Sequential action of antibacterial effectors in Dictyostelium discoideum phagosomes

Abstract

Mammalian professional phagocytic cells ingest and kill invading microorganisms and prevent the development of bacterial infections. Our understanding of the sequence of events that results in bacterial killing and permeabilization in phagosomes is still largely incomplete. In this study, we used the Dictyostelium discoideum amoeba as a model phagocyte to study the fate of the bacteria Klebsiella pneumoniae inside phagosomes. Our analysis distinguishes three consecutive phases: bacteria first lose their ability to divide (killing), then their cytosolic content is altered (permeabilization), and finally their DNA is degraded (digestion). Phagosomal acidification and production of free radicals are necessary for rapid killing, membrane‐permeabilizing proteins BpiC and AlyL are required for efficient permeabilization. These results illustrate how a combination of genetic and microscopical tools can be used to finely dissect the molecular events leading to bacterial killing and permeabilization in a maturing phagosome.

Following phagocytosis, internalized Klebsiella pneumoniae are killed within a few minutes. Phagosomal acidification and production of free radicals are necessary for rapid killing, and membrane‐permeabilizing proteins BpiC and AlyL are required for efficient permeabilization. Digestion of the bacterial content, in particular of bacterial DNA, is then initiated and extends over a period of 30 to 60 min.

Abbreviations

AlyL

amoeba lysozyme‐like protein L

A.U.

arbitrary units

Bac

bacteria

BpiC

bactericidal/permeability‐increasing protein C

DAPI

4′,6‐diamidino‐2‐phenylindole

DNA

deoxyribonucleic acid

FITC

fluorescein isothiocyanate

GFP

green fluorescent protein

KO

knock‐out

LB

lysogeny broth

LPS

lipopolysaccharide

NADPH

reduced nicotinamide adenine dinucleotide phosphate

Nox

NADPH oxidase

PCR

polymerase chain reaction

SB

Sorensen's buffer

WT

wild‐type

PAF

paraformaldehyde

1. INTRODUCTION

Phagocytic cells are an essential element of innate immunity in the human body (Rosales & Uribe‐Querol, 2017). Professional phagocytic cells like macrophages and neutrophils ingest and kill invading microorganisms and prevent the development of bacterial infections. Within these cells, as the early phagosome matures into a phagolysosome, it acquires luminal and membrane constituents of lysosomes. Many potential antibacterial mechanisms have been identified in maturing phagosomes, and they generally fall into four main categories (Uribe‐Querol & Rosales, 2017). First, reactive oxygen species and reactive nitrogen species are produced by phagosomal enzymes like the NADPH oxidase 2 (NOX2). Second, the phagosome ionic composition is profoundly modified by ionic pumps and transporters, such as the proton v‐ATPase which ensures acidification of the phagosome. Many ions, like protons or zinc may participate directly or indirectly (e.g., by activating other bactericidal mechanisms) in intracellular killing. Third, a variety of lysosomal enzymes degrade specific bacterial elements. Fourth, peptides and proteins delivered to the phagosome permeabilize bacterial membranes (Jauslin et al., 2021; Nash et al., 2006).

In view of the large number of putative antibacterial mechanisms, our understanding of the sequence of events that results in bacterial destruction in phagosomes is still largely incomplete (Hurst, 2012). Even our vocabulary is uncertain: in different publications, the term “bacterial killing” may refer to very diverse events, such as the permeabilization of bacterial membranes (Riazanski et al., 2020), the partial or complete destruction of various bacterial constituents (Jauslin et al., 2021), or the loss of the bacterial capacity to grow (Lelong et al., 2011). While these diverse events are clearly linked, their precise role in the overall process and the underlying molecular mechanisms remain to be established. One of the most precise studies on that theme proposes that in alveolar macrophages, early non‐acidic phagosomes are the site where bacteria are permeabilized by a NOX‐dependent mechanism, while delivery of lysosomal enzymes and acidification initiates a second phase devoted to digestion of the bacterial membrane (Riazanski et al., 2020). The limitations in our understanding are compounded by the fact that different phagocytic cells (e.g., neutrophils, macrophages or amoebae) may make use of different antibacterial mechanisms and that different species of bacteria are probably killed by largely different mechanisms (Hurst, 2012; Jauslin et al., 2021).

Dictyostelium discoideum is a soil amoeba which continuously ingests, kills and digests a wide variety of microorganisms to ensure its feeding (Cosson & Lima, 2014). To the best of our knowledge, D. discoideum makes use of an array of antibacterial mechanisms very similar to those used by mammalian phagocytes (Dunn et al., 2017). It is amenable to biochemical analysis (e.g., identification of bacteriolytic proteins in vitro) (Guilhen et al., 2020), to genetic analysis (e.g., analyzing the phenotype of specific KO mutant cells) (Jauslin et al., 2021) and to live microscopy (e.g., visualize phagocytosis of bacteria) (Leiba et al., 2017). In this study we analyzed bacterial killing, permeabilization and digestion in D. discoideum phagosomes and determined the role of several key gene products at various stages in this process.

2. RESULTS

2.1. In D. discoideum phagosomes, permeabilization of K. pneumoniae precedes their digestion

For the sake of clarity, we adopted a strict nomenclature in this study. The choice of words is largely arbitrary, and is not meant to imply a priori anything about the underlying mechanisms. Bacteria are called alive if they can divide, dead if they cannot, and killing refers to the loss of the bacterial ability to divide. Permeabilization of GFP‐expressing bacteria is defined as the loss of their intracellular GFP fluorescence. The equivalence between loss of GFP fluorescence and bacterial permeabilization is hammered out in the Discussion section. Finally, bacterial digestion designates the loss of bacterial integrity evidenced when DAPI‐stained bacteria gradually disaggregate and lose their DAPI staining. Of these three parameters (killing, permeabilization, digestion), our previous analysis focused on the loss of bacterial GFP fluorescence (=bacterial permeabilization) which was generally observed a few minutes after ingestion of non‐pathogenic K. pneumoniae bacteria into D. discoideum phagosomes (Jauslin et al., 2021; Leiba et al., 2017).

To better define the sequence of events in D. discoideum phagosomes, we first addressed the question of whether loss of bacterial GFP fluorescence was synonymous with loss of the whole bacterial structure. For this, we labeled GFP‐expressing bacteria with DAPI, a membrane‐permeable fluorescent dye that is expected to persist in killed bacteria as long as their DNA is not digested (Hoff, 1988; Johnson & Criss, 2013a). The observed DAPI fluorescence was decreased by fluorescence energy transfer to GFP, imposing a significant level of correction for all the results shown in Figure 1 (See Supplementary Figure S1). In individual ingested bacteria, DAPI staining persisted after extinction of GFP fluorescence, and was lost gradually over a much longer period of time (>30 min) (Figure 1a). The levels of GFP and DAPI fluorescence were quantified in individual ingested bacteria, and two typical curves are shown (bac1 and bac2; Figure 1b). For these two bacteria, loss of GFP fluorescence occurred abruptly 4 min (bac1) and 20 min (bac2) after ingestion of the bacteria. On the contrary, in these two instances, DAPI fluorescence declined gradually 30 min after ingestion, and was fully lost approximately 60 min after ingestion (Figure 1b). 150 ingested bacteria were observed and we always observed that DAPI fluorescence persisted after extinction of GFP. The average fluorescence levels were determined and plotted as a function of time after phagocytosis in WT D. discoideum (Figure 1c), and this analysis confirmed that GFP fluorescence disappeared faster than DAPI fluorescence (Figure 1c). To precisely determine the relationship between these two events (GFP extinction, DAPI extinction), DAPI fluorescence was plotted as a function of time following GFP extinction for each bacterium (Figure 1d). This analysis showed that the DAPI fluorescence did not vary significantly in the 10 min preceding extinction of GFP fluorescence. DAPI fluorescence started to decrease approximately 20 min after GFP extinction, and was completely lost approximately 60 min after GFP extinction (Figure 1d). Overall, these observations clearly establish that permeabilization of ingested bacteria (=GFP extinction) occurred with a half‐life of approximately 10 min after ingestion, and was followed by a much slower gradual digestion phase (=DAPI extinction) extending over a period of approximately 1 h. Together these observations indicate that in phagosomes, bacterial permeabilization always precedes their digestion.

FIGURE 1.

Permeabilization and digestion of Klebsiella pneumoniae in Dictyostelium discoideum phagosomes. To visualize ingestion and intracellular digestion of individual bacteria, D. discoideum cells were incubated for 2 h with GFP‐expressing K. pneumoniae stained with DAPI. (a) Successive images of a WT D. discoideum cell ingesting (t = 0), destroying (t = 8 min) and digesting (t = 30 min) an individual K. pneumoniae bacterium. Scale bar 10 μm. (b) In each image, the fluorescence intensity of GFP and DAPI was measured, normalized to t = 0 and represented as a function of time following ingestion. The measured DAPI fluorescence was corrected to take into account the fluorescence energy transfer between DAPI and GFP molecules (see Supplementary Figure S1). Two examples of internalized bacteria are shown (bac1 and bac2). Loss of GFP fluorescence occurred abruptly 4 min (Bac1) and 20 min (Bac2) after ingestion of the bacteria. DAPI fluorescence declined gradually 30 min after ingestion and was fully lost approximately 60 min after ingestion. (c) The average GFP and DAPI fluorescence levels were determined in 150 bacteria and plotted as a function of time after phagocytosis. DAPI fluorescence persisted after extinction of GFP in all cases. (d) The same DAPI fluorescence measurements shown in (c) were plotted as a function of time normalized to the time of GFP extinction for each bacterium. DAPI fluorescence did not vary significantly in the 10 min preceding extinction of GFP fluorescence and started decreasing approximately 20 min after GFP extinction (n = 150). A.U, Arbitrary units.

2.2. Kil1 and Kil2 are necessary for efficient digestion

We next tested whether extinction of DAPI fluorescence was slowed down in kil1 KO and kil2 KO mutants, both of which have previously been shown to destroy ingested bacteria very inefficiently (Benghezal et al., 2006; Jauslin et al., 2021; Lelong et al., 2011). Kil1 is a Golgi sulfotransferase involved in the maturation of lysosomal enzymes (Benghezal et al., 2006). Kil2 is a phagosomal P‐type ATPase that has been proposed to transport magnesium ions from the cytosol to the phagosomal lumen (Lelong et al., 2011). As previously reported, we observed that extinction of bacterial GFP was slower in kil1 KO cells (Figure 2a), in kil2 KO cells (Figure 2b) and in kil1kil2 double KO cells (Figure 2c) than in WT cells. In addition, DAPI fluorescence persisted longer in kil1 KO cells (Figure 2a), in kil2 KO cells (Figure 2b) and in kil1kil2 double KO cells (Figure 2c) than in WT cells. In these experiments, the observed DAPI fluorescence was decreased by fluorescence energy transfer to GFP, imposing a significant correction of the measured values (See Supplementary Figure S1). To avoid the need for correction and to confirm our results, bacteria that did not express GFP were used, and DAPI fluorescence did persist longer in kil1 KO, kil2 KO and double kil1kil2 KO than in WT cells (Figure 2d). Together these experiments indicate that efficient permeabilization and digestion of ingested bacteria requires the activity of both Kil1 and Kil2.

FIGURE 2.

Efficient permeabilization and digestion of bacteria requires the activity of both Kil1 and Kil2. Dictyostelium discoideum cells were incubated with GFP‐expressing Klebsiella pneumoniae stained with DAPI and imaged as described in the legend to Figure 1. Extinction of GFP fluorescence and DAPI fluorescence was slower in kil1 KO (a), kil2 KO (b) and kil1kil2 double KO (c) cells than in WT cells (N = 2 independent experiments; n = 60 ingested bacteria per condition). In these experiments, bacteria expressing GFP were labeled with DAPI and the DAPI fluorescence was decreased by fluorescence energy transfer to GFP, imposing a significant level of correction (Supplementary Figure S1). (d) To avoid the need for compensation, bacteria that did not express GFP were used, and identical results were obtained (N = 4 independent experiments WT: n = 120; kil1 KO: n = 60; kil2 KO: n = 60; kil1kil2 KO: n = 60).

2.3. Killing of K. pneumoniae precedes their permeabilization in phagosomes

We next attempted to define if extinction of GFP fluorescence was synonymous with loss of bacterial viability, and if not, which one of the two events occurred first. In order to assess simultaneously the disappearance of bacterial fluorescence and the loss of their ability to grow, we first visualized by light microscopy for 25 min D. discoideum cells mixed with K. pneumoniae (phase 1) (Figure 3a). The medium was then abruptly changed to LB containing a small percentage of paraformaldehyde, detergent (Triton X100) and kanamycin. The temperature was raised to 37°C and bacterial growth and division were observed for 15 h (phase 2) (Figure 3a). The medium used during phase 2 killed D. discoideum cells as evidenced by the observation that they irreversibly stopped moving within seconds and rounded up. However, the cell structure was sufficiently preserved to prevent the release of ingested bacteria from dead D. discoideum cells (Supplementary Movie S1). On the contrary, these conditions did not affect bacterial viability as evidenced by the fact that addition of paraformaldehyde and Triton X100 did not slow down growth of K. pneumoniae in LB (Supplementary Figure S2). A vast majority of uningested bacteria grew and divided during phase 2 (91 ± 0.74%; mean ± SEM; N = 58 independent experiments, n = 2320 bacteria) confirming that bacterial viability was not decreased in phase 2. We expected each ingested bacteria to fall into one of seven theoretical categories. Intraphagosomal bacteria that were still fluorescent at the end of phase 1 were classified into category A if they divided during phase 2, category B if they did not divide during phase 2 but did not lose their fluorescence, or category C if they did not divide and lost their fluorescence during phase 2 (Figure 3a,b). Intraphagosomal bacteria that lost GFP fluorescence during phase 1 were classified into category D if they did not divide during phase 2, category E if they did not recover GFP fluorescence but divided during phase 2, category F if they recovered their fluorescence but did not divide during phase 2, or category G if they recovered their fluorescence and divided during phase 2 (Figure 3a,b). We performed 58 independent experiments and visualized 3344 phagocytic events where WT D. discoideum cells ingested a K. pneumoniae bacterium (Figure 3a). For 2240 ingested bacteria, GFP fluorescence was still detected at the end of phase 1 (referred to as GFP‐ON in Figure 3a; categories A + B + C). For 1104 ingested bacteria, GFP fluorescence went off during phase 1 (referred to as GFP‐OFF in Figure 3a). Division of non‐fluorescent bacteria during phase 2 can be visualized by phase contrast microscopy (Supplementary Figure S3), or by fluorescence microscopy if they recover GFP fluorescence. We never observed GFP‐OFF bacteria dividing (category E), recovering fluorescence (category F) or recovering fluorescence and dividing (category G). Accordingly, all 1104 GFP‐OFF bacteria were classified in category D. This observation indicates that all GFP‐OFF bacteria are incapable of dividing, that is, are dead. On the other hand, the 2240 GFP‐ON bacteria were seen to fall into the three proposed categories: during phase 2, 400 of them divided (category A), 132 did not divide but remained fluorescent (category B), and 1708 did not divide and lost their fluorescence (category C) (Figure 3a,b). This result indicates that ingested bacteria that are still fluorescent can be either alive (category A) or dead (categories B and C). Together these observations indicate that in phagosomes, bacterial death (= loss of the ability to divide) always precedes their permeabilization (= loss of GFP fluorescence).

FIGURE 3.

In Dictyostelium discoideum phagosomes, Klebsiella pneumoniae bacteria are first killed then permeabilized. (a) Schematic representation of the experimental procedure: D. discoideum cells were mixed with GFP‐expressing K. pneumoniae and imaged every 30 s for 25 min (phase 1). The medium was then changed to LB containing kanamycin, paraformaldehyde (0.002%) and triton X100 (0.002%) and the temperature raised to 37°C. These conditions killed and permeabilized D. discoideum cells without disrupting fully their internal structure and did not affect bacterial viability. Bacteria were imaged further every 10 min during 15 h (phase 2). Bacteria were classified in seven categories (A–G). The number of events observed in each category is listed on the right (N = 58 independent experiments). (b) Representative images showing bacteria classified in categories A, B, C, and D, as well as extracellular bacteria. White arrowheads indicate the time of phagocytosis. Scale bar 10 μm. (c) Loss of GFP fluorescence (=bacterial permeabilization) and of bacterial ability to grow (=bacterial killing) were plotted as a function of time following ingestion. The data shown combines data from 58 independent experiments (n = 3344 phagocytic events).

For each bacterium, during phase 1, the time of ingestion and the time of GFP extinction were recorded. This allowed us to establish the kinetics of GFP fluorescence extinction (= bacterial permeabilization) during phase 1 (Figure 3c, ■). To determine the kinetics of bacterial killing, we also calculated in the same experiments the percentage of ingested bacteria that were still capable of dividing at a given time after phagocytosis (Figure 3c, ▼). In WT D. discoideum cells, 50% of bacteria were killed 2 min after ingestion, while it took approximately 9 min for 50% of them to lose their fluorescence (Figure 3c). These observations indicate that the killing of ingested K. pneumoniae precedes their intracellular permeabilization in D. discoideum phagosomes. Note that, while it is relatively easy to visualize bacterial permeabilization (i.e., GFP extinction), determining the kinetics of bacterial killing requires data from a large number of independent experiments.

2.4. Kil1 and Kil2 are necessary for efficient killing

We next assessed in the same manner the killing and permeabilization of K. pneumoniae in kil1 KO and kil2 KO cells. Previous experiments established that both bacterial permeabilization and bacterial killing are slower in kil1 KO and kil2 KO cells than in WT cells (Benghezal et al., 2006; Jauslin et al., 2021; Lelong et al., 2011). When bacterial death and permeabilization were recorded simultaneously as described in Figure 3, we also observed that both killing and permeabilization were significantly delayed in kil1 KO cells (Figure 4a) and in kil2 KO cells (Figure 4b) compared to WT cells. In kil1‐kil2 double KO cells, killing of ingested bacteria was even slower than in individual kil1 and kil2 KO cells (Figure 4c). These observations confirm that the procedure developed in this study measures accurately and simultaneously bacterial killing and bacterial permeabilization, and that Kil1 and Kil2 are both required for efficient intracellular killing and permeabilization of K. pneumoniae (Figure 4 and Supplementary Figure S4a).

FIGURE 4.

Kil1 and Kil2 are necessary for efficient killing of bacteria. Bacterial killing and permeabilization in phagosomes were analyzed as described in Figure 3. Both bacterial killing and permeabilization were slower in kil1 KO cells (a), kil2 KO cells (b) and kil1kil2 double KO cells (c) than in WT cells. To obtain meaningful comparisons, only experiments carried out in parallel on the same day were used for direct comparisons of WT and mutant cells. (a: N = 5 independent experiments, WT n = 376 phagocytic events, kil1 n = 412; b: N = 8, WT n = 535, kil2 n = 566; c: N = 7, WT n = 526 kil1kil2 n = 631). The comparison between WT and mutants was performed for each 3 min time interval using the total number of bacteria (N _CatA + N _CatB + N _CatC + N _CatD) and the total number of live bacteria (N _CatA). *p < .05, Fisher's exact test.

2.5. AlyL, BpiC, NoxA, and acidic pH are necessary for efficient permeabilization of K. pneumoniae in phagosomes

Three effector proteins were previously shown to participate in the permeabilization of K. pneumoniae: AlyL, one of D. discoideum lysozyme, the BpiC bactericidal permeability‐increasing protein and NoxA, the main superoxide‐producing NADPH oxidase in D. discoideum (Jauslin et al., 2021). Genetic inactivation of the corresponding genes led to a slower permeabilization of K. pneumoniae bacteria, although in the original study the defect observed in noxA KO cells was not statistically significant (Jauslin et al., 2021). We observed virtually identical results in this new set of experiments: in alyL KO cells, bpiC KO cells and noxA KO cells, permeabilization of K. pneumoniae was significantly slower than in WT cells (Supplementary Figure S4b). In addition, NH₄Cl increases the pH in acidic lysosomes and phagosomes (Marchetti et al., 2009), and as observed previously (Jauslin et al., 2021), NH₄Cl slowed down permeabilization of K. pneumoniae (Supplementary Figure S4b). Note that NH₄Cl only partially raises the phagosomal pH in D. discoideum endocytic compartments (Marchetti et al., 2009). While this observation indicates that the very acidic pH of phagosomes is important for efficient permeabilization of bacteria, a complete loss of phagosomal acidification would presumably have a more profound effect on bacterial permeabilization.

2.6. NoxA and acidic pH are required for efficient killing of bacteria, BpiC and AlyL are not

We then analyzed the intracellular killing and permeabilization of K. pneumoniae in alyL KO cells, bpiC KO cells and noxA KO cells as well as in cells exposed to NH₄Cl (Wiegand et al., 2011). Note that a significant variability is apparent when comparing the kinetics of killing observed in WT cells in different sets of experiments. For this reason, the killing kinetics in mutant cells were only compared with kinetics in WT cells analyzed the same day. The kinetics of intracellular killing were identical in alyL KO cells, in bpiC KO cells and in WT cells (Figure 5a,b, ▼). On the contrary, bacterial permeabilization was significantly slower in alyL KO cells and bpiC KO cells than in WT cells (Figure 5a,b, ■). These results indicate that AlyL and BpiC participate in K. pneumoniae permeabilization but not in their intracellular killing in D. discoideum phagosomes. A different phenotype was observed in noxA KO cells and in cells incubated with NH₄Cl: in these cells, both the permeabilization of K. pneumoniae and their killing were slower than in WT cells (Figure 5c,d), indicating that in the minutes following phagocytosis, production of superoxide by NoxA and rapid phagosomal acidification are both necessary for efficient killing of ingested bacteria. While live bacteria persisted longer in phagosomes in these mutant cells, they were eventually all killed after 15 min in noxA KO cells and 6 min in NH₄Cl‐treated cells, indicating that redundant killing mechanisms can ensure killing of ingested bacteria even when one effector is inactivated.

FIGURE 5.

AlyL, BpiC, NoxA and acidic pH are necessary for efficient permeabilization of Klebsiella pneumoniae in phagosomes but only NoxA and acidic pH are required for efficient killing. Bacterial killing and permeabilization in phagosomes were measured as described in Figure 3. Genetic inactivation of alyL (a) and bpiC (b) delayed bacterial permeabilization but not bacterial death (a: N = 14 independent experiments, WT n = 661, alyL n = 690; b: N = 10, WT n = 616, bpiC n = 666). Genetic inactivation of noxA (c) and addition of NH₄Cl (d), delayed both bacterial killing and permeabilization (c: N = 11, WT n = 689, noxA n = 730; d: N = 4, WT n = 334, NH₄CL n = 396). The statistical analysis between wild type and mutants were performed for each 3 min period using the total number of events analized at each time period (N _CatA + N _CatB + N _CatC + N _CatD) and the total number of regrowth events (N _CatA). *p < .05, Fisher's exact test.

2.7. Phagosomal acidification precedes permeabilization of bacteria

Phagosomes are rapidly acidified following ingestion of bacteria by D. discoideum cells. In order to determine when acidification of phagosomes is initiated, we used specific anti‐LPS antibodies (Crespo‐Yanez & Ayadi, 2022) to decorate the surface of K. pneumoniae bacteria with pH‐sensitive FITC fluorescence. Following bacterial ingestion, the FITC fluorescence decreased rapidly, indicating that phagosomes acidified within 2–3 min of their formation (Figure 6a). Similar kinetics of acidification were seen in all the mutants analyzed in the current study (Figure 6b,c) with only a minor delay (≈ 1 min) in kil1 KO cells and kil2 KO cells. Note that extinction of FITC fluorescence indicates with exquisite precision the initiation of phagosome acidification. It does not provide an accurate measurement of the very acidic pH (<3) reached in D. discoideum phagosomes (Bodinier et al., 2020; Marchetti et al., 2009) since FITC fluorescence is almost fully quenched at pH 5. This explains why even NH₄Cl treatment did not produce significant changes in the acidification kinetics in phagosomes (Figure 6c): as described earlier NH₄Cl only partially increases the pH of lysosomal compartments in D. discoideum (Marchetti et al., 2009), an effect not sufficient to prevent the quenching of FITC. Together these observations indicate that phagosome acidification is initiated a few minutes after formation of phagosomes in both WT and D. discoideum mutant cells.

FIGURE 6.

Rapid acidification of Klebsiella pneumoniae‐containing phagosomes. To visualize the acidification of the phagosome lumen, we imaged Dictyostelium discoideum cells mixed with FITC labeled K. pneumoniae. (a) Representative images showing a K. pneumoniae coated with FITC‐labeled antibodies ingested by a D. discoideum cell. The white arrowhead indicates the time of phagocytosis. FITC extinction took place 3 min after ingestion. Scale bar 10 μm. (b) Loss of FITC fluorescence was measured as a function of time following ingestion in WT and noxA KO D. discoideum cells. FITC fluorescence decreased with a half‐life of approximately 3 min (N = 4, n = 120 for each cell type). (c) The half‐life of FITC fluorescence in phagosomes was determined in D. discoideum (WT, kil1 KO, kil2 KO, noxA KO, alyL KO and bpiC KO cells) and in WT cell with the addition of NH4Cl. Each point represents the result of an independent experiment. The mean and S.E.M. are indicated. *p < .05, Kruskal–Wallis test. (WT: N = 13 independent experiments; kil1 KO: N = 9; kil2 KO: N = 7; NH₄Cl: N = 6; noxA KO: N = 4; alyL KO: N = 4; bpiC KO: N = 4).

In order to determine whether the initiation of phagosome acidification takes place before, after or during permeabilization of bacteria, we used K. pneumoniae bacteria expressing a red fluorescent protein mCherry, which is destroyed in phagosomes with kinetics very similar to those observed for GFP‐expressing bacteria (Supplementary Figure S5). We then observed the ingestion of these bacteria after labeling their surface with FITC‐coupled antibodies (Figure 7a). We observed 100 ingestion events, and in every case the disappearance of FITC fluorescence preceded the disappearance of mCherry fluorescence (Figure 7a,b column A). These observations indicate that the acidification of D. discoideum phagosomes precedes the permeabilization of the ingested bacteria.

FIGURE 7.

Acidification of phagosomes precedes permeabilization of Klebsiella pneumoniae. (a) In order to measure in the same experiment the permeabilization of K. pneumoniae (i.e., the loss of cytosolic mCherry) and the acidification of the phagosome lumen (i.e., the loss of surface FITC fluorescence), we mixed Dictyostelium discoideum cells with FITC labeled K. pneumoniae and recorded ingestion, FITC extinction and permeabilization of K. pneumoniae. In the example shown the FITC fluorescence disappeared 2 min after ingestion, and mCherry fluorescence 37 min after ingestion (indicated with a star). (b) We counted the number of instances where FITC fluorescence disappeared after (G1), before (G2) or simultaneously with (G3) permeabilization of bacteria. We observed in one experiment the ingestion of 100 bacteria coated with FITC‐labeled antibodies, and ingestion of 52 bacteria labeled by direct coupling with FITC. Loss of FITC fluorescence always preceded loss of mCherry fluorescence, indicating that permeabilization of bacteria occurred after acidification of phagosomes.

Binding of antibodies to the bacterial surface may, in principle, alter recognition of the bacteria by the D. discoideum cell and the maturation of the phagosomes. To exclude this possibility, we assessed the acidification dynamics using bacteria that were labeled by direct linkage of FITC to the bacterial surface. We observed 52 ingestion events and like with antibody‐coated bacteria, extinction of FITC always preceded permeabilization of bacteria visualized by extinction of mCherry (Figure 7b column B). Together these experiments clearly indicate that acidification of phagosomes is initiated before the permeabilization of bacteria.

3. DISCUSSION

As described in the Graphical Abstract, in this work, we dissected the evolution of bacteria in D. discoideum phagosomes into three successive phases: killing, permeabilization, and digestion. Shortly after ingestion (2–3 min), bacteria are killed by a process requiring phagosomal acidification, the activity of the Kil2 putative magnesium pump, and production of superoxide by the NoxA NADPH oxidase. During the second phase, within 10 min, the action of lysosomal proteins, in particular AlyL and BpiC leads to the abrupt extinction of the GFP contained in bacteria. AlyL is a lysozyme capable of digesting bacterial peptidoglycans between the two bacterial membranes, but its antibacterial activity is mainly due to its membrane‐permeabilization domain, composed of four amphipathic helices (Jauslin et al., 2021). BpiC interacts with bacterial LPS and increases bacterial permeability. Since both AlyL and BpiC target components of the bacterial envelope and act by permeabilizing bacterial membranes, the second phase in the evolution of bacteria is most likely a permeabilization phase. The sudden extinction of the intra‐bacterial GFP fluorescence is presumably caused by the loss of bacterial membrane integrity, exposing the bacterial cytosol to the very acidic phagosomal pH and to lysosomal enzymes. The fact that permeabilization of ingested bacteria is also slower in noxA KO cells and in ammonium chloride‐treated cells than in WT cells may indicate either that free radicals and phagosomal acidity also participate at a later stage in the permeabilization of ingested bacteria, or that slowing down killing delays the onset of permeabilization, for example by delaying delivery of lysosomal enzymes to phagosomes. The third phase reflects the gradual digestion of bacterial components, in particular DNA, by lysosomal enzymes. It extends over a period of 1 h. All three phases of bacterial processing in the phagosome are slowed down in kil1 KO and kil2 KO cells, stressing the importance of these two gene products in the overall bactericidal function of the phagosome. Kil1 is a sulfotransferase and sulfation is a common trait of many lysosomal enzymes, potentially necessary both for their lysosomal targeting and for their enzymatic activity (Benghezal et al., 2006). Kil2 is a putative magnesium pump present in the phagosomal membrane and its absence was previously shown to decrease the proteolytic activity in phagosomes (Lelong et al., 2011). Since magnesium is a cofactor for many enzymes it seems likely that Kil2, like Kil1, is required for the function of many different antibacterial proteins. It is thus not surprising to observe that both Kil1 and Kil2 are necessary at all stages of the phagosomal evolution.

In our study, we used a model phagocytic cell (D. dictyostelium amoebae) and specific methods for analysis to follow the fate of GFP‐expressing K. pneumoniae (bacterial regrowth to determine viability; GFP extinction to measure permeabilization of the bacteria, DAPI to measure integrity of bacterial DNA). In general terms, our conclusions are similar to those reached in a previous study of intracellular killing of GFP‐expressing Pseudomonas aeruginosa in mammalian macrophages (Riazanski et al., 2020). This earlier study identified two distinct phases in phagosomes. First, bacteria lost their GFP fluorescence and were simultaneously permeabilized, and this process required the production of free radicals by NADPH oxidase. Second, bacteria were delivered to acidic phagolysosomes and exposed to digestive enzymes. The permeabilization phase observed in this study is presumably equivalent to the permeabilization of bacteria measured in our study by the loss of their GFP fluorescence, and which was indeed slowed down in the absence of noxA activity. In addition, we report that loss of membrane‐permeabilizing AlyL or BpiC, also delayed bacterial permeabilization. Moreover, our results indicates that killing of bacteria can occur prior to their permeabilization and that it requires the activity of noxA and phagosomal acidification. Together these two studies indicate that killing, permeabilization and digestion of bacteria in phagosomes are remarkably similar in D. discoideum amoebae and in mammalian macrophages. This observation suggests that conclusions drawn from studies in model D. discoideum phagocytes are also relevant in mammalian macrophages. The main bactericidal proteins identified so far in D. discoideum (NoxA, AlyL, BpiC) have clear orthologues in mammalian macrophages (Nox2, lysozyme, Bpi) and it seems likely that our conclusions regarding their sequential roles can be extrapolated to mammalian macrophages.

It seems likely that the classification of phagosome maturation into three phases is still an oversimplification. The development of more refined methods would certainly allow to dissect with more precision the events occurring in phagosomes. For example, a better time resolution and more precise indicators may allow to determine whether acidification of the phagosome and production of free radicals occur simultaneously or consecutively. Finally, this work was entirely focused on how a non‐pathogenic K. pneumoniae strain (Lima et al., 2018) is destroyed in maturing phagosomes. The tools developed in this study can also be used to determine how pathogenic strains of K. pneumoniae behave in the phagocytic pathway, and how specific bacterial features allow pathogens to escape killing and permeabilization.

4. EXPERIMENTAL PROCEDURES

4.1. Cell culture and strains

D. discoideum cells were cultured in HL5 medium (Froquet et al., 2009) at 21°C and subcultured twice a week to maintain a maximal cellular density of 10⁶ cells/ml. D. discoideum mutant strains used in this study were derived from the DH1‐10 subclone (Cornillon et al., 2000) of the D. discoideum strain DH1 (Caterina et al., 1994), referred to in this study as wild‐type (WT).

The kil1, kil2, kil1kil2, alyL, noxA, and bpiC knockout strains described previously (Jauslin et al., 2021) were created by deleting a portion of the gene of interest in WT cells and replacing it with a blasticidin resistance (BSR) cassette.

The K. pneumoniae strain KpGe (Lima et al., 2018) was transfected with a plasmid conferring resistance to kanamycin (Bodinier et al., 2020) and constitutively expressing codon‐optimized yeast‐enhanced GFP (yEGFP) (pZAZ‐prGFP) or mCherry (pZE27MC3). They were grown at 37°C in LB medium supplemented with kanamycin (50 mg/L).

4.2. Assessing intracellular digestion of bacteria by live microscopy

GFP‐expressing K. pneumoniae were labeled with DAPI: 4′,6‐diamidino‐2‐phenylindole, dihydrochloride dissolved in deionized water (dH2O) to a stock concentration of a 5 mg/ml (Invitrogen #D1306), as described previously (Johnson & Criss, 2013b). Briefly, bacteria were grown for 16 h. Bacteria (1 ml) were washed once in 1 ml of Soerensen's phospate Buffer (SB: 2 mM Na₂HPO₄, 14.7 mM KH₂PO₄, pH 6) supplemented with 100 mM Sorbitol (SBS). Bacteria were resuspended in 1 ml of SBS and then diluted at 1/100 in SBS to a final volume of 1 ml. 50 μg of DAPI were added and the suspension was incubated during 20 min at room temperature in the dark. Bacteria were washed again with 1 ml of SBS, and 150 μl of the suspension was transferred to a glass‐bottom well (μ‐slide 8‐well, IBIDI). 7 × 10⁵ D. discoideum cells were washed with 1 ml of SBS, resuspended in 1 ml of SBS and 100 μl were added to each well. The mixed bacteria and D. discoideum cells were allowed to sediment for 10 min at 21°C. To image the whole cell volume at each timepoint, an image (brightfield, DAPI and GFP fluorescence) was taken in five successive focal planes with a step size of 3 μm every 30 s for 2 h with a Nikon eclipse Ti2 widefield time‐lapse microscope equipped with a DS‐Qi2 camera. The NIS software was used to extract the images, and Fiji to compile and analyze movies. A series of images started when an individual bacterium was ingested by a D. discoideum cell and ended when the fluorescence of the phagocytosed bacteria was fully extinguished. Along each series, the GFP and DAPI fluorescence intensities of the bacteria were measured, normalized to the fluorescence at the time of ingestion and plotted as a function of time. When using GFP‐expressing bacteria, the DAPI fluorescence was decreased by Fluorescence Energy Transfer to GFP imposing a significant level of correction (Supplementary Figure S1).

4.3. Combined analysis of intracellular killing and permeabilization of bacteria

The aim of this procedure was to assess simultaneously bacterial death (= loss of its ability to grow) and bacterial permeabilization (= extinction of GFP fluorescence). Preliminary tests revealed that rapid growth of K. pneumoniae in our experimental setup could only be observed reproducibly when using exponentially growing bacteria, as opposed to an overnight culture of bacteria (Supplementary Figure S2). Hence GFP‐expressing K. pneumoniae grown overnight were diluted 1/1000 in 2 ml of fresh LB with kanamycin and cultured during 6 h at 37°C, when they reached exponential growth (Supplementary Figure S6). They were then washed and mixed with D. discoideum cells in SBS in a glass‐bottom well as described above. When indicated, SBS was supplemented with 40 mM NH₄Cl. The cells were then imaged, and the images analyzed as described above. During the first phase of the experiment, ingestion of bacteria and extinction of GFP fluorescence were assessed for 25 min. After 25 min, SBS was removed and replaced by LB medium containing 0.002% (v/v) Triton X‐100, 0.002% (w/v) paraformaldehyde and kanamycin (50 μg/ml). This medium was optimized to kill D. discoideum cells without lysing them and to allow growth of K. pneumoniae. During the second phase of the experiment, cells were incubated at 37°C and imaged every 10 min for 15 h to visualize growth and division of bacteria. The NIS software was used to extract the images, and Fiji software was used to compile and analyze movies.

We defined seven theoretical categories to classify our observations. Intraphagosomal bacteria that were still fluorescent at the end of phase 1 were classified into category A (if they grew during phase 2), category B (if they did not grow during phase 2 but did not lose their fluorescence) or category C (if they did not grow and lost their fluorescence during phase 2). Intraphagosomal bacteria that were not fluorescent at the end of phase 1 where classified into category D (if they did not grow during phase 2 and did not recover their fluorescence), category E (if they grew without recovering their fluorescence during phae 2), category F (if they recovered their fluorescence but did not grow during phase 2) or category G (if they recovered their fluorescence and grew during phase 2).

For kinetic analysis, for each bacterium time 0 was the time of its phagocytosis, the time of residency inside the phagosome was recorded as well as the moment at which GFP fluorescence was lost. The kinetics of bacterial permeabilization (= GFP extinction) during phase 1 were determined using the Kaplan–Meier estimator on Prism version 7.0.a Graphpad. These values were plotted as the % of GFP‐positive bacteria and represent the probability of being GFP‐positive (P_GFP) as a function of time. Using the events of categories A, B, and C we calculated the kinetics of killing. For each 3‐min time interval, the probability of growth (P _growth) was calculated by counting the number of events in category A (N _CatA) divided by the total number of events (N _CatA + N _CatB + N _CatC) and multiplied by the probability of being GFP‐positive (P _GFP). The probability of growth was corrected by the percentage of growth of extracellular bacteria (P _{growth_EC}),

$$P_{\text{growth}}=\left(\frac{N_{\text{CatA}}}{N_{\text{CatA}}+N_{\text{CatB}}+N_{\text{CatC}}}\times P_{\mathrm{GFP}}\right)/P_{\text{growth}\_\mathrm{EC}.}$$

This calculation revealed the percentage of bacteria capable of dividing at a given time after phagocytosis.

The statistical analysis between wild type and mutants were performed for each 3 min time interval using the total number of events analyzed (N _CatA + N _CatB + N _CatC + N _CatD) and the total number of regrowth events (N _CatA). *p < .05, Fisher's exact test.

4.4. Combined analysis of intraphagosomal acidification and permeabilization of bacteria

The acidification of a phagosome containing a K. pneumoniae bacteria was determined by labeling the surface of K. pneumoniae with pH‐sensitive FITC. This labeling was achieved by following two methods: (1) coating with FITC‐labeled antibodies. Briefly, bacteria were washed in 1 ml of SBS and resuspended in 1 ml of SBS. Bacteria were diluted in SBS at 1/100 and then incubated for 10 min with 5 μg/ml of the AI516‐mouseFc antibody (ABCD Antibodies; abcd‐antibodies.com) recognizing K. pneumoniae LPS (Crespo‐Yanez & Ayadi, 2022), washed once with 1 ml of SBS, resuspended in 400 μl of SBS and then incubated for 10 min with FITC‐conjugated goat anti‐mouse IgG secondary antibody (1/200; Invitrogen A16067), then washed 3 times with 1 ml of SBS and resuspended in 200 μl of SBS. After that, 150 μl of the suspension was deposited in a glass‐bottom well (μ‐slide 8‐well, IBIDI). (2) To couple directly FITC to the bacterial surface, bacteria were washed once in PBS (phosphate buffered saline, pH 7.4) and resuspended in 1 ml of PBS, and then they were diluted in PBS at 1/100 and incubated for 30 min with 0.25 mg/ml of fluorescein‐5‐isothiocyanate (stock solution dissolved in anhydrous dimethil sulfoxide at 5 ml/ml; ThermoFisher Scientific #F1906) at room temperature in the dark, washed once with 1 ml of PBS containing 40 mM NH₄Cl and twice with PBS, then once with SBS, and finally they were resuspended in 1 ml of SBS. 150 μl of the FITC‐labeled bacteria were deposited in a glass‐bottom well (μ‐slide 8‐well, IBIDI), then mixed 2.3 × 10⁵ D. discoideum cells, and allowed to sediment for 10 min and imaged as described above.

AUTHOR CONTRIBUTIONS

Xènia Crespo‐Yanez: Conceptualization; Investigation; Writing ‐ original draft; Methodology; Validation; Visualization; Writing ‐ review & editing; Formal analysis; Software; Data curation. Joseph Oddy: Writing ‐ review & editing; Formal analysis; Conceptualization; Investigation. Otmane Lamrabet: Conceptualization; Investigation; Formal analysis; Writing ‐ review & editing. Tania Jauslin: Conceptualization; Investigation; Writing ‐ review & editing; Formal analysis. Anna Marchetti: Formal analysis. Pierre Cosson: Conceptualization; Funding acquisition; Writing ‐ original draft; Validation; Writing ‐ review & editing; Project administration; Supervision; Data curation; Resources

CONFLICT OF INTEREST

The authors declare no competing or financial interests.

ETHICS APPROVAL AND CONSENT TO PARTICIPATE

Not applicable.

Supporting information

Figures S1‐S6

Movie S1

Crespo‐Yanez, X. , Oddy, J. , Lamrabet, O. , Jauslin, T. , Marchetti, A. & Cosson, P. (2023). Sequential action of antibacterial effectors in Dictyostelium discoideum phagosomes. Molecular Microbiology, 119, 74–85. 10.1111/mmi.15004

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request and/or will be available in a public repository that issues datasets with DOIs.