Effector-mediated subversion of proteasome activator (PA)28αβ enhances host defense against Legionella pneumophila under inflammatory and oxidative stress conditions

Legionella pneumophila is a natural pathogen of amoebae that causes Legionnaires’ Disease in immunocompromised individuals via replication within macrophages. L. pneumophila virulence and intracellular replication hinges on hundreds of Dot/Icm-translocated effector proteins, which are essential for biogenesis of the replication-permissive Legionella-containing vacuole (LCV). However, effector activity can also enhance mammalian host defense via effector-triggered immunity. The L. pneumophila effector LegC4 is important for virulence in amoebae but enhances host defense against L. pneumophila in the mouse lung and, uniquely, within macrophages activated with either tumor necrosis factor (TNF) or interferon (IFN)-γ. The mechanism by which LegC4 potentiates cytokine-mediated host defense in macrophages is unknown. Here, we found that LegC4 enhances cytokine-mediated phagolysosomal fusion with Legionella-containing vacuole (LCV) and binds host proteasome activator (PA)28α, which forms a heterooligomer with PA28β to facilitate ubiquitin-independent proteasomal degradation of oxidant-damaged (carbonylated) proteins. We found that oxidative stress was sustained in the presence of LegC4 and that the LegC4 restriction phenotype was relieved in PA28αβ-deficient macrophages and in the lungs of mice in vivo. Our data also show that oxidative stress is sufficient for LegC4-mediated restriction in macrophages producing PA28αβ. PA28αβ has been traditionally associated with antigen presentation; however, our data support a novel mechanism whereby effector-mediated subversion of PA28αβ enhances cell-autonomous host defense against L. pneumophila under inflammatory and oxidative stress conditions. This work provides a solid foundation to evaluate induced proteasome regulators as mediators of innate immunity.


Introduction
Many intracellular bacterial pathogens cause disease by replicating within host macrophages [1]. However, macrophages activated by T H 1 proinflammatory cytokines, such as tumor necrosis factor (TNF) and interferon (IFN)-γ, are potently and broadly microbicidal. Activated macrophages and inflamed tissues are in a state of altered protein homeostasis (proteostasis), owing in part to robust production of reactive oxygen species (ROS), which cause collateral damage to cellular proteins. To cope with this oxidative stress, cytokine signaling and ROS upregulate proteasome activator (PA)28αβ, which engages proteolytic 20S proteasome core particles (CPs) to facilitate rapid ubiquitin-independent degradation of oxidant-damaged proteins [2]. Impaired proteasome activity and proteotoxic stress trigger compensatory upregulation of lysosomal degradation pathways, which are central to cell-autonomous immunity within activated macrophages [3]. PA28αβ has been traditionally associated with antigen presentation and adaptive immunity [4,5]; however, a role for PA28αβ in cell-autonomous antimicrobial innate host defense has not been established.
Legionella pneumophila is a natural pathogen of amoebae and accidental human pathogen that causes Legionnaires' Disease pneumonia in immunocompromised individuals by replicating within alveolar macrophages [6]. L. pneumophila replication within macrophages is a direct consequence of extensive co-evolution with natural host amoebae [7][8][9]; however, L. pneumophila rarely transmits person-to-person and is rapidly cleared by acute pathogen-centric innate immune responses in healthy hosts [10][11][12]. To establish its intracellular niche, the Legionella-containing vacuole (LCV), L. pneumophila translocates hundreds of virulence factors, called effector proteins, into infected host cells by a Dot/Icm type IV secretion system (T4SS) [6,13,14]. Effectors are essential for virulence, but detection of their activity within host cells can induce pro-inflammatory cytokine production via effector-triggered immunity (ETI) [10,[15][16][17]. Pro-inflammatory cytokines activate and potentiate the antimicrobial capacity of infected macrophages and are central to host defense against L. pneumophila [18,19]. Thus, L. pneumophila is a well-established model pathogen to study mechanisms of pathogen-centric and effector-mediated antimicrobial innate immunity.
We discovered that the L. pneumophila effector LegC4 induces a unique mechanism of effector-mediated host defense by potentiating the antimicrobial activity of activated macrophages [20]. LegC4 confers a fitness disadvantage on L. pneumophila in a mouse model of Legionnaires' Disease, likely resulting from elevated levels of pro-inflammatory cytokines in the lung since a growth defect is only observed within cultured macrophages when activated with exogenous TNF or IFN-γ [20,21]. The evolutionary basis for retention of LegC4 by L. pneumophila stems from its importance for replication within its natural host, Acanthamoeba castellanii [21]. Thus, LegC4 confers quantifiable opposing phenotypes in natural (protozoa) and 'accidental' (mammalian) hosts, which is noteworthy since the majority of L. pneumophila effectors are functionally redundant in laboratory models and have no obvious growth phenotypes [21,22]. The specific augmentation of cytokine-mediated restriction by LegC4 represents a potentially novel mechanism of ETI, but how LegC4 functions and potentiates host defense against L. pneumophila is unknown.
Here, we investigated the mechanism of LegC4 restriction within macrophages and found central roles for PA28αβ, phagolysosomal fusion, and oxidative stress. LegC4 enhanced cytokine-dependent phagolysosomal fusion with LCVs and perpetuated oxidative stress. LegC4 bound host PA28α in vitro and changed the subcellular localization of PA28α within infected and transfected cells. LegC4 restriction was relieved in PA28αβ-deficient (Psme1/2 -/-) BMDMs and the fitness advantage associated with ΔlegC4 mutant bacteria in vivo was abolished in Psme1/2 -/mice. The abundance of oxidant-damaged proteins was increased in the presence of LegC4, and oxidative stress was sufficient to impair replication of LegC4-producing L. pneumophila in BMDMs producing PA28αβ. Importantly, L. pneumophila replication was attenuated within Psme1/2 -/-BMDMs irrespective of LegC4, suggesting that impaired PA28αβ activity is sufficient to augment host defense in macrophages and is the mechanism driving LegC4 restriction. Overall, our data support a model whereby PA28α targeting by a translocated effector protein perturbs proteostasis and enhances cell-autonomous restriction of L. pneumophila within activated macrophages. These data are the first to uncover a role for PA28αβ in innate host defense against an intracellular bacterial pathogen and reveal a novel mechanism by which the antimicrobial activity of macrophages can be augmented under inflammatory and oxidative stress conditions.

LegC4 confers a fitness disadvantage on L. pneumophila in cytokineactivated macrophages and in a mouse model of Legionnaires' Disease
We previously reported that LegC4 confers a fitness disadvantage on L. pneumophila in intranasal C57BL/6 mouse models of Legionnaires' Disease (LD) and within cytokine-activated macrophages [20,21]. C57BL/6 mice are restrictive to L. pneumophila due to flagellin (FlaA)mediated activation of the NAIP5/NLRC4 inflammasome [23,24]. Thus, to study FlaA-independent host responses within C57BL/6 mice, we leverage FlaA-deficient L. pneumophila (ΔflaA) strains. We found that L. pneumophila ΔflaAΔlegC4 (herein called ΔΔ) mutant strains outcompeted parental ΔflaA strains in the lungs of C57BL/6 wild-type (WT) mice [20] but a growth phenotype for L. pneumophila ΔΔ was only observed within cultured mouse primary bone marrow-derived macrophages (BMDMs) activated with exogenous recombinant (r) TNF or rIFN-γ [20]. However, autocrine and paracrine TNF produced by cultured BMDMs is sufficient to attenuate L. pneumophila strains induced to overexpress legC4 from a complementing multicopy plasmid in naïve WT BMDMs [20]. Importantly, this phenotype was specific to intracellularly growing bacteria since L. pneumophila growth in broth in vitro is unaffected by overexpression of legC4 [20]. Thus, LegC4 confers a growth defect on L. pneumophila in a dose-dependent manner within cytokine-activated macrophages.
We confirmed our previous results and found that overexpression of legC4 was sufficient to impair L. pneumophila replication within WT BMDMs since L. pneumophila ΔΔ (plegC4) replication was significantly attenuated compared to both the L. pneumophila ΔflaA parental strain and L. pneumophila ΔΔ harboring an empty plasmid vector (pEV) (Fig 1A). Impaired replication was dependent on TNF signaling since overexpression of legC4 did not impair growth within BMDMs unable to produce TNF (Myd88 -/-; Fig 1B) or signal through TNFR1 (Tnfr1 -/-; Fig 1C). IFN-γ signaling is sufficient for LegC4 restriction [20]; however, the LegC4 restriction phenotype was retained within Ifngr1 -/-BMDMs (Fig 1D), indicating that the relatively low levels of IFN-γ secreted by macrophages are not sufficient for restriction. Importantly, we found that LegC4 restriction is not specific to mouse macrophages since both L.
Cytokine signaling drives LegC4 growth defects in cultured BMDMs, but the role of cytokine signaling in vivo is unclear. We evaluated bacterial fitness by calculating competitive index values for mice infected with a 1:1 mixture of L. pneumophila ΔflaA and either L. pneumophila ΔΔ (pEV) or ΔΔ (plegC4) for 48 h (see Materials & Methods). We confirmed our previous results showing that LegC4 confers a dose-dependent fitness defect on L. pneumophila in the lungs of WT mice since L. pneumophila ΔflaA was outcompeted by L. pneumophila ΔΔ (pEV) but was able to outcompete L. pneumophila ΔΔ (pJBlegC4; Fig 1F) [20]. We also confirmed that loss of TNFR1 is insufficient to abolish the legC4 mutant fitness advantage ( Fig 1F) and obtained similar results in the lungs of IFNGR1-and MyD88-deficient mice ( Fig 1F). However, the fitness phenotype associated with loss LegC4 was less pronounced in the lungs of Myd88 -/mice compared to WT, Tnfr1 -/and Ifngr1 -/mice (Fig 1F), which is likely due to the relatively low levels of MyD88-independent IFN-γ produced in the L. pneumophila-infected lung [25]. Furthermore, the fitness disadvantage associated with overexpression of legC4 was less pronounced in the lungs of MyD88-, IFNGR1-and TNFR1-deficient mice compared to WT mice ( Fig 1F). These data support our previous work showing that LegC4 augments redundant cytokine-mediated pathogen restriction mechanisms [20]. Overexpression of legC4 is clearly not necessary to observe a phenotype; however, based on the narrow phenotypic window in BMDMs activated with exogenous cytokine, this infection model is useful for our goal to define how LegC4 potentiates cytokine-mediated restriction within macrophages

Pharmacological inhibition of the H + -type vacuolar (v)ATPase abolishes LegC4 restriction in BMDMs
We subsequently tested whether lysosomal acidification contributes to LegC4-mediated restriction by quantifying L. pneumophila replication within BMDMs treated with Bafilomycin A1 (BAF). BAF is a pharmacological inhibitor of the H + -type vacuolar (v)-ATPase, which drives vacuolar acidification and phagolysosomal fusion [32]. Thus, we tested a role for vAT-Pase activity in LegC4 restriction by quantifying L. pneumophila replication within BAFtreated BMDMs. LCV acidification is important for late stages of the L. pneumophila lifecycle [33]; however, intracellular replication is minimally impacted within BMDMs treated with a low concentration (�12.5 nM) of BAF [34]. We found that L. pneumophila ΔΔ (plegC4) replication did not differ from L. pneumophila ΔflaA and ΔΔ (pEV) strains within BMDMs treated with 10 nM BAF (Fig 2C, left). However, as expected, L. pneumophila ΔΔ (plegC4) replication was significantly attenuated compared to ΔflaA and ΔΔ (pEV) within vehicle (DMSO)-treated BMDMs (Fig 2C, right). These data further support a role for lysosomes in LegC4-mediated restriction.

LegC4 interacts with host proteasome activator (PA)28α
To gain insight into the mechanism of LegC4 function, we identified LegC4 binding partners by yeast two-hybrid (Y2H) using LegC4 as bait and 30,000 clones of fragmented genomic (g) DNA from mouse splenocytes as prey (Hybrigenics, Inc., see Materials & Methods). Our screen revealed interactions between LegC4 and several regulators of eukaryotic proteasomes, including proteasome activator (PA)28α, PA28γ, and Ecm29 (S1 Table). The highest-confidence interactor was PA28α with 28 individual clones, all of which encoded a minimal overlapping region in the PA28α C-terminal domain (amino acid residues 105-249) (S1 Table and S1 Fig). These data suggest a direct interaction between LegC4 and PA28α, which is conserved in amoebae and upregulated by cytokine signaling in macrophages.
We sought to validate our Y2H data using reciprocal affinity purification, co-immunoprecipitation (IP), and confocal microscopy. We evaluated direct binding between recombinant PA28α and LegC4 by reciprocal affinity chromatography on GST-or His 6 -Myc-tagged fusion proteins produced in E. coli. Using glutathione agarose beads, we found that GST-LegC4 retained Myc-PA28α ( Fig 3A) and GST-PA28α retained Myc-LegC4 ( Fig 3B). PA28α did not interact with either GST alone or the unrelated L. pneumophila effector Lgt1 (Fig 3A and 3B), known to target eukaryotic elongation factor 1A [35,36], which indicates that PA28α does not bind non-specifically to L. pneumophila effectors. To determine if PA28α and LegC4 interact in mammalian cells, we co-transfected HEK293T cells to transiently produce epitope-tagged 3xFLAG-LegC4 and either PA28α-Myc or GFP-PA28α fusion proteins. Ectopic expression of PA28α was necessary in these experiments since it is expressed only in response to cytokine signaling [37,38]. Using α-FLAG and α-GFP pull-downs, we found that PA28α-Myc coimmunoprecipitated with 3xFLAG-LegC4 ( Fig 3C) and that 3xFLAG-LegC4 co-immunoprecipitated with GFP-PA28α (Fig 3D), respectively. Together, these data show that LegC4 binds PA28α.
We further evaluated the biological relevance of LegC4-PA28α binding by evaluating protein localization in cells by confocal microscopy. Ectopically produced 3xFLAG-LegC4 colocalized with GFP-PA28α within transiently transfected HeLa cells since calculated Pearson Correlation Coefficients (PCC) indicated a significant positive correlation between 3xFLA-G-LegC4 and GFP-PA28α (Mean PCC = 0.798 ± 0.054) compared to 3xFLAG-LegC4 and GFP (Mean PCC = 0.546 ± 0.196) (Fig 3E and 3F). Interestingly, LegC4 modulated PA28α subcellular localization since GFP-PA28α localized to the nucleus in the absence of LegC4 [ Fig  3E and 3G, arrowhead, (i)] but was cytosolically localized within cells co-expressing LegC4 [ Fig 3E and 3G, arrows, (ii)] [39]. We subsequently tested whether PA28α localization is influenced by Dot/Icm-translocated LegC4 within L. pneumophila-infected BMDMs. WT BMDMs were infected with L. pneumophila ΔflaA, ΔΔ, or ΔdotA for 7.5 h, at which time a replicationcompetent LCV is established. Fixed cells were permeabilized and stained with α-L. pneumophila and α-PA28α antibodies and Hoechst (DNA) and imaged by confocal microscopy. Endogenous PA28α is constitutively produced in immune cells so can be imaged in BMDMs. We found that PA28α was primarily cytosolic within BMDMs infected with L. pneumophila ΔflaA, but localized robustly to the nucleus of BMDMs infected with either L. pneumophila ΔΔ or ΔdotA (Fig 3H, white arrowheads). Interestingly, PA28α was present on LCVs harboring L. pneumophila ΔΔ but not ΔflaA or ΔdotA control strains (Fig 3H, blue arrowheads and inset). Although altered PA28α localization mediated by translocated LegC4 in naïve infected BMDMs was not sufficient for LegC4 restriction or phagolysosomal fusion in naïve WT BMDMs (Figs 1A and 2B), these data further support a biologically relevant interaction between LegC4 and PA28α, which has not previously identified as a target of bacterial effectors.

LegC4 does not confer a fitness disadvantage in the lungs of Psme1/2 -/mice
Since loss of PA28αβ was sufficient to suppress LegC4-mediated growth attenuation in BMDMs, we hypothesized that LegC4 would not confer a fitness disadvantage on L. pneumophila in the lungs of Psme1/2 -/mice. We infected Psme1/2 -/mice with a 1:1 mixture of L. pneumophila ΔΔ (pEV) and ΔflaA and calculated competitive index values from lung homogenates at 48 h post-infection. Strikingly, L. pneumophila fitness was unaffected by LegC4 in the lungs of Psme1/2 -/mice (CI = 1.09 ± 0.793); Fig 4C). In aggregate, our data suggest that PA28αβ is important for LegC4-mediated restriction in cytokine activated BMDMs and the mouse lung in vivo.

The abundance of carbonylated proteins in macrophages is increased in the presence of LegC4 under oxidative stress conditions
PA28αβ is expressed constitutively in cells of the innate immune system and upregulated to maintain the integrity of the proteome under oxidative stress conditions [2,38,40,42]. PA28αβ engages proteolytic proteasome 20S core particles (CPs) to promote rapid ubiquitin-independent proteolysis of oxidant-damaged (carbonylated) proteins [4,43], the abundance of which increases commensurately with the abundance of reactive oxygen species (ROS) [44,45]. 20S CPs can degrade carbonylated proteins, but their efficiency is substantially increased when in complex with PA28αβ [2,43]. Concomitantly, there is a strong negative correlation between PA28αβ activity and carbonylated protein abundance under oxidative stress conditions [43]. Our Y2H data suggest that LegC4 binds the C-terminal domain of PA28α, which is important for PA28αβ docking on proteasome CPs [46][47][48]. Thus, we rationalized that carbonylated proteins would be more abundant in the presence of LegC4 under oxidative stress conditions. To directly evaluate LegC4, we generated RAW 264.7 cells stably transfected with 3xFLAG-LegC4 (RAW-LegC4) or empty plasmid vector (RAW-Cntrl). We confirmed 3xFLAG-LegC4 production by confocal microscopy (Fig 5A) and Western blot (Fig 5B). Immunoprecipitation was necessary to visualize LegC4 since it is produced by the macrophages at low abundance.

PLOS PATHOGENS
PA28αβ-and effector-mediated host defense However, the abundance of ectopically produced LegC4 was sufficient to attenuate replication of L. pneumophila ΔflaA (Fig 5C). Attenuated growth was observed at 48 h post-infection, similar to the kinetics of LegC4 restriction within BMDMs (see Fig 1A), which corroborates our microscopy data suggesting that ectopically-produced LegC4 is functional (see Fig 3G and 3H) and demonstrates that RAW-LegC4 cells are suitable to evaluate LegC4 activity. RAW-LegC4 and -Cntrl cells were incubated in the presence or absence of 10 μM H 2 O 2 for 24 h and protein carbonyls in cell lysates were quantified by ELISA. We used H 2 O 2 since it is membrane permeable, stable in solution, and biologically relevant since L. pneumophila-infected macrophages in the lung would also be exposed to neutrophil-generated extracellular ROS [26,49]. Indeed, H 2 O 2 -treated RAW-LegC4 cells contained significantly more protein carbonyls than RAW-Cntrl cells (Fig 5D). Under native conditions, the concentration of carbonylated proteins was unaffected by LegC4 (Fig 5D), which suggests that LegC4 does not itself induce oxidative stress. The abundance of protein carbonyls was unaffected by H 2 O 2 in RAW-Cntrl cells,

PLOS PATHOGENS
PA28αβ-and effector-mediated host defense likely because upregulation of PA28αβ over the duration of this experiment was sufficient for turnover of damaged proteins.
We subsequently tested whether LegC4 increases the abundance of protein carbonyls within L. pneumophila-infected BMDMs. Tnfr1 -/-BMDMs were used to control for any LegC4-mediated differences in TNF secretion [20] and, consequently, intracellular ROS production [50]. Tnfr1 -/-BMDMs were infected with L. pneumophila ΔflaA, ΔΔ (pEV), or ΔΔ (plegC4) for 4 h in the presence or absence of 50 μM H 2 O 2 and protein carbonyls were quantified by Western blot and densitometry. We found that carbonylated proteins were more

PLOS PATHOGENS
PA28αβ-and effector-mediated host defense abundant within BMDMs infected with L. pneumophila ΔΔ (plegC4) compared to L. pneumophila ΔflaA and ΔΔ (pEV)-infected and uninfected BMDMs when oxidative stress was induced with H 2 O 2 (Fig 5E and 5F). The abundance of protein carbonyls was not affected by LegC4 within naïve untreated BMDMs, which further suggests that LegC4 itself does not induce oxidative stress. Protein carbonyl abundance was not affected by either LegC4 or L. pneumophila infection within untreated (UT) BMDMs (Fig 5E and 5F), likely because the amount of ROS produced in response to L. pneumophila infection under these conditions is insufficient for detectable differences in protein carbonyls by Western blot. Together, these data support our hypothesis that LegC4 perturbs proteostasis in macrophages under oxidative stress conditions and suggest that LegC4 may function by impairing PA28αβ activity.

Oxidative stress is sufficient for LegC4 restriction in PA28αβ-producing BMDMs
Our current data suggest role(s) for PA28αβ and oxidative stress in LegC4 restriction; thus, we hypothesized that oxidative stress would be sufficient for LegC4-mediated restriction within BMDMs producing PA28αβ. To test this, we quantified replication of L. pneumophila ΔflaA and ΔΔ within WT and Psme1/2 -/-BMDMs treated with 20 μM H 2 O 2 to induce oxidative stress. Indeed, oxidative stress was sufficient for LegC4 restriction since L. pneumophila ΔΔ replicated to significantly higher levels than L. pneumophila ΔflaA in WT BMDMs (Fig 6A), showing that oxidative stress is sufficient for LegC4 restriction. Importantly, the growth advantage associated with ΔlegC4 mutation was abolished in PA28αβ-deficient BMDMs, but replication of LegC4-producing bacteria was unaffected (Fig 6A). These data show that LegC4 impairs L. pneumophila replication under oxidative stress conditions via a mechanism involving PA28αβ.

L. pneumophila replication is impaired in Psme1/2 -/-BMDMs irrespective of LegC4
Our data suggest that LegC4 restriction is mediated by subversion of PA28αβ. Thus, it is possible that the dose-dependency of LegC4 in naïve BMDMs results from stoichiometric differences in its abundance relative to PA28αβ. Thus, we rationalized that L. pneumophila replication would be suppressed in naïve Psme1/2 -/-BMDMs irrespective of LegC4 and found that, indeed, both L. pneumophila ΔflaA and ΔΔ strains were attenuated for replication within

Neither TNF production nor cytotoxicity are potentiated by loss of PA82αβ in L. pneumophila-infected BMDMs
Our data suggest that LegC4 enhances phagolysosomal fusion with LCVs via subversion of PA28αβ activity and that loss of PA28αβ is sufficient to impair L. pneumophila intracellular replication. However, impaired bacterial growth may also result from increased TNF production or death of infected BMDMs. We quantified TNF in the supernatants of WT and

Inflammasome activation is dispensable for LegC4-mediated restriction
Our data suggest that LegC4 perturbs proteostasis and potentiates oxidative stress. Sustained oxidative and proteotoxic stress and proteostasis perturbations can activate the NLRP3 inflammasome, which culminates in activation of caspase-1, an effector caspase responsible for pyroptotic cell death and inflammatory cytokine secretion. However, pharmacological inhibition of the NLRP3 inflammasome with MCC950 did not affect LegC4 restriction since L. pneumophila ΔΔ (plegC4) was still attenuated for replication compared to L. pneumophila ΔΔ (pEV) and ΔflaA within WT BMDMs (S4A Fig). We also found that LegC4 restriction was preserved within BMDMs derived from caspase-1-deficient (Casp1 -/-) mice (S4B Fig). Within cultured macrophages, pyroptotic cell death is responsible for L. pneumophila restriction downstream of inflammasome activation. Our data suggest that LegC4 does not enhance LDH release from infected BMDMs (S3 Fig); however, pyroptotic cell death is more pronounced within BMDMs primed with toll-like (TLR) receptor agonists [51,52]. Since TLR2 is important for host defense against L. pneumophila [25], we evaluated LDH release from L. pneumophila-infected BMDMs pretreated with the TLR2 agonist PAM 3 CSK 4 . Similar to our previous results (S3 Fig), LDH release from BMDMs primed with PAM 3 CSK 4 was unaffected by LegC4 (S4C Fig). These data suggest that LegC4-mediated restriction occurs independently of NLRP3-and caspase-1-dependent pyroptotic cell death and support an overarching model whereby subversion of PA28αβ by LegC4 augments lysosomal pathogen killing under inflammatory and oxidative stress conditions.

Discussion
The Dot/Icm effector LegC4 is bona fide L. pneumophila virulence factor in a natural host amoeba but confers a dose-dependent fitness disadvantage on L. pneumophila in cytokine-

PLOS PATHOGENS
PA28αβ-and effector-mediated host defense activated macrophages and in a mouse model of Legionnaires' Disease [20,21]. LegC4 restriction represents a unique mode of effector-mediated immunity since its activity augments existing inflammatory responses in macrophages. In this study, we have uncovered mechanistic insights into LegC4 restriction and a new role for PA28αβ in cell-autonomous host defense against L. pneumophila. We propose a model whereby subversion of host PA28αβ by LegC4 perturbs proteostasis under oxidative stress conditions, resulting in increased lysosome biogenesis and phagolysosomal fusion with LCVs ( S5 Fig). This model is supported by our data showing that (1) LegC4 enhances cytokine-mediated phagolysosomal fusion with LCVs; (2) LegC4 binds and modulates the subcellular localization of PA28α; (2) PA28αβ is required for the effects of LegC4 in BMDMs and in vivo; (3) PA28αβ substrates are more abundant in the presence of LegC4; (4) oxidative stress is sufficient for LegC4 restriction; and (5) loss of PA28αβ was restrictive to L. pneumophila irrespective of LegC4. This work is the first to identify PA28α as a target for a bacterial effector protein and a role for PA28αβ in innate and effector-mediated host defense against an intracellular bacterial pathogen.
L. pneumophila is an accidental pathogen of humans that rarely transmits between infected individuals. Thus, an evolutionary basis for the LegC4-PA28α interaction can be explained by LegC4 targeting the PA28α homolog in amoebae. Indeed, the natural host A. castellanii, in which loss of legC4 confers a growth defect [21], encodes a PA28α homolog (AcPA28) that shares 34% identity with mouse PA28α. Sequence identity between PA28α and AcPA28 is primarily in the PA28α C-terminal domain, which is important for engaging 20S CPs and was contained in all PA28α clones identified in our Y2H screen (S1 Table). Since PA28αβ substrates were more abundant in the presence of LegC4 (Fig 5), we postulate that LegC4 binding impairs PA28-CP complex formation. The function of AcPA28 has not been described; however, it is likely a proteasome regulator based on structural studies showing that mammalian PA28α and evolutionary distant PA28s are similar and all associate with 20S CPs [47,48]. The benefit of inhibiting PA28 in the natural host is unclear, but sequestering PA28 from CPs may preserve the integrity of the constitutive 26S proteasome, a central player in L. pneumophila's virulence strategy [53]. We are currently investigating the molecular mechanism by which LegC4 modulates PA28 activity and its role in L. pneumophila virulence in natural host amoebae.
Virulent L. pneumophila establish a replicative LCV by avoiding endosomal maturation of their phagosome. The current dogma dictates that lysosomal fusion with LCVs at early time points during infection (� 16 h) and L. pneumophila intracellular replication are mutually exclusive [33,54]. We found that plasmid expression of legC4 increased the abundance of LAMP1 on LCVs in within WT BMDMs, similar to what is seen for the avirulent Dot/Icmdeficient strain. LegC4-mediated phagolysosomal fusion was suppressed in Tnfr1 -/-BMDMs, but significantly more LAMP1+ LCVs were present relative to cells infected with L. pneumophila ΔflaA and ΔΔ (pEV) control strains. This is intriguing since plasmid expression of legC4 does not affect L. pneumophila replication within Tnfr1 -/-BMDMs [20] (Fig 1D). Based on our model that phagolysosomal fusion results from oxidative stress (S5 Fig), we postulate that the TNFR1-independent increase in LAMP1+ LCVs within cells infected with the legC4 overexpressing strain may result from ROS produced in response to phagocytosis or pattern-recognition receptor signaling [55,56]. A mechanistic connection between phagolysosomal fusion and PA28αβ can be explained by compensatory upregulation of lysosomal degradation pathways resulting from impaired proteasome activity and oxidative stress [3, 57,58]. However, it is unlikely that LegC4 functions exclusively through PA28α since modest LegC4-mediated restriction was retained in Psme1/2 -/-BMDMs (Fig 4A and 4B). A potential explanation for this is that LegC4 may target other host proteins when PA28α is absent, such as Ecm29 or PA28γ, which were identified in our Y2H screen (S1 Table). Ecm29 is a particularly interesting

PLOS PATHOGENS
PA28αβ-and effector-mediated host defense candidate since it disassembles 26S proteasomes under oxidative stress conditions to liberate free 20S CPs that can then associate with PA28αβ [59][60][61]. Our data suggest that lysosomes are central to LegC4 restriction, but the underlying mechanism(s) of phagolysosomal fusion with LCVs and the role of PA28αβ remain to be fully elucidated.
Lysosomal degradation pathways, including autophagy, are cytoprotective under inflammatory and oxidative stress conditions and central to cytokine-mediated cell-autonomous pathogen restriction. Lysosomal restriction within IFN-γ-activated macrophages involves immunity-related GTPases (IRGs), which localize to and facilitate lysosomal trafficking to pathogen-containing phagosomes [62]. Although IRGs may contribute to restriction of LegC4-producing L. pneumophila strains within IFN-γ-activated macrophages [63], IFNGR1 signaling was dispensable for LegC4 restriction (Fig 1E and 1F), which suggests that IRGs are not central drivers of LegC4-mediated L. pneumophila restriction. It is tempting to speculate that autophagy contributes to LegC4 restriction since autophagy genes are upregulated to clear protein aggregates and protect against TNF-induced death in myeloid cells [64]. Autophagy also prevents aberrant inflammasome-mediated pyroptotic cell death below a certain threshold of inflammasome agonist [65]. The NLRC4 inflammasome is dispensable for LegC4-mediated restriction [20,21] and we also found no role for the NLRP3 inflammasome (S4 Fig), which is activated by both oxidative and proteotoxic stress [66]. Whether autophagy contributes to LegC4-mediated restriction is unknown and warrants investigation since several L. pneumophila effectors impair global autophagy in host cells [67][68][69].
Using several complementary approaches, we found that LegC4 binds PA28α and influences its localization in transfected and infected cells. The change in PA28α localization supports an interaction with LegC4, but the underlying mechanism by which it occurs is unclear. We were unable to detect Dot/Icm-translocated LegC4 by confocal microscopy within infected BMDMs, likely because the abundance is below the limit of detection, which is a common challenge in the study of bacterial effectors that do not localize to organellar membranes [70,71]. Our data suggest that the interaction between LegC4 and PA28α may drive localization changes in transfected cells; however, the mechanism is likely more complex since presumed stoichiometric imbalances between PA28α and Dot/Icm-translocated LegC4 make it unlikely that binding is sufficient. Thus, in infected macrophages, LegC4-mediated changes in PA28α localization may result from a need for additional nuclear PA28αβ to clear proteins damaged by the oxidative burst in the presence of LegC4. However, the extent of oxidative damage under these conditions is clearly insufficient to impair L. pneumophila replication since endogenous LegC4 does not affect bacterial replication within naïve BMDMs. Interestingly, we also found endogenous PA28α on LCVs harboring LegC4-deficient L. pneumophila, which raises the possibilities that LegC4 binding or activity sequesters PA28α in the cytosol away from proteasome complexes associated with the LCV [63,72]. Although changes in PA28α localization support its functional modulation by LegC4, the molecular underpinnings and consequences of LegC4-PA28α binding, especially based on stoichiometric differences in protein abundance during infection, on PA28αβ activity and pathogen restriction require further investigation.
A role for PA28αβ and oxidative stress in LegC4 restriction was also revealed by the sufficiency of hydrogen peroxide (H 2 O 2 ) for impaired growth in WT BMDMs. H 2 O 2 is routinely used to induce oxidative stress in laboratory models since both superoxide (O 2 •-) and hydroxyl (OH • ) radicals are unstable in solution. Within immune cells, phagocytosis, pro-inflammatory cytokines, and PRR signaling increase production of ROS by phagocyte NADPH oxidase (NOX2) and mitochondria [50,55,56,[73][74][75]. Intracellular ROS contribute to cell-autonomous pathogen restriction but ROS are also released into the extracellular space via plasma membrane-bound NOX2 or by passive diffusion across membranes (H 2 O 2 ) [49,76]. Interestingly,

PLOS PATHOGENS
PA28αβ-and effector-mediated host defense ROS play a negligible role in host defense against L. pneumophila in cultured naïve macrophages, likely owing NOX2 exclusion from LCVs [26,77,78]. However, in the lung, ROS are produced abundantly by neutrophils and neutrophil ROS is important for host defense against L. pneumophila [26]. Thus, infected alveolar macrophages are exposed to both extracellular ROS produced by lung neutrophils and infection-and cytokine-induced intracellular ROS [49,79,80]. When we treated BMDMs with exogenous H 2 O 2 to recapitulate the in vivo environment, endogenous LegC4 was sufficient to impair L. pneumophila replication ( Fig 6A); thus, the fitness disadvantage associated with LegC4 in vivo may be due to oxidative stress. Furthermore, it is tempting to speculate that the dose-dependency of LegC4 restriction may correlate with the abundance of ROS. ROS are microbicidal, but this activity is unlikely to be involved in LegC4 restriction since (1) L. pneumophila has multiple complimentary strategies to protect against oxidative stress and can withstand up to 2 mM H 2 O 2 in vitro (orders of magnitude greater than intracellular concentrations) [78,[81][82][83]; and (2) replication of LegC4-deficient L. pneumophila was unaffected by H 2 O 2 treatment (Fig 6A and 6B). While our data do not preclude roles for additional host factors or cytokine-driven cellular processes in LegC4 restriction of L. pneumophila, they support a central role for oxidative stress.
Our data suggest that oxidative stress is sustained in the presence of LegC4 since oxidatively damaged (carbonylated) proteins were more abundant. Carbonylated proteins are used as a biomarker for oxidative stress since proteins and nascent polypeptides are highly susceptible to oxidative damage [44,45]. Importantly, increased protein carbonyl abundance is also indicative of the inability of cellular proteasomes and chaperones to keep pace with the rate of damage [84]. Thus, it is conceivable that impaired PA28αβ activity is indeed responsible for increased protein carbonyl abundance in the presence of LegC4. Nascent polypeptides and misfolded proteins are more susceptible to oxidative damage [85]; thus, impaired co-translational chaperone activity, such as inactivation of Hsp70 by the effector LegK4 [86], may also contribute to carbonylated protein abundance in L. pneumophila-infected macrophages. Accumulation of heavily oxidized polypeptides results in formation of aggregates, which are impervious to proteasomal degradation and must be cleared by lysosomal degradation pathways. We think that a global increase in carbonylated proteins drives increased phagolysosomal activity in the presence of LegC4; however, all proteins are not equally susceptible to oxidative damage and individual oxidized species may be involved. For example, the actin cytoskeleton, which plays a critical role in vesicular transport, is deeply influenced by cellular redox status [87]. Filamentous (F)-actin is involved in phagolysosomal fusion with intracellular L. pneumophila and pathogen restriction downstream of the caspase-11 inflammasome [88][89][90][91]. Caspase-11 promotes phagolysosomal fusion with intracellular pathogens by mediating phosphorylation and inactivation of the actin depolymerization factor cofilin [88,91]. Phosphorylated cofilin is highly susceptible to oxidative damage, and its oxidation may therefore enhance phagolysosomal fusion with LCVs [92].
In summary, we have uncovered a novel mechanism of effector-mediated host defense and a new role for PA28αβ and oxidative stress in cell-autonomous pathogen restriction. Interestingly, PA28αβ is targeted by several viral pathogens, likely as a means to suppress antigen presentation on MHC class I [93][94][95]. Impaired antigen presentation is unlikely to play a role in L. pneumophila restriction since phenotypes are observed in cultured macrophages and during acute lung infection, prior to initiation of an adaptive immune response. This work is the first to show bacterial pathogen targeting of PA28α and suggest that suppression of PA28αβ activity and aberrant proteostasis may be a pathogen-triggered host defense mechanism to potentiate existing inflammatory responses. The ability to enhance the antimicrobial potential of activated macrophages has exciting implications for development of host-centric immunotherapeutics and future studies will provide further insight into the ubiquity and mechanism(s) of pathogen restriction induced by proteotoxic stress and suppression of PA28αβ activity.

Ethics statement
All experiments involving animals were approved by the Kansas State University and Michigan State University Institutional Animal Care and Use Committees (KSU Protocols: 4022, 4501, and 4386; MSU protocol: PROTO202200365) and performed in compliance with the Animal Welfare Act and National Institutes of Health guidelines. Animals were euthanized in accordance with American Veterinary Medical Association Guidelines.

Mice and bone marrow-derived macrophages
Mice on C57BL/6 background were purchased from the Jackson Laboratories (Bar Harbor, Maine). Wild-type, Tnfr1 -/-, Ifngr1 -/-, and Myd88 -/mice were purchased from live repositories and Psme1/2 -/mice were recovered from cryopreserved embryos and bred for homozygotes at the Jackson Laboratories (Bar Harbor, ME). In-house colonies were maintained in specific pathogen-free conditions at Kansas State University and Michigan State University.
Bone marrow was harvested from seven-to twelve-week-old mice as described [98]. BMDMs were generated by differentiation in RPMI supplemented with 20% heat-inactivated fetal bovine serum (HI-FBS) (Biowest, Riverside, MO) and 15% L929 cell supernatant for 6 days prior to seeding for infection. Femurs from C57BL/6 Casp1 -/mice were a gift from Dr. Russell Vance (University of California, Berkeley).

Competitive index
Six-to twelve-week-old sex-and age-matched mice were anesthetized with ketamine/xylazine and infected via the intranasal route as previously described [20]. Mixed bacterial inoculums (1:1) containing a total of 5 x 10 6 bacteria were diluted and plated on selective medium (10 μg mL -1 chloramphenicol for plasmid selection). At 48 h post-infection, mice were euthanized and homogenates from extracted lung tissue were plated on selective media, as described [20].

Recombinant protein production and affinity chromatography
E. coli BL21 (DE3) overnight cultures were sub-cultured into fresh LB medium containing the appropriate antibiotics at 1:100 and grown with shaking for 3h at 37˚C. Recombinant protein production was induced with 1mM IPTG and cultures were grown overnight with shaking at 16˚C. Bacteria were harvested by centrifugation at 1,400 r.c.f. for 10 min at 4˚C, washed with ice-cold PBS and incubated in bacterial lysis buffer [50mM Tris pH 8.0, 100mM NaCl, 1mM EDTA, 200μg mL -1 lysozyme, and complete protease inhibitor (Roche)] for 30 min on ice followed by addition of 2 mM DTT. Bacteria were sonicated and lysates were clarified by centrifugation at 21,000 r.c.f. for 15 min at 4˚C, as described [100]. Clarified lysates harboring GSTfusion proteins were incubated with magnetic glutathione agarose beads (Pierce) for 1hr at 4˚C with rotation and washed twice in Glutathione wash buffer [125mM Tris pH 7.4, 250mM NaCl, 1mM DTT, 1mM EDTA]. Clarified lysates harboring His 6 -Myc-fusion proteins were added to the beads and incubated at 4˚C with rotation for 1hr. Beads were washed twice with Glutathione wash buffer and proteins were eluted by boiling in Laemmli sample buffer. Proteins were separated by SDS-PAGE followed by Coomassie brilliant blue staining and Western blot (see below).

Cell culture and transfections
HEK 293T, HeLa, and RAW 264.7 cells (gifts from Dr. Craig Roy, Yale University) were maintained at 37˚C and 5% CO 2 in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% HI-FBS. THP-1 cells were purchased from the American Type Culture Collection (ATCC) and maintained in RPMI supplemented with 10% HIFBS and differentiated with 100 nM phorbal 12-myristate 13-acetate (PMA) for 4 days prior to infection with L. pneumophila (see below). All cell lines were used between passage 4-20.
HEK 293T cells were transfected with purified plasmid DNA (Table A in S1 Text) for 24 h using calcium phosphate as described [101]. HeLa cells were transfected for 24 h using jet-PRIME transfection reagent according to manufacturer's guidelines. Two hours before transfection, HeLa cells were washed and incubated in low-serum media (DMEM/4% HIFBS).

PLOS PATHOGENS
PA28αβ-and effector-mediated host defense Media were replaced with DMEM 10% HIFBS four hours post-transfection and assayed 24 h post-transfection.
To generate RAW 264.7 cells with stable plasmid integrations, cells were transfected with pcDNA::3xflag-legC4 or pcDNA::3xflag vector (Table A in S1 Text) using a Nucleofector 2b electroporator (Lonza, Basel, Switzerland) and cultured over 14 days with Zeocin selection (200-1000 μg mL -1 ). 3xFLAG-LegC4 expression was confirmed by Western blot and confocal microscopy using a-FLAG antibodies (see below) and stable RAW 264.7 cells were maintained in culture in the presence of 200 μg mL -1 Zeocin. For L. pneumophila infections, cells were seeded one day prior to infection and infected in the absence of Zeocin.

Immunoprecipitation
Transiently transfected HEK 293T cells or RAW 264.7 cells were washed with phosphate-buffered saline and lysed in ice-cold NP-40 buffer [1% non-iodet P40 (v/v), 20 mM Tris pH 7.5, 150 mM NaCl, 10 mM Na 4 P 2 O 7 , 50 mM NaF, complete protease inhibitor (Roche)]. Lysates were clarified and proteins were immunoprecipitated using α-FLAG M2 magnetic beads or protein G-conjugated Dynabeads (ThermoFisher) pre-incubated with rabbit α-GFP antibody (Abcam, ab6556), as indicated and according to manufacturer's instructions. Input samples were collected from cell lysates prior to incubation with beads. Beads and input samples were resuspended in 3x Laemmli sample buffer for SDS-PAGE and Western blot analysis.

Protein carbonyl assays
Protein carbonyls were quantified using the OxiSelect Protein Carbonyl ELISA Kit (Cell Biolabs) following manufacturer's instructions. Briefly, RAW 264.7 cells were seeded at 2x10 6 in a 10 cm tissue culture dishes in DMEM 10% HIFBS. The next day, media were aspirated, and cells were incubated in DMEM/2.5% HIFBS in the presence or absence of 10 μM H 2 O 2 . After 24 h, cells were washed twice in PBS and lysed in ice-cold 500 μL of detergent-free lysis buffer [25 mM HEPES pH 7.4, 150 mM NaCl, 10 mM MgCl 2 , 1 mM EDTA, 2% glycerol (v/v)] supplemented with complete protease inhibitor cocktail (Roche). Cell lysates were scraped into a pre-chilled microcentrifuge tube and incubated with agitation for 30 min at 4˚C followed by sonication at 40% intensity three times in 10 second intervals. Lysates were clarified by centrifugation at 4˚C for 20 min at 13,000 r.c.f. Supernatants were transferred to a fresh tube, snap frozen in liquid nitrogen for 2 min and stored at -80˚C until use. Protein concentrations were quantified using a Coomassie Plus (Bradford) Protein Assay (Pierce) and diluted to 10 μg mL -1 . Samples and standards were adsorbed to a 96-well protein binding plate and ELISA was performed according to manufacturer's instructions. Absorbance at 450 nm was quantified on a BioTek Epoch2 microplate reader.

PLOS PATHOGENS
PA28αβ-and effector-mediated host defense Protein carbonyls were visualized using the OxiSelect Protein Carbonyl Immunoblot Kit (Cell Biolabs). Tnrf1 -/-BMDMs were seeded in 24-well plates at 2.5x10 5 one day prior to infection with indicated L. pneumophila strains at an MOI of 50 in the presence or absence of 10 μM H 2 O 2 . One hour after infection, cells were washed 3x with PBS -/-, media were replaced and cells were incubated for an additional 3 h in the presence or absence of H 2 O 2 . Cells were washed in ice-cold PBS and lysed in 120 μL ice-cold NP-40 lysis buffer (see above). Lysates were diluted in 3x Laemmli sample buffer and boiled for 10 min. Proteins were separated on a 4-20% gradient SDS-PAGE gel and transferred to PVDF membrane using a BioRad TransBlot semi-dry transfer apparatus. Membranes were processed according to manufacturer's instructions using 10 min wash steps. Blots were stripped and re-probed with rabbit α-β-actin antibody (1:1000; Cell Signaling Technology) followed by goat-α-rabbit-HRP (1:5000; ThermoFisher). Blots were visualized as described above and densitometry was performed using Fiji ImageJ and Adobe Photoshop software.
Yeast two-hybrid analysis. Yeast two-hybrid screening was performed by Hybrigenics Services, S.A.S., Evry, France (http://www.hybrigenics-services.com). The legC4 coding sequence was PCR-amplified from pSN85::legC4 [20] and cloned into pB66 as a C-terminal fusion to Gal4 DNA-binding domain (Gal4-C4). The construct was checked by sequencing and used as a bait to screen a random-primed Mouse Spleen library constructed into pP6. pB66 derives from the original pAS2ΔΔ vector [102] and pP6 is based on the pGADGH plasmid [103].
Sixty million clones (6-fold the complexity of the library) were screened using a mating approach with YHGX13 (Y187 ade2-101::loxP-kanMX-loxP, mat-α) and CG1945 (mat-a) yeast strains as previously described [102]. 303 His+ colonies were selected on a medium lacking tryptophan, leucine and histidine, and supplemented with 2mM 3-aminotriazole to handle bait autoactivation. The prey fragments of the positive clones were amplified by PCR and sequenced at their 5' and 3' junctions. The resulting sequences were used to identify the corresponding interacting proteins in the GenBank database (NCBI) using a fully automated procedure. A confidence score (PBS, for Predicted Biological Score) was attributed to each interaction as previously described [104].

Immunofluorescence microscopy
To quantify LAMP1+ LCVs in naïve macrophages, 1x10 5 BMDMs were seeded on poly-Llysine (PLL)-coated glass coverslips in 24-well plates and infected with L. pneumophila at a multiplicity of infection (MOI) of 25 in triplicates. At 1h post-infection, cells were washed 3x in PBS to remove extracellular bacteria and cells were incubated for an additional 8 h in fresh media was added. Quantification of LAMP1+ LCVs in rTNF-treated BMDMs was performed as described but media were supplemented with 25 ng mL -1 rTNF throughout the course of infection, which was allowed to proceed for an additional 5 h after washing in PBS. Coverslips were fixed with 4% paraformaldehyde (PFA; Thermo Scientific), permeabilized in ice cold methanol, and blocked with blocking buffer [0.1% saponin (w/v), 1% HIFBS (v/v), 0.5% bovine serum albumin (BSA; w/v) in PBS]. Coverslips were stained with 1:1,000 rabbit α-L. pneumophila (Invitrogen, PA17227), 1:2,000 rat α-LAMP1 (Developmental Studies Hybridoma Bank) primary antibodies and 1:500 Alexa 488-conjugated goat α-rabbit and Alexa594-conjugated goat α-rat secondary antibodies (ThermoFisher) in blocking buffer. Nuclei were stained with Hoechst (ThermoFisher) at 1:2,000. Coverslips were mounted on glass slides with ProLong Gold Antifade Mountant (ThermoFisher). LAMP1+ LCVs were scored on a Leica DMiL LED inverted epifluorescence microscope (n = 300 cells/strain) and the researcher scoring the samples were blind to their identity.

L. pneumophila intracellular growth curves
To quantify L. pneumophila intracellular replication, macrophages were seeded in 24-well tissue culture plates and infected in triplicates at an MOI of 1 the next day. Differentiated BMDMs were seeded at 2.5x10 5 in seeding media (RPMI, 10% HIFBS, 7.5% L929 cell supernatant), THP-1 cells were seeded at 5x10 5 and differentiated with 100 nM PMA in RPMI 10% HIFBS for three days. Cells were washed with PBS and incubated in the absence of PMA for one day prior to infection. Stable RAW 264.7 cells were seeded at 2.5x10 5 in DMEM 2.5% HIFBS. At 1 h post-infection, cells were washed 3x with PBS followed by addition of fresh media. Macrophages were lysed in sterile water and colony forming units (CFU) were enumerated at the indicated time points as described [20]. Where indicated, 25 ng mL -1 rTNF (Gibco), 20-50 μM H 2 O 2 (VWR), 10 nM bafilomycin A1 (BAF; ApexBio), or 1 μM MCC950 (ApexBio) were added at the time of infection and maintained throughout. DMSO was used as a vehicle control where indicated.

Enzyme-linked immunosorbent assay (ELISA)
To quantify TNF secretion, 2.5 x 10 5 BMDMs were seeded in 24-well tissue culture plates and infected with the indicated L. pneumophila strains at an MOI of 10. After 1 h of infection, cells were washed with PBS, and media were replaced. Supernatants were at 8 h post-infection and used fresh or stored at -20˚C for up to 1 week and TNF was quantified using mouse TNF-α ELISA MAX kit (BioLegend) following manufacturer's instructions. Absorbance at 450 nm was quantified on a BioTek Epoch2 microplate reader.

Cytotoxicity assay
To quantify cell death, lactate dehydrogenase (LDH) activity in cell supernatants was quantified. BMDMs were seeded in triplicates at 2.5x10 5 in a 24-well tissue culture plate for 24 h and infected with the indicated L. pneumophila strains at an MOI of 10 in 500 μL of seeding media (RPMP/10% HIFBS/7.5% L929 cell supernatant) and incubated for the indicated times. Supernatants were transferred to a 96-well plate and centrifuged at 200 r.c.f. for 10 min. LDH was quantified using the CytoTox 96 Non-Radioactive Cytotoxicity Assay (Promega) according to manufacturer's instructions. Absorbance at 490 nm was quantified on a BioTek Epoch2 microplate reader and percent cytotoxicity was calculated by normalizing absorbance values to cells

PLOS PATHOGENS
PA28αβ-and effector-mediated host defense treated with lysis buffer. Where indicated, cells were primed with 1 μM PAM 3 CSK 4 (Tocris) for 24 h prior to infection and plasmid expression of legC4 was induced with 1 mM IPTG.

Statistical analysis
Statistical analyses were performed with GraphPad Prism 9 software using either Students' ttest, or two-way ANOVA, as indicated, with a 95% confidence interval. Data are presented as mean ± standard deviation (s.d.) or standard error of the mean (s.e.m.) and, unless otherwise indicated, statistical analyses were performed on samples in triplicates.

PLOS PATHOGENS
PA28αβ-and effector-mediated host defense stress in activated macrophages via intracellular ROS production and by passive diffusion of membrane-permeable extracellular ROS (H 2 O 2 ). (B) ROS indiscriminately carbonylate amino acid side chains, which perturbs protein folding and function, and turnover of these damaged proteins is mediated by PA28αβ-CP proteasomes. (C) LegC4 translocated into L. pneumophila-infected macrophages binds PA28α and modulates its activity by an unknown mechanism. (D) Impaired proteasome activity leads to formation of carbonylated protein aggregates, which are impervious to proteasomal degradation and trigger upregulation of lysosome biogenesis and fusogenic activity. (E) Increased phagolysosomal fusion with the Legionella-containing vacuole (LCV) may result from global increases in lysosomal degradation to maintain proteostasis under oxidative stress conditions. Image created with Biorender.com. (TIF) S1 Table. Clones of proteasome regulators identified as LegC4 interacting partners by yeast two-hybrid. (XLSX) S1 Text. Supplemental tables and references. The supplemental text contains a list of Legionella strains (Table A), plasmids (Table B), and oligonucleotide primers (