Siglec-8 antibody reduces eosinophils and mast cells in a transgenic mouse model of eosinophilic gastroenteritis (original) (raw)

EG and EoE tissues from patients have elevated numbers of both mast cells and eosinophils. Of the EGIDs, EoE is the most prevalent and well characterized, whereas much less is known about EG and EGE. To gain greater understanding of the biology driving EG, we procured fresh gastric tissue from patients with EG or nondiseased control subjects without a history of GI disease (Supplemental Table 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.126219DS1). Tissues were digested into single-cell suspensions, and immune cells were characterized by flow cytometry (Figure 1A). To confirm that mast cells and eosinophils were correctly identified in our gating strategy, we examined known surface markers for these cells within the individual gating windows. Mast cells (CD45+7AAD–SSChiCD24–CD16–) expressed Siglec-8 and CD117 (Supplemental Figure 1A), and eosinophils (CD45+7AAD–SSChiCD24+CD16–) expressed Siglec-8, CCR3, and CD11b (Supplemental Figure 1A). To further validate the gating strategy, we sorted mast cells and eosinophils from their respective windows, followed by May-Grunwald Giemsa and H&E staining, respectively. Cells sorted from the mast cell window routinely contained metachromatic granules that were characteristic of mast cells (Supplemental Figure 1B). Similarly, more than 95% of cells sorted from the eosinophil window displayed a bilobed nucleus and eosin-pink granules that resembled human eosinophils (Supplemental Figure 1B). These data demonstrate that the gating strategy used in Figure 1A correctly identified mast cells and eosinophils in GI tissue.

EG and EoE patient tissues have significantly increased numbers of eosinophFigure 1

EG and EoE patient tissues have significantly increased numbers of eosinophils and mast cells compared with nondiseased control tissue. (A) Flow cytometry gating strategy used to identify immune cells, including eosinophils and mast cells, in human stomach tissue from patients with EG. Percentage of (B) eosinophils (CD45+7AAD–SSChiCD16–CD24+) and (C) mast cells (CD45+7AAD–SSChiCD16–CD24–) present in nondiseased (black) or EG (gray) stomach tissue identified using the gating strategy in A. (D) Percentage of neutrophils, T cells, monocytes, DCs, and other (B cells, NK cells, macrophages, basophils) in nondiseased (black) or EG (gray) stomach tissue using the gating strategy shown in A. Percentage of (E) eosinophils and (F) mast cells present in nondiseased (black) or EoE (gray) esophageal tissue identified using the gating strategy in A. The percentage of cells was derived from the CD45+ viable population. Data are plotted as mean ± SEM for n = 7 nondiseased stomach tissue and n = 4 nondiseased esophageal tissue; n = 4 EG, n = 3 EG + EoE, and n = 3 EoE patients. *P < 0.05; **P < 0.01 by Mann-Whitney U test.

As expected, EG tissues had significantly increased numbers of eosinophils compared with nondiseased tissue (Figure 1B). In addition, we found that mast cells were increased by approximately 5-fold in EG tissue compared with nondiseased tissue (Figure 1C). These data were confirmed in EG tissue using additional flow cytometry surface markers for mast cells and eosinophils (Supplemental Figure 1, C–E). Interestingly, mast cells were elevated to a similar extent as eosinophils in EG patient tissues (8.9% vs. 9.3% of all CD45+ cells, respectively). In contrast, the percentage of neutrophils and monocytes was reduced in EG tissue compared with nondiseased control tissue, whereas other immune cells, including T cells, remained unchanged (Figure 1D).

To determine whether these observations could be extended to other EGIDs, we processed and characterized esophageal tissue from patients with EoE. Consistent with findings in EG tissue, as well as previously published findings, mast cells and eosinophils were significantly increased to a similar extent compared with nondiseased esophageal tissue (Figure 1, E and F, and refs. 1216). These data demonstrate that both eosinophils and mast cells are markedly and proportionally elevated in both EG and EoE patient tissue and support our flow cytometry–based approach to quantitatively assess immune cells in fresh human tissue.

Eosinophils and mast cells in EG and EoE tissue are in an activated state. To characterize the state of activation of eosinophils and mast cells from EG tissue, we examined the expression of surface markers associated with activation by flow cytometry (Figure 2, A and B, and Supplemental Figure 2A). As anticipated, Siglec-8 was selectively expressed on mast cells and eosinophils from both EG and nondiseased stomach tissues (Figure 2, C and D, and Supplemental Figure 2B). In contrast with Siglec-8, IL-5Rα was minimally expressed on EG and nondiseased tissue eosinophils and mast cells (Figure 2C and Supplemental Figure 2C). In addition, tissue eosinophils from patients with EG displayed significantly higher expression of the activation markers, CD11b and CD49d, compared with nondiseased tissue eosinophils, consistent with an increased activation state (Figure 2C and refs. 1719). Moreover, mast cells from patients with EG displayed significantly increased expression of the degranulation and activation markers CD63 (LAMP3) and CD107a (LAMP1), suggesting an activated and degranulating state (Figure 2D and ref. 20). Consistent with atopy and high serum IgE levels reported for patients with EG (21), EG tissue mast cells also displayed significantly higher levels of surface IgE and FcεRI (Figure 2D and Supplemental Figure 2D).

Mast cells and eosinophils from EG and EoE patient tissues are highly activFigure 2

Mast cells and eosinophils from EG and EoE patient tissues are highly activated compared with nondiseased control cells. (A) Dot plot of eosinophils in EG patient tissue identified by CD45+7AAD–CD117–CD16–CCR3+SSChi cells. Histogram of EG eosinophils labeled for analysis of surface expression of Siglec-8, IL-5Rα, CD11b, or CD49d or a fluorescence minus 1 (FMO) negative control (gray). (B) Dot plot of mast cells in EG patient tissue identified by CD45+7AAD–CD117+FcεRI+ cells. Histogram of EG mast cells labeled for analysis of surface expression of Siglec-8, CD107a, CD63, or IgE or an FMO negative control (gray). (C) Expression as shown by ΔMFI of Siglec-8, IL-5Rα, CD11b, and CD49d on stomach eosinophils from nondiseased controls (black) or patients with EG (gray). (D) Expression as shown by ΔMFI of the mast cell activation and degranulation markers, CD63, CD107a, and IgE, on stomach mast cells from nondiseased controls (black) or patients with EG (gray). Data are plotted as mean ± SD for n = 5–6 nondiseased stomach tissue; n = 2 EG, and n = 3 EG + EoE. *P < 0.05; **P < 0.01 by Mann-Whitney U test.

Last, to determine whether increased mast cell activation occurred in other EGIDs, we examined the same markers on mast cells from EoE patient tissue. Compared with nondiseased esophageal mast cells, EoE mast cells also had significantly higher expression of CD107a, CD63, and IgE (Supplemental Figure 2E). These data demonstrate that eosinophils and mast cells are found in elevated numbers and in an activated state in EG patient tissue. Furthermore, we show that mast cells from EoE patient tissues are also activated and display increased levels of cell surface markers associated with degranulation, such as CD107a. These findings suggest that increased mast cell activation may be a shared feature across EGIDs and support a role for activated eosinophils and mast cells in EGID disease pathogenesis.

Siglec-8–transgenic mice express Siglec-8 on eosinophils, mast cells, and basophils. Based on the data above demonstrating elevation and activation of eosinophils and mast cells in EG patient tissue, we set out to evaluate the activity of an anti–Siglec-8 mAb in a murine disease model of EG and EGE. Like other members of the CD33-related Siglecs, Siglec-8 appears to have evolved recently, with close homologs found only in some primate species (11, 22). Therefore, to examine the in vivo activity of anti–Siglec-8 mAb, we generated a human Siglec-8–expressing transgenic mouse, distinct from previously generated mice (e.g., ROSA26-Siglec-8–knockin mice) (23, 24). Instead, Siglec-8–transgenic founder mice were generated via the pronuclear microinjection of DNA (Supplemental Figure 3A) as described in Methods. Transmission of the Siglec-8 transgene was successful in 2 of the chimeric founders’ progeny (lines 335 and 307), and the transgenic mouse line selected (line 335) contained a single-copy insertion of the Siglec-8 gene (Supplemental Figure 3, B–D). Siglec-8–transgenic mice did not display differences between males or females compared to wild-type (WT) littermates in body weight, weights of major organs, absolute or relative percentages of blood cell types, blood chemistries, or coagulation (Supplemental Figure 3E). Examination of peripheral blood leukocytes (PBLs) and peritoneal lavage (PL) cells from these mice showed high levels of Siglec-8 on the cell surface of mast cells, eosinophils, and basophils in PBLs and PL (Supplemental Figure 4, A–C), consistent with the selective expression of Siglec-8 on human immune cells (8, 11, 25). Siglec-8 was also found on eosinophils and mast cells in other tissues, including GI, lung, and skin (data not shown), demonstrating expression on both connective tissue and mucosal mast cells. In contrast, Siglec-8 was not detected on lymphocyte, neutrophil, monocyte, or macrophage cell populations (Supplemental Figure 4, A–C).

To determine whether certain functional properties of Siglec-8 were preserved in the transgenic mice, such as ligand-induced internalization (26), we engaged the Siglec-8 receptor in vivo by dosing with AK002-G4 (humanized IgG4 [hIgG4] anti–Siglec-8 mAb; ref. 8) or isotype-matched control mAb and examined the extent of receptor internalization. Compared with isotype control mAb–treated mice, anti–Siglec-8 mAb treatment induced Siglec-8 internalization but not Siglec-F internalization on peripheral blood eosinophils as expected (Supplemental Figure 4, D and E). Finally, insertion of the Siglec-8 gene in these transgenic mice had no effect on endogenous levels of Siglec-F expression on eosinophils (Supplemental Figure 4E and data not shown).

OVA sensitization and intragastric challenge induces EG and EGE in Siglec-8–transgenic mice. Next, we adapted a previously published experimental study design to create a Siglec-8–expressing murine model of EG and EGE (27, 28). Eosinophils were identified in GI tissue preparations as viable CD45+Lin–Ly6G–CD11b+CCR3+Siglec-F+ cells by flow cytometry (Supplemental Figure 5A). GI eosinophils displayed robust Siglec-8 expression, consistent with transgene expression (Supplemental Figure 5A). Systemic sensitization and repeated intragastric challenge with OVA (Figure 3A) induced significant eosinophilic infiltration into the stomach, small intestine, and MLNs compared with sham-treated mice at study takedown on day 39 that resembled EG and EGE (Figure 3, B–G). Consistent with an allergen-specific response, OVA-sensitized and -challenged mice had increased serum levels of OVA-specific IgE and IgG1 compared with sham-treated mice (Figure 3, H and I).

Systemic sensitization and intragastric challenge with OVA induces EG and EFigure 3

Systemic sensitization and intragastric challenge with OVA induces EG and EGE in Siglec-8–transgenic mice. (A) Schematic of EG and EGE mouse model in Siglec-8–transgenic mice. Mice were systemically sensitized with OVA in aluminum hydroxide adjuvant (alum) on days 0 and 14, followed by 6 intragastric OVA challenges starting on day 28 until day 39. IP, intraperitoneal. (B) Representative flow cytometry contour plots of stomach eosinophils and (C) the percentage of eosinophils in the stomach in sham- or OVA-administered mice on day 39 quantified by flow cytometry. (D) Representative flow cytometry contour plots of duodenal eosinophils and (E) the percentage of eosinophils in the duodenum in sham- or OVA-administered mice on day 39 quantified by flow cytometry. (F) Representative flow cytometry contour plots of MLN eosinophils and (G) the percentage of eosinophils in the MLNs in sham- or OVA-administered mice on day 39 quantified by flow cytometry. (H and I) Serum levels of OVA-specific IgE or IgG1 in sham-treated mice (black) or mice sensitized and challenged with OVA (gray) on day 39. The percentage of eosinophils is derived from the CD45+ viable cell population. Data are plotted as mean ± SEM (n = 8–10 mice/group) and are representative of 3 experiments. *P < 0.05; **P < 0.01 by Mann-Whitney U test.

Anti–Siglec-8 mAb reduces OVA-induced eosinophilic infiltration in the stomach and intestinal tissues. To determine the timing for therapeutic treatment with an anti–Siglec-8 antibody in our EG and EGE mouse model, we examined eosinophils in the periphery, stomach, small intestine, and MLNs after the third OVA challenge on day 32 (Supplemental Figure 6). OVA-sensitized and -challenged mice had significantly increased eosinophils in the stomach, MLNs, and blood compared with sham-treated mice on day 32 (Supplemental Figure 6, A–C). Surprisingly, OVA-exposed mice did not have significantly increased eosinophils in the small intestine at this time point (Supplemental Figure 6D), suggesting the stomach and MLNs are the primary sites of eosinophil infiltration while the small intestine is secondary. Having established EG- and EGE-like disease on day 32, we selected this time point for therapeutic dosing of the anti–Siglec-8 mAb.

Therapeutic administration of an anti–Siglec-8 mAb (mIgG2a) on day 32 led to a significant reduction of eosinophils in the stomach, small intestine, and MLNs at study takedown on day 39, compared with isotype control mAb–treated mice (Figure 4, A–D, and Supplemental Figure 7, A–C). Eosinophil numbers in mice treated with anti–Siglec-8 mAb were not completely eliminated in the tissue but rather reduced to levels seen in sham-treated mice. In addition to a reduction in tissue eosinophils, mice treated with anti–Siglec-8 mAb had a significant decrease in peripheral blood eosinophils, consistent with the known antibody-dependent cell-mediated cytotoxicity (ADCC) activity of this antibody isotype and subclass (29), compared with sham- and control mAb–treated mice (Figure 4E and Supplemental Figure 7D). To confirm the OVA-induced intestinal eosinophilia seen by flow cytometry, we quantified the mRNA levels of major basic protein (MBP), an eosinophil granule protein, in the small intestine. Mice treated with anti–Siglec-8 mAb had decreased expression of MBP down to background levels too, similar to the pattern seen with eosinophils in the small intestine compared with control mAb–treated mice (Supplemental Figure 8A). These data demonstrate that anti–Siglec-8 mAb treatment reduced OVA-induced intestinal tissue eosinophilia in our EG and EGE mouse model.

Administration of an anti–Siglec-8 mAb reduces eosinophils in GI tissues inFigure 4

Administration of an anti–Siglec-8 mAb reduces eosinophils in GI tissues in mice with EG and EGE. (A) Representative flow cytometry dot plots of stomach tissue eosinophils in mice treated with sham, OVA and isotype control mAb, or OVA and anti–Siglec-8 mAb. The percentage of eosinophils on day 39 in the (B) stomach, (C) duodenum, (D) MLNs, and (E) peripheral blood quantified by flow cytometry in sham-treated mice (black) or mice sensitized and challenged with OVA and dosed with either an isotype control mAb (gray) or anti–Siglec-8 mAb (blue). The percentage of eosinophils is derived from the CD45+ viable cell population. Data are plotted as mean ± SEM (n = 6–7 mice/group) and are representative of 3 experiments. **P < 0.01 by 1-way ANOVA with Tukey’s multiple-comparisons test.

Anti–Siglec-8 mAb reduces OVA-induced mast cell accumulation in the stomach and intestinal tissues. We also examined mast cell infiltration in the EG and EGE mouse model. Mast cells were identified in GI preparations as viable CD45+Lin–CD117+FcεRI+ cells by flow cytometry and expressed Siglec-8 (Supplemental Figure 5B). As seen with stomach tissue eosinophils, significantly increased mast cells were found in the stomach but not small intestine by day 32 in OVA-administered mice compared with sham-treated mice (Supplemental Figure 6, E and F). On day 39, significantly increased mast cell numbers were seen in the stomach, small intestine, and MLNs in OVA-sensitized and challenged mice compared with sham-treated mice (Figure 5, A–D; black vs. gray). Therapeutic treatment with an anti–Siglec-8 mAb led to a significant reduction in the percentage of mast cells in the stomach and small intestine on day 39 compared with control mAb–treated mice (Figure 5, A–C, and Supplemental Figure 7, E and F), albeit not quite back to baseline levels. Similar effects were seen with MLN mast cells but at very low overall cell counts (Figure 5D and Supplemental Figure 7G). To confirm the differences in OVA-induced mast cell infiltration seen by flow cytometry, we quantified the mRNA levels of mast cell protease 1 (MCPT1) in the small intestine. Mice treated with anti–Siglec-8 mAb had significantly reduced expression of MCPT1 in a pattern similar to changes in mast cell levels, confirming decreased mast cell numbers in the small intestine compared with control mAb–treated mice (Supplemental Figure 8B).

Administration of an anti–Siglec-8 mAb reduces mast cells in GI tissues inFigure 5

Administration of an anti–Siglec-8 mAb reduces mast cells in GI tissues in mice with EG and EGE. (A) Representative flow cytometry dot plots of stomach tissue mast cells in mice treated with sham, OVA and isotype control mAb, or OVA and anti–Siglec-8 mAb. The percentage of mast cells on day 39 in the (B) stomach, (C) duodenum, and (D) MLN quantified by flow cytometry in sham-treated mice (black) or mice sensitized and challenged with OVA and dosed with either an isotype control mAb (gray) or anti–Siglec-8 mAb (blue). The percentage of stomach (E) eosinophils or (F) mast cells on days 32, 34, and 39 in mice treated with sham (black), OVA and isotype control mAb (gray), or OVA and anti–Siglec-8 mAb (blue) quantified by flow cytometry. The percentage of mast cells is derived from the CD45+ viable cell population. Data are plotted as mean ± SEM (n = 6–7 mice/group for BD and n = 4–6 mice/group for E and F) and are representative of 3 experiments. *P < 0.05; **P < 0.01 by 1-way ANOVA with Tukey’s multiple-comparisons test (BD) or 2-tailed t test with Holm-Šídák’s posttest (E and F).

Stomach eosinophils and mast cells are differentially reduced after anti–Siglec-8 mAb treatment. The cell-specific activity of Siglec-8 on eosinophils and mast cells has been well characterized in vitro and ex vivo (810). To evaluate whether the reduction of mast cells and eosinophils in mice treated with anti–Siglec-8 mAb reflected the known activity of Siglec-8 in vivo, we analyzed stomach tissue on day 32 (before mAb treatment), day 34 (2 days after mAb treatment), and day 39 in our EG and EGE mouse model. As was seen previously, OVA-challenged mice displayed elevated mast cells and eosinophils in the stomach on day 32 compared with sham-treated mice (Figure 5, E and F). On day 34, 2 days after mAb treatment, mice dosed with an anti–Siglec-8 mAb had significantly decreased eosinophils in the stomach compared with isotype control mAb–treated mice, whereas mast cells decreased only nominally (Figure 5, E and F). On day 39, both stomach eosinophils and mast cells were significantly reduced in mice treated with anti–Siglec-8 mAb; however, the magnitude of decrease seen with eosinophils was greater than that of mast cells (Figure 5, E and F). These data suggest that anti–Siglec-8 mAb treatment differentially reduces eosinophils and mast cells in the GI tract, consistent with the unique cell-specific activities of Siglec-8 in eosinophils and mast cells.

The faster and more extensive reduction of tissue eosinophils seen in mice treated with anti–Siglec-8 mAb compared with mast cells on days 34 and 39 suggest that Siglec-8 mAb treatment may directly decrease eosinophils in GI tissue. To evaluate this, we collected and cultured dissociated ex vivo stomach tissue from OVA-challenged mice on day 39 overnight in the presence of either an anti–Siglec-8 mAb or isotype control mAb, followed by analysis of eosinophils by flow cytometry. Anti–Siglec-8 mAb treatment of dissociated stomach tissue led to significantly fewer eosinophils compared with isotype control mAb–treated tissue (Supplemental Figure 8, C–E). Similarly, AK002 directly reduced human tissue eosinophils in ex vivo lung tissue (8). These data suggest that Siglec-8 mAb treatment directly reduces GI tissue eosinophils, consistent with the known apoptotic activity of Siglec-8.

Anti–Siglec-8 mAb reduces OVA-induced inflammation in the intestine and serum. Upon activation with IgE and allergen, mast cells and subsequently eosinophils elicit inflammatory allergic effects via production of chemokines that drive a type 2 immune response. To evaluate these responses and effects of anti–Siglec-8 mAb treatment in the EG and EGE mouse model, we quantified the mRNA expression of known mediators implicated in driving type 2 inflammation in the small intestine. We did not observe increased gene expression of the eosinophil-recruiting chemokine, CCL11, in OVA-challenged mice or detectable expression of the canonical Th2 mediators, IL-4, IL-5, and IL-13, in the intestine (Supplemental Figure 9A). However, the expression of CCL17 (TARC), CCL2 (MCP1), and CCL5 (RANTES) were increased upon OVA challenge in the intestine and decreased with anti–Siglec-8 mAb treatment compared with isotype control–treated mice (Figure 6, A–C). Furthermore, we measured MLN weight as a surrogate for intestinal inflammation (30). Mice sensitized and challenged with OVA had increased MLN weight compared with sham-treated mice on day 39, and, consistent with the decreased inflammatory signature in the intestine, mice treated with an anti–Siglec-8 mAb had significantly reduced MLN weights compared with isotype control–treated mice (Supplemental Figure 9B).

Mice treated with anti–Siglec-8 mAb display reduced expression of OVA-inducFigure 6

Mice treated with anti–Siglec-8 mAb display reduced expression of OVA-induced type 2 immune–associated inflammatory cytokines and chemokines in intestinal tissue and serum. Quantitative PCR (qPCR) gene expression analysis of (A) CCL17, (B) CCL2, and (C) CCL5 in the duodenum at day 39 of study in sham-treated mice (black) or mice sensitized and challenged with OVA and dosed with either an isotype control mAb (gray) or anti–Siglec-8 mAb (blue). (D and E) CCL2 and IL-9 levels in serum in sham-treated (black) or OVA-treated (gray) mice on day 28 (before first OVA challenge) and days 32, 34, and 39. (FH) CCL2, IL-9, and CXCL1 levels in serum in mice treated with sham (black), OVA and isotype control mAb (gray), and OVA and anti–Siglec-8 mAb (blue) (n = 5 mice/group). Graphs are plotted as mean ± SEM (n = 6–8 mice/group) and are representative of 3 experiments. *P < 0.05; **P < 0.01 by 1-way ANOVA with Tukey’s multiple-comparisons test (AC) or 2-tailed t test with Holm-Šídák’s posttest (D and E).

To evaluate systemic changes in mice challenged with intragastric OVA, we examined the expression of cytokines and chemokines in the serum throughout the challenge phase on days 28, 32, 34, and 39. Serum levels of known eosinophil chemokines and cytokines, such as CCL11 and IL-5, were similar in OVA-challenged and sham-treated mice (Supplemental Figure 9, C and D). In contrast, the levels of CCL2, IL-9, and CXCL1 increased throughout the challenge phase in mice exposed to OVA (Figure 6, D and E, and Supplemental Figure 9E). Consistent with the reduction of mast cells and eosinophils in GI tissues, serum levels of CCL2, CXCL1, and IL-9 in OVA-challenged mice on day 39 were significantly reduced with anti–Siglec-8 mAb therapeutic treatment (Figure 6, F–H). These data demonstrate that the anti–Siglec-8 mAb reduced the expression of several OVA-induced inflammatory mediators associated with eosinophil- and mast cell–driven inflammation.