Ceramide-1-phosphate is a regulator of Golgi structure and is co-opted by the obligate intracellular bacterial pathogen Anaplasma phagocytophilum

ABSTRACT Many intracellular pathogens structurally disrupt the Golgi apparatus as an evolutionarily conserved promicrobial strategy. Yet, the host factors and signaling processes involved are often poorly understood, particularly for Anaplasma phagocytophilum, the agent of human granulocytic anaplasmosis. We found that A. phagocytophilum elevated cellular levels of the bioactive sphingolipid, ceramide-1-phosphate (C1P), to promote Golgi fragmentation that enables bacterial proliferation, conversion from its non-infectious to infectious form, and productive infection. A. phagocytophilum poorly infected mice deficient in ceramide kinase, the Golgi-localized enzyme responsible for C1P biosynthesis. C1P regulated Golgi morphology via activation of a PKCα/Cdc42/JNK signaling axis that culminates in phosphorylation of Golgi structural proteins, GRASP55 and GRASP65. siRNA-mediated depletion of Cdc42 blocked A. phagocytophilum from altering Golgi morphology, which impaired anterograde trafficking of trans-Golgi vesicles into and maturation of the pathogen-occupied vacuole. Cells overexpressing phosphorylation-resistant versions of GRASP55 and GRASP65 presented with suppressed C1P- and A. phagocytophilum-induced Golgi fragmentation and poorly supported infection by the bacterium. By studying A. phagocytophilum, we identify C1P as a regulator of Golgi structure and a host factor that is relevant to disease progression associated with Golgi fragmentation. IMPORTANCE Ceramide-1-phosphate (C1P), a bioactive sphingolipid that regulates diverse processes vital to mammalian physiology, is linked to disease states such as cancer, inflammation, and wound healing. By studying the obligate intracellular bacterium Anaplasma phagocytophilum, we discovered that C1P is a major regulator of Golgi morphology. A. phagocytophilum elevated C1P levels to induce signaling events that promote Golgi fragmentation and increase vesicular traffic into the pathogen-occupied vacuole that the bacterium parasitizes. As several intracellular microbial pathogens destabilize the Golgi to drive their infection cycles and changes in Golgi morphology is also linked to cancer and neurodegenerative disorder progression, this study identifies C1P as a potential broad-spectrum therapeutic target for infectious and non-infectious diseases.

proteins and lipids from the endoplasmic reticulum (ER) are packaged into coat protein complex II (COPII)-coated vesicles that are delivered through the ER-Golgi intermedi ate compartment (ERGIC) to the cis-Golgi cisternae and migrate through the stack to the trans-Golgi network (TGN) for delivery to other organelles or out of the cell (Fig. 1).In retrograde trafficking, COPI vesicles transport misfolded proteins from the Golgi through the ERGIC to the ER (1,2).Golgi matrix proteins, including Golgi reassembly stacking proteins (GRASPs) and golgins, maintain Golgi structure.Golgins tether vesicles to cisternal membranes (2).Two recent studies confirmed that, contrary to their name, GRASPs do not hold the cisternae into stacks but instead link the stacks into the Golgi ribbon (3,4).GRASP65 (GRASP of 65 kDa), concentrated in the cis-Golgi, and GRASP55, present in the medialand trans-Golgi, perform non-redundant roles in ribbon linking (5,6) (Fig. 1).Both also critically influence Golgi vesicle transport (3,7).
The Golgi is a dynamic organelle that undergoes morphological changes under certain physiologic conditions (2).Knockout or acute degradation of GRASP55 and GRASP65 leads to ribbon uncoupling (3,4,7); and this structural change has been proposed to indirectly result from the imbalance of cargo transport caused by GRASP depletion (3).GRASP phosphorylation also promotes loss of ribbon integrity.Cyclindependent kinase 1 phosphorylation of GRASP55 and GRASP65 at the onset of mitosis leads to ribbon unlinking (2,(8)(9)(10).c-Jun N terminal kinase (JNK) and protein kinase Cα (PKCα ) phosphorylation of GRASP65 and GRASP55, respectively, also induce ribbon disruption.Cell division control protein 42 (Cdc42) modulates Golgi morphology and Golgi-to-ER retrograde trafficking upstream of JNK (11)(12)(13).In its active GTP-bound state, Cdc42 associates with cellular membranes and binds COPI, which culminates in retrograde traffic disruption and Golgi destabilization (14,15).GTP-bound Cdc42 also accelerates Golgi anterograde trafficking (16).Moreover, PKCα prompts Cdc42 membrane localization (17,18), indicating that, in addition to directly inducing Golgi disruption by phosphorylating GRASP55 (19), it indirectly modulates Golgi stability through Cdc42.Golgi morphology is also altered under certain pathologic conditions including inflammation, cancer, and neurodegeneration (1,20).Importantly, Golgi destabilization does not impair transport function but rather yields more membrane surfaces for vesicle budding and thereby accelerates anterograde transport (2,21), a phenomenon that ties in with the influential roles of GRASP55 and GRASP65 on vesicle transport and Golgi structural integrity (3,4,7).
In this study, we investigated the hypothesis that increases in C1P levels induced by A. phagocytophilum disrupt Golgi structure to bolster infection.Elevated C1P levels were found to benefit the bacterium during infection of myeloid cells and mice, and the sphingolipid disrupts Golgi structure by promoting Cdc42 membrane-binding, PKCα phosphorylation of GRASP55, and JNK phosphorylation of GRASP65.Cells overexpress ing phosphorylation-resistant versions of GRASP55 and GRASP65 are inhibited for both C1P-and A. phagocytophilum-induced Golgi fragmentation and poorly support A. phagocytophilum infection.Overall, C1P is a major regulator of Golgi morphology that can be exploited for pathogenic endosymbiont success.

CERK-derived C1P is critical for A. phagocytophilum intracellular replication, infection cycle progression, and productive infection in mice
To determine if CERK-derived C1P is required for A. phagocytophilum infection, HL-60 and RF/6A cells infected in the presence of NVP231 were examined at 24 hpi by qPCR.Decreases in the bacterial DNA load of approximately 98% and 50% were observed in NVP231-treated RF/6A and HL-60 cells, respectively (Fig. 2E; Fig. S1D).To assess the importance of CERK-derived C1P for the pathogen's conversion from the replicative RC morphotype to the infectious DC form, the experiment was repeated to assess the expression of aph1235, which encodes a DC-specific marker that is induced when A. phagocytophilum undergoes RC-to-DC conversion between 24 and 32 h (73).Whereas aph1235 expression increased over the time course for control cells, its levels were pronouncedly reduced in NVP231-treated cells (Fig. 2F; Fig. S1E).Next, CERK was downregulated in RF/6A cells by siRNA, followed by incubation with DC organisms.CERK downregulation induced a several-fold reduction in the bacterial DNA load (Fig. 2G).To define the relevance of CERK to A. phagocytophilum infection in vivo, CERK −/− or wild-type mice were inoculated with DC organisms after which the bacterial load in the peripheral blood was determined by qPCR.In wild-type mice, bacterial load peaked by day 12, followed by a gradual decline to nearly undetectable levels by day 28 (Fig. 2H). A. phagocytophilum DNA was barely detectable at all time points in CERK −/− mice.
To further validate the importance of C1P to A. phagocytophilum infection, we investigated the effects of increasing C1P levels on bacterial-induced Golgi fragmenta tion, the Anaplasma load, and ApV expansion.C1P levels were elevated by siRNA-medi ated downregulation of CPTP as confirmed by qRT-PCR (Fig. 3A) and UPLC-ESI-MS/MS analysis (Fig. 3B).Elevating C1P levels enhanced A. phagocytophilum-induced Golgi fragmentation (Fig. 3C and D), and concomitantly increased the bacterial DNA load and ApV number per cell (Fig. 3E and F).Collectively, these data demonstrate that CERK-derived C1P is critical for A. phagocytophilum to promote Golgi fragmentation, phagocytophilum DC organisms (I) or left uninfected (U).At 24 h, C1P levels were analyzed using UPLC-ESI-MS/MS with the two main chain lengths of C1P depicted.One-way ANOVA with Tukey's post hoc test was used to test for significant differences in D-e-C 14:0 C1P and D-e-C 16:0 C1P levels among the conditions.
(C and D) CERK-derived C1P is required for A. phagocytophilum-induced Golgi fragmentation.RF/6A cells were treated as in (B).At 24 h post-infection, the cells were fixed, immunolabeled with TGN46 antibody (Alexa Fluor 594 secondary, red), stained with DAPI (blue) to visualize host cell nuclei and bacterial nucleoids, and examined using laser scanning confocal microscopy (LSCM).Merged fluorescence images are shown (C).The regions that are denoted by hatched lined boxes are magnified in the insets that are demarcated by solid lined boxes.Scale bar, 10 µm.(D) Fluorescence micrographs were examined to determine the mean (±SD) percentage of cells with dispersed Golgi and mean (±SD) Golgi area.Data are representative of four independent experiments in which 50 cells were examined per condition each time.Statistical analysis was performed using one-way ANOVA with Tukey's post hoc test.(E and F) CERK inhibition inhibits A. phagocytophilum infection and RC-to-DC conversion.RF/6A cells were treated as in (B).qRT-PCR and the 2 −ΔΔCT method were used to measure the bacterial load as relative A. phagocytophilum 16S rRNA gene (aph16S)-to-human β-actin expression at 24 h post-infection (E) and relative aph1235-to-aph16s expression at 24, 28, and 32 h post-infection (F).Data are indicative of four separate experiments.Statistical analysis was performed using an unpaired, two-tailed t test with Welch's correction for panel (E) and using one-way ANOVA with Tukey's post-hoc test for Panel F. (G) siRNA downregulation of CERK abrogates A. phagocytophilum infection.RF/6A cells were transfected with CERK siRNA (siCERK) or non-targeting control siRNA (siNT), followed by incubation with A. phagocytophilum DC organisms 48 h later.At 24 h post-infection, the bacterial load was determined using qRT-PCR.Data are representative of three independent experiments.Statistical analysis was performed using an unpaired, two-tailed t test with Welch's correction.(H) A. phagocytophilum fails to productively infect CERK −/− mice.Wild-type (CERK +/+ ) and CERK −/− mice were injected intraperitoneally with 1 × 10 8 A. phagocytophilum DC organisms.Peripheral blood samples collected on day 0 (prior to infection) and the indicated days post-infection were analyzed by qPCR and the 2 −ΔΔCT method to measure the relative A. phagocytophilum aph16S gene to murine DNA levels.Data are representative of two independent experiments each conducted with four to five male and female mice per group (total of 9-10 mice per group).Statistical analysis was performed using repeated measures ANOVA *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.which, in turn, is important for it to replicate, progress through its infection cycle, and productively infect mammalian hosts.
To examine if Cdc42-regulated Golgi disruption benefits A. phagocytophilum infection, siCdc42 or siNT-treated RF/6A cells (Fig. 6A) were incubated with DC organisms.At 24 hpi, Golgi fragmentation was inhibited by 16% and TGN immunosignal in the ApV lumen was reduced by 22% in Cdc42 knockdown cells (Fig. 6B through D).Concom itantly, Cdc42 downregulation had no effect on the bacterial load, but significantly, albeit modestly, impaired ApV expansion (Fig. 6E and F).Also, Cdc42 downregulation nearly abrogated A. phagocytophilum expression of APH0032 (Fig. 6A), a secreted effector protein that is upregulated and localizes to the ApV membrane during ApV maturation prior to DC release (25,71,72).Thus, while Cdc42 downregulation does not inhibit A. phagocytophilum replication, it reduces TGN anterograde traffic into the ApV, which is linked to an inhibition in ApV maturation and hence progression of the pathogen's infection cycle.
To examine if the C1P-mediated PKCα/Cdc42/JNK pathway induces GRASP55 and GRASP65 phosphorylation, HUVECs were treated with siCdc42, Gö6976, or SP600125 independently or in conjunction with siCPTP.Threonine-and serine-phosphorylated proteins recovered by immunoprecipitation were subjected to Western blot analysis to assess total and phosphorylated forms of GRASP55 and GRASP65.Cdc42, PKCα, and JNK inhibition each markedly reduced C1P-induced phosphorylation of GRASP55 and GRASP65 (Fig. 8A through C).Taken together, these results show that C1P induces the PKCα/Cdc42/JNK signaling pathway that ultimately results in GRASP phosphorylation and subsequent disruption of the Golgi apparatus.

A. phagocytophilum requires GRASP55 and GRASP65 phosphorylation to promote Golgi fragmentation and for optimal infection
Because C1P stimulates signaling that culminates in GRASP phosphorylation and Golgi destabilization, and since A. phagocytophilum promotes Golgi fragmentation as a promicrobial strategy (25), we evaluated if the bacterium induces Golgi structural changes in a GRASP phosphorylation-dependent manner.RF/6A cells were transfected to express GRASP55-GFP and GRASP65-GFP, phosphorylation-resistant mutants thereof, or GFP alone, followed by incubation with DC organisms.At 24 h, the percentage of A. phagocytophilum infected cells co-expressing GRASP55-TT225,249AA-GFP and GRASP65-TSS216,220,277AAA-GFP that had fragmented Golgi was reduced by at least 70% versus controls (Fig. 8D and E).Moreover, ApVs in cells expressing phosphorylation-resistant GRASP55 and GRASP65 harbored markedly fewer bacteria than in cells co-expressing GFP-tagged wild-type GRASP55 and GRASP65 or GFP (Fig. 8D and F).Therefore, A. phagocytophilum perturbs Golgi morphology in a GRASP phosphorylation-dependent manner, which, in turn, is critical for optimal proliferation within its vacuolar niche.

DISCUSSION
Obligate intracellular pathogens are master cell biologists that exploit host processes to their advantage and can therefore serve as useful tools for elucidating unrecognized eukaryotic cellular pathways.By studying A. phagocytophilum, we revealed a role for CERK-derived C1P as a regulator of Golgi structure.C1P induces signaling through PKCα, Cdc42, and JNK that leads to GRASP phosphorylation and, consequently, perturbation of Golgi morphology (Fig. 8).When these findings are considered together with prior reports that (i) fragmented Golgi exhibit accelerated anterograde trafficking (1), (ii) GRASP phosphorylation or depletion increases Golgi anterograde trafficking (7,107), (iii) MVBs receive anterograde traffic from the TGN (45), and (iv) A. phagocytophilum not only lives in a pathogen-modified MVB but also parasitizes TGN-derived vesicles that are delivered into the ApV lumen (25,44), the essentiality of C1P to the bacterium's infection cycle becomes clear.By elevating host cell C1P levels and coopting the C1P-induced signaling axis, A. phagocytophilum upregulates TGN vesicle delivery into its parasitopho rous vacuole, which increases the bacterial load, promotes ApV maturation, and is key for RC-to-DC conversion (Fig. 9).Pharmacologic inhibition and siRNA-mediated downregula tion of CERK pronouncedly lower A. phagocytophilum levels and bacterial expression of the DC marker, aph1235.Notably, treatment with siRNA targeting Rab10, which mediates anterograde trafficking from the TGN, has the same effect (25).The importance of C1P to A. phagocytophilum is recapitulated in vivo, as it fails to infect CERK −/− mice.Conversely, raising C1P levels via CPTP downregulation increases bacterial levels and ApV numbers per cell.Because each ApV derives from the entry of a single A. phagocytophilum organism (43), the increase in ApVs per cell could be due to the acceleration of the infection cycle leading to more reinfection events.Alternatively, it could result from a Golgi fragmentation-associated increase in the trafficking of receptors critical for A. phagocytophilum entry to the cell surface.How A. phagocytophilum elevates C1P levels is unclear, but potential unexplored or understudied mechanisms include activating CERK, decreasing C1P transport, or suppressing C1P catabolism.
D-e-C14:0 C1P and D-e-C16:0 C1P levels are elevated in both A. phagocytophilum infected cells and cells in which CPTP has been depleted using siRNA.There have been no reports for specific chain lengths of C1P having variable functions.Any differential acyl chain lengths for C1P are currently surmised to follow the available substrate ceramide for CERK and could be indicative of cellular location (54,56,57,108), but this has not been strongly interrogated.To date, studies have been hampered by the low levels of C1P in many cell types allowing for the main forms of C1P, which are D-e-C14:0 C1P and D-e-C16:0 C1P, to be appropriately quantitated (50,51,54,109).
CERK-derived C1P activates Cdc42 to enhance its translocation from the cytoplasm to cellular membranes.Downregulation of Cdc42 abrogates the effect of C1P on the induction of Golgi destabilization showing that Cdc42 is a key factor in the down stream signaling that C1P modulates.Moreover, Cdc42 downregulation also inhibits (p-Thr) and serine-phosphorylated (p-Ser) proteins immunoprecipitated (IP) from duplicate whole-cell lysates (WCLs) were Western-blotted and probed with antibodies specific for GRASP55 and GRASP65, respectively.WCLs were probed with GRASP55 and GRASP65 antibodies as loading controls.Levels of phospho-GRASP55 (P-G55) and phospho-GRASP65 (P-G65) were assessed as the mean (±SD) normalized ratio of p-G55:GRASP55 (G55) and p-G65:GRASP65 (G65) densitometric signals.Data are representative of four independent experiments.Statistical analysis was performed using one-way ANOVA with Tukey's post hoc test.(D-F) GRASP phosphorylation is critical for A. phagocytophilum-induced Golgi fragmentation and optimal A. phagocytophilum fitness.RF/6A cells expressing WT GRASP55-GFP and WT GRASP65-GFP, phosphorylation-resistant GRASP55-GFP (TT225,249AA) and phosphorylation-resistant GRASP65-GFP (TSS216,220,274AAA), or GFP were infected with A. phagocytophilum DC organisms.(D) At 24 h, the cells were fixed, immunolabeled with TGN46 antibody ApV maturation as shown by a lack of APH0032 induction and reduction in ApV area.A functional role of APH0032 has not been discerned.Whether Cdc42 downregulation impairs the expression of APH0032 exclusively or other unidentified AVM-localized A. phagocytophilum effectors that are coincidentally expressed with APH0032 during the infection cycle is unknown.If the latter is true, then Cdc42 could play a significant role in influencing ApV maturation and A. phagocytophilum pathogenesis.Additionally, Cdc42 regulation of actin-related processes, which are targeted by numerous intracellu lar pathogens (110)(111)(112), cannot be ruled out as contributing to A. phagocytophilum intracellular fitness.Why Cdc42 downregulation modestly inhibits Golgi fragmentation in RF/6A cells, but pronouncedly does so in HUVECs and why inhibiting or downregulating CERK strongly impairs this mechanism in both cell lines is unclear.Specifically, our findings suggest that either Golgi fragmentation regulates only the ApV maturation process of the A. phagocytophilum infection cycle or C1P can facilitate Golgi fragmenta tion in RF/6A cells by additional pathways independent of Cdc42.PKCα is known to directly phosphorylate GRASP55 (19), which would induce Golgi ribbon uncoupling.This alternate/cooperating mechanism of GRASP phosphorylation would be Cdc42-inde pendent and explain the significant, but minimal effect of Cdc42 downregulation on A. phagocytophilum-induced Golgi fragmentation in RF/6A cells.Indeed, activation of PKCα at the Golgi apparatus could bypass the need for Cdc42 activation to induce Golgi instability and directly induce GRASP55 phosphorylation.This report supports both plausible mechanisms.Moreover, it suggests a cooperative role between PKCα direct phosphorylation of GRASP55 and the indirect signaling cascade of PKCα activation of Cdc42 and subsequent GRASP phosphorylation (Fig. 8).
Prior to this report, whether A. phagocytophilum infection activates JNK signaling was unknown.Herein, we oriented the signal transduction pathway from PKCα to Cdc42 to JNK downstream of the pathogen inducing an increase in CERK-derived C1P.We further showed that a key biological mechanism modulated by JNK that drives the A. phagocy tophilum infection cycle is GRASP65 phosphorylation and ensuing Golgi fragmentation.JNK −/− mice are highly resistant to A. phagocytophilum infection, and this has been linked to the repressive role of JNK on CD1d-restricted natural killer T cells production of IFNγ, a cytokine that is critical for clearing the infection in mice (113).In lieu of our findings, it is expected that cis-Golgi uncoupling induced by the JNK-GRASP65 axis would be inhibited in JNK −/− mice, which would at least partially impair the pathogen's ability to parasitize anterograde traffic and synergize with IFNγ-mediated clearing.
How CERK-derived C1P activates PKCα to facilitate subsequent activation of Cdc42 and JNK is not known, but PKCα possesses a C2-domain (114,115).C1P binds and activates group IVA phospholipase A 2 via this enzyme's C2 domain facilitating its translocation to the Golgi (49-52, 57, 58, 61, 116, 117).This gives rise to the hypothesis that C1P analogously binds PKCα at its C2-domain to translocate PKCα to the Golgi.The possibility that C1P associates with PKCα may explain the conundrum as to how classical PKCs translocate to different membranes in response to specific agonists.For example, classical PKCs prefer the phosphatidylserine (PS)-rich plasma membrane when acutely activated versus internal membranes.Alternatively, sustained activation of classical PKCs such as PKCα induces the translocation of these enzymes to a pericentriolar region/ centrioles, a site of mitotic signaling, and Golgi fragmentation is a key mechanism in mitosis (118)(119)(120)(121)(122)(123)(124)(125).The link between C1P, PKCα, and Golgi fragmentation established herein may explain the known role of C1P in cellular proliferation (46,55,69).
Inducing changes in Golgi morphology is a broadly conserved strategy among diverse pathogens.Yet, the underlying mechanisms are not well understood and only a few targeted host cell factors are known (22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40).Many of those that have been identified are Golgi matrix and tethering proteins.Chlamydia trachomatis promotes cleavage of golgin-84 (31).Legionella pneumophila uses its effectors LegA15 to dislocate p115 from the Golgi and SdeA to phosphoribosyl-ubiquitinate GRASP55 and GRASP65 to prevent their oligomerization (23,38).Whereas Chlamydia and Legionella directly promote Golgi disassembly by targeting proteins responsible for maintaining the organelle's integrity, A. phagocytophilum indirectly does so through CERK-derived C1P.Human cytomegalovirus induces GRASP65 phosphorylation by an unknown means to fragment the Golgi as an essential step in infectious particle production (30).SARS-CoV-2 triggers Golgi dispersion by downregulating GRASP55 to accelerate viral trafficking and release (29,36,126).Interestingly, interaction of the SARS-CoV-2 spike protein with host angiotensin-convert ing enzyme 2 that mediates invasion is enhanced by heparan sulfate, the synthesis of which is increased by GRASP depletion (127)(128)(129).Furthermore, a major cause of Golgi morphological changes in Alzheimer's disease is GRASP65 phosphorylation (107).Given that SARS-CoV-2 can cause a neuropathological phenotype reminiscent of Alzheimer's Disease as well as clinical brain fog, it has been speculated that SARS-CoV-2 neuropathol ogy might be caused by viral-induced GRASP phosphorylation (127).As the host factors that drive Golgi fragmentation associated with the progression of other neurodegenera tive diseases, inflammatory disorders, and cancers are incompletely defined (1,20), the potential contribution of C1P to these conditions is worth considering.
Our study provides the foundation to explore C1P as an antimicrobial target to treat HGA as well as other infectious and non-infectious diseases whose pathologies are associated with GRASP-dependent Golgi structural changes.Importantly, CERK inhibitors are well tolerated in mice (50).As precedent for host-directed therapeutics against A. phagocytophilum, tricyclic antidepressants that inhibit aSMase, a key enzyme for the bacterium's intra-MVB/TGN parasitic lifestyle, halt A. phagocytophilum infection (66,67).Future studies are needed to determine how C1P activates both Cdc42 and PKCα along with the mechanism by which A. phagocytophilum elevates CERK-derived C1P levels as little is known regarding CERK activation and sustained C1P levels in specific cellular topologies.In closing, our work reveals a previously unappreciated role for C1P in regulating Golgi dispersal, a cellular pathology linked to the progression of diverse diseases and identifies the bioactive sphingolipid as a host factor that is critical for A. phagocytophilum pathogenesis.

Immunofluorescence microscopy
Cells were fixed with 4% (vol/vol) paraformaldehyde (Electron Microscopy Sciences) in phosphate-buffered saline (PBS) for 20 min, followed by washing three times with ice-cold PBS.The cells were permeabilized with 0.25% (vol/vol) Triton X-100 (Fisher Scientific) in PBS and washed three times with ice-cold PBS.The cells were blocked with 10% (vol/vol) goat serum in PBS for 30 min or 5% (vol/vol) BSA in PBS for 30 min, followed by probing with primary antibodies diluted in 1% (vol/vol) goat serum or 1% (vol/vol) BSA for 90 min.Primary antibodies were rabbit anti-TGN46 [Novus Biologicals (RRID:AB_10011762); 1:500], Calreticulin [Thermo Fisher Scientific (PA3-16862); 1:200], GFP [Thermo Fisher Scientific (A-21311); 1:500], and GM130 [Novus Biological (RRID:AB_2916095); 1:200].After washing with PBS, samples were incubated with secondary antibodies conjugated to Alexa Fluor fluorochromes (Invitrogen) in 1% (vol/vol) BSA for 1 h.DAPI (4′,6′-diamidino-2-phenylindole, Vesterfield) was used per the manufacturer's instructions to stain host nuclei and A. phagocytophilum nucleoids.Coverslips were mounted using Prolong Gold Anti-fade reagent (Invitrogen) and imaged at room temperature.Samples were imaged using a Keyence BZX-800 microscope; a TCS SP8 microscope (Leica Microsystems) affixed with an Andor iXon Life 888 EMCCD camera (Oxford Instruments) and a 63× water-immersion objective with 1.2 numeric aperture; or a Zeiss LSM laser scanning confocal microscope (Zeiss).For quantifying increases in Golgi area in fluorescence micrographs obtained using the Keyence BZX-800 microscope, ImageJ was used to measure Golgi area in pixels from 50 cells per condition.For quantifying increases in Golgi area in fluorescence micrographs obtained using the TCS SP8 microscope, LAS X (Leica Microsystems) software (version 3.7.4.23463) was used to measure Golgi area in μm 2 from 25 cells per condition in triplicate.ApV lumen TGN46 fluorescence intensity was determined using ImageJ.The A. phagocytophilum load in ApVs per 25 GFP-positive cells per condition and ApV area were assessed by examining fluorescence micrographs obtained using the TCS SP8 microscope using LAS X software.

Statistical analysis
Statistical analyses were performed using the Prism 8.0 software package (GraphPad, San Diego, CA).Statistical significance was set at P values of <0.05.models are available from the Chalfant Laboratory at the University of Virginia through a material transfer agreement Virginia Commonwealth University.

ADDITIONAL FILES
The following material is available online.

FIG 2
FIG 2 CERK-derived C1P regulates Golgi morphological changes induced by and is critical for A. phagocytophilum infection.(A) Schematic of sphingolipid metabolism.Ceramide, center of sphingolipid metabolism, is generated by either de novo synthesis by ceramide synthases (CERS) or by catabolism of sphingomyelin (SM) via a sphingomyelinase (SMase) [acid sphingomyelinase (aSMase) is depicted].Ceramide is phosphorylated by CERK to generate C1P, which is transported by CPTP to the plasma membrane and other organelles, where it is catabolized by lipid phosphatases (LPP).(B) CERK-derived C1P is induced by A. phagocytophilum infection.RF/6A cells were pretreated with either NVP231 (400 nM) or control (0.001% DMSO) for 1 h, followed by incubation with A.

FIG 3
FIG 3 Elevating C1P levels via CPTP siRNA-mediated downregulation enhances Golgi fragmentation and benefits A. phagocytophilum infection.(A and B) CPTP downregulation raises C1P levels.RF/6A cells were transfected with CPTP siRNA (siCPTP) or siNT.At 48 h, the cells were incubated with A. phagocytophilum DC organisms.At 24 h post-infection, qRT-PCR and the 2 −ΔΔCT method were used to assess relative CPTP-to-β-actin expression (A), while C1P levels were measured using UPLC-ESI-MS/MS (B).Data in (A) and (B) are representative of four and three independent experiments, respectively.Statistical analysis was performed using an unpaired, two-tailed t test with Welch's correction for panel (A) and one-way ANOVA with Tukey's multiple comparisons test for panel (B).(C and D) Elevating C1P levels enhances Golgi dispersal in uninfected and A. phagocytophilum infected cells.RF/6A cells were treated with siNT or siCPTP for 48 h were incubated with A. phagocytophilum DC organisms (Infected, I) or not (Uninfected, U).At 24 h post-infection, the cells were fixed, immunolabeled with antibody against TGN46 (Alexa Fluor 594 secondary, red), stained with DAPI (blue) to visualize host cell nuclei and bacterial nucleoids, and examined by LSCM.Merged fluorescence images are shown (C).Scale bar, 10 µm.(D) Fluorescence micrographs were examined to determine the mean (±SD) percentage of cells with fragmented Golgi and mean (±SD) Golgi area.Data are representative of four independent experiments in which 50 cells were examined per condition each time.Statistical analysis was performed using one-way ANOVA with Tukey's post hoc test.(E and F) CPTP downregulation increases the A. phagocytophilum load and the number of ApVs per cell.RF/6A cells were treated as in (A).(E) At 24 h post-infection, qRT-PCR and the 2 −ΔΔCT method were used to measure the bacterial load as relative A. phagocytophilum 16S rRNA gene (aph16S)-to-β-actin expression.(F) Fluorescence micrographs were examined to determine the mean (±SD) number of ApVs per cell.Data are representative of at least three independent experiments.Statistical analyses were performed using an unpaired, two-tailed t test with Welch's correction.*P < 0.05; **P < 0.01; ***P < 0.001; and ****P < 0.0001.

FIG 5 FIG 6
FIG 5 C1P induces Cdc42 membrane association, which is required for C1P-mediated Golgi morphological changes.(A) C1P induces Cdc42 membrane association.HUVECs were transfected with siNT, siCPTP, siCERK, or siCERK/siCPTP, followed by subcellular fractionation at 72 h.Cytosolic (CE) and membrane (ME) fractions were analyzed by Western blotting to determine the localization of Cdc42 as the mean (±SD) normalized ratio of Cdc42:voltage-dependent anion channel (VDAC, ME associated protein) and Cdc42: heat shock protein 90 (HSP90, CE associated protein) densitometric signals.Data are representative of four independent experiments.Statistical analysis was performed using one-way ANOVA, followed by Tukey's post hoc test.(B) Cdc42 downregulation prevents C1P-induced Golgi fragmentation.HUVECs were transfected with siRNAs as (A).At 72 h, the cells were fixed, immunolabeled with TGN46 antibody (Alexa Fluor 594 secondary, red), stained with DAPI (blue) to label host cell nuclei, and imaged with immunofluorescence microscopy.Scale bar, 20 µm.Fluorescence micrographs were examined to determine the mean (±SD) percentage of cells with dispersed Golgi and mean (±SD) Golgi area.Data are representative of four independent experiments in which 50 cells were examined per condition each time.Statistical analysis was performed using a one-way ANOVA with Tukey's post hoc test.***P < 0.001 and ****P < 0.0001.

FIG 7
FIG 7 PKCα and JNK are required for C1P-stimulated changes in Golgi morphology.(A) C1P induces PKCα and JNK phosphorylation.HUVECs were transfected with siNT, siCPTP, siCERK, or siCERK/siCPTP.At 72 h, phosphorylated PKCα (p-PKCα), p-JNK, total PKCα, and total JNK levels were analyzed by Western blotting.Levels of p-PKCα and p-JNK were assessed as the mean (±SD) normalized ratios of p-PKCα:PKCα and p-JNK:JNK densitometric signals.(B) PKCα and JNK inhibition blocks C1P-induced Golgi dispersal.HUVECs transfected with siNT or siCPTP cells were treated at 48 h post-siRNA addition with 3 µM Gö6976 or 25 µM SP006125.At 24 h post-treatment, the cells were fixed, immunolabeled with antibody against TGN46 (Alexa Fluor 594 secondary, red), stained with DAPI (blue) to visualize host cell nuclei, and examined by LSCM.Scale bar, 20 µm.Fluorescence micrographs were examined to determine the mean (±SD) percentage of (Continued on next page)

FIG 7 (FIG 8
FIG 7 (Continued)cells with fragmented Golgi and mean (±SD) Golgi area.Data presented are representative of four independent experiments in which 50 cells were examined per condition each time.(C) Cdc42 downregulation inhibits C1P-induced JNK phosphorylation, but not PKCα.HUVECs were transfected with siNT, siCPTP, siCdc42 or siCdc42/CPTP.At 72 h, p-PKCα, p-JNK, total PKCα, and total JNK levels were analyzed by Western blotting and densitometry to determine the mean (±SD) normalized ratios of p-PKCα:PKCα and p-JNK:JNK.(D and E).PKCα is upstream of JNK in C1P-mediated signaling.HUVECs that had been treated with siNT or siCPTP were treated with Gö6976 (D) or SP006125 (E).At 24 h post-treatment, the mean (±SD) normalized ratios of P-PKCα:PKCα and p-JNK:JNK were determined.Data are representative of four independent experiments.Statistical analyses were performed using one-way ANOVA with Tukey's post hoc test.*P < 0.05; **P < 0.01; ***P < 0.001; and ****P < 0.0001.

(FIG 9
FIG 9 Model.A. phagocytophilum elevates host cell C1P levels, which induces signaling through PKCα, Cdc42, and JNK that leads to GRASP55 and GRASP65 phosphorylation that, in turn, promotes uncoupling of the Golgi ribbon.The resulting increase in anterograde trafficking of TGN-derived vesicles into the ApV, a pathogen-modified MVB, benefits the A. phagocytophilum infection cycle.