Anaplasma phagocytophilum AnkA binds to granulocyte DNA and nuclear proteins

Blackwell Science, LtdOxford, UKCMICellular Microbiology 1462-5814Blackwell Publishing Ltd, 200468743751Original ArticleJ. Park et al.AnkA binds to host DNA and nuclear proteins

Cellular Microbiology (2004) 6(8), 743–751

doi:10.1111/j.1462-5822.2004.00400.x

Anaplasma phagocytophilum AnkA binds to granulocyte DNA and nuclear proteins

Jinho Park,1† Kee Jun Kim,2 Kyoung-seong Choi,1 Dennis J. Grab2 and J. Stephen Dumler1* 1 Division of Medical Microbiology, Department of Pathology, The Johns Hopkins University School of Medicine, Ross Research Building, 720 Rutland Avenue, Baltimore, Maryland, USA. 2 Division of Infectious Diseases, Department of Pediatrics, The Johns Hopkins University School of Medicine, Baltimore, Maryland, USA. Summary Human granulocytic anaplasmosis (HGA) is caused by the obligate intracellular bacterium Anaplasma phagocytophilum. The bacterium infects, survives, propagates in, and alters neutrophil phenotype, indicating unique survival mechanisms. AnkA is the only known A. phagocytophilum component that gains access beyond neutrophil vacuoles and is transported to the infected host cell nucleus. The ability of native and recombinant AnkA to bind DNA and nuclear proteins from host HL-60 cells was assessed by the use of immunoprecipitation after cisdiamminedichloroplatinum (cis-DDP) DNA-protein crosslinking, by probing uninfected HL-60 cell nuclear lysates for AnkA binding, and by recovery and sequence analysis of immunoprecipitated DNA. AnkA binds HL-60 cell DNA as well as nuclear proteins of approximately 86, 53 and 25 kDa, whereas recombinant A. phagocytophilum Msp2 or control proteins do not. DNA immunoprecipitation reveals AnkA binding to a variety of target genes in the human genome, including genes that encode proteins with ATPase, tyrosine phosphatase and NADH dehydrogenase-like functions. These data indicate that AnkA could exert some effect on cells through binding to protein:DNA complexes in neutrophil nuclei. Whether AnkA binding leads to neutrophil functional alterations, and how such alterations might occur will depend upon definReceived 22 December, 2003; revised 27 February, 2004; accepted 2 March, 2004. *For correspondence. E-mail [email protected]; Tel. (+1) 410 955 8654; Fax (+1) 443 287 3665. †Present address: Department of Veterinary Internal Medicine, College of Veterinary Medicine, Chonbuk National University, Jeonju, Jeonbuk 561–756, Korea. © 2004 Blackwell Publishing Ltd

itive identification of binding partners and associated metabolic and biochemical pathways. Introduction Anaplasma phagocytophilum is a tick-borne, obligate intracellular bacterium that infects neutrophils of mammals, including man (Dumler et al., 2001). The unique niche of this bacterium implies that it has developed specialized systems for survival and propagation. It is now well documented that A. phagocytophilum alters neutrophil function by inducing an ‘activated-deactivated’ phenotype, wherein infected neutrophils are incapable of generating phagocyte oxidase, are defective at adhering to and transmigrating across activated endothelium, and are defective at phagocytosis and microbicidal activity. Yet, infection activates cells for production of proinflammatory responses with chemokine secretion, degranulation and release of metalloproteases (Woldehiwet, 1987; Banerjee et al., 2000; Klein et al., 2000; Choi et al., 2003; Park et al., 2003a). The mechanisms that underlie control of these various cellular processes by the bacterium are increasingly studied and appear to involve regulation at the level of host cell mRNA, via signal transduction, and perhaps by effects upon transcription factors (Banerjee et al., 2000; Carlyon et al., 2002; Kim and Rikihisa, 2002; Lin and Rikihisa, 2003). AnkA (also known as Ank) is a unique protein of A. phagocytophilum that is high molecular weight (160 kDa), not associated with the bacterial membrane, but is found localized within nuclei of infected granulocyte hosts, presumably after secretion and transport through at least three distinct membranes (Caturegli et al., 2000). Although its close ultrastructural association with condensed chromatin and the presence of multiple ankyrin repeat domains tempts speculation about a role in affecting gene transcription or transcription factors, the actual binding targets in the nucleus are not known. We hypothesized that AnkA would bind to DNA or DNA–protein targets of granulocyte nuclei, and tested this by analysing these binding properties in vitro. The results of these analyses provide strong evidence of a DNA binding role for AnkA, with a potential contribution by protein, and lead to further speculation about the functional nature of AnkA–host cell DNA interactions.

744 J. Park et al.

Fig. 1. Detection sensitivity of A. phagocytophilum protein immunoblots using monoclonal antibodies IE3 for AnkA (left) and 20B4 for Msp2 (right). Lanes 1 through 6 contain decreasing concentrations of A. phagocytophilum protein (1, 1000 ng; 2, 200 ng; 3, 40 ng; 4, 8 ng; 5, 1.6 ng; 6, 0.32 ng), and lane 7 contains 3 mg uninfected HL-60 cell protein. Note the 25-fold greater sensitivity for detection of Msp2 versus AnkA in whole A. phagocytophilum lysates. AnkA usually presents with 2–3 high molecular weight bands (Caturegli et al., 2000), and the various bands detected with Msp2 mAb represent monomers and oligomers (Park et al., 2003c).

Results and discussion Anaplasma phagocytophilum AnkA but not Msp2 localizes into HL-60 cell nuclei Previous ultrastructural localization indicates that AnkA is found in the nucleus of infected HL-60 cells in close proximity to condensed chromatin (Caturegli et al., 2000). Additional evidence of this localization and evidence that excludes A. phagocytophilum whole cell contamination of nuclear preparations in these in vitro experiments was first attempted. Thus, a quantitative protein immunoblotting method to identify A. phagocytophilum membraneassociated Msp2 or non-membrane AnkA was used to assess whether detection of the latter was caused by contamination of nuclei by whole bacteria. With evaluation of whole cell density gradient-purified A. phagocytophilum, Msp2 was detected using mAb 20B4 in as little as 8 ng of total protein. However, in the same preparation of whole A. phagocytophilum, the detectable quantity of AnkA using mAb IE3 after loading ≥200 ng of A. phagocytophilum total protein was 25-fold lower (Fig. 1). These experiments were repeated at least three times with identical results. When nuclei from A. phagocytophilum-infected HL-60 cells were prepared and examined using this method, only AnkA was detected (Fig. 2) confirming its translocation into HL-60 cell nuclei and the absence of contaminating of A. phagocytophilum in nuclear preparations. AnkA binds to DNA, including some HL-60 cell genomic DNA fragments To determine if AnkA binds to DNA or demonstrates specificity for DNA from certain sources, recombinant AnkA

binding was assessed in electrophoretic mobility shift assays (EMSA). From approximately 100 pUC18 HL-60 DNA clones screened for rAnkA, rMsp2, and BSA binding, only a few bound rAnkA, and none bound control proteins. The rAnkA-calmodulin binding protein (–cbp) fusion most significantly altered the mobility of both the pCR4-Aprrs (cloned A. phagocytophilum 16S rRNA gene) and the EcoRI-digested insert component of one clone from the pUC18-HL-60 cell genomic DNA library, although a lesser degree of binding to some pUC18 alone was also demonstrated (Fig. 3). Recombinant AnkA did not alter mobility of control pCAL-ankA or pCAL-kan r indicating that the binding of rAnkA-cbp resulted from specific interactions within the rrs or HL-60 cell insert DNA. Control proteins (BSA and rMsp2-cbp) did not bind to any of the DNA constructs used, underscoring the relative specificity of AnkA for these forms of DNA. The specificity of HL-60 cell DNA for rAnkA-cbp fusion protein was further shown with concentration-dependent electrophoretic mobility shifts (Fig. 4). The lack of any staining intensity change when incubated with other DNA substrates confirms the absence of DNase activity in rAnkA-cbp. Unfortunately, the clones were unstable, lost thereafter, and unavailable for subsequent DNA sequence analysis. Immunoprecipitated AnkA-HL-60 DNA Identification of the DNA that binds to rAnkA-cbp was attempted by immunoprecipitation; however, inconsistent recovery (data not shown) prompted evaluation of DNA

Fig. 2. Protein immunoblot detection of AnkA but not Msp2 in nuclear lysates from A. phagocytophilum-infected HL-60 cells. The presence of AnkA confirms its nuclear localization and the lack of Msp2 confirms the absence of contaminating A. phagocytophilum in nuclear preparations. Left panel, protein immunoblot using monoclonal antibody IE3 to detect AnkA; right panel, protein immunoblot using monoclonal antibody 20B4 to detect Msp2. Lane 1, control density gradient-purified cell-free A. phagocytophilum; lane 2, control uninfected HL-60 cell nuclei; lane 3, A. phagocytophilum-infected HL-60 cell nuclei. Note the presence of AnkA and absence of Msp2 in nuclear preparations from infected cells. © 2004 Blackwell Publishing Ltd, Cellular Microbiology, 6, 743–751

AnkA binds to host DNA and nuclear proteins 745 Fig. 3. Electrophoretic mobility shift assays (EMSA) demonstrate that recombinant AnkAcbp fusion protein, but not recombinant Msp2cbp fusion protein or bovine serum albumin (BSA) binds to EcoRI digested-HL-60 cell genomic DNA cloned into pUC18 (pUC18-HL60) and to A. phagocytophilum rrs (16S rRNA gene) cloned in pCR4 TOPO plasmid vector (pCR4-Ap rrs). No binding to pUC18, ankA in the pCAL vector (pCAL-ankA), or a kanamycin resistance gene in pCAL vector (pCAL-kan r ) is demonstrated. The DNA and plasmid preparations labelled at the top were mixed with the following proteins: lane 1, no protein; lane 2, BSA; lane 3, rAnkA-cbp; lane 4, rMsp2-cbp. Note the absence of bands in several rAnkAcbp lanes owing to the inability of the large complexes to enter the gel.

Fig. 4. Binding to rAnkA-cbp fusion protein and electrophoretic mobility shift of EcoRI-digested pUC18-HL-60 DNA is dependent upon rAnkA protein concentration. All reactions contain 40 ng pUC18/HL-60 DNA, without protein (DNA only), without protein in binding buffer (binding buffer), or with rAnkA-cbp protein quantities added as indicated.

specifically crosslinked to native AnkA by cisdiamminedichloroplatinum (cisplatin or cis-DDP) (Ferraro et al., 1996; Jenke et al., 2002; VanderWaal et al., 2002). When native AnkA was immunoprecipitated from A. phagocytophilum-infected HL-60 cells, DNA was only coimmunoprecipitated from cis-DDP-treated preparations (Fig. 5). No DNA was co-immunoprecipitated (i) without cis-DDP crosslinking (ii) when using Msp2 mAb 20B4, or (iii) with isotype-matched control mAb, underscoring the specificity of the AnkA-HL-60 cell DNA interactions. To further demonstrate this association, DNA sheared by sonication after cis-DDP crosslinking resulted in a broad ‘smear’ of lower molecular weight complexes that was difficult to resolve. Likewise, when the mobility of AnkA detected in protein SDS-PAGE immunoblots was examined, the typical 160 kDa protein was not observed after immunoprecipitation of cis-DDP-crosslinked AnkA-HL-60 cell DNA complexes, owing to the large complexes that poorly penetrate the gel (Fig. 6). This concept is affirmed as increasing concentrations and time intervals of DNase treatment allowed increasing AnkA detection at its native molecular size, as demonstrated in controls without cisDDP DNA-protein crosslinking. These data confirm a specific ability of native A. phagocytophilum AnkA to associate with DNA from the infected cells in vitro. © 2004 Blackwell Publishing Ltd, Cellular Microbiology, 6, 743–751

rAnkA binds to HL-60 cell nuclear proteins In addition to DNA binding properties, the association with chromatin shown in ultrastructural studies (Caturegli et al.,

Fig. 5. HL-60 DNA is co-immunoprecipitated with AnkA from A. phagocytophilum-infected HL-60 cell DNA crosslinked by cis-DDP. Ethidium bromide was used to stain the agarose gels to detect DNA bound to AnkA. Sonication was performed before immunoprecipitation to assess the effect of fragmentation on AnkA-DNA electrophoretic mobility. Treatments are indicated as above the figure. Lanes 1 through 5 were not sonicated; lanes 6 through 9 were sonicated. Note the single high molecular weight DNA band immunoprecipitated with AnkA mAb (lane 3) not present with other mAbs or uninfected cells. Also note the broad DNA smear in lane 8 that results from immunoprecipitation of AnkA bound to sheared HL-60 cell DNA.

746 J. Park et al. with rAnkA-cbp or control proteins. In this overlay blot analysis, only rAnkA-cbp reacted with whole HL-60 cell proteins or HL-60 cell nuclear proteins, regardless of whether detection was attempted using AnkA or Msp2 mAbs, AnkA monospecific polyclonal antibody, or polyclonal A. phagocytophilum antibody (Fig. 7). Although a 53 kDa protein was better detected in whole HL-60 cell preparations, it and proteins of 86 and 25 kDa also bound rAnkA in nuclei, suggesting that AnkA binds predominantly to nuclear proteins.

Fig. 6. HL-60 cell DNA is co-immunoprecipitated with AnkA from A. phagocytophilum-infected HL-60 cell DNA crosslinked by cis-DDP. Protein immunoblots were used to detect AnkA bound to DNA. Lane 1, uninfected HL-60 cell protein only; lane 2, AnkA mAb immunoprecipitation after cis-DDP-treatment of A. phagocytophilum-infected cells – note the band > 185 kDa (arrowhead); lane 3, AnkA mAb immunoprecipitation after 15 min with 1 U DNase and cis-DDP-treatment of A. phagocytophilum-infected cells – note the moderate shift of bands ≥ 185 kDa to 160 kDa or less; lane 4, AnkA mAb immunoprecipitation after 30 min with 3 U DNase and cis-DDP-treatment of A. phagocytophilum-infected cells – note the marked shift in bands to 160 kDa or less, similar to no cis-DDP control (lane 5); lane 5, AnkA mAb immunoprecipitation of non-cis-DDP-treated A. phagocytophilum-infected cells (no cis-DDP control); lane 6, protein G/AnkA mAb only; lane 7, purified A. phagocytophilum protein only.

2000) also suggests a potential interaction with chromatin or nuclear proteins, or with DNA–protein complexes. To examine the possibility that AnkA can also bind proteins, protein blots of whole HL-60 cells and nuclei were probed

Analysis of DNA immunoprecipitated from cis-DDP-treated, A. phagocytophilum-infected HL-60 cells To determine whether AnkA of infected HL-60 cell nuclei binds to specific genomic sequences, DNA recovered from cis-DDP-treated infected cells by immunoprecipitation with AnkA mAb was cloned for sequencing. To assure reproducibility, DNA recovery, cloning, and sequence analysis was performed on three occasions. Because Msp2 mAbs and isotype-matched control mAbs did not immunoprecipitate any DNA from chromatin, water was used as a negative control, and no recombinant clones were recovered. Of 17 plasmid clones analysed, insert sizes ranged from 73 to 382 bp, each with no obvious similarities or convincing consensus sequences after CLUSTALX alignment. To determine if specific oligonucleotides not obvious with CLUSTALX alignment might be present among the 17 cloned sequences, BLASTN analysis was conducted and revealed nine groups of similar oligoFig. 7. rAnkA, but not rMsp2, binds to HL-60 whole cell and nuclear lysates. HL-60 cell components (left) or nuclear proteins (right) were separated by SDS-PAGE and transferred to nitrocellulose. Individual nitrocellulose strips were reacted with rAnkA, rMsp2, bovine serum albumin (BSA), or no protein (np). Bound A. phagocytophilum proteins were detected by reaction with A. phagocytophilum mAbs (Ank mAb IE3 – lanes 1 and 5, left panel, and lanes 1 and 2, right panel, or Msp2 mAb 20B4 – lane 3 left panel, lanes 3 and 4 right panel), with rabbit polyclonal rAnkA antiserum (lanes 2 and 6 left panel), rabbit polyclonal A. phagocytophilum antiserum (lane 4 left panel). Note the presence of rAnkA binding strongly to 53 kDa and weakly to 86 kDa HL-60 whole cells, and strongly to 86- and 25 kDa and weakly to 53 kDa nuclear components. Binding with a 185 kDa protein is observed in whole cells only (left panel, lanes 1 and 2), and with a 25 kDa protein in nuclear lysates (right panel, lane 1) among others.

© 2004 Blackwell Publishing Ltd, Cellular Microbiology, 6, 743–751

© 2004 Blackwell Publishing Ltd, Cellular Microbiology, 6, 743–751

(98%) (89%) 100% (100%) (99%) (100%)

e-100 2e-67 4e-93 4e-87 6e-59 2e-93

similar to NADH dehydrogenase subunit 2 similar to beta-glucuronidase precursor (Beta-G1) LOC285505: hypothetical protein Near hypothetical protein FLJ11331 In Genome Scan model Hs16–10655-33-15-6 LOC349926: similar to protease 1 [Pneumocystis carinii f. sp. carinii ] Near RAS p21 protein activator 3

1p36.33 5q21.1 4q32.1 4q26 16q12.2 13q34

gene with protein product, function unknown gene with protein product, function unknown mRNA only Predicted and EST GTPase-activator protein for Ras-like GTPases

NT_034471 NT_034772 NT_016606 NT_016354 NT_010498.13 NT_024498

NT_024524 NM_000701 e-118 3e-55 277/303 (91%) 131/142 (92%)

195/197 176/196 175/175 169/169 122/123 180/180 T11-JHP T14-JHP T15-JHP T21-JHP

JH-N3

T7-JHP

e-164 e-151 e-135 e-131 (100%) (97%) (94%) (93%) 302/302 295/302 286/302 284/303

T10-JHP

inositol signalling system

Cation transport ATPase–inorganic ion transport and metabolism Cytochrome c oxidase subunit III 13q14.13 1p13

AF007118 AJ421032 NT_011519 NT_011903 Signal transduction–phosphatidyl

21p11 13q12.11 22q11.21 Yq11.223

NT_007933 cDNA only 7q32 e-150 279/279 (100%) JH-T3

Near UniGene Cluster Hs.353002 Near Genome Scan model Hs7–8090-33-512-11 TPTE – Transmembrane phosphatase with tensin homology TPIP – TPTE and PTEN homologous inositol lipid phosphatase LOC284853: similar to putative protein-tyrosine phosphatase TPTE LOC286573: TPTEP transmembrane phosphatase with tensin homology pseudogene LOC220115: hypothetical protein tyrosine phosphatase ATP1A1: ATPase, Na+/K + transporting, alpha 1 polypeptide

Locus description Chromosome Locus E-value Identity (% homology) Locus

Table 1. BLASTN (GenBank) and MEGABLAST (Human genome) analysis of immunoprecipitated DNA complexed to A. phagocytophilum AnkA after cis-DDP treatment.

nucleotides identified in 2–3 clones for each group. However, as the range of sizes identified varied from 11 to 13 nucleotides, the E-values were high (0.005–0.24). No single sequence longer than several bases was common among the clones. Although cis-DDP crosslinks proteins that are within 4 Å of DNA at the time of the reaction (VanderWaal et al., 2002), it may not specifically link AnkA to its definitive target DNA binding sequences and instead might allow binding in proximity to the target. Thus, each individual sequence was used as a BLASTN query of both GenBank and the human genome. Of the 17 sequences examined, significant similarities were identified in the human genome for eight clones (Table 1). Several clones identified regions within known or putative genes on multiple chromosomes, including clones that identified genes with known or predicted ATPase, tyrosine phosphatase, and NADH dehydrogenase-like functions. One clone identified five distinct loci on chromosomes 13, 21, 22 and the Y chromosome with between 91 and 100% homology for TPTE, a transmembrane phosphatase with tensin homology involved in signal transduction. Another clone identified putative genes on chromosomes 1 and 5 similar to NADH dehydrogenase subunit 2. It is unclear why some cloned sequences were not identified in the human genome database, but this absence could reflect the altered genomic content of the leukemic HL-60 cells. The significance of AnkA binding to or near these genes or associated nuclear proteins is not known. Duplicate or multiple clones that represented the same chromosomal target were not identified, suggesting that AnkA may bind to multiple regions on several chromosomes. There are several potential explanations for the lack of recovery of uniform chromosome loci with this chromatin immunoprecipitation method. First, AnkA could bind nonspecifically to chromatin or DNA, but this seems unlikely given the lack of binding by other control proteins. Second, AnkA may bind to specific DNA sequences, but chromatin immunoprecipitation after cis-DDP treatment may not provide sufficient resolution to detect the nature of the binding sites. Third, AnkA could bind to DNA motifs widely distributed throughout the chromosome or to chromatin via protein interactions with other DNA-binding proteins. It would be difficult to predict the pattern of DNA recovery for these circumstances as many DNA binding sites, histone proteins, and transcription or regulatory factor proteins could also yield a non-uniform pattern. Of interest, the predicted structure for AnkA is similar to that of several transcription factor regulatory proteins such as IkBa and pINK4 family proteins (data not shown). Regardless, it appears that other novel approaches will be needed to definitively identify the specific structures to which AnkA binds before its function in the neutrophil nucleus is discerned.

Locus accession no.

AnkA binds to host DNA and nuclear proteins 747

748 J. Park et al. Obligate intracellular bacterial infections often involve macrophages and epithelial cells, although other cells such as fibroblasts are sometimes infected as well (reviewed in Hornef et al., 2002; Rosenberger and Finlay, 2003). Among professional phagocytes, productive infection of neutrophils by bacteria has only been demonstrated for species within the family Anaplasmataceae (Dumler et al., 2001; Carlyon and Fikrig, 2003; Rikihisa, 2003). Most notable in this group is A. phagocytophilum, the causative agent of granulocytic anaplasmosis (formerly granulocytic ehrlichiosis). How this bacterium manages to circumvent destruction involves more than simple inhibition of phagosome-lysosome fusion, as an increasing number of functional changes in neutrophils is reported (Woldehiwet, 1987; Banerjee et al., 2000; Klein et al., 2000; Carlyon et al., 2002; Carlyon and Fikrig, 2003; Choi et al., 2003; Park et al., 2003a; Rikihisa, 2003). However, given the manipulation of neutrophil function with A. phagocytophilum infection, interactions with transcription factors, signal transduction pathways, other metabolic and biochemical pathways, perhaps even direct regulation of gene transcription can be anticipated. Bacterial manipulation of host cells is an increasingly recognized phenomenon Introduction of bacterial components into host cells can be achieved through type III and type IV secretion mechanisms. Under these circumstances, bacterial proteins can directly interact with host signal transduction pathways or actin polymerization in the cytosol, and by introduction of DNA into the host nucleus for a single species, Agrobacterium tumefaciens (Nagai and Roy, 2003; Waterman and Holden, 2003). However, introduction of bacterial proteins into the host cell nucleus is a rarely recognized phenomenon, and could present bacteria with novel ways to influence host cells for bacterial survival, such as with endosymbiont X-bacteria regulation of S-adenosylmethionine synthetase expression in Amoeba proteus (Pak and Jeon, 1997; Jeon and Jeon, 2003). Whether AnkA acts in a similar manner is not known. Anaplasma phagocytophilum exerts effects on neutrophils at many levels, including manipulation of phagocyte oxidase by downregulation of mRNA for Rac2 and gp91phox, potential proteolytic degradation of p22phox, induction of chemokine release presumably by induction of gene transcription, induction of degranulation and release of granule contents including proteolytic enzymes and metalloproteases, induction of surface integrin and Igfamily adhesion molecules, and inhibition of apoptosis (Woldehiwet, 1987; Banerjee et al., 2000; Klein et al., 2000; Yoshiie et al., 2000; Carlyon et al., 2002; Choi et al., 2003). Moreover, complex neutrophil functions are altered by infection, including loss of selectin-mediated endothe-

lial cell adhesion, reduced transendothelial cell migration, prolonged activation and degranulation, and reduced phagocytic and microbicidal activity (Woldehiwet, 1987; Choi et al., 2003; Park et al., 2003). Whereas a single common theme that can explain all of these diverse effects is unlikely to be found, the identification of DNAand protein-binding AnkA in the nucleus of infected cells suggests a role in influencing the neutrophil to permit survival and replication within. Whether the physical presence of AnkA bound directly to DNA, proteins, or both is involved with any changes in neutrophil or granulocyte function is still not known. Limited data now suggest that A. phagocytophilum infection directly impacts the transcription of critical genes that could alter neutrophil function and biological response to favor intracellular infection and survival (Carlyon et al., 2003; Rikihisa, 2003). The simple interactions as observed here by EMSA would not seem to provide sufficient specificity to guide the alterations observed with A. phagocytophilum infection. However, such interactions and specificity could occur by direct interactions with transcription factors or interference with their DNA binding properties (often the result of heterodimeric or multimeric interactions), alterations in signal transduction, changes in nuclear-cytoplasmic flux of DNAbinding proteins, or alterations in binding of proteins and DNA by nuclear matrix proteins, among many hypothetical mechanisms. Although the specific means for these alterations is not known, further investigation may help to identify which specific molecules are targeted by A. phagocytophilum and AnkA, and how these interactions lead to bacterial manipulation of the neutrophil host.

Experimental procedures Anaplasma phagocytophilum Anaplasma phagocytophilum Webster strain was propagated in HL-60 cells (human promyelocytic leukaemia cell line) in RPMI 1640 medium supplemented with 1% fetal bovine serum and 2 mM L-glutamine (Goodman et al., 1996; Asanovich et al., 1997). Cell-free A. phagocytophilum was purified by renografin density gradient centrifugation from heavily infected HL-60 cells, as previously described (Chen et al., 1994; Asanovich et al., 1997). The presence of intact bacteria was confirmed by Romanowsky staining (HEMA 3, Biochemical Science, Swedesboro, NJ) and indirect fluorescent-antibody assay (IFA). The protein concentration of purified A. phagocytophilum was determined by the BCA method (Pierce chemical, Rockford, IL).

Nuclear protein extracts Nuclei and nuclear proteins were prepared from A. phagocytophilum-infected and -uninfected HL-60 cells using the Nuclear Extraction Kit (Panomics, Redwood, CA) as per manufacturer’s instructions. Briefly, 5 ¥ 106 cells were incubated with hypotonic buffer A mix (100 mM HEPES pH 7.9, 100 mM KCl, 100 mM © 2004 Blackwell Publishing Ltd, Cellular Microbiology, 6, 743–751

AnkA binds to host DNA and nuclear proteins 749 EDTA, 1 mM DTT, 0.4% IGEPAL, and protease inhibitor cocktail) for 10 min on ice and cells were then disrupted by vortexing. The nuclei were separated from the lysed cellular components by centrifugation (15 000 g) for 3 min at 4∞C and then resuspended in buffer B (100 mM Hepes, pH 7.9, 2 M NaCl, 5 mM EDTA, 50% glycerol, 10 mM DTT, protease inhibitor cocktail) and lysed by vigorous vortexing. After rocking on ice for 2 h, the supernatant containing nuclear proteins was separated from particulates by centrifugation (15 000 g for 5 min at 4∞C). Protein content was determined by the BCA method as described above.

SDS-PAGE and immunoblot analysis To determine the sensitivity of AnkA protein detection compared to Msp2 protein, cell-free A. phagocytophilum was serially diluted, separated by SDS-PAGE, and immunoblotted using AnkA monoclonal antibody (mAb) IE3 or Msp2 mAb 20B4 (Park et al., 2003b). To demonstrate the exclusive presence of AnkA in nuclei and to exclude the possibility of contamination with whole A. phagocytophilum, the protein concentration of the nuclear lysates was measured and identical quantities of the same nuclear prepration were loaded for electrophoresis and immunoblotted for the detection of AnkA and Msp2 protein as above. Immunoblots were also used to demonstrate changes in electrophoretic mobility of AnkA after cis-DDP cross-linking or DNase treatments (Ferraro et al., 1996; Jenke et al., 2002; VanderWaal et al., 2002). For immunoblots, alkaline phosphatase-labelled goat anti-rabbit or mouse IgG (Kirkegaard and Perry Laboratories, Gaithersburg, MD) was used as a secondary antibody. BCIP-NBT (Sigma Chemical, St Louis MO) was used as substrate for alkaline phosphatase and colour development. Isotypematched control monoclonal antibodies or control rabbit serum were purchased from Sigma (Sigma Chemical, St Louis, MO). All immunoblot experiments were repeated at least three times.

Recombinant AnkA and Msp2 Recombinant AnkA (rAnkA) and recombinant Msp2 (rMsp2) were produced as fusion proteins with calmodulin-binding protein as previously described (Caturegli et al., 2000; Caspersen et al., 2002; Park et al., 2003b); both retained reactivity in protein immunoblots with AnkA mAb IE3, rabbit polyclonal anti-rAnkA, Msp2 mAb 20B4, and rabbit polyclonal anti-A. phagocytophilum respectively (Caturegli et al., 2000). In brief, both recombinant fusion proteins were produced in E. coli and purified by calmodulin affinity chromatography.

Plasmid DNA construction The A. phagocytophilum rrs plasmid clone pCR4-Aprrs was made by cloning the 919 bp A. phagocytophilum 16S rRNA gene PCR product into the pCR4-TOPO vector, as suggested by the manufacturer (Invitrogen, Carlsbad, CA). A small insert (approximately 1000 bp) plasmid library (pUC18-HL-60) was prepared using EcoRI-digested genomic DNA from uninfected HL-60 cells cloned into the pUC18 vector. A total of approximately 40 random clones from this library were assessed for binding to rAnkA and control proteins. A single clone that demonstrated rAnkA binding was utilized for experiments. Similarly, pCAL-ankA (A. phagocytophilum ankA gene in the pCAL vector) and pCAL-kanr (kana© 2004 Blackwell Publishing Ltd, Cellular Microbiology, 6, 743–751

mycin resistance gene in pCAL vector) were constructed and prepared as previously described (Caturegli et al., 2000). All plasmid clones were expanded and their DNAs were purified (Wizard Plus Minipreps DNA purification System, Promega, Madison, WI) and assessed for the correct size insert after digestion with EcoRI.

Electrophoretic mobility shift assays (EMSA) rAnkA, rMsp2, bovine serum albumin (BSA), or no protein were incubated with pCR4-Aprrs, pUC18, pUC18-HL-60, pCAL-ankA, or pCAL-kanr plasmid DNA in TPEN (10 mM Tris, 2 mM K3PO4, 1 mM EDTA, 50 mM NaCl) protein-DNA binding buffer for 40 min at 37∞C. After incubation, the DNA–protein complexes were resolved on 0.6% agarose gels and EtBr-stained. Shifts in mobility of DNA were analysed using a UV transilluminator and video documentation system (Gel Doc 2000, Bio-Rad, USA).

Analysis of AnkA and HL-60 cell DNA interactions by cis-DDP cross-linking Anaplasma phagocytophilum–infected HL-60 cells were washed with Hank’s buffer and treated with the specific DNA-protein cross-linker, cis-DDP (Sigma Chemical, St Louis, MO) or with buffer only, for 2 h at 37∞C in Hank’s buffer as previously described (Ferraro et al., 1996; Jenke et al., 2002). HL-60 cells were then harvested by centrifugation and lysed (5 M urea, 2 M guanidine hydrochloride, 2 M sodium chloride, 200 mM potassium phosphate, pH 7.5) for 30 min on ice. Identical aliquots were subsequently subjected to DNA shearing by brief sonication (for DNA analysis) or treated with 3 U, 1 U, or no DNase (for immunoblot analysis). AnkA-DNA complexes were immunoprecipitated with AnkA mAb IE3 or with isotype-matched control mAb precleared with Sepharose-protein G beads. All immunoprecipitated beads were washed three times with PBS. The HL-60 genomic DNA in immunoprecipitated sonicated or non-sonicated AnkA-DNA complexes was isolated as modified from Jenke et al. (2002) using the QIAgen genomic DNA preparation kit (Qiagen, Valencia, CA). The DNA was resolved in 1% agarose gels. In parallel experiments, the immunoprecipitated DNase-treated or untreated AnkA-DNA complexes were separated by SDS-PAGE on 10% gels and transferred to nitrocellulose for protein immunoblotting with AnkA mAb IE3 as described above.

Cloning and sequence analysis of cis-DDP-treated AnkA-immunoprecipitated HL-60 genomic DNA fragments To determine if AnkA binds to specific DNA sequences, DNA recovered after AnkA immunoprecipitation from cis-DDP-treated A. phagocytophilum-infected HL-60 cell nuclei was cloned and sequenced (Ferraro et al., 1996; VanderWaal et al., 2002). To do so, DNA fragments of