The late endosomal adaptor p14 is a macrophage host-defense factor against Salmonella infection

Recently, a novel primary immunodeficiency syndrome was identified and traced back to a homozygous single point mutation in the human LAMTOR2 (p14) gene, causing reduced p14 protein expression and subsequently severe neutropenia and aberrant lysosomal function in cells of both, the innate and adaptive immune system (Bohn et al., 2007). Consequently, affected patients suffer from recurrent respiratory infections with Streptococcus pneumoniae, which occur secondary due to defective intracellular organelle fusion in neutrophils, B and T cells. Thus, we proposed that p14 might be of importance in host defense against intracellular pathogens because p14 regulates late endosomal biogenesis and activation of local extracellular signal-regulated kinase (ERK). ERK activation takes place in response to bacterial infections and is involved in the induction of the oxidative and nitrosative stress as major antimicrobial immune defense mechanisms (Bogdan, 2001).

Macrophages are in the first line of host defense against pathogenic microbes. These professional phagocytes have evolved a repertoire of antimicrobial mechanisms based on the formation of toxic radicals, including NADPH phagocytic oxidase (phox), which is responsible for the generation of reactive oxygen species (ROS), and inducible nitric oxide (NO) synthase (iNOS), which generates reactive nitrogen species (RNS). The phagolysosome as the ultimate microbicidal organelle is designed for the containment of pathogens followed by radical mediated killing of certain intracellular microorganisms. Despite its essential importance for endosomal fusion events, a putative role of p14 in macrophage effector functions and host defense against intracellular pathogens has not been investigated before.

In co-evolution with mammalian hosts, bacterial pathogens have evolved sophisticated strategies to manipulate and escape from the host antimicrobial defense system. Salmonella enterica serovar Typhimurium (Salmonella Typhimurium) for instance, is a facultatively intracellular bacterium, which can enter and replicate within a variety of host cells in a special membrane-bound compartment, the Salmonella-containing vacuole (SCV) causing systemic diseases in mice. Salmonella delivers a cohort of effector proteins into the host cell cytosol via the bacterial type III secretion system, which is encoded on separate genomic regions; the Salmonella pathogenicity islands 1 (SPI-1) and 2 (SPI-2), respectively. SPI-1-encoded proteins are crucial for bacterial invasion and early SCV maturation, whereas SPI-2 effectors expressed a few hours after invasion are implicated in late SCV maturation, intracellular bacterial survival and replication. SCV biogenesis has been extensively studied in a variety of host cells (professional phagocytic and non-phagocytic cells) and is defined by continuous and dynamic interactions with the host endosomal system. Initially, the SCV associates transiently with markers of early endosomes, such as early endosomal antigen 1 (EEA1), which are then rapidly replaced by late endosomal markers, including Rab family GTPase 7 (Rab7), lysosome-associated membrane protein 1 (LAMP1) and the vacuolar ATPase (Méresse et al., 1999; Steele-Mortimer et al., 1999).

The scaffold complex consisting of p14, MEK1 partner (MP1) and MAPK and ERK kinase 1 (MEK1) is localized to late endosomes where it is required for the efficient activation of ERK signaling (Wunderlich et al., 2001; Teis et al., 2002; Kurzbauer et al., 2004). Furthermore, conditional gene disruption revealed a specific and essential function of the endosomal scaffold complex p14–MP1–MEK1 in vivo by regulating late endosomal traffic and cellular proliferation (Teis et al., 2006).

Because endosomal fusion events and mitogen activated protein kinase (MAPK) signaling in general are also essential for macrophage function and immunity against intracellular microbes, we investigated the specific role of p14 in Salmonella-infected murine macrophages. To test whether p14 functions as a host defense factor, we took advantage of two different cell systems, bone-marrow-derived macrophages (BMDMs) and mouse embryonic fibroblasts (MEFs), both deficient in p14, which served as an in vitro model for professional phagocytic and non-phagocytic cells, respectively. Additionally, we performed Salmonella infection studies in vivo using conditional knockout mice expressing the Cre recombinase under the control of the lysozyme M promoter, which allows the specific deletion of Lamtor2 (LMCp14^−/−) in the monocyte or macrophage cell lineage.

This study reveals a crucial role for the adaptor protein p14 in host defense against Salmonella Typhimurium. p14 deficiency correlates with the loss of three essential requirements to battle Salmonella infections. First, p14 is needed for the efficient activation of ERK on late endosomes. Second, p14 is required for the accurately timed transport of the Salmonella through the endolysosomal system. Third, the loss of p14 is accompanied by a defect in the activation of NADPH oxidase and iNOS, two central antimicrobial effector systems. Thus, p14 supports the accurate transport of the Salmonella through the endolysosomal system thereby regulating bacterial replication in both, professional phagocytes and in non-phagocytic cells in vitro and helps mice to successfully battle Salmonella infections in vivo.

To investigate the specific role of p14 in murine macrophages during infection with intracellular pathogens, we established an infection model using BMDMs from LMCp14^+/+ and LMCp14^−/− mice and the facultatively intracellular pathogen Salmonella Typhimurium. First, we performed colony forming unit (CFU) assays to examine the intracellular bacterial load in isogenic wild-type (wt, p14^+/+) and p14^−/− macrophages 6 hours and 24 hours after Salmonella infection (Fig. 1A). Although no statistically significant difference in the number of CFUs was observed at 6 hours after infection, 24 hours after infection, p14^−/− macrophages (Fig. 1A, black bar) displayed a fivefold (P=0.0016, unpaired Student's t-test) increase in the number of intracellular Salmonella. Immunofluorescence microscopy of this time point revealed an accumulation of Salmonella in p14^−/− macrophages. Notably, we could observe a substantially smaller size, probably reflecting partial degradation of Salmonella, in wt but not in p14^−/− macrophages (Fig. 1B). Furthermore, general differences in the arrangement of the actin cytoskeleton were noticed already in uninfected wt and p14^−/− macrophages (Fig. 1B, 0h). However, the altered actin cytoskeleton of p14^−/− macrophages had no effect on early phagocytic events (supplementary material Fig. S1A,C), which are mainly dependent on actin dynamics. The accumulation of Salmonella in p14^−/− macrophages could be confirmed by cryo-electron microscopy (EM). In p14 macrophages, vacuolar compartments contained only singular, more or less degraded Salmonella (Fig. 1C, upper panel, asterisks) after 6 h of Salmonella infection, together with the fluid tracer bovine serum albumin (BSA)–gold, as a result of their regular fusions with default endocytic pathways. By contrast, p14^−/− macrophages constantly displayed characteristic clusters of Salmonella within huge SCVs (Fig. 1C, lower panel, SCVs outlined by arrowheads) that contained clearly less BSA–gold than observed in the controls. Furthermore, a quantification of the cellular bacterial loads after 24 hours of infection showed that 60% of p14 deficient macrophages displayed four or more Salmonella per cell compared with only 19% in the wild-type macrophages (Fig. 1D, black bar).

Because the intracellular bacterial numbers depend on the relative contributions of bacterial replication and bacterial elimination by the host, we determined the bacterial load by comparing a replication-deficient Salmonella strain (SPI-2 mutant) with its congenic wt Salmonella strain (Fig. 1E). At the early time point, 6 hours of Salmonella infection, the number of wt or SPI-2-deficient bacterial colonies showed no difference between wt and p14^−/− macrophages. However, after 24 hours of infection, the differences in the bacterial load between p14^−/− and p14 macrophages observed with the wt bacterial strain disappeared following infection with the Salmonella SPI-2 mutant strain, suggesting that p14 affects the bacterial life cycle at late stages when bacterial replication within macrophages takes place (Fig. 1E).

Alternatively, we infected non-phagocytic cells, p14^f/– and p14^−/− MEFs, with Salmonella to distinguish the contribution of bacterial replication from host-mediated killing of pathogens. Consistent with the results from the primary macrophages, p14^−/− MEFs showed a fivefold (P=0.0008, unpaired Student's t-test) increase in the bacterial load after 24 hours of Salmonella infection compared with wt (p14^f/–) MEFs (Fig. 2A, black bars). This observation was also confirmed by immunofluorescence staining in which p14^−/− MEFs displayed higher loads of intracellular Salmonella after 24 hours of infection (Fig. 2B). Notably, in p14^−/− MEFs, we observed general changes in the organization of the actin cytoskeleton independent of Salmonella infection (Fig. 2B, top panel), which had no influence on Salmonella uptake (supplementary material Fig. S1D). On the basis of these results, we conclude that p14 is one of the components of the cellular host defense system required to limit intracellular Salmonella replication.

Next, we analyzed whether the increased intracellular bacterial numbers seen in p14^−/− cells resulted from an enhanced bacterial uptake. Previous studies have shown that cytoskeletal proteins such as actin, as well as structural proteins including the multi-domain protein IQGAP1 (Brandt et al., 2007; Brown et al., 2007), are involved in the formation of the phagocytic cup. Therefore, the localization of these proteins in Salmonella-infected p14^+/+ and p14^−/− macrophages was examined by confocal microscopy (supplementary material Fig. S1A). On the basis of these data, p14 appears not to be involved in phagocytic cup formation. After 10 to 30 minutes of infection, EEA1 is recruited to the Salmonella-containing compartment, which is relevant for the subsequent maturation of the compartment (Steele-Mortimer et al., 1999). Confocal images of Salmonella-infected p14^+/+ and p14^−/− macrophages revealed no differences in the transport to EEA1 positive compartments (supplementary material Fig. S1B). Additionally, quantification of Salmonella uptake (number of bacteria per cell) after 10 minutes of exposure revealed no differences in p14^−/− versus p14^+/+ macrophages (supplementary material Fig. S1C). In agreement, the uptake of Salmonella in non-phagocytic cells (MEFs) was also not dependent on the presence of p14 (supplementary material Fig. S1D). Thus, the p14–MP1 scaffold complex is not required for the uptake or the early trafficking of Salmonella in primary macrophages.

Previous studies have revealed that ERK signaling plays an important role at several phases during Salmonella infection. On the one hand, host cells activate ERK through Toll-like receptors upon infection (Zu et al., 2009), and ERK signaling triggers the production of several pro-inflammatory mediators (Brereton et al., 2009; Kasper et al., 2010). On the other hand, Salmonella expresses effector proteins to subvert ERK-mediated signaling (Rosenberger and Finlay, 2002; Lin et al., 2003). Therefore, we wanted to investigate whether p14 regulates ERK activation in macrophages upon Salmonella infection, and whether p14 interferes with the transport of Salmonella to late endosomal compartments.

As a first step, we demonstrated the presence of the scaffold complex p14–MP1 on isolated late phagolysosomes by flotation of these compartments using latex beads (supplementary material Fig. S2A). Both proteins, p14 and MP1, were only detectable on late phagolysosomes from p14^+/+ and p14^+/− macrophages but were absent in p14^−/− macrophages. Additionally, immuno-electron microscopy (EM) confirmed the localization of the p14–MP1 anchoring protein p18 (supplementary material Fig. S2B, arrows) (Nada et al., 2009) at the limiting membrane of SCVs in p14^−/− macrophages and MEFs.

As a result of the presence of the p14–MP1 scaffold on late endosomes or phagosomes, we next performed pulse–chase experiments and stained p14^+/+ and p14^−/− macrophages for Salmonella, LAMP1 (late endosomes) and phosphorylated ERK and examined the intracellular transport of bacteria by confocal microscopy (Fig. 3A). After 10 minutes of infection, ERK activation was similar in both p14^+/+ and p14^−/− macrophages, and Salmonella had not yet reached LAMP1-positive compartments. However, upon 60 minutes of chase, Salmonella were transported to LAMP1-positive compartments simultaneously causing an activation of ERK in p14^+/+ macrophages (Fig. 3A, middle panel, left). By contrast, p14 deficiency delayed the acquisition of LAMP1 on Salmonella-containing compartments. Thus, Salmonella only reached this compartment after a pulse of 360 minutes (Fig. 3A, bottom panel, right), whereas in wt controls, Salmonella entered within the first hour of infection. Concomitantly, no ERK activation on LAMP1-positive compartments could be observed in p14^−/− macrophages.

Immunoblot analysis confirmed a decrease in ERK activation in Salmonella infected p14^−/− macrophages after 180 and 360 minutes upon continuous Salmonella exposure compared with p14^+/+ macrophages (Fig. 3B). These data suggest that late endosomal ERK activation triggered by the p14–MP1 scaffold complex is correlated with the timed transport of Salmonella to LAMP1-positive structures. Thus, the absence of p14 from late endosomes or phagosomes delays but does not completely block the fusion of Salmonella-containing vesicles with LAMP1-positive late endosomes or phagosomes (Fig. 3A, bottom panel; supplementary material Fig. S3).

During maturation, phagosomes acquire a full arsenal of antimicrobial features, including the assembly of the NADPH oxidase. One key step during activation is the translocation of the cytoplasmic subunit p47 to the membrane of the bacteria-containing compartment. Interestingly, earlier studies have shown that ERK could phosphorylate the p47 subunit, thereby initiating its translocation (Dewas et al., 2000). Therefore, we performed confocal microscopy of infected macrophages and stained for Salmonella, p47-phox and LAMP1 upon 60 minutes of Salmonella exposure (Fig. 3C). Although p47-phox was efficiently recruited to the Salmonella phagosome (LAMP1 positive) in wt macrophages, p47-phox failed to target the Salmonella-containing compartment (LAMP1 negative) in p14^−/− macrophages (Fig. 3C).

These observations suggest that p14 is needed for the full activation of ERK on Salmonella-containing vesicles, thereby initiating the assembly of the NADPH oxidase. The loss of the endosomal ERK signaling is accompanied by a delay of Salmonella trafficking to LAMP1-positive structures but does not block the fusion of the Salmonella-containing compartments with late endosomes or lysosomes.

The late SCV, in which bacterial replication takes place, is enriched in LAMPs but is reported to lack general markers of the phagolysosome, such as the mannose-6-phosphate receptor (CI-6MPR) or lysosomal hydrolases, including Cathepsin D (CathD) (Holden, 2002; Smith et al., 2007). To determine whether p14 is required for the transport of Salmonella to phagolysosomal compartments, p14^+/+ and p14^−/− macrophages were labeled for Salmonella and CathD or CI-M6PR after 24 hours of infection and examined by confocal microscopy (Fig. 4A). Interestingly, whereas in control macrophages Salmonella-containing compartments were strongly positive for both endolysosomal markers, CathD and CI-6MPR were generally less prominent or absent on structures containing Salmonella in p14-deficient macrophages. Immuno-EM confirmed these observations. CI-M6PR and LAMP1 clearly localized to SCVs of p14^+/+ macrophages, whereas, in p14^−/− macrophages, CI-M6PR labeling appeared strongly reduced or even absent in these vacuoles, despite its presence in neighboring compartments (supplementary material Fig. S3).

One of the main characteristics of a phagolysosomal compartment is its acidic environment, which is essential for the activation of lysosomal hydrolases such as CathD. To address whether Salmonella reaches acidic, degrading compartments, we incubated wt and p14^−/− macrophages with the weak base DAMP according to an established protocol (Orci et al., 1987). Immuno-EM showed distinct, specific DNP-labeling in both, wt and p14^−/− macrophages, indicating an acidic environment of these compartments (supplementary material Fig. S4). Concomitantly, by means of immuno-EM we observed strong cathepsin D labeling on Salmonella-containing vacuoles in wt macrophages, but a clearly reduced labeling intensity in p14^−/− macrophages (supplementary material Fig. S4).

The infection by Salmonella results in the massive remodeling of the endosomal system of the eukaryotic host cell and the formation of long tubular endosomal compartments, so-called Salmonella-induced filaments (SIFs). The formation and maintenance of SIFs in a perinuclear region is a highly dynamic process that is required for efficient intracellular proliferation of Salmonella. In both, wt and p14^−/− macrophages, the formation of juxtanuclear SIFs could be observed as shown by cryo-EM (supplementary material Fig. S5). Because of the highly dynamic process of SIF extension and contraction during different phases of Salmonella infection (Rajashekar et al., 2008) and the methods applied in this study, we were not able to detect differences in the number of SIFs or in their specific subcellular localization.

Phagocytic cells also form NO to exert antimicrobial activity against intracellular pathogens, apart from the contributions of NADPH oxidase and the endolysosomal system to host resistance (Bogdan, 2001). By investigating the expression of the NO generating enzyme iNOS by quantitative reverse transcription PCR we found a significant decrease in iNOS mRNA levels in p14^−/− compared with p14^+/+ macrophages at 6 hours (P=0.004, unpaired Student's t-test) and 24 hours (P=0.006, unpaired Student's t-test) after Salmonella infection (Fig. 4B). These results were confirmed by immuno-blot analysis with macrophages infected for 3, 6 and 24 hours (Fig. 4C). Indeed, p14-deficient macrophages exhibited a severe defect in the induction of iNOS protein expression at 6 and 24 hours after infection compared with their wild-type counterparts.

Next, we asked whether p14 plays a general role in the biogenesis of phagolysosomes or if its function is limited to Salmonella infections. Therefore, we performed pulse chase experiments using Alexa-Fluor-488-labeled yeast (zymosan), stained p14^+/+ and p14^−/− macrophages for LAMP1 (late endosomes) and examined the intracellular transport of yeast by confocal microscopy (supplementary material Fig. S7A). After 6 hours of chase, yeast particles were found to accumulate in LAMP1-positive structures demonstrating reduced degradation in p14^−/− macrophages compared with wt counterparts. Furthermore, EM showing yeast (Saccharomyces cerevisiae) and the fluid phase marker BSA within the same vacuolar compartment suggests that the defect in degradation in p14^−/− macrophages does not result from a fusion defect (supplementary material Fig. S7B). These experiments indicate that p14 plays a general role in the biogenesis of phagosomes that is not restricted to Salmonella infections. Thus, p14 is required for the function and induction of the three major antimicrobial systems comprising the phagolysosome, NADPH oxidase and iNOS.

Because p14-deficient cells provide a better niche for Salmonella replication in vitro, we were interested whether LMCp14^−/− mice were able to control a Salmonella infection in vivo. Therefore, C57BL/6 as well as LMCp14^+/+ and LMCp14^−/− mice on the same genetic background were inoculated intraperitoneally (i.p.) with 200 CFUs of Salmonella. The median survival time of LMCp14^−/− mice was markedly reduced compared with LMCp14^+/+ or C57BL/6 mice (96 versus 150 or 156 hours, P<0.001 when compared by log-rank test; Fig. 5A).

For further analysis, mice were inoculated i.p. with 500 CFUs of Salmonella and sacrificed after 72 hours. Upon histopathological examination LMCp14^+/+ animals exhibited several well-circumscribed inflammatory loci in liver (Fig. 5B, top panel) and spleen (supplementary material Fig. S6A, top panel). By contrast, LMCp14^−/− mice showed large necrotic loci in the liver. These inflammatory loci were largely formed by granulocytes and monocytes, which were surrounded by recruited macrophages. Immunofluorescence analysis of cryosections of liver (Fig. 5B) and spleen (supplementary material Fig. S6A) from infected mice showed that Mac-1 (monocytes)-, Gr-1 (granulocytes)- and F4-80 (macrophages)-positive cells were also recruited to these inflammatory loci. By contrast, LMCp14^−/− mice displayed large necrotic loci in the liver (Fig. 5B, top panel), which were structurally less distinct, though positive for monocytes and macrophages (Fig. 5B). In the spleen of the LMCp14^−/− mice, many small inflammatory loci were formed, which were identified as being mainly monocytes, granulocytes and macrophages (supplementary material Fig. S6A).

Intriguingly, monocytes, granulocytes and macrophages in liver (Fig. 5B, bottom panel) and spleen (supplementary material Fig. S6A) from LMCp14^−/− mice displayed higher numbers of intracellular Salmonella than their wt counterparts. This observation was also reflected by a significant increase (P=0.001, unpaired Student's t-test of log-transformed results) of bacterial colonies per gram liver tissue as assessed by CFU assays (Fig. 5C) in LMCp14^−/− mice versus LMCp14^+/+ mice. A similar increase of bacterial colonies per gram tissue was also found in the spleen, although less pronounced (supplementary material Fig. S6B). Additionally, iNOS induction was impaired in vivo, as shown in spleen lysates from Salmonella-infected LMCp14^−/− mice compared with control mice (Fig. 5D).

In summary, we demonstrated that the late endosomal adaptor protein p14 is required for three major lines of host defense. First, p14-mediated ERK activation on late endosomes is needed for the delivery of Salmonella into the phagolysosome. Second, p14 is required for the translocation of the p47 subunit to late endosomes, which is essential for the activation of the NADPH oxidase and third, for the induction of the late antimicrobial effector pathway through iNOS-mediated formation of NO. Thus, p14-mediated signaling is required to fight Salmonella infections in vitro and in vivo.

The outcome of an infectious disease is critically dependent on the balance between host resistance and bacterial virulence factors. In this study, we introduce the late endosomal scaffold complex p14/MP1 as one of the cellular host defense mechanisms against the intracellular pathogen Salmonella Typhimurium. The endosomal adaptor protein p14 impedes the replication of intracellular Salmonella in professional phagocytes by promoting its sorting into the phagolysosome. The loss of p14 resulted in impaired control of intracellular bacterial proliferation at later stages of infection (24 hours), when bacterial effectors (SPI-2) important for survival and replication are expressed. Accordingly, a replication-deficient SPI-2 mutant Salmonellastrain did not behave differently in p14^−/− and p14^+/+ macrophages. In MEFs, which are unable to actively kill Salmonella because of the low expression of antimicrobial effector molecules, highly elevated numbers of intracellular Salmonella were found in p14-deficient cells compared with their wild-type counterparts. Importantly, we could exclude an increased uptake of Salmonella as direct cause of higher bacterial loads both in p14-deficient macrophages and MEFs.

Salmonella is known to reside in a vacuolar compartment, the SCV, a protected niche, in which it can replicate. The trafficking of Salmonella and thus the biogenesis of SCVs is defined by a continuous interaction of Salmonella with the host endosomal system, which is controlled in a time- and space-dependent manner. Indeed, we found that the loss of p14 interferes with the kinetics of late SCV formation. Although the formation of late SCVs was not completely blocked by the loss of p14, macrophages deficient in p14 showed a delay in the acquisition of LAMP1 to the Salmonella-containing compartment accompanied by the loss of late endosomal ERK activation (Fig. 6). In fact, studies showed that a delay in SCV trafficking, similar to retention of Rab5 or Rab7 downstream effectors, already delays the transport to lysosomes and might favor the formation of a replication-friendly niche for Salmonella (Hashim et al., 2000; Marsman et al., 2004). A defect in the transport of activated cell surface receptors, such as epidermal growth factor receptor (EGFR), to late endosomes was observed previously by our group. Therefore, late endosomal sorting is a specific function of the p14–MP1 signaling complex (Teis et al., 2006). Thus, the delay of Salmonella transport in p14-deficient cells might open a time window giving Salmonella a chance to proliferate in a more replication-friendly niche.

The ERK signaling pathway is an important mediator during Salmonella infection and is targeted by both the host cell and the bacterium. It has been shown that ERK signaling is not important for phagocytic uptake of Salmonella (Procyk et al., 1999), but is required at later stages of infection because MEK kinase inhibition increased the number of intracellular Salmonella (Rosenberger and Finlay, 2002). Additionally, several Salmonella effector proteins, such as SpvC and SptP, located on SPI-1 and SPI-2, respectively, are known to inhibit ERK activation facilitating enhanced pathogen replication in SCV or promoting pathogen virulence (Lin et al., 2003; Mazurkiewicz et al., 2008; Humphreys et al., 2009). Here, we demonstrate that in particular the late endosomal ERK activation mediated by the p14–Mp1 complex is required for the timed transport of the Salmonella to late phagolysosomes.

ERK activation plays an important role for NADPH phox activation, too. Phox is composed of two major membrane proteins, gp91-phox and p22-phox, which comprise the heterodimeric flavocytochrome b558 and three cytosolic proteins, p47-phox, p67-phox and Rac (reviewed by Fang, 2004). Phosphorylation of the cytosolic subunit p47 by ERK enables its translocation and thereby its interaction with the membrane bound flavocytochrome b558, which is important for the function of the NADPH oxidase (Dewas et al., 2000; Qian et al., 2008). In fact, we observed that the translocation of the cytosolic p47-phox to the Salmonella-containing compartment was inhibited in p14-deficient macrophages, which might be caused by the inefficient late endosomal ERK activation in these cells (Fig. 6). Accordingly, it has been shown that both MEK kinase and phox activities impair bacterial replication (Rosenberger and Finlay, 2002). Furthermore, mice lacking p91 or p47 subunits are more susceptible to Salmonella infections, and macrophages from these mice fail to efficiently kill the ingested bacteria (Mastroeni et al., 2000; Vazquez-Torres et al., 2000a; van Diepen et al., 2002).

With respect to the interaction with the lysosomal system, the SCV apparently deviates from the default phagosome maturation pathway. This late stage in SCV maturation is not fully understood, yet. On the one hand, studies using primary macrophages demonstrated that SCVs do fuse with the lysosomal compartment (Mills and Finlay, 1998; Garvis et al., 2001; Mukherjee et al., 2002). On the other hand, however, studies indicated that despite expressing the phagosomal marker LAMP1, SCVs do segregate and avoid fusion with mature lysosomes to establish a replicative niche for the Salmonella. Therefore, late SCVs lack certain late endosomal or lysosomal markers, such as the CI-M6PR, or lysosomal hydrolases, such as Cathepsin D (Garcia-del Portillo and Finlay, 1995; Holden, 2002; Smith et al., 2007). Generally, Salmonella occurring in lysosomal-marker-positive SCVs can be regarded as those that failed to escape from host cell defense, leading to SCV–lysosome fusion and subsequent bacterial killing. Although the loss of p14 does not completely block the fusion of SCVs with the degrading compartments, it impairs Salmonella transport and locates bacteria preferentially into a compartment with reduced CI-M6PR and CathD labeling compared with wild-type cells. This delayed transport appears to open a time window for Salmonella to proliferate in a more replication-friendly environment, resulting in higher intracellular bacterial numbers (Fig. 6).

Macrophages kill Salmonella in a dynamic process that requires the generation of both ROS and RNS. Accordingly, we found reduced iNOS mRNA and protein expression in p14-deficient macrophages. This antimicrobial defense mechanism, relevant for Salmonella elimination within macrophages, is initiated by a plethora of signaling cascades, including MAPK (Caivano, 1998; Vazquez-Torres et al., 2000b) signaling pathways. Thus, the inefficient activation of iNOS in p14^−/− macrophages might result from the reduced ERK activation at late Salmonella-containing compartments (Fig. 6). Similarly, iNOS-knockout mice were more susceptible to Salmonella infection, and macrophages from these animals displayed a decreased ability to kill intracellular Salmonella (Mastroeni et al., 2000; Vazquez-Torres et al., 2000a). Interestingly, it has been shown that RNS synthesized during high NO fluxes can repress the phoP–phoQ system responsible for survival of intracellular Salmonella (Bourret et al., 2009). Thus, the low expression levels of iNOS and subsequently of NO in p14^−/− macrophages might favor the survival and replication of Salmonella.

The importance of the late endosomal scaffold complex p14–Mp1 for host defense is further emphasized by the increased mortality of the LMCp14^−/− mice upon Salmonella infection. The high bacterial loads in the liver and spleen of these mice resulted from an uncontrolled replication of Salmonella in p14-deficient macrophages in vivo. Furthermore, the importance of p14 in host defense is also reflected by the inefficient killing of extracellular pathogens by phagocytic cells from patients having reduced p14 protein levels (Bohn et al., 2007).

Here, we demonstrated that the late endosomal adaptor protein p14 supports the host defense mechanism against Salmonella Typhimurium infection by contributing to the accurate trafficking of the Salmonella within the host endosomal system. The adaptor p14 is needed for the full activation of late endosomal ERK on Salmonella-containing compartments, thereby initiating the assembly of the NADPH oxidase and its immediate delivery to the phagolysosomal system. Conversely, the loss of p14 opens a time window for Salmonella to remain in a replication-friendly niche (with low CathD and CI-M6PR levels), in which the late antimicrobial effector enzyme iNOS is not yet appropriately induced. Thus, p14-mediated signaling supports the accurate trafficking of Salmonella through the endosomal system in both professional phagocytes and in non-phagocytic cells in vitro, and helps mice to successfully combat Salmonella infections in vivo (Fig. 6).

The balance between pathogen degradation and replication depends on a remarkably complex interplay between effector proteins from both the Salmonella and the host. On the host side, several proteins, including kinases such as Akt (Kuijl et al., 2007) or PtdIns 5-kinase (Kerr et al., 2010), as well as several Rab GTPases (Drecktrah et al., 2007) or cell surface receptors, such as SLAM (Berger et al., 2010), were identified as components of the host defense mechanism that inhibit SCV maturation and Salmonella replication. However, Salmonella has evolved sophisticated strategies to manipulate host defense mechanisms (Nairz et al., 2010). The coordinated expression of certain virulence genes allows bacteria to maintain contacts with the host endosomal and signal transduction system to establish a replicative niche and assure their own survival. Owing to the complexity of the interplay between host and Salmonella, some open questions still remain. Therefore, the identification of new interaction points between host and Salmonella is a priority for investigating new therapeutics.

Neutrophils from patients with LAMTOR2 (p14) gene mutations displayed a severe deficiency in killing the extracellular pathogens Pseudomonas aeruginosa and Escherichia coli (Bohn et al., 2007). Here, we demonstrate that p14–Mp1-mediated signaling is also required for the defense against the intracellular pathogen Salmonella Typhimurium in primary macrophages and in fibroblasts. Based on our findings, the p14–MP1 complex is involved in the accurate transport of intracellular pathogens through the endolysosomal system, thereby preventing uncontrolled bacterial replication. Although further studies are needed to dissect the relative contributions of p14-initiated antimicrobial mechanisms, we could show for the first time that the scaffold complex p14–MP1 is required to control Salmonella replication by regulating accurate Salmonella trafficking correlated with late endosomal ERK activation, which in turn, might initiate the antimicrobial effectors NADPH oxidase and iNOS. Thus, the late endosomal adaptor p14 is an essential component of the cellular host defense machinery in the fight against intracellular pathogens.

ERK1/2 and phosphorylated ERK1/2 antibodies were purchased from Cell Signaling. The anti-mouse CD107a (LAMP1), iNOS, Gr-1 and Mac-1 antibodies were obtained from BD Pharmingen, and antibodies against EEA1 and IQGAP1 were from Santa Cruz. Anti-P14 and anti-MP1 antibodies were described previously (Teis et al., 2002). The anti-CSA-1 (to detect Salmonella) antibody was purchased from KPL and the anti-F4/80 antibody was from Serotec. Antibodies against actin and α-tubulin and Hoechst XXXXX stain were obtained from Sigma. The secondary antibodies (Alexa Fluor 488, Alexa Fluor 568 and Alexa Fluor 647), Alexa-Fluor-568–phalloidin and Alexa-Fluor-488–zymosan were obtained from Invitrogen.

Conditional deletion of p14 in lysozyme-M-positive cells was performed by crossing p14^f/f mice (Teis et al., 2006) with LysMCre transgenic mice (Clausen et al., 1999) and by further breeding to generate p14f/f;LysMCre (LMCp14^−/−) animals.

Age- and sex-matched mice were sacrificed, bone marrow cells were recovered from tibiae and femora and erythrocytes were lysed. After 24 hours of incubation, non-adherent cells were collected and cultured on non-tissue culture dishes for selection in RPMI supplemented with 10% fetal calf serum (FCS), 50 IU/ml penicillin and 30% L929 conditioned medium as a source of CSF-1. Cells used in assays were grown for 7–10 days in culture and were transferred to tissue culture plates overnight before use (n=number of independent BMDM preparations per genotype). MEFs were grown in high-glucose DMEM (PAA) supplemented with 50 IU/ml penicillin, 50/ml streptomycin and 10% FCS (PAA). Generation of p14^f/– and p14^−/− MEFs was described previously (Teis et al., 2006).

Cells were lysed in lysis buffer [50 mM Tris-HCl (pH 7.5), 200 mM NaCl, 1% Triton X-100, 10% glycerol, 0.5 mM EDTA, 0.5 mM EGTA, 2 mM Na₃OV₄, 50 mM NaF and proteinase inhibitors (aprotinin, pepstatin, leupeptin, Pefabloc SC)], lysates were separated by SDS-PAGE, blotted and probed with the respective antibodies.

Immunofluorescence analysis of MEFs and macrophages was performed as previously described (Taub et al., 2007). Unless otherwise indicated, pulse–chase experiments were performed as follows. Before infection, cells were incubated in complete medium without antibiotics for at least 5 hours. BMDMs (7.5×10⁵) and MEFs (1×10⁵) were infected at a multiplicity of infection (MOI) of 10. After 10 minutes (or 1 hour) of Salmonella incubation (pulse), cells were washed with complete medium without antibiotics but supplemented with 0.05% gentamicin (50 mg/ml, PAA) and incubated for the indicated time periods (chase) in order to kill all extracellular bacteria. Cells were washed with phosphate buffered saline (PBS) containing gentamicin and immediately fixed on ice for 10 minutes with 4% paraformaldehyde, permeabilized in 0.2% Triton X-100 and blocked in gelatin containing blocking buffer for 30 minutes. Cells were incubated for 1.5 hours with the primary antibody, washed five times and incubated for 40 minutes with the secondary antibody and mounted in Mowiol (Calbiochem).

Isolated liver and spleen pieces were embedded in optimal cutting temperature Tissue-Tek on dry ice. Cryosections (4 μm) were air-dried and fixed in 4% paraformaldehyde. After washing, sections were blocked in blocking buffer (15% goat serum, 1% BSA) supplemented with 0.2% Triton X-100 for 30 minutes at room temperature (RT) and then incubated for 1 hour with the primary antibody, washed and incubated with the secondary antibody. Sections were counterstained with Hoechst 33258 (Sigma) and mounted in Mowiol.

Sample processing and immunogold labeling were performed as described previously (Taub et al., 2007). Briefly, for morphology cells were cultured on sapphire coverslips and incubated with Salmonella (MOI: 10) for 6 or 24 hours (or with yeast for 3 hours) plus 5-nm-gold-labeled BSA for the last 30 minutes of Salmonella incubation. Subsequently, samples were processed by high-pressure freezing, freeze-substitution and resin embedding. For immuno-EM of thawed cryosections (from conventionally cultured, chemically fixed BMDMs and MEFs), the following primary antibodies were used: rat LAMP1 (1D4B, from the Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA), rabbit CI-MPR (courtesy of Bernard Hoflack, Biotechnology Center of the TU Dresden, Dresden, Germany), rabbit cathepsin D (courtesy of Kurt v. Figura, Georg-August-University of Göttingen, Göttingen, Germany and Stefan Höning, Institute for Biochemistry I, University of Cologne, Köln, Germany), rabbit p18=LAMTOR1 (HPA002997, from Atlas Antibodies), rabbit DNP (Invitrogen). DNP-labeling was used to localize DAMP accumulating within acidic compartments according to published methods (Orci et al., 1987).

Preparation of total RNA and quantification of mRNA expression by qRT-PCR was performed as previously described. Probes used for Taqman qRT-PCR carried 5′FAM and 3′BHQ1 labels. Sequences of primers and probes spanned exon–intron boundaries: mu-iNOS: 5′-CAGCTGGGCTGTACAAACCTT-3′, 5′-CATTGGAAGTGAAGCGTTTCG-3′, 5′-CGGGCAGCCTGTGAGACCTTTGA-3′, mu-Hprt: 5′-GACCGGTCCCGTCATGC-3′, 5′-TCATAACCTGGTTCATCATCGC-3′, 5′-ACCCGCAGTCCCAGCGTCGTC-3′ (Nairz et al., 2009a).

Primary BMDMs or MEFs were used for in vitro infection assays. Wild-type Salmonella enterica serovar Typhimurium strain ATCC14028 was used for all experiments and grown under sterile conditions in LB broth (Sigma) to logarithmic phase (5 hours), unless otherwise indicated. The congenic Salmonella SPI-2 mutant was generated as described previously and grown in parallel to its parental wild-type strain for 16–18 hours in LB broth to late log state (Vazquez-Torres et al., 2000b). Before infection, cells were incubated in complete medium without antibiotics for at least 5 hours. For CFU experiments, BMDMs (7.5×10⁵) and MEFs (1×10⁵) were infected at a multiplicity of infection (MOI) of 10. After 1 hour of incubation with Salmonella, cells were washed with complete medium without antibiotics but supplemented with 0.05% gentamicin (50 mg/ml) and incubated for another 3 hours to kill all extracellular bacteria. Afterwards, gentamicin levels were reduced to 0.025%. After the indicated time period (6 hours and 24 hours) cells were washed with PBS, lysed in 0.5% deoxycholic acid (Sigma) and plated under sterile conditions in appropriate dilutions onto LB agar plates. Each experiment was performed three times for each time point.

All in vivo experiments were performed with good animal practice according to the guidelines of the relevant local and national animal welfare bodies. All animal experiments were approved by the advisory board for animal experimentation of the Medical University of Innsbruck and by the Austrian Ministry for Science and Research. Mice were kept under specific pathogen-free conditions at the central animal facilities of the Medical University of Innsbruck. Sex- and age-matched mice were infected i.p. with 500 CFUs of Salmonella diluted in 200 μl of PBS and were sacrificed 72 hours after infection. Mice were monitored twice daily for signs of illness. Bacterial load of spleen and liver was determined by serial dilutions of organ homogenates on LB agar under sterile conditions and the number of bacteria per gram tissue was calculated. For statistical analysis, results were log-transformed before comparison by the unpaired Student's t-test. Additionally, dissected liver and spleen pieces were embedded in optimal cutting temperature Tissue-Tek on dry ice and subjected to immunofluorescence analysis. For survival studies, sex- and age-matched mice were infected i.p. with 200 CFUs of Salmonella diluted in 200 μl of PBS (Nairz et al., 2009b). Mice were monitored twice daily until death, and after 336 hours, the experiments were terminated.

BMDMs from LMCp14^+/+, LMCp14^+/− and LMCp14^−/− mice were incubated with blue latex beads (0.80 μm, 1:200) for 6 hours. After 16 hours, cells were washed and incubated for another 2 hours before LPL preparation. Cells were washed and scraped with ice-cold PBS. All solutions used were supplemented with proteinase inhibitor cocktail. Next, cells were harvested, homogenized (27G needle) in homogenization buffer [250 mM sucrose, 3 mM imidazole (pH 7.4), 1 mM EDTA]. After centrifugation, the supernatant was loaded onto a sucrose gradient [2 ml of 62% sucrose, diluted (1:2, 62% sucrose) sample, 2 ml of 35% sucrose, 2 ml of 25% sucrose, remainder 10% sucrose solution]. Samples were centrifuged for 1 hour 15 min at 34,000 r.p.m. at 4°C. The blue interface was collected and washed with buffer [3 mM imidazole, 1 mM EDTA] by centrifuging for another 50 minutes at 34,000 r.p.m. at 4°C. The blue pellet was resuspended in hot protein sample buffer and subjected to SDS-PAGE.

In addition to the primary acquisition by the Zeiss software, images were converted to Adobe Photoshop CS3. Brightness, contrast or tonal value was improved, and figures were arranged with CorelDraw X5 and exported as jpeg files. Confocal image analysis was performed using a Zeiss LSM 510 confocal laser-scanning microscope. Epi-fluorescence image analysis was performed using a Zeiss Axiovert Imager M1.

All animal experiments were approved by the advisory board for animal experimentation of the Medical University of Innsbruck and by the Austrian ministry for Science and Research based on the Austrian Animal Testing Act of 1988 (permit number BMWF-66.011/0114-II/3b/2011).

We are grateful to David Teis for critically discussing the manuscript, Ferric C. Fang for providing us with SPI-2 deficient Salmonella Typhimurium strain and Bernard Hoflack and Stefan Höning for providing us antibodies against CI-MP6R and cathepsin D, respectively. We thank Karin Gutleben, Barbara Witting and Hannes Ebner for excellent technical assistance with immuno-electron microscopy.