Tyrosine Sulfation of Native Mouse Psgl-1 Is Required for Optimal Leukocyte Rolling on P-Selectin In Vivo

Background We recently demonstrated that tyrosine sulfation is an important contributor to monocyte recruitment and retention in a mouse model of atherosclerosis. P-selectin glycoprotein ligand-1 (Psgl-1) is tyrosine-sulfated in mouse monocyte/macrophages and its interaction with P-selectin is important in monocyte recruitment in atherosclerosis. However, whether tyrosine sulfation is required for the P-selectin binding function of mouse Psgl-1 is unknown. Here we test the function of native Psgl-1 expressed in leukocytes lacking endogenous tyrosylprotein sulfotransferase (TPST) activity. Methodology/Principal Findings Psgl-1 function was assessed by examining P-selectin dependent leukocyte rolling in post-capillary venules of C57BL6 mice transplanted with hematopoietic progenitors from wild type (WT→B6) or Tpst1;Tpst2 double knockout mice (Tpst DKO→B6) which lack TPST activity. We observed that rolling flux fractions were lower and leukocyte rolling velocities were higher in Tpst DKO→B6 venules compared to WT→B6 venules. Similar results were observed on immobilized P-selectin in vitro. Finally, Tpst DKO leukocytes bound less P-selectin than wild type leukocytes despite equivalent surface expression of Psgl-1. Conclusions/Significance These findings provide direct and convincing evidence that tyrosine sulfation is required for optimal function of mouse Psgl-1 in vivo and suggests that tyrosine sulfation of Psgl-1 contributes to the development of atherosclerosis.


Introduction
Atherosclerosis is a chronic inflammatory disease of the arterial wall [1,2]. It is initiated by vascular endothelial injury that leads to endothelial dysfunction and intramural accumulation of oxidized LDL. This causes the elaboration of signalling molecules and induction of adhesion receptors that promotes recruitment of monocytes into the vessel wall, a dominant factor in the initiation and progression of atherosclerosis [3].
We recently examined the importance of tyrosine sulfation in the development of atherosclerosis in a model in which lethallyirradiated Ldlr2/2 mice were rescued with hematopoietic progenitors lacking tyrosylprotein sulfotransferase (TPST) activity [4]. We observed substantial reductions in aortic root lesion size and the number of macrophages in lesions in hyperlipidemic Ldlr2/2 recipients transplanted with TPST deficient progenitors compared to controls. These data indicate that tyrosine sulfation of one or more proteins expressed in hematopoietic cells has a major impact on the development of atherosclerosis. The identities and the relative importance of the tyrosine-sulfated proteins involved are unknown. However, P-selectin glycoprotein ligand-1 (Psgl-1), along with the chemokine receptors Ccr2, Ccr5, and Cx3cr1 are likely candidates [4,5].
Psgl-1 is a homodimeric mucin that is broadly expressed on hematopoietic cells [6]. In mice lacking P-selectin or Psgl-1, leukocyte rolling is virtually absent in a model of trauma induced P-selectin expression in post-capillary venules [7,8]. Thus, Psgl-1 is the major physiologic ligand for P-selectin [9]. Psgl-1 is also a key player in the development of atherosclerosis. Psgl-1 expressed on Ly-6C hi monocytes is a major mediator of monocyte recruitment into atherosclerotic lesions in mice, and aortic root lesions are <40% smaller in hyperlipidemic ApoE2/2; Selplg2/2 mice compared to ApoE2/2 mice [10]. In addition, transient P-selectin or Psgl-1 blockade using mAbs reduces macrophage influx and neointima formation in a model of arterial injury in ApoE2/2 mice [11].
Structure-function relationships for human PSGL-1 have been defined in great detail. The P-selectin binding site spans <15 residues near the N-terminus of the mature polypeptide, it contains sulfotyrosine residues at positions 5, 7, and 10 and a core 2 O-glycan terminating with sialyl-Le x linked to Thr16 [12,13,14,15]. Together these structural features are both necessary and sufficient for P-selectin binding. However, for mouse Psgl-1 the structure-function relationships are not as clearly defined. Like human PSGL-1, the P-selectin binding site of mouse Psgl-1 is near the N-terminus as defined by function blocking mAbs, but its amino acid sequence is considerably different than that of human PSGL-1 ( Fig. 1) [16,17].
In contrast to the substantial evidence for the importance of Oglycosylation, the importance of tyrosine sulfation of mouse Psgl-1 for P-selectin binding has not been definitely demonstrated. Indeed, it was only recently that mouse Psgl-1 was shown to be tyrosine-sulfated [4]. Tyr13 and Tyr15 are the only possible sulfation sites because they are the only tyrosine residues in the extracellular domain. The only study addressing the potential importance of tyrosine sulfation for mouse Psgl-1 is that of Xia et al who examined the function of recombinant mouse Psgl-1 expressed in Chinese hamster ovary (CHO) cells stably expressing human FucTVII and human C2GlcNAcT-I in vitro [21]. They observed that TyrRPhe substitution at position 13, but not at position 15, impaired P-selectin binding and rolling of CHO cells on P-selectin in vitro, suggesting that Tyr13 was sulfated. However, the authors noted that their results should be interpreted cautiously because amino acid substitutions might impair function indirectly.
We therefore sought to directly examine the functional importance of tyrosine sulfation of native mouse Psgl-1 by testing its function in leukocytes lacking endogenous TPST activity in vivo. To accomplish this, mice were transplanted with hematopoietic progenitors from mice lacking both the Tpst1 and Tpst2 genes, the only TPSTs expressed in mice, and the rolling behaviour of TPST deficient leukocytes was examined in wellcharacterized, physiologically relevant assays. We found that TPST deficient leukocytes roll on P-selectin in vivo and in vitro. However, rolling of TPST deficient leukocytes was less efficient than wild type leukocytes despite equivalent surface expression of Psgl-1.

Ethics statement
All procedures involving vertebrate animals were reviewed and approved by the Institutional Animal Care and Use Committee at the Oklahoma Medical Research Foundation (Protocol #W0070).

Hematopoietic transplantation
Tpst1;Tpst2 double knockout (Tpst DKO) mice were generated and characterized as previously described [22,23]. These mice have severely impaired post-natal viability. Therefore, fetal livers were used as the source of hematopoietic progenitors. Lethallyirradiated B6.SJL-Ptprc a Pep3 b /BoyJ recipients (B6.SJL, The Jackson Laboratory, Stock #002014) were transplanted with E15.5 fetal liver cells from wild type 129S6 or Tpst DKO mice, which are in the 129S6 background as described previously [4]. These groups are abbreviated as WTRB6 and Tpst DKORB6, respectively. All studies were conducted 16-24 weeks after transplantation. Complete blood counts were determined at the time of experimentation as previously described [4].

Intravital microscopy
Mice were anesthetized, placed on a warmed microscope stage, and a catheter was placed in the left carotid artery for injections and blood sampling. Exteriorization of the cremaster muscle was used to induce P-selectin-dependent leukocyte rolling [24,25]. The cremaster muscle was mounted on an observation portal and continuously bathed with Hank's balanced salt solution or 131.9 mM NaCl, 18 mM NaHCO 3 , 4.7 mM KCl, 2.0 mM CaCl 2 and 2 mM MgSO 4 , pH 7.2 equilibrated with 79% N 2 and 16% CO 2 and 5% O 2 at 36uC. All data collection was completed within 20 min of exteriorization of the cremaster muscle.
Observations of post-capillary venules were made using a Nikon Eclipse E600-FN microscope equipped with a water immersion objective (40x/0.80 W). Images were recorded using a CCD camera (DC-330E, Dage-MTI) and centerline velocities (v CL ) were measured using an optical doppler velocimeter (Microvessel Velocity OD-RT, CircuSoft Instrumentation). Vessel diameter and the distance leukocytes rolled were determined from recorded images using a digital image processing system (SGI O2 workstation running Inovision ISEEH v5.24 software) and freezeframe advancing.
Rolling flux fractions were calculated by dividing leukocyte rolling flux, defined as the number of rolling leukocytes passing a line perpendicular to the vessel axis over a period of 1 min, by total leukocyte flux estimated as WBC N v b N p N (d/2) 2 , where WBC is total leukocyte count, v b is mean blood flow velocity (v CL N 0.625) and d is vessel diameter [26]. Rolling velocities for 10 leukocytes passing a line perpendicular to the vessel axis were measured in the same venules as rolling flux fractions. Leukocytes were analyzed for a period of 1 s (30 frames). Mean rolling velocity was calculated by dividing the distance travelled by the elapsed time.
Leukocyte interaction with the vessel wall was considered as rolling and not free flowing when velocities were below the critical velocity estimated as v crit = v b N e N (2 -e), where e is the ratio of the leukocyte diameter to vessel diameter [27]. The leukocyte diameter is taken to be 7 mm [28]. Wall shear rates (c w ) were estimated as c w = 4.9 (8 v b /d) where 4.9 is a correction factor obtained from velocity profiles determined using microparticle image velocimetry in microvessels [29,30,31].

Parallel plate rolling assays
Polystyrene 35-mm dishes were coated with anti-human IgM Fc mAb (20 mg/ml, clone MH15-1, Accurate Chemical & Scientific) in HBSS overnight at 4uC. Dishes were washed with HBSS, 0.1% human serum albumin (HSA), blocked with HBSS, 1% HSA for 2 h and then incubated for 1 h at 37uC with media from COS-7 cells transfected with plasmids encoding mouse P-selectin/IgM or mouse CD45/IgM chimera. The plasmids were from by John B. Lowe (University of Michigan) and the conditioned media was kindly provided by Dr. Lijun Xia and John Michael McDaniel (Oklahoma Medical Research Foundation). P-selectin site densities were determined using 125 I-labeled RB40.34 [32]. In some experiments dishes were pre-incubated with blocking P-selectin mAb RB40.34 (20 mg/ml, 1 h).
Bone marrow cells were flushed from femurs and passed through a 40 mm filter. Erythrocytes were lysed and cells were pelleted and resuspended in HBSS, 0.5% HSA at 0.5610 6 cells/ml. Cells were drawn through a parallel plate flow chamber (GlycoTech) using a PHD 200 syringe pump (Harvard Apparatus). Rolling leukocytes were observed using a Zeiss Axiovert 200 microscope equipped with a 20x/0.3 Ph1 objective. After 5 minutes, images were recorded using a CCD camera (XC-77, Hamamatsu Photonics) and analyzed using the image processing system described above. For each experiment, rolling leukocytes were analyzed in 4 fields in a vertical line perpendicular to the direction of flow. The number of rolling cells was converted to cells/mm 2 and the mean rolling velocities of 10 cells in each field were calculated by dividing the distance travelled by the elapsed time.

Statistical Analysis
Differences in rolling velocity and rolling flux fraction were determined using independent samples t-tests using SPSS (SPSS for Mac, rel. 18.0). In addition to the p-values for the t-test, we present the effect size, Cohen's d, which measures the magnitude of the difference between the two group means expressed in terms of standard deviation. A Cohen's d value $0.8 represents a large effect size [33]. All tests were two-tailed and an a #0.05 was set for statistical significance. All results are represented as the mean 6 S.E.M.

Hematopoietic reconstitution
To assess the efficiency of reconstitution of Tpst DKO hematopoiesis in B6.SJL recipients, complete blood counts and the percentage of donor (CD45.2 + ) cells were determined at the time of experimentation at 16-24 weeks after transplantation. We observed that the total leukocyte, neutrophil, lymphocyte, monocyte, erythrocyte and platelet counts were normal and that there were no significant differences between the two transplant groups (Table 1).
For in vivo rolling and P-selectin binding studies using peripheral blood, 96.260.7% (n = 10) of circulating neutrophils and monocytes in WTRB6 mice and 94.761.2% (n = 17) in Tpst DKORB6 mice were CD45.2 + at the time of experimentation. For in vitro rolling and P-selectin binding studies using bone marrow leukocytes, 97.460.5% (n = 4) of bone marrow leukocytes in WTRB6 mice and 96.561.2% (n = 6) in Tpst DKORB6 mice were CD45.2 + at the time of experimentation.
Mice appeared normal with no clinical signs of graft vs. host disease (i.e. diarrhea). Furthermore, body weights of the WTRB6 group (22.060.4 g, n = 13) were similar to the Tpst DKORB6 group (23.160.5 g, n = 13). Taken together, these data demonstrate efficient reconstitution of donor hematopoiesis in the B6.SJL recipients, confirming the histocompatibility of the donor-recipient pair.

P-selectin-dependent leukocyte rolling in vivo
P-selectin-dependent leukocyte rolling in post-capillary venules was induced by exteriorization of the cremaster muscle. Rolling flux fractions were quantitated in 49 venules in 7 WTRB6 mice and 39 venules in 7 Tpst DKORB6 mice and leukocyte rolling velocities were determined for an average of 8.5 rolling cells in each venule ( Fig. 2A and B). Hemodynamic and microvascular parameters including venule diameter, centerline velocity and wall shear rates were comparable in WTRB6 and Tpst DKORB6 mice ( Table 2). We observed that rolling flux fractions were lower in Tpst DKORB6 venules (8.361.1%, n = 39 venules) compared to WTRB6 venules (18.061.8%, n = 49 venules). We also observed that leukocyte rolling velocities were higher in Tpst DKORB6 venules (88.664.8 mm/s) compared to WTRB6 venules (33.161.8 mm/s). Statistical analysis showed that rolling flux fraction in Tpst DKORB6 venules were significantly lower (p,0.0001, d = 0.96) and rolling velocities were significantly higher in the Tpst DKORB6 group compared to the WTRB6 group (p,0.0001, d = 2.4).
In some experiments, blocking mAbs to P-selectin or Psgl-1 were administered after initial data collection and venules were reexamined for rolling leukocytes. In each of these experiments observations were made before and immediately after mAb administration in a single venule and then in an additional 2-7 venules in each animal. We observed that rolling leukocytes, defined as those with velocities less than v crit , were undetectable after administration of 10 mg of P-selectin mAb RB40.34 in both WTRB6 and Tpst DKORB6 mice (n = 3). In a separate series of experiments, rolling leukocytes were also undetectable in WTRB6 and Tpst DKORB6 mice (n = 3) after administration of 10 mg of Psgl-1 mAb 4RA10.

P-selectin-dependent leukocyte rolling in vitro
To study P-selectin-dependent rolling in a more defined system, leukocyte rolling was observed on P-selectin coated dishes in a parallel plate flow chamber. The number of rolling leukocytes and leukocyte rolling velocities were quantitated in three independent experiments comparing leukocytes harvested from wild type and Tpst DKORB6 mice. Bone marrow leukocytes were harvested and drawn over dishes coated with mouse P-selectin/IgM at a shear stress of 1 dyn/cm 2 as described in Methods.
For wild type leukocytes, we observed 176615 rolling cells/ mm 2 , whereas for Tpst DKORB6 leukocytes we observed only 9766 rolling cells/mm 2 (n = 12 fields from 3 independent paired experiments) (Fig. 3A). Statistical analysis showed that the number of rolling cells in the Tpst DKORB6 group was significantly lower than the WT group (p,0.0001, d = 2.1).
In the same experiments, the velocities of 10 individual leukocytes in each of the 4 fields observed were measured. We found that wild type leukocytes rolled with a mean velocity of 1.660.1 mm/s. In contrast, Tpst DKORB6 leukocytes had mean rolling velocities of 2.460.1 mm/s (Fig. 3B). Statistical analysis showed that rolling velocities were significantly higher in the Tpst DKORB6 group compared to the WT group (p,0.0001, d = 2.3). No detectable leukocyte rolling was observed on dishes coated with mouse CD45/ IgM or when dishes coated with P-selectin/IgM were pre-incubated with the P-selectin blocking mAb RB40.34 (data not shown).

P-selectin binding and Psgl-1 expression
P-selectin binding to Psgl-1 on neutrophils in peripheral blood was determined using flow cytometry. Neutrophils were gated based on their forward and orthogonal light scattering properties and on donor origin (CD45.2 + ). We observed that the mean fluorescence intensity (MFI) of P-selectin binding to WTRB6 cells was 1,3226138 (n = 3) and 557664 for Tpst DKORB6 cells (n = 7) (Fig. 4A & B). This difference is highly significant (p = 0.016). Pre-incubation of cells with P-selectin blocking mAb RB40.34 or Psgl-1 blocking mAb 4RA10 completely blocked binding of the P-selectin/IgM to levels equivalent to that for CD45/IgM (data not shown).
These samples were also analyzed by flow cytometry using the Psgl-1 blocking mAb 2PH1, to investigate whether the reduced rolling of Tpst DKO cells and reduced binding of P-selectin to Tpst DKO neutrophils was due to altered surface expression of Psgl-1. We observed that MFI of 2PH1 binding to neutrophils from WTRB6 mice (3,4886169, n = 3) and Tpst DKORB6 mice were indistinguishable (3,3226123, n = 7, p = 0.47) (Fig. 4C & D). We also examined Psgl-1 expression on leukocytes from a rare Tpst DKO mouse at post-natal day 14 using 4RB12, a non-blocking antibody to Psgl-1. 4RB12 binding to peripheral blood neutrophils and bone marrow leukocytes from the Tpst DKO mouse was indistinguishable from an age-matched wild type mouse examined in parallel (data not shown).

Discussion
We recently reported that transplantation of Ldlr2/2 mice with Tpst DKO hematopoietic progenitors drastically attenuated  development of atherosclerosis [4]. This result indicated that tyrosine sulfation of one or more proteins expressed in hematopoietic cells has a major impact on the development of atherosclerosis. Psgl-1 is one likely candidate because it is known to be tyrosine-sulfated in the mouse and its role in monocyte recruitment in atherosclerosis is well established [4,10,11]. We therefore sought to directly examine the functional importance of tyrosine sulfation for Psgl-1 in vivo.
To address this question, mice were transplanted with hematopoietic progenitors from mice lacking endogenous TPST activity and the rolling behaviour of TPST deficient leukocytes was examined in a well-characterized model of trauma induced Pselectin expression in post-capillary venules in the cremaster muscle. We observed that significantly fewer TPST deficient leukocytes rolled in post-capillary venules in the Tpst DKORB6 group compared to wild type leukocytes in the WTRB6 group and TPST deficient leukocytes rolled at significantly higher velocities than wild type leukocytes. These observations were confirmed in two well-defined in vitro assay systems. First, in a parallel plate adhesion assay, fewer TPST deficient leukocytes rolled on P-selectin and they rolled at higher velocities compared to wild type leukocytes under physiologically relevant shear stress. In addition, in a flow cytometry assay, binding of fluid-phase Pselectin to TPST deficient peripheral blood and bone marrow leukocytes was significantly lower than binding to wild type leukocytes. Importantly, we showed that impaired rolling in vivo and in vitro and impaired binding of fluid-phase P-selectin to TPST deficient leukocytes was not due to differences in surface expression of Psgl-1. Finally, antibody blocking experiments in Tpst DKORB6 mice showed that anti-Psgl-1 mAb abolished Pselectin dependent rolling in vivo. Taken together these observations demonstrate that tyrosine sulfation enhances the binding capacity of mouse Psgl-1, but is not an absolute requirement for Psgl-1 function in vivo. These findings, in conjunction with our previous report that atherosclerosis is attenuated in hyperlipidemic Ldlr2/2 mice with Tpst DKO hematopoiesis, suggest that tyrosine sulfation of Psgl-1 may contribute to lesion development.
It is formally possible that impaired rolling of TPST deficient leukocytes is due to differences in O-glycosylation of Psgl-1 compared to wild type leukocytes. However, it is difficult to envision how O-glycosylation, that occurs in an earlier Golgi compartment, could be impacted by the presence or absence of tyrosine sulfation that occurs in the trans-Golgi network [34,35].
In our studies, leukocyte rolling was completely abrogated by injection of a blocking P-selectin antibody. This is consistent with published data that leukocyte rolling in this model is entirely dependent on P-selectin expression on the post-capillary venules [25]. Psgl-1 is the predominant ligand for P-selectin is the early phases (,30 min) after trauma-induced inflammation in the mouse cremaster. However, previous studies indicate that a minor component of P-selectin-dependent rolling in this model is Psgl-1independent [16,19,36,37]. For example, Yang et al reported that rolling flux fraction was severely reduced but detectable in Psgl-1 deficient animals (1.2%) compared wild type controls (20.9%) [8]. In addition, Sperandio et al reported that administration of the anti-Psgl-1 mAb 4RA10 to wild type mice reduced rolling flux fraction from 27% to 8% and increased rolling velocities from 44 to 110 mm/sec. In our study, rolling was abolished by 4RA10 in both WTRB6 and Tpst DKORB6 mice. Thus, we do not detect a Psgl-1-independent component of P-selectin-dependent rolling that has been reported by others.
Our findings provide strong support for previous in vitro observations by Xia et al, who examined the effects of site-directed mutagenesis and sodium chlorate on mouse Psgl-1 function in CHO cells stably expressing human FucT-VII and C2GlcNAcT-I   [21]. Sodium chlorate inhibits synthesis of the sulfate donor PAPS and therefore blocks the action of all sulfotransferases [38]. They reported that mutagenesis of Tyr13, but not Tyr15, to Phe or incubation of cells with sodium chlorate impaired, but did not abolish P-selectin binding and rolling of the transfected CHO cells. Although this implicates Tyr13 as a potential site for sulfation, these observations do not prove that Tyr13 is sulfated and that Tyr15 is not, because Tyr to Phe substitution(s) might impair function indirectly by altering the protein conformation or may affect sulfate addition at the nearby non-mutated tyrosine. Thus, further studies are necessary to directly determine the precise location and stoichiometry of sulfation.
In summary, we examined the functional role for tyrosine sulfation of mouse Psgl-1 using physiologically relevant assay systems in a unique model in which mice were transplanted with hematopoietic progenitors from mice lacking TPST activity. This model enabled examination of mouse Psgl-1 function in a native mouse leukocyte modified by endogenous mouse glycosyltransferases without altering the amino acid sequence of the protein. Our studies provide direct and convincing evidence that tyrosine sulfation is required for optimal function of mouse Psgl-1 in vivo.