Xpf suppresses the mutagenic consequences of phagocytosis in Dictyostelium

The DNA damage response is highly conserved and prevents the accumulation of deleterious DNA damage after exposure to environmental mutagens. However, organisms and cell types show enormous variability in their susceptibility to mutagen exposure. For instance, in vertebrates, certain lineages such as blood stem cells are highly sensitive to DNA damage, whereas others such as muscle cells are resistant to the same mutagens (Meijne et al., 1991; Rossi et al., 2007). On a quite different scale are organisms such as Deinoccocus radiodurans and Dictyostelium discoideum, which are highly resistant to mutagens (Deering, 1968; Zahradka et al., 2006; Zhang et al., 2009). It is clear that these organisms have evolved DNA damage responses capable of rapidly and efficiently repairing extensive DNA damage (Hudson et al., 2005). However, an important question remains: why have they evolved such effective DNA repair? One possibility is that both D. radiodurans and Dictyostelium have unusual life cycles in that they can survive in dormant desiccated states. Such suspended existence could lead to the accumulation of extensive DNA damage that must be repaired to resume growth. Another possibility is that they are exposed to heavy mutagenesis as a consequence of their life cycle or niche, including their food being a source of mutagens. In the wild, Dictyostelium feed on bacteria by phagocytosis. The ingested microorganism is trapped in a phagolysosome where it is ultimately killed and degraded in a process resembling that employed by professional phagocytes such as macrophages (Cosson and Soldati, 2008). Here, we show that Dictyostelium amoebae use the DNA repair nuclease Xpf to protect their genome from mutagens released during the consumption of bacteria, revealing an unanticipated role of DNA repair in bacterial phagocytosis.

In the wild, the unicellular amoeba Dictyostelium discoideum is found in the soil litter feeding predominantly on bacteria and dividing by binary fission (Hohl and Raper, 1963; Weeks and Weijer, 1994). In the laboratory, this organism can be propagated either on agar plates coated with a Klebsiella aerogenes bacterial lawn, or it can be grown in axenic medium (Fey et al., 2007). We have previously reported that this organism is highly resistant to the mutagen and DNA-crosslinking agent cisplatin (Zhang et al., 2009). This resistance is under genetic control because Dictyostelium deficient in the excision repair nuclease Xpf (xpf⁻) are hypersensitive to this DNA-crosslinking agent. However, whenever we propagated the xpf⁻ strain we noted a profound growth defect on K. aerogenes plates (Fig. 1A,B). To address whether this defect was due to a loss of viability, we quantified the colony-forming efficiency. As seen in Fig. 1C, the plating efficiency of Xpf-deficient cells was greatly reduced when grown on K. aerogenes. In contrast, this strain grew as well as wild-type (Ax2) Dictyostelium in axenic medium (Fig. 1D,E). Furthermore, the cloning efficiency of xpf⁻ in axenic medium was comparable with that of Ax2 (Fig. 1F), indicating that the vegetative growth defect is specifically associated with feeding on K. aerogenes but not with axenic medium. Thus, these data suggest that the xpf⁻ strain struggles to proliferate when utilizing live K. aerogenes as a nutritive source.

We next tested whether this growth defect could be observed with other known bacterial food sources (Escherichia coli, Micrococcus luteus and Bacillus subtilis). xpf⁻ amoebae showed a growth defect on all of the three quite distinct bacteria to varying degrees (Fig. 2A,B; Fig. S1A). We then whether this effect depended on live food or whether it could also be observed with dead bacteria. We therefore heat-inactivated K. aerogenes for 20 min at 121°C. This preparation could still support Dictyostelium growth but it is not as nutritious as living K. aerogenes (Fig. 2A,B; Fig. S1A). However, heat-inactivated K. aerogenes was also toxic to xpf⁻ amoebae, suggesting that the bacteria do not need to be metabolically active at the time of ingestion to affect xpf⁻ amoebae.

In order to feed on bacteria, Dictyostelium must internalize their prey and then digest it to release nutrients (Cosson and Soldati, 2008; Lelong et al., 2011). This whole process contrasts to when Dictyostelium grows in axenic medium, where nutrient uptake occurs through a process called macropinocytosis (Bloomfield et al., 2015; Cardelli, 2001). We therefore determined whether the xpf⁻ strain is competent at ingesting and subsequently digesting bacteria. Ax2 and xpf⁻ strains were pulsed with bacteria for varying times and the number of bacterial colony-forming units (CFU) per amoeba was scored for both the total bacteria bound and internalized by Dictyostelium (Lelong et al., 2011). The CFU number did not markedly differ between wild-type and xpf⁻ Dictyostelium, strongly indicating that xpf⁻ cells do not present a defect in bacterial killing (Fig. 2C,D). We confirmed the bacterial phagocytic proficiency by quantifying the uptake of tetramethylrhodamine isothiocyanate (TRITC)-labeled yeast (Fig. 2E) (Rivero and Maniak, 2006). Finally, when bacteria are ingested by the amoeba, the nascent phagosome incorporates the membrane protein p80 – a reliable marker of the endocytic pathway (Ravanel et al., 2001). Ax2 and xpf⁻ strains were therefore pulsed with GFP-expressing K. aerogenes and were then visualized by confocal microscopy. Immunofluorescent detection of p80 was colocalized with GFP-containing vesicles. These results show that in both Ax2 and xpf⁻ cells, GFP-labeled K. aerogenes fuse with p80-containing vesicles, causing a decrease in the GFP signal, denoting bacterial digestion (Fig. 2F). Supporting this observation, the clearance in the supernatant of a suspension where both Dictyostelium and K. aerogenes are co-incubated was similar between the Ax2 and xpf⁻ strains (Fig. 2G). From this set of experiments, we can conclude that whereas the xpf⁻ strain is sensitive to K. aerogenes, this sensitivity was not due to defective uptake or digestion of bacterial food.

Xpf is an endonuclease that cuts out damaged DNA caused by UV irradiation and interstrand crosslinking agents (Ahmad et al., 2008; Enzlin and Scharer, 2002). We therefore set out to establish whether this endonuclease activity is required for Dictyostelium to effectively utilize bacteria as a food source. Xpf, and in particular its nuclease motif carrying a crucial metal-binding aspartic acid residue, is highly conserved between humans (where the enzyme is also known as ERCC4) and Dictyostelium (Fig. 3A) (Enzlin and Scharer, 2002). Consequently, we transfected Ax2 and xpf⁻ Dictyostelium cells with plasmids for expression of GFP fusions of either wild-type Dictyostelium Xpf (Xpf) or a nuclease-dead mutant form where the key aspartic acid residue in the nuclease motif was mutated to an alanine residue (Xpf-D771A). All transfected strains expressed the recombinant Xpf (Fig. S1B), but only wild-type Xpf and not the nuclease-dead mutant rescued the growth defect of xpf⁻ cells on K. aerogenes bacterial lawns (Fig. 3B; Fig. S1B,C). To extend our analysis, we then tested whether two other DNA repair nuclease-deficient Dictyostelium strains [Mus81 (mus81⁻) and Fan1 (fan1⁻)] were also susceptible to growth inhibition on K. aerogenes plates. In fact, neither of these mutants exhibited a growth defect on K. aerogenes plates (Fig. 3C). However, Dictyostelium has robust DNA repair systems and it has been described as a γ-ray-resistant organism (Deering, 1968; Hudson et al., 2005). Accordingly, we investigated mutants in Xpf-related DNA repair pathways for their contribution to tolerance of bacterial mutagens and found that Dictyostelium knockouts in the translesion synthesis DNA-repair polymerase Rev3 (rev3^–) and the global nucleotide excision repair (NER) gene xpc (xpc⁻) showed a mild growth defect, whereas a fncD2⁻ strain, which has an inactive Fanconi anaemia DNA repair pathway, showed comparable growth on K. aerogenes bacterial lawns to the wild-type strain Ax2 (Fig. S1D). Taken together, our results show that sustained growth on bacterial plates specifically requires the nuclease activity of Xpf.

The Xpf endonuclease repairs damaged DNA; hence, we reasoned that in its absence, damaged DNA would be expected to accumulate and might eventually lead to mutagenesis. Direct approaches to detect DNA damage were unsuccessful. However, we measured mutagenesis frequency in Dictyostelium by assessing their resistance to methanol (Garcia et al., 2002; Podgorski and Deering, 1980). This mutagenic reporter system relies on the fact that methanol is toxic to Dictyostelium because of its conversion by the enzyme Catalase A (CatA) into toxic formic acid (Garcia et al., 2002). Inactivation of the catA gene leads to a failure to convert methanol into formic acid and hence confers resistance to this alcohol (Fig. 4A). We therefore set up a methanol mutagenesis assay to determine the frequency of resistance to methanol in Ax2 and xpf⁻ strains after exposure to K. aerogenes (Fig. 4B). Briefly, we first took Ax2 and xpf⁻ cells and expanded them from single clones into six-well plates in axenic medium, and then plated these out on methanol-containing plates. In parallel, we took the same two strains and plated them out onto K. aerogenes plates, picked single colonies, and re-plated on new K. aerogenes agar plates. We then scraped the entire population from a single plate and briefly expanded them in individual wells within a six-well plate; the expanded population was then plated onto methanol plates. Methanol resistance (as number of resistant colonies per 10⁶ viable cells) was then determined on methanol agar plates. A clone that acquired a catA mutation early during growth in any differential condition will show an elevated number of methanol-resistant colonies, but by repeating the assay many times with independent cultures, this fluctuation assay captures the mutation frequency (Luria and Delbruck, 1943). The data in Fig. 4C show that both Ax2 and xpf⁻ Dictyostelium developed few methanol-resistant clones when propagated in axenic medium. In contrast, when propagated on K. aerogenes plates, the xpf⁻ strain shows a striking induction of methanol-resistant colonies compared to Ax2, indicating that Xpf prevents mutagenesis in this growth condition. Next, we sought to determine the mutational pattern underlying these mutagenic events and thus amplified, cloned and sequenced the catA gene from Ax2 and xpf⁻ methanol-resistant clones. However, the pattern of mutations did not differ greatly between the two strains (Fig. S2), although it is important to note that the mutational pattern observed here might be biased towards gene-disrupting mutations, which are more likely to cause enzyme inactivation than point mutations.

It is thus very likely that the reason xpf⁻ amoebae fail to thrive in the presence of a bacterial food source is due to the accumulation of DNA damage. Xpf participates in several DNA repair pathways, including homologous recombination, NER and the Fanconi pathway (Manandhar et al., 2015), which might explain why its role is so fundamental for phagocytic growth. We can only speculate as to the mechanism responsible for causing DNA damage. The simplest explanation is that genotoxins are produced by the ingested bacteria, although it is unlikely to be an exotoxin given that heat-inactivated bacteria were still toxic. It is plausible that the genotoxin might be an integral bacterial part that is not inactivated by heat sterilization, or it could be a substance generated as a consequence of ‘digesting’ this food source. Another intriguing possibility comes from the resemblance of Dictyostelium to neutrophils, which kill ingested bacteria by using respiratory burst activity (Chen et al., 2007; Cosson and Soldati, 2008; Zhang et al., 2016). Neutrophil killing generates a battery of reactive molecules such as hypochlorus acid and reactive oxygen species, which are known to be highly mutagenic (Knaapen et al., 2006). It is thus conceivable that the mutagen forms part of an amoebal immune response to bacteria. An intriguing observation is that bacteria differ in their potency to inhibit growth. Exploring which factors determine these differences might give insight into the nature of genotoxicity. Finally, this work highlights the intricate manner by which nutritional sources might stimulate mutagenesis. Consuming food is essential to (heterotrophic) life, but, as this work highlights, often comes at considerable mutational cost against which the organism must defend itself.

All Dictyostelium strains were routinely grown at 22°C in HL5 (axenic medium) supplemented with streptomycin (200 µg/ml). Bacterial lawn plates were made by spreading 300 µl of an overnight bacterial culture on SM-agar plates and pictures were taken after 5 days at 22°C. For Micrococcus luteus, 0.5 l of stationary-phase culture was pelleted and resuspended in 50 ml, and then 1 ml of this culture was spread on SM-agar plates. Pictures were taken after 9 days. Comparison between two groups was performed by using a t-test in Prism software.

The parental strain was the Kay laboratory version of Ax2, according to the following nomenclature: Ax2 (wild type), HM1403 (xpf⁻), HM1464 (mus81⁻), HM1253 (fncD2⁻), HM1456 (xpc⁻), HM1351 (rev3⁻) (Zhang et al., 2009) and HM1515 (fan1⁻; DDB_G0267916). The colony-forming efficiency in HL5 was determined by sorting one cell per well into 96-well plates. After 20 days, the number of confluent wells per plate was scored and represented as a percentage. The colony-forming efficiency on bacterial plates was scored by plating 25 and 50 viable Dictyostelium cells on K. aerogenes lawns and counting the colonies after 5 days for Ax2, and 6 days for xpf⁻. Growth profiles in axenic medium were obtained using a Vi-cell analyzer (Beckman Coulter). Doubling time was determined as described previously (Fey et al., 2007). Cloning of Xpf was carried out using primers described in Table S1 in pDM317. The Fan1-knockout was made using a pLPBLP-targeting vector constructed by using the primers listed in Table S1.

Phagocytosis and killing of bacteria were analyzed as described previously (Benghezal et al., 2006). Phagocytosis of fluorescent TRITC (tetramethylrhodamine isothiocyanate)-labeled yeast was based on a published protocol (Rivero and Maniak, 2006). Clearance of bacteria was followed by reading optical density in the supernatant of a phosphate buffer suspension initially containing 10⁶ Dictyostelium and 1×10⁸ K. aerogenes cells.

Endocytosis and intracellular bacteria were visualized by incubating 25×10⁶ GFP-expressing K. aerogenes from an overnight culture with 5×10⁵ Dictyostelium cells in 500 µl of HL5 medium. An anti-p80 H161 monoclonal antibody (2 µg/ml; Pierre Cosson, Department of Cell Physiology and Metabolism, University of Geneva Medical School, Geneva, Switzerland) was used together with an Alexa-Fluor-647-coupled secondary antibody (Mercanti et al., 2006; Ravanel et al., 2001). Images were acquired on a Zeiss LSM710 confocal microscope and processed in ImageJ.

Ax2 or xpf⁻ cells were grown in HL5 to 10⁶ cells/ml. From this broth, Dictyostelium were plated in order to get isolated clones both in HL5-containing 96-well plates and on K.-aerogenes-coated plates. The isolated clones (n=41 from HL5, and n=43 for Ax2 or n=72 for xpf⁻ from K. aerogenes plates) were expanded in HL5. Then, the clonal population was lifted, and up to 5×10⁶ viable amoebae were plated onto 3% methanol-containing SM-agar plates. After 6 days, the number of methanol-resistant colonies per plate was scored and plotted relative to the colony-forming efficiency on K. aerogenes plates (Garcia et al., 2002; Podgorski and Deering, 1980). Finally, methanol-resistant clones were grown in HL5 and the catA gene was cloned into pTOPO for sequencing.

We thank Pierre Cosson for providing the anti-p80 antibody and GFP-expressing K. aerogenes, and for his advice on phagocytosis assays. We also thank Felix Dingler for the critical reading of the manuscript.