Skip main navigation

Glycosphingolipids on Human Myeloid Cells Stabilize E-Selectin–Dependent Rolling in the Multistep Leukocyte Adhesion Cascade

Originally published, Thrombosis, and Vascular Biology. 2016;36:718–727



Recent studies suggest that the E-selectin ligands expressed on human leukocytes may differ from those in other species, particularly mice. To elaborate on this, we evaluated the impact of glycosphingolipids expressed on human myeloid cells in regulating E-selectin-mediated cell adhesion.

Approach and Results—

A series of modified human cell lines and primary neutrophils were created by targeting UDP-Glucose Ceramide Glucosyltransferase using either lentivirus-delivered shRNA or CRISPR-Cas9-based genome editing. Enzymology and mass spectrometry confirm that the modified cells had reduced or abolished glucosylceramide biosynthesis. Glycomics profiling showed that UDP-Glucose Ceramide Glucosyltransferase disruption also increased prevalence of bisecting N-glycans and reduced overall sialoglycan expression on leukocyte N- and O-glycans. Microfluidics-based flow chamber studies demonstrated that both the UDP-Glucose Ceramide Glucosyltransferase knockouts and knockdowns display ≈60% reduction in leukocyte rolling and firm adhesion on E-selectin bearing stimulated endothelial cells, without altering cell adhesion to P-selectin. Consistent with the concept that the glycosphingolipids support slow rolling and the transition to firm arrest, inhibiting UDP-Glucose Ceramide Glucosyltransferase activity resulted in frequent leukocyte detachment events, skipping motion, and reduced diapedesis across the endothelium. Cells bearing truncated O- and N-glycans also sustained cell rolling on E-selectin, although their ability to be recruited from free fluid flow was diminished.


Glycosphingolipids likely contribute to human myeloid cell adhesion to E-selectin under fluid shear, particularly the transition of rolling cells to firm arrest.


P-(CD62P), E-(CD62E), and L-selectin (CD62L) constitute a family of C-type lectins that initiate the multistep process of leukocyte recruitment at sites of inflammation.13 Among these, P- and E-selectin are expressed on stimulated endothelial cells, and they mediate the direct capture of leukocytes to the inflamed vessel. L-selectin is expressed on leukocytes, and this facilitates secondary capture via leukocyte–leukocyte adhesion. All members of the selectin family bind cell-surface glycoconjugates expressed on blood leukocytes with rapid on- and off-rates in a calcium-dependent manner.46 Such high on-rates enable the capture or tethering of leukocytes from flowing blood. Following this, a succession of rapid bond formation and breakage events results in cell rolling on the vessel wall. Rolling provides residence time during which leukocytes can be activated by chemokines expressed on the inflamed endothelium and also via signaling through selectin-ligand bonds. This then leads to firm adhesion and leukocyte diapedesis across the vascular endothelium. Similar to leukocyte adhesion during inflammation, selectin-mediated binding also regulates lymphocyte homing, hematopoietic stem cell migration, and cancer metastasis.

Our knowledge regarding the glycoconjugates that bind selectins during inflammation comes, in large part, from studies using transgenic mice or human blood. Such studies have demonstrated that the sialyl Lewis-X (sLeX) tetrasaccharide attached to a core-2 O-glycan at the N-terminus of P-selectin glycoprotein ligand-1 (PSGL-1, CD162) is a major ligand for P- and L-selectin in both humans and mice.79 Although only 2% of the overall O-glycans of human PSGL-1 contain the sLeX core-2 O-glycan,10 in myeloid cells, this structure constitutes ≈20% of the carbohydrates at Thr-57 on the N-terminus of this glycoprotein.11 With regard to E-selectin, the physiological carbohydrate ligands likely differ between mice and humans.1,3 Notably, the rolling of mouse neutrophils on E-selectin is protease-sensitive but this is not the case for human leukocytes.12,13 Further, although the E-selectin ligands in mouse neutrophils have largely been identified to include PSGL-1, E-selectin ligand-1, and CD44,14 the major players in humans remain unknown.1,3 In this regard, PSGL-1 is a relatively minor ligand for human E-selectin, and there is no homolog for E-selectin ligand-1 in humans.1,15 Also, functional sialofucosylated CD44 glycoforms do not exist in mature leukocytes.16 Other candidate E-selectin glycoconjugate ligands in humans include CD43, integrin Mac-1, L-selectin, and glycosphingolipids (GSLs).13 The absolute and relative importance of these macromolecules to human leukocyte adhesion remains unknown.

Among the E-selectin ligands, the current study evaluated the potential importance of GSLs during selectin-dependent myeloid cell adhesion because proof supporting their functional importance is incomplete. In this regard, previous studies have suggested that a group of sialofucosylated lactosylceramides called myeloglycans can bind E-selectin.17,18 These GSLs contain 4 or more N-acetyllactosamine (LacNAc) repeats, with one or more internal fucose residues.18 NeuAcα2,3Galβ1,4GlcNAcβ1,3[Galβ1,4(Fucα1,3)GlcNAcβ1,3]2[Galβ1,4GlcNAcβ1,3]2Galβ1,4GlcβCer represents a prototypic myeloglycan selectin-ligand.19,20 In these studies, myeloglycans extracted from human promyelocytic HL-60 cells (HL-60s)19 and also primary human neutrophils20 were shown to support the rolling of E-selectin expressing Chinese Hamster Ovary cells under shear. Additionally, inhibition of glycolipid biosynthesis by the small molecule P4 (D-threo-1-phenyl-2-palmitoylamino-3-pyrrolidino-1-propanol) for 24 hours partially (≈50%) reduced neutrophil binding to immobilized E-, but not P-, selectin under static conditions.20 However, these studies were not conducted under fluid shear, a necessary condition to demonstrate the physiological relevance of the GSL ligands.21 Further, pharmacological treatments can have nonspecific effects, and the short-term (24 hours) treatment may leave residual GSLs on cells that did not turn over rapidly.

In the current study, we developed shRNA-based knockdown and clustered regularly interspaced short palindromic repeats/clustered regularly interspaced short palindromic repeat–associated 9 (CRISPR-Cas9)–based knockout strategies to disrupt lactosylceramide biosynthesis because this provides a specific strategy to evaluate the ability of the GSLs to function as E-selectin ligands. To this end, we targeted UDP-Glucose Ceramide Glucosyltransferase (UGCG)20,22,23 because mammalian hematopoietic cells primarily contain glucosylceramides with a neolacto core (Galβ1–4GlcNAcβ1–3Galβ1–4Glcβ1-Cer).24 This strategy was applied to HL-60 promyelocytic leukocytes, HL-60s differentiated to neutrophils, and primary neutrophils derived from human hematopoietic stem cells (hHSCs). The results show that UGCG disruption both prevents stable human granulocyte rolling and reduces diapedesis across inflamed endothelial cells. Thus, GSL binding to endothelial E-selectin may control the transition from myeloid cell rolling to firm arrest at sites of inflammation.

Materials and Methods

Materials and Methods are available in the online-only Data Supplement.


Human but Not Mouse Leukocyte Adhesion to E-Selectin Is Pronase-Insensitive

We sought to confirm that human leukocyte and HL-60 rolling on E-selectin is pronase (a mixture of proteases)-insensitive12,13 (Figure 1). Here, mouse neutrophil rolling and adhesion on E-selectin (stimulated human umbilical vein endothelial cells [HUVECs]) was abrogated by pronase (Figure 1A). This was not the case for either primary human neutrophils (Figure 1B) or HL-60 cells (Figure 1C). In both Figure 1B and 1C, pronase decreased the number of adherent cells without altering cell rolling density. On recombinant P-selectin, the rolling of all cell types was abolished by pronase (Figure 1D–1F). Blocking mAbs against human (KPL1) and mouse (2PH1) PSGL-1 abrogated rolling on P- (bottom row) but not E-selectin. Thus, the N-terminus of PSGL-1 does not contain a major E-selectin ligand (top row). Additional controls confirmed that the molecular interactions in Figure 1A–1C could be completely blocked by anti-E-selectin mAbs (HAE-1f or P2H3), and cell binding in Figure 1D–1F was abrogated by anti-P-selectin mAb G1 (data not shown7,25,26).

Figure 1.

Figure 1. Rolling of human and mouse neutrophils on E- and P-selectin: 2×106/mL wild-type (WT) mouse neutrophils (A and D), human neutrophils (B and E), or HL-60 cells (C and F) were perfused over substrates bearing either E- (top) or P-selectin (bottom) at 1 dyne/cm2. Substrates were composed of either E-selectin-expressing interleukin (IL)-1β-stimulated human umbilical vein endothelial cells (AC) or physisorbed recombinant P-selectin (DF). In some cases, the cells were treated with 1 mg/mL pronase for 1 h at 37°C before perfusion. Individual bars in each panel present the rolling (dark) and adherent (light) cell density 2 minutes after the start of perfusion. 2PH1 and KPL-1 are blocking mAbs against murine and human P-selectin glycoprotein ligand-1 (PSGL-1), respectively. * and † represent statistically significant differences for rolling and adherent cells, respectively (P<0.05 with respect to all other treatments, except *s and †s are not different from each other).

To evaluate the effectiveness of pronase on reducing cell surface glycoprotein levels, the binding of a panel of mAbs to human neutrophils and HL-60s was evaluated (Table I in the online-only Data Supplement). Here, the binding of some of the mAbs like anti-PSGL-1 was dramatically reduced by pronase, whereas others like anti-CD45 and mAb HECA-452 were only partially (≈60%) reduced. Cell surface Mac-1/CD11b was reduced by 62% to 87%, and this may account for the reduced firm adhesion noted in Figure 1B and 1C. Overall, differences between mouse and human leukocyte rolling on E-selectin were confirmed. This effect could be because of GSL ligands that are unique to human cells, residual glycoprotein E-selectin ligands that were not digested by pronase, or the exposure of new functional epitopes in human cells that were previously cryptic.

Silencing UGCG in HL-60s

ShRNA was applied to silence UGCG on HL-60s to address the shortcomings of the pronase approach (Figure 2). This caused a 93% reduction in UGCG mRNA expression as measured by quantitative real-time polymerase chain reaction (Tables II and III in the online-only Data Supplement). Stable UGCG knockdown HL-60s (ie, UGCG¯HL-60) with red fluorescence were then obtained by inserting a DsRed reporter in the lentiviral vector (Figure 2A). Reduction in UGCG enzyme activity in these DsRed-positive UGCG¯HL-60s was confirmed by enzymology (Figure 2B). Here, the C6-NBD-Cer substrate was converted to C6-NBD-GlcCer on addition of UDP-Glc and wild-type (WT) HL-60 cell lysates (lane 1). The thin-layer chromatography Rf values of 0.6 and 0.79 for C6-NBD-GlcCer and C6-NBD-Cer matched previous reports.27 In the UGCG¯HL-60s, UGCG enzymatic activity was reduced by ≈80% (lane 2). No product was observed when either the cell lysate was absent (lane 5) or when GDP-Fuc replaced UDP-Glc (lane 6). Using flow cytometry (Figure 2C), the UGCG¯HL-60s displayed an ≈75% reduction in sLeX and related epitopes recognized by mAb HECA-452 (left panel) and ≈35% reduction in mAb CSLEX-1 binding (middle).

Figure 2.

Figure 2. Silencing UDP-Glucose Ceramide Glucosyltransferase (UGCG) in HL-60s. A, pLKO.1 lentiviral vector with DsRed reporter and UGCG shRNA was used to create stable UGCG¯HL-60 cells. B, Thin-layer chromatography of UGCG enzyme assay shows activity reduction in the UGCG¯HL-60 cell lysate (lane 2) compared with wild-type (WT) HL-60 (lane 1). C6-NBD-GlcCer product standards serve as positive control (lanes 3 and 4). Negative controls either lack cell lysate (lane 5) or contain GDP-Fuc in place of UDP-Glc (lane 6). C, Flow cytometry histograms showing cell-surface expression of cutaneous lymphocyte antigen (CLA)/HECA-452 (left), sialyl Lewis-X (sLeX)/CD15s/CSLEX-1 (middle), and Lewis-X/CD15 (right). Black empty and shaded histograms correspond to WT and UGCG¯HL-60s. Dashed empty histograms correspond to isotype controls. Silencing UGCG reduced cell surface CLA and sLeX expression. D and E, Rolling and adherent cell density (D) and cumulative rolling velocity distribution (E) data for WT and UGCG¯HL-60s perfused over interleukin (IL)-1β-stimulated human umbilical vein endothelial cells at 1 dyn/cm2. Statistical symbols are same as Figure 1. Cells with reduced UGCG activity displayed ≈60% reduced rolling and adherent cell density and ≈2-fold higher rolling velocity. Pronase further reduced the number of adherent UGCG¯HL-60s. Number in brackets present (number of experiments/total number of cells analyzed).

UGCG¯HL-60 rolling on P-selectin under fluid shear was identical to WT HL-60s (Figure I in the online-only Data Supplement). Thus, the core-2 sLeX at the N-terminus of PSGL-1 remained intact in the knockdown cells. In contrast, UGCG silencing reduced HL-60 rolling density on E-selectin bearing–stimulated HUVECs by ≈60% (Figure 2D) and augmented cell rolling velocity by 2-fold (Figure 2E). Anti-PSGL-1 mAb KPL-1 abolished UGCG¯HL-60 rolling on P-selectin but not HUVECs. Pronase reduced the number of adherent UGCG¯HL-60s on HUVECs without affecting cell rolling density. The residual rolling could be either because of incomplete UGCG silencing or pronase insensitive selectin-ligands. Overall, disrupting GSL biosynthesis reduced cell rolling on E-selectin.

Genome Editing and Glycome Analysis of UGCG-KOs

Genome editing allows complete deletion of specific enzyme activity unlike RNAi, which is incomplete.25,28 HL-60 knockouts (KO) were thus generated using CRISPR-Cas9 method (Figure 3). The UGCG-KO cell line thus created contained a 16bp chromosomal deletion as seen in the sequencing results (Figure 3A). Only one UGCG allele was sequenced here because UGCG is present on chromosome-9, and spectral karyotyping of HL-60 demonstrates the loss of the second copy because of a chromosome-9 translocation.29 The absence of off-target editing was also confirmed by sequencing all potentially similar exonic off-target sites. Consistent with this, enzymology-based thin-layer chromatography demonstrates the complete loss of UGCG activity in the UGCG-KO cell lysates (Figure II in the online-only Data Supplement). Loss of glycolipids was also confirmed using liquid chromatography-mass spectrometry–based lipidomics analysis30,31 (Figure III in the online-only Data Supplement). Here, C16-, C24-HexCer, and C16-, C24-(Hex)2Cer were detected in WT HL-60s, but these were completely depleted in the UGCG-KOs.

Figure 3.

Figure 3. UDP-Glucose Ceramide Glucosyltransferase (UGCG) knockout (KO) HL-60s. A, Schematic of clustered regularly interspaced short palindromic repeats/clustered regularly interspaced short palindromic repeat–associated 9 (CRISPR-Cas9) editing vector used to create knockout HL-60 cells lacking UGCG. Top row presents wild-type (WT) HL-60 UGCG sequence, while bottom row shows 16bp chromosomal deletion in the UGCG-KO cells. Bold underlined text highlights the guide RNA sequence, and the bold italicized text denotes the protospacer adjacent motif (PAM/NGG). Only one UGCG allele was sequenced because UGCG is present on chromosome-9, and spectral karyotyping of HL-60 demonstrates the loss of the second copy because of the chromosome-9 translocation. B, Flow cytometry histograms present cell surface expression of cutaneous lymphocyte antigen (CLA)/HECA-452, sialyl Lewis-X (sLeX)/CSLEX-1, LeX/CD15, and VIM-2/CD65s epitopes on WT (black empty histogram) and UGCG-KO HL-60s (gray filled). Dashed empty histogram is isotype control. Mean±SEM data for WT and KO cells is presented in inset. C, L-PHA-FITC (left), Mal-II-biotin (middle), and PNA-FITC (right) lectin binding to WT (black empty) and UGCG-KO (gray filled) HL-60s measured using flow cytometry. Before PNA-FITC staining, the cells were desialylated using α2,3/6/8/9 neuraminidase from A. ureafaciens. In all panels, dashed histograms correspond to cells alone without lectins. Cells incubated with Mal-II-biotin were subsequently detected using α-biotin-FITC. Hatched peaks indicate negative controls: WT cells treated with neuraminidase (middle) and without neuraminidase (right). UGCG knockouts display reduction in CLA and sLeX epitopes (*P<0.05 with respect to WT), but no major change in lectin binding. D, Partial MALDI-TOF MS (matrix-assisted laser desorption/ionization time-of-flight mass spectrometry) spectra of permethylated N-glycans of WT (upper) and UGCG-KO (lower) HL-60 cells. Spectra are from the 50% acetonitrile fraction. Red peaks correspond to decreased sialylated structures, whereas green peaks correspond to increased bisected structures in the UGCG-KO HL-60s (lower). The increase in bisected N-glycans in the UGCG-KO is supported by comparing the relative intensity of ions at m/z 3748 to 3503 (blue peaks). In WT, the ratio of these peaks is 0.356. In the UGCG-KO, it is 1.183, thus suggesting an ≈232% increase in bisected structures in the UGCG-KO HL-60s. Full spectra are shown in Figure IV in the online-only Data Supplement. E, MALDI-TOF MS spectra of permethylated O-glycans of WT (top) and UGCG-KO (bottom) HL-60 cells. Spectra are from the 35% acetonitrile fraction. Annotated structures are according to the Consortium for Functional Glycomics guidelines. All molecular ions are [M+Na]+. Putative structures are based on composition, tandem MS/MS (data not shown), and biosynthetic knowledge. Cartoons that include sugar symbols outside a bracket have not been unequivocally defined. Letters m and M in bold characters indicate minor and major abundances, respectively. L-PHA indicates Phaseolus vulgaris Leucoagglutinin; Mal-II, Maackia Amurensis lectin-II; and PNA lectin, Peanut lectin.

UGCG-KO HL-60s displayed ≈80% reduction in cutaneous lymphocyte antigen/HECA-452 (Figure 3B, top left), ≈50% reduction in sLeX expression (top right), and a small decrease in LeX expression (bottom left). These findings are similar to the UGCG¯HL-60 results (Figure 2C). MAb VIM-2 expression was unchanged, suggesting that this reagent may also recognize glycans on non-GSL glycoconjugates32 (Figure 3B, bottom right). In lectin-binding studies, the UGCG-KOs did not show substantial changes in N-glycan structures as measured using L-PHA (Phaseolus vulgaris Leucoagglutinin) binding, or overall sialylation using both Mal-II (Maackia Amurensis lectin-II) and PNA lectin (Peanut lectin; Figure 3C).

The decrease in the HECA-452 epitope upon knocking-out/down UGCG prompted detailed mass spectrometric profiling of HL-60-associated N- (Figure 3D; Figures IV–VI in the online-only Data Supplement) and O-glycans (Figure 3E). Here, MALDI-TOF MS (matrix-assisted laser desorption/ionization time-of-flight mass spectrometry) revealed a complex N-glycan profile in WT HL-60s consisting of multi-antennary structures decorated with varying levels of fucose and sialic acid (NeuAc) similar to previous work25 (Figure 3D, upper panel; Figure IV in the online-only Data Supplement). Disrupting UGCG resulted in less sialylated structures and a concomitant increase in bisected N-glycans (Figure 3D lower panel; Figure IV in the online-only Data Supplement). The knockdown UGCG¯HL-60s also displayed an intermediate relative abundance of the sialylated and bisected N-glycans (Figure V in the online-only Data Supplement). MALDI-TOF MS and MALDI-TOF/TOF MS/MS (Figure VI in the online-only Data Supplement) analysis of N-glycans subjected to endo-β-galactosidase digestion revealed the extent of polylactosamine repeats (poly-LacNAcs) and defined the terminal epitopes on extended antennae. Here, compared with undecorated LacNAc (m/z 722), we noted a partial decrease in sialylated LacNAc (m/z 1083, 55%), sLeX (m/z 1257, 33%), and VIM-2 (m/z 1706, 15%) on knocking-out UGCG (Figure VI in the online-only Data Supplement, left versus right panels; Table IV in the online-only Data Supplement). Poly-LacNAcs (m/z 518) and LeX terminal epitopes (m/z 896 and 1519) remained unaltered. No major difference was detected in the relative abundance of multibranched N-glycans except for an ≈30% decrease in tetra-antennary N-glycans in the UGCG-KO HL-60s (Table V in the online-only Data Supplement).

The O-glycan profile of the WT HL-60s consisted of core-1 (m/z 895 and 1256) and core-2 (m/z 983, 1344, 1518, and 1879) structures (Figure 3E, top panel).25 In the UGCG-KO HL-60s, overall, similar structures were observed, however, accompanied by a decreased sialylation (Figure 3F, bottom panel). In this regard, we noted a decrease in the double sialylated-Tn antigen (m/z 1256) over the sialylated-Tn antigen (m/z 895), a decrease of the double sialylated core 2 O-glycan (m/z 1705) over the monosialylated core 2 O-glycan (m/z 1344), and a reduction in core-2 O-glycans carrying the sLeX terminal epitope (m/z 1879; Figure 3D, top versus bottom panels). Thus, a decrease in N- and O-glycan sialylation accompanied UGCG silencing/deletion.

UGCG-KO HL-60s Exhibit Skipping Motion on Stimulated Endothelial Cells

The effect of UGCG disruption on cell rolling and adhesion was quantified (Figure 4; Movie A in the online-only Data Supplement). Here, consistent with the silencing approach, the UGCG-KO HL-60s displayed a 50% reduction in cell rolling and 70% reduction in firm adhesion density on stimulated HUVECs where cell adhesion is E-selectin-dependent (Figure 4A). Pronase did not alter the total number of interacting cells, though it decreased the proportion of adherent cells and increased the median rolling velocity of both the WT and the UGCG-KO HL-60s by ≈1.5- to 2-fold (Figure 4B). UGCG-KOs treated with pronase exhibited a 3- to 4-fold higher median rolling velocity compared with the WT HL-60s. Similar partial reduction in cell rolling density and 3-fold increase in cell rolling velocity was observed when the knockout cells were perfused over recombinant E-selectin substrates (Figure VII in the online-only Data Supplement).

Figure 4.

Figure 4. E-selectin-mediated rolling of UGCG-KO HL-60s. A and B, Rolling and adherent cell density (A) and rolling velocity distribution (B) data for UGCG-KO HL-60s measured under conditions in Figure 2D and 2E, respectively. Statistical symbols are same as in Figure 1. In addition, ** denotes statistically significant difference (P<0.05) in cell rolling with respect to all other treatments. UGCG-KO HL-60s display ≈60% reduced rolling and adherent cell density, with ≈3-fold higher median rolling velocity compared with WT HL60s. Pronase further increased rolling velocity. C, Individual columns present total number of cell tethers formed in a 4-minutes interval. Number of tethers did not vary significantly with cell type or on pronase treatment. D, Tethered cells (normalized to 100%) were classified into (1) adherent, (2) rolling, or (3) detached cells as defined in Methods. Cell detachment was prominent in UGCG-KO HL-60 cells (for detached cells: ‡P<0.05 with respect to all treatments except ‡s are not different from each other, and #P<0.05 with respect to all other treatments). Number of adherent cells is also highest for WT HL60s (†P<0.05 with respect to all treatments except †s are not different from each other). Instantaneous rolling velocity (E) and cumulative distance travelled with time (F) for 5 representative cells for each treatment. WT HL-60s rolled stably at 3 to 15 μm/s, whereas all the other cell types show abrupt increases in rolling velocities as seen in the intermittent peaks ≤250 μm/s. Detachment events are depicted by up arrows in F. UGCG indicates UDP-Glucose Ceramide Glucosyltransferase.

Detailed tethering analysis on stimulated HUVECs confirmed that the disruption of GSL biosynthesis may reduce the stability of leukocyte rolling (Figure 4C–4F). Here, the number of tethers formed was similar for all treatments (Figure 4C). Among these cells, only 5% cell detachment was observed in the case of WT HL-60s with 65% of the tethered cells eventually transitioning to firm arrest over 4 minutes (Figure 4D). For all other treatments, the number of adherent cells decreased dramatically. Both pronase treatment and knocking out UGCG caused 50% to 60% of the tethered cells to detach in the experimental window. Detachment was even higher (≈80%) when pronase was applied to the UGCG-KO HL-60s. The additive effect of pronase and UGCG deletion was also noted on following individual cell trajectories (Figure 4E and 4F). Here, the WT HL-60s exhibited robust rolling with low, stable velocities (top panels). Pronase (second panel) and UGCG gene disruption (third from top) often increased instantaneous rolling velocities to 200 to 250 μm/s, with frequent cell skipping motion leading to detachment events. Thus, the cumulative distance travelled by these cells was higher. The strength of cell interactions was further decreased on pronase treatment of UGCG-KO HL-60s (Figure 4F). Similar experimental results as with the UGCG-KOs were also observed on performing tethering analysis on the UGCG¯HL-60s (Figure VIII in the online-only Data Supplement).

UGCG Disruption Reduced Cell Transmigration

HL-60s were differentiated toward neutrophils using DMSO to determine whether reduced UGCG activity affects leukocyte transmigration across HUVEC monolayers (Figure 5; Movie B in the online-only Data Supplement). During such differentiation, HL-60 cell size decreased to more closely resemble the size of primary human neutrophils. CD11b expression on the WT HL-60s also increased 5-fold over 5 days, with cell differentiation proceeding markedly faster for the UGCG-KOs and UGCG¯HL-60s (Figure 5A). Although the sLeX expression on the undifferentiated UGCG-KO was low compared with WT-HL-60s, the HECA-452 binding/sLeX expression was comparable for both cells on differentiation. Further, LeX levels were reduced for all cell types at day 5. On differentiation, the WT, UGCG¯HL-60s, and UGCG-KOs all acquired an ability to transmigrate across the endothelium.

Figure 5.

Figure 5. Studies with differentiated HL-60 cells. HL-60s were differentiated toward granulocytes by culturing cells in growth medium containing 1.3% DMSO for 5 days. A, Cell surface glycan and glycoprotein expression over the course of differentiation. B, Rolling and adherent cell density data for differentiated cells binding to stimulated human umbilical vein endothelial cell monolayers at 1 dyne/cm2 with or without pronase treatment. Experimental method and statistical symbols are identical to those in Figure 1. C and D, % Transmigration, under shear (C) and static (no flow, D) conditions. Plot quantifies the % of total interacting cells in a field-of-view that transmigrated during a 20 minutes interval. ‡ and # denote P<0.05 with respect to all treatments, except these symbols are not different from each other. Increased cell detachment of UGCG-KO and UGCG¯HL-60s reduced transmigration under shear. UGCG indicates UDP-Glucose Ceramide Glucosyltransferase.

Both the undifferentiated and differentiated HL-60s displayed rolling and firm adhesion interactionsm though the fraction of cells transitioning to firm arrest was higher for the differentiated cells. This is presumably because of the smaller size and higher Mac-1 expression levels of the differentiated cells (Figure 1C versus Figure 5B). An ≈50% to 60% reduction in rolling and adherent cell density was observed for both the differentiated UGCG and UGCG-KO HL-60s compared with WT. Although pronase only partially reduced the number of adherent cells in the case of differentiated WT HL-60s, it reduced the total number of interacting cells by ≈80% to 90% in the case of the differentiated cells lacking UGCG (both knockdowns and knockouts).

In static transmigration assays, differentiated HL-60 transmigration was reduced by anti-CD18 blocking mAb IB4 consistent with previous data,33 and it was abolished by pronase (Figure 5D). The anti-E-selectin mAb P2H3 partially reduced transmigration efficacy, although the effect did not reach statistical significance. In addition, both the UGCG knockdowns and knockouts displayed reduced transmigration by 50% to 60% under static conditions. The effect of UGCG disruption on leukocyte transmigration was more prominent under hydrodynamic shear (65% to 80% reduction), likely because of the reduced recruitment of these cells on E-selectin-expressing stimulated HUVECs under flow (Figure 5C).

UGCG Disruption Reduces the Recruitment of Primary Neutrophils Derived From hHSCs

To determine the role of the GSLs in regulating primary human myeloid cell adhesion, lentiviral gene silencing was performed on CD34+ hHSCs while they were being differentiated toward neutrophils (Figure 6). This typically resulted in ≈25% to 35% red fluorescence cells because both the control and the UGCG shRNA virus carried a DsRed reporter (Figure 6A). Microfluidics-based cell adhesion assays then followed the transduced, fluorescent control (DsRed-Neu) and knockdown (UGCG¯-Neu) cells. Here, on interleukin-1β-stimulated HUVEC monolayers, an ≈50% reduction of cell adhesion was robustly observed in hHSC-derived UGCG¯ neutrophils compared with the DsRed+ controls (Figure 6B). Thus, similar to the HL-60s, attenuation of UGCG activity reduced primary granulocyte adhesion under physiological flow.

Figure 6.

Figure 6. Neutrophils derived from human hematopoietic stem cells (hHSCs). hHSCs were differentiated to mature neutrophils over 12 days. UGCG¯-Neu were generated using lentivirus carrying a DsRed reporter and UGCG shRNA. Control cells generated using a vector lacking the shRNA was called DsRed-Neu. A, Transduction efficiency quantified based on % cells that were DsRed-positive (bottom right quadrant). B, Rolling and adherent cell density data for DsRed+ neutrophils derived from hHSCs on interleukin (IL)-1β-stimulated human umbilical vein endothelial cells. * and † denote statistically lower rolling and adherent cell density on E-selectin on silencing UGCG (P<0.05 with respect to WT). Data are presented for hHSCs derived from 4 different human donors. UGCG indicates UDP-Glucose Ceramide Glucosyltransferase.


Methods to disrupt E-selectin-mediated cell adhesion and the identification of physiological human E-selectin ligands is important, both in the context of inflammatory diseases and hematopoietic stem cell homing.1,3 This is because leukocyte rolling on E-selectin results in slower rolling velocities compared with L- and P-selectin because of low molecular binding on- and off-rates.26,34,35 Such slow rolling enhances leukocyte–endothelium contact, facilitating enhanced cell activation, the conversion of rolling cells to firm arrest, and diapedesis. Additionally, unlike the human ortholog, the murine Selp gene is uniquely responsive to inflammatory cytokines like tumor necrosis factor-α and interleukin-1β because it has binding sites for multiple transcription factors, including NF-κB (nuclear factor kappa-light-chain-enhancer of activated B cells) and ATF-2 (activating transcription factor 2).36 Because our understanding regarding the relevance of selectins largely comes from murine disease models where P-selectin has an exaggerated role, it is possible that current literature underestimates the contributions of E-selectin to human ailments. Finally, the physiological E-selectin ligands in humans are unknown, and their definition is clinically significant as it can identify new anti-inflammatory drug targets.

This work suggests that the GSLs are likely to be critical E-selectin ligands specific to human leukocytes. Under the most stringent conditions, the demonstration that a particular molecule is a functional selectin-ligand requires that it satisfy 4 criteria (modified from Varki21). First, the ligand should be expressed in the right place at the right time. This is satisfied by the GSLs because they are expressed on the human myeloid cell surface and they bear various sialofucosylated glycans.18,20,25

Second, it should recognize selectins with high affinity, as demonstrated previously for E-selectin–GSL interactions.1820 These investigators extracted a group of relatively low abundance, sialofucosylated GSLs collectively termed myeloglycans from both primary human neutrophils and HL-60s. These monosialogangliosides contained extended 5 to 6 lactosamine repeat units, terminal α2,3-sialylation, and multiple internal fucose residues. When reconstituted as a substrate in parallel-plate flow chambers, the myeloglycans supported the rolling of E-selectin expressing Chinese Hamster Ovary cells under flow.

Third, the deletion of these molecules in relevant cell types should alter selectin-mediated cell adhesion. In this regard, although previous studies used small molecule inhibitors to reduce leukocyte GSL content,20 adhesion assays under fluid shear could not be performed after pharmacological treatment because human peripheral blood neutrophils are short-lived and the cells adopted an irregular shape when cultured. This limitation is addressed here using precise gene silencing and genome editing to reduce UGCG activity in human myeloid cells. Although the UGCG mRNA levels and enzyme activity were reduced by 93% and 80%, respectively, in the knockdowns, enzyme activity was abolished in the knockouts. Similar changes in cell surface glycan expression and rolling was observed using both the knockouts and knockdowns, suggesting that the findings reported here are robust. Similar observations were also made with different cell types, including WT HL-60s, differentiated HL-60s, and primary neutrophils derived from hHSCs.

With respect to the detailed functional role of the GSLs, UGCG disruption did not affect the leukocyte tethering rates on E-selectin bearing–stimulated HUVECs under flow. However, although WT HL-60s displayed continuous slow rolling, both the knockouts and knockdowns displayed skipping motion that was characterized by high intermittent instantaneous velocities. Here, leukocytes released from the stimulated HUVEC monolayer were recaptured downstream in a fraction of the cases. The unstable motion of the UGCG-deficient cells resulted in ≈50% to 60% reduction in the number of rolling and adherent cells on the HUVEC monolayer. Median rolling velocity was also increased by 2-fold. These data are consistent with the notion that the GSLs promote slow rolling on the endothelium because they reside closer to the cell membrane compared with the glycoproteins which extend further away. Such slow rolling may promote the transition of tethered and rolling cells to firm arrest.

The fourth and most stringent requirement to demonstrate that a given molecule is a functional selectin ligand is that function blocking mAbs or other specific reagents that can perturb one molecular interaction without altering other bystanders must be developed. To date, the only selectin-ligand to fully satisfy this criterion is PSGL-1 for which function blocking mAbs are available.37 To determine whether similar specificity can be established for the UGCG knockout/down approach, we performed both flow cytometry with functionally relevant mAbs/lectins and comprehensive glycomic profiling of N-/O-glycans. Here, the glycans in the WT and KO cells were qualitatively identical, albeit with some quantitative differences. First, although PHA-L lectin binding did not show a change in the pattern of N-glycan branching, glycomics profiling revealed an increased prevalence of bisected N-glycans. Based on quantitative real-time polymerase chain reaction, this change can be explained by a decrease in the mRNA levels of the branching GlcNActransferases (GnT-IV and -V) on UGCG disruption, which may account for the apparent increase in activity of the bisecting enzyme, GnT-III (Figure IX in the online-only Data Supplement). Second, knocking out UGCG reduced the binding of 2 mAbs that bind sLeX and related epitopes on O-glycans, N-glycans and GSLs, HECA-4523840 (70% reduction), and CSLEX-141 (35% reduction). A reduction in N- and O-linked sialoglycan expression was also noted using glycomics profiling. Thus, alterations in the glycolipid pathway have systems-level effects that perturb the output of other apparently unrelated pathways. Additional studies are necessary to more precisely define the regulation of these glycosylation gene regulatory networks.42 This may explain why the perturbation of glycolipid biosynthesis results in a more profound reduction in transendothelial migration compared with the blockade of E-selectin-binding function.

To more precisely determine whether the reduced cell adhesion properties of the UGCG-deficient cells are directly caused by the contribution of the GSLs to leukocyte adhesion or whether they are an indirect effect because of changes in O-/N-glycans, we created additional CRISPR-Cas9 HL-60-KO variants where the biosynthesis of O- and N-glycans was abrogated by knocking out the T-synthase chaperone COSMC and GlcNAcT-I gene MGAT1 (G. Stolfa et al, in preparation). These dual knockouts lacking O- and N-glycans displayed reduced cell surface sLeX expression. Although these cells were not recruited onto E-selectin substrates efficiently under continuous fluid shear, they could sustain cell rolling if captured under static conditions before a ramped increase in applied shear. Taken together with the UGCG-KO data presented in the current article, these observations support a role for the GSLs in sustaining the slow rolling of human myeloid cells. Because of this, the abrogation of GSL biosynthesis reduced the transition of the rolling cells to firm arrest and subsequent extravasation.

In conclusion, human genetic perturbation studies in the current article show that the GSLs are likely to function as E-selectin ligands that facilitate the transition of rolling human myeloid cells to firm arrest. It would be interesting to determine whether these same macromolecules also contribute to other selectin-mediated cell adhesion processes in humans, like hematopoietic stem cell homing, lymphocyte adhesion, and tumor metastasis.

Nonstandard Abbreviations and Acronyms


glycosphingolipids or glycolipids




P-selectin glycoprotein ligand-1


sialyl Lewis-X


UDP-Glucose Ceramide Glucosyltransferase


The online-only Data Supplement is available with this article at

Correspondence to Sriram Neelamegham, PhD, 906 Furnas Hall, State University of New York, Buffalo, NY 14260. E-mail


  • 1. Zarbock A, Ley K, McEver RP, Hidalgo A. Leukocyte ligands for endothelial selectins: specialized glycoconjugates that mediate rolling and signaling under flow.Blood. 2011; 118:6743–6751. doi: 10.1182/blood-2011-07-343566.Google Scholar
  • 2. Vestweber D, Blanks JE. Mechanisms that regulate the function of the selectins and their ligands.Physiol Rev. 1999; 79:181–213.Google Scholar
  • 3. Mondal N, Buffone A, Neelamegham S. Distinct glycosyltransferases synthesize E-selectin ligands in human vs. mouse leukocytes.Cell Adh Migr. 2013; 7:288–292. doi: 10.4161/cam.24714.Google Scholar
  • 4. Beauharnois ME, Lindquist KC, Marathe D, Vanderslice P, Xia J, Matta KL, Neelamegham S. Affinity and kinetics of sialyl Lewis-X and core-2 based oligosaccharides binding to L- and P-selectin.Biochemistry. 2005; 44:9507–9519. doi: 10.1021/bi0507130.Google Scholar
  • 5. Wild MK, Huang MC, Schulze-Horsel U, van der Merwe PA, Vestweber D. Affinity, kinetics, and thermodynamics of E-selectin binding to E-selectin ligand-1.J Biol Chem. 2001; 276:31602–31612. doi: 10.1074/jbc.M104844200.Google Scholar
  • 6. Mehta P, Cummings RD, McEver RP. Affinity and kinetic analysis of P-selectin binding to P-selectin glycoprotein ligand-1.J Biol Chem. 1998; 273:32506–32513.Google Scholar
  • 7. Buffone A, Mondal N, Gupta R, McHugh KP, Lau JT, Neelamegham S. Silencing α1,3-fucosyltransferases in human leukocytes reveals a role for FUT9 enzyme during E-selectin-mediated cell adhesion.J Biol Chem. 2013; 288:1620–1633. doi: 10.1074/jbc.M112.400929.Google Scholar
  • 8. Xia L, Ramachandran V, McDaniel JM, Nguyen KN, Cummings RD, McEver RP. N-terminal residues in murine P-selectin glycoprotein ligand-1 required for binding to murine P-selectin.Blood. 2003; 101:552–559. doi: 10.1182/blood-2001-11-0036.Google Scholar
  • 9. Sako D, Comess KM, Barone KM, Camphausen RT, Cumming DA, Shaw GD. A sulfated peptide segment at the amino terminus of PSGL-1 is critical for P-selectin binding.Cell. 1995; 83:323–331.Google Scholar
  • 10. Wilkins PP, McEver RP, Cummings RD. Structures of the O-glycans on P-selectin glycoprotein ligand-1 from HL-60 cells.J Biol Chem. 1996; 271:18732–18742.Google Scholar
  • 11. Lo CY, Antonopoulos A, Gupta R, Qu J, Dell A, Haslam SM, Neelamegham S. Competition between core-2 GlcNAc-transferase and ST6GalNAc-transferase regulates the synthesis of the leukocyte selectin ligand on human P-selectin glycoprotein ligand-1.J Biol Chem. 2013; 288:13974–13987. doi: 10.1074/jbc.M113.463653.Google Scholar
  • 12. Bochner BS, Sterbinsky SA, Bickel CA, Werfel S, Wein M, Newman W. Differences between human eosinophils and neutrophils in the function and expression of sialic acid-containing counterligands for E-selectin.J Immunol. 1994; 152:774–782.Google Scholar
  • 13. Larsen GR, Sako D, Ahern TJ, Shaffer M, Erban J, Sajer SA, Gibson RM, Wagner DD, Furie BC, Furie B. P-selectin and E-selectin. Distinct but overlapping leukocyte ligand specificities.J Biol Chem. 1992; 267:11104–11110.Google Scholar
  • 14. Hidalgo A, Peired AJ, Wild MK, Vestweber D, Frenette PS. Complete identification of E-selectin ligands on neutrophils reveals distinct functions of PSGL-1, ESL-1, and CD44.Immunity. 2007; 26:477–489. doi: 10.1016/j.immuni.2007.03.011.Google Scholar
  • 15. Levinovitz A, Mühlhoff J, Isenmann S, Vestweber D. Identification of a glycoprotein ligand for E-selectin on mouse myeloid cells.J Cell Biol. 1993; 121:449–459.Google Scholar
  • 16. Merzaban JS, Burdick MM, Gadhoum SZ, Dagia NM, Chu JT, Fuhlbrigge RC, Sackstein R. Analysis of glycoprotein E-selectin ligands on human and mouse marrow cells enriched for hematopoietic stem/progenitor cells.Blood. 2011; 118:1774–1783. doi: 10.1182/blood-2010-11-320705.Google Scholar
  • 17. Tiemeyer M, Swiedler SJ, Ishihara M, Moreland M, Schweingruber H, Hirtzer P, Brandley BK. Carbohydrate ligands for endothelial-leukocyte adhesion molecule 1.Proc Natl Acad Sci U S A. 1991; 88:1138–1142.Google Scholar
  • 18. Stroud MR, Handa K, Salyan ME, Ito K, Levery SB, Hakomori S, Reinhold BB, Reinhold VN. Monosialogangliosides of human myelogenous leukemia HL60 cells and normal human leukocytes. 2. Characterization of E-selectin binding fractions, and structural requirements for physiological binding to E-selectin.Biochemistry. 1996; 35:770–778. doi: 10.1021/bi952461g.Google Scholar
  • 19. Handa K, Stroud MR, Hakomori S. Sialosyl-fucosyl Poly-LacNAc without the sialosyl-Lex epitope as the physiological myeloid cell ligand in E-selectin-dependent adhesion: studies under static and dynamic flow conditions.Biochemistry. 1997; 36:12412–12420. doi: 10.1021/bi971181t.Google Scholar
  • 20. Nimrichter L, Burdick MM, Aoki K, Laroy W, Fierro MA, Hudson SA, Von Seggern CE, Cotter RJ, Bochner BS, Tiemeyer M, Konstantopoulos K, Schnaar RL. E-selectin receptors on human leukocytes.Blood. 2008; 112:3744–3752. doi: 10.1182/blood-2008-04-149641.Google Scholar
  • 21. Varki A. Selectin ligands: will the real ones please stand up?J Clin Invest. 1997; 99:158–162. doi: 10.1172/JCI119142.Google Scholar
  • 22. Seito N, Yamashita T, Tsukuda Y, Matsui Y, Urita A, Onodera T, Mizutani T, Haga H, Fujitani N, Shinohara Y, Minami A, Iwasaki N. Interruption of glycosphingolipid synthesis enhances osteoarthritis development in mice.Arthritis Rheum. 2012; 64:2579–2588. doi: 10.1002/art.34463.Google Scholar
  • 23. Ichikawa S, Hirabayashi Y. Glucosylceramide synthase and glycosphingolipid synthesis.Trends Cell Biol. 1998; 8:198–202.Google Scholar
  • 24. Symington FW, Hedges DL, Hakomori S. Glycolipid antigens of human polymorphonuclear neutrophils and the inducible HL-60 myeloid leukemia line.J Immunol. 1985; 134:2498–2506.Google Scholar
  • 25. Mondal N, Buffone A, Stolfa G, Antonopoulos A, Lau JT, Haslam SM, Dell A, Neelamegham S. ST3Gal-4 is the primary sialyltransferase regulating the synthesis of E-, P-, and L-selectin ligands on human myeloid leukocytes.Blood. 2015; 125:687–696. doi: 10.1182/blood-2014-07-588590.Google Scholar
  • 26. Zhang Y, Neelamegham S. Estimating the efficiency of cell capture and arrest in flow chambers: study of neutrophil binding via E-selectin and ICAM-1.Biophys J. 2002; 83:1934–1952. doi: 10.1016/S0006-3495(02)73956-8.Google Scholar
  • 27. Ichikawa S, Hirabayashi Y. Genetic approaches for studies of glycolipid synthetic enzymes.Methods Enzymol. 2000; 311:303–318.Google Scholar
  • 28. Cong L, Ran FA, Cox D, Lin S, Barretto R, Habib N, Hsu PD, Wu X, Jiang W, Marraffini LA, Zhang F. Multiplex genome engineering using CRISPR/Cas systems.Science. 2013; 339:819–823. doi: 10.1126/science.1231143.Google Scholar
  • 29. Liang JC, Ning Y, Wang RY, Padilla-Nash HM, Schröck E, Soenksen D, Nagarajan L, Ried T. Spectral karyotypic study of the HL-60 cell line: detection of complex rearrangements involving chromosomes 5, 7, and 16 and delineation of critical region of deletion on 5q31.1.Cancer Genet Cytogenet. 1999; 113:105–109.Google Scholar
  • 30. Atilla-Gokcumen GE, Bedigian AV, Sasse S, Eggert US. Inhibition of glycosphingolipid biosynthesis induces cytokinesis failure.J Am Chem Soc. 2011; 133:10010–10013. doi: 10.1021/ja202804b.Google Scholar
  • 31. Atilla-Gokcumen GE, Muro E, Relat-Goberna J, Sasse S, Bedigian A, Coughlin ML, Garcia-Manyes S, Eggert US. Dividing cells regulate their lipid composition and localization.Cell. 2014; 156:428–439. doi: 10.1016/j.cell.2013.12.015.Google Scholar
  • 32. Macher BA, Buehler J, Scudder P, Knapp W, Feizi T. A novel carbohydrate, differentiation antigen on fucogangliosides of human myeloid cells recognized by monoclonal antibody VIM-2.J Biol Chem. 1988; 263:10186–10191.Google Scholar
  • 33. Smith CW, Rothlein R, Hughes BJ, Mariscalco MM, Rudloff HE, Schmalstieg FC, Anderson DC. Recognition of an endothelial determinant for CD 18-dependent human neutrophil adherence and transendothelial migration.J Clin Invest. 1988; 82:1746–1756. doi: 10.1172/JCI113788.Google Scholar
  • 34. Kunkel EJ, Ley K. Distinct phenotype of E-selectin-deficient mice. E-selectin is required for slow leukocyte rolling in vivo.Circ Res. 1996; 79:1196–1204.Google Scholar
  • 35. Puri KD, Finger EB, Springer TA. The faster kinetics of L-selectin than of E-selectin and P-selectin rolling at comparable binding strength.J Immunol. 1997; 158:405–413.Google Scholar
  • 36. Liu Z, Miner JJ, Yago T, Yao L, Lupu F, Xia L, McEver RP. Differential regulation of human and murine P-selectin expression and function in vivo.J Exp Med. 2010; 207:2975–2987. doi: 10.1084/jem.20101545.Google Scholar
  • 37. Moore KL, Patel KD, Bruehl RE, Li F, Johnson DA, Lichenstein HS, Cummings RD, Bainton DF, McEver RP. P-selectin glycoprotein ligand-1 mediates rolling of human neutrophils on P-selectin.J Cell Biol. 1995; 128:661–671.Google Scholar
  • 38. Ohmori K, Fukui F, Kiso M, Imai T, Yoshie O, Hasegawa H, Matsushima K, Kannagi R. Identification of cutaneous lymphocyte-associated antigen as sialyl 6-sulfo Lewis X, a selectin ligand expressed on a subset of skin-homing helper memory T cells.Blood. 2006; 107:3197–3204. doi: 10.1182/blood-2005-05-2185.Google Scholar
  • 39. Picker LJ, Michie SA, Rott LS, Butcher EC. A unique phenotype of skin-associated lymphocytes in humans. Preferential expression of the HECA-452 epitope by benign and malignant T cells at cutaneous sites.Am J Pathol. 1990; 136:1053–1068.Google Scholar
  • 40. Wagers AJ, Stoolman LM, Kannagi R, Craig R, Kansas GS. Expression of leukocyte fucosyltransferases regulates binding to E-selectin: relationship to previously implicated carbohydrate epitopes.J Immunol. 1997; 159:1917–1929.Google Scholar
  • 41. Fukushima K, Hirota M, Terasaki PI, Wakisaka A, Togashi H, Chia D, Suyama N, Fukushi Y, Nudelman E, Hakomori S. Characterization of sialosylated Lewisx as a new tumor-associated antigen.Cancer Res. 1984; 44:5279–5285.Google Scholar
  • 42. Liu G, Neelamegham S. Integration of systems glycobiology with bioinformatics toolboxes, glycoinformatics resources, and glycoproteomics data.Wiley Interdiscip Rev Syst Biol Med. 2015; 7:163–181. doi: 10.1002/wsbm.1296.Google Scholar


The binding of specific glycoconjugates expressed on human myeloid cell surfaces to E-selectin on endothelial cells controls the progress of a variety of human inflammatory diseases. The identification of such glycans on human leukocytes is important because inhibitors that block their binding to E-selectin can result in novel anti-inflammatory therapies. Using genetic approaches (lentiviral shRNA and clustered regularly interspaced short palindromic repeats/clustered regularly interspaced short palindromic repeat–associated 9 [CRISPR-Cas9] knockouts), the current article unambiguously demonstrates that sialofucosylated glycosphingolipids expressed on the surface of human, but not murine, myeloid cells can bind E-selectin under physiological fluid shear conditions. Such glycosphingolipid binding to E-selectin likely controls the rate at which human leukocytes of the myeloid lineage transition from rolling interactions on the vessel wall to firm arrest at sites of inflammation. Blocking glycosphingolipid biosynthesis results in leukocytes exhibiting skipping or unstable rolling interactions and reduced diapedesis across the endothelium.