bacteria would take longer to adapt to hostile host environments such as the lower respiratory tract and produce the necessary determinants required for survival such as pneumolysin. Along this line, our observation that biofilm pneumococci down-regulate pneumolysin allows for speculation that biofilm pneumococci stop producing factors that elicit a strong inflammatory response during biofilm formation within the nasopharynx. Presumably, this would promote long-term colonization by modulating the immune response. This concept is supported by the finding that invasive serotypes of S. pneumoniae
colonize the nasopharynx for a shorter duration than non-invasive serotypes 
. The reduction in pneumolysin and PcpA levels during biofilm growth also suggests that immunization with pneumolysin or PcpA would also have a modest effect against colonization but might still protect against disseminated (i.e. planktonic) disease. In contrast, antibodies against CbpA and PsrP, which are up-regulated during biofilm growth, might deter nasopharyngeal colonization and thereby promote species replacement, such as with Staphylococcus aureus
, in immunized individuals. As such the differential production of protein vaccine candidates during biofilm versus planktonic growth should be an important consideration in the design of any future protein vaccine against S. pneumoniae
or other bacterial pathogens 
. Of note this concept is consistent with findings by Oggioni et al.,
showing altered pneumococcal virulence gene expression occurred during sessile bacterial growth on fixed surfaces versus planktonic 
Finally, the enhanced expression of psrP
and its accessory proteins was in agreement with our previous studies that showed PsrP contributes to robust biofilm formation 
. While unconfirmed microarray and qRT-PCR data implies that the type I pilus might be down regulated during biofilm production; this would be surprising as the pilus of Group A Streptococci and Group B Streptococci have been shown to play an important role in biofilm formation 
. Of note, microarrays did not reveal enhanced cbpA
expression, however, it was determined by qRT-PCR and immunoblot that CbpA levels were increased. This emphasizes the necessity for validation of RNA data with protein studies and raises the possibility that other determinants are also altered. Thus a proteomics approach is warranted to address this gap. Of note, our microarray findings are in stark contrast to those by Allegrucci et al.
, who found a dramatic increase in the number of detectable biofilm proteins when examining 2-dimensional gels of a serotype 3 isolate 
. A possible explanation for this discrepancy is the accumulation of dead bacteria and their proteins in a mature biofilm. This would be an inseparable and confounding factor in any proteomic analysis of mature biofilms.
In summary, we observed a dramatic enhancement in the ability of biofilm pneumococci to attach to host cells as well as a dramatic reduction in their ability to cause invasive disease. Notably, biofilm pneumococci colonized the nasopharynx normally. As the vast majority of S. pneumoniae do not cause invasive disease, it is most likely that these biofilm related changes occur so as to facilitate long-term colonization of the nasopharynx rather than promote development of invasive disease. Therefore, and based on the available information, we suggest that the ability to form robust biofilms is not required for virulence, but instead contributes towards long-term colonization and transmission of the pneumococcus.