Tregs Suppress GvHD at the Periphery and Unleash the Gvl Effect in the Bone Marrow

Posttransplant relapse is still a major cause of treatment failure in high-risk acute leukemia. Evidence from experimental bone marrow transplantation (BMT) showed coinfusion of conventional T lymphocytes (Tcons) with regulatory T lymphocytes (Tregs) suppressed lethal Graft-versus-Host Disease (GvHD) without impairing the Graft-versus-Leukemia (GvL) effect (Edinger et al., Nat Med. 2003;9:1144-1150). In HLA haploidentical transplantation for acute leukemia we demonstrated that donor-derived natural Tregs, coinfused with Tcons, protected recipients against GvHD (Di Ianni et al., Blood 2011, 117:3921-3928) and largely prevented posttransplant leukemia relapse, as only 5% of evaluable patients relapsed (Martelli et al., Blood 2014, 124:638-44).

The mechanism by which the GvL effect is maintained in the absence of GvHD is still unknown. In humans, naïve CD45RA+ Tregs express CXCR4 and preferentially localize to the bone marrow while memory CD45RO+ Tregs display lower CXCR4 expression and home to the periphery (Booth et al., J Immunol. 2010;184:4317-4326). Since peripheral blood Tregs to be used for adoptive immunotherapy are CD45RO+, we hypothesized that GvL without GvHD might be due to unopposed Tcon alloreactivity in the bone marrow combined with regulated T cell alloreactivity at the periphery. Immunodeficient NSG mice (a total of 40 for each experimental group) received human myeloid or lymphoblastic leukemia (3x10⁶) and HLA mismatched Tregs/Tcons (3x10⁶). Mice that received leukemia and Tcons (without Tregs) cleared leukemia but died of GvHD. Human T cells harvested from their BM, spleen and liver were predominantly (90%) CD8+ and displayed potent alloreactivity (by 51-chromium release at an E:T ratio of 5:1) against human leukemia (lysis = 50% ± 15), autologous to leukemia PHA blasts (lysis = 60% ± 10) and mouse Con A blasts (lysis = 40% ± 8). Mice that received human leukemia and Tcons plus Tregs were rescued from leukemia and survived without GvHD. Human T cells harvested from spleen and liver were composed of CD8+ T cells (40%) and CD4+/FOXP3- T cells (60%). Despite their FOXP3 negativity, purified CD4+ T cells had retained their regulatory function as they inhibited mixed lymphocyte reaction (by 50% at a Tcon/Treg ratio of 1:2). Purified CD8+ T cells displayed no alloreactivity against human leukemia, PHA blasts and mouse Con A blasts. In contrast, human T cells harvested from BM were still predominantly CD8+ and still displayed potent alloreactivity against human leukemia and autologous to leukemia PHA blasts and mouse ConA blasts, suggesting Tcons had retained their alloantigen recognition against human and mouse MHC. Finally, in mice treated with Tregs alone, human T cells were recovered only in the spleen and liver, they displayed a CD4+/FOXP3- phenotype and inhibited mixed lymphocyte reaction. No human T cells were found in the bone morrow, thus showing human CD45RO+ Tregs do not home to the bone marrow.

In conclusion, Treg/Tcon adoptive immunotherapy mediates GvL effect in the absence of GvHD because Tcons that home to the bone marrow exert unopposed alloantigen recognition, while Tcons that home to the periphery are blocked by Tregs.