To determine if GFP-Rab10 localizes to the ApV in a host cell type besides myeloid cells (Huang et al., 2010a), we assessed for the phenomenon in infected RF/6A endothelial cells using laser scanning confocal microscopy (LSCM). RF/6A cells were also selected for this purpose because, in addition to being permissive for A. phagocytophilum infection, they are large and flat, which makes them ideal for imaging the ApV (Munderloh et al., 2004; Sukumaran et al., 2011; Beyer et al., 2014). Similar to that previously observed for infected HL-60 cells (Huang et al., 2010a), GFP-Rab10-positive vesicles surrounded ApVs (Fig. 1A). Additionally, GFP-Rab10 was observed within ApVs encircling A. phagocytophilum organisms. GFP alone remained diffuse in the cytosol of infected cells. Given these observations for GFP-Rab10, we hypothesized that endogenous Rab10-positive vesicles are delivered into the ApV. Accordingly, we screened infected RF/6A cells with antibodies against endogenous Rab10 and the bacterial outer membrane protein (OMP), Asp14 (14 kDa A. phagocytophilum surface protein) (Kahlon et al., 2013) or APH0032, which is a bacterial effector that is pronouncedly expressed late (24 to 32 h) during the infection cycle and localizes to the A. phagocytophilum occupied vacuolar membrane (AVM) (Huang et al., 2010b). In addition to producing a vesicular labelling pattern in the cytosol and at the peripheries of ApVs, endogenous Rab10 signal surrounded individual intravacuolar bacteria in ring-like labelling patterns and partially colocalized with Asp14 signal (Fig. 1B and C). Z-stack analysis and three-dimensional (3D) rendering of a representative ApV confirmed the presence of Rab10-positive vesicles within the vacuole associated with intravacuolar bacteria (Fig. 1D and Movie S1). Thus, Rab10-positive vesicles are recruited to the ApV and delivered into its lumen where they associate with A. phagocytophilum organisms. Notably, a minor portion of APH0032-positive ApVs lacked lumenal Rab10 signal, as exemplified by the ApV in the lower-right corners of the images presented in Fig. 1C and D. This observation suggests that delivery of Rab10-positive vesicles into the ApV may be a temporal event that occurs during the period of APH0032 expression.

Given that Rab10 directs exocytic traffic from the TGN, we investigated if the ApV associates with the Golgi apparatus and if the Rab10-positive vesicles delivered into its lumen are TGN derived. Screening A. phagocytophilum infected RF/6A cells with antibodies against markers for the cis-Golgi (GM130 and GolgB1), medial-Golgi (GolgB1) and trans-Golgi (TGN46) revealed ApVs closely apposed to and in many instances wrapped by the Golgi (Fig. 2). Signal for TGN46 but not GM130 or GolgB1 was detected within ApV lumens, indicating specific delivery of TGN-derived vesicles or fragments. TGN46 labelling inside ApVs partially colocalized with Asp14 signal (Fig. 3A). Z-stack imaging and 3D rendering of a representative ApV further verified that the presence of TGN46-positive vesicles closely associated with intravacuolar A. phagocytophilum organisms (Fig. 3B and C and Movie S2) and was highly similar to the observed proximal association of Rab10-positive vesicles near the surfaces of the bacteria (Fig. 1B–D and Movie S1). As a complementary approach, organelles from uninfected and A. phagocytophilum-infected HL-60 cells were fractionated using a density gradient centrifugation method that keeps the ApV intact (Niu et al., 2012). Immunoblot analysis revealed altered distribution of TGN46 in the fractions of infected versus uninfected cells (Fig. 3D). Specifically, TGN46 co-migrated to the fraction in which the A. phagocytophilum OMP, P44 (Truchan et al., 2013), was most abundant.

Given the abundance of TGN vesicles/fragments delivered into the ApV, we examined Golgi integrity during A. phagocytophilum infection. ApVs were demarcated using an antibody against the cytoskeletal protein, vimentin, which localizes to the AVM (Sukumaran et al., 2011). In uninfected host cells and host cells with low A. phagocytophilum loads (one to three ApVs per cell), the Golgi remained intact, and intravacuolar bacteria were TGN46 positive (Fig. S1A and B). However, bacterial load-dependent fragmentation of the Golgi was observed for cells having medium (4–10 ApVs per cell) and high bacterial loads (≥ 11 ApVs per cell) (Fig. S1B). Additionally, even though the cis-Golgi, medial-Golgi and trans-Golgi were each fragmented and dispersed throughout host cells with high bacterial loads, only a TGN46 signal was detected within ApVs (Fig. S1B and C). This finding reinforces that A. phagocytophilum selectively targets TGN-derived exocytic traffic. Furthermore, in cells with high A. phagocytophilum loads, not all of the ApV lumens were visually TGN46 positive (Fig. S1B), which is likely due to the bacterial demand for TGN vesicles exceeding the supply. Collectively, these data support the findings that the ApV interacts with the Golgi apparatus and intercepts TGN-derived traffic and that TGN-derived vesicles or fragments are delivered into the ApV lumen.

The partial colocalization of Rab10 and TGN46 signals with Asp14-associated and 4′,6-diamidino-2-phenylindole (DAPI)-associated signals of A. phagocytophilum observed using LSCM suggested that Rab10-positive trans-Golgi-derived vesicles are present within the ApV lumen in close proximity to intravacuolar A. phagocytophilum bacteria. LSCM has a lateral resolution of approximately 200 nm (Allen et al., 2014). Within cells, and presumably within the ApV, macro-molecular associations occur in spatial distances less than 200 nm. Therefore, structured illumination microscopy, a form of super-resolution microscopy, which has a lateral resolution of approximately 100 nm (Allen et al., 2014), was employed to more accurately gauge whether Rab10 and TGN46 signals were associated with intravacuolar A. phagocytophilum organisms. Supporting the phenomena observed via LSCM, SIM imaging revealed the Rab10 signal to exhibit a vesicle-like pattern that was diffuse in the host cell cytosol and a more clustered pattern that partially localized with Asp14-positive and DAPI-positive A. phagocytophilum organisms (Fig. 4A). Likewise, the TGN46 signal pronouncedly colocalized with bacterial-associated Asp14 and DAPI signals (Fig. 4B). These data indicate that Rab10 and TGN46 are within 100 nm of the surfaces of A. phagocytophilum organisms.

Anaplasma phagocytophilum lacks genes that are necessary for sphingolipid synthesis (Dunning Hotopp et al., 2006). Given that the TGN is enriched in sphingolipids (van Meer, 1998) and TGN vesicles accumulate in the ApV, we examined for the delivery of sphingolipids into the ApV lumen using BODIPY Texas Red (TR) ceramide. This fluorescently labelled lipid accumulates in the Golgi, where it is modified into fluorescently labelled sphingolipids that subsequently traffic to the plasma membrane via TGN-derived secretory vesicles (Pagano et al., 1991). Infected host cells were incubated with BODIPY TR ceramide and examined using LSCM. BODIPY TR ceramide-positive vesicles were observed surrounding ApVs and within ApV lumen associating with bacteria in ring-like labelling patterns (Fig. 5A). Because the infection was asynchronous and all A. phagocytophilum bacteria were TR positive, we inferred that BODIPY TR-derived sphingolipids associated with both RC and DC forms. Furthermore, ultra performance liquid chromatography–electrospray ionization–tandem mass spectrometry (UPLC-ESI-MS/MS) analyses detected the presence of multiple ceramide, sphingomyelin and monohexyl sphingolipid subspecies in A. phagocytophilum DC organisms that had been naturally released from infected HL-60 cells, but not in media from uninfected control HL-60 cells (Fig. 5B–E). These results substantiate that host cell sphingolipids that are typically found in the TGN are delivered into the ApV lumen and are incorporated by A. phagocytophilum.

To better understand the relevance of Rab10-positive TGN vesicle delivery to A. phagocytophilum intracellular development, we examined if the vesicles specifically associate with the DC or RC form. Infected RF/6A cells were screened with Rab10 or TGN46 antibody together with an antibody against the A. phagocytophilum DC-specific protein, APH1235 (Troese et al., 2011; Mastronunzio et al., 2012). An RC-specific marker has yet to be identified. As previously reported (Troese et al., 2011; Mastronunzio et al., 2012), APH1235-positive A. phagocytophilum organisms were detected bound to host cell surfaces and as intracellular clusters, presumably within vacuoles (Fig. 6A). Strikingly, Rab10 and TGN46 were nearly undetectable on APH1235-positive bacteria. This observation indirectly suggests that RC but not DC organisms predominantly associate with TGN vesicles.

To further investigate this phenomenon, we assessed Rab10 and TGN46 localization within infected cells by LSCM over a 36 h synchronized infection time course. We used LSCM to examine APH0032 and APH1235 expression kinetics to verify that the infection cycle proceeded as expected. APH0032 expression/AVM localization was examined for the additional reason that one of our initial experiments suggested that Rab10 vesicle import into the ApV is a temporal event that occurs coincident with APH0032 expression (Fig. 1C and D). Consistent with previous observations (Huang et al., 2010b), the percentages of APH0032-positive AVMs pronouncedly rose from 8 to 28 h (Fig. 6B and C). Also as previously reported (Troese et al., 2011; Mastronunzio et al., 2012), the percentages of A. phagocytophilum organisms expressing APH1235 dropped from 100% at 4 h to nearly 0% between 12 and 24 h and then pronouncedly increased again to more than 80% by 36 h (Fig. 6C). The localization of TGN46 and Rab10 around and within ApVs occurred synchronously. Rab10-positive vesicles and TGN vesicles were detected surrounding ApVs beginning at 16 h (Fig. 6B and C). Both markers were observed within ApVs beginning at 24 h and peaked at 28 h, a time point at which nearly 100% of ApVs were Rab10 and TGN46 positive. After 28 h, intravacuolar Rab10 and TGN46 signals waned considerably. Notably, the time points at which Rab10 and TGN46 signals associated with the ApV and A. phagocytophilum organisms preceded those when the bacteria were APH1235 positive. The coincident and temporal associations of Rab10 and TGN46 signals with A. phagocytophilum organisms indirectly suggested that Rab10 and TGN46 antibodies did not non-specifically label the bacteria. To verify that this was the case, host cell-free RC and DC organisms were screened via LSCM. Both RCs and DCs were stained by DAPI and labelled by P44 antibody (Fig. S2). Rab10 and TGN46 antibodies labelled only the RC morphotype. Collectively, these data demonstrate that delivery of Rab10-positive TGN vesicles into the ApV and their association with intravacuolar bacteria precedes A. phagocytophilum conversion from the replicative RC form to the infectious DC form.

To define the significance of Rab10 to A. phagocytophilum TGN parasitism, we assayed for TGN vesicle recruitment to and delivery into the ApV in infected host cells in which Rab10 had been knocked down. For this experiment, we used HEK-293T cells because they are susceptible to A. phagocytophilum infection and have a high transfection efficiency (≥ 75%) relative to RF/6A and HL-60 cells (<20%) (Beyer et al., 2014). Furthermore, GFPR-ab10 and TGN46 localize to and within the ApV, respectively, in these cells (Fig. S3A and B). HEK-293T cells were treated with small interfering RNA (siRNA) targeting Rab10, glyceraldehyde 3-phosphate dehydrogenase (GAPDH) or non-targeting siRNA. Knockdown of target protein expression was confirmed by Western blot (Fig. 7A). Rab10 knockdown had no effect on Golgi apparatus integrity, as TGN46 labelling revealed the organelle to be perinuclear, condensed and structurally intact (Fig. S3C). Treated cells were infected with A. phagocytophilum followed by LSCM examination. The percentage of ApVs with lumenal TGN46 signal was reduced by approximately fourfold in Rab10 siRNA-targeting versus non-targeting siRNA-treated HEK-293T cells (Fig. 7C). Also, whereas the Golgi apparatus was fragmented and barely detectable in infected control cells, it remained intact in infected cells in which Rab10 had been knocked down (Fig. 7B). Furthermore, the percentage of infected cells and the number of ApVs per cell were reduced by approximately twofold and threefold, respectively, in Rab10-knockdown cells compared with control cells (Fig. 7D and E). Taken together, these data suggest that Rab10 is critical for TGN recruitment to the ApV, for TGN vesicle uptake by the ApV and for productive A. phagocytophilum infection.

Because the A. phagocytophilum load was pronouncedly reduced in host cells in which Rab10 had been knocked down (Fig. 7D and E), yet many of these cells were observed to harbour numerous late-stage ApVs (Fig. 7B), we reasoned that acquisition of Rab10-positive TGN vesicles was not important for bacterial replication but was instead critical for the production of infectious progeny. To test this hypothesis, media from A. phagocytophilum-infected HEK-293T cells that had been treated with Rab10, GAPDH or non-targeting siRNA was collected at 48 h post-infection and added to naïve cells. In untreated cells, this media contains naturally liberated, infectious A. phagocytophilum DC organisms (Beyer et al., 2014). If siRNA treatment inhibited RC-to-DC conversion, then fewer DC bacteria would be released into the media and, consequently, would produce a lower level of infection in recipient cells. At 24 h, the percentage of infected host cells and the number of ApVs per cell was more than twofold and nearly fourfold lower, respectively, in cells that had been incubated with media from infected Rab10-knockdown cells relative to those that had received media from infected non-targeting siRNA-treated control cells (Fig. 8A and B). To confirm that Rab10 knockdown inhibited RC-to-DC transition, the experiment was repeated and quantitative reverse transcription (qRT)-PCR was performed on RNA collected at 24, 28 and 32 h post-inoculation to assess for aph1235 expression. The selected time points are those during which A. phagocytophilum TGN acquisition, aph1235 up-regulation (Troese et al., 2011;Mastronunzio et al., 2012) and RC-to-DC transition (Troese and Carlyon, 2009) would normally occur. As expected (Troese et al., 2011; Mastronunzio et al., 2012), aph1235 expression rose over the time course in host cells that had been treated with non-targeting siRNA (Fig. 8C). In contrast, aph1235 transcript levels did not increase at any time point in host cells in which Rab10 had been knocked down. Thus, Rab10-dependent acquisition of TGN vesicles is critical for aph1235 up-regulation, RC-to-DC conversion and the production of A. phagocytophilum infectious progeny.

Because Rab10-positive vesicles are recruited to the ApV in a guanine nucleotide-independent, tetracycline-sensitive manner (Huang et al., 2010a) and are delivered into the lumen where they associate with A. phagocytophilum organisms, we reasoned that at least one pathogen-encoded protein contributes to these processes. To identify potential A. phagocytophilum Rab10-interacting proteins, we incubated whole-cell lysates of infected and uninfected HL-60 cells with glutathione sepharose beads coated with glutathione-S-transferase (GST)-tagged Rab10 or GST alone. A 26 kDa protein that was specifically isolated from infected cells by the GST-Rab10 beads (Fig. 9A) was identified by mass spectrometry as APH0111 (Fig. 9B). This protein is annotated as UMPK (Dunning Hotopp et al., 2006), a nucleotide scavenger enzyme that catalyses the conversion of uridine monophosphate to uridine diphosphate (Yan and Tsai, 1999). To validate that UMPK and Rab10 interact and to evaluate the specificity of the interaction, whole-cell lysates of Escherichia coli expressing Halo-tagged UMPK were incubated with beads coated with GST alone, GST-Rab10, GST-Rab11 or GST-Rab35. Rab11 and Rab35 were included because they also localize to the AVM (Huang et al., 2010a). Only GST-Rab10-coupled beads precipitated recombinant UMPK (Fig. 10A).

To assess if UMPK colocalizes with Rab10 in living cells and in a guanine nucleotide-independent manner, HeLa cells were cotransfected to express Halo-UMPK and GFP-tagged wild-type Rab10, GTP-locked Rab10Q68L, GDP-locked Rab10T23N or guanine nucleotide binding defective Rab10N122I and examined by LSCM. HeLa cells coexpressing Halo-UMPK and GFP alone or one of several additional GFP-tagged Rabs that are known to be recruited to the ApV (Rab1A, Rab4A, Rab11A, Rab14, Rab22A and Rab35) or not (Rab5 and Rab7) (Huang et al., 2010a) served as controls. Consistent with the guanine nucleotide-independent specificity of UMPK for Rab10, coexpression of Halo-UMPK with each GFP-tagged Rab10 isoform yielded aggregative vesicular labelling patterns where Halo and GFP signals completely overlapped (Fig. 10B). Of the other Rab GTPases, Halo-UMPK poorly localized with Rab35 and did not colocalize with any other Rab (Fig. 10B and C). Halo and GFP signals did not colocalize in cells that expressed either Halo alone and Halo-UMPK and GFP alone (Fig. S4A) or GFP-Rab10 (Fig. S4B). Halo-tagged E. coli UMPK and GFP-Rab10 also did not colocalize in transfected cells (Fig. S4C), further indicating the specificity of the A. phagocytophilum UMPK-Rab10 interaction. To confirm that UMPK is capable of binding Rab10 in living cells, Halo-UMPK was coexpressed with and assessed for the ability to precipitate GFP-tagged Rab10 isoforms. All GFP-tagged Rab10 isoforms, but neither GFP nor GFP-Rab7, coprecipitated with Halo-UMPK (Fig. 10D). Taken together, these data identify A. phagocytophilum UMPK as a Rab10-specific ligand that binds the GTPase in a guanine nucleotide-independent manner.

Uridine monophosphate kinase is a 244-amino-acid protein with a predicted molecular mass of 26 kDa and two transmembrane domains that putatively present amino acids 1 to 100 and 155 to 244 on the bacterial surface (Figs 9B and 11A and B). Our previous evaluation of the A. phagocytophilum surface proteome identified UMPK as a likely surface protein (Kahlon et al., 2013). Mining the UMPK tryptic peptides recovered in that study (Kahlon et al., 2013) revealed that they mapped to the protein’s two predicted surface-exposed regions (Fig. 9B). To verify that UMPK is a surface protein, A. phagocytophilum-infected RF/6A cells and human neutrophils were screened with antibody specific for a peptide corresponding to predicted surface-exposed UMPK residues, 189–195 (Figs 9B, 11A and B and S5A), together with APH0032 antibody. In both infected host cell types, UMPK_182–195 antibody labelled intravacuolar bacteria, but not the APH0032-positive AVM (Fig. 11C). Next, to determine if UMPK residues 182–195 are an immunoaccessible domain that is exposed on the bacterial outer membrane, the surfaces of intact host cell-free A. phagocytophilum organisms were treated with proteinase K, a proven approach for verifying surface presentation of A. phagocytophilum and Chlamydia trachomatis OMPs (Wang et al., 2006; Ojogun et al., 2012; Kahlon et al., 2013; Seidman et al., 2014). Western-blotted lysates of proteinase K-treated and vehicle-treated DC bacteria were screened with UMPK_182–195 antiserum. Positive control antiserum targeted a confirmed surface-exposed epitope of Asp55 (55 kDa A. phagocytophilum surface protein) (Ge and Rikihisa, 2007), while negative control antiserum targeted APH0032, which is not an OMP (Huang et al., 2010b). Following surface proteolysis, neither UMPK_182–195 nor Asp55 was detected (Fig. 11D). APH0032 was amply detected in lysates of both treated and vehicle control-treated bacteria. Immunofluorescence examination of A. phagocytophilum-infected host cells at multiple time points post-infection revealed that the bacterium constitutively expressed UMPK (Fig. S5B). Thus, UMPK is an A. phagocytophilum surface-localized Rab10-specific ligand that is expressed throughout the course of infection.

Uninfected and A. phagocytophilum (NCH-1 strain)-infected human promyelocytic HL-60 cells [CCL-240; American Type Culture Collections (ATCC), Manassas, VA] and RF/6A rhesus monkey choroidal endothelial cells (CRL-1780; ATCC) were cultured as described previously (Huang et al., 2010b). HeLa 229 epithelial cells (CCL-1.2; ATCC) were grown in Roswell Park Memorial Institute 1640 (Invitrogen, Carlsbad, CA) supplemented with 10% foetal bovine serum (FBS; Gemini Bio-Products, Sacramento, CA) at 37°C with 5% CO₂. HEK-293T cells were cultured in Dulbecco’s modified Eagle’s medium with l-glutamine, 4.5 g l⁻¹ of d-glucose and 100 mg l⁻¹ of sodium pyruvate (Invitrogen) supplemented with 10% FBS, 1× minimal essential medium non-essential amino acids (Invitrogen) and 15 mM of HEPES (Affymetrix, Cleveland, OH) at 37°C with 5% CO₂. Human neutrophils were isolated from peripheral blood from healthy donors using Polymorphprep (Axis-Shield; Greiner Bio-One, Frickenhausen, Germany) as described previously (Carlyon et al., 2004). The protocol (HM11407) for obtaining donor blood for the purpose of isolating neutrophils has been reviewed and approved by the Virginia Commonwealth University Institutional Review Board with respect to scientific content and compliance with applicable research and human subjects regulations.

Previously described constructs encoding GFP-tagged and GST-tagged Rab proteins were kind gifts from Marci Scidmore (Cornell University) and have been described previously (Rzomp et al., 2003; Cortes et al., 2007; Huang et al., 2010a). APH0111 (UMPK) DNA was synthetically generated and cloned into the SgfI and PmeI sites of the pFN21K vector mammalian expression vector by Promega Corporation (Madison, WI). The insert was subcloned into the SgfI and PmeI sites of the E. coli expression vector pFN18A (Promega) using the Flexi system transfer kit (Promega). The pFN21K control vector that was used for expressing the Halo tag alone was constructed by digesting the original pFN21K vector with SgfI and PmeI to remove the barnase gene. The digested vector was religated with a double-stranded DNA bridge that was generated using oligonucleotides 5′-CGCGTAAGGGTAGGTTT-3′ (sense strand) and 5′-AAACGTACCCTTACGCGAT-3′ (antisense strand). DNA encoding the E. coli UMPK gene was amplified from Alpha-Select Gold Efficiency E. coli (Bioline, Taunton, MA) using primers 5′-ATATGCGATCGCCGCTACCAATGCAAAACCCGTCTATAAACGCATTC-3′ (sense strand; spacer nucleotides are italicized; SgfI site is underlined) and 5′-AACTGTTTAAACTTATTCCGTGATTAAGTCCCTTCTTTTTCACC-3′ (antisense strand; spacer nucleotides are italicized; PmeI site is underlined) and cloned into the SgfI and PmeI sites of the pFN21K. DNA encoding the A. phagocytophilum UMPK gene was amplified from NCH-1 DNA using primers 5′-GACCGCGATCGCCGATTGTAATGTTGTGGAGAATGTGCGTTATTCTA-3′ (sense strand; spacer nucleotides are italicized; SgfI site is underlined) and 5′-GTCGGTTTAAACCTATCCAGATATAGTAGTAAACGTACCCC-3′ (antisense strand; spacer nucleotides are italicized; PmeI site is underlined) and ligated into pFN21K or pFN18A. All inserts were amplified with Platinum Pfx DNA polymerase (Invitrogen, Grand Island, NY), and the sequences verified by DNA sequencing.

BL21 E. coli (Novagen, EMD Millipore, Billerica, MA) transformed with a plasmid encoding GST-Rab10 were grown to an optical density at 600 nm of 0.4 and induced with 100 µM of isopropyl β-d-1-thiogalactopyranoside at 16°C for 16 h. GST-Rab10 was solubilized from inclusion bodies using 1% sarkosyl in Brymora bacterial lysis buffer [10 mM Tris–HCl (pH 7.4), 250 mM NaCl, 1% (vol/vol) Triton x-100, 2.5 mM MgCl₂] and combined with the soluble fraction. GST-Rab10 was immobilized on glutathione sepharose beads (GE Healthcare Life Sciences, Little Chalfont, UK) with the addition of 2% Triton X-100 and 20 mM 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate. Fifty million uninfected or A. phagocytophilum-infected HL-60 cells were lysed in radioimmunoprecipitation assay buffer [10 mM Tris-Cl (pH 8.0), 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate and 140 mM NaCl]. The lysates were pre-cleared with glutathione beads to eliminate non-specific binding and then mixed with 5 mg of GST or GST-Rab10 glutathione beads for 2 h at 4°C. The beads were washed and boiled, and the eluates were resolved on a 20 cm acrylamide tris-glycine 10–20% gradient gel (Jule Inc., Milford, CT) using the Protean II XL Cell electrophoresis rig (BioRad, Hercules, CA). The gel was stained with SYPRO Ruby (Invitrogen). The approximately 26 kDa band of interest that was exclusively recovered from the A. phagocytophilum-infected host cell lysate was excised and submitted to the W. M. Keck Facility at Yale University (New Haven, CT) for mass spectroscopy and database searching.

Cells for immunofluorescence assays were grown and infected on #1½ 12 × 12 mm glass coverslips for LSCM (Electron Microscopy Sciences, Hatfield, PA) or #1½ HP (i.e. 0.17 ± 0.005 mm thick) high-performance glass coverslips (Zeiss, Thornwood, NY) for SIM. They were fixed in 4% paraformaldehyde (Electron Microscopy Sciences) for 30 min followed by permeabilization with 0.5% Triton X-100 for 10 min and washed in 1× phosphate-buffered saline (PBS). In some cases, host cell-free A. phagocytophilum RC and/or DC organisms were recovered from infected HL-60 cells as described (Carlyon, 2005) and fixed to glass slides. The cells or host cell-free bacteria were screened with antibodies for immunofluorescence microscopy as previously described (Huang et al., 2010c). Commercial primary antibodies were against TGN46 (Sigma-Aldrich, St. Louis, MO), GM130 (BD Biosciences, San Jose, CA), GolgB1 (Sigma-Aldrich), HaloTag (Promega), GFP (Invitrogen), Rab10 (Sigma-Aldrich) and vimentin (Abcam, Cambridge, MA). Antibodies against Asp14, APH1235, APH0032 and P44 were described previously (Huang et al., 2010b; Troese et al., 2011; Kahlon et al., 2013). Affinity-purified rabbit polyclonal antiserum targeting a peptide corresponding to A. phagocytophilum UMPK amino acids 182 to 195 was generated by New England Peptide (Gardner, MA). Secondary antibodies conjugated to Alexa Fluor fluorochromes were obtained from Invitrogen. BODIPY ceramide/bovine serum albumin complex (Invitrogen) was added to infected RF/6A cells per manufacturer’s instructions. Coverslips were mounted with Prolong Gold Anti-fade reagent with or without DAPI (Invitrogen). Images were obtained using an Olympus BX51 microscope affixed with a disc-spinning unit (Olympus, Center Valley, PA) or a Zeiss LSM 700 laser scanning confocal microscope. For SIM, coverslips were stained with 1 µg ml⁻¹ of DAPI in 1× PBS for 5 min and mounted with Prolong Gold Anti-fade reagent without DAPI (Invitrogen). Images were obtained using a Nikon N-SIM superresolution microscope. LSCM and SIM were performed at the VCU Microscopy Facility, which is supported, in part, by funding from NIH-NINDS Center core grant (5P30NS047463) and NIH-NCI Cancer Center Support Grant (P30 CA016059). 3D movies were generated using Volocity Image Analysis Software (Perkin Elmer, Waltham, MA).

HeLa cells were seeded onto 12 × 12 mm coverslips (Electron Microscopy Sciences) and cotransfected with plasmid DNA encoding Halo alone, Halo-A. phagocytophilum UMPK, Halo-E. coli UMPK or GFP-UMPK, and various GFP-Rabs or GFP alone was processed and labelled using Lipofectamine 2000 (Invitrogen). HaloTag TMR Direct ligand (Promega) was added to wells containing cells transfected to express Halo-tagged protein. After 24 h, the cells were processed and labelled, where applicable, as described earlier, and examined by immunofluorescence microscopy for colocalization of Halo and GFP signals.

Neutrophils and HL-60 cells were infected with A. phagocytophilum DC organisms released from infected HL-60 cells by sonication as previously described (Troese and Carlyon, 2009). RF/6A or HEK-293T cells were infected with A. phagocytophilum DC bacteria that had been naturally released from infected RF/6A cells into the culture media as follows. Adherent host cells to be infected were seeded onto 12 mm coverslips (Electron Microscopy Sciences) and overlain with 200 µl of A. phagocytophilum-laden media from heavily infected RF/6A cells. The supernatant was spun down on top of the cells at 1000 × g for 3 min followed by a 1 h incubation at 37°C with 5% CO₂. The cells were washed with 1× PBS to remove unbound bacteria, and fresh media was added. The cells were returned to 37°C with 5% CO₂ for 24 or 48 h or specific time points, after which the cells were fixed in 4% paraformaldehyde and examined by immunofluorescence microscopy.

Uninfected or infected HL-60 cells of 2 × 10⁷ were washed with ice-cold 1× PBS two times followed by one wash in cold homogenization buffer (HB; 250 mM sucrose, 10 mM Tris–HCl, pH 7.4,1 mM EDTA). The cells were suspended in 1 ml of cold HB with protease inhibitors (Roche) and transferred to a type B dounce homogenizer (Gerresheimer Kimble Chase LLC, Vineland, NJ). The cells were homogenized for 20 to 30 strokes until ≥ 90% of the host cells were lysed, as verified by the trypan blue exclusion assay. The homogenate was centrifuged at 500 × g for 5 min to pellet the nuclei and unbroken cells. The supernatant was overlain on top of a 5%, 15% or 25% continuous Opti-prep (Sigma-Aldrich) gradient. The gradient was centrifuged at 200,000 × g for 3 h in a Beckman Coulter (Indianapolis, IN) Optima XE-100 ultracentrifuge. Twelve 1 ml fractions were collected from the top of the gradient. Protein was concentrated from the fractions by trichloroacetic acid precipitation. Equal volumes of fractions 1 to 9 were immunoblotted using TGN46, GM130 and P44 antibodies.

Cell lysates or gradient centrifugation fractions were analysed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot as previously described (Troese et al., 2011). Commercially available primary antibodies targeted β-actin (Santa Cruz Biotechnology, Santa Cruz, CA), GAPDH (Sigma-Aldrich), GFP (Invitrogen), HaloTag (Promega), TGN46 (Sigma-Aldrich) and Rab10 (Sigma-Aldrich). Rabbit antisera against A. phagocytophilum proteins P44 (IJdo et al., 1999) and APH0032 (Huang et al., 2010b) were described previously. Rabbit antibody targeting a surface-exposed epitope of Asp55 (Ge and Rikihisa, 2007) was a kind gift from Yasuko Rikihisa (Ohio State University, Columbus, OH).

Media from 3 × 10⁷ A. phagocytophilum-infected (≥ 90%) HL-60 cells, which contained DC bacteria that had been naturally released, was spun at 300 × g twice to pellet host cells. The supernatant was spun at 1000 × g to remove host cellular debris. Next, the supernatant was spun at 5200 × g for 10 min to pellet DC organisms. The resulting pellet was washed twice with ice-cold PBS. All spins were performed at 4°C. As a background control, the media from 3 × 10⁷ uninfected HL-60 cells was processed and analysed by UPLC-ESI-MS/MS in parallel. The A. phagocytophilum DC and control samples were examined for the presence of sphingolipids using UPLC-ESI-MS/MS as previously described (Simanshu et al., 2013; Wijesinghe et al., 2014). Three separate biological replicates were examined in triplicate.

Seeded onto 12 mm glass coverslips (Electron Microscopy Sciences) were 4 × 10⁵ HEK-293T cells. After 16 to 20 h, 80 µl of 5 µM ON-TARGETplus human Rab10 siRNA SMARTpool (GE Dharmacon, Lafayette, CO) was mixed with 320 µl of media and added to the wells. Non-targeting or GAPDH-targeting siRNA (GE Dharmacon) was added to control wells. After 72 h, 200 µl of media containing A. phagocytophilum DC organisms that had been released from infected RF/6A cells was added, and the bacteria were spun on the cells as described earlier. Additionally, cells from one well of each siRNA treatment were harvested for Western blot analysis to confirm knockdown. At 48 h post-infection, cells were harvested for quantitative PCR, processed for microscopy analyses, or the DC-laden media was added to naïve HEK-293T cells to detect the presence of infectious progeny.

To analyse A. phagocytophilum load after siRNA treatment, DNA was isolated from host cells using the DNeasy Blood and Tissue Kit (Qiagen, Boston, MA). Fifty nanograms of DNA was analysed with primers specific for A. phagocytophilum 16S rDNA and host cell β-actin using SsoFast EvaGreen Supermix (BioRad) as previously described (Ojogun et al., 2012). To detect APH1235 transcript at 24, 28 and 32 h post-infection, cDNA was generated and analysed by qRT-PCR as previously described (Kahlon et al., 2013).

HeLa cells were seeded into a T-75 flask per pulldown and cotransfected with 12 µg each of Halo-UMPK (pFN21K) and GFP-Rab constructs using Lipofectamine 2000 (Invitrogen) per the manufacturer’s instructions. After 24 h, the cells were trypsinized from the flask, spun down at 300 × g and subjected to one freeze–thaw cycle before complete lysis with Mammalian lysis buffer containing protease inhibitors (Promega). The pulldown was performed according to Promega’s HaloTag Mammalian Pull-Down and Labeling Systems Technical Manual (TM#342) with the addition of 2.5 mM MgCl₂ in the binding and wash buffers. The eluates were resolved by SDS-PAGE, transferred to nitrocellulose and probed with GFP antibody (Invitrogen) as described (Troese et al., 2011).

BL21(DE3) E. coli (Novagen, EMD Millipore) transformed with the pFN18A Halo-UMPK construct were grown to an optical density at 600 nm of 0.4 and induced with 100 µM isopropyl β-d-1-thiogalactopyranoside at 25°C for 4 h. A culture of 50 ml culture was pelleted and lysed with Bugbuster (Novagen, EMD Millipore) containing EDTA-free Complete Protease Inhibitor (Roche, Indianapolis, IN). The lysate was divided equally, and each aliquot was incubated with 5 mg each of GST, GST-Rab10, GST-Rab11 or GST-Rab35 glutathione beads overnight on a rotator at 4°C. The beads were washed and boiled, and the eluates were resolved by SDS-PAGE and immunoblotted with HaloTag antibody (Promega).

Putative transmembrane domains in A. phagocytophilum UMPK were identified using TMPred (http://www.ch.embnet.org/software/TMPRED_form.html). Surface-exposed proteins of A. phagocytophilum DC organisms were removed by proteolysis as previously described (Ojogun et al., 2012), except that the bacteria were treated with 4 mg ml⁻¹ of proteinase K for 45 min at 25°C. We previously reported recovering UMPK as an A. phagocytophilum surface protein using a surface biotinylation-affinity chromatography approach (Kahlon et al., 2013). To provide the previously unpublished sequences of the peptides recovered in this study, we mined the peptide spectra from this analysis and denoted them in Fig. S5.

One-way analysis of variance or Student’s t-test was performed using the Prism 5.0 software package (GraphPad; San Diego, CA) to assess statistical significance as described. Statistical significance was set at P ≤ 0.05.