HeLa229 cells were purchased from the American Type Culture Collection (ATCC; Manassas, VA, USA) and cultured in Dulbecco’s Modified Eagle’s Medium (DMEM; HyClone, Logan, UT, USA) supplemented with 10% heat-inactivated fetal bovine serum (FBS; Gibco, Grand Island, NY, USA) at 37°C with 5% CO₂. C. trachomatis L2 (L2/434/Bu) was purchased from the ATCC, and a C. trachomatis L2 CPAF-deficient strain was kindly provided by Dr. Guangming Zhong (University of Texas Health Science Center, San Antonio, TX, USA), which originated by Dr. Raphael Valdivia (Duke University, Durham, NC, USA) (Snavely et al., 2014). The C. trachomatis strains were propagated in HeLa229 cells and purified as described previously (Caldwell et al., 1981). Aliquots of purified chlamydial elementary bodies (EBs) were stored in sucrose-phosphate-glutamate (SPG) storage buffer (220 mM sucrose, 12.5 mM phosphate, and 4 mM l-glutamic acid, pH 7.2-7.4) at −80°C until use.

HeLa229 cells were seeded in 6-well plates at a density of 6×10⁵ cells per well and were cultured overnight prior to chlamydial inoculation. After the medium was removed, C. trachomatis, diluted in 250 μL SPG, was directly inoculated onto the cell monolayers at an appropriate multiplicity of infection (MOI) as indicated for individual experiments for 2 h at 37°C with 5% CO₂. An equal amount of SPG without C. trachomatis was used to mock infect HeLa229 cells as a control. During chlamydial inoculation, 6-well plates were rocked slowly every 30 min to ensure the inoculum fully contacted with the cell monolayers. Following infection, the inoculum was removed and replaced with DMEM supplemented with 10% FBS. In the penicillin inhibition experiment, 100 U/mL penicillin G (Solarbio, Beijing, China) was added to the medium at the same time. Cells were cultured at 37°C in a 5% CO₂ incubator. At 0, 12, 24, 36 and 44 h post-infection, cells were washed twice with 1× phosphate-buffed saline (PBS), scraped off the plate, and then suspended in 250 μL PBS per well. The cell suspensions were transferred into 1.5-mL Eppendorf tubes, vortexed with 3-mm glass beads for 1 min, and then centrifuged at 500 g for 10 min at 4°C. The supernatants were collected and used in C5 cleavage assays and hemolysis assays.

Reactions were carried out at 37°C with a reaction volume of 30 μL in a 1.5-mL Eppendorf tube containing 400 nM purified human C5 (Complement Technologies, Tyler, TX, USA) mixed with 25 μL C. trachomatis-HeLa229 cell lysates (C. trachomatis infection group), mock-infected HeLa229 cell lysates (mock infection group), PBS (C5 control group) or EBs. In the inhibition assay, C. trachomatis-HeLa229 cell lysates were pre-mixed with 400 μM nafamostat mesylate (Abcam, Cambridge, MA, USA) for 10 min, or with 500 μM lactacystin (Glpbio, Montclair, CA, USA) for 30 min respectively, both at 37°C. Reactions were terminated at various time points as indicated for the individual experiments by immediately placed on ice. Cleavage products were subjected to Western blot analysis, enzyme-linked immunosorbent assay (ELISA), and C-terminal sequencing based on liquid chromatography-tandem mass spectrometry (LC-MS/MS).

Cleavage products (28 μL) were mixed with 7 μL 5× loading buffer (Beyotime Biotechnology, Shanghai, China) and boiled at 95°C for 5 min. Next, 10 μL samples were resolved by SDS-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to polyvinylidene difluoride (PVDF) membranes (Merck Millipore, Billerica, MA, USA), and blocked with 5% BSA in Tris-buffered saline (containing 0.5% Tween 20) for 60 min before incubation with the appropriate antibodies. Primary antibodies used were mouse monoclonal antibody against C5/C5a (Cat# ab11876, Abcam, 1:1000 dilution) to recognize the fragment of C5 containing C5a; mouse monoclonal antibody clones 100a against CPAF (kindly provided by Dr. Guangming Zhong, 1:50 dilution) to recognize the C-terminal 35-kDa fragment of CPAF; and biotin goat polyclonal antibodies against C. trachomatis (Cat# ab20387, Abcam, 1:1000 dilution) to recognize the 40-kDa chlamydial major outer membrane protein (MOMP). The bound primary antibodies were incubated with corresponding HRP-conjugated secondary antibodies (Cat# SA00001-0 and SA00001-1, ProteinTech Group, Chicago, IL, USA, 1:5000) and detected through enhanced chemiluminescence by using Luminata Classico Western HRP Substrate (Merck Millipore, Billerica, MA, USA).

Quantitative analysis of C5a in the cleavage products were analyzed using a commercially available human C5a ELISA kit (CUSABIO BIOTECH, Wuhan, China) according to the manufacturer’s instructions. A 1:1000 dilution was selected as the optimum dilution ratio of the sample based on the results of pre-experiments.

The three-dimensional X-ray crystallographic structure of human C5 protein (PBD ID: 3CU7) was obtained from protein data bank (PBD) database. The molecular dynamics simulation of docked complexes was carried out using the Desmond module of Schrödinger software (Schrödinger Release, version 2020). Here, OPLS3e force field was used to initiate the molecular dynamics simulation, and the system was solvated using TIP3 water model. The neutralization of the system was performed by adding counter ions. Energy minimization of the entire system was performed using OPLS3e, as it is an all-atom type force field. The geometry of water molecules, the bond lengths and the bond angles of heavy atoms were restrained using the SHAKE algorithm. Simulation of the continuous system was executed by applying periodic boundary conditions and long-range electrostatics was maintained by the particle mesh Ewald method. The equilibration of the system was performed using NPT ensemble with temperature at 300 k and pressure at 1.0 bar. The coupling of temperature-pressure parameters was done using the Berendsen coupling algorithm. On post-preparation of the system, the production run was performed for 50 ns with a time step of 1.2 fs and trajectory recording was done for every 50 ps summing up to the recording of 1000 frames. The calculation of the Root Mean Square Deviation (RMSD) was conducted for the backbone atoms and was analyzed graphically to understand the nature of protein-ligand interactions.

The lytic activity of C5b-9 was compared with that of C5b_Ct-9 by using established chicken erythrocyte lysis assays with modifications (Rawal and Pangburn, 1998; Krisinger et al., 2012). Briefly, to produce C5b,6 or C5b_Ct,6, 50% normal human serum (NHS; Solarbio, Beijing, China) was incubated with PBS (control group), 10 nM cobra venom factor (CVF; BOYAO, Shanghai, China), C. trachomatis-HeLa229 cell lysates, or both C. trachomatis-HeLa229 cell lysates and 10 nM CVF for the indicated time at 37°C, respectively. Tubes containing chicken erythrocytes (3×10⁷) in a final volume of 270 μL gelatin veronal-buffered saline with EDTA (GVBE; Complement Technologies) were prepared and kept on ice. An aliquot (30 μL) of the produced C5b,6 or C5b_Ct,6 was added to the chicken erythrocytes and the mixtures were incubated for 30 min at 37°C. Unlysed cells were removed by centrifugation for 5 min at 3000 g, then the amount of hemoglobin released from the lysed erythrocytes was quantitated by measuring the absorbance of the supernatant at 414 nm. The background signal was established with a control containing 15 μL NHS, 15 μL PBS, and 270 μL GVBE, but no chicken erythrocytes. Chicken erythrocytes incubated with Pierce™ IP Lysis Buffer (Thermo Fisher Scientific) was used as a positive control of 100% lysis.

RIPA Lysis and Extraction Buffer (Thermo Fisher Scientific) was used for sample lysis and protein extraction. The amount of protein was quantified with the BCA Protein Assay Kit (Bio-Rad, Hercules, CA, USA). Protein digestion by trypsin was performed according to filter-aided sample preparation (FASP) procedure (Wiśniewski et al., 2009). The digest peptides of each sample were desalted on Empore™ SPE Cartridges C18 (Sigma-Aldrich, St. Louis, MO, USA), concentrated by vacuum centrifugation and reconstituted in 40 µl of 0.1% (v/v) formic acid. LC-MS/MS analysis was performed on a Q Exactive mass spectrometer (Thermo Fisher Scientific) that was coupled to Easy nLC (Thermo Fisher Scientific). The MS raw data for each sample were combined and searched using the MaxQuant 1.5.3.17 software for identification and quantitation analysis.

The three-dimensional structure of CPAF protein (PBD ID: 3DOR) was obtained from PBD database. The three-dimensional structure of NM was obtained in the PubChem (https://pubchem.ncbi.nlm.nih.gov/). The NM and CPAF structures were processed by PyMOL (version 2.3) and AutoDockTools (version 1.5.6). Molecular docking was performed using Autodock 4 (Morris et al., 2009) to verify the binding activity of NM and CPAF. The final docking structure of the complex was evaluated according to the binding energy. The docking results were visualized in PyMOL.

Data are presented as means ± standard deviation (SD). All data were analyzed with GraphPad Prism 8 (GraphPad Software Inc., La Jolla, CA, USA) using two-tailed unpaired Student’s t tests and one-way ANOVA with Dunnett’s multiple comparisons test if three or more groups were compared. P <0.05 was considered statistically significant.

To explore the mechanism of C3-independent C5 activation induced by C. trachomatis infection, a co-incubation system containing purified human C5 (400 nM) and C. trachomatis-HeLa229 cell lysates (MOI=1) or mock infected HeLa229 cell lysates (mock-infected control) or PBS (C5 control) was established. This system mimics the process of contact between the contents of lysed epithelial cells and C5 in body fluids during genital C. trachomatis infection in the absence of C3. An anti-C5a antibody was utilized to detect the fragments of C5 cleavage products encompassing C5a. After incubation for 5 to 120 min, Western blot analysis revealed that co-incubation of C5 and C. trachomatis-HeLa229 cell lysates yielded a previously undescribed C5-derived ~38-kDa fragment encompassing C5a ( Figure 1A , Lines 2-7). This band was not detected in the C5 control ( Figure 1A , Line 1) or the mock-infected control ( Figure 1A , Line 8) after incubation for 120 min. Unlike the traditional C5 cleavage pattern ( Figure 1B ), an 11-kDa C5a band was not detected in any of the samples, despite optimization of Western blot conditions for this small molecular weight protein.

To investigate whether the C5a generated in co-incubation system was below the limitation of detection of Western blot analysis, an ELISA was subsequently performed. The ELISA showed that there was a certain amount of C5a in the C5 control group ( Figure 1C ). This could be due to the only 95% purity of the human C5 protein and the possible presence of a small amount of C5a impurities in it. Thus, the concentration of C5a in the C5 control group was taken as the baseline in the following comparison. In the C. trachomatis infection group, an increase in C5a content was not detected after incubation for 5 min (P > 0.05). However, after incubation for 15 to 120 min, the concentration of C5a in the C. trachomatis infection group significantly increased (15 min: P < 0.05; 30 min: P < 0.01; 60 min, 90 min, and 120 min: P < 0.0001) compared with that of the C5 control group. Moreover, the concentration of C5a gradually increased over the incubation time. In contrast, the concentration of C5a in the mock infection group was not statistically different (P > 0.05) to that of the C5 control group, congruent with Western blot results. This suggested that the process of incubation itself or the presence of HeLa229 cells lysates alone did not produce the C5 cleavage effect, and that cleavage of C5 observed in the C. trachomatis infection group was induced by chlamydial infection. The formation rates of C5a were calculated based the concentration of human C5 protein (400 nM) added into the co-incubation system. After 120 min of incubation, the formation rate of C5a was only 10.98 ± 0.54% for the C. trachomatis infection group.

To identify the amino acid sequence of the previously undescribed C5-derived ~38-kDa fragment encompassing C5a in Western blot analysis, C-terminal sequencing based on LC-MS/MS was performed. Excised bands from SDS-PAGE gels containing the ~38-kDa fragment were digested by trypsin, and half of the obtained peptides were further digested by Glu-C. The two groups of peptides were analyzed by LC-MS/MS and the obtained peptide information was correlated with the amino acid sequence of human C5 (NCBI search number: P01031.4) for comparison. Sequences of peptides identified by trypsin digestion or trypsin & Glu-C digestion are shown in Table 1 . The peptides closest to the C-terminal were both IPLDLVPK and corresponded to I963-K970 of C5. As the ~38-kDa fragment encompassing C5a (the N-terminal of C5 α-chain) was detected with the anti-C5a antibody, the sequence of the novel ~38-kDa fragment corresponded to T678-K970 of C5.

By combining the results from Western blot analysis, ELISA, and C-terminal sequencing based on LC/MS-MS, the unique cleavage mechanism of C5 induced by C. trachomatis infection could be predicted ( Figure 2A ). In the traditional C5 cleavage pattern, C4b2a3b or C3bBb3b, as C5 convertases, rapidly hydrolyze C5 at R751 in the α-chain of C5 and release C5a and C5b. However, in the distinct C5 cleavage pattern induced by C. trachomatis infection, the K970 site was first cleaved efficiently. During physiological conditions, the C5-derived ~38-kDa fragment detected in Western blot analysis and still connected with C5 α-chain and C5 β-chain by disulfide linkages, which resembles C5, is herein referred to as C5_Ct. The time at which the band corresponding to the C5-derived ~38-kDa fragment was detected by Western blot analysis was earlier than the C5a level elevated in the ELISA, suggesting that cleavage of the R751 site was more slowly induced by C. trachomatis infection compared with cleavage of the K970 site. After releasing C5a, C5_Ct was modified into a three-chained disulfide-linked molecule that resembled C5b, herein referred to as C5b_Ct.

A molecular dynamics simulation was conducted to predict the conformational change of C5 after the cleavage of the K970 site. The RMSD was stable after 40 ns and converged at 3.8 Å ( Figure 2B ). The superposition of C5 and C5_Ct ( Figure 2C ) showed that the overall conformational structure of C5_Ct did not undergo a significant change after breakage of the peptide bond between residues 970 and 971, which indicated that C5_Ct might be similar to C5 in the process of C5b_Ct-9 assembly. However, in the locality of peptide-bond cleavage between residues 970 and 971, the electrostatic surface was no longer as smooth as it was before the fracture ( Figures 2D, E ). The next part of the investigation therefore focused on whether this change made a difference to the lytic activity of C5b-9 and C5b_Ct-9.

During the genital infection of Chlamydia, lysis of infected host cells and excessive tissue damage might be associated with overactivation of the complement system, especially due to the enhanced lytic activity of MAC (i.e., C5b-9 and C5b_Ct-9). There were two possible causes for the enhanced overall MAC function during C. trachomatis infection: 1) C. trachomatis infection promoted the generation of MAC, resulting in the total number of MAC increasing. 2) The lytic ability of a single MAC enhanced after the modification from C5b-9 to C5b_Ct-9, due to structural changes. To verify these hypotheses, NHS was mixed with C. trachomatis-HeLa229 cell lysates or/and C5 convertase in a co-incubation system, and the lytic activity of these products was determined by chicken erythrocyte lysis assays ( Figure 3A ).

Firstly, evaluation of whether C. trachomatis infection directly increased the amount of MAC was conducted. The product of NHS and PBS incubation served as a control group to verify the background lytic activity of NHS. Compared with the control group, the product of NHS and C. trachomatis-HeLa229 cell lysate incubation was not statistically different after 15, 60, and 120 min, respectively (P > 0.05), which suggested that C. trachomatis infection failed to generate enough MAC, due to slow cleavage of the R751 site and lack of C5b or C5b_Ct.

Next, the difference in the lytic ability between C5b-9 and C5b_Ct-9 was determined. Excessive C5 convertase was used for cleavage at the R751 site that would release enough C5b to assemble C5b-9. Since C5 convertase C3bBb3b has a short half-life (1.5-3 min), CVF was mixed with NHS to assemble CVF,Bb, a more stable C5 convertase that has a longer half-life (~7 h) and similar kinetic properties to C3bBb3b with an indistinguishable Km and k_cat, for rapid cleavage of C5 at R751 (DiScipio et al., 1983; Rawal and Pangburn, 2000; Krisinger et al., 2012). As predicted, the product of NHS and CVF incubation showed a significantly stronger lytic activity than that of the control group at all incubation times (P < 0.001), and the lysis rate of erythrocytes was almost constant over the incubation period ( Figure 3B ). These findings suggested that CVF,Bb had cleaved most C5 into C5b in a short time and an abundance of C5b-9 was assembled. Incubation of C. trachomatis-HeLa229 cell lysates together with CVF and NHS slightly but statistically significantly increased the product lytic activity compared with the product of NHS and CVF incubation (P < 0.001). Taken together, these results revealed that C5b_Ct-9 exhibited more lytic activity than C5b-9.

To explore the correlation between the developmental cycle of the Chlamydia and the cleavage of C5, the C. trachomatis-HeLa229 cell lysates (MOI=1) collected at 0 h, 12 h, 24 h, 36h and 44 h post-infection were co-incubated with purified C5 (400 nM) for 120 min. The results ( Figure 4A ) showed that the lysates collected at 24 h post-infection yield a faint band of ~38-kDa fragment. The significant reduction of the bands of C5α chain demonstrated that C5 was largely cleaved by the lysates collected at 36 h and 44 h post-infection. Since each developmental cycle of C. trachomatis L2 is ~48 h, these results indicated that only the chlamydial infection in the middle and late stage of the developmental cycle had the cleavage effects on C5.

Chlamydia has a unique biphasic lifestyle involved in the alternate between two morphologically distinct forms, an infectious non-replicative elementary body (EB), and a replicative, non-infectious reticulate body (RB) (O’Connell and Ferone, 2016). After the invasion into host cells, EBs differentiate into RBs to grow and divide. When replication is completed, RBs convert back to EBs asynchronously in the middle and late stage of the developmental cycle. To determine whether chlamydial EBs was the effective component in the C. trachomatis-HeLa229 cell lysates that had directly cleavage activity on C5, purified chlamydial EBs, with the same infectivity of 10⁸ IFU/mL as the C. trachomatis-HeLa229 cell lysates (MOI=1) used in previous experiments described above, were co-incubated with purified C5 (400 nM) for 30 min. The result ( Figure 4B ) showed that EB failed to cause the cleavage of C5. Since the purification process of EBs removed host cell components and soluble proteins from lysates, this observation suggested that the unknown C3-independent C5 convertase induced by C. trachomatis infection was a chlamydial secretory protein or a newly protein derived from host cells, synthesized in the middle and late stage of chlamydial infection.

In order to further explore the relationship between the middle and late stage of chlamydial infection and the cleavage of C5, a model of penicillin-induced persistence of infection with C. trachomatis reported previously (Dille et al., 2014) was used. HeLa229 cells infected with C. trachomatis L2 (MOI=1) with or without addition of 100 U/mL Penicillin G. Lysates collected at 44 h post-infection were co-incubated with purified C5 (400 nM) for 120 min. The results ( Figure 4C ) showed that penicillin treatment significantly reduced the consumption of the band of C5α chain. The expression of CPAF was also reduced by penicillin treatment, which was consistent with the result reported by Dille (Dille et al., 2014). In this model, the invasion and stimulation of Chlamydia to host cells are mostly retained, but due to environmental stress, RBs convert to aberrant bodies instead of EBs. Thus, the result suggested the middle and late stage of chlamydial infection is closely related to C5 cleavage, and also demonstrated that the invasion of Chlamydia didn’t induce host cells to secrete the new protease that cleave C5.

To identify this unknown C3-independent C5 convertase occurring in the middle and late stage of chlamydial infection, a mass spectrometry-based label-free quantitative proteomics analysis of the lysates at 24 h post-infection was conducted. Compared with the mock infection group as a control, there were 57 human proteins and 259 chlamydial proteins only identified in the infection group but not the mock infection group (shown in Table S1 ). No known C5 convertases, such as conventional C5 convertases (C4b2a3b or C3bBb3b), thrombin, trypsin, kallikrein, plasmin, factor XIIa and so on, in the differentially expressed proteins. To our knowledge, all known C5 convertases are serine proteases (Rawal and Pangburn, 2001). According to annotated function in the protein database, three serine proteases in the differentially expressed proteins were identified as candidate C5 convertases: Tripeptidyl-peptidase 1 (TPP1), CPAF and cHtrA. Interestingly, the result also showed that Glia-derived nexin, a serine protease inhibitor derived from host cells, was induced by C. trachomatis infection, with inhibitory activity toward thrombin, trypsin, and urokinase that are all C3-independent C5 convertases. It seemed that the host response tended to inhibit rather than upregulate C5 convertase activity during the chlamydial infection.

Among 3 candidate C5 convertases identified by proteomics analysis, CPAF, a chlamydial secreted serine protease, with the potential to cleave or degrade a wide range of host proteins including complement factor C3 and factor B (Zhong, 2009; Yang et al., 2016), showed some correlations with C3-independent C5 activation in the experiments described above. On the one hand, several previous researches reported that although CPAF protein was detected inside inclusions as early as 12 h post-infection, clear CPAF secretion and activity were only detected 24 h after infection (Zhong et al., 2001; Huang et al., 2008). The expression profile of activated CPAF was consistent with the time points during the chlamydial infection that activated C5. On the other hand, Zhong et al. demonstrated that no CPAF or CPAF activity was contained in purified EBs (Zhong et al., 2001) that showed no cleavage effects on C5 in our experiment. Therefore, CPAF was considered the candidate with the highest priority for C3-independent C5 convertase during chlamydial infection.

To further investigate the correlation between CPAF and C5 activation, a C. trachomatis L2 CPAF-deficient (CPAF^−/−) strain was used for the preparation of C. trachomatis CPAF^−/−-HeLa229 cell lysates for use in the co-incubation system for C5 cleavage. The MOI was increased to 3 to ensure an adequate amount of CPAF^−/− strain was present in the co-incubation system due to CPAF mutants producing less infectious yield than wild-type C. trachomatis (Snavely et al., 2014). Western blot analysis ( Figure 5A ) revealed distinct bands corresponding to the C5-derived ~38-kDa fragment in the wild-type infection group after incubation for 5 and 15 min. The bands gradually became thinner with prolonged incubation time. C. trachomatis CPAF^−/−-HeLa229 cell lysates incubated with C5 did not produce the ~38-kDa fragment after incubation from 5 to 120 min. In both the wild-type group and the CPAF^−/− group, bands corresponding to MOMP were detected, which verified that the two co-incubation systems contained adequate amounts of C. trachomatis. In the CPAF^−/− group, there was no band corresponding to CPAF as expected. This combination of results indicated that CPAF participated in a C3-independent C5 activation during chlamydial infection, but not distinguished between the two possibilities: 1) CPAF directly cleaved C5. 2) An unknown protein derived from host cell upregulated due to CPAF, and then cleaved C5.

Lactacystin, as a CPAF inhibitor reported previously (Zhong et al., 2001; Huang et al., 2008), was pre-mixed with C. trachomatis-HeLa229 cell lysates followed by co-incubation with purified C5 protein. Western blot analysis ( Figure 5B ) showed that lactacystin had no significant inhibitory effect on C5 cleavage triggered by 1:1 dilution of C. trachomatis-HeLa229 cell lysates, but slightly reduced consumption of the band corresponding to the C5α chain (115 kDa) in the group of 1:2 dilution of C. trachomatis-HeLa229 cell lysates.

NM, a synthetic serine protease inhibitor with a good safety profile, is widely used to inhibit the activity of the complement system (Watkins et al., 1989; Schwertz et al., 2008). To evaluate whether NM could inhibit the C3-independent C5 activation triggered by chlamydial infection, NM (400 μM) was pre-mixed with different dilutions of C. trachomatis-HeLa229 cell lysates followed by co-incubation with purified C5 protein. Western blot analysis revealed that NM completely blocked the C3-independent C5 activation triggered by different dilutions of C. trachomatis-HeLa229 cell lysates ( Figure 6A ), suggesting that NM had an inhibitory effect on the C5 activation induced by chlamydial infection.

To further explore whether there was a reasonable conformation between NM and CPAF, a molecular docking with NM and CPAF was conducted. Molecular docking programs perform a search algorithm in which the conformation of the ligand is evaluated recursively until the convergence to the minimum energy is reached (Pagadala et al., 2017). The binding energy between NM and CPAF was calculated to predict their affinity. A binding energy lower than 0 indicates that two molecules combine spontaneously and that smaller binding energies lead to more stable conformations. The results showed that the binding energy between NM and CPAF was -7.34 kcal/mol, suggesting a strong and stable binding between NM and CPAF. The NM-CPAF complex obtained from precise docking in Autodock and visualized in PyMOL, unraveling the in silico binding inhibitory mechanisms. According to previous research reports, the active site of CPAF is a water-mediated catalytic triad containing His105, Ser499 and Glu558 (Huang et al., 2008). NM specifically bound to the active site of CPAF ( Figures 6B, C ). The hydrophilic parts at both ends of NM formed hydrogen bonds with Ser377, Ser499, and Ala524 ( Figure 6D ), which played important roles in stabilizing the conformation of NM-CPAF complex. A stable occupation of NM in the active site of CPAF near two important catalytic residues, His105 and Ser499 ( Figure 6D ), may explain the inhibition of NM on CPAF. Due to the strong interaction of NM with active site of CPAF, NM might be considered as an effective CPAF inhibitor. Combined with the fact that NM inhibited the cleavage of C5 induced by C. trachomatis-HeLa cells lysates, these docking results also enhance the likelihood that CPAF interacts directly with C5.