Modulating endothelial cells with EGFL7 to diminish aGVHD after allogeneic bone marrow transplantation in mice

Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is used to treat patients with high-risk/refractory hematological malignancies and/or blood disorders. Unfortunately, acute graft-versus-host disease (aGVHD) is the principal complication of allo-HSCT.¹  aGVHD is mediated by alloreactive donor T lymphocytes and occurs in response to differences in major and/or minor histocompatibility antigens expressed by recipient cells.^2,3  Despite the use of standard GVHD prophylaxis regimens such as calcineurin inhibitors and other agents, 30% to 75% of allo-HSCT patients eventually develop aGVHD.^4,5  Patients with severe and/or steroid-refractory aGVHD have a poor long-term prognosis and a mortality rate reaching 70% to 80%.^6-9  Thus, alternative treatments for controlling excessive aGVHD are needed.

Epidermal growth factor-like domain 7 (EGFL7) is an ∼30-kDa secreted protein important for angiogenesis and neurogenesis.^9-11  EGFL7 inhibits endothelial cell (EC) activation by proinflammatory cytokines through a negative feedback loop.¹²  In a murine model of multiple sclerosis, EGFL7 reduced neuroinflammation by binding avβ3 integrins and preventing T-cell infiltration.¹³  Thus, unlike immunosuppressive therapies that target immune cell function, EGFL7 acts primarily on blood vessels to reduce immune cell infiltration. Given the importance of lymphocyte homing in aGVHD pathology,^14,15  we hypothesized that blocking EC activation after allo-HSCT could reduce GVHD severity. This study is the first to demonstrate EGFL7 as a potential novel treatment of allo-HSCT patients that develop aGVHD.

C57BL/6.SJL (cat. no. 002014), B6D2F1 (cat. no. 100006), BALB/c (cat. no. 000651), and C57Bl/6J (cat. no. 000664) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). All animals were housed at Maisonneuve-Rosemont Hospital and The Ohio State University animal facilities. Animal studies were performed in accordance with the Maisonneuve-Rosemont Hospital Animal Care Committee and Institutional Animal Care and Use Committee (The Ohio State University). Recombinant human EGFL7 (rEGFL7) was purchased from Peprotech (Rocky Hill, NJ).

B6D2F1 mice were irradiated (2 × 600 rad, 3 hours apart), and 10⁷ bone marrow (BM) cells from C57BL/6 donor mice (H-2^b) were injected IV into lethally irradiated B6D2F1 recipients (H-2^b/d) along with 3 × 10⁶ T cells from B6.SJL (allogeneic) or B6D2F1 (syngeneic) mice. T cells were isolated using the T-cell enrichment kit (Stemcell Technologies). Mice were monitored daily and scored thrice a week for clinical severity of acute GVHD using a scoring system modified from Cooke et al.¹⁶  Vehicle (phosphate-buffered saline [PBS]) or rEGFL7 (10 µg) was administered intraperitoneally daily to mice, starting when signs of GVHD appear (∼17 to 21 days post–stem cell transplantation). Mice with a clinical score of 7 or more were considered very sick and euthanized according to approved animal protocols.

All other methods are described in the supplemental data.

Current therapies for aGVHD rely largely on the inhibition of immune cell activation and effector functions.¹⁷  Here, we evaluated how EGFL7, an inhibitor of EC activation, can reduce aGVHD. We used a parent into F1 (B6→B6D2F1) mouse model to evaluate the therapeutic benefit of EGFL7 treatment (supplemental Figure 1A). Mice at day +21 post-HSCT showed evidence of mild hunching posture and dull fur. Because rEGFL7 administration could represent an alternative second-line therapy, we initiated rEGFL7 treatment at day +21 post–bone marrow transplantation (BMT) to determine the effect of rEGFL7 on active, established disease rather than preventing the development of aGVHD. After 7 days of rEGFL7 treatment, mice showed marked improvement in posture, fur appearance, and activity. This was accompanied by a stop or reduction in body weight loss in rEGFL7-treated mice in contrast to PBS-treated mice that continued to lose weight (Figure 1A). Importantly, rEGFL7 treatment results in lower clinical scores and prolonged survival of mice after allo-HSCT compared with PBS controls (Figure 1B-C). We validated this result in a second mouse model of aGVHD, B6→Balb/c, showing increased survival of rEGFL7-treated mice compared with PBS controls (supplemental Figure 2A-C). The gut and liver are typically affected by aGVHD.^18,19  Visual examination of the colon revealed the presence of swelling and a reduction in the length in the PBS group compared with the rEGFL7-treated mice (supplemental Figure 1B-C). Histopathology analysis revealed a higher amount of leukocyte infiltration in both large intestine and liver of the PBS group compared with rEGFL7-treated mice (Figure 1D; supplemental Figure 1D). Large intestine of PBS-treated mice showed focal to diffuse areas of villous atrophy with a dense amount of mixed inflammatory infiltrate in the lamina propria composed predominantly of lymphomononuclear cells and neutrophils. In addition, PBS-treated mice showed a more profound loss of mucous-secreting goblet cells in the mucosal epithelium along with enhanced crypt epithelial cell apoptosis (>1 apoptotic body per crypt), and focal crypt abscess formation. In contrast, rEGFL7-treated mice showed normal integrity of mucosal epithelium with negligible to mild degree of lamina propria inflammation and a lesser number of crypt epithelial cell apoptotic bodies (Figure 1D). In the liver, rEGFL7-treated mice showed a fewer number and limited extent of portal tracts involvement by inflammatory cell infiltrates, smaller degree of bile duct injury, lobular necroinflammatory activity, and vascular endotheliitis (supplemental Figure 1D). Thus, reduction in leukocyte infiltration in the gut and liver is consistent with an overall lower aGVHD histopathological score in rEGFL7-treated mice compared with the PBS-treated group (Figure 1E; supplemental Figure 1E). Finally, we performed immunofluorescence microscopy to assess EC and T-lymphocyte populations in the intestine. We found that rEGFL7 treatment resulted in increased proliferation (Ki-67⁺) of intestine ECs (CD31⁺CD45⁻) (Figure 1F; supplemental Figure 1F) with an associated decrease in infiltrating CD4⁺ and CD8⁺ T cells (Figure 1G; supplemental Figure 1G). In addition, we found that rEGFL7 treatment resulted in a decrease in the expression of VCAM1 on CD31⁺ ECs in the gut, suggesting that EGFL7 can reduce EC activation (supplemental Figure 1H-I). EC damage occurs during GVHD,^20,21  and based on these data, we show that rEGFL7 treatment not only reduces EC activation but also perhaps promotes EC repair, leading to reduction in T-cell infiltration in the gut and liver, resulting in GVHD improvement.

aGVHD insults to primary and secondary lymphoid organs impair immune reconstitution after allo-HSCT.^22-25  Treatment with rEGFL7 tends to increase T-cell (Figure 2A), B-cell, and dendritic cell (DC) counts in the spleen (supplemental Figure 3A-B) and BM (supplemental Figure 3C-E) compared with PBS-treated mice. rEGFL7 therapy also tends to increase thymocyte counts with normalization of double-positive and single-positive CD4⁺ and CD8⁺ thymocyte populations in rEGFL7 compared with PBS-treated mice (Figure 2B-C). Variability in immune reconstitution may be related to GVHD insults prior to initiating EGFL7 treatment, but additional studies are needed. Thus, increases in T-cell, B-cell, and dendritic cell counts are consistent with a reduction in aGVHD severity in rEGFL7-treated mice. However, it is also possible that the effect of rEGFL7 on immune reconstitution could result from direct effects on these cells/tissues, thus contributing to decreased disease severity. Part of the success of an allogeneic transplant is the ability to sustain the graft-versus-leukemia (GVL) effect important for eliminating minimal residual disease.^3-5  To study the impact of rEGFL7 treatment on GVL, we transplanted a luciferase-transduced murine mastocytoma P815 cell line along with or without alloreactive-B6 splenocytes. Recipients of allogeneic splenocytes were treated daily with PBS or rEGFL7 starting at day +3 until day +10 post-BMT. Mice transplanted with mastocytoma P815 cells without allogeneic splenocytes all succumbed to leukemia. In mice that received allogeneic splenocytes, treatment with rEGFL7 did not interfere with GVL effects as seen by decreased leukemia burden similar to PBS controls (Figure 2D-E). Splenocytes were isolated from mice at time of death, and flow cytometric analysis confirmed the absence of GFP⁺ P815 cells (GFP⁺) in the spleen of rEGFL7-treated mice (Figure 2F-G), further validating the GVL-sparing effect of rEGFL7. Importantly, further investigation is warranted to evaluate the impact of rEGFL7 on relapse post–allo-HSCT, as EGFL7 effect has been documented on human AML cells.

In conclusion, we identified EGFL7 as a potential therapy to reduce/treat aGVHD. Although immunosuppressive therapies typically impair immune cell activation/functions, the effectiveness of EGFL7 therapy to treat GVHD relies largely on the inhibition of EC activation and access of immune cells to inflamed tissues.¹¹  However, we cannot exclude the possibility that EGFL7 might have a direct effect on alloreactive T lymphocytes. This work represents the first proof of concept that EGFL7 therapy can diminish/treat aGVHD and provides a novel perspective about modulating ECs to reduce GVHD.

This work was supported by grants from the Cancer Research Society of Canada 24380 (M. Guimond), Cancer Research Society and the Leukemia Lymphoma Society of Canada 840026 (M. Guimond), NCI K12 Paul Calabresi Award K12 CA133250-11 (P.R.), and from the American Cancer Society RSG-18-170-01-LIB (A.M.D.). A.A. has received the Andy & Léna Chabot Cancer Research Society Studentship.

Contribution: A.M.D., M.M.M., C.G., N.E.S., S. Kalyan, M. K., R.K., M. Goulard, S. Kolovich, A.R., E.N., C.-E.B., F.D., and A.A. performed GVHD experiments and contributed to data interpretation; R.G. provided scientific input to experiments and reviewed the manuscript; and M. Guimond, A.M.D., and P.R. designed the project, analyzed data, and wrote the manuscript.

Conflict-of-interest disclosure: M. Guimond and A.M.D. hold a revisionary patent for the use of EGFL7 in the treatment of GVHD. The remaining authors declare no competing financial interests.

Correspondence: Martin Guimond, 5415 Blvd de l’Assomption, Montréal, QC H1T 2M4, Canada; e-mail: martin.guimond@umontreal.ca or martin.guimond@gmail.com; and Adrienne Dorrance, 460 W 12th Ave, BRT 890, Columbus, OH 43210; e-mail: adrienne.dorrance@osumc.edu.