Intracellular Parasitism by the Human Granulocytic Ehrlichiosis Bacterium Through the P-Selectin Ligand, PSGL-1

See allHide authors and affiliations

Science  02 Jun 2000:
Vol. 288, Issue 5471, pp. 1653-1656
DOI: 10.1126/science.288.5471.1653


Human granulocytic ehrlichiosis (HGE) is a febrile tick-borne illness caused by a recently discovered intracellular bacterium remarkable for its tropism for professionally phagocytic neutrophils. Monoclonal antibodies against the P-selectin binding domain of the leukocyte P-selectin glycoprotein ligand, PSGL-1, prevented HGE cell binding and infection, as did enzymatic digestion of PSGL-1. Furthermore, simultaneous neoexpression in nonsusceptible cells of complementary DNAs for both PSGL-1 and its modifying α-(1,3) fucosyltransferase, Fuc-TVII, allowed binding and infection by HGE. Thus, the HGE bacterium specifically bound to fucosylated leukocyte PSGL-1. Selectin mimicry is likely central to the organism's unique ability to target and infect neutrophils.

Human granulocytic ehrlichiosis (HGE) is an emerging tick-borne infection that causes an acute febrile illness, which is often severe, and frequently coinfects individuals with Lyme disease (Borrelia burgdorferi), also carried byIxodes ticks (1, 2). In the few years since its recognition, the disease has been reported in several hundred individuals, primarily in the regions of the north central and northeastern United States, where Lyme disease is endemic, and seropositivity has been reported in several other countries. The etiologic agent is an obligate intracellular bacterium found in the peripheral blood granulocytes of infected patients and is closely related, or identical, to Ehrlichia equi and E. phagocytophila (1, 3), which infect horses and ruminants, respectively. Several pathogens of humans have life cycles that involve permissive infection of monocytes, but the HGE agent is noteworthy in its tropism for and growth within professionally phagocytic granulocytes. The as of yet unnamed etiologic organism of HGE was recently isolated with the promyelocytic leukemia cell line, HL60, and it has since been shown that the organism can replicate in other myeloid cell lines and in normal human bone marrow progenitor cells (3–5). The tropism of this organism for leukocytes and its ability to evade normal phagocytic pathways and lysosomal fusion (6) suggest that the organism uses specific receptor-mediated pathway(s) for cellular adhesion and entry.

Although intracellular infection by bacteria is an area of intense investigation, the entry of pathogens into permissive pathways in normally hostile phagocytic cells remains poorly understood (7). Some organisms infect a variety of cells and bind to molecules that are widely distributed in the host (e.g.,Yersinia species and Toxoplasma gondii to β-1 integrins) (8, 9). Others are able to use specific leukocyte molecules in their adhesion to monocytes (e.g.,Legionella pneumophila, Leishmania donovanii, andMycobacterium tuberculosis through binding to complement receptors) (10, 11). The receptors responsible for ehrlichial binding and entry into host cells were unknown.

Leukocyte cell surface expression of the sialyl Lewis x tetrasaccharide (sLex or CD15s) and the expression of the α-(1,3) fucosyltransferase, Fuc-TVII, responsible for its terminal fucosylation are linked to cell susceptibility to HGE infection (12). However, Fuc-TVII has multiple substrates, sLex itself modifies a wide variety of proteins and lipids, and antibodies to sLex (anti-sLex) did not block HGE adhesion to susceptible cells (12), suggesting that sLexalone is not a specific receptor for the organism. We hypothesized that the HGE receptor is a Fuc-TVII–modified cell surface glycoprotein expressed on neutrophils. The linkage of infection to the expression of sLex also suggested that the HGE receptor might be a selectin ligand.

We examined the effects of monoclonal antibodies (mAbs) against sLex itself and known sLex-modified or related mucin-modified proteins, including CD43, CD16, CD24, and the major identified selectin ligand of neutrophils (13,14), PSGL-1, on HGE binding (15). As shown by fluorescence-activated cell sorting (FACS) analysis (16) of binding of fluoresceinated bacteria, mAb PL1, which is directed against the NH2-terminus of PSGL-1 (residues 8 to 21) and blocks P-selectin binding to cells (17), blocked the binding of HGE to HL60 cells in a dose-dependent manner (Fig. 1A). These results were confirmed with three different, geographically diverse strains of HGE. mAb KPL-1, which binds to the tyrosine sulfate motif of PSGL-1 and blocks binding to P-selectin (18), also blocked binding of bacteria to HL60 cells, whereas mAb PL2, which binds to a membrane proximal epitope on PSGL-1 and does not block binding to P-selectin (17), did not. mAb PL1 also blocked HGE binding to human peripheral blood neutrophils (Fig. 1B). In contrast, numerous other mAbs, including three directed against the β-1 integrin subunit CD11b (e.g., Leu 15; Fig. 1B), as well as isotype control immunoglobulins, had little or no effect.

Figure 1

mAbs against the P-selectin binding domain of P-selectin glycoprotein ligand-1 (PSGL-1) block binding of the HGE bacterium to susceptible cells. (A) Dose-response curve of inhibition of binding of fluoresceinated HGE bacteria (expressed as the percentage of control binding observed in the presence of isotype control IgG1) to HL60 cells by anti–PSGL-1 mAb PL1. Marked inhibition was observed at concentrations as low as 2 μg/ml. (B) mAbs PL1 and KPL-1 (10 μg/ml), which both bind to the PSGL-1 P-selectin binding site, block HGE binding to HL60 cells and human blood polymorphonuclear neutrophils (PMN), whereas mAb PL2, against a nonbinding PSGL-1 domain (see text), and the anti-CD11b mAb Leu 15 do not.

To confirm that these results represent inhibition of specific binding, we performed identical studies by indirect immunofluorescent microscopy (IFA) allowing the direct visualization and enumeration of adherent bacteria (12). mAbs KPL-1 and PL1 prevented HGE binding to either HL60 cells or neutrophils. For example, HL60 cells in the presence of control immunoglobulin G, subclass 1 (IgG1), bound a median of 5.0 bacteria per cell (Fig. 2A, mean 5.0) compared with 1 bacteria per cell bound (mean 1.1) in the presence of KPL-1 (10 μg/ml) (Fig. 2C) (P < 0.001, Wilcoxan sum of ranks test).

Figure 2

Inhibition of HGE binding and infection by anti–PSGL-1. HL60 cells were preincubated for 1 hour at 37°C with isotype control IgG1 (A and B) or with the anti–PSGL-1 mAb KPL-1 (C and D) and then challenged with HGE (13). (A) and (C) show cells removed and washed immediately after challenge to assess bacterial binding. Bacteria (labeled yellow-green by indirect immunofluorescence) avidly bind to the HL60 cell surface (A), but this binding is prevented by KPL-1 (C). After 48 hours of cultivation, samples from the same experiments were evaluated for the presence of intracellular bacterial colonies indicative of bacterial replication. The majority of cells preincubated with control antibody are highly infected (B), but there is no infection of the cells pretreated with KPL-1 (D) (scale bar, 10 μm).

If binding to PSGL-1 by HGE is required for bacterial entry, then the prevention of adhesion by blocking mAbs would be expected to reduce intracellular infection. Preincubation of HL60 cells with KPL-1 (10 μg/ml), but not control IgG1, prevented infection, with a mean of 0.1% of KPL-1–treated cells infected when observed 48 hours later (Fig. 2D) compared with 87% of control cells (Fig. 2B; P < 0.001, Fisher Exact Test). Similarly, incubation of granulocytes with KPL-1 (10 μg/ml) prevented infection by cell-free HGE bacteria (1% of cells compared with 18.3% infected in IgG1 controls at 48 hours;P < 0.001). HGE is probably transmitted directly from cell to cell, and such transfer in vitro is more efficient than the use of cell-free bacteria. Hence, we investigated whether KPL-1 could prevent cell-to-cell transfer from highly infected HL60 cells. Again, treatment of neutrophils with anti–PSGL-1 mAb reduced subsequent infection, even by large numbers of highly infected HL60 cells (38.5% of control IgG1-treated neutrophils infected compared with 4% of those treated with KPL-1; P < 0.001).

To extend our findings with mAbs, we used neoexpression of PSGL-1 on a normally nonexpressing leukocyte cell line to discover whether or not HGE would then bind to and enter such cells. Given our previous findings that cellular susceptibility to HGE infection correlates with sLex expression, we also investigated whether Fuc-TVII, which is needed for sLex synthesis in leukocytes (19), is required to glycosylate PSGL-1 for effective HGE binding to occur.

The B lymphoblastoid cell line BJAB (20) expresses neither PSGL-1 nor Fuc-TVII/sLex and does not significantly bind to or become infected by the HGE agent. Using cDNA for PSGL-1 or Fuc T-VII, we generated stable transfectants of BJAB that expressed either PSGL-1 or Fuc-TVII/sLex, or both (21). These transfectants were then each challenged with bacteria, and their ability to bind organisms and become infected was determined. Untransfected BJAB cells rarely bound organisms (mean 0.4 and median 0 bacteria per cell) and did not become infected (Fig. 3, A and B). Transfectants expressing PSGL-1 alone also did not bind bacteria or become significantly infected. Transfectants expressing Fuc-TVII alone bound slightly more bacteria (e.g., mean of 1.1 organisms per cell, median of 1; P= NS) than parental BJAB cells but did not become susceptible to infection. However, dual transfectants expressing both PSGL-1 and Fuc-TVII bound HGE (mean 3.4 and median 2 organisms per cell; P< 0.001 compared with parental BJAB cells; Fig. 3C) and became infected (mean 79% at 48 hours, compared with 0.5% of parental BJAB;P < 0.001; Fig. 3D). Furthermore, HGE binding to and subsequent infection of the dual transfectants could be prevented by preincubating cells with the anti–PSGL-1 mAb KPL-1. Identical results were obtained by FACS analysis of binding of fluorescent bacteria to parental and transfected cells. Thus, coexpression of PSGL-1 and Fuc-TVII was necessary and sufficient for HGE binding to and infection of BJAB cells.

Figure 3

Neoexpression of PSGL-1 and sLexconfers on BJAB cells the ability to bind the HGE bacterium and become infected. The B lymphoblastoid cell line, BJAB, does not normally express either PSGL-1 or the α-(1,3) fucosyltransferase, Fuc-TVII, which fucosylates sLex, a tetrasaccharide that modifies PSGL-1. Parental BJAB cells neither bind HGE bacteria (A) nor become infected (B). However, BJAB cells stably transfected to express both PSGL-1 and Fuc-TVII bind HGE (C) and become infected (D) (scale bar, 10 μm).

The binding of PSGL-1 to P- and L-selectins is abolished by proteolytic cleavage by the O-sialyloglycoprotease ofPastuerella hemolytica. We examined whether pretreatment of cells with this enzyme would reduce HGE binding (22). Pretreatment of the PSGL-1– and sLex-expressing dual transfectant BJAB cells with this enzyme at concentrations of 0.06 to 0.24 mg/ml substantially reduced HGE binding in a dose-dependent manner (e.g., median of 1 bacteria per cell bound after pretreatment with 0.24 mg/ml compared with 7.5 bacteria per cell in controls;P < 0.001). Finally, to examine the possibility that the HGE bacterium might have itself acquired a known selectin [either the gene(s) or the selectin moieties themselves during growth in or exit from the cell] that binds to leukocyte PSGL-1, we pretreated bacteria with mAbs against P-, L-, or E-selectins as well as with a polyclonal antibody against P-selectin in an attempt to either block cell binding or to stain selectin(s) on the surface of HGE organisms (15). These antibodies had no effect on HGE binding and did not bind to components of the HGE bacteria.

The evidence presented here demonstrates that the agent of HGE uses PSGL-1 as its leukocyte receptor or as a key initial receptor component. Bacterial binding and infection were prevented by mAbs against the selectin binding region of PSGL-1, and binding was reduced by proteolytic destruction of PSGL-1. The nonpermissive BJAB cell line, when transfected to express both PSGL-1 and Fuc-TVII/sLex, avidly bound HGE and became infected. Proper glycosylation of PSGL-1 is likely essential for this binding to occur, as evidenced by the failure of BJAB cells expressing PSGL-1 without Fuc-TVII to bind the HGE agent and by previous studies showing that Fuc-TVII/sLexexpression correlates with HGE binding to hematopoietic cell lines (5, 12). The neoexpression of Fuc-TVII/sLex alone, without PSGL-1, slightly enhanced HGE binding (but did not allow infection). In these respects, and in the failure of the PL2 mAb to block binding, the organism's interactions with leukocytes and related cell lines closely resemble the interaction of P-selectin with PSGL-1.

Specific cell surface receptors for intracellular organisms have seldom been identified, none previously on granulocytes. The avidity of HGE for α-(1,3) fucosylated PSGL-1 is likely important in its tropism for phagocytes, which richly express this glycosylated molecule on their surface (13, 18). Engagement of PSGL-1 by P-selectin is known to induce signal transduction (23,24), and such events have been shown to be important in the entry into cells of such pathogens as Yersinia andSalmonella species (25, 26). That signal transduction occurs after HGE binding is supported by our recent findings that HGE infection markedly up-regulates β-chemokine as well as tissue factor secretion (27, 28). Whether PSGL-1–mediated signaling by HGE directly results in endocytosis of the bacterium or triggers changes in expression of other molecules to allow subsequent entry is unknown. However, HGE binding to PSGL-1 likely allows it either to escape or modify a neutrophil phagocytic pathway and thus survive and multiply. Better understanding of this pathway could help define new leukocyte mechanisms that might be manipulated to up- or down-regulate phagocyte responses. Studies to identify the bacterial ligand for PSGL-1 also have important implications. This ligand represents a potential vaccinogen against HGE infection and could also provide a molecular model for the development of compounds to serve as nonselectin ligands for PSGL-1 that could then be used as pharmacologic inhibitors of deleterious cell adhesion–mediated events (29).

  • * To whom correspondence should be addressed. E mail: goodmanj{at}


View Abstract

Navigate This Article