Rapid Pathogen-Induced Apoptosis: A Mechanism Used by Dendritic Cells to Limit Intracellular Replication of Legionella pneumophila

Dendritic cells (DCs) are specialized phagocytes that internalize exogenous antigens and microbes at peripheral sites, and then migrate to lymphatic organs to display foreign peptides to naïve T cells. There are several examples where DCs have been shown to be more efficient at restricting the intracellular replication of pathogens compared to macrophages, a property that could prevent DCs from enhancing pathogen dissemination. To understand DC responses to pathogens, we investigated the mechanisms by which mouse DCs are able to restrict replication of the intracellular pathogen Legionella pneumophila. We show that both DCs and macrophages have the ability to interfere with L. pneumophila replication through a cell death pathway mediated by caspase-1 and Naip5. L. pneumophila that avoided Naip5-dependent responses, however, showed robust replication in macrophages but remained unable to replicate in DCs. Apoptotic cell death mediated by caspase-3 was found to occur much earlier in DCs following infection by L. pneumophila compared to macrophages infected similarly. Eliminating the pro-apoptotic proteins Bax and Bak or overproducing the anti-apoptotic protein Bcl-2 were both found to restore L. pneumophila replication in DCs. Thus, DCs have a microbial response pathway that rapidly activates apoptosis to limit pathogen replication.


Introduction
Macrophages and dendritic cells (DCs) are the sentinels of the innate immune system. They are key in sensing infection and activating downstream antimicrobial responses [1,2]. These professional phagocytes are activated following stimulation of patternrecognition receptors, such as transmembrane Toll-like receptors (TLRs) and cytoplasmic nucleotide-binding domain and leucinerich repeat containing receptors (NLRs) by pathogen-associated molecular patterns (PAMPs). Signaling through these receptors induces the expression and secretion of proinflammatory cytokines, chemokines and other antimicrobial defense molecules [3][4][5][6][7].
Bacterial pathogens that are able to infect and establish residence within macrophages and DCs provide a unique challenge to the innate immune system, as many pathogens have evolved virulence factors that subvert the cellular processes of these cells. One such pathogen is Legionella pneumophila, the etiological agent of the severe pneumonia known as Legionnaires' disease [8,9]. L. pneumophila is able to infect alveolar macrophages and modulate transport of the phagosome in which it resides to avoid fusion with endosomes and lysosomes [10]. L. pneumophila has the ability to recruit vesicles in transit between the endoplasmic reticulum (ER) and Golgi apparatus and use these vesicles to remodel the L. pneumophilacontaining vacuole (LCV) to create a unique ER-derived vacuole that supports intracellular replication [10][11][12][13][14][15][16][17]. Modulation of intracellular transport of the LCV requires a functional type IV secretion system (TFSS) encoded by the dot and icm genes, which translocates bacterial effectors directly into the host cytosol [18][19][20][21]. Many of the translocated effector proteins engage host factors involved in vesicular transport and assist in LCV transport [22][23][24][25][26][27][28][29]. L. pneumophila mutants defective in the Dot/Icm system do not replicate intracellularly, as they are unable to modulate intracellular transport and occupy a more conventional phagosome that undergoes rapid endocytic maturation [19,30].
Although L. pneumophila has evolved sophisticated strategies to overtake phagocytic host cells, the mammalian innate immune system is able to efficiently control bacterial infection and replication. Responses controlled by the TLR adaptor protein MyD88 effectively clear L. pneumophila from the lungs of infected mice [31][32][33]. NLRs also contribute to the detection and control of infection. Naip5 (Birc1e), an NLR encoded within the Lgn1 locus, limits replication of L. pneumophila in mouse macrophages [34][35][36][37]. Naip5 is activated by a Dot/Icm-dependent signaling event that presumably involves the delivery of the bacterial protein flagellin into the host cell cytosol [38][39][40]. Naip5 in conjunction with the NLR protein Ipaf activates caspase-1, which limits L. pneumophila replication in macrophages by inducing a pro-inflammatory cell death pathway known as pyroptosis [37][38][39][40].
Naip5 control of caspase-1 activation does not seem to be the only cellular mechanism used by innate immune cells to control L. pneumophila replication. In DCs infected with L. pneumophila, although phagosomes containing bacteria are able to mature into ER-derived organelles, bacterial replication is limited [41]. DCs are still able to process and present L. pneumophila antigens on MHC class II molecules, and de novo synthesis of L. pneumophila proteins inside DCs is critical for maximal stimulation of CD4 + T cells. This indicates that restriction of L. pneumophila replication could be important to the ability of DCs to present bacterial antigens to T cells and direct subsequent adaptive immune responses [41]. Interestingly, DCs are able to limit the intracellular replication of several other pathogens that are capable of replicating in macrophages, such as Listeria monocytogenes, Mycobacterium tuberculosis and Salmonella enterica Serovar Typhimurium [42][43][44][45].
Thus, it appears that there are inherent differences between DCs and macrophages with respect to their abilities to restrict replication of intracellular pathogens. We show here that one of these differences involves the ability of DCs to rapidly activate a cell intrinsic apoptotic cell death pathway in response to the intracellular pathogen L. pneumophila.

Results
Canonical pathogen surveillance pathways are not required for restriction of L. pneumophila replication by DCs Signaling through TLRs in macrophages results in enhanced phagocytosis and phagosome fusion with lysosomes [46]. Thus, innate immune recognition of L. pneumophila could activate cellular processes that control bacterial replication in DCs. Cells deficient in the adapters MyD88 or Rip2 were used to interfere with the TLR and Nod signaling pathways respectively, to determine whether L. pneumophila replication in DCs is restricted by activation of signaling pathways controlled by innate immune receptors. Replication of L. pneumophila was not detected in DCs derived from A/J mice, which are defective for Naip5 signaling, or from A/J-derived mice deficient in either MyD88 or Rip2 ( Figure 1A). By contrast, exponential replication of L. pneumophila occurred in the macrophages derived from these mice ( Figure 1A). L. pneumophila intracellular replication was not observed in DCs derived from mice deficient in both MyD88

Author Summary
The immune system is designed to identify microbes that enter the body and elicit responses that prevent the replication and dissemination of these organisms. Dendritic cells play an important role in regulating host immunity to pathogens. Their phagocytic capacity enables DCs to internalize and destroy most microbes, and the ability of DCs to migrate to specialized lymphoid organs is important for inducing antigen-specific immunity. Here, we analyzed interactions between DCs and Legionella pneumophila, a bacterial pathogen that can subvert phagocytic host cell functions to create a vacuole that permits intracellular replication. We found that L. pneumophila infection rapidly induced DCs to commit cell death through apoptosis. Rapid apoptosis was not observed after infection of macrophages, which are the phagocytic cells that support L. pneumophila replication in the lungs of infected animals. Using cells derived from knockout mice, we found that DCs deficient in the proteins Bax and Bak, which are essential for induction of the apoptosis pathway, were unable to restrict the intracellular replication of L. pneumophila. Likewise, overproduction of Bcl-2, which is a negative regulator of apoptosis, resulted in DCs that were permissive for L. pneumophila replication. These data indicate DCs have the ability to rapidly undergo apoptosis when infected with a microbe capable of replicating intracellularly, and this response effectively prevents pathogen replication. We hypothesize that this response may be designed to interfere with the migration of infected DCs through the lymphatic system, which would prevent DCs from serving as a ''Trojan Horse'' that transports pathogenic microbes from peripheral sites to central organs. and Trif (data not shown), indicating that the lack of both of these TLR adaptor proteins did not restore L. pneumophila intracellular replication in DCs. Thus, DC restriction of L. pneumophila replication does not require TLR signaling through MyD88 and Trif or Nod1/2 signaling through Rip2.
Mouse macrophages restrict L. pneumophila replication by inducing a cell death pathway controlled by Naip5 and caspase-1 [37]. Mouse macrophages become permissive for L. pneumophila replication if they are homozygous for the permissive Naip5 gene encoded in the A/J mouse or if caspase-1 is absent [35][36][37]. Intracellular replication of L. pneumophila was examined in DCs derived from Naip5-deficient mice to determine if the Naip5 protein produced by A/J-derived DCs retained an activity sufficient to restrict replication. The Naip5 2/2 DCs did not support replication of L. pneumophila ( Figure S1). It remained possible that proteins other than Naip5 might activate a caspase-1dependent pathway that prevented L. pneumophila replication in DCs from A/J mice. To test this possibility, L. pneumophila replication was measured in DCs derived from caspase-1-deficient mice homozygous for the A/J Naip5 allele (Casp1 2/2 ). L. pneumophila replication was not detected in Casp1 2/2 DCs, whereas, L. pneumophila replication was similar in BMMs from these same Casp1 2/2 and caspase-1-sufficient mice (Casp1 +/+ ) ( Figure 1B). Thus, Naip5 and caspase-1 are not required for DC restriction of L. pneumophila replication.

DC apoptosis occurs rapidly after L. pneumophila infection
Although caspase-1-mediated cell death was not required for DCs to restrict the replication of L. pneumophila, it remained possible that another cell death pathway could be important for this process. Thus, we examined whether apoptosis occurred upon L. pneumophila infection of DCs. TdT-mediated dUTP-biotin nick end-labeling (TUNEL) analysis was performed on DCs infected for 6 hours with either wild type (WT) L. pneumophila or the isogenic DdotA strain that has a nonfunctional Dot/Icm secretion system. Examination of DCs that had internalized WT L. pneumophila revealed that 37% were TUNEL positive (Figure 2A and 2B, top panel). Only 1% of DCs containing the DdotA strain were TUNEL positive (Figure 2A and 2B, top panel). The majority of DCs were TUNEL positive following induction of apoptosis with staurosporine (staur), a broadspectrum protein kinase inhibitor ( Figure 2B, bottom panel). Similar results were obtained using Casp1 2/2 DCs ( Figure S2), indicating that the absence of caspase-1 did not prevent apoptosis in DCs infected with L. pneumophila.
Macrophages and DCs were infected with WT L. pneumophila to compare the kinetics of apoptosis. At 1-hour post infection, infected DCs became TUNEL positive, whereas, TUNEL-positive macrophages were not apparent until12-hours post infection ( Figure 2C). In addition to using TUNEL staining, the kinetics of apoptosis was determined by measuring caspase-3/7 activity in DCs and macrophages after L. pneumophila infection. At 4-hours post infection there was a significant Dot/Icm-dependent increase in caspase-3/7 activity in DC extracts, but not in corresponding macrophage extracts ( Figure S3). A significant increase in Caspase-3/7 activity was not observed for macrophages until 11-hours post infection ( Figure S3). Thus, apoptosis in DCs was induced by L. pneumophila with faster kinetics than in similarly infected macrophages.
Caspase-3-mediated effector responses are induced by L. pneumophila after DC infection Caspase-3 mediates many of the downstream effector responses in the apoptotic cell death pathway, including fragmentation of DNA in the nucleus [47]. Caspase-3-deficient mice (Casp3 2/2 ) were used to determine whether DNA fragmentation induced after L. pneumophila infection of DCs was due to induction of the apoptotic cell death pathway. TUNEL analysis performed on DCs derived from Casp3 2/2 and Casp3 +/+ mice 6 hours after infection with L. pneumophila revealed that 57% of the infected Casp3 +/+ DCs were TUNEL positive, whereas only 9.5% of Casp3 2/2 DCs infected with WT L. pneumophila were TUNEL positive ( Figure 3A, left panel and 3B). Both Casp3 2/2 and Casp3 +/+ DCs infected with the DdotA strain showed minimal TUNEL staining ( Figure 3A, right panel and 3B). Thus, L. pneumophila infection of DCs rapidly activates downstream components of the apoptotic cell death pathway.

Caspase-3 is involved in DC restriction of L. pneumophila replication
To determine whether activation of the apoptotic cell death pathway was important for DC restriction of L. pneumophila replication, DCs from A/J-derived Casp3 2/2 and Casp3 +/+ mice infected with WT L. pneumophila were examined by fluorescence microscopy. The efficiency of L. pneumophila internalization determined 2 hours after infection was equivalent for Casp3 2/2 and Casp3 +/+ DCs ( Figure 4A, top panel). When DCs were examined 10 hours after infection, there was a significant increase in the percentage of infected Casp3 2/2 DCs that contained vacuoles supporting L. pneumophila replication (R.V.) (19%) compared to Casp3 +/+ DCs (6%) ( Figure 4A, bottom panel). Representative images in Figure 4A show that the number of L. pneumophila in vacuoles that supported replication was higher in Casp3 2/2 DCs, and that most of the infected Casp3 +/+ DCs had condensed or fragmented nuclei. These data were corroborated by determining colony-forming units (CFUs) over time. There was roughly a 10-fold increase in L. pneumophila CFUs 72 hours after Casp3 2/2 DCs were infected with WT L. pneumophila compared to a slight decrease in CFUs recovered from Casp3 +/+ DCs at 72 hours ( Figure 4C). DCs eliminated the DdotA strain with equal efficiency. Macrophages derived from these mice were infected in parallel. The infected Casp3 +/+ macrophages had normal nuclei ( Figure 4B) and supported L. pneumophila replication to similar levels as the Casp3 2/2 macrophages ( Figure 4B and 4D). These data indicate that caspase-3 plays a role in restricting L. pneumophila replication in DCs, but not macrophages.

Cell death mediated by Bax and Bak restricts L. pneumophila replication in DCs
Bax and Bak play a central role in regulating apoptosis. When activated by members of the BH3-only protein family, Bax and Bak create a channel in the membrane of mitochondria that releases cytochrome c. This results in activation of the apoptosome and the subsequent activation of effector caspases, such as caspase-3 [48][49][50][51]. DCs derived from C57BL/6 (B6) and from mice deficient in Bak (Bak 2/2 ) or both Bax and Bak (Bax 2/2 Bak 2/2 ) were analyzed to determine if Bax and Bak have a role in cell death induced by L. pneumophila. TUNEL analysis demonstrated that WT L. pneumophila induced equivalent levels of cell death in DCs derived from B6 and Bax 2/2 Bak 2/2 mice ( Figure 5A), suggesting that the Naip5dependent pathway of cell death remained functional in DCs. A L. pneumophila strain containing an in-frame deletion of the flaA gene encoding flagellin was used to bypass Naip5-mediated cell death [38][39][40]. A dramatic reduction in cell death was observed for Bax 2/2 Bak 2/2 DCs infected with L. pneumophila DflaA ( Figure 5A). Measurements of caspase-3/7 activity following infection of DCs confirmed that Bax and Bak were required for induction of apoptosis by L. pneumophila DflaA (Table 1). Thus, L. pneumophila independently induces DC cell death by a Bax/Bak-dependent pathway and a Naip5-dependent pathway.
Replication of WT L. pneumophila was not detected in either Bak 2/2 or Bax 2/2 Bak 2/2 DCs ( Figure 5B), which is consistent with the Naip5-mediated pathway being operational in these cells. L. pneumophila DflaA replicated to similar levels as WT L. pneumophila in DCs derived from Casp3 2/2 mice homozygous for the A/J Naip5 allele ( Figure S4), indicating that eliminating flagellin does not significantly enhance the capacity of L. pneumophila to replicate in DCs with a genetic defect in the Naip5 signaling pathway. DCs from Bax 2/2 Bak 2/2 mice supported replication of L. pneumophila DflaA, whereas, replication of L.
The development of vacuoles containing replicating L. pneumophila DflaA was evaluated in DCs derived from B6, Casp3 2/2 and Bax 2/2 Bak 2/2 mice. Vacuoles containing replicating L. pneumophila DflaA were detectable in both Casp3 2/2 , and Bax 2/2 Bak 2/2 DCs at 8-hours post infection ( Figure 6A). Large vacuoles containing .10 L. pneumophila DflaA were frequent in the Bax 2/2 Bak 2/2 DCs at 12-hours post infection, but were found less frequently in the Casp3 2/2 DCs ( Figure 6A). Although Casp3 2/2 DCs exhibited enhanced resistance to cell death induced by L. pneumophila DflaA, they were not as resistant to cell death as the Bax 2/2 Bak 2/2 DCs ( Figure 6B), which likely explains why the Bax 2/2 Bak 2/2 DCs were slightly more permissive for replication of L. pneumophila DflaA at 12-hours post infection compared to the Casp3 2/2 DCs. These data indicate L. pneumophila activation of the intrinsic cell death pathway in DCs is sufficient to limit intracellular replication.

Bcl-2 overproduction antagonizes restriction of L. pneumophila replication by DCs
Bcl-2 is a pro-survival protein that regulates apoptosis [52,53]. Overexpression of pro-survival proteins such as those from the Bcl-2 family can block mitochondria membrane permeabilization and prevent apoptosis [54][55][56]. DCs from transgenic mice expressing human BCL2 under the control of the CD68 promoter (Tg(bcl2) 535rm) (Jamieson & Medhzitov, unpublished data) were used to determine whether overproduction of Bcl-2 could interfere with the ability of DCs to restrict L. pneumophila replication. Immunoblot analysis confirmed that both macrophages and DCs derived from Tg(bcl2) 535rm mice produced human Bcl-2, and that overproduction of Bcl-2 did not affect the levels of Bax and Bak in these cells ( Figure 7A). Replication of WT L. pneumophila was not observed in Tg(bcl2) 535rm DCs, presumably because these cells produce a functional Naip5 protein ( Figure 7B). Replication of the DflaA strain was observed in the Tg(bcl2) 535rm DCs, but not in control DCs from B6 mice ( Figure 7B). Single cell analysis confirmed replication of the DflaA strain in Tg(bcl2) 535rm DCs ( Figure 7C). At 10-hours post infection, 21% of the DflaA-infected Tg(bcl2) 535rm DCs had large vacuoles containing replicating L. pneumophila, and most of the infected Tg(bcl2) 535rm DCs were devoid of apoptotic features, such as condensed and fragmented nuclei, that were observed in infected control DCs derived from B6 mice ( Figure 7C). TUNEL staining confirmed that the Tg(bcl2) 535rm DCs were more resistant to apoptosis after infection by L. pneumophila DflaA compared to control B6 DCs ( Figure 7D). Thus, Bcl-2 overproduction limited DC apoptosis in response to L. pneumophila and resulted in enhanced intracellular replication.
DCs have a unique ability to efficiently restrict L. pneumophila replication by apoptosis Macrophages derived from Bax 2/2 Bak 2/2 and Tg(bcl2) 535rm mice were used to determine whether rapid induction of programmed cell death as a mechanism to restrict L. pneumophila replication was an exclusive property displayed by DCs. Replication of WT L. pneumophila was restricted by the Bax 2/ 2 Bak 2/2 macrophages and Tg(bcl2) 535rm macrophages as efficiently as control B6 macrophages ( Figure 8A and 8B). When the DflaA strain was used to bypass Naip5-mediated growth restriction, bacterial replication was not enhanced in the Bax 2/2 Bak 2/2 macrophages or Tg(bcl2) 535rm macrophages compared to control B6 macrophages ( Figure 8A and 8B). Single cell analysis confirmed these growth curve results, and showed that Bax and Bak function was not required for Naip5-mediated growth restriction of WT L. pneumophila and had no measurable effect on limiting the growth of the DflaA strain in macrophages ( Figure 8C).
Previous studies have shown that macrophages infected with a L. pneumophila mutant deficient in the effector protein SdhA undergo rapid cell death by an unknown pathway [57]. This pneumophila DflaA strain to generate L. pneumophila DflaA, sdhA::kan. Elimination of Bax and Bak did not restore replication of L. pneumophila DflaA, sdhA::kan in macrophages ( Figure 9A) and the L. pneumophila DflaA, sdhA::kan strain was unable to replicate in Bax 2/ 2 Bak 2/2 DCs ( Figure 9B). After infection by L. pneumophila DflaA, sdhA::kan, cell death levels measured by TUNEL staining were similar in Tg(bcl2) 535rm macrophages and control B6 macrophages ( Figure 9C). The L. pneumophila DflaA, sdhA::kan strain also induced cell death in DCs derived from Tg(bcl2) 535rm mice ( Figure 9D). Thus, the L. pneumophila sdhA mutant phenotype was similar in both macrophages and DCs, which indicates that SdhA is necessary to prevent L. pneumophila from killing both macro- phages and DCs by a pathway that does not require Bax and Bak function.

Discussion
Two cell death pathways were found to restrict L. pneumophila replication in DCs. The first pathway was described previously in macrophages and involved activation of Naip5 by a process requiring L. pneumophila flagellin [37][38][39][40]. It had been shown clearly that stimulation of Naip5 by L. pneumophila flagellin results in the activation of caspase-1 [37][38][39][40], which is a critical mediator of pyroptosis. Recent data indicate that Naip5 activation of caspase-1 also results in the activation of caspase-7 [58], and that Naip5-dependent activation of caspase-7 is important for restriction of L. pneumophila replication in mouse macrophages. Many details of the Naip5 signaling pathway remain to be determined, including the full repertoire of proteins required for Naip5mediated cell death and all the cell types capable of restricting the replication of L. pneumophila by this pathway. Our data help to answer some of these questions by showing that components of the Naip5 pathway required for flagellin sensing and downstream effector responses are functioning in DCs. Additionally, the observation that overproduction of Bcl-2 or elimination of Bax and Bak did not affect restriction of WT L. pneumophila replication in macrophages and DCs with a functional Naip5 protein provides evidence that this pathway is not functionally dependent on the mitochondrial pathway of apoptosis. Thus, both macrophages and DCs have the capacity to undergo Naip5-dependent pyroptosis. In addition to restricting pathogen replication, activation of caspase-1 during this response generates bioactive IL-1b and IL-18 to stimulate additional antimicrobial responses and promote the recruitment of other immune cells [59][60][61][62]. This suggests that pyroptosis is a general innate immune response mediated by both macrophages and DCs to initiate early pro-inflammatory events at the site of microbial infection.
A second cell death pathway, which involved Bax and Bak regulation of caspase-3 activation, was found to efficiently restrict L. pneumophila replication in DCs. When the pyroptosis pathway was inactivated, either by using DCs with a defective Naip5 allele or by using L. pneumophila that had the gene encoding flagellin deleted, the cell death pathway regulated by Bax and Bak was as efficient as the pyroptosis pathway at restricting replication of L. pneumophila. A similar number of replicating L. pneumophila were contained in vacuoles in DCs deficient in Bax and Bak at 10-hours post infection ( Figure 5C) when compared to macrophages  ( Figure 8C). Additionally, the number of L. pneumophila recovered from DCs deficient in caspase-3 was similar after 24-hours of infection when compared to macrophages (Figure 4). Because the addition of bacteria stimulates the maturation of DCs in culture, and mature DCs become non-phagocytic, L. pneumophila replication in cultured DCs was not amplified by reinfection. This explains why replication subsided after L. pneumophila exited infected DCs at 24-hours post infection, but continued over a 72-hour period in macrophages ( Figure 4). Thus, rapid activation of the intrinsic cell death pathway appears to be the primary mechanism by which DCs from permissive strains of mice restrict the intracellular replication of L. pneumophila. L. pneumophila was capable of replication in DCs deficient in caspase-3; however, DCs deficient in both Bax and Bak were more permissive. This suggests that deletion of Bax and Bak more acutely blocks the apoptotic pathway, perhaps because other effector caspases can compensate for caspase-3 deficiency.
Consistent with this explanation, Bax 2/2 Bak 2/2 mice have severe developmental defects and most die perinatally, whereas, Casp3 2/2 mice are viable and have fewer developmental defects [63][64][65]. Accordingly, L. pneumophila infection induced the mitochondrial pathway of apoptosis in Casp3 2/2 DCs, but the absence of caspase-3 was sufficient to delay cell death for a long enough period of time that vacuoles containing replicating L. pneumophila were detected. By contrast, apoptosis was not induced upon L. pneumophila infection of Bax 2/2 Bak 2/2 DCs and in the absence of cell death L. pneumophila was able to replicate for a longer period of time as was indicated by an increase in the number of large vacuoles containing over 10 bacteria. These data also suggest that cell death, as opposed to another activity mediated specifically by caspase-3, was sufficient to restrict L. pneumophila replication.
The finding that overproduction of Bcl-2 resulted in enhanced bacterial replication in DCs supports the hypothesis that the mitochondrial pathway of apoptosis is important for restriction of L. pneumophila replication in DCs. Bcl-2 functions as a negative regulator of Bax and Bak function, preventing their activation and insertion into the mitochondrial membrane [66,67]. Thus, the observation that Bcl-2 overproduction phenocopies a deficiency in Bax and Bak indicates that L. pneumophila infection of DCs triggers a cell-autonomous response that activates the mitochondrial pathway of apoptosis, leading to restriction of intracellular bacterial proliferation.
Previous studies in macrophages and macrophage-like cells have demonstrated that L. pneumophila is capable of activating the mitochondrial pathway of apoptosis [68][69][70][71]; however, our data indicate that the timing of this response is different in DCs compared to macrophages. In macrophages the response is slower, and morphological signs of apoptosis were typically not observed in cells until the late stages of infection after robust bacterial replication had occurred. Host cell apoptosis induced by L. pneumophila in both macrophages and DCs required a functional Dot/Icm secretion system, but not bacterial replication. This suggests that apoptosis is activated in response to either direct activities of bacterial effector proteins translocated by the Dot/Icm system or by host cell disturbances that are caused by the cumulative actions of multiple effector proteins.
The balance of pro-apoptotic to anti-apoptotic factors is important in the regulation of the mitochondrial pathway of apoptosis. Microbial infection affects this balance both by triggering the activation of pro-apoptotic factors and by inducing expression of anti-apoptotic proteins [49,[72][73][74]. For many nonpathogenic bacteria, these two events are balanced and apoptosis is prevented. The added stress on cells infected with pathogenic microbes, however, will typically result in apoptosis unless the pathogen has the ability to alter the function of proteins involved in regulating cell death [75,76]. Thus, differences in the expression of Bcl-2 family members or in the functioning of effector proteins could account for the faster kinetics of apoptosis in DCs compared to macrophages following L. pneumophila infection.
Two effector proteins translocated into host cells by the L. pneumophila Dot/Icm system have been implicated in preventing cell death. The effector protein SidF appears to interfere with the function of pro-apoptotic Bcl-2 family members BNIP3 and Bcl-Rambo [77]. Although macrophages infected with a sidF mutant show increased apoptosis 14-hours after infection, this increase in apoptosis does not impact bacterial replication greatly [77]. By contrast, the effector SdhA is required to prevent macrophage cell death during infection by a mechanism that is not understood, and the cell death induced by an sdhA mutant greatly reduces bacterial replication in macrophages [57]. We found that the sdhA mutant induced cell death in both macrophages and DCs, and that this cell death pathway was not inhibited by Bcl-2 over-expression or elimination of Bax and Bak. Additionally, intracellular growth of the sdhA mutant was not restored in macrophages deficient in caspase-3 (data not shown). Thus, both macrophages and DCs are equally susceptible to cell death induced by the sdhA mutant, and the cell death pathway triggered by the sdhA mutant does not require several of the central components of the apoptosis pathway. These data are consistent with there being an intrinsic difference between macrophages and DCs with respect to their ability to activate the mitochondrial cell death pathway in response to L. pneumophila.  In addition to L. pneumophila, there are many other reports demonstrating that DCs are able to restrict the replication of pathogens capable of growing within macrophages [42][43][44][45]. DCs are very proficient at migrating from peripheral tissues to the host lymphatic system following exposure to maturation stimuli, such as encounters with microbes. Because of this property, it has been suggested that DCs can function as a ''Trojan Horse'' capable of systemic dissemination of pathogens internalized at peripheral sites of infection [42,44,78]. Here we show that rapid cell death is one mechanism DCs use to avoid being subverted by an intracellular pathogen. In addition to preventing pathogen replication and dissemination, apoptotic DCs harboring intracellular pathogens would become substrates for phagocytosis by neighboring DCs and macrophages, and most mechanisms used by intracellular pathogens to subvert host cellular function would be ineffective as long as the pathogen were residing in an apoptotic cell. Thus, apoptotic bodies containing pathogens would be degraded in lysosomes, resulting in the release of pathogen-derived molecules that could stimulate innate immune receptors and trigger adaptive responses by being presented on the cell surface in association with host MHC proteins. Based on these data, we hypothesize that rapid pathogen-induced apoptosis by DCs is an important innate immune response to intracellular pathogens.

Bacterial cultures
L. pneumophila serogroup 1 strain, Lp01 [18], an isogenic dotA mutant strain (DdotA), and a flagellin-deficient mutant strain (DflaA) [79] were cultured on charcoal yeast extract agar (CYE) [80] for 2 days prior to use in experiments. The DflaA, sdhA::kan strain was cultured on CYE with 10 mg/mL kanamycin. The plasmid pAM239 was used to produce DSred or GFP in the L. pneumophila strains indicated [81]. For experiments utilizing bacteria expressing DSred or GFP, L. pneumophila was grown on plates supplemented with chloramphenicol (6.25 mg/ml), and DSred or GFP expression was induced after infection by adding IPTG (0.2 mM) to the tissue culture medium.

Macrophage and dendritic cell cultures
Bone-marrow derived macrophages (BMMs) were prepared as described previously with some modifications [85]. Briefly, bone marrow was collected from the femurs and tibiae of mice. Cells were plated on non-tissue culture-treated dishes and incubated at 37uC in RPMI-1640 containing 20% heat-inactivated fetal bovine serum (FBS), 30% macrophage colony-stimulating factor (M-CSF)-conditioned medium, and 1% penicillin-streptomycin. On day 7, cells were harvested and resuspended in RPMI 1640 containing 10% FBS and 15% M-CSF-conditioned medium. Cells were then plated in 24-well tissue culture-treated plates and incubated at 37uC. Bone marrow derived-DCs (BMDCs) were prepared as described in Lutz et al. [86]. Modifications were as follows. Cells were plated on non-tissue culture-treated dishes and incubated at 37uC in RPMI-1640 supplemented with 10% heatinactivated FBS, 50 mM 2-mercaptoethanol, 1% penincillinstreptomycin and 1% GM-CSF (DC medium). Cells were harvested and used on day 10.

Intracellular replication assays
Intracellular replication of L. pneumophila in BMMs was measured as described previously [79] and modified slightly for DCs. L. pneumophila was added to DCs at a multiplicity of infection (MOI) of 20. The plates were centrifuged at 150 g for 5 minutes (min) and then incubated at 37uC for 30 min. Cells were removed from the wells and DCs were positively selected on magnetic columns using anti-CD11c-coated magnetic beads (Miltenyi Biotech). To remove extracellular bacteria, DCs were washed 36 with PBS containing 2 mM EDTA and 0.5% BSA while bound to the column. DCs were eluted and 2610 5 DCs were added to individual wells in 48-well plates. Adherent and nonadherent DCs were taken from individual wells and lysed with sterile H 2 O at the indicated times after infection, and these fractions were pooled with the culture supernatants. Dilutions from the pooled fractions were plated on CYE agar to determine bacterial CFUs. Data are the mean CFUs recovered from three independent wells6SD. Bacterial replication was calculated by determining the fold increase in CFUs.
Single cell immunofluorescence assays to measure L. pneumophila uptake and formation of vacuoles containing replicating bacteria (RV) L. pneumophila uptake and intracellular growth in both Casp3 2/2 and Casp3 +/+ DCs was performed as previously described [41]. Intracellular replication in B6, Bak 2/2 , and Bax 2/2 Bak 2/2 DCs was performed following the same protocol described previously with some modifications to the immunofluorescence staining [41]. Briefly, after permeabilization for 15 min at room temperature (R.T.) in RPMI containing 0.05% saponin, coverslips were incubated for 1 h at R.T. in permeabilization solution containing anti-MHC II I-A b+d+q , I-E d+k antibody (TIB 120; American Type Culture Collection (ATCC), Rockville, MD). Coverslips were washed 36 in RPMI containing 0.05% saponin. Coverslips were incubated 45 min at R.T. with Alexa Fluor 568-conjugated goat anti-rat (Invitrogen-Molecular Probes) in permeabilization solution and then washed 36 with PBS. Coverslips were mounted on slides and examined by fluorescence microscopy. TIB 120 staining of MHC II was used to identify DCs. Assays to measure uptake and formation of vacuoles containing replicating L. pneumophila in BMMs were conducted similarly [14]. Data are represented by the mean number of cells observed in three independent coverslips.

TUNEL staining
DCs previously selected by CD11c magnetic beads were infected with L. pneumophila at an MOI of 25 or treated for 5 h with staurosporine (1 mg/ml) and assayed for nuclear DNA fragmentation by TUNEL with the in situ cell death detection kit (Roche). Samples were then analyzed by fluorescence microscopy and all data points represent the average number of TUNEL positive cells6SD obtained from three independent coverslips.

Immunoblotting
BMMs and DCs were directly lysed in SDS-PAGE sample buffer. Lysates were separated by SDS-PAGE, and proteins were transferred (Wet Transfer Cell; Bio-Rad) at 100 V for 1 h to Immobilon P membranes (Millipore) in transfer buffer (50 mM Tris, 40 mM glycine, and 10% methanol). Membranes were blocked for 1 h at 25uC in Tris-buffered saline (TBS), 5% nonfat dry milk, and 0.1% Tween-20. Membranes were incubated with primary antibody overnight at 5uC and incubated with horseradish peroxidase-conjugated secondary antibody 1 h at R.T. Rabbit antihuman Bcl-2, rabbit anti-Bax, and rabbit anti-Bak (Cell Signaling Technology) were used. Western Lightning Chemiluminescence Reagent Plus (Perkin Elmer) was used for antibody detection.

Caspase-3/7 activity
Macrophages and DCs were plated in 96 well plates at a concentration of 5610 4 cells/well. Cells were infected with L. pneumophila at an MOI of 50, incubated at 37uC for 1, 2, 4, 6 and 11 hours and then frozen at 220uC to lyse the cells. Caspase-3/7 activity was measured using the Apo-One Homogeneous caspase-3/7 kit (Promega). Relative fluorescence units (RFU) measured at each time point is proportional to the amount of caspase-3/7 activity. All data points represent the average values6SD obtained from three wells assayed independently.