Effector-mediated subversion of proteasome activator (PA)28 ab enhances lysoso- mal targeting of Legionella pneumophila within cytokine-activated macrophages

16 Legionella pneumophila is a natural pathogen of protozoa that causes Legionnaires’ Disease pneumonia via replication within macrophages using hundreds of translocated effector proteins. 18 In metazoans, effectors can also enhance pathogen clearance via effector-triggered immunity. The effector LegC4 confers a fitness disadvantage on L. pneumophila uniquely within cytokine-acti- 20 vated macrophages, but the mechanism is unknown. Here, we demonstrate that LegC4 restriction 21 occurs via subversion of proteasome activator (PA)28 ab , which is induced by cytokines and func- 22 tions to resolve oxidative stress. LegC4 impaired resolution of oxidative stress and LegC4-medi- 23 ated restriction was abolished within PA28 ab -deficient macrophages. Impaired proteasome ac- 24 tivity upregulates lysosomal degradation pathways and, indeed, subversion PA28 ab by LegC4 en- 25 hanced lysosomal fusion with the Legionella -containing vacuole. PA28 ab has been traditionally 26 associated with antigen presentation; however, our data support a new model whereby subversion 27 of PA28 ab enhances macrophage cell-autonomous immunity against L. pneumophila . This work 28 provides a solid foundation to evaluate induced proteasome regulators as mediators of innate im- 29 munity. 30 that LegC4-mediated restriction occurs independently of This study supports a novel role for proteasome activator (PA)PA28 ab in effector-mediated defense against an intracellular bacterial pathogen. Our data show that LegC4 binds PA28 ab , subverts resolution of oxidative proteotoxic stress and promotes lysosomal restriction of L. pneu- mophila . Thus, we propose a model whereby prolonged oxidative proteotoxic stress resulting LegC4 subversion PA28 ab enhances lysosomal targeting of L. pneumophila within cyto- macrophages Fig S4 ). This model is supported by our data showing that (1) LegC4 binds PA28 a ; (2) PA28 ab is required for the effects of LegC4 on host cells; (3) LegC4 impairs resolution of oxidative proteotoxic stress; and (4) that oxidative stress and phagolysosomal fusion contribute to LegC4-mediated restriction of L. pneumophila. This work is the first describe as a target for a bacterial effector protein and a role for PA28 ab in effector-mediated host defense against an intracellular bacterial pathogen.


32
Many intracellular bacterial pathogens cause disease by replicating within host macrophages 33 (Mitchell et al., 2016). However, macrophages activated by TH1 proinflammatory cytokines, such 34 as tumor necrosis factor (TNF) and interferon (IFN)-g, are potently and broadly microbicidal. Ac-35 tivated macrophages are in a state of altered protein homeostasis (proteostasis), owing in part to 36 robust production of reactive oxygen species (ROS), which cause collateral damage to cellular 37 proteins. To cope with oxidative stress, cytokine signaling induces expression of proteasome acti-38 vator (PA)28ab, which engages proteolytic 20S proteasome core particles (CPs) and facilitates 39 degradation of oxidant-damaged proteins (Pickering et al., 2010). Impaired proteasome activity 40 causes compensatory upregulation of lysosomal degradation pathways, which are central to cell-41 autonomous immunity within activated macrophages (Dikic, 2016). PA28ab has been tradition-42 ally associated with antigen presentation and adaptive immunity (Graaf et al., 2011;Preckel et al., 43 1999); however, the contribution of PA28ab to antimicrobial innate host defense has not been 44 investigated.

46
Legionella pneumophila is a natural pathogen of amoebae and the etiological agent of Legion-47 naires' Disease, a severe inflammatory pneumonia of the elderly and immunocompromised that 48 results from uncontrolled bacterial replication within alveolar macrophages. L. pneumophila rep-49 lication within macrophages has resulted from extensive co-evolution with their protozoan hosts 6 since LegC4 confers a comparable fitness disadvantage on L. pneumophila within human THP-1 110 macrophages (Fig 1A-B). Further support and utility of the LegC4 overexpression model was ev-111 idenced by loss of LegC4-mediated restriction within BMDMs unable to produce or respond to 112 endogenous TNF (Myd88 -/and Tnfr1 -/-, respectively) (Fig 1C, 1D) (Ngwaga et al., 2019). IFN-g 113 signaling is sufficient for LegC4 restriction (Ngwaga et al., 2019); however, the LegC4 restriction 114 phenotype was retained within Ifngr1 -/-BMDMs (Fig 1E), indicating that the relatively low levels 115 of IFN-g secreted by macrophages are not sufficient for restriction.

116
We subsequently evaluated how cytokine signaling impacts the fitness of LegC4-deficient 117 bacteria in vivo. We confirmed our previous results showing that LegC4 is deleterious to L. pneu-118 mophila within the lungs of WT and TNFR1-deficient mice (Fig 1F). We also found that LegC4 119 still conferred a fitness disadvantage on L. pneumophila within the lungs of MyD88-and IFNGR1-120 deficient mice (Fig 1F). However, the fitness advantage associated with loss of LegC4 was less 121 pronounced in the lungs of Myd88 -/mice compared to WT, Tnfr1 -/and Ifngr1 -/- (Fig 1F), which 122 is mainly due to the relatively low levels of MyD88-indpendent IFN-g produced in the L. pneu-123 mophila-infected lung (Archer and Roy, 2006). These data support our previous work showing 124 that LegC4 augments redundant cytokine-mediated pathogen restriction mechanisms (Ngwaga Table S1). The highest-confidence interactor was PA28a 7 with 28 individual clones, all of which encoded a minimal overlapping region in the PA28a C-136 terminal domain (amino acid residues 105-249) ( Table S1). These data suggest a direct interac-137 tion between LegC4 and host PA28a, which is conserved in amoebae and induced by cytokines in 138 macrophages.

139
We initially validated the PA28a-LegC4 interaction by reciprocal co-immunoprecipitation 140 and immunofluorescence microscopy on ectopically produced epitope tagged fusion proteins. Ec-141 topic expression of PA28a was necessary in these experiments since it is expressed only in re-142 sponse to cytokine signaling (Fabunmi et al., 2001;Seifert et al., 2010). We found that PA28a-

150
( Fig 2E). Together, these data show that LegC4 interacts with host PA28a, which has not been 151 previously identified as a target of bacterial effectors.

153
LegC4-mediated restriction of L. pneumophila is abolished in PA28ab-deficient 154 BMDMs. Based on the interaction between PA28a and LegC4, we tested the hypothesis that 155 PA28a is important for LegC4-mediated restriction of L. pneumophila within BMDMs. PA28a-156 deficient (Psme1 -/-) mice are not commercially available, so we evaluated the role of PA28a using 8 interaction of PA28a with 20S proteolytic proteasome core particles (CPs) (Wilk et al., 2000). We 161 found that PA28ab was necessary for LegC4-mediated restriction of L. pneumophila since the no 162 growth attenuation was associated with legC4 overexpression compared to control strains within 163 Psme1/2 -/-BMDMs (Fig 2F). Thus, our data suggest that PA28ab activity is important for LegC4-164 mediated restriction and support a model whereby LegC4 binds and regulates the function of

176
Our yeast two-hybrid data suggest that LegC4 binds the CTD of PA28a, which is involved 177 in PA28ab docking on proteolytic proteasome core particles and consequent proteasomal degra-178 dation of oxidant damaged (carbonylated) proteins. We rationalized that LegC4-meditated inhi-179 bition of PA28ab activity would impair degradation of carbonylated proteins and tested this by 180 quantifying protein carbonyls within macrophages. We first tested this within uninfected RAW 181 264.7 cells stably producing 3xFLAG-LegC4 (RAW-LegC4). Stable production of 3xFLAG-LegC4 182 was confirmed by confocal microscopy (Fig 3A) and Western blot (Fig 3B). Immunoprecipita-183 tion was necessary to visualize LegC4 since it is produced by macrophages at low abundance.

184
However, this amount of ectopically produced LegC4 was sufficient to attenuate L. pneumophila 9 ∆flaA replication when compared to vector-transfected control (RAW-Cntrl) cells (Fig 3C). At-186 tenuated growth was observed at 48 h post-infection, similar to the kinetics of LegC4 restriction 187 within BMDMs (Fig 1A), which suggests that RAW-LegC4 cells are suitable to evaluate whether 188 LegC4 impacts the concentration of carbonylated proteins within macrophages under oxidative 189 stress conditions.

190
We quantified the abundance of carbonylated proteins within lysates of uninfected RAW-

191
LegC4 and -Cntrl cells in the presence or absence H2O2 to induce oxidative stress. Under native 192 conditions, the concentration of carbonylated proteins within RAW-LegC4 and -Cntrl cells did 193 not differ (Fig 3D). However, under oxidative stress conditions, the concentration of protein car-194 bonyls in RAW-LegC4 cells was significantly greater than control cells (Fig 3D). These data sug-  BMDMs, in which LegC4 is not restrictive (see Fig 1C). Tnfr1 -/-BMDMs were infected for 4 h in 202 the presence or absence of H2O2 and protein carbonyls were visualized by Western blot. We found 203 that the abundance of carbonylated proteins was increased within H2O2-treated BMDMs infected 204 with L. pneumophila ∆flaA∆legC4 (plegC4) relative to ∆flaA and empty vector control strains 205 (Fig 3E). Since no LegC4-mediated differences were observed within untreated Tnfr1 -/-, it is un-206 likely that the increase in protein carbonyls observed is due to a LegC4-mediated upregulation 207 global ROS production by infected BMDMs. Together, these data support our hypothesis that 208 LegC4 impairs resolution of oxidative proteotoxic stress within macrophages.

209
Based on our data showing that LegC4 impairs resolution of oxidative proteotoxic stress, 210 we hypothesized that oxidative stress is sufficient for LegC4-mediated restriction within BMDMs.

10
To test this hypothesis, we quantified L. pneumophila replication within Tnfr1 -/-BMDMs in the 212 presence or absence of H2O2 to induce oxidative stress. We found that replication of the L. pneu-213 mophila ∆flaA∆legC4 (plegC4) strain was significantly attenuated within H2O2-treated Tnfr1 -/-214 BMDMs compared to L. pneumophila ∆flaA and ∆flaA∆legC4 empty vector (pEV) control strains 215 (Fig 3F). These differences represent a 78% decrease in colony forming units (CFU), which is 216 striking since experimental limitations necessitated a low concentration (10 µM) of H2O2. It is 217 unlikely that growth attenuation was due to direct L. pneumophila killing by H2O2 since (1) repli-218 cation of control L. pneumophila strains did not differ between H2O2-treated and untreated     Fig 1D); thus, we tested whether growth differences were due to increased TNF secre- or ∆∆ (pEV) compared to WT BMDMs but no differences in secretion from BMDMs infected with 249 L. pneumophila ∆∆ (plegC4) (Fig S1B). These data suggest that LegC4-and PA28ab-mediated 250 growth phenotypes are not due to enhanced TNF production by infected BMDMs.

254
These data suggest neither LegC4 nor PA28ab restriction phenotypes are a consequence of host 255 cell death.

258
Our current data suggest that LegC4 may impair PA28ab activity, but how this leads to bacterial

268
We initially tested the hypothesis that LegC4 increases LAMP1 localization to LCVs by immuno-

278
We subsequently tested the hypothesis that phagolysosomal fusion is important for

295
Inflammasome activation is dispensable for LegC4-mediated restriction. Our current 296 data suggest that LegC4-mediated restriction occurs via lysosomal fusion with LCVs under oxida-297 tive proteotoxic stress conditions. However, sustained oxidative stress and proteostasis perturba-298 tions can activate the NLRP3 inflammasome, which culminates in activation of caspase-1, an ef-299 fector caspase responsible for pyroptotic cell death and inflammatory cytokine secretion. How-300 ever, pharmacological inhibition of the NLRP3 inflammasome with MCC950 did not affect LegC4-301 mediated restriction since L. pneumophila ∆∆ (plegC4) was still attenuated for replication com-302 pared to L. pneumophila ∆∆ (pEV) and ∆flaA within WT BMDMs (Fig S3A). We also found that 303 LegC4-mediated restriction was preserved within BMDMs derived from caspase-1-deficient 304 (Casp1 -/-) mice (Fig S3B). Within cultured macrophages, pyroptotic cell death is responsible for 305 L. pneumophila restriction downstream of inflammasome activation. Our data suggest that LegC4 306 does not enhance LDH release from infected BMDMs (Fig S3C); however, pyroptotic cell death 307 is more pronounced within BMDMs primed with toll-like (TLR) receptor agonists (Bergsbaken et  shares 34% identity with mouse PA28a. Sequence identity between PA28ab and AcPA28 is pri-333 marily in the C-terminal domain, which is important for engaging 20S CPs and was contained in 334 all PA28a clones identified in our yeast two-hybrid screen (see Table S1). These data have led us   Fig 3F). This suggests that the con-

425
Bone marrow was harvested from seven-to twelve-week-old mice as described (Case and

500
Supernatants were transferred to a fresh tube, snap frozen in liquid nitrogen for 2 min and stored 501 at -80˚C until use. Protein concentrations were quantified using a Coomassie Plus (Bradford)

502
Protein Assay (Pierce) and diluted to 10 µg mL -1 . Samples and standards were adsorbed to a 96-503 well protein binding plate and ELISA was performed according to manufacturer's instructions.

590
To quantify cellular metabolic activity, an MTT colorimetric assay was used. BMDMs were 591 seeded in a 96-well tissue culture plate at 1x10 5 and infected the next day with the indicated L.