Insertion and correct polymerization of the membrane attack complex (MAC) into the outer membrane is paramount for killing of Gram-negatives such as Klebsiella pneumoniae by the complement system 6. Although the link between expression of LPS O-antigen and resistance to MAC-mediated killing has been established for Klebsiella 25,29,47, we here show that specifically the LPS O1-antigen of Klebsiella prevents polymerization of C9 into functional MAC, thereby preventing killing.
Our findings confirm previous data showing that Klebsiella pneumoniae expressing an O1-antigen are more resistant to MAC-mediated killing than those expressing an O2a-antigen. Using site-directed mutagenesis, we could pinpoint MAC resistance to not only the presence of the O1-antigen, but we could also show that specifically the O1-cap (D-galactan II repeating units) is required to create a MAC-resistant phenotype.
Our results confirm the previously reported discrepancy in serum-resistance between the O1 and O2a serotypes, namely that O1-antigen strains are more resistant to MAC-mediated killing than O2-antigen strains 20,25,36. While the O1-antigen and its structure have been known for more than thirty years 48, the two genes responsible for the synthesis of D-galactan II, wbbY and wbbZ, respectively, have been only recently characterized 38,39. While wbbY has been shown to be an essential glycosyltransferase for the synthesis of D-galactan II, the role of wbbZ, a pyruvyltransferase, remains understudied. Additionally, these two genes are located between transposable elements and might have gotten acquired via lateral gene transfer 35.
Despite their susceptibility to serum, strains expressing an O2-antigen are commonly found among Klebsiella serotypes causing infections in the blood stream 24. This could be due to the fact that the O2-antigen seems to be less likely recognized by antibodies due to shielding by capsule and therefore more resistant against recognition by antibodies and opsonophagocytic uptake 24,35.
Next to showing the role of O-antigen as determinant of MAC-resistance, we also show that deletion of capsule (in strain KpO1_1) has no influence on the bacterial capacity to survive exposure to MAC, emphasizing the role of O-antigen as determinant of MAC-resistance. The contribution of capsule to serum resistance is still a matter of discussion. While we can confirm earlier findings that capsule is not a determinant of serum resistance 19, the contrary has been reported as well, as Álvarez et al. have shown that in absence of O-antigen, capsule becomes the main determinant of serum resistance 40,49. Some reports suggest that the bacterial production of capsule can increase upon external stress factors 50. However, in general, resistance is mostly linked to hyper-encapsulation, especially with K-types K1 or K2 51,52. It has been speculated that O-antigen might influence retention of capsule polysaccharides 53, but the role of capsule is thought to be of bigger importance eliciting resistance to opsonophagocytic killing 41.
Our mechanistic studies show that expression of O1-antigen does not block early steps of complement activation. Instead, expression of the O-antigen leads to increased deposition of C3b and C5 conversion and therefore high levels of bacterium-bound C5b6 and even C5b-9 complexes. However, deposited C5b-9 complexes are not functional on O1-antigen strains, because the complexes are unable to correctly insert into the bacterial outer membrane and polymerize into a functional pore. Our finding that deposition of C3b was not impaired by the presence of an O-antigen, is in disagreement with previous reports, where serum-resistant K. pneumoniae LPS smooth strains (O1-strains) deposited less C3 than their serum-sensitive LPS rough (O-antigen negative strains) 47. We can, however, confirm earlier findings that the O1-antigen does not lead to evasion of complement activation 19. Surprisingly, in presence of O1-antigen, we observed higher levels of C3b deposition, which are also reflected in the higher amount of C5 converted to C5a and C5b. Presence of the shorter O2a-antigen (O1-cap deletion) still leads to high C5 conversion, while deletion of the entire O1-antigen leads to drastically lower C5 conversion rates. This is partly in line with previous findings, that serum-resistant Gram-negatives which express an LPS O-antigen can potently convert C5 21. However, it seems the shorter O2-antigen is enough to trigger the increased C5 conversion, even though it does not lead to MAC-resistance. The strong release of pro-inflammatory C5a in response to LPS constitutes a great risk factor for an anaphylactic shock, as has been shown in mice 54.
The potent C5 conversion further correlated with deposition of C9, the terminal protein of the MAC. Surprisingly, the amount of detectable C9 on the bacterial surface did not correlate with killing. We could show that expression of the O1-antigen prevents polymerization of C9 into a bactericidal pore. We can link this mechanism not only to MAC-resistant strains, but to the presence of the O1-antigen specifically. In presence of the full O1-antigen, detectable levels of C9 are highest, while C9 polymerization and insertion into the bacterial membrane are impaired. As a consequence, sMAC is released. While the shorter O2a-antigen might not be sufficient to prevent MAC-mediated killing, it nevertheless decreases the efficiency of how the MAC can form on the bacterial surface, whereas the O1-antigen prevents MAC-formation altogether. The underlying mechanism might be explained by the excessive consumption of complement components by C5 convertases resulting in MAC pre-pores. We speculate that the LPS O1-antigen, by activating more C5 convertases, changes the local ratio of C5 to C9 to conditions where only a suboptimal amount of C9 molecules are available per C5b6 pre-pore complex 21. Unable to correctly anchor, insert into the membrane and to further polymerize, these MAC pre-pores would get released into the bacterial supernatant as soluble MAC. Doorduijn et al. have reported a similar finding in E. coli, where presence of an O-antigen prevented polymerization of C9 and lead to increased C5 conversion and sMAC generation 21,22.
In the case of a shorter O2-antigen, convertase activity will still be increased and lead to partial release of MAC precursors as sMAC. However, the O2a-antigen gives less protection against membrane insertion of the MAC, and some pores form correctly and penetrate the outer membrane, thereby killing the bacteria. These findings highlight the importance of the O1-antigen for MAC-resistance in K. pneumoniae. It might also serve as a possible explanation why O1-bearing K. pneumoniae are the most common serotype of clinical relevance 35. Potentially, these data suggest a multi-faceted role of the O1-antigen during an infection. Not only does the O1-antigen render K. pneumoniae non-susceptible to MAC-killing, but the increased release of anaphylatoxin C5a also pose a high inflammatory burden during an ongoing infection. The release of soluble MAC components in both cases might serve as a biomarker of ongoing infections with K. pneumoniae.
In light of the rise of multidrug resistant hypervirulent Klebsiella strains 55, understanding how K. pneumoniae resists exposure to the complement system and killing via the MAC is of increasing importance. This study gives molecular insights into how expression of the LPS O-antigen helps K. pneumoniae withstand direct killing through the MAC.