Francisella tularensis virulence relies on a conserved putative catalytic triad within the Type VI secretion system component PdpC
Jeanette E Bröms, Igor Golovliov, Athar Alam, Shaochun Zhu, André Mateus, Thomas Henry, Anders Sjöstedt
TL;DR
This study finds a key amino acid triad in the PdpC protein of Francisella tularensis that is crucial for its ability to cause disease but not for protein secretion.
Contribution
The study identifies a conserved catalytic triad in PdpC that is essential for Francisella virulence but not secretion.
Findings
Mutations in the PdpC triad impaired phagosomal escape and intracellular replication.
The triad is essential for PdpC effector function but not for T6SS-dependent secretion.
Equivalent mutations in F. novicida confirmed the triad's importance.
Abstract
Gram-negative bacteria utilize type VI secretion systems (T6SS) for microbial competition and host interaction. While most pathogens rely on the canonical T6SSi, Francisella species uniquely possess T6SSii. The highly virulent human pathogen Francisella tularensis harbors a distinct T6SSii variant that includes pdpC, encoding a putative effector protein. Bioinformatic analysis revealed a conserved amino acid triad in PdpC, homologous to motifs found in Make Caterpillars Floppy family toxins. To investigate the functional relevance of this triad, site-directed mutagenesis was performed in the live vaccine strain of F. tularensis, substituting each residue with alanine. Mutants showed impaired phagosomal escape, reduced intracellular replication, and marked attenuation in the mouse infection model. Equivalent mutations introduced into F. novicida, a model for T6SS-mediated secretion,…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 10
Figure 11- —French National Research Agency10.13039/501100001665
- —Swedish Research Council10.13039/501100004359
- —Kempestiftelserna10.13039/501100007067
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsBacillus and Francisella bacterial research · Yersinia bacterium, plague, ectoparasites research · Vibrio bacteria research studies
Introduction
Gram-negative bacteria utilize a variety of secretion systems to interact with target cells (Green and Mecsas 2016). Among these, the most ubiquitous system is the type VI secretion system (T6SS), which exists in nearly all Gram-negative pathogens and are present in more than 25% of all bacterial genomes (Allsopp and Bernal 2023). Individual bacteria can harbor between one and six distinct T6SS clusters, each typically composed of a core cluster of around a dozen genes essential for assembly and function, though most clusters contain 15–20 genes in total (Hernandez et al. 2020). The accessory genes vary considerably between species and clusters, with many encoding effector molecules (Allsopp and Bernal 2023). The effector functions are highly diverse and encompass both potent antibacterial activity as well as distinct activities against target eukaryotic cells, including cell lysis, cytoskeleton immobilization, dampening of the inflammatory response, or manipulation of intracellular signaling pathways (Hernandez et al. 2020). Mechanistically, these effects are achieved via membrane disruption, inhibition of cell wall synthesis, nucleic acid degradation, or inhibition of protein synthesis. Some substrates do not act on cellular targets, but are instead secreted to sequester essential nutrients from the environment (Yang et al. 2021). Thus, compared to other virulence mechanisms, the T6SS confers an unprecedented versatile machinery that allows the pathogen to effectively outcompete microbial rivals, colonize hosts, modulate immune responses, and rapidly adapt to or withstand adverse environmental conditions.
The T6SS can be classified into major groups based on phylogenetic relationships of core components and the conservation of accessory proteins. Most common Gram-negative pathogens possess T6SSs of the T6SS^i^ type, which is further divided into six subtypes (Russell et al. 2014, Li et al. 2015). Some T6SSs, however, are more divergent and form separate types. An example is T6SS^ii^, found exclusively in the genus Francisella (Bröms et al. 2010, Russell et al. 2014). T6SS^ii^ represents a phylogenetic outlier and comprises the highly virulent human pathogen Francisella tularensis, as well as a broad range of species adapted to diverse environments—ranging from aquatic pathogens and tick endosymbionts to opportunistic pathogens affecting immunocompromized individuals (Challacombe et al. 2017, Kumar et al. 2020). Francisella tularensis is a facultative intracellular bacterium, a lifestyle that appears common across the genus Francisella (Sjöstedt 2006, Degabriel et al. 2023). A key feature of the intracellular lifestyle is the presence of a T6SS (Spidlova and Stulik 2017, Clemens et al. 2018, Degabriel et al. 2023). The secretion system is encoded by the so-called Francisella pathogenicity island (FPI) and comprises a cluster of 16–19 genes (Clemens et al. 2018, Degabriel et al. 2023). The core genes are present in essentially all Francisella genomes analyzed to date (Kumar et al. 2020). Based on FPI gene content and organization, three major T6SS groups can be identified. The first includes all F. tularensis subspecies and rare human pathogens, such as Francisella novicida (Kumar et al. 2020). The second group includes isolates from a wide variety of habitats, e.g. symbiotic bacteria, aquatic bacteria, and bacteria from cooling water systems. The third group includes isolates causing disease exclusively in immunocompromized individuals. The most notable difference between the groups is the absence of the pdpC and pdpE genes in the latter two groups (Kumar et al. 2020). To date, no homologs of these genes have been identified outside the Francisella genus.
PdpC has been much studied and found to be important for the intracellular life style and virulence of F. tularensis, whereas no function has been assigned to PdpE (Bröms et al. 2011, Lindgren et al. 2013, Long et al. 2013, Lindgren et al. 2014, Uda et al. 2014, Eshraghi et al. 2016, Ozanic et al. 2016, Brodmann et al. 2021, Cantlay et al. 2022). One rationale to study PdpC was the finding that the attenuated phenotype of a spontaneous PdpC-truncated mutant in the prototypic F. tularensis subspecies tularensis strain SCHU S4 could be reversed by complementation with pdpC (Lindgren et al. 2014). Similarly, another spontaneously attenuated strain of SCHU S4 regained virulence after serial passage in mice. The only difference between the attenuated and virulent isolates was a single nucleotide substitution in pdpC, which restored expression of the full-length protein and, consequently virulence (Uda et al. 2014). Consistent with these findings, a pdpC mutant of SCHU S4 showed much impaired intramacrophage growth and was avirulent in mice (Long et al. 2013). Further supporting the importance of PdpC, a ΔpdpC mutant of the live vaccine strain (LVS) of F. tularensis exhibited no intramacrophage replication and was markedly attenuated in the mouse model (Lindgren et al. 2013). Likewise, a pdpC deletion mutant of F. novicida was highly attenuated in mice, although it demonstrated only marginal impairment of intracellular growth in murine cells (Chou et al. 2013).
While critical for virulence in several Francisella subspecies, it was demonstrated that PdpC is not required for T6SS assembly in F. novicida (Brodmann et al. 2021). Mass-spectrometry studies on F. novicida identified PdpC as one of five FPI-encoded proteins secreted via the T6SS. PdpC was hypothesized to be an effector rather than a structural component, since disruption of pdpC did not affect the secretion of the other four proteins (Eshraghi et al. 2016). Remarkably, in the presence of PdpC, the remaining four putative effectors were found to be redundant with regard to intramacrophage growth in human cells (Eshraghi et al. 2016). Despite these extensive investigations, the molecular function of PdpC remains unknown. To address this, we performed a bioinformatic analysis and identified a conserved amino acid triad within PdpC, with similarity to a catalytic triad present in toxins of the Make Caterpillars Floppy (MCF) family (Agarwal et al. 2015). We provide substantial experimental evidence demonstrating that this triad of PdpC is essential for the virulence of F. tularensis.
Materials and methods
Bacterial strains, plasmids, and growth conditions
The bacterial strains and plasmids utilized in this study are detailed in Table S1. Escherichia coli strains were cultured at 37°C in Luria–Bertani (LB) broth or on LB agar plates. Francisella tularensis LVS and F. novicida U112 were grown either in liquid Chamberlain’s medium, or on modified GC (MC) agar at the same temperature. For sucrose-based selection, both F. tularensis LVS and F. novicida U112 were plated on modified Mueller–Hinton (MH) agar supplemented with 10% sucrose. When necessary, antibiotics were added at the following concentrations: carbenicillin (Cb; 100 µg/ml), tetracycline (Tet; 10 µg/ml), kanamycin (Km; 50 µg/ml for E. coli, 10 µg/ml for F. tularensis or F. novicida), or chloramphenicol (Cm; 25 µg/ml for E. coli, 2.5 µg/ml for F. tularensis and 8 µg/ml for F. novicida).
Construction of amino acid-substituted mutants in F. tularensis LVS
All PCR products were initially cloned into the pCR4-TOPO TA cloning vector (450 071, Thermo Fischer scientific) to facilitate sequencing (Eurofins MWG Operon) prior to downstream cloning steps. Single amino acid substitutions within pdpC were introduced in cis on the LVS chromosome as described below. An LVS-derived insert (6.6 kb) used for in cis complementation of LVS ΔpdpC (Lindgren et al. 2013) was cloned in several steps into a modified pBluescript KS+ (Stratagene). The MCC (multiple cloning cassette) of this vector had been altered by replacing the SacI site with XhoI via the QuikChange site-directed mutagenesis kit (200 518, Agilent Technologies) using primers pBSmut_F2: 5´-GAC TCA CTA TAG GGC GAA TTG GAT CTC GAG CGC GGT GGC GGC CGC TCT AGA AC-3´ and pBSmut_R2: 5´-GTT CTA GAG CGG CCG CCA CCG CGC TCG AGA TCC AAT TCG CCC TAT AGT GAG TC-3´. A 120-bp PCR fragment was then amplified from the modified vector using primers pBS mutSacI_F: 5´-CTC GAG CGA GCT CGC GGC CGC TCT AGA ACT AGT GG-3´ (XhoI) and pBS_R: 5´-AGG GAA CAA AAG CTG GGT ACC-3´ (KpnI). Following XhoI/KpnI digestion, this fragment was introduced into the modified vector to reintroduce a SacI-site downstream of XhoI, resulting in the modified pBluescript KS (pJEB1193), which harbored the MCC restriction sites: XhoI-SacI-BamHI-XhoI-ApaI-KpnI.
Next, a fragment incorporating the internal BamHI-site of pdpC and downstream 2881 bp sequence was PCR amplified using primers PdpC_CD_F: 5´-GGA TCC TTG TAT AAC CTA TTA TCA AC-3´ (BamHI) and PdpC_CD_R: 5´-GGG CCC TAA TAG GAC TGT TTC TGA ACT AAA G-3´ (ApaI) and cloned as a BamHI/ApaI fragment into pJEB1193. Similarly, a fragment incorporating the internal SacI site of pdpC and 2,793 bp upstream sequence was amplified using primers PdpC_AB_F: 5´-CTC GAG AAA ACT CTT ATT GAA AAA ATT GAA GT-3´ (XhoI) and PdpC_AB_R2: 5´-GAG CTC AAG AAG CCA GGA AG-3´ (SacI) and subsequently introduced as a XhoI/SacI fragment. The resulting construct, pJEB1196, contained a copy of pdpC lacking the 901 bp region between the endogenous SacI and BamHI sites. To restore the full-length gene, the deleted region was replaced with either the wild-type (WT) pdpC sequence or variants containing single amino acid substitutions, generated by overlap PCR (for primers, see Table S2). These fragments were introduced into pJEB1196 using SacI/BamHI cloning. The complete pdpC fragments, including flanking regions, were then excised with XhoI and ApaI and ligated into XhoI/ApaI-cut pDM4 (Milton et al. 1996). The resulting suicide vectors were introduced into S17-1λpir by electroporation and used for conjugation with LVS ΔpdpC. Allelic exchange enabled the replacement of one of the deleted pdpC copies with either the WT or mutant version. To replace both copies, the procedure was repeated. In all cases, successful recombination was confirmed by PCR screening.
Construction of ΔpdpC and amino acid-substituted mutants in F. novicida
Primer sets used for generating the pdpC deletion mutant in F. novicida U112 are listed in Table S2. Amplified fragments were first inserted into the pCR4-TOPO TA vector before proceeding with further cloning. To construct the pdpC deletion mutant, 1240 bp regions upstream and downstream of the target gene were sequentially cloned into pBluescript SK+ (Stratagene) using XhoI/BamHI and BamHI/SacI restriction sites, respectively. The resulting vector, pJEB1210, contained a 2510 bp fragment representing the PdpC Δ6–1321 deletion, with flanking regions joined at a BamHI site. The fragment was then cloned into XhoI/SacI-digested pDM4, generating suicide vector pJEB1221 (pΔpdpC). The vector was introduced into S17-1λpir by electroporation to facilitate conjugation with F. novicida U112, using sucrose selection to promote allelic exchange. Homologous recombination between the plasmid and chromosome enabled the generation of the ΔpdpC mutant. Successful recombination events were confirmed by PCR screening.
To generate single amino acid substitutions in F. novicida U112, the following strategy was used. An integration cassette designated gro_Km, was synthesized by GenScript, and included the F. tularensis gro promoter and a F. tularensis codon usage-optimized Km resistance gene derived from Tn5. The cassette was flanked with SalI and SacI restriction sites at the 5 and 3 ends, respectively, and cloned into pBluescript SK+ (pBSSK+) resulting in the construct denoted pBSSK+/gro_Km. Subsequently, a BglII*/PstI fragment containing the sacB gene was excised from pMP-719 (LoVullo et al. 2009), blunt-ended and inserted downstream of the Km^R^ gene into SmaI-digested pBSSK+/gro_Km. The resulting gro_Km_sacB cassette was then cloned to BamHI-digested and blunt-end treated pJEB1210, carrying the PdpC Δ6–1321 deletion (above). The resulting plasmid, pBSSK+/ΔpdpC: gro_Km/sacB*, was introduced to F. novicida by electroporation and after a double crossover event, Km^R^ clones were selected, generating the F. novicida ΔpdpC: gro_Km_sacB strain.
Single amino acid substitutions were introduced using overlapping PCR. Internal forward and reverse primers were designed with overlapping sequences containing the desired mutations near the center. External primers were used to amplify ∼900 bp regions upstream and downstream of the pdpC deletion. After purification, the overlapping fragments were combined in a second PCR using only the external primers to generate full-length amplicons. For WT pdpC, the same external primers were used with chromosomal DNA as template. After purification, PCR products were introduced to the F. novicida ΔpdpC: gro_Km_sacB strain by electroporation and transformants were selected on MH agar supplemented with 10% sucrose. To confirm the loss of the gro_Km_sacB cassette, sucrose-resistant clones were plated on MH agar with or without Km. Clones that were both sucrose-resistant and Km-sensitive were subjected to sequencing to verify the presence of the intended amino acid substitutions.
Quantitative real-time PCR
The procedure for bacterial RNA extraction, cDNA synthesis, and qPCR were performed as previously described (Bröms et al. 2009). Primer sequences are included in Table S2. Negative controls were included by omitting either the template or reverse transcriptase during cDNA synthesis. Each reaction was carried out in triplicate using two independent RNA preparations. Data acquisition was performed using the 7900HT Sequence Detection System (Applied Biosystems) and its associated software. Gene expression levels were normalized to the F. tularensis housekeeping gene tul4 (FTL0421) and compared to the corresponding genes in LVS. Relative expression was calculated using the comparative Ct (ΔΔCt) method, and fold changes were determined using the 2^-ΔΔCt formula. Statistical significance was assessed using paired two-tailed Student´s t-tests.
Western blot analysis
Unless otherwise specified, bacterial lysates were prepared from cultures grown on modified GC agar that were mixed with Laemmli sample buffer, boiled, and separated on 12%–15% sodium dodecyl sulfate (SDS)-polyacrylamide gels (Bio-Rad). Proteins were then transferred to nitrocellulose membranes (1 620 115, Bio-Rad) using a Trans-Blot Turbo Transfer System (Bio-Rad). The membranes were probed with the following primary antibodies: mouse anti-IglC (1:2000), and rabbit anti-PdpC (1:4000) (NR-3196 and NR-4379 respectively; BEI Resources). Secondary horseradish peroxidase (HRP)-conjugated antibodies were goat anti-mouse (1:2000) and donkey anti-rabbit (1:10 000) (S2005 Santa Cruz Biotechnology and NA934V GE Healthcare, respectively). Detection was performed using the Enhanced Chemiluminescence (ECL) system (RPN 2232, GE Healthcare).
In vitro secretion assay
F. novicida strains, grown overnight on MC agar, were subcultured into tryptic soy broth (TSB) medium supplemented with 0.1% glucose, 0.1% cysteine, and 5% KCl to an initial OD_600_ of 0.15. Cultures were incubated until reaching an OD_600_ of 1.5, at which point both cell pellets and filter-sterilized supernatant containing secreted proteins were collected. Supernatants (approximately 4 ml) were precipitated using 20% TCA on ice for 30 min, then centrifuged at 15 000 × g for 15 min at 4°C. The resulting precipitate was used for either Western blot or mass spectrometry analysis. For Western blot analysis, the precipitate was washed with ice-cold acetone, dried, and then dissolved in SDS-PAGE sample loading buffer.
Sample preparation for mass spectrometry
Peptide digestion was performed using a modified SP3 protocol (Hughes et al. 2014, Hughes et al. 2019). Briefly, precipitated proteins were resuspended in lysis buffer (2% SDS, 20 mM Tris(2-carboxyethyl)phosphine (TCEP) hydrochloride) and boiled at 95°C for 10 min. SpeedBeads magnetic carboxylate modified particles (1:1 v/v mix of hydrophilic and hydrophobic types; GE45152105050250 and GE65152105050250; Sigma Aldrich) were washed four times using liquid chromatography mass spectrometry (LC-MS) grade water and added to each sample in binding buffer (f.c. 50% ethanol, 2.5% formic acid). Samples were incubated at RT with shaking (500 rpm, 15 min), then transferred to a 0.22 µm filter plate (MSGVN2210, Sigma Aldrich). Unbound material was removed by centrifugation at 1000 × g, and beads retained on the filter were washed four times with 70% ethanol. Trypsin was added in digestion solution (100 mM HEPES pH7.5, 5 mM chloroacetamide, 1.2 mM TCEP) at a ratio of 1 µg trypsin per 25 µg protein to each sample. Digestion was carried out overnight at RT with shaking (500 rpm). Peptides were collected by centrifugation at 1000 × g, and remaining peptides were eluted from the beads with 10 µl of 2% dimethyl sulfoxide (DMSO) and pooled. Peptides were desalted by use of an Oasis HLB plate (186001828BA, Waters) using the manufacturer’s protocol and then dried.
Liquid chromatography tandem mass spectrometry (LC-MS/MS)
Dried peptides were dissolved with 0.1% formic acid in LC-MS water. One µg of peptides per sample was analyzed by LC-MS/MS using a Vanquish Neo UHPLC system (Thermo Scientific). Peptides were first trapped on a PEPMAP NEO C18 trap column (5 µm, 300 µm × 5 mm; Thermo Scientific) for desalting and concentration, followed by separation on a nano-EaseTM M/Z HSS C18 T3 analytical column (100 Å, 1.8 µm, 75 µm × 250 mm; Waters). The separation and elution were carried out over 90 min using a gradient starting from mobile phase A (water with 0.1% formic acid) to 8% mobile phase B (80% acetonitrile with 0.1% formic acid) over 4 min, followed by an increase to 27% B over 60 min, and then to 40% B over the next 13 min. The gradient was then increased to 80% B for 0.1 min and held for 4 min, followed by a decrease to 2% B in 0.5 min. Finally, column equilibration was performed.
Data were acquired on an Orbitrap Exploris 480 mass spectrometer (Thermo Scientific) operated in a data-dependent acquisition (DDA) mode. The following parameters were used: Survey scans were recorded over a mass range of m/z 375–1500 at a resolution of 120 000. The RF lens was set to 40%, and the normalized automatic gain control (AGC) target was 300%. A maximum cycle time of 2 s was applied to control the number of precursors selected for MS/MS analysis. Precursor ions with charge states of 2–6 charges were selected for fragmentation. Dynamic exclusion was enabled to prevent repeated selection of the same precursors within 35 s. MS scans were acquired at a resolution of 15 000 (at m/z 200), with the AGC target set to auto. The isolation window was set to 1.4 m/z. Fragmentation was performed using Higher Energy Collisional Dissociation (HCD) with a normalized collision energy (NCE) of 30. Isotopic peaks were excluded from MS/MS analysis.
MS/MS data analysis
Raw MS data was searched against the F. novicida U112 UniProt FASTA database (proteome identifier: UP000000762) using FragPipe (version 18). Label-free quantification was achieved using LFQ-MBR workflow. Proteins identified as contaminants and decoys were excluded from further analysis. Statistical analysis was performed using the online software Fragpipe-Analyst (Hsiao et al. 2024). Default settings were applied, except that variance-stabilizing normalization was selected for normalization. Perseus-type imputation was used to handle missing values and false discovery rate (FDR) correction was achieved using the Benjamini–Hochberg method, designated as the adjusted P-value. For visualization, volcano plots were generated using R (version 4.2.2).
The MS data has been deposited to the ProteomeXchange Consortium via the PRIDE partner repository (Perez-Riverol et al. 2022), with the dataset identifier PXD059583.
Cultivation and infection of macrophages
J774 macrophages (ATCC TIB-67) were maintained in DMEM (GIBCO BRL) supplemented with 10% heat-inactivated FBS (GIBCO). One day before infection, cells were seeded in tissue culture plates using DMEM containing 10% FBS. Next day, cells were washed, fresh medium was added, and cells were allowed to recover for 30 min before infection. All infections were performed using a multiplicity of infection (MOI) of 200, except for the transmission electron microscopy (TEM) analysis (MOI 1000).
Intracellular replication in macrophages
To evaluate intracellular growth of F. tularensis or F. novicida, macrophages were infected for 2 h, followed by three washes and treatment with 5 µg/ml gentamicin for 30 min to eliminate extracellular bacteria (defined as time zero). At 0, 24, and 48 h post-infection, the macrophage monolayers were lysed using PBS containing 0.1% deoxycholate. The lysates were serially diluted and plated on modified GC-agar to determine viable counts. Statistical comparisons of bacterial growth relative to LVS or F. novicida U112 were performed using a two-sided t-test with equal variance.
LDH release assay
The lactate dehydrogenase (LDH) release assay was conducted as described previously (Bröms et al. 2011). Cells were infected as outlined in “Intracellular Replication in Macrophages” (above), and culture supernatants were collected at 0, 24, and 48 h post-infection. LDH levels in the supernatants were measured to assess cytopathogenicity. Results are shown as means ± SD from three or four wells in one representative experiment. Uninfected cells lysed in PBS with 0.1% deoxycholate served as a positive control, with this value considered 100% lysis. Sample absorbance was expressed as a percentage of the control value.
Intracellular immunofluorescence assay
To evaluate phagosomal escape, GFP-expressing F. tularensis was used for cell infections, as described previously (Bröms et al. 2011). Cells were stained for the LAMP-1 glycoprotein, and colocalization of GFP-labeled F. tularensis and LAMP-1 was analyzed using an epifluorescence microscope (Axioskop2; Carl Zeiss MicroImaging GmbH). In two separate experiments, 50 bacteria per slide were scored from a total of five glass slides per strain.
Transmission electron microscopy
Infection and sample preparation for TEM have been described in detail elsewhere (Bröms et al. 2011). Sections were examined using a JEOL JEM 1230 Transmission Electron Microscope (Jeol Ltd). Membrane integrity was assessed by analyzing at least 100 bacteria from two different sections. Bacteria were categorized based on the condition of the phagosomal membrane: (i) intact, (ii) slightly damaged (< 50% membrane integrity affected), (iii) highly damaged (> 50% membrane integrity affected), or (iv) no residual membrane.
Mouse infection
For infection experiments, mice obtained from Charles River, Germany were used. To evaluate the virulence of each strain, female C57BL/6 mice (n = 6) were infected subcutaneously using an inoculum of 50 μl of saline. Previously published data has indicated that a sample size of six animals/group was sufficient to determine statistical differences (Forslund et al. 2006). There was a maximum of four mice per cage. The experiments were not blinded. The order of treatments was randomized. To confirm the infectious dose, aliquots of the diluted suspensions were plated on GC-agar to determine the number of CFU injected. Mice were observed twice daily for symptoms of severe infection. Animals exhibiting irreversible signs of morbidity were humanely euthanized by CO_2_ asphyxiation, in accordance with local guidelines adapted from the NIH´s “Guideline for Pain and/or Distress in Laboratory Animals” (NIH 2023). Based on our experience, these mice were at most 24 h away from death, and the time to death was estimated under this assumption. For the determination of the mortality, all data points were included. All animal experiments were approved by the Local Ethical Committee on Laboratory Animals in Umeå, Sweden (A36-19) and adhere to the ARRIVE Essential guidelines.
Bioinformatic analyses
PSI-Blast performed after exclusion of Thiothrix/Francisella group (taxid:72 273) sequences, was utilized to identify regions in PdpC homologous to protein regions of other species. A region from a Legionella gratiana protein (WP_058499943.1) was further analyzed by the iterative search algorithm Jackhmmer (Finn et al. 2015) to identify distant homologs and conserved domains across bacterial taxa. The C-terminus including the kinase domain of the L. gratiana protein (starting at residue 1253) was excluded from this analysis. In addition, the full-length PdpC sequence was analyzed using SMART (Simple Modular Architecture Research Tool) to predict domain architecture and Phyre to assess potential structural homology and secondary structure elements.
Structural modeling and mutational impact analysis
The three-dimensional structure of the PdpC protein from F. tularensis LVS was predicted using AlphaFold2 (Jumper et al. 2021). The full-length amino acid sequence was submitted to the AlphaFold-pipeline, and the best predicted structure (rank-1) based on the predicted Local Distance Difference Test (pLDDT) score was selected for further analysis. To assess the impact of specific point mutations on protein stability and flexibility, the wild-type structure was modified using DynaMut and DynaMut2 (Rodrigues et al. 2018, Rodrigues et al. 2021). These tools estimate the change in Gibbs free energy (ΔΔG) to predict the stabilizing or destabilizing effect of each point mutation (Rodrigues et al. 2018). The Dynamut´s ENCoM-based approach was used to calculate the change in vibrational entropy (ΔΔSVib) to predict differences in protein flexibility. PDB files were visualized using the UCSF ChimeraX program (Meng et al. 2023).
Results
Bioinformatic identification of an MCF-like domain in PdpC
Several forms of bioinformatic analyses were conducted to gain insight into the function of PdpC. Whereas the Smart and Phyre tools revealed no, or only low-confidence hits, PSI-BLAST identified a region within PdpC with distinct similarity to a hypothetical Legionella gratiana protein (WP_058499943.1) (Fig. 1). Using the L. gratiana protein as a reference in a Jackhmmer search, we searched for more distantly related homologs to explore whether the shared region is part of a broader, evolutionarily conserved domain. The results revealed a domain, amino acids 791–1058 of the L. gratiana protein, with homology to toxins, effectors, and uncharacterized proteins and this region is conserved in numerous other species (Fig. 2). The homologous regions all correspond to the MCF (Makes Caterpillars Floppy) domain, which is characterized by a consensus sequence denoted MCF-SHE. This includes a conserved catalytic triad, postulated to confer protease activity (Agarwal et al. 2015). This triad is conserved not only in PdpC and the L. gratiana protein, but also in the broader set of the partially homologous proteins shown in Fig. 2.
PSI-BLAST alignment between Francisella PdpC and a protein from Legionella gratiana (ID: WP_058499943.1). The alignment highlights a central region corresponding to residues 791–1109 of the 1741 amino acid L. gratiana protein, which share homology with a central region (residues 445–779) of the 1328 amino acid Francisella PdpC protein. The kinase domain of WP_058499943.1 is located at the C-terminus (residues 1253–1504) and is therefore not part of the alignment. Gaps introduced to optimize the alignment are indicated by dashes. The alignment shows 26% sequence identity and an E-value of 1e-08, indicating statistically significant similarity. The residues of the putative catalytic triad in PdpC are indicated by boxes.
Multiple sequence alignment of proteins containing the MCF1-SHE domain. Proteins are labeled by species name. Columns with fully conserved serine (S), histidine (H), or glutamic acid (E) residues are highlighted in red. Conserved substitutions are color-coded based on physicochemical properties: green for hydrophobic residues, bright blue for negatively charged residues, and dull blue for small alcoholic residues. Alignment was performed using Clustal Omega and visualized with MView. Residues mutated in this study are indicated with asterisks.
AlphaFold2 modeling of PdpC demonstrates low global and local reliability
As no crystal structure of PdpC is publicly available, a three-dimensional model of PdpC from F. tularensis LVS was generated using AlphaFold2 (Fig. 3A). The protein comprises a single polypeptide chain of 1 328 amino acids with a predicted globular architecture. The N-terminal region (residues 1–162) features beta sheets, while the rest of the protein is dominated by alpha helices and loops. Per-residue confidence was assessed using pLDDT scores, visualized as a color-coded 3D model (Fig. S1). Most regions displayed very low confidence. To further evaluate global reliability, we examined confidence metrics: the best-ranked model had a pTM (predicted Template Modelling) score of 0.29 and a PAE (maximum predicted alignment error) of 31.8 Å, reflecting uncertainty in the relative orientation of structural regions (Fig. S2). Overall, the poor prediction quality likely results from the absence of homologous templates in the Protein Data Bank, highlighting the uniqueness of the PdpC protein. Despite this limitation, a close-up view of the spatial arrangement of the conserved amino. i.e. Gly-662, His-705, and Thr-710, revealed that they are positioned in close proximity (Fig. 3B), consistent with the presence of a functional catalytical triad.
Predicted tertiary structure of PdpC from F. tularensis LVS generated by AlphaFold. (A) The LVS PdpC protein sequence (accession number: XWT32315) was used to generate an AlphaFold2 structure model using ColabFold. Molecular visualization was performed using UCSF ChimeraX. The structural prediction with highest score is shown. (B) The amino acid residues constituting the putative catalytic triad are highlighted, indicating their spatial arrangement within the folded structure.
Generation and expression of amino acid-specific pdpC mutants of LVS
To assess the functional importance of the three conserved amino acids in PdpC, we generated alanine-substitution mutants in cis, a strategy that preserves native expression and regulation. Alanine is widely used for functional studies because it is non-bulky, chemically inert, and mimics the secondary structure preferences of many amino acids (Morrison and Weiss 2001).
LVS (F. tularensis subsp. holarctica) was utilized as host strain since it carries duplicated FPI copies, essentially identical to those in virulent F. tularensis strains, enabling biologically relevant comparisons. Four complemented strains were successfully obtained in the LVS ΔpdpC mutant: ΔpdpC/WT, ΔpdpC/S661A, ΔpdpC/H709A, and ΔpdpC/E713A. Since in cis complementation is challenging and random with regard to which copy is being replaced, we only managed to retrieve one strain with both deletion copies replaced by the desired pdpC mutation, resulting in strain ΔpdpC/H709A 2c (Table S1). This strain was tested together with the single copy H709A strain in all subsequent assays and they showed identical phenotypes (data not shown).
To further analyze the functional role of the region demonstrating similarities to MCF proteins, an additional set of pdpC mutants was created in cis: ΔpdpC/G662A, ΔpdpC/H705A, and ΔpdpC/T710A. Gly-662, highly conserved among MCF proteins and present in PdpC, has been implicated in the catalytic activity of VopK, a Type III effector protein of Vibrio cholerae (Bankapalli et al. 2015), and is located near the serine residue of the putative catalytic triad (Fig. 2). His-705, in contrast, is a non-conserved residue close to the triad (Fig. 2) and was included to test the hypothesis that variable residues are less likely to be critical for protein function. Thr-710 represents a semi-conserved amino acid; while serine is most common at this position in related proteins, threonine occurs in some cases, including PdpC (Fig. 2).
RT-PCR analysis confirmed the absence of pdpC transcripts in the deletion mutant and expression in all complemented strains. The strain with two copies of the H709A variant exhibited expression levels similar to LVS, whereas strains with single-copy point mutations showed approximately two- to three-fold lower expression, comparable to the WT-complemented strain (data not shown). At the protein level, all site-directed mutants produced PdpC at levels comparable to the WT-complemented strain, with two exceptions: the E713A mutant exhibited markedly reduced protein levels, while the H709A 2c strain showed elevated levels similar to those observed in the parental LVS strain (Fig. S3). This is likely explained by the presence of two copies of the mutated pdpC gene, resulting in an increased gene dosage.
Impact of substitution mutations on the PdpC structure
To assess whether the catalytic triad point mutations affected protein stability and flexibility, we utilized DynaMut, a structure-based prediction tool (Rodrigues et al. 2018), employing the AlphaFold2-predicted PdpC structure (Fig. 3A). The predicted changes in stability (ΔΔG) for the S661A, H709A and E713A mutants were −0.0, -−0.12, and −0.1 kcal·mol⁻¹, respectively, suggesting potential mild destabilization (Table S3). In contrast, the predicted changes in vibrational entropy (ΔΔSvib) for S661A, H709A, and E713A were 0.121, 1.240, and 0.332 kcal·mol⁻¹·K⁻¹, respectively, indicating increased molecular flexibility in the mutant proteins compared to the wild-type protein (Table S3). The S661A mutation, residing in a loop region, had minimal structural impact and did not alter Ser-661 interactions with neighboring residues (Fig. 4A). In contrast, the H709A mutation resulted in significant rearrangement of local interactions, including the loss of two polar contacts and the formation of a new hydrogen bond with Ser-706, the loss of two polar, one van der Waals, and one hydrophobic interaction with Met-703, and the formation of a new hydrophobic interaction with Glu-713. The disruption of polar and aromatic interactions likely reduces structural rigidity, potentially destabilizing the local fold and contributing to the increased predicted flexibility. Furthermore, the mutation may abolish potential catalytic or ligand-binding functions of His-709 (Table S3, Fig. 4B). Similarly, the Glu-713 to Ala substitution caused a reorganization of local interactions. This included the formation of a new hydrophobic interaction with His-709, accompanied by the loss of all three polar interactions with it. A new polar interaction with Met-716 was also observed. Overall, this mutation reduces the polar bonding capacity and alters the local interaction landscape, potentially increasing flexibility, although to a lesser extent than the H709A mutant (Table S3, Fig. 4C).
Structural analysis of PdpC point mutants from F. tularensis LVS using DynaMut2. Close-up views show residue interactions in the predicted structure, before (left) and after (right) Alanine-substitution for each of the six mutants (A–F). Color-coded dashes indicate polar bonds (orange), hydrophobic interactions (green), Van der Waals forces (blue), hydrogen bonds (red) and steric clashes (pink). Atom coloring: oxygen (red), nitrogen (blue), sulfur (yellow), carbon (gray).
The predicted changes in stability (ΔΔG) for two of the mutants outside the catalytic triad, i.e. G662A and T710A, were 0.568 and 0.736 kcal·mol⁻¹, respectively, suggesting a potential mild stabilizing effect (Table S3). Additionally, the predicted changes in vibrational entropy (ΔΔSvib) for G662A and T710A were −0.129 and −0.471 kcal·mol⁻¹·K⁻¹, respectively, indicating reduced molecular flexibility in these mutant proteins. In contrast, H705A exhibited a different pattern. Both its ΔΔG (0.146 kcal·mol⁻¹) and ΔΔSvib (1.815 kcal·mol⁻¹·K⁻¹) were positive, indicating increased stability and molecular flexibility (Table S3). This is somewhat atypical; generally, a positive ΔΔG is associated with destabilization, while a negative ΔΔSVib (ENCoM) suggests increased flexibility or entropy in the mutant relative to the wild type. However, cases where both values are positive do occur (Capriotti et al. 2008, Li et al. 2020). The Gly-662 to Ala substitution introduces a methyl side chain, enabling additional interactions not possible with glycine. This change results in a new hydrophobic interaction with Gln-666, potentially enhancing structure stability and reducing flexibility (Fig. 4D). Similarly, the substitution of His-705 to Ala resulted in the formation of a new hydrogen bond with Phe-702 and a van der Waals interaction with Met-703, contributing to the increased stability of the H705A mutant (Table S3, Fig. 4E). The Thr-710 to Ala substitution also reorganized local interactions with neighboring amino acids, notably shifting a hydrogen bond from Ile-714 to Glu-713, which further stabilized the structure and decreased its flexibility (Table S3, Fig. 4F).
Together, these data suggest that the catalytic triad mutations, particularly H709A and E713A, induce local structural rearrangements that increase molecular flexibility, potentially compromising PdpC function. In contrast, mutations outside the triad, such as G662A and T710A, appear to stabilize the protein and reduce flexibility.
Impact of pdpC mutations on phagosomal escape, intramacrophage growth, and cytopathogenicity
Phagosomal escape is the initial step of the intracellular life cycle of F. tularensis, and mutants that lack a functional T6SS are unable to escape from the phagosome (Bröms et al. 2010, Degabriel et al. 2023). To assess escape, we used epifluorescence microscopy to examine bacterial colocalization with LAMP-1 in J774 macrophage-like cells. At both 2 and 6 h post-infection, LVS, WT and the H705A and T710A mutant strains escaped efficiently, whereas G662A, H709A and E713A mutants displayed no escape, similar to ΔpdpC and the negative control strain, ΔiglC (Fig. 5). Notably, S661A escaped with delayed kinetics (∼ 50% at 2 h versus 70%–80% for other escaping strains) (Fig. 5).
Colocalization of GFP-labeled F. tularensis strains with LAMP-1. J774 murine macrophage-like cells were infected for 2 h with GFP-expressing F. tularensis strains at an MOI of 200. Following washing, cells were incubated for an additional 2 or 6 h. Fixed and immunolabeled samples were examined for colocalization between GFP-tagged bacteria and the lysosomal marker LAMP-1 using an epifluorescence microscope. Data are presented as mean values with standard deviation from one representative experiment out of two. Asterisks indicate statistically significant differences in colocalization compared to LVS (, P ≤ 0.05; **, P ≤ 0.01; **, P ≤ 0.001) according to a Student’s t-test.
Transmission electron microscopy (TEM) confirmed these findings: LVS, WT, H705A, and T710A demonstrated very similar levels of escape at 2 and 6 h, while S661A again displayed delayed escape (Fig. 6 and Fig. S4). In contrast, G662A, H709A, E713A, ΔpdpC and ΔiglC, all remained predominantly confined within phagosomes (Fig. 6 and Fig. S4).
Phagosomal membrane integrity following infection with F. tularensis. J774 cells were infected with F. tularensis strains at an MOI of 1000 for 2 h, washed, followed by a 6-h incubation before fixation and analysis by transmission electron microscopy (TEM). Representative TEM images were acquired using a JEOL JEM 1230 microscope. Black arrows indicate vacuolar membranes enclosing intracellular bacteria. Scale bar: 0.5 μm.
Replication at 24 and 48 h mirrored escape patterns. The WT, S661A, H705A, and T710A complemented strains replicated efficiently, similar to LVS (Fig. 7A and B), while ΔiglC, ΔpdpC and mutants G662A, H709A, and E713A showed no or minimal replication (Fig. 7A and B; P < 0.001). The cytopathogenic effects closely correlated to the degree of intracellular replication: infection with ΔiglC, ΔpdpC, G662A, H709A, and E713A mutants caused no cytopathogenic effect, whereas S661A, H705A, and T710A induced marked effects, although delayed compared to LVS (Fig. 7C and D).
*Intracellular replication and cytopathogenic effects of F. tularensis strains in J774 cells. Cells were infected with various F. tularensis strains at an MOI of 200 for 2 h. Panels A and B show viable bacterial counts (log₁₀ CFU) at 0, 24, and 48 h post-infection following gentamicin treatment and a 30 min recovery. Cells were lysed using PBS with 0.1% sodium deoxycholate, and viable bacteria were quantified by plating. Data represents mean ± SD from triplicate samples in one representative experiment of two. Statistical differences from the parental LVS strain were determined using a two-tailed t-test with equal variance (*P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001). Panels C and D display cytopathogenicity measured by LDH release at the same time points, expressed as a percentage of the LDH activity in lysed, uninfected control cells. Bars represent mean ± SD from triplicate wells in one of two representative experiments. Asterisks indicate statistically significant differences compared to uninfected controls at the corresponding time point (*P ≤ 0.05; **P ≤ 0.01; **P ≤ 0.001).
Together, these findings show that conserved residues in the analyzed region are critical for the intracellular life cycle of F. tularensis, with certain mutations causing complete loss of function, while others result in partial defects.
Site-directed mutations in pdpC reveal distinct virulence phenotypes in vivo
Intracellular growth and cytopathogenicity are important factors for Francisella pathogenicity in vivo (Bosio 2011). To determine whether point mutations affected virulence, C57BL/6 mice were infected via the intradermal route with LVS, ΔpdpC, or the various in cis-complemented mutants. Dose and mortality data, summarized in Table S4, were as follows: at a dose of 7.9 × 10^7^ CFU (approximately 4 × LD_50_, (Forslund et al. 2006), LVS caused 100% mortality. Similarly, a dose of 6.8 × 10^7^ CFU of the WT-complemented mutant resulted in 100% mortality, as did doses of 7.2 × 10^7^ CFU and 8.9 × 10^7^ CFU of H705A and T710A mutants, respectively. In contrast, no mice died after infection with ΔpdpC (6.2 × 10^8^ CFU), or any of the mutants expressing S661A (5.5 × 10^7^ CFU), G662A (6.1 × 10^8^ CFU), H709A (9.5 × 10^8^), or E713A (6.8 × 10^8^ CFU). In addition, the H709A mutant harboring two copies of the mutated gene was also included. Also, this mutant was highly attenuated since all mice survived a dose of 7.1 × 10^8^ CFU.
To further investigate the intermediate phenotype of the S661A mutant observed in phagosomal escape assays, an additional experiment was performed using a higher dose (9.4 × 10⁸ CFU). In this case, 67% of the mice died, indicating partial virulence. In contrast, 100% of mice infected with LVS died at a much lower dose, 7.9 × 10^7^ CFU.
Altogether, the site-directed mutants demonstrated variable phenotypes in the mouse model that closely correlated with their ability to escape the phagosome. Thus, H709A, E713A, and G662A mutants, all defective in phagosomal escape, were highly attenuated even at doses ≥30-fold the LD_50_ of LVS. The S661A mutant that demonstrated delayed escape but normal intracellular replication, was also attenuated, but retained partial virulence at high doses. In contrast, the H705A and T710A mutants, both demonstrating normal escape and replication, were fully virulent, as was the WT-complemented strain.
The catalytic triad is important for intracellular growth and cytopathogenicity of F. novicida
To investigate whether the conserved catalytic triad also plays an important role in another subspecies, we constructed ΔpdpC together with the corresponding mutants in cis in F. novicida U112, a commonly employed model for the more virulent subspecies of Francisella. The selected mutations − S658A, H706A, and E710A − correspond to S661A, H709A, and E713A in the LVS strain.
The intracellular replication of the mutants was evaluated in J774 macrophages. All strains exhibited similar uptake, but ΔpdpC and all substitution mutants were significantly attenuated (P < 0.05 to < 0.001) for intracellular replication in comparison to the parental strain, F. novicida U112, or to the WT-complemented strain (Fig. 8A). Substitution mutants H706A and E710A showed a significant net growth of 2.5–2.7 log_10_ CFU as did ΔpdpC, while the WT-complemented strain and U112 replicated around 3.7–3.8 log_10_ CFU after 24 h (Fig. 8A). The S658A mutant exhibited an intermediate phenotype with a net growth of 3.1 log_10_ CFU (Fig. 8A). The observed cytopathogenic effects in part reflected the attenuated phenotype of the mutants. Specifically, infection with the amino acid-substituted mutants H706A and E710A, or the ΔpdpC deletion mutant resulted in little LDH release, although it remained significantly elevated compared to the ΔiglC mutant (P < 0.001). In contrast, infection with the S658A mutant resulted in intermediate levels, while high levels were observed during infection with the WT-complemented strain and U112 (Fig. 8B).
*Intracellular proliferation and cytotoxicity of F. novicida strains in J774 macrophages. (A) Bacterial replication was assessed by determining viable counts (log₁₀ CFU) at 0 and 24 h post-infection following gentamicin treatment. Cells were lysed using PBS with 0.1% sodium deoxycholate, and colony-forming units were enumerated by plating. (B) Cytotoxic effects were evaluated by measuring LDH release at 24 and 48 h time points, expressed as a percentage relative to fully lysed, uninfected control cells. Data represent mean ± SD from triplicate samples in one of two representative experiments. Statistical comparisons were made using a two-tailed t-test assuming equal variance (*P ≤ 0.05; **P ≤ 0.01; **P ≤ 0.001).
Collectively, these findings demonstrate that the conserved catalytic triad in PdpC is functionally important for both intracellular replication and cytopathogenicity in Francisella. While site-directed mutations in PdpC of F. novicida significantly impaired these processes, the mutants retained partial activity. In contrast, two out of the three corresponding substitutions in the LVS background completely abolished replication and cytotoxicity. This suggests that although the catalytic triad is important in both subspecies, PdpC overall plays a more critical role in LVS, indicating subspecies-specific differences in its contribution to virulence.
T6SS-mediated secretion by F. novicida pdpC site-directed mutants
F. novicida U112 is particularly suitable for studying T6SS-mediated secretion in vitro due to its responsiveness to KCl induction (Clemens et al. 2015), a feature missing in LVS. To investigate the secretion profiles of the F. novicida site-directed mutants, two complementary approaches were employed. First, Western blot analysis was used to assess the secretion of IglC and PdpC after growing the bacteria in medium supplemented with KCl, leading to activation of T6S. As expected, the ΔdotU mutant, which lacks a functional T6SS (Bröms et al. 2012), did not secrete either IglC or PdpC, confirming its role as a negative control. In contrast, all pdpC mutants, including the deletion mutant and the point mutants, secreted IglC at levels comparable to the WT-complemented strain and F. novicida U112 (Fig. 9). This indicates that neither the absence of PdpC nor the introduced mutations in the PdpC protein impaired IglC secretion. Importantly, WT PdpC and all PdpC point mutants were expressed and secreted at levels comparable to the U112 strain, suggesting that the amino acid substitutions did not interfere with the expression or secretion of the PdpC protein (Fig. 9).
Western blot analysis of FPI-associated protein expression and secretion in F. novicida. The indicated F. novicida strains were cultured in TSB supplemented with 5% KCl. Protein expression (pellet fractions) and secretion (cleared culture supernatants) were analyzed by SDS-PAGE followed by immunoblotting using specific antisera against PdpC or IglC. The band marked with an asterisk represents a non-specific signal detected by the PdpC antiserum. The experiment was performed in triplicate, and one representative experiments is shown.
To complement the targeted analysis provided by Western blotting, we next adapted a previously published protocol for analyzing F. novicida-secreted proteins using MS (Eshraghi et al. 2016) and the same set of F. novicida strains. This approach enables the identification and quantification of a wide range of secreted proteins, allowing for a more comprehensive assessment of secretion profiles. In total, the analysis identified 508 F. novicida proteins (Table S5), most of which were detected at low levels and showed no significant variation between the tested strains. The fold-change and P-values for strain comparisons to U112 are summarized in Table S5. As expected, U112 secreted high levels of the known T6SS substrates IglC, VgrG, PdpA, PdpC, PdpD, OpiA, and OpiB1, none of which were detected in the ΔdotU culture supernatant, based on a significance threshold of log_10_ P ≥1.5 and log_2_ fold-change ≥5.0 (Fig. 10 and Table S5). While other differences in protein abundance were observed, none exceeded a five-fold change relative to ΔdotU (Table S5). When comparing the protein profiles of WT PdpC or the PdpC point mutants to that of F. novicida U112, or to each other, all differences were below five-fold (Table S5 and data not shown).
Mass spectrometry (MS)-based analysis of FPI protein secretion in the absence of DotU in F. novicida. Volcano plot showing differential protein abundance in culture supernatants from a ΔdotU mutant and the wild-type F. novicida U112 strain. Several known FPI substrates, including OpiA, OpiB1, PdpA, VgrG, PdpC, IglC, and PdpD, are highlighted. Thresholds for significance were set at log₂ fold change ≥ 5.0 and –log₁₀ adjusted P-value ≥ 1.5.
While PdpC was essentially absent in the culture medium of ΔpdpC, it was secreted at high levels by all other mutants (Fig. 11 and Table S5). Consistent with earlier findings using alternative secretion assays, the absence of PdpC did not affect the secretion of other substrates (Eshraghi et al. 2016) (Table S5).
Mass spectrometry (MS)-based analysis of FPI protein secretion in the absence of PdpC in F. novicida. Volcano plot showing differential protein abundance in culture supernatants from a ΔpdpC mutant and the wild-type F. novicida U112 strain. Thresholds for significance were set at log₂ fold change ≥ 5.0 and –log₁₀ adjusted P-value ≥ 1.5.
Together, the Western blot and MS results confirm that the T6SS remains functional in all pdpC mutants. The amino acid substitutions do not impair the secretion of PdpC itself, IglC or other T6SS substrates. Thus, the observed defects in intracellular growth and cytopathogenicity of the catalytic triad mutants are not due to general defects in T6SS-mediated secretion.
Discussion
The role of PdpC has been much investigated due to its unique presence in F. tularensis—the only human pathogen within the genus—and its proposed role as an effector of the Francisella T6SS. Despite numerous investigations, any specific effector mechanism(s) executed by the protein is still enigmatic. Here, we provide evidence that four conserved amino acid; three of which show similarities to a triad homologous found in MCF toxins, are very important not only for the function of PdpC, but also for the virulence of F. tularensis LVS in the mouse model. Notably, these mutations do not impair secretion of other T6SS effectors, Thus, the data highlight the critical role of these amino acids in the function of PdpC. Our findings align with previous studies highlighting PdpC´s essential contribution to the high virulence of the prototypic subspecies tularensis strain SCHU S4 (Long et al. 2013, Lindgren et al. 2014, Uda et al. 2014) and its identification as a secreted protein in F. novicida, where it is considered the most important effector (Eshraghi et al. 2016). Together, these results reinforce the importance of PdpC for the virulence of all human pathogenic Francisella subspecies.
To investigate the roles of individual amino acids within the putative catalytic triad, we employed the rigorous strategy of in cis complementation in the LVS strain. LVS was chosen because its FPI region is essentially identical to that of highly virulent F. tularensis strains, ensuring the broader relevance of our findings. A challenge with LVS, however, is its duplicated FPI, which complicates genetic manipulation in cis. Nevertheless, we opted for this type of complementation in order to preserve native regulatory control of protein expression. While complementation in trans typically results in high protein levels, it often results in non-physiological expression or toxicity, and therefore fail to achieve phenotypical complementation, as observed in various biological systems (Alon 2006, Makanae et al. 2013). Our findings that WT, H705A and T710A restored an LVS-like phenotype support the notion that complementation in cis led to physiologically relevant expression, enabling phenotypical complementation.
To evaluate the functional significance of individual residues within PdpC, we replaced some with alanine. These included the four conserved residues S661, H709, E713, and G662, of which the first three were identified as showing similarity to the catalytic region of the MCF toxins (Agarwal et al. 2015). Functional assays revealed that mutants H709A, E713A, and G662A failed to escape the phagosome, whereas the S661A strain demonstrated delayed escape. In contrast, the H705A and T710A mutants, targeting non-conserved or semi-conserved residues, showed degrees of phagosomal escape similar to the LVS strain. These differences were reflected in intracellular replication: mutants lacking phagosomal escape showed minimal replication in J774 cells, whereas the others replicated similarly to LVS. The mouse model further distinguished phenotypes, since H709A, E713A, and G662A were attenuated by at least 30-fold, whereas S661A demonstrated intermediate attenuation (∼ four-fold). In contrast, H705A and T710A retained virulence comparable to that of LVS.
Our data demonstrated differences in the role of PdpC between LVS and F. novicida. While certain amino acid-substituted LVS mutants more or less lacked intracellular replication, the corresponding F. novicida mutants replicated more than 100-fold—highlighting a reduced dependency on PdpC in the latter subspecies. The data agree with previously published data showing that deletion of pdpC in LVS or the highly virulent SCHU S4 strain leads to severe impairment of intracellular growth in mouse bone marrow-derived macrophages and J774 cells (Lindgren et al. 2013, Long et al. 2013, Lindgren et al. 2014). In contrast, F. novicida ΔpdpC mutants demonstrated only minor replication defects in vitro, in agreement with our findings, whereas they showed a 100-fold attenuation in the mouse model (Ludu et al. 2008, Chou et al. 2013). Notably, F. novicida remains lethal in the Galleria mellonella model, even in the absence of PdpC (Brodmann et al. 2021). A prominent difference between F. novicida and other F. tularensis subspecies is the presence of a functional PdpD protein in the former. In F. novicida, PdpD is secreted and has been shown to contribute to virulence, whereas in other F. tularensis subspecies, the protein is truncated and likely non-functional (Chou et al. 2013, Eshraghi et al. 2016, Brodmann et al. 2021). Since PdpD and PdpC may have overlapping functions, it is likely that the presence of a functional PdpD in F. novicida partly compensates for the absence of PdpC. This redundancy could explain why deletion of pdpC impacts intracellular replication less in F. novicida than in F. tularensis.
A key feature of F. tularensis virulence is its ability to suppress host inflammatory responses, earning it the designation of a stealth pathogen (Telepnev et al. 2003, Sjöstedt 2006). Using the LPS-induced TNF release assay, we previously demonstrated that LVS infection effectively blocks TNF production, while a T6SS-deficient iglC mutant induce intermediate levels. Notably, ΔpdpC and all in cis-complemented point mutants mirror the LVS phenotype [(Lindgren et al. 2013) and data not shown], suggesting that TNF suppression remains intact in the absence of PdpC. This likely reflects the continued secretion of other T6SS effectors secreted in the absence of PdpC, despite the intraphagosomal location.
Two key features of F. tularensis pathogenicity are phagosomal escape and the secretion of T6SS effectors, both of which are essential for intracellular replication and likely for modulating host signaling (Bröms et al. 2010, Celli and Zahrt 2013, Clemens et al. 2018, Degabriel et al. 2023). To better define the T6SS-specific secretome, we performed MS-based analysis, similar to previous studies (Eshraghi et al. 2016, Ledvina et al. 2018). This approach, which requires in vitro T6SS activation, was applied to F. novicida grown in KCl-supplemented medium (Clemens et al. 2015). Proteins detected in F. novicida U112 and in cis-complemented strains, but absent in the T6SS-deficient dotU mutant, were considered T6SS-dependent. Consistent with earlier findings, we identified IglC, VgrG, PdpA, PdpC, PdpD, OpiA, and OpiB1 as secreted proteins. Importantly, the secretomes of amino acid-substituted PdpC mutants were similar to F. novicida U112, indicating that neither deletion, nor functional disruption of PdpC significantly alters T6SS-mediated secretion. These results support the role of PdpC as a T6SS effector without affecting the overall secretome.
Previous studies have identified distinct and partly overlapping roles for PdpC, PdpD, OpiA, and OpiB in virulence, with evidence of redundancy depending on the infection models (Eshraghi et al. 2016, Brodmann et al. 2021). For instance, OpiA-dependent phenotypes were only evident in the absence of PdpC, and similar patterns were seen for OpiB and PdpD. However, combined deletion of all four effectors resulted in a phenotype resembling T6SS-deficient strains, suggesting partial functional overlap. Interestingly, each effector alone was sufficient to kill Galleria mellonella larvae, indicating that redundancy and effector importance are host-specific (Brodmann et al. 2021).
While MS-based approaches provide a broad view of the secretome, they may miss low-abundance substrates. Alternative methods, such as reporter fusion assays, have identified additional candidates (e.g. IglE, IglF, IglI, IglJ, and PdpE) (Bröms et al. 2012). However, these methods require bacterial replication and are unsuitable for non-replicating PdpC mutants. Notably, PdpE was previously identified as a secreted substrate in complex with PdpC, though it is not required for intracellular growth and PdpE mutants show no phenotype in various models (Bröms et al. 2011, Lindgren et al. 2014, Adams et al. 2024 ).
Previous studies have provided limited insight into the molecular functions of F. tularensis T6SS substrates. To date, only OpiA has been functionally characterized as a phosphatidylinositol 3-kinase that delays phagosomal maturation (Ledvina et al. 2018). Although two of the three OpiB proteins, all structurally similar to OpiA and containing putative cysteine protease and ankyrin domains, were identified as secreted (Eshraghi et al. 2016), no phenotypes have been reported for OpiB mutants. A major challenge in elucidating effector functions is the lack of observable phenotypes in single-gene mutants, with the notable exception of PdpC (Eshraghi et al. 2016). Our identification of a critical active site in PdpC that is essential for both its function and F. tularensis virulence represents a significant step toward understanding this key effector.
Although the conserved triad in PdpC resembles motifs found in other proteins across species, its precise function remains enigmatic. In a preprint (Liu et al. 2025: bioRxiv), Liu et al. reported a cryo-EM structure of PdpC. Ideally, the impact of our mutations should be reassessed in the context of this structure; however, it is not yet available in the Protein Data Bank, thereby precluding such modeling. The same study also demonstrated that PdpC selectively binds phospholipids, suggesting that future work should investigate whether the mutations that reduce virulence also impair phospholipid binding.
The presence of a conserved triad often suggests catalytic potential—commonly associated with proteases, lipases, or transferases, but without biochemical or structural data, this remains speculative. The conservation of these residues across diverse species implies evolutionary pressure to maintain this region, suggesting functional importance. While this does not confirm enzymatic activity, it supports the idea that the triad contributes to a conserved mechanism.
Interestingly, the number of identified Francisella effectors remains remarkably low (Eshraghi et al. 2016), especially when compared to other intracellular pathogens such as Legionella pneumophila and Coxiella burnetii, which rely on hundreds of effectors, many of which have redundant roles (Qiu and Luo 2017). The reliance of F. tularensis on a small set of T6SS effectors may reflect an evolutionary trend toward genome reduction and specialization. In fact, comparative genomic analyses have shown that F. tularensis has undergone significant genome streamlining relative to experimental Francisella species, such as F. novicida and F. philomiragia. This includes the loss of metabolic and regulatory genes, expansion of insertion sequences, and reduced recombination—all hallmarks of adaptation to a restricted intracellular niche (Larsson et al. 2009). Notably, the duplication of the FPI in F. tularensis may represent a compensatory mechanism to maintain virulence despite a reduced genome. These features are consistent with patterns observed in many intracellular pathogens and may help explain the pathogen’s dependence on a limited, but highly specialized set of effectors, with PdpC playing a central role in its virulence strategy.
Supplementary Material
xtag009_Supplemental_Files
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Adams N, Hulvey D, Khalid N et al. Pathogenicity determinant protein E is a Francisella type VI secretion system effector protein that modulates host death. bio Rxiv. 2024. 10.1101/2024.11.25.624902 preprint: not peer reviewed. · doi ↗
- 2Agarwal S, Agarwal S, Biancucci M et al. Induced autoprocessing of the cytopathic makes caterpillars floppy-like effector domain of the Vibrio vulnificus MARTX toxin. Cell Microbiol. 2015;17:1494–509. 10.1111/cmi.12451.25912102 PMC 4583801 · doi ↗ · pubmed ↗
- 3Allsopp LP, Bernal P. Killing in the name of: T 6SS structure and effector diversity. Microbiology (Reading). 2023;169:1–22. 10.1099/mic.0.001367. · doi ↗
- 4Alon U. An Introduction to Systems Biology: Design Principles of Biological Circuits. Boca Raton,FL, USA: Taylor & Francis Group, 2006. 10.1201/9781420011432. · doi ↗
- 5Bankapalli LK, Mishra RC, Singh B et al. Identification of critical amino acids conferring lethality in Vop K, a type III effector protein of Vibrio cholerae : lessons from Yeast Model System. P Lo S One. 2015;10:e 0141038. 10.1371/journal.pone.0141038.26488395 PMC 4619451 · doi ↗ · pubmed ↗
- 6Bosio CM. The subversion of the immune system by Francisella tularensis. Front Microbio. 2011;2:9. 10.3389/fmicb.2011.00009. · doi ↗
- 7Brodmann M, Schnider ST, Basler M. Type VI secretion system and Its effectors Pdp C, Pdp D, and Opi A contribute to Francisella Virulence in Galleria mellonella Larvae. Infect Immun. 2021;89:e 0057920. 10.1128/IAI.00579-20.33875476 PMC 8208517 · doi ↗ · pubmed ↗
- 8Bröms JE, Lavander M, Meyer L et al. Igl G and Igl I of the Francisella pathogenicity island are important virulence determinants of Francisella tularensis LVS. Infect Immun. 2011;79:3683–96.21690239 10.1128/IAI.01344-10PMC 3165494 · doi ↗ · pubmed ↗