Polymerization of C9 enhances bacterial cell envelope damage and killing by membrane attack complex pores.

Complement proteins can form membrane attack complex (MAC) pores that directly kill Gram-negative bacteria. MAC pores assemble by stepwise binding of C5b, C6, C7, C8 and finally C9, which can polymerize into a transmembrane ring of up to 18 C9 monomers. It is still unclear if the assembly of a polymeric-C9 ring is necessary to sufficiently damage the bacterial cell envelope to kill bacteria. In this paper, polymerization of C9 was prevented without affecting binding of C9 to C5b-8, by locking the first transmembrane helix domain of C9. Using this system, we show that polymerization of C9 strongly enhanced damage to both the bacterial outer and inner membrane, resulting in more rapid killing of several Escherichia coli and Klebsiella strains in serum. By comparing binding of wildtype and 'locked' C9 by flow cytometry, we also show that polymerization of C9 is impaired when the amount of available C9 per C5b-8 is limited. This suggests that an excess of C9 is required to efficiently form polymeric-C9. Finally, we show that polymerization of C9 was impaired on complement-resistant E. coli strains that survive killing by MAC pores. This suggests that these bacteria can specifically block polymerization of C9. All tested complement-resistant E. coli expressed LPS O-antigen (O-Ag), compared to only one out of four complement-sensitive E. coli. By restoring O-Ag expression in an O-Ag negative strain, we show that the O-Ag impairs polymerization of C9 and results in complement-resistance. Altogether, these insights are important to understand how MAC pores kill bacteria and how bacterial pathogens can resist MAC-dependent killing.

Introduction

Complement proteins in human serum play a crucial role in fighting off invading bacteria. Activation of the complement cascade ultimately results in the assembly of membrane attack complex (MAC) pores that can directly kill Gram-negative bacteria [1-3]. MAC assembly is initiated when recognition molecules, such as antibodies and lectins, bind to bacteria and recruit early complement proteins [4]. This triggers a proteolytic cascade that deposits convertase enzymes on the bacterial surface [5]. These convertases convert complement component C5 into C5b, which initiates the assembly of a large ring-shaped MAC pore that damages the bacterial cell envelope [3,6,7]. Although MAC pores can efficiently kill complement-sensitive bacteria, some bacterial pathogens can survive killing by MAC pores [8-11]. Therefore, studying how MAC pores kill bacteria is important to understand how the complement system prevents infections and how bacterial pathogens resist killing by MAC pores.

MAC pores assemble in a stepwise manner [12]. When a surface-bound convertase converts C5 into C5b, C5b immediately binds to C6 to form the C5b6 complex [13,14]. Direct binding of C7 to C5b6 anchors the MAC precursor to the membrane [15]. Next, C8 binds to membrane-anchored C5b-7, which triggers structural rearrangements in C8 that result in insertion of a transmembrane β-hairpin into the membrane [16,17]. Finally, C9 binds to C5b-8 to form C5b-91, after which C9 self-polymerizes to form a transmembrane ring consisting of C5b-8 and up to 18 monomers of C9 with an inner diameter of 11 nm [18,19].

Although C5b-8 can already cause small 1–2 nm lesions in the membrane of erythrocytes and liposomes without polymeric-C9 [17,20], it is still unclear if C9 polymerization is required to sufficiently damage the complex bacterial cell envelope to kill Gram-negative bacteria. On bacteria, MAC pores assemble on the outer membrane (OM), which largely consists of lipopolysaccharide (LPS). The O-antigen (O-Ag) of LPS can vary in length between bacterial strains and species [21,22], and this has frequently been associated with complement-resistance [8,9,23,24]. Apart from the OM, the Gram-negative cell envelope also consists of a cytosolic inner membrane (IM) and a periplasmic peptidoglycan layer [25]. We recently developed methods to separately study OM and IM damage in time [14,26,27], which highlighted that C9 is required to efficiently trigger damage to the IM and to allow passage of antimicrobial enzymes in serum through the OM. Although the presence of polymeric-C9 has been associated with cell envelope damage [28], direct evidence to what extent polymerization of C9 contributes to bacterial cell envelope damage by MAC pores is limited. In addition, a direct causal link between bacterial killing and polymerization of C9 has still not been established, mainly because tools to specifically prevent polymerization of C9 have been limited [29]. Recently, Spicer et al. demonstrated that ‘locking’ the first transmembrane helix (TMH-1) domain of C9 could prevent polymerization of C9 [30]. Here, we used this ‘locked’ C9 to study how C9 polymerization contributes to bacterial cell envelope damage and killing by MAC pores.

In this paper, we show that polymerization of C9 enhanced the efficiency by which MAC pores damage both the OM and IM, which ultimately resulted in faster killing of several Escherichia coli and Klebsiella strains. This study therefore highlights that polymerization of C9 is required to form MAC pores that efficiently damage the bacterial cell envelope and kill bacteria. Moreover, we found that polymerization of C9, but not binding of C9 to C5b-8, was impaired on several complement-resistant E. coli strains that survive killing by MAC pores. Finally, we show that the expression of LPS O-Ag impairs polymerization of C9 and results in resistance to MAC-dependent killing. This study therefore also provides insights into how bacterial pathogens resist MAC-dependent killing.

Results

Locking the TMH-1 domain of C9 strongly impairs its capacity to polymerize, without preventing binding to C5b-8 on E. coli

To study the contribution of C9 polymerization to bacterial killing by the MAC, we wanted to use a system in which C9 can bind to C5b-8, but cannot polymerize. Spicer et al. recently demonstrated that the TMH-1 domain of C9 has a crucial role in C9 polymerization [30]. When C9 binds to C5b-8, structural rearrangements in C9 trigger unfurling of the TMH-1 () and TMH-2 to form a transmembrane β-hairpin. Although both TMH domains of C9 insert into the membrane, unfurling of the TMH-1 domain also exposes an elongation surface that allows a subsequent C9 to bind (). This ultimately results in the assembly of a polymeric-C9 ring that forms the MAC pore together with C5b-8 (. Based on this crucial role of the TMH-1 domain in polymerization, Spicer et al. designed a C9 TMH-1 ‘lock’ mutant (C9TMH-1 lock) in which the TMH-1 domain was linked to β-strand 4 of the MACPF/CDC domain via an intramolecular cysteine bridge. This lock prevents unfurling of the TMH-1 domain, and thus prevents both the formation of a transmembrane β-hairpin () and binding of a subsequent C9 (). Reducing the cysteine bridge with DTT can unlock the TMH-1 domain and restore its capacity to polymerize ().

C9TMH-1 lock was recombinantly expressed and site-specifically labelled with a fluorophore via sortagging, as was done previously for wildtype C9 (C9wt) [14]. Fluorescent labelling was comparable between both C9TMH-1 lock and C9wt (), which means that the fluorescence of both proteins can be directly compared in our assays. C9TMH-1 lock showed impaired lysis of sheep erythrocytes compared to C9wt, which could be restored by reducing C9TMH-1 lock with DTT (). Moreover, fluorescent C9TMH-1 lock was used to distinguish SDS-stable polymeric-C9 from monomeric-C9 by SDS-PAGE, which is frequently used as a read-out for C9 polymerization [31]. C9TMH-1 lock did not form polymeric-C9 together with preassembled C5b6 (pC5b6), C7 and C8, whereas C9wt did (). These data confirm that the capacity of C9TMH-1 lock to form polymers is strongly impaired, although it has to be noted that due to the detection limit of polymeric-C9 by SDS-PAGE we cannot state that polymerization of C9 is completely prevented.

Although Spicer et al. already demonstrated that locking the TMH-1 domain prevents polymerization of C9 [30], we wanted to confirm that this did not prevent binding of C9 to C5b-8. E. coli MG1655 bacteria were incubated in C8-depleted serum to activate complement and label them with MAC precursor C5b-7 (). C5b-7 labelled bacteria were washed to remove remaining serum components and incubated with C8 and C9wt or C9TMH-1 lock to further assemble the MAC. Both C9wt and C9TMH-1 lock bound to C5b-7 in a C8-dependent manner as measured by flow cytometry (), but C9wt binding was 10-fold higher than C9TMH-1 lock. Since the amount of C5b-8 on the surface, as measured by C6-FITC binding, was comparable for both C9wt and C9TMH-1 lock (), the relative difference in C9 binding suggested a difference in polymerization of C9 (). SDS-PAGE confirmed that only monomeric-C9 was detected on bacteria incubated with C9TMH-1 lock (). Reducing C9TMH-1 lock with DTT increased C9 binding 3-fold compared to C9TMH-1 lock without DTT () and resulted in the detection of polymeric-C9 on bacteria by SDS-PAGE (). Both C9 binding and polymeric-C9 detection were approximately 2- to 3-fold lower compared to C9wt with DTT (), suggesting that reducing C9TMH-1 lock only partially restored its capacity to polymerize compared to C9wt. Finally, an antibody that recognizes a neo-epitope exposed in polymeric-C9, which is frequently used for the detection of MAC pores [32], specifically detected bacteria incubated with C9wt and C9TMH-1 lock with DTT, but not with C9TMH-1 lock without DTT (). In summary, these data indicate that C9TMH-1 lock can bind to C5b-8 on E. coli, but that its capacity to polymerize is strongly impaired.

Fig 1

C9TMH-1 lock binds to C5b-8, but its capacity to form polymers is impaired on E. coli.

a) Schematic overview how C9 polymerizes. Binding of C9 (blue) to C5b-8 (grey) triggers unfurling of the TMH-1 domain of C9 (green) and subsequent insertion into the membrane (I). Unfurling of TMH-1 exposes the elongation surface of C9 to bind a subsequent C9 monomer (II). Ultimately, this results in the formation of a C9 polymer (III). Locking the TMH-1 domain of C9 (orange, C9TMH-1 lock) with an intramolecular cysteine bridge (red) prevents unfurling of the TMH-1 domain (IV). This also prevents exposure of the elongation surface of C9TMH-1 lock and subsequent polymerization (V). Reducing the cysteine bridge of C9TMH-1 lock with DTT restores its polymerizing capacity (VI). b) Schematic overview how E. coli MG1655 were labelled with MAC. Bacteria were labelled with C5b-7 by incubating them in 10% C8-depleted serum for 30 minutes. Bacteria are washed and next incubated with 10 nM C8 for 15 minutes. Finally, 20 nM of Cy5-labelled C9wt or C9TMH-1 lock is added in the presence or absence of 10 mM DTT for 30 minutes. c) Binding of Cy5-labelled C9wt or C9TMH-1 lock to bacteria treated as described in b and measured by flow cytometry. d) Binding of Cy5-labelled C9 in c was divided by the fluorescence of bacteria labelled with C8 + C9TMH-1 lock to calculate the relative binding difference as indication for polymerization of C9. e) Bacterial cell pellets were analyzed by SDS-PAGE for in-gel fluorescence of Cy5-labelled C9wt or C9TMH-1 lock to distinguish monomeric-C9 from polymeric-C9. f) Bacteria were washed after DTT incubation and next stained with AF488-labelled mouse anti-polymeric-C9 antibody (aE11) and staining was measured by flow cytometry. Flow cytometry data are represented by geoMFI values of the bacterial population. The SDS-PAGE image is representative for at least three independent experiments. Data represent individual values of three independent experiments with mean +/- SD. Statistical analysis was done on log-transformed data (10log for c and f, 2log for d) using a paired one-way ANOVA with Tukey’s multiple comparisons’ test. Significance was shown as * p ≤ 0.05, ** p ≤ 0.005, **** p ≤ 0.0001.

Polymerization of C9 enhances bacterial killing by MAC pores

We then assessed if polymerization of C9 is important for bacterial killing by MAC pores. A DNA dye (Sytox) that cannot permeate an intact IM was used to measure the percentage of cells with IM damage by flow cytometry, which we have previously shown to be a sensitive read-out for bacterial killing [14]. Adding C9wt to C5b-8 labelled bacteria resulted in IM damage in a dose-dependent manner, reaching 100% Sytox positive cells from above 3 nM C9 (). For C9TMH-1 lock, IM damage was impaired and did not increase above 30% Sytox positive cells at 100 nM C9 (). Moreover, bacterial viability was determined by counting colony forming units (CFU’s) and decreased only 10-fold for C9TMH-1 lock compared to C5b-8 alone, whereas C9wt decreased bacterial viability at least a 1,000-fold (). Reducing C9TMH-1 lock with DTT restored its capacity to damage the IM () and kill bacteria () relative to C9wt in reducing conditions. However, reducing C9wt or C9TMH-1 lock did diminish IM damage () and bacterial killing () compared to C9wt in non-reducing conditions. Unlocking the TMH-1 domain after C9TMH-1 lock has bound C5b-8, and the remaining soluble C9TMH-1 lock was washed away, did not further increase IM damage ().

Polymerization of C9 enhances bacterial killing by MAC pores.

E. coli MG1655 was labelled with C5b-7 by incubating in 10% C8-depleted serum for 30 minutes. Bacteria were washed and next incubated with 10 nM C8 for 15 minutes. a-b) A concentration range of C9wt or C9TMH-1 lock was added in the absence or presence of 10 mM DTT for 30 minutes. a) Sytox was used to determine the percentage of cells that have a damaged bacterial IM by flow cytometry as read-out for bacterial killing. b) Bacterial viability was determined by counting colony forming units (CFU’s) and calculating the proportion of viable cells for 100 nM C9 condition (a) compared to C5b-7 labelled bacteria in buffer. The horizontal dotted line represents the detection limit of the assay. c-d) C5b-8 labelled bacteria were incubated for 30 minutes with 20 nM C9wt and a concentration range of C9TMH-1 lock. Bacteria were stained with AF488-labelled mouse anti-polymeric-C9 aE11-antibody (c) and Sytox to determine the percentage of cells that has a damaged bacterial IM (d) by flow cytometry. e) Bacterial cell pellets were analyzed by SDS-PAGE for in-gel fluorescence of Cy5-labelled C9wt to distinguish monomeric-C9 from polymeric-C9. From left to right, the molar excess of C9TMH-1 lock from left to right is: 0 (C9wt only), 0.016, 0.05, 0.14, 0.46, 1.1, 3.3 and 10. The SDS-PAGE image is representative for at least three independent experiments. Flow cytometry data are represented by geoMFI values or cell frequencies of the bacterial population. Data represent mean values +/- SD (a, c and d) of three independent experiments or individual values (c) of four independent experiments with mean +/- SD. Statistical analysis was done on 10log-transformed data using a paired one-way ANOVA with Tukey’s multiple comparisons’ test (b). Significance was shown as * p ≤ 0.05, ** p ≤ 0.005.

Polymerization of C9wt () and subsequent IM damage () could be inhibited by C9TMH-1 lock in a dose-dependent manner. Flow cytometry revealed that polymeric-C9 already decreased by 50% when the amount of C9TMH-1 lock was still 10-fold lower than the amount of C9wt (). This suggests that C9TMH-1 lock can interfere at multiple stages in the assembly of a polymeric-C9 ring. However, IM damage was only fully inhibited when there was 10-fold more C9TMH-1 lock than C9wt (). This indicates that little polymeric-C9 is required to damage the IM. This raised the question whether few ring-shaped MAC pores with 18 C9 monomers, or arc-shaped MAC pores with fewer C9 monomers, as observed for other homologous pore forming proteins [33,34], can already damage the IM. SDS-PAGE revealed a similar decrease in polymeric-C9 () as with staining by flow cytometry (). No incomplete C9 polymers were detected by SDS-PAGE, although it is unclear if these potential incomplete C9 polymers are SDS-stable. Altogether our data primarily suggest that polymerization of C9 enhances bacterial killing by MAC pores.

Polymerization of C9 increases OM damage

Since MAC pores assemble on the OM [14], we also wondered how C9 polymerization contributes to OM damage. First, OM damage was compared for C9wt and C9TMH-1 lock by measuring leakage of periplasmic mCherry (perimCherry, 22 kDa) through the OM of MG1655 using flow cytometry [14]. Adding C9wt to C5b-8 labelled bacteria resulted in rapid leakage of perimCherry within 5 minutes (). By contrast, no more perimCherry leaked out with C9TMH-1 lock compared to C5b-8 alone within 60 minutes (), or when the TMH-1 domain was unlocked after C9TMH-1 lock had bound C5b-8 and the remaining C9TMH-1 lock was washed away (). This suggests that polymerization of C9 is required to cause leakage of periplasmic proteins through the OM.

Polymerization of C9 increases OM damage by MAC pores.

E. coli MG1655 (grey) was labelled with C5b-7 by incubating in 10% C8-depleted serum for 30 minutes. Bacteria were washed and next incubated with buffer, 10 nM C8 with 20 nM C9wt or C9TMH-1 lock to measure damage to the bacterial OM. a) periplasmic mCherry (perimCherry, red) leakage was measured at different time points by flow cytometry and represented as relative perimCherry fluorescence compared to t = 0. b) C5b-7 labelled bacteria were incubated with a concentration range of C8 and 20 nM C9wt or C9TMH-1 lock supplemented with 3 μg/ml nisin for 30 minutes. Nisin influx through OM was measured using Sytox to determine the percentage of cells that have a damaged bacterial IM as read-out for bacterial killing by flow cytometry after 30 minutes. c) DiOC2 enters the periplasm when the OM is permeable and shifts from green to red fluorescence in cells with an intact inner membrane (IM) potential. DiOC2 shifts back to green fluorescence when the IM potential is lost. Flow cytometry data are represented by geoMFI values of the bacterial population. Data represent mean +/- SD of three independent experiments (a and b). The multiwell plate-reader assay (c) is shown by one representative experiment that has been repeated at least three times.

We also assessed if C9 polymerization affects the influx of extracellular molecules through the OM. We have previously shown that perturbation of the OM by MAC pores can sensitize Gram-negative bacteria to the antibiotic nisin (3.4 kDa), which normally cannot pass through the OM of Gram-negative bacteria [26]. A 1,000-fold more C5b-8 was required to sensitize bacteria to nisin with C5b-8 alone or C5b-8 + C9TMH-1 lock compared to C5b-8 + C9wt (), suggesting that polymerization of C9 strongly enhanced damage to the OM. Finally, we measured the influx of DiOC2 (0.5 kDa) through the OM in the presence or absence of polymeric-C9. DiOC2 is a green fluorescent dye that shifts to red fluorescence when it is incorporated in membranes with a membrane potential, which in the case of E. coli is the IM. With C9TMH-1 lock, the increase in red:green fluorescence ratio was delayed compared to C9wt, suggesting that polymerization of C9 enhanced the influx of DiOC2 through the OM (). When bacteria die, the IM potential is lost, which results in a drop of red fluorescence. Loss of IM potential, initiated at the peak in red:green ratio, was also delayed with C9TMH-1 lock compared to C9wt (). Interestingly, C9TMH-1 lock did cause more rapid influx of DiOC2 () and nisin () through the OM compared to C5b-8 alone. This suggests that C9TMH-1 lock can slightly increase OM damage. Nonetheless, together these data highlight that polymerization of C9 increases damage to the OM.

An excess of C9 is required for efficient polymerization of C9 on bacteria

We next used C9TMH-1 lock to study the efficiency of C9 polymerization on bacteria when the amount of available C9 was limited. E. coli MG1655 were labelled with a constant amount of C5b-8 while the amount of available C9wt and C9TMH-1 lock was limited. Binding of C9wt was comparable to C9TMH-1 lock when the amount of available C9 was below 1 nM (). In addition, below 1 nM C9 no polymeric-C9 was detected via antibody staining () and SDS-PAGE () for both C9wt and C9TMH-1 lock. This suggests that there is little polymerization when the amount of available C9 is limited. As the amount of available C9 was increased above 1 nM C9, binding of C9wt continued to increase up to 10-fold higher than C9TMH-1 lock (), whereas binding of C9TMH-1 lock saturated (). This coincided with the detection of polymeric-C9 via antibody staining () and SDS-PAGE () for C9wt, whereas only monomeric-C9 was detected for C9TMH-1 lock (). In addition, perimCherry only leaked through the OM when the concentration of C9wt was above 1 nM (). Since leakage of perimCherry is dependent on the presence of polymeric-C9 (), these data suggest that C9 does not efficiently polymerize when the amount of available C9 is limited. Finally, when the amount of surface-bound C5b-8 was varied by titrating C8, the difference in C9wt binding compared to C9TMH-1 lock binding decreased when an excess of C8 was added (). perimCherry leakage () and SDS-PAGE () confirmed that an excess of C8 decreased the relative abundance of polymeric-C9 compared to monomeric-C9. This suggests that an increase in the amount of C5b-8 on the bacterial surface impairs polymerization of C9. In conclusion, these data suggest that an excess of C9 is required for efficient polymerization of C9 on bacteria.

Fig 4

An excess of C9 is required for efficient polymerization of C9 on bacteria.

E. coli MG1655 was labelled with C5b-7 by incubating in 10% C8-depleted serum for 30 minutes. Bacteria were washed and next incubated with 10 nM C8 and a concentration range of Cy5-labelled C9wt or C9TMH-1 lock for 30 minutes to measure binding of C9. a) Binding of Cy5-labelled C9wt or C9TMH-1 lock to bacteria measured by flow cytometry. b) The relative increase in C9wt binding compared to bacteria labelled with C9TMH-1 lock was calculated as indication for C9 polymerization. c) Bacteria were stained with AF488-labelled mouse anti-polymeric-C9 aE11-antibody and staining was measured by flow cytometry. Bacterial cell pellets were analyzed by SDS-PAGE for in-gel fluorescence of Cy5-labelled C9wt (d) or Cy5-labelled C9TMH-1 lock (e) to distinguish monomeric-C9 from polymeric-C9. f) periplasmic mCherry (perimCherry) leakage was measured by flow cytometry and represented as relative perimCherry fluorescence compared to t = 0. Flow cytometry data are represented by geoMFI values of the bacterial population. Data represent mean +/- SD of three independent experiments (a, b, c and f). SDS-PAGE images are representative for at least three independent experiments.

Polymerization of C9 enhances bacterial cell envelope damage and killing in serum

So far, we have looked at the effect of C9 polymerization on the MAC in the absence of other serum components. Next, we wanted to see if polymerization of C9 also enhances bacterial cell envelope damage and killing in a serum environment. E. coli MG1655 was incubated in 3% C9-depleted serum with C9wt or C9TMH-1 lock and influx of DiOC2 or Sytox were measured over time in a plate-reader to measure OM and IM damage respectively. OM damage () preceded IM damage (), but both were delayed by 30 minutes with C9TMH-1 lock compared to C9wt. Also, bacterial viability was a 1,000-fold higher for C9TMH-1 lock compared to C9wt after 30 minutes (). These data suggest that polymerization of C9 enhances bacterial cell envelope damage and killing in serum.

Interestingly, bacteria were killed with C9TMH-1 lock to a comparable level as C9wt after 90 minutes. This was not observed with C9TMH-1 lock in the absence of serum (), which indicated that the presence of serum affects MAC-dependent killing, even in the absence of polymerized C9. In serum, complement activation is still ongoing, which could ultimately result in more MAC pores on the surface of bacteria and explain why viability still decreases in time. Indeed, stopping further MAC formation, by adding C5 conversion inhibitor OmCI after 30 minutes, prevented bacterial killing in C9-depleted serum with C9TMH-1 lock over a 100-fold (). This suggests that ongoing conversion of C5 and subsequent MAC formation is primarily responsible for the increase in bacterial killing over time. Serum also contains bactericidal enzymes that can more efficiently pass through a damaged OM, such as lysozyme (14.7 kDa) and type IIa secreted phospholipase 2A (PLA, 14.5 kDa) [27]. Both lysozyme and PLA decreased bacterial viability when C5b-8 labelled bacteria were incubated with C9TMH-1 lock compared to C5b-8 alone (). This suggests that these serum proteins also contribute to MAC-dependent killing in serum in the absence of polymerized C9. Nonetheless, taken together our data mainly highlight that polymerization of C9 enhances the efficiency by which MAC pores damage the bacterial cell envelope and kill bacteria in serum.

Polymerization of C9 enhances cell envelope damage for several E. coli and Klebsiella strains in serum

We wondered if our findings on E. coli MG1655 could also be extrapolated to other E. coli strains and other Gram-negative species. First, the effect of C9 polymerization on OM and IM damage was measured for three other complement-sensitive E. coli strains (BW25113, MC1061 and 547563.1) in serum. To compare strains, time points were interpolated at which the red:green fluorescence ratio of DiOC2 reached half the value of the peak (t1/2peak, shown in ) as measure for OM damage, and at which half the maximum Sytox value was reached (t1/2maximum, shown in ) as measure for IM damage. Both OM damage (Figs and ) and IM damage (Figs and ) were delayed with C9TMH-1 lock compared to C9wt for all tested E. coli strains (MG1655, BW25113, 547563.1 and MC1061). C9TMH-1 lock did increase OM damage and IM damage compared to C9-depleted serum alone for three out of four E. coli strains (Figs , ). For one strain, 547563.1, C9TMH-1 lock only slightly enhanced IM damage, since C9-depleted serum alone already damaged the IM of these bacteria (Figs and ). OM damage (Figs and ) and IM damage (Figs and ) were also delayed with C9TMH-1 lock compared to C9wt for two clinical Klebsiella isolates (Klebsiella variicola 402 and Klebsiella pneumoniae 567880.1). C9TMH-1 lock did not cause IM damage within 90 minutes for Klebsiella, suggesting that IM damage by MAC pores for Klebsiella was more dependent on polymerization of C9 than for E. coli. In conclusion, these data suggest that C9 polymerization enhances cell envelope damage for multiple complement-sensitive E. coli and Klebsiella strains in serum.

Fig 5

Polymerization of C9 enhances bacterial cell envelope damage and killing in serum.

E. coli MG1655 was incubated in 3% C9-depleted serum supplemented with a physiological concentration (= 25 nM) of C9wt or C9TMH-1 lock for 90 minutes. As negative control, 25 μg/ml C5 conversion inhibitor OmCI was added to the serum. a) OM damage was measured by DiOC2 influx, which was determined by the shift in red:green fluorescence ratio over time in a multiwell plate-reader assay. b) IM damage was measured by Sytox influx over time in a multiwell plate-reader assay. c) Bacterial viability was determined at different time points by counting colony forming units (CFU’s) and calculating the proportion of viable cells compared to t = 0. At t = 30 (vertical dotted line), buffer was added (closed symbols) or OmCI (open symbols) to stop MAC formation. The horizontal dotted line represents the detection limit of the assay. d-e) Other E. coli strains (BW25113, MC1061, 547563.1) and K. variicola 402 were also incubated in 3% C9-depleted serum supplemented with C9wt or C9TMH-1 lock for 90 minutes. K. pneumoniae 567880.1 was incubated in 10% C9-depleted serum because of less efficient activation of the complement cascade and was therefore also supplemented with 80 nM C9wt or C9TMH-1 lock. d) OM damage was represented for all strains by the time when DiOC2 influx reached half the value of the peak (t1/2peak, shown in a for C9wt). e) IM damage was represented for all strains by the time when Sytox fluorescence reached half the maximum value (t1/2maximum, shown in b for C9wt). Multiwell plate-reader assays (a and b) are shown by one representative experiment that has been repeated at least three times. Data represent mean +/- SD of three independent experiments (c, d and e).

Polymerization of C9 is impaired on complement-resistant E. coli that express LPS O-Ag

In our previous study we have identified E. coli isolates that survive killing by MAC pores because MAC does not stably insert into the OM [15]. Here, we wanted to see if C9 polymerizes on these complement-resistant E. coli strains (clinical isolates 552059.1, 552060.1 and 567705.1). Binding of C9TMH-1 lock in 3% C9-depleted serum was comparable between all strains, suggesting that the total amount of C5b-8 on the surface was comparable for complement-sensitive and complement-resistant E. coli (). However, for complement-resistant E. coli binding of C9wt was similar to C9TMH-1 lock, indicating that C9 did not polymerize (). By contrast, binding of C9wt was 5- to 15-fold higher than C9TMH-1 lock for all four complement-sensitive E. coli (). Adding up to 10-fold more C9 slightly increased the difference between C9wt binding and C9TMH-1 lock 2- to 3-fold on complement-resistant 552059.1 (), but this difference was still 5-fold lower compared to complement-sensitive MG1655. Although this indicated that some C9 polymerized on 552059.1, bacteria were still not killed (). These data suggest that polymerization of C9, but not binding of C9 to C5b-8, is impaired on complement-resistant E. coli.

Polymerization of C9 is impaired on complement-resistant E. coli that express LPS O-Ag.

a-c) Complement-resistant (552059.1, 552060.1 and 567705.1) and complement-sensitive (MG1655, BW25113, MC1061 and 547563.1) E. coli strains were incubated in 3% C9-depleted serum supplemented with a physiological concentration (= 25 nM) of Cy5-labelled C9wt or C9TMH-1 lock for 30 minutes to measure C9 binding by flow cytometry. a) Binding of Cy5-labelled C9TMH-1 lock to bacteria. b) The relative increase in C9wt binding compared to bacteria labelled with C9TMH-1 lock was calculated as indication for C9 polymerization. c) The relative increase in C9wt binding compared to bacteria labelled with C9TMH-1 lock was compared for complement-sensitive MG1655 and complement-resistant 552059.1 at different C9 concentrations. d-e) CGSC7740 wildtype and wbbL+ were incubated in 3% C9-depleted serum similarly as E. coli strains in a-c). d) Binding of Cy5-labelled C9TMH-1 lock to bacteria. e) The relative increase in C9wt binding compared to bacteria labelled with C9TMH-1 lock was calculated as indication for C9 polymerization. Flow cytometry data are represented by MFI values of the bacterial population. Data represent individual values with mean +/- SD of three (a, b and c) or four (d and e) independent experiments. Statistical analysis was done on log-transformed data (2log for c and e, 10log for d) using a paired one-way ANOVA with Tukey’s multiple comparisons’ test (a-c) or a paired t-test (d and e). Significance was shown as * p ≤ 0.05.

We wondered if the LPS O-Ag is responsible for impairing polymerization of C9. The LPS O-Ag has frequently been associated with complement-resistance [23], although it remains unclear how. All three complement-resistant strains express LPS O-Ag (), whereas only one out of four complement-sensitive strains used in this study expresses LPS O-Ag. This indicated that the LPS O-Ag could be involved in impairing polymerization of C9. To directly study if O-Ag impairs polymerization of C9, we restored O-Ag expression in a complement-sensitive E. coli K12 strain (CGSC7740) that does not express O-Ag [35]. This strain lacks O-Ag because of an IS5-element that inactivates the wbbL gene that encodes an essential rhamnose transferase required for the expression of O-Ag [36]. Removal of this IS5-element restored O-Ag expression in CGSC7740 wbbL+ () and resulted in resistance to MAC-dependent killing in 5% serum (). Binding of C9TMH-1 lock in C9-depleted serum was comparable between CGSC7740 wbbL+ and the wildtype strain (), suggesting that complement was activated in a comparable manner. However, binding of C9wt was only 2.5-fold higher than C9TMH-1 lock on CGSC7740 wbbL+, whereas C9wt bound 12-fold more compared to C9TMH-1 lock on the wildtype strain (). This suggests that polymerization of C9 is impaired on CGSC7740 wbbL+. In summary, these data highlight that expression of O-Ag results in resistance to MAC-dependent killing and impairs polymerization of C9.

Discussion

Understanding how MAC pores damage the bacterial cell envelope and kill bacteria is important to understand how the complement system prevents infections. Although it was still unclear whether a completely assembled MAC pore is needed to kill bacteria, we here show that it is important to efficiently damage the bacterial cell envelope and rapidly kill multiple Gram-negative bacterial strains and species.

Our findings suggest that bacteria are killed more rapidly when C9 polymerizes, both by MAC alone as well as in a serum environment. Previous reports have already suggested that the absence of C9 delays bacterial killing [37], and have correlated the presence of polymeric-C9 to bacterial killing [28,38]. Our study extends on these insights, since it provides direct evidence that polymerization of C9 enhances bacterial killing by using a system in which C9 can bind to C5b-8 without polymerizing. Although Spicer et al. had already demonstrated that locking the TMH-1 domain of C9 could prevent polymerization of C9 [30], we extend on these insights by confirming that this does not affect binding of C9 to C5b-8 with direct binding experiments on E. coli. Apart from bacterial killing, our study also suggests that polymerization of C9 more efficiently damages both the OM and IM of the bacterial cell envelope. Polymerization of C9 strongly enhanced passage of small molecules through the OM and was required for passage of periplasmic proteins through the OM. Our findings therefore extend on previous reports that have suggested that polymeric-C9 is required for efficient cell envelope damage [27,28]. OM damage preceded IM damage, corresponding with our earlier findings [14]. Here, we add to these insights showing that OM damage precedes the loss of IM potential, which is a widely accepted characteristic of cell death.

Our data indicate that little polymeric-C9 already strongly enhanced bacterial killing and cell envelope damage. However, it remains unclear if a full polymeric-C9 ring of 18 C9 molecules is required. The aE11 antibody that is used for the detection of polymeric-C9 can also detect soluble MAC [32], which only contains up to 3 C9 molecules [39]. Although SDS-PAGE only revealed a single band for C9 polymers, it is still unknown if this band contains a mixture of C9 polymers of different sizes or if smaller C9 polymers are SDS-stable. Homologous pore-forming proteins of the MACPF/CDC family, such as suilysin and perforin, are known to form arc-shaped pores instead of ring-shaped pores [33,34]. It is therefore possible that arc-shaped MAC pores with only a few C9 molecules could be formed on bacteria and could enhance cell envelope damage and bacterial killing. Arc-shaped MAC pores have previously been observed on model-lipid membranes [17]. However, previous analyses on the bacterial OM with atomic force microscopy (AFM) primarily revealed the presence of ring-shaped MAC pores [14,15]. This suggests that even if arc-shaped MAC pores can be formed and have bactericidal capacity, it remains questionable if they are formed under physiological conditions.

Based on our results, we hypothesize that extensive OM damage by MAC pores is driving bacterial killing. Binding of C8 to C5b-7 allows passage of small molecules through the OM (), but these lesions have minimal effect on the IM and bacterial viability. The presence of C9TMH-1 lock slightly increased damage to both the OM and IM. Based on the structure of monomeric-C9 [30], it is highly unlikely that the other TMH domain (TMH-2) of C9 can insert into the membrane when the TMH-1 domain is locked. However, we can never fully exclude that some C9TMH-1 lock still polymerizes. If there is some residual polymerization, it seems likely that this would mainly result in arc-shaped and not ring-shaped MAC pores, which could explain why C9TMH-1 lock primarily enhances passage of smaller molecules (such as DiOC2) through the OM compared to proteins such as perimCherry. However, our data do not show any direct evidence that C9TMH-1 lock polymerized. We therefore hypothesize that binding of C9 to C5b-8 without polymerization of C9 could affect the stability by which C8 is inserted into the OM, which ultimately could affect cell envelope damage (). Nonetheless, altogether our data highlight that polymerization of C9 drastically enhanced the damage of both the OM and IM damage and rapidly killed bacteria ().

Assembly of MAC pores enhances OM damage and bacterial killing.

Schematic overview how MAC pores assemble on E. coli and their effect on outer membrane damage and bacterial killing. I) C8 (orange) binds to membrane-anchored C5b-7 (grey) and subsequently inserts transmembrane β-hairpins into the bacterial outer membrane (OM), which causes small lesions in the OM. II) Binding of C9 (blue) to C5b-8 without polymerization of C9 slightly increases the OM damage and bacterial killing. III) C9 polymerizes and forms a transmembrane polymer, which drastically increases OM damage and rapidly kills bacteria.

Our study primarily emphasizes that polymerization of C9 strongly enhances bacterial cell envelope damage and killing. How OM damage by MAC pores destabilizes the IM and kills bacteria remains unclear. The OM is an essential load-bearing membrane that confers stability to the bacterial cell [40]. The extent of OM damage by MAC pores could therefore determine whether a cell can cope with the increase in osmotic pressure. OM damage by MAC pores could even directly interfere with osmoregulation of the cell, as has been suggested for antimicrobial peptides that damage the OM of E. coli [41]. Moreover, since metabolism and growth-phase have been associated with sensitivity to killing by MAC pores [42,43], OM damage could also induce a stress response that affects the capacity of a cell to survive envelope stress. C9-depleted serum alone damaged the IM for one E. coli strain, even though the amount of C5b-8, indicated by the binding of C9TMH-1 lock, and OM damage was comparable to other E. coli strains. On the other hand, for Klebsiella strains IM damage appeared more dependent on polymerization of C9 than for E. coli strains. These differences between strains and species suggest that the capacity of a bacterium to cope with OM damage-related stress is a crucial determinant for bacterial killing by MAC pores. Further research looking directly at the cellular response of bacteria would be necessary to better understand how OM damage by MAC pores causes cell death.

Our data also suggest that an excess of C9 is required for efficient polymerization of C9. Joiner et al. already demonstrated that limiting the C9 concentration decreased the ratio of C9:C7 on the surface [38]. We further extend on these insights, because our data suggest that binding of C9 to C5b-8 is comparable or even favored over binding to C9 in the nascent polymeric-C9 ring (C5b-9x). This suggests that the ratio between the generated C5b-8 and the amount of available C9 is important to ensure assembly of complete MAC pores. MacKay et al. have previously demonstrated that C9 binds comparably to C5b-8 or C5b-91 by adding C9 to E. coli labelled with C5b-8 at 0°C [29]. By contrast, Parsons et al. recently suggested that binding and C9 to C5b-8 and subsequent insertion into the membrane is a kinetic bottleneck in the assembly of complete MAC pores [44]. However, Parsons et al. looked at MAC assembly in the presence of an excess of C9, which we here show to affect the assembly kinetics of MAC pores. Moreover, experiments in this study were done on model lipid-membranes that were not labelled with convertases, which also could have influenced the assembly of MAC pores [14]. Similar time-resolved AFM experiments on bacteria could provide more insight into the kinetics of the assembly of complete MAC pores on the bacterial surface.

Finally, our study highlights that polymerization of C9 is impaired on E. coli that resist MAC-dependent killing. Impaired polymerization of C9 correlated with the presence of LPS O-Ag in the tested E. coli strains. Moreover, restoring O-Ag expression in an O-Ag negative strain resulted in impaired polymerization of C9 and resistance to MAC-dependent killing. The presence and length of LPS O-Ag have previously been correlated with survival in serum [23]. Our data extend on these findings suggesting that the O-Ag results in complement-resistance by inhibiting polymerization of C9. Although LPS O-Ag modifications can also affect earlier steps in the complement cascade [9,10,45-47], this did not seem to play a major role as the amount of C5b-8 on the surface was comparable between complement-sensitive and -resistant E. coli. How the O-Ag affects polymerization of C9 remains unclear. The O-Ag could affect the distance of the nascent MAC to the OM and thereby the accessibility of hydrophobic patches in the OM [8]. Our previous study demonstrated that C9 is not efficiently inserted into the OM of the same complement-resistant E. coli used in this study [15], because C9 was more sensitive to trypsin digestion compared to complement-sensitive E. coli. We therefore hypothesize that the presence of O-Ag results in MAC assembly further away from the OM, which prevents C9 from inserting into the OM and forming a polymeric-C9 ring. Future experiments looking directly at the distance of MAC components to the OM could help further understand how the O-Ag affects MAC assembly on complement-resistant E. coli.

In conclusion, our study provides insight into how MAC pores damage the bacterial cell envelope and kill Gram-negative bacteria. Moreover, our study highlights how bacteria resist killing by MAC pores. These fundamental insights are important to understand how the complement system prevents infections and how bacteria escape killing by the immune system. Ultimately, these insights may guide the development of future immune therapy against bacteria.

Materials and methods

Serum and complement proteins

Serum depleted of complement components C5, C8 or C9 was obtained from Complement Technology. Serum was thawed and aliquoted, but not subjected to any further freeze-thaw cycles. Preassembled C5b6 (pC5b6) and C8 were also obtained from Complement Technology. His-tagged complement components C5, C6 and C7 were expressed in HEK293E cells at U-Protein Express as described previously [15]. OmCI was produced in HEK293E cells at U-Protein Express as well and purified as described before [48]. Monoclonal mouse-anti polymeric-C9 (aE11, kindly provided by T. Mollness and P. Garred) was randomly labelled with NHS-Alexa Fluor AF488 (ThermoFisher) according to manufacturer’s protocol. Lysozyme was obtained from Raybio and recombinant type IIa secreted phospholipase 2A (PLA) was kindly provided by Gérard Lambeau [49].

Expression and purification of C9

C9wt and C9TMH-1 lock (F262C V405C) were cloned into the vector pcDNA34 (Thermo Fisher Scientific) that was modified with a NheI/NotI multiple cloning site. This vector contains a cystatin-S signal peptide and was further modified to encode for the expression of a C-terminal AAA-3x(GGGGS)-LPETGG-HHHHHH tag. gBlocks (Integrated DNA Technologies) containing codon optimized C9 sequences were cloned via Gibson assembly into the NheI/NotI digested pcDNA34 and transformed into Top10F’ E. coli. After verification of the correct sequence, the plasmids were used to transfect EXPI293F cells. EXPI293F cells were grown in EXPI293 medium (Life Technologies) in culture filter cap Erlenmeyer bottles (Corning) on a rotation platform (125 rotations/min) at 37°C, 8% CO2. One day before transfection, cells were diluted to 2x106 cells/ml. The next day, cells were diluted to 2x106 cells/ml using SFM4Transfx-293 medium, containing UltraGlutamine I (VWR International) prior to transfection using PEI (Polyethylenimine HCl MAX; Polysciences). 0.5 μg DNA/ml cells, containing 50% empty (dummy) vector, was added to Opti-MEM (1:10 of total volume; Gibco) and gently mixed. After adding 1 μg/ml PEI in a PEI/DNA (w/w) ratio of 5:1, the mixture was incubated at room temperature for 20 min and added dropwise to cells while manually rotating the culture flask. After 3.5 hours, 1 mM valproic acid (Sigma) was added. After 5 days of expression, the cell supernatant was collected by centrifugation and filtration (0.45 μm). Cell supernatant was diafiltrated over a 30 kDa membrane on a Quixstand (GE healthcare) to Tris/NaCl buffer (50 mM Tris/500 mM NaCl at pH 8.0). Proteins were finally loaded on a Ni2+-loaded HiTrap HP Chelating column (GE healthcare) in Tris/NaCl buffer supplemented with 40 mM imidazole and eluted with 150 mM imidazole. Final purification was done by size-exclusion chromatography (SEC) on a Superdex 200 Increase column (GE Healthcare) on an Akta Explorer (GE Healthcare) with PBS. The concentration of proteins was determined by measuring absorbance at 280 nm and verified by SDS-PAGE.

Site-specific fluorescent labelling of MAC components

C6 and C9 were labelled with fluorescent probes as described previously [14,15]. 50 μM of protein with C-terminal LPETGG-His tag was incubated with 25 μM His-tagged sortase-A7+ [50] and 1 mM GGG-substrate in Tris/NaCl buffer (50 mM Tris/300 mM NaCl at pH 7.8) for two hours at 4°C. GGGK-FITC (Isogen Life Science) was used for C6-LPETGG-His and GGGK-azide (Genscript) for C9-LPETGG-His. Sortagged proteins were purified on a HisTrap FF column (GE Healthcare), which captures protein that was not sortagged and still contains a His-tag. FITC-labelled C6 was directly purified by SEC on a Superdex 200 Increase column on the Akta Explorer with PBS. GGG-azide labelled proteins were concentrated to 25 μM on a 30 kDa Amicon Tube (Merck Millipore) in Tris/NaCl buffer and next labelled with 100 μM DBCO-Cy5 (Sigma Aldrich) via copper-free click chemistry for 3h at 4°C. Finally, Cy5-labelled proteins were also purified by SEC on a Superdex 200 Increase column with PBS. Labelling of the proteins was monitored during SEC by measuring absorbance at 280 nm (protein), 488 nm (FITC) and 633 nm (Cy5) nm and finally verified by SDS-PAGE by measuring in-gel fluorescence with LAS4000 Imagequant (GE Healthcare).

Bacterial strains

Unless otherwise specified, the common laboratory E. coli strain MG1655 was used in our experiments. For experiments where leakage of periplasmic mCherry (perimCherry) was measured, MG1655 was used transformed with pPerimCh containing a constitutively expressed perimCherry previously used in [14]. Other laboratory E. coli strains that were used in this study included BW25113 and MC1061. Clinical isolates, namely E. coli 547563.1, 552059.1, 552060.1, 567705.1, Klebsiella variicola 402 and Klebsiella pneumoniae 567880.1, were obtained from the clinical Medical Microbiology department at the University Medical Center Utrecht. CGSC7740 wildtype and wbbL+ were kindly provided by Benjamin Sellner (Biozentrum, University of Basel). CGSC7740 wbbL+ was constructed by replacing the IS5-element present in the wbbL+ gene with a sacB-kan cassette to select for kanamycin-resistance. The sacB-kan cassette was then replaced with wildtype wbbL without the IS5-element and selected by counter selection on sucrose.

Bacterial growth

For all experiments, bacteria were plated on Lysogeny Broth (LB) agar plates. Single colonies were picked and grown overnight at 37°C in LB medium. For MG1655 transformed with pPerimCh, LB was supplemented with 100 μg/ml ampicillin. The next day, subcultures were grown by diluting at least 1/30 and these were grown to mid-log phase (OD600 between 0.4–0.6). Once grown to mid-log phase, bacteria were washed by centrifugation three times (11000 rcf for 2 minutes) and resuspended to OD 1.0 (~1 x 109 bacteria/ml) in RPMI (Gibco) + 0.05% human serum albumin (HSA, Sanquin).

Complement labelling and serum bactericidal assays

For MAC-specific bactericidal assays, bacteria were labelled with C5b-7 as described previously [15]. In short, bacteria (~1 x 108 bacteria/ml) were incubated with 10% C8-depleted serum (v/v) for 30 minutes at 37°C, washed three times and resuspended in RPMI-HSA. C5b-7 labelled bacteria (~5 x 107 bacteria/ml) were incubated for 30 minutes at 37°C with 10 nM C8 and 20 nM C9, unless stated differently. When C9 was reduced with 10 mM dithiothreitol (DTT), 20 nM C8 was added to bacteria (~5 x 107 bacteria/ml) 15 minutes before C9 was added to allow binding of C8 to C5b-7 at RT. Washing steps were done by pelleting bacteria at 11,000 rcf for 2 minutes and washing with RPMI-HSA. For serum bactericidal assays, bacteria (~5 x 107 bacteria/ml) were incubated with 3% C9-depleted serum (v/v) supplemented with physiological concentrations of C9 (100% serum ± 1 μM) for 30 minutes at 37°C, unless stated differently. KP880.1 was incubated in 10% C9-depleted serum (v/v) with the corresponding physiological concentration of C9. Blocking of C5 conversion in serum was done with 25 μg/ml OmCI as final concentration. For assays where nisin (Handary, SA, Brussels) was added, 3 μg/ml was used as final concentration.

Flow cytometry

Complement-labelled bacteria (~5 x 107 bacteria/ml) were incubated with 2.5 μM of Sytox Blue Dead Cell stain (Thermofisher). Samples were diluted to ~2.5 x 106 bacteria/ml in RPMI-HSA and subsequently analyzed on a MACSquant VYB (Miltenyi Biotech) for Sytox and perimCherry fluorescence. Polymeric-C9 deposition was measured by incubating bacteria (2.5 x 107 bacteria/ml) with 6 μg/ml monoclonal AF488 labelled mouse-anti polymeric-C9 (aE11) for 30 minutes at 4°C. For C9 binding in serum, bacteria were stained with 1 μM Syto9 (Thermofisher) to exclude serum noise events. Samples were next diluted to ~2.5 x 106 bacteria/ml in 1.1% paraformaldehyde (v/v) and subsequently analyzed on the BD FACSVerse flow cytometer for Cy5 and aE11-AF488 or Syto9 fluorescence. Flow cytometry data was analysed in FlowJo V.10. Bacteria were gated on forward scatter and side scatter. In serum, an additional trigger was placed on Syto9 fluorescence. Sytox positive cells were gated such that the buffer only control had <1% positive cells.

Polymeric-C9 detection by SDS-PAGE

Bacterial labelled with complement components were collected by spinning bacteria down 11,000 rcf for 2 minutes and subsequently washing cell pellets twice in RPMI-HSA. Cell pellets were resuspended and diluted 1:1 in SDS sample buffer (0.1M Tris (pH 6.8), 39% glycerol (v/v), 0.6% SDS (m/v) and bromophenol blue) supplemented with 50 mg/ml DTT and placed at 95°C for 5 minutes. Samples were run on a 4–12% Bis-Tris gradient gel (Invitrogen) for 75 minutes at 200V. Gels were imaged for 10 minutes with increments of 30 seconds on the LAS4000 Imagequant (GE Healthcare) for in-gel Cy5 fluorescence. Monomeric-C9 and polymeric-C9 were distinguished by size, since monomeric-C9 runs at 63 kDa and polymeric-C9 is retained in the comb of the gel.

Bacterial viability assay

Bacteria were treated with MAC components or serum as described above. Next, colony forming units (CFU) were determined by making serial dilutions in PBS (100, 1.000, 10.000 and 100.000-fold). Serial dilutions were plated in duplicate on LB agar plates and incubated overnight at 37°C. The next day, colonies were counted and the corresponding concentration of CFU/ml was calculated.

Multi-well fluorescence assays

Bacteria (~5 x 107 bacteria/ml) added to RPMI-HSA supplemented with 1 μM of Sytox Green Dead Cell stain (Thermofisher) or 30 μM DiOC2 (PromoCell). Bacteria were next incubated with MAC components, serum and/or nisin as described above. Fluorescence was measured every 60 seconds on a Clariostar platereader (BMG labtech). Sytox green fluorescence was measured using an excitation wavelength of 484–15 nm and emission wavelength of 527–20. DiOC2 fluorescence was measured using an excitation wavelength 484–15 nm and emission wavelength of 527–20 (green) and 650–24 (red). For DiOC2, the red fluorescence was divided by the green fluorescence to determine the red:green fluorescence ratio. t1/2peak for DiOC2 was interpolated at half the value of the peak, which was calculated by subtracting the background ratio value at t = 0 from the peak ratio value and dividing by two. t1/2maximum for Sytox was interpolated at half the value of the maximum Sytox value, which was calculated by subtracting the fluorescence value at t = 0 from the maximum fluorescence value and dividing by two.

LPS O-antigen Silver staining

E. coli strains were typed for LPS O-Ag by Silver staining after SDS-PAGE based on [51,52]. In short, bacteria were scraped from blood agars plates, resuspended in PBS and incubated at 56°C for 60 minutes. Cell pellets were next deproteinated with 400 μg/ml proteinase K for 90 minutes and diluted in 2x Laemli buffer with 0.7 M beta-mercaptoethanol. Cell pellets were run on a 4–12% BisTris gel as described above and fixed overnight in fixing buffer (40% ethanol (v/v) + 4% (v/v) glacial acetic acid). The gel was oxidized for 5 minutes in fixing buffer supplemented with 0.6% (m/v) periodic acid. The gel was then stained for 15 minutes with freshly prepared 0.3% silver nitrate in 0.125 M sodium hydroxide and 0.3% ammonium hydroxide (v/v). Finally, the gel was developed for 7 minutes in developer solution (0.25% citric acid (m/v) + and 0.2% formaldehyde (v/v)). In between steps, the gel was washed three times with MilliQ. Lipid A and LPS core were distinguished from LPS O-Ag based on size.

Data analysis and statistical testing

Unless stated otherwise, graphs are comprised of at least three biological replicates. Statistical analyses were performed in GraphPad Prism 8 and are further specified in the figure legends.

Labelling C9 with Cy5 and validating that locking the TMH-1 domain impairs polymerizing capacity.

a) SDS-PAGE of a concentration range of Cy5-labelled C9wt and C9TMH-1 lock (in μg/ml). Both InstantBlue staining (top) and in-gel Cy5 fluorescence (bottom) were shown. b) Sheep erythrocytes (4%) labelled with rabbit anti-sheep IgM were incubated in 2% C9-depleted serum for 30 minutes. Next, erythrocytes were washed and incubated with a concentration range of C9wt or C9TMH-1 lock in the presence or absence of 10 mM DTT. The percentage of lysed erythrocytes was calculated by adding MilliQ as 100% lysis or buffer as 0% lysis control after incubation with C9-depleted serum. c) 20 nM pC5b6, 20 nM C7, 20 nM C8 was incubated with 100 nM Cy5-labelled C9wt or C9TMH-1 lock. SDS-PAGE was done to distinguish monomeric-C9 from polymeric-C9 by in-gel Cy5 fluorescence. (d) C6-FITC binding to E. coli MG1655 measured by flow cytometry. Bacteria were labelled with C5b-7 by incubating them in 10% C8-depleted serum supplemented 12 μg/ml C6-FITC with for 30 minutes. Bacteria were washed and next incubated with 10 nM C8 for 15 minutes. Finally, 20 nM of Cy5-labelled C9wt or C9TMH-1 lock was added in the presence or absence of 10 mM DTT for 30 minutes. Flow cytometry data are represented by geoMFI values of the bacterial population. Data represent mean +/- SD (b) or individual values with mean +/- SD (d) of three independent experiments. Statistical analysis was performed using a paired one-way ANOVA with Tukey’s multiple comparisons’ test. For d, data were 10log-transformed. Significance was shown as * p ≤ 0.05, ** p ≤ 0.005, **** ** p ≤ 0.0001.

(TIF)

Click here for additional data file.

Unlocking TMH-1 domain after C9 has bound C5b-8 does not affect cell envelope damage.

E. coli MG1655 was labelled with C5b-7 by incubating in 10% C8-depleted serum for 30 minutes. Bacteria were washed and next incubated with 10 nM C8 and 50 nM C9wt or C9TMH-1 lock for 15 minutes. Next, bacteria were washed (@) and incubated with buffer or 5 mM DTT for 30 minutes. Bacteria were analyzed by flow cytometry for Sytox influx to determine inner membrane damage (a) and leakage of periplasmic mCherry (perimCherry) for outer membrane damage (b). Flow cytometry data are represented by relative geoMFI values of the bacterial population compared to bacteria in buffer. Data represent individual values with mean +/- SD of three independent experiments.

(TIF)

Click here for additional data file.

Polymerization of C9 enhances nisin influx.

E. coli MG1655 was labelled with C5b-7 by incubating in 10% C8-depleted serum for 30 minutes. Bacteria were washed and next incubated with 10 nM C8 and 20 nM of C9wt or C9TMH-1 lock supplemented with or without 3 μg/ml nisin to measure passage of nisin through the OM. Nisin influx was determined by measuring Sytox influx over time in a multi-well plate-reader assay. The graph shows one representative experiment that has been repeated at least three times.

(TIF)

Click here for additional data file.

Increasing the amount of surface-bound C5b-8 impairs polymerization of C9.

E. coli MG1655 was labelled with C5b-7 by incubating in 10% C8-depleted serum for 30 minutes. Bacteria were washed and next incubated with a concentration range of C8 and 3 nM (a,b,d) or 1 nM (c) Cy5-labelled C9wt or C9TMH-1 lock for 30 minutes. a) Binding of Cy5-labelled C9wt or C9TMH-1 lock to bacteria measured by flow cytometry. b) The relative increase in C9wt binding compared to bacteria labelled with C9TMH-1 lock was calculated as indication for C9 polymerization. c) periplasmic mCherry (perimCherry) leakage was measured after 30 minutes by flow cytometry and represented as relative perimCherry fluorescence compared to t = 0. d) Bacterial cell pellets were analyzed by SDS-PAGE for in-gel fluorescence of Cy5-labelled C9wt to distinguish monomeric-C9 from polymeric-C9. Flow cytometry data are represented by geoMFI values of the bacterial population. Data represent mean +/- SD of three independent experiments. The SDS-PAGE image is representative for at least three independent experiments.

(TIF)

Click here for additional data file.

C9TMH-1 lock increases passage of lysozyme and phospholipase IIa-2A through the outer membrane compared to C5b-8 alone.

E. coli MG1655 was labelled with C5b-7 by incubating in 10% C8-depleted serum for 30 minutes. Bacteria were washed and next incubated with 10 nM C8 and 20 nM of C9wt or C9TMH-1 lock. a) Bacterial viability was determined at different time points by counting colony forming units (CFU’s) and calculating the proportion of viable cells compared to t = 0. b) E. coli MG1655 bacteria were incubated with 10% C8 depleted serum for 30 minutes. After washing, bacteria were incubated with buffer, 10 nM C8 or 10 nM C8 + 20 nM C9 in the presence of 5 μg/ml lysozyme (lys) or 0.3 μg/ml recombinant type IIa secreted phospholipase 2A (PLA). Bacterial viability was determined after 90 minutes of incubation by counting CFU’s and calculating the proportion of viable cells compared to t = 0. The horizontal dotted line represents the detection limit of the assay. Data represent mean +/- SD (b) or individual values with mean +/- SD (a) of three independent experiments. Statistical analysis was done on 10log-transformed data (b) using a paired one-way ANOVA with Tukey’s multiple comparisons’ test. Significance was shown as ** p ≤ 0.005.

(TIF)

Click here for additional data file.

Polymerization enhances cell envelope damage on multiple Gram-negative strains.

E. coli strains (BW25113, MC1061, 547563.1) and K. variicola 402 were incubated in 3% C9-depleted serum supplemented with a physiological concentration (= 25 nM) of C9wt or C9TMH-1 lock for 90 minutes. K. pneumoniae 567880.1 was incubated in 10% C9-depleted serum supplemented with 80 nM C9wt or C9TMH-1 lock. a) OM damage was measured by DiOC2 influx, which was determined by the shift in red:green fluorescence ratio over time in a multiwell plate-reader assay. b) IM damage was measured by Sytox influx over time in a multiwell plate-reader assay. Graphs represent one representative experiment that has been repeated at least three times.

(TIF)

Click here for additional data file.

LPS O-Antigen expression confers complement-resistance.

a) Complement-resistant E. coli 552059.1 was incubated in 3% C9-depleted serum supplemented with 25, 75 or 250 nM C9wt, C9TMH-1 lock or 25 μg/ml OmCI. Bacterial viability was determined by counting colony forming units (CFU’s) per ml. The horizontal dotted line represents the detection limit of the assay. b) Complement-sensitive (MG1655, BW25113, MC1061 and 547563.1) and complement-resistant (552059.1, 552060.1, 567705.1) E. coli strains were typed for the presence of LPS O-Antigen (O-Ag) via Silver staining. LPS-core was distinguished from LPS O-Ag based on size. c) CGSC7740 wildtype (wt) and wbbL+ were typed for the presence of LPS O-Ag via Silver staining as done for b. d) CGSC7740 wt and wbbL+ were incubated in 5% normal human serum (NHS) supplemented with and without 25 μg/ml OmCI. Bacterial viability was determined by counting CFU’s and calculating the proportion of viable cells compared to t = 0. The horizontal dotted line represents the detection limit of the assay. SDS-PAGE images are representative for at least two independent experiments. Data represent individual values with mean +/- SD of three independent experiments (b and d).

(TIF)

Click here for additional data file.

13 Jul 2021

Dear Dr Bardoel,

Thank you very much for submitting your manuscript "Polymerization of C9 enhances bacterial cell envelope damage and killing by membrane attack complex pores" for consideration at PLOS Pathogens. As with all papers reviewed by the journal, your manuscript was reviewed by members of the editorial board and by several independent reviewers. In light of the reviews (below this email), we would like to invite the resubmission of a significantly-revised version that takes into account the reviewers' comments.

Thank you for submitting your manuscript to PLOS Pathogens. We have now finally received the reviews, which unfortunately were a bit late. As you will see the comments were many and ranged broadly.

As the reviewers found also merit in your work and manuscript, we would like to invite a thoroughly revised manuscript for a second review. However, we can not guarantee publication, unless the requirement by the reviewers are met appropriately. I sincerely hope that you will find the expert comments helpful.

We cannot make any decision about publication until we have seen the revised manuscript and your response to the reviewers' comments. Your revised manuscript is also likely to be sent to reviewers for further evaluation.

When you are ready to resubmit, please upload the following:

[1] A letter containing a detailed list of your responses to the review comments and a description of the changes you have made in the manuscript. Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

[2] Two versions of the revised manuscript: one with either highlights or tracked changes denoting where the text has been changed; the other a clean version (uploaded as the manuscript file).

Important additional instructions are given below your reviewer comments.

Please prepare and submit your revised manuscript within 60 days. If you anticipate any delay, please let us know the expected resubmission date by replying to this email. Please note that revised manuscripts received after the 60-day due date may require evaluation and peer review similar to newly submitted manuscripts.

Thank you again for your submission. We hope that our editorial process has been constructive so far, and we welcome your feedback at any time. Please don't hesitate to contact us if you have any questions or comments.

Sincerely,

Seppo Meri

Guest Editor

PLOS Pathogens

Christoph Tang

Section Editor

PLOS Pathogens

Kasturi Haldar

Editor-in-Chief

PLOS Pathogens

​orcid.org/0000-0001-5065-158X

Michael Malim

Editor-in-Chief

PLOS Pathogens

orcid.org/0000-0002-7699-2064

***********************

Dear Authors

Thank you for submitting your manuscript to PLOS Pathogens. We have now finally received the reviews, which unfortunately were a bit late. As you will see the comments were many and ranged broadly.

As the reviewers found also merit in your work and manuscript, we would like to invite a thoroughly revised manuscript for a second review. However, we can not guarantee publication, unless the requirement by the reviewers are met appropriately. I sincerely hope that you will find the expert comments helpful.

With best regards

Seppo Meri

Guest Editor

PLOS Pathogens

Reviewer's Responses to Questions

Part I - Summary

Please use this section to discuss strengths/weaknesses of study, novelty/significance, general execution and scholarship.

Reviewer #1: In this manuscript the authors make use of the ingenious “locked C9” reported by the Dunstone group in 2018 to explore whether MAC-mediated bacterial killing requires C9 polymerisation. The manuscript is very well written and easy to read – for which this reviewer is grateful. I do have some concerns about the presentation and interpretation of the interesting data; these are listed below.

Reviewer #2: This is a very well written paper that uses innovative tools to evaluate the role of C9 in killing of gram-negatives. This paper unifies several older observations and also lends new insights into the role of C9 in killing gram negatives. Differences across strains and species with respect to how MAC is assembled and inserted have been elucidated. The experiments all have stringent controls. The data have been presented very clearly and the discussion is excellent.

Reviewer #3: This work closely follows on from a previous Plos pathogens paper that describes how the MAC facilitates pore formation of the outer membrane of E. coli and delivery of other factors that damage cell wall/membrane integrity. This particular manuscript explores whether the MAC needs to form a full pore or not and how incomplete pore formation affects function although some of the research described as kinetics contradicts published SM, time-resolved work. There is also exploration of how the MAC might be inhibited in complement resistant E. coli. Overall there is an interesting set of experiments but it is difficult to explain as the regain of function of the lock mutant in reducing conditions is not perfect and the DTT does affect the wtC9 function probably due to the partial loss of the normal disulphide bonds required for folding and stability of C9. This data needs to be presented and interpreted in a more upfront manner with the interpretation that the mutant is not completely inactive.

**********

Part II – Major Issues: Key Experiments Required for Acceptance

Please use this section to detail the key new experiments or modifications of existing experiments that should be absolutely required to validate study conclusions.

Generally, there should be no more than 3 such required experiments or major modifications for a "Major Revision" recommendation. If more than 3 experiments are necessary to validate the study conclusions, then you are encouraged to recommend "Reject".

Reviewer #1: In the abstract and elsewhere the authors state that there has been, prior to the availability of the “locked C9” reagent, “no robust way to specifically prevent polymerization of C9”. I refer the authors to the excellent 1994 paper from McKay and Dankert that uses temperature shifts to do exactly this in the context of bacterial killing. No doubt, the current work builds on this observation but they should fully acknowledge it.

At the risk of being pedantic, the description of the MAC in lines 57/58 does not match the description in the references used; in these Bubeck’s group show clearly that the pore is composed of 22 staves, 18 from C9 and one each from C6, C7, C8a and C8b. Similarly pedantic – but important I think; Spicer demonstrated rather than suggested the lock C9 (line 71).

The first dataset explores the impact of C9 polymerisation. Although I usually hate cartoons in figures, 1a and 1b are nice and likely to be of help to non-experts. Supp1b shows minimal (but not zero) ShE lysis using the locked C9; the Spicer paper found no haemolytic activity in the locked protein and the model proposed would suggest no effect – do the authors think that this is real? Figures c/d show binding of C9wt and C9lock to C5b-8, replicating what was shown in Spicer et al, albeit in the context of E coli, clarify that this is confirmatory. Figures 1e and S1c show C9 monomer/polymer in slot blots with no markers or other orientations. Here and elsewhere I strongly urge that full gels/blots with clear markers are used. Next, the authors test bacterial lysis using a neat method they have established; Fig 2a shows weak but significant activity in the C9lock – again, given the model proposed, any activity is surprising. Do the authors have an explanation of this? The impact of releasing the lock on lysis and viability is clear in 2b/c. The dose-dependent inhibition of poly-C9 formation shown in fig 2d is intriguing. Spicer showed very similar data in a different assay system and reached a similar conclusion – that the C9lock could bind and stall polymerisation at different stages; the current data though do beg the question of how many unfolded C9 molecules are needed for the ae11 epitope – given that the mAb detects TCC (2 or 3 C9s) likely not many! Do lower Mr ae11-stained bands appear? Are these sub-pore complexes SDS-labile? The bacterial lysis kinetics in figure 2e look very different – more of a threshold effect and kicking in at equimolar and above – state that this shows “that only very few C9 polymers are required to damage the IM”; however, it could equally show that even complexes with only a few C9s are lytic, the converse of the hypothesis.

Here and elsewhere I wonder why the authors have not washed after binding C9lock then unlocking in order to properly explore the impact of insertion of that first C9 – seems an obvious addition to many of the experiments that would distinguish effects of C5b-8, C5b-8-C9lock and C5b-8-C9unlock.....

OM lysis is next explored; 3a/b both show that C5b-8/C9wt causes OM damage while C5b-8/C9lock does not (as expected); a surprising omission is testing the impact of lock release – binding C9lock, washing and unlocking to test impact of C5b-8/C9 (single C9). Fig 3c shows a surprising residual activity in C9lock – state that this “suggests that binding of C9 to C5b-8 in the absence of polymerization slightly increases OM damage”. Again, this doesn’t fit the model – what might the mechanism be?

Next the authors explore binding and polymerisation of C9wt and C9lock to C5b-8-coated bugs. By titrating the C9s they show an initial similar increase in binding with dose up to an inflexion point where the C9lock is saturated but C9wt continues to bind. Only above this inflexion point is poly-C9 seen in bacterial pellets – although monomeric C9 is also only seen above this point and no data are shown for C9lock, presumably because neither mono- or poly-C9 can be detected; the data in 4d do not support their statement that SDS-PAGE shows only monomeric C9 at lower concentrations. The authors suggest this indicates that “C9 starts polymerizing when all C5b-8 complexes have bound one C9 molecule”; they note that this interpretation directly contradicts the Parsons study which shows, albeit in an artificial membrane system, that capture of the first C9 is rate-limiting and thereafter further C9 recruitment is rapid. This is an important clarification but I’m not sure that the data fully support it. It seems equally logical to suggest that at low doses all C9 offered is bound, either directly to C5b-8 (only that for C9lock) or both to C5b-8 and C5b-9 (for C9wt); when the C5b-8 sites are saturated the curves diverge, as is seen. Neither poly- nor mono-C9 binding is detected at the low C9 inputs as shown in fig 4d (note comment above regarding slotblots). The strong statement that “these data suggest that polymerization of C9 is rate-limiting in the assembly of complete MAC pores on bacteria” is not well supported by the data – kinetic analyses of the sort described in the Parsons study are likely needed.

Figure 5 shows killing of bacteria in C9-depleted serum supplemented with C9wt of C9lock to explore the requirement for C9 polymerisation for MAC effects on bacteria. These straightforward experiments confirm that polymerisation of C9 in the MAC is essential for efficient bacterial lysis/killing. The findings are complicated by the observation that killing in the presence of C9lock is delayed rather than prevented; they suggest that this is due to continued MAC deposition (supported by OmCI block) and/or presence of other bactericidal serum components. It would be interesting to confirm that there is no “lock-release” in these prolonged serum incubations that might enable polymerisation. The final dataset extends the key observation of requirement for polymerisation for efficient killing to other bacteria and also makes the interesting observation that C9 polymerisation (not C5b-8 formation or first C9 binding) is limited on resistant strains.

Reviewer #2: none

Reviewer #3: In this sense there is a lot of conceptual overlap with the previous Plos pathogens paper (ref 31 in this manuscript) which shows delivery of other factors. An explanation in the introduction from the authors about then potential overlap and how this is original research would be appreciated. I think that it is really important to put this upfront in the paper as there is a sense that there is redundancy in this manuscript with ref 31.

Figure 1: I find the diagram in figure 1A is confusing as it depicts C9 as binding to C5b-8 and then moving away from C5b-8 as a monomer. The monomer changes shape and then recruits a second monomer of C9. This is not actually how the Membrane attack complex forms and would confuse all readers regardless of whether they are experts or not. This needs to be fixed. The term polyC9 should only be used for the unusual characteristic of human C9 whereby it can form a homomeric ring that looks essentially like a pore but is really a non-physiological phenomenon. This figure needs major work.

Lines 103-5: The author's only use bar graphs to demonstrate the loss of activity for the disulfide trap mutant compared to wtC9. Whilst it might appear to be a clean cut binary style observation that there is either no activity in non-reducing conditions and activity in reducing conditions, when the original paper (Spicer et al.) is looked at, there is a lower LD50 for the reduced disulphide C9 compared to wtC9. Under no circumstances is it acceptable to have only bar graphs of activity for cytolytic protein activity. The original titration based activity profiles (done as replicates) should be presented for this type of data. This will provide a more qualitative view of the activity profile of the proteins being used and is a more honest way of showing the regain of function in the given reducing conditions. Curves should go to a relevantly high concentration of C9 as it is not uncommon for there to be some residual activity. The authors show a loss of wtC9 activity in the presence of DTT but this is not observed in the original Spicer paper so, again, the raw titration based curves would be important here.

Lines 113-123 and Figure 1c, e: I find that the use of the log scale on the Y-axis of Fig 1c dilutes the visual perception of the differences in binding of the trap mutant with and without DTT. In effect the binding of the lock mutant in the DTT is half of wt C9 in DTT. I would say from this data that the C9 lock mutant is half as good at binding/polymerising to the C5b-8 complex and maybe this should be stated as such. This interpretation is also supported by Fig 1e with a much smaller “polyC9” band which was correctly interpreted by the authors. What is missing is also the binding of aE11 to the reduced C9 lock mutant in Figure 1f to quantitate the formation of polymerised C9 and a more honest, quantitated statement of the amount of C9 polymerisation in the reduced form of the C9 lock mutant. I.e. it is not a perfect regain of wt-like activity and should be interpreted as such.

Figure 2a,b: It would be easier for the reader to interpret the experiments correctly if you combined Figures 2 a and b.

Lines 142-144, figure 2d,e: I find this to be a really important observation that could be expanded. Is it possible to interpret that arcs are able to allow the damage of the inner membrane at this stage of the manuscript? And refer to the background of data of arcs which is quite extensive for the MACPF/CDC family.

Lines 149 - 55, Figure 3a: the y-axis is not a linear-scale not a log scale. Not consistent with the other experiments. Is this experiment compared to E. coli not expressing peri-mCherry for background scattering as the baseline? Overall this is a weak observation but that is fine. The weak change may suggest that even with wtC9 there are only a few full pores that are capable of passaging mCherry as opposed to the passage of nisin and this should be noted in the results interpretation. Statistics need to be assigned to state it this is a significant observation or not as it looks like it is not. Maybe present the nisin work which is more significant before the mCherry work and state clearly the potential for arc formation as a possible explanation.

Lines 168-72: Given that there is potential background pore formation for the C9 lock mutant which should be explained in a more upfront way (i.e. Figure 1c), then a better/alternative explanation for Fig 3c is that there is still some background MAC formation that allows the passage of DiOC2. Again I think this is perfectly logical. There is also evidence of leaking membranes with the C5b-8 complex which may explain the black triangle curve.

Discussion: There definitely needs to be discussion about the role of arcs or incomplete MAC pores that can allow some passage of some molecules (DiOC2 and nisin) but not mCherry. And this should be reflected in the model presented in Figure 7 (just add an arc between step iii and step IV.)

Lines 189-90: “Altogether, these data suggest that polymerization of C9 is rate-limiting in the

assembly of complete MAC pores on bacteria.” This needs to be more clearly stated that the C9:C9 interaction is slower than the initial C8:C9 interaction. Having said this, it is impossible to reconcile this whole E.coli, ensemble data with published single molecule, time resolved data (AFM work published in ref 30) that provides more reliable evidence of the opposite kinetic observation? With this kind of ensemble data that relies on titration of protein then it isn’t really a true kinetic study.

Figure 4d: Would be complete with the Western blot of the C9 lock mutant as well.

Lines 235 - 262: There is not enough data to support the interpretation of the impaired C9 polymerisation on complement resistant E. coli being due to the improper anchoring of C5b-7. Further explanation or experiments would be needed for this section and it seems rather immature.

**********

Part III – Minor Issues: Editorial and Data Presentation Modifications

Please use this section for editorial suggestions as well as relatively minor modifications of existing data that would enhance clarity.

Reviewer #1: The discussion is clear, although some claims made need to be tempered and some areas suggested as novel more clearly described as confirmations of the Spiller findings – for example, that C9lock binding to C5b-8 is as efficient as WT (already in the Spiller paper which goes way beyond “suggesting”). The suggestion that binding of C9 somehow increases killing capacity of C5b-8 is repeated but how this off-model effect might be explained is not discussed. Although some of the rather strong historical data on MAC killing of bacteria is mentioned (Joiner and Brown – though only a single, albeit important, paper), some highly relevant, particularly from Dankert (see above) and Taylor, data are not included and would strengthen the discussion.

Reviewer #2: A minor editorial point – there is no reference to Fig 6d in the text – I think lines 258 and 261 refer to Fig 6d (and not 6c)?

Reviewer #3: Lines 109-10: “These data confirm that the capacity of C9TMH-1 lock to form polymers is impaired, and can be reversed by reducing the cysteine bridge lock.” Figure S1c does not show the “C9TMH1 lock” forming polyC9 in reducing conditions (in SDS-PAGE) and so this statement has not been experimentally qualified. This is later covered by analysis of Fig1e so move to that section?

Lines 138-9: “Reducing C9TMH-1 lock with DTT restored its capacity to damage the IM (Fig. 2b)” Add at the end of this phrase the following or equivalent “relative to wtC9 in reducing conditions which had overall reduced activity compared to wtC9 in non-reducing conditions”

Figure 2c figure legend line 666: is this 3 or 4 “individual values (c) of three independent experiments with mean +/- SD. Also the SD bar for the far right bar looks truncated?

Line 243 and other parts of methods. Make sure that you have appropriate units after each number full stop for example 10% c5-depleted serum should have volume per volume after the 10%

Line 456 why was there no tenfold dilution?

**********

PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

Figure Files:

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, . PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at .

Data Requirements:

Please note that, as a condition of publication, PLOS' data policy requires that you make available all data used to draw the conclusions outlined in your manuscript. Data must be deposited in an appropriate repository, included within the body of the manuscript, or uploaded as supporting information. This includes all numerical values that were used to generate graphs, histograms etc.. For an example see here on PLOS Biology: http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.1001908#s5.

Reproducibility:

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

14 Sep 2021

Submitted filename: Response letter.docx

Click here for additional data file.

20 Oct 2021

Dear Dr Bardoel,

We are pleased to inform you that your manuscript 'Polymerization of C9 enhances bacterial cell envelope damage and killing by membrane attack complex pores' has been provisionally accepted for publication in PLOS Pathogens.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. A member of our team will be in touch with a set of requests.

Please note that your manuscript will not be scheduled for publication until you have made the required changes, so a swift response is appreciated.

IMPORTANT: The editorial review process is now complete. PLOS will only permit corrections to spelling, formatting or significant scientific errors from this point onwards. Requests for major changes, or any which affect the scientific understanding of your work, will cause delays to the publication date of your manuscript.

Should you, your institution's press office or the journal office choose to press release your paper, you will automatically be opted out of early publication. We ask that you notify us now if you or your institution is planning to press release the article. All press must be co-ordinated with PLOS.

Thank you again for supporting Open Access publishing; we are looking forward to publishing your work in PLOS Pathogens.

Best regards,

Seppo Meri

Guest Editor

PLOS Pathogens

Christoph Tang

Section Editor

PLOS Pathogens

Kasturi Haldar

Editor-in-Chief

PLOS Pathogens

​orcid.org/0000-0001-5065-158X

Michael Malim

Editor-in-Chief

PLOS Pathogens

orcid.org/0000-0002-7699-2064

***********************************************************

There were comments from reviewer 3, especially concerning the binding of "locked" C9 to C5b-8 in relation to preventing C9 polymerization, and whether this was already demonstrated first by the current authors or in the paper by Spencer et al. There is also a general comment on using an appropriately wide range of concentrations when performing lysis assays. These comments can be acknowledged, but do not - in my view - necessitate another round of revision.

Reviewer Comments (if any, and for reference):

Reviewer's Responses to Questions

Part I - Summary

Please use this section to discuss strengths/weaknesses of study, novelty/significance, general execution and scholarship.

Reviewer #1: The authors have addressed the majority of my concerns in the revision. A few minor niggles remain but are not sufficient to justify a further revision.

Reviewer #3: This is a thorough study of MAC assembly, specifically C9 binding, in the context of the bacterial surface. Overall there are some great tools used in this study and is a step closer in under the mechanism of action of the MAC on bacteria.

**********

Part II – Major Issues: Key Experiments Required for Acceptance

Please use this section to detail the key new experiments or modifications of existing experiments that should be absolutely required to validate study conclusions.

Generally, there should be no more than 3 such required experiments or major modifications for a "Major Revision" recommendation. If more than 3 experiments are necessary to validate the study conclusions, then you are encouraged to recommend "Reject".

Reviewer #1: None.

Reviewer #3: Normally I would not too critical at this stage but it appears that the authors continue to over-embellish at times. There is a major point raised by reviewer 1, Point 5.

Line 120 "Although Spicer et al. already demonstrated that locking the TMH-1 domain prevents polymerization of C9 [30], we wanted to confirm that this did not prevent binding of C9 to C5b-8." And line 305 “Our study extends on these insights, since it provides direct evidence that polymerization of C9 enhances bacterial killing by using a system in which C9 can bind to C5b-8 without polymerizing. Although Spicer et al. had already demonstrated that locking the TMH-1 domain of C9 could prevent polymerization of C9 [30], we extend on these insights by confirming that this does not affect binding of C9 to C5b-8 with direct binding experiments on E. coli.”

Technically the FACS is still only looking at the binding of the added C9 to the cells with implication that it is binding to the MAC and is not really a direct protein:protein binding experiment or a SP experiment. I would say that this overall result is only a slight enhancement on the lysis activity only based work presented in the Spicer paper where the competition assay and the structure suggest that the first C9 will bind (locked or WT).

But I also accept that this at the next stage of revision and will take guidance from the editorial team.

ANSWER 19 (also reviewer 1, point 4)- We have added the full original experiment which included a concentration range of C9wt and C9TMH-1 lock (S1b). Spicer et al. measured only at one C9 concentration in a different experimental set-up, which makes it possible that they did not pick up this small haemolytic effect of C9TMH-1 lock.

The reviewer thanks the authors for adding the full experiment. Spicer et al use a different type of assay where they observe lysis over time which has more information compared to standard assays but, as the authors point out, do not do this on a range of concentrations. So, it would be good to point out to both the authors and the editors that in future all lysis activity assays (regardless of the type) should use a suitable range of concentrations. This should become standard for high quality journals, such as PLoS Pathogens, for future publications. It would be good if the pore forming protein community ensures that this is the standard.

**********

Part III – Minor Issues: Editorial and Data Presentation Modifications

Please use this section for editorial suggestions as well as relatively minor modifications of existing data that would enhance clarity.

Reviewer #1: None

Reviewer #3: (No Response)

**********

PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #3: No

4 Nov 2021

Dear Dr Bardoel,

We are delighted to inform you that your manuscript, "Polymerization of C9 enhances bacterial cell envelope damage and killing by membrane attack complex pores," has been formally accepted for publication in PLOS Pathogens.

We have now passed your article onto the PLOS Production Department who will complete the rest of the pre-publication process. All authors will receive a confirmation email upon publication.

The corresponding author will soon be receiving a typeset proof for review, to ensure errors have not been introduced during production. Please review the PDF proof of your manuscript carefully, as this is the last chance to correct any scientific or type-setting errors. Please note that major changes, or those which affect the scientific understanding of the work, will likely cause delays to the publication date of your manuscript. Note: Proofs for Front Matter articles (Pearls, Reviews, Opinions, etc...) are generated on a different schedule and may not be made available as quickly.

Soon after your final files are uploaded, the early version of your manuscript, if you opted to have an early version of your article, will be published online. The date of the early version will be your article's publication date. The final article will be published to the same URL, and all versions of the paper will be accessible to readers.

Thank you again for supporting open-access publishing; we are looking forward to publishing your work in PLOS Pathogens.

Best regards,

Kasturi Haldar

Editor-in-Chief

PLOS Pathogens

​orcid.org/0000-0001-5065-158X

Michael Malim

Editor-in-Chief

PLOS Pathogens

orcid.org/0000-0002-7699-2064