Sialylation-dependent regulation of intestinal PMN TEpM in vitro and in vivo. Given the heavily sialylated nature of PMN surface glycoproteins and previous work showing mobilization of sialidases from intracellular granules to the cell surface of activated PMN (29, 30), studies were performed to determine effects of exogenous free Sia on PMN TEpM. Dose response analyses revealed that exposure of human PMN to 5–10 mM Sia reduced N-formyl-L-methionyl-Leucyl-L-phenylalanine–driven (fMLF-driven) migration across T84 intestinal epithelial cell (IEC) monolayers in the physiologically relevant basolateral to apical direction (Figure 1, A and B). At a concentration of 5 mM, Sia reduced detectable PMN numbers in the apical chamber by ≥ 80% compared with 5 mM control sugar galactose (Gal) (Figure 1C). We next determined effects of Sia on PMN migration in a system with no epithelial cells. For these assays, PMN chemotaxis to 100 nM fMLF across collagen-coated transwells was assessed. In contrast to effects observed during TEpM, exposure to 5 mM Sia had no effect on PMN chemotaxis across collagen-coated transwell filters (Figure 1D). These results suggest that Sia specifically interferes with critical PMN-epithelial adhesion interactions during TEpM.

Sialidase inhibition reduces PMN TEpM in vitro and in vivo. (A) T84 intestinal epithelial cells were cultured to confluency as inverted monolayers on 3 μm porous polycarbonate filters (Transwell). (B) Human PMNs incubated with 0.5 mM to 10 mM Sia were placed in the upper chamber of transwell filters and induced to migrate into the bottom chamber in response to 100 nM fMLF. Migration was quantified by MPO assay. Data are expressed as mean ± SEM and were analyzed by 1-way ANOVA followed by Bonferroni post hoc testing (n = 3 independent donors, ***P < 0.001). (C) Human PMN were exposed to 5 mM Sia or 5 mM Gal before assessment of TEpM as described above. Data are expressed as mean ± SEM and were analyzed by 1-way ANOVA followed by Bonferroni post hoc testing (n = 4 independent donors, ****P < 0.0001). (D) Effect of 5 mM Sia or 5 mM Gal on PMN migration across collagen-coated transwells to 100 nM fMLF. Data are expressed as mean ± SEM and were analyzed by 1-way ANOVA followed by Bonferroni post hoc testing (n = 3 independent donors). (E and F) Number of murine PMN recruited into the lumen of proximal colon loops following luminal injection of LTB4 ± 5 mM Sia or 5 mM Gal. Data are mean ± SEM and were analyzed by 1-way ANOVA followed by Bonferroni post hoc testing (n = 2 independent experiments with 4–5 mice per group, *P < 0.05). (G) Human PMNs incubated with 1–5 mM 2-DN, 5 mM KDO, or 5 mM Sia in the upper chamber of transwell filters were induced to migrate into the bottom chamber in response to 100 nM fMLF. Migration was quantified by MPO assay. Data are expressed as mean ± SEM and were analyzed by 1-way ANOVA followed by Bonferroni post hoc testing (n = 3 independent donors, ****P < 0.0001).

In vitro experiments were extended to animal studies to see if Sia had similar effects on TEpM in vivo. In these studies, we used a previously established proximal colon loop model that enables quantitative and spatiotemporal studies of leukocyte trafficking across colonic mucosa in response to luminally administered chemoattractants (31). This in vivo surgical model facilitates assessment of PMN at multiple stages of transmigration across intestinal mucosa, including lamina propria and epithelial-associated PMN as well as those that have reached the colonic lumen. Analysis of in vivo murine PMN migration into the proximal colon in response to a solution of luminally applied (leukotriene B4) LTB4 revealed that coinjection of 5 mM Sia along with LTB4 resulted in a ≥ 60% decrease in the number of PMN reaching the intestinal lumen, relative to mice injected with the control sugar Gal (Figure 1, E and F). Given that observed Sia mediated decreases in PMN migration across intestinal mucosa, we hypothesized that addition of free Sia was likely acting as a competitive inhibitor of sialidase activity, thus preventing removal of terminal Sia residues from the cell surface. To confirm that negative effects on TEpM observed with Sia were mediated by sialidase inhibition, PMN were exposed to the pan sialidase inhibitor N-Acetyl-2,3-dehydro-2-deoxyneuraminic acid (2-DN). As can be seen in Figure 1G, incubation of human PMN with 2-DN inhibited TEpM in a dose-dependent fashion, with 5 mM inhibiting migration by ≥ 75% (P < 0.0001) compared with 5 mM 2-keto-3-deoxyoctonate ammonium salt (KDO), a molecule with similar charge and structure as 2-DN but without sialidase inhibitory activity (28). To determine if decreases in TEpM observed with sialidase inhibition were the result of PMN aggregation, PMN exposed to 5 mM Sia, 2-DN, or relevant controls were examined by light microscopy. Incubation of PMNs with sialidase inhibitors or control sugars did not result in significant levels of aggregation compared with cells incubated with the known agglutinating agent wheat germ agglutinin (WGA) (Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.167151DS1), demonstrating that Sia- or 2-DN–induced inhibition of PMN TEpM in vitro and in vivo is not the result of PMN aggregation. We next determined if inhibition of sialidase activity altered intestinal epithelial barrier function. Exposure of T84 IECs to sialidase inhibitors or controls did not significantly change transepithelial electrical resistance, suggesting that Sia- and 2-DN–mediated decreases in PMN TEpM are not the result of altered epithelial permeability (Supplemental Figure 1B).

Removal of α2-3 Sia from specific PMN glycoproteins facilitates TEpM. Given the reduction in PMN epithelial trafficking mediated by global inhibition of sialidase activity, we analyzed the extent of surface sialylation of human PMN before and after TEpM (Figure 2A). As a common terminating sugar for longer oligosaccharide chains, Sia connects to underlying Gal residues via α2-3 or α2-6 linkages (32, 33). Therefore, we analyzed effects of TEpM on PMN surface expression of α2-3– and α2-6–linked Sia. Interestingly, surface expression of α2-3–linked Sia —as detected by fluorescein isothiocyanate–conjugated (FITC-conjugated) Maackia Amurensis Lectin II (MALII) — decreased by ≥ 40% on postmigrated PMN (Figure 2, B and C). In contrast. a ≥ 60% increase in surface expression of α2-6 Sia — detected by FITC conjugated Sambucus Nigra lectin (SNA) — was observed on PMN that had undergone TEpM (Figure 2, D and E). Data, therefore, demonstrate specific loss of α2-3 sialylation from the surface of PMN that have migrated across IECs. To confirm the requirement for α2-3 sialidase activity during PMN epithelial trafficking, functional effects of either an α2-3 sialidase inhibitor (3’ sialyllactose [3’SL]) or an α2-6–specific sialidase inhibitor (6’SL) were examined (Figure 2F). As can be seen in Figure 2G, inhibition of α2-3 sialidase activity by 3’SL inhibited human PMN TEpM by ≥ 60%, while incubation with 6’SL had no significant effect on PMN epithelial trafficking. Analysis of in vivo migration of murine PMN into the proximal colon in response to a solution of luminally applied LTB4 revealed that coinjection of 5 mM 3’SL along with LTB4 resulted in a ≥ 55% decrease in the number of PMN reaching the intestinal lumen (Figure 2H). In contrast, coincubation of LTB4 with 6’SL had no significant effect on trafficking of murine PMN into the colon. Taken together, results demonstrate the requirement for specific sialidase-mediated removal of α2-3 Sia during PMN trafficking across colonic epithelium in vitro and in vivo.

Sialidase-mediated removal of α2-3 Sia is required during PMN TEpM in vitro and in vivo. (A–E) Levels of surface expression of α2-3– and α2-6–linked Sia were assessed before and after PMN TEpM by flow cytometry using FITC-conjugated MAL II or FITC-conjugated SNA. Data are expressed as mean ± SEM and were analyzed by 1-way ANOVA followed by Bonferroni post hoc testing (n = 3 independent PMN donors, ***P < 0.001). (F and G) Human PMNs incubated with 5 mM 3’SL or 5 mM 6’SL in the upper chamber of transwell filters were induced to migrate into the bottom chamber in response to 100 nM fMLF. Migration was quantified by MPO assay. Data are expressed as mean ± SEM and were analyzed by 1-way ANOVA followed by Bonferroni post hoc testing (n = 4 independent donors, **P < 0.01, ***P < 0.001). (H) Number of murine PMN recruited into the lumen of proximal colon loops in vivo following luminal injection of LTB4 ± 5 mM 3’SL or 5 mM 6’SL. Data are mean ± SEM and were analyzed by 1-way ANOVA followed by Bonferroni post hoc testing (n = 3 independent experiments with 3-5 mice per group, **P < 0.01).

CD11b/CD18 is the major PMN glycoprotein decorated with α2-3 Sia, and removal of Sia facilitates CD11b activation. Given the observation that specific removal of α2-3–linked Sia promotes PMN TEpM, experiments were performed to identify human PMN glycoproteins that are preferentially decorated with α2-3– or α2-6–linked Sia by Western blotting with biotinylated MALII or SNA. Three major α2-3 sialylated glycoproteins were identified with molecular weights ranging from 80 to 180 kDa (Figure 3A). Western blotting also confirmed decreased expression of α2-3 Sia by PMN that had migrated across intestinal epithelial monolayers (migrated lane compared with nonmigrated lane, Figure 3A). In contrast to the limited number of α2-3 sialylated glycoproteins observed, immunoblotting with SNA revealed numerous glycoproteins ranging in molecular weight from 15 to 160 kDa that were decorated with α2-6 sialylation. Furthermore, there was no decrease in levels of α2-6 sialylation observed in transmigrated PMN (Figure 3B).

Sialidase inhibition prevents conformational activation of PMN CD11b/CD18. (A and B) Lysates from PMN before or after TEpM were immunoblotted with biotinylated MALII or SNA. Data shown are representative of PMN from 3 independent donors. (C) α2-3 Sia containing glycoproteins were pulled from human PMN lysates by a MALII column and subjected to tryptic digestion and LC-MS/MS analysis. Table shows accession numbers, protein names, and number of tryptic peptides identified. (D) In total, 10 μg CD11b/CD18 immunopurified from human PMN was immunoblotted with biotinylated MALII or SNA. (E–G) Flow cytometry of human PMN following stimulation with 100 nM fMLF ± 5 mM Sia, Gal, 2-DN, or KDO using FITC-conjugated CBRM1/5. Data are expressed as mean ± SEM and were analyzed by 1-way ANOVA followed by Bonferroni post hoc testing (n = 4 PMN donors, ***P < 0.001, **P < 0.01, *P < 0.05).

Liquid chromatography tandem mass spectrometry (LC-MS/MS) of glycoproteins isolated by MALII affinity chromatography identified the major carrier of α2-3 sialylation in human PMN lysates to be integrin αM (33 tryptic peptides identified) and β2 integrin (32 tryptic peptides identified) (Figure 3C). These glycoproteins represent both heteromeric subunits of the β2 integrin CD11b/CD18. Western blotting of CD11b/CD18 protein immunopurified from human PMN confirmed α2-3 sialylation and α2-6 sialylation of both CD11b and CD18 subunits (Figure 3D).

It has been previously demonstrated that transition of CD11b/CD18 from a bent conformation into an open extended state facilitates high-affinity integrin binding interactions (Figure 3E) (34). Therefore, we determined whether removal of α2-3 Sia plays a role in CD11b/CD18 conformational activation utilizing flow cytometric analyses and an antibody specific for the active or extended form of CD11b/CD18 (CBRM1/5). As can be seen in Figure 3F, exposure of human PMN to 100 nM fMLF resulted in significantly enhanced CD11b activation. Importantly, exposure of PMN to sialidase inhibitors (5 mM Sia or 5 mM 2-DN) prevented fMLF-mediated increases in CD11b activation (Figure 3G). In contrast, incubation of PMN with controls (Gal and KDO) did not interfere with fMLF-mediated increases in CBRM1/5 binding or CD11b activation (Figure 3G). Taken together, these data suggest that sialidase-mediated removal of α2-3 from CD11b/CD18 plays an important role in CD11b/CD18 conformational activation, thus promoting adhesive interactions during PMN TEpM.

Sialylation regulates PMN phagocytosis, degranulation, and superoxide release. Given the important role sialyation plays in PMN TEpM, we examined the effect of inhibiting sialidase activity on other critical CD11b/CD18-mediated PMN inflammatory functions. Effects of sialidase inhibition on PMN degranulation in response to potent stimuli latrunculin B (LaB) and fMLF were evaluated. As expected, and shown in Figure 4, A and B, incubation with 1.25 μM LaB followed by 5 μM fMLF resulted in degranulation, as detected by increased surface expression of markers of primary (CD63) and secondary (CD66b) granules on the surface of human PMN. Importantly, coincubation of PMN with 5 mM Sia or 2-DN significantly reduced LaB and fMLF induced degranulation. In contrast, incubation of PMN with the control sugar Gal or KDO did not reduce degranulation induced by LaB/fMLF treatment. Significant LaB/fMLF-mediated increases in surface expression of markers of primary granules (CD63) and secondary granules (CD15) were also observed in murine PMN (Figure 4, C and D). As was observed for human PMN, coincubation of murine PMN with Sia or 2-DN (Figure 4, C and D) — but not controls (Gal or KDO) — resulted in decreased surface expression of CD63 and CD15 after LaB/fMLF stimulation (Figure 4, C and D). Taken together, these data demonstrate robust sialylation-dependent regulation of human and murine PMN degranulation responses.

Sialidase inhibition prevents degranulation and ROS release in human and murine PMN. (A and B) Human PMN were exposed to 5 mM Sia, 5 mM Gal, 5 mM 2-DN, or 5 mM KDO for 30 minutes at 37°C, followed by stimulation with 1.25 μM LaB and 5 μM fMLF to induce degranulation before assessment of surface expression of CD66b and CD63 by flow cytometry. Data shown are fold-change in mean fluorescence intensity (MFI) comparing treatment with sialidase inhibitors against relevant control (Gal or KDO). Data are expressed as mean ± SEM and were analyzed by 1-way ANOVA followed by Bonferroni post hoc testing (n = 4–6 PMN donors, ***P < 0.001, ****P < 0.0001). (C and D) Murine PMN were exposed to 5 mM Sia, 5 mM Gal, 5 mM 2-DN, or 5 mM KDO for 30 minutes at 37°C followed by stimulation with 1.25 μM LaB and 10 μM fMLF to induce degranulation before assessment of surface expression of CD66b and CD63 by flow cytometry. Data shown are fold-change in mean fluorescence intensity (MFI) comparing treatment with sialidase inhibitors against relevant control (Gal or KDO). Data are expressed as mean ± SEM and were analyzed by 1-way ANOVA followed by Bonferroni post hoc testing for PMN isolated from 3–5 mice (**P < 0.01, ***P < 0.001). (E) Human PMN incubated with 5 mM Gal, 5 mM Sia, 5 mM 2-DN, or 5 mM KDO were exposed to 100 μM cytochrome C. Reduction of cytochrome C in response to 500 nM fMLF was measured by quantifying changes in absorbance at 550 nm at 2, 5, 10, 15, 20, and 60 minutes. Data are fold change in absorbance relative to time 0, are expressed as mean ± SEM, and were analyzed by 1-way ANOVA followed by Bonferroni post hoc testing (n = 3 independent human PMN donors, *P < 0.05, **P < 0.01, ***P < 0.001). (F) Murine PMN incubated with 5 mM Gal, 5 mM Sia, 5 mM 2-DN, or 5 mM KDO were exposed to 100 μM cytochrome C. Reduction of cytochrome C in response to 1 μM fMLF was measured by quantifying changes in absorbance at 550 nm at 2, 5, 10, 15, 20, and 60 minutes. Data are fold change in absorbance relative to time 0; data are expressed as mean ± SEM and were analyzed by 1-way ANOVA followed by Bonferroni post hoc testing (n = 3 mice, *P < 0.05, **P < 0.01, ***P < 0.001).

PMN oxidative burst responses, while crucial for host defense against invading microbes, are also implicated in PMN-associated tissue damage in numerous inflammatory disorders. As can be seen in Figure 4, E and F, exposure of PMN to the bacterial peptide fMLF resulted in robust superoxide generation (as measured by quantifying reduction of cytochrome C) between 5 and 60 minutes of stimulation. Importantly, coincubation of human or murine PMN with 5 mM Sia or 2-DN significantly decreased fMLF-induced superoxide release at all time points measured between 5 and 60 minutes relative to indicated controls (Figure 4, E and F). In addition to degranulation and reactive oxygen species (ROS) production, phagocytosis is an essential tool in the PMN antimicrobial arsenal. Therefore, we determined effects of sialidase inhibition on PMN phagocytosis. In contrast to inhibitory effects observed for degranulation and superoxide release, flow cytometric analyses demonstrated that exposure of human PMN to 5 mM Sia or 2-DN significantly increased PMN phagocytosis of fluosphere beads relative to relevant controls (Supplemental Figure 2A). A similar increase in PMN phagocytosis of fluorescent beads was observed for murine PMN incubated with 5 mM Sia or 2-DN relative to indicated controls (Supplemental Figure 2B). Given the delayed clearance of PMN in inflamed mucosal tissues under pathologic conditions, we next examined effects of sialidase inhibition on PMN apoptosis. Incubation of human PMN (Supplemental Figure 2C) or murine PMN (Supplemental Figure 2D) with 5 mM Sia had no significant effect on PMN apoptosis levels, as measured by flow cytometry quantification of annexin V+ cells. Taken together, these data suggest that sialidase-dependent removal of surface Sia residues is a key driver regulating multiple PMN inflammatory effector functions, including epithelial transmigration, degranulation, and superoxide release.

Sialylation regulates signaling downstream of β2 integrin in human and murine PMNs. It is well appreciated that activated CD11b/CD18 mediates PMN functions through outside-in signaling via Syk (35). We thus investigated effects of sialyation on Syk signaling in PMN. For these experiments, fMLF-stimulated human and murine PMN were treated with sialidase inhibitors for varying time points, followed by Western blot and probing for changes in Syk activity using antibodies against well-characterized inhibitory (Thr323) and activating (Thr525/526) phosphorylation sites on Syk (Figure 5A) (36, 37). Importantly, coincubation of fMLF-stimulated human PMN with 5 mM Sia or 2-DN resulted in a significant increase in phosphorylation of the inhibitory SykTyr323 site between 15 and 30 minutes (Figure 5, B, D, and E). In contrast, no phosphorylation of this inhibitory site was observed in PMN stimulated with fMLF plus controls (Gal or KDO). In addition to phosphorylation at SykTyr323, exposure of PMN to sialidase inhibitors significantly decreased fMLF-mediated phosphorylation of SykTyr525/526 activation sites between 15 and 60 minutes (Figure 5, B–E). In contrast, consistent increases in SykThr525/526 phosphorylation were observed in fMLF-stimulated human PMN exposed to Gal or KDO controls with maximal 4-fold increases in Syk activation observed at 60 minutes. It has been previously reported that p38 MAPK signaling is activated downstream of Syk in PMNs (35). We observed that fMLF-mediated activation of p38 MAPK at Thr180 and Tyr182 in human PMN is significantly decreased by sialidase inhibitors (Sia or 2-DN) between 30 and 60 minutes of stimulation (Figure 5, B–E). Consistent with what was observed for human PMN, inhibition of sialidase activity significantly decreased fMLF-mediated activation of Syk (increased SykTyr323/decreased SykTyr525/526) in murine PMN at time points between 15 and 30 minutes (Figure 6, A–D). In contrast, robust Syk activation was observed in murine PMN stimulated with fMLF in the presence of controls (Gal or KDO). Sialidase inhibition in murine PMN also resulted in significant decreases in fMLF-induced p38 MAPK activation at time points between 30 and 60 minutes (Figure 6, A–D). These results demonstrate Sia-dependent regulation of Syk-p38 MAPK signaling in activated human and murine PMN. Taken together, these data demonstrate that desialylation activates PMN CD11b/CD18, which signals through Syk and p38 MAPK to upregulate TEpM, superoxide release and degranulation responses (Figure 7).

Sialidase inhibition blocks fMLF-mediated Syk and downstream p38 MAPK activation in human PMN. (A) Model showing activating and inhibitory phosphorylation sites on CD45 and downstream regulation of p38 MAPK. (B and C) Lysates from human PMN stimulated over a 60-minute time course with 100 nM fMLF plus 5 mM Gal or 5 mM Sia (B), or 5 mM KDO or 5 mM 2-DN (C), were immunoblotted for total Syk, SykTyr323, SykTyr525/526, total p38 MAPK, or p38Thr180/Tyr182. Blots shown are representative of 3 independent PMN donors. (D and E) Densitometry analyses of phosphorylation of Syk and p38 MAPK following 5, 15, 30, 45, and 60 minutes of stimulation normalized to band intensity at time 0. Data are expressed as mean ± SEM from 3 independent human PMN donors and were analyzed by 1-way ANOVA followed by Bonferroni post hoc testing. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Sialidase inhibition blocks fMLF mediated Syk and downstream p38 MAPK activation in murine PMN. (A and B) Lysates from murine PMN stimulated over a 60-minute time course with 200 nM fMLF plus 5 mM Gal or 5mM Sia (A), or 5 mM KDO or 5 mM 2-DN (B), were immunoblotted for total Syk, SykTyr323, SykTyr525/526, total p38 or p38Thr180/Tyr182. Blots shown are representative of PMN isolated from 3 to 4 mice. (C and D) Densitometry analyses of phosphorylation of Syk and p38 MAPK following 5, 15, 30, 45, and 60 minutes of stimulation normalized to band intensity at time 0. Data are expressed as mean ± SEM and were analyzed by 1-way ANOVA followed by Bonferroni post hoc testing (n = 3–4 murine PMN donors, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).

PMN activation model. Model showing how PMN activation results in surface mobilization of intracellular sialidases resulting in desialylation and activation of CD11b/CD18, which signals through Syk and p38 MAPK to drive PMN inflammatory effector functions including TEpM, degranulation and ROS release.