Allogeneic hematopoietic cell transplantation (allo-HCT) is the only curative treatment option for many malignant diseases due to the beneficial immunological graft-versus-leukemia (GVL) effect [1]. However, (acute) graft-versus-host disease ((a)GVHD) continues to be a major complication after allo-HCT. This syndrome is caused by donor T cells that attack the intestinal tract, liver, and skin [2]. Early diagnosis remains challenging and to date mainly relies on clinical symptoms and histopathology. When clinical symptoms are readily detected, response to therapy is rarely beneficial to the patient. Therefore, a predictive test to identify patients at risk before the visible onset of aGVHD is highly desirable.

A detailed understanding of the pathophysiological processes including the kinetics of alloreactive T cell priming, migration, and effector mechanisms after allo-HCT may provide clues to identify patients that are likely to develop aGVHD. Recently, in vivo bioluminescence imaging (BLI) of T cells that carried the firefly luciferase (luc+) gene in a major histocompatibility complex (MHC) mismatched mouse model helped define a distinct aGVHD initiation and effector phase [3]. Donor T cells first migrated into secondary lymphoid organs (SLOs), where they were activated by interacting with host antigen presenting cells (APCs). This interaction led to the upregulation of certain homing receptors which the cells used for target tissue infiltration [3]. The subsequent effector phase, taking place in peripheral aGVHD target tissues, started 3 days later when alloreactive T cells left the SLOs and migrated via the peripheral blood (PB) to their respective target organs, mainly the gastrointestinal tract (GIT), the liver, and the skin [4–6].

In the present work, we asked whether it is possible to detect alloreactive donor T cells in the PB early after allo-HCT in a clinically relevant minor histocompatibility antigen (miHAg) mismatch mouse model. We characterized donor T cells according to the expression of two well-described homing receptors, α4β7 integrin and P-selectin ligand. Based on these observations we succeeded in effectively treating mice with rapamycin on first alloreactive cell detection in the PB to timely prevent aGVHD exacerbation.

Understanding the precise timing of T cell engraftment and cell migration patterns is necessary to further investigate the phenotypic and functional changes within the lymphocyte population in order to develop a predictive test to identify patients at high risk for aGVHD. To study the dynamic pathophysiologic events leading to the initiation of aGVHD we first compared survival, weight change, and clinical aGVHD scoring in a clinically relevant model of miHAg mismatch allo-HCT (B6, H-2^b > BALB/b, H-2^b) to a well-defined MHC major mismatched aGVHD model (B6, H-2^b > BALB/c, H-2^d) [5]. Initially, all miHAg mismatched recipients lost and regained weight similar to syngeneic control mice (Figure 1A) and did not display any clinical signs of aGVHD within the first 2 weeks after allo-HCT (Figure 1B). By day +15, miHAg mismatched BALB/b recipients started to lose weight again and by day +21 mice presented the first signs of diarrhea followed by ruffled fur and hunchback. However, most animals survived until the end of the experiments at day +30 (Figure 1C). All MHC major mismatched BALB/c recipients developed signs of a hyperacute form of GVHD (weight loss, ruffled fur, diarrhea, hunched posture, and lethargy) and most animals succumbed between day +6 and day +14 after HCT (Figure 1A). Transfer of BM cells alone rescued all recipients from myeloablative irradiation except for one BM control mouse that likely did not engraft (Figure 1C). Syngeneic albino B6 recipients (H-2^b) did not develop any signs of aGVHD (Figure 1B), regained their body weight, and survived until the end of the experiments (Figure 1A,C). Daily in vivo BLI measurements confirmed the same spatiotemporal shift of aGVHD initiation to effector phase in both allogeneic models but a diverging signal increase after miHAg mismatch allo-HCT. These early events in aGVHD initiation are in accordance with a seminal study by Korngold and colleges [6]. In contrast, in syngeneic HCT recipients, homeostatic T cell proliferation led to a moderate whole body as well as single organ signal increase compared to allogeneic recipients (see Additional file 1A,B).

Next, we closely monitored donor T cell migration patterns in the clinically relevant miHAg mismatch model to address whether it is possible to detect alloreactive donor T cells in the PB early after allo-HCT. Therefore, we analyzed PB samples of transplanted mice daily by multiparameter flow cytometry (Figure 2A). By employing the well-established MHC major mismatch model of hyperacute GVHD we detected donor T cell mobilization from SLOs to the PB as early as day +4 after allo-HCT. Donor CD4⁺ and CD8⁺ T cell frequencies peaked between day +6 and +8 and decreased until mice succumbed to aGVHD (Figure 2A, left panel). Similarly, after miHAg allo-HCT, donor T cells first mobilized to the PB by day +4, but in contrast to the hyperacute model, elevated numbers of migrating donor CD4⁺ and CD8⁺ T cells remained readily detectable throughout a period of 12 days before cell numbers declined and mice started to show first signs of aGVHD (by day +21) (Figure 2A, middle panel). T cell chimerism analysis after miHAg mismatch allo-HCT revealed that after day +6 more than 90% of PB CD8⁺ and CD4⁺ T cells were of donor T cell origin (Figure 2B). Of note, the detection of CD90.1⁺ T cells only included transferred donor T cells but not donor bone marrow derived CD90.2⁺ T cells. This explains why after day +15 CD90.1⁺ donor T cells apparently decline. However, this cannot be ascribed to an increase of host type T cells but rather to an increase of donor bone marrow CD90.2⁺ T cells that could not be detected by our type of chimerism analysis (Figure 2B). In syngeneic recipients donor T cell numbers slowly but constantly increased suggesting homeostatic T cell reconstitution of the immune system (Figure 2A, right panel). Histopathological analyses of aGVHD target organs confirmed the donor T cell infiltration in these organs. By day +6 after MHC major mismatch allo-HCT allogeneic T cells markedly infiltrated target organs resulting in massive tissue damage. In contrast, miHAg mismatch allo-HCT recipients displayed low intestinal T cell infiltration and only mild histopathological damage on day +6. However, by day +21, donor T cells massively infiltrated small and large bowel of miHAg mismatch allo-HCT recipients that by this time displayed the first clinical signs of aGVHD. The tissues of syngeneic controls displayed only mild T cell infiltrates and no tissue damage at this time (see Additional file 2A,B). As aGVHD target organ infiltration depends on the appropriate cellular homing receptors [3] we considered whether alloreactive T cells could be more precisely characterized by the expression of the homing receptors α4β7 integrin and P-selectin ligand, implicated in intestinal and skin aGVHD manifestation, respectively [12, 13]. Therefore, we examined the dynamic changes in their expression daily after HCT to test a possible correlation between their upregulation and the identified phase of massive cell migration (Figure 2C,D). Until day +3, only few donor T cells circulated in the PB in all three models and these cells were mostly negative for the receptors tested. On day +6, donor T cells of the MHC major mismatch model upregulated α4β7 integrin and P-selectin ligand compared to a minor upregulation of those receptors after miHAg allo-HCT. Importantly, in the latter model upregulation of α4β7 integrin and P-selectin ligand peaked at day +11 (45% α4β7⁺ CD8⁺ and 22% P-lig⁺ CD8⁺ donor T cells, Figure 2D). After day +11, numbers of cells expressing these homing receptors slowly declined correlating with a drop in absolute cell numbers. In contrast, donor T cells in syngeneic transplanted mice expressed only low levels of α4β7 integrin and P-selectin ligand over time.

In line with our results, two clinical retrospective studies indicate a correlation between homing receptor upregulation and the subsequent development of aGVHD symptoms. Chen and colleagues found at least tenfold upregulation of α4β7 integrin on naïve as well as memory T cell subsets before the onset of clinically apparent intestinal GVHD compared to cutaneous or no aGVHD in a case-controlled study including 59 patients [14]. Another study with 33 patients found an association of cutaneous lymphocyte antigen (CLA, a P-selectin glycoprotein ligand-1)-upregulation on PB T cells and the onset of severe aGVHD [15]. Furthermore, in a recent clinical study Chen et al. could correlate the detection of α4β7⁺ memory T cells with the appearance of intestinal aGVHD in patients [16]. However, these studies did not address how soon before aGVHD onset and for how long these T cell surface receptors could be detected.

Different functional studies in murine models across major or minor MHC mismatches confirmed that α4β7 or ß7 negative donor T cells cause less aGVHD morbidity and mortality compared to WT T cells mainly due to reduced homing to the intestinal tract while still exhibiting the GVT effect [12, 17]. Likewise, ameliorated aGVHD was observed in P-selectin-deficient mice, which underscores the importance of the interaction between P-selectin and its ligands that are highly expressed on migrating effector T cells [13]. In our experiments we also found significantly higher α4β7 integrin and P-selectin ligand expression on migrating allogeneic T cells than in syngeneic controls. The expression of these molecules on PB donor T cells correlates with the identified cell migration phase and strongly associates with aGVHD induction in allo-HCT suggesting their potential usefulness as predictive markers. Recently, research efforts have focused on the identification of reliable predictive markers to identify patients at risk before aGVHD onset with some potential biomarkers being investigated thoroughly [18, 19]. Different combinations of these biomarkers have been used to predict treatment response and survival of aGVHD patients, but a predictive marker panel to reliably identify patients at risk for aGVHD is still lacking.

Figure 3 summarizes the dynamic events of aGVHD pathophysiology after miHAg mismatch allo-HCT. The described donor T cell migration phase with the cells being highly positive for α4β7 integrin and P-selectin ligand opens a potential diagnostic window for a timely therapeutic intervention prior to the onset of clinically apparent aGVHD, which did not develop before day +21 in this model.

Therefore, we asked whether this time frame from the first detection of alloreactive T cells suffices to effectively prevent aGVHD onset. Subsequently, we started to treat mice with rapamycin [20, 21] as soon as we could detect alloreactive T cells in the PB (day +6) (Figure 4A). A total of 10 days of rapamycin administration strongly reduced the total body BLI signal intensity of treated mice compared with vehicle controls. By day +11, treated mice already showed much less overall body signal predominantly originating from the spleen, skin, and BM compartment. Abdominal signals were very low compared with the controls. Immunofluorescence stainings of donor T cell infiltrates in aGVHD target organs confirmed reduced CD90.1⁺ donor T cells in rapamycin-treated mice compared to controls on day +11 after miHAg mismatch allo-HCT (see Additional file 3A-D). Quantification of BLI images revealed a significant reduction of total body signal intensities by day +11 and day +15 after allo-HCT (Figure 4B). Treated mice showed a 7-fold lower signal by day +11 and a 13-fold lower signal by day +15 compared with vehicle controls. By the end of treatment (day +15) the signal from the treated group slightly increased but the benefit of the treatment remained apparent with a fourfold lower BLI signal compared to control mice. Importantly, rapamycin treatment resulted in the recovery from already visible aGVHD symptoms occurring by day +21 (Figure 4C) that constantly improved thereafter. All mice treated in a timely fashion with rapamycin survived (Figure 4C).

We propose that the identified migration phase of alloreactive T cells may serve as a potential diagnostic window for early and timely targeted therapeutic interventions. In addition to the described surface markers, α4β7 integrin and P-selectin ligand, combinations with other markers appear attractive. Further investigations of suitable surface receptor combinations to precisely identify alloreactive PB T cells could be complemented by detecting polypeptides in the urine of patients as early indicators for aGVHD. Kaiser et al. identified two peptides of the leukotriene A4 hydrolase and of serum albumin as possible biomarkers [22]. Furthermore, tumor necrosis factor receptor 1 (TNFR1), interleukin 2 receptor alpha (IL-2Ra), IL-8, and hepatocyte growth factor (HGF) as plasma biomarkers relevant for aGVHD effectively discriminated between patients with and without aGVHD [23] and elafin functioned as a biomarker for skin aGVHD patients [24]. Different combinations of these biomarkers have been used to predict treatment response and survival of aGVHD patients [18, 19]. We envision combining the described homing receptor panel with these biomarkers in order to develop a reliable predictive test to identify patients at risk before disease onset.

Encouraging for the clinical allo-HCT setting, the directed, preemptive rapamycin treatment starting immediately at the first detection of alloreactive donor T cells significantly improved aGVHD symptoms and granted survival in all miHAg allo-HCT recipients. Similarly, preemptive treatment with other immunomodulators and, particularly, drugs altering T cell trafficking [25] appear highly attractive, but may exert their biggest benefit if individuals at high risk for aGVHD can be identified in a timely manner before the first clinical symptoms appear.

In the present study, we demonstrate in a clinically relevant miHAg allo-HCT mouse model that it is feasible to detect migrating alloreactive donor T cells for an extended time period of 2 weeks and, importantly, before aGVHD symptoms become apparent. As the prediction of aGVHD is highly relevant to the clinical outcome, it may be beneficial to closely monitor the kinetics of T cell engraftment, expansion, and homing receptor expression in allo-HCT patients. Our preclinical findings may have implications for the development of a predictive blood test identifying patients at risk for aGVHD and thereby giving physicians the chance for a timely and targeted therapeutic intervention.