SEARCH

SEARCH BY CITATION

Keywords:

  • calcineurin;
  • CTLA-4;
  • graft-versus-host disease;
  • humanized mice;
  • NOD.scid IL-2Rγ−/−

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

Dysregulated T cells are a hallmark of several autoimmune and inflammatory diseases; thus, models to study human T cells in vivo are advantageous, but limited by lacking insight into human T cell functionality in mice. Using non-obese diabetic (NOD), severe combined immunodeficient (SCID) or recombination activating gene-1 (RAG1)−/− and interleukin-2 receptor gamma-chain (IL-2Rγ)−/− mice reconstituted with human peripheral blood mononuclear cells (PBMCs), we have studied the mechanisms of human T cell expansion and activation in mice. Injection of human PBMCs into mice caused consistent xeno-engraftment with polyclonal expansion and activation of functional human T cells and production of human cytokines. Human T cell expansion coincided with development of a graft-versus-host disease (GVHD)-like condition observed as weight loss, multi-organ immune infiltration and liver damage. CD8+ T cells alone were sufficient for expansion and required for disease development; in contrast, CD4+ T cells alone expanded but did not induce acute disease and, rather, exerted regulatory capacity through CD25+CD4+ T cells. Using various anti-inflammatory compounds, we demonstrated that several T cell-activation pathways controlled T cell expansion and disease development, including calcineurin-, tumour necrosis factor-α and co-stimulatory signalling via the CD80/CD86 pathway, indicating the diverse modes of action used by human T cells during expansion and activation in mice as well as the pharmacological relevance of this model. Overall, these data provide insight into the mechanisms used by human T cells during expansion and activation in mice, and we speculate that PBMC-injected mice may be useful to study intrinsic human T cell functions in vivo and to test T cell-targeting compounds.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

Dysregulated T cell responses are a hallmark of several autoimmune and inflammatory disorders [1-5] and modulation of T cell activation, expansion or effector function in vivo could be considered a reasonable predictor of efficacy for treatment strategies against such diseases. Unfortunately, studies of human T cells are restricted largely to in-vitro experiments, and mouse T cells remain the model of choice for T cell studies in vivo. However, development of novel targeted therapies, such as monoclonal antibodies, is often challenged by the lack of cross-reactivity or target incompatibility across species necessitating surrogate compounds for preclinical testing.

Attempts to create robust mouse models with functional components of the human immune system have, for several decades, been an attractive goal to bridge these problems [6-8]. Historically, these models have suffered from low human chimerism with poor functionality of the engrafted cells. However, recent advances in transgenic mouse models have demonstrated that non-obese diabetic (NOD).severe combined immunodeficient (SCID) or NOD.recombination-activating gene (RAG)1−/− mice carrying a mutation in the interleukin (IL)-2 receptor common-gamma chain (IL-2Rγc−/−) show highly improved engraftment of human immune cells [8-10]. In particular, it has been described that T cells from human peripheral blood mononuclear cells (PBMCs) can engraft and expand and eventually give rise to a xenogeneic graft-versus-host disease (GVHD) in these mice [9, 10]. Although this model does not recapitulate a complete human immune system, these findings imply a model to study intrinsic mechanisms regarding human T cell activation, expansion and effector function in vivo. Indeed, recent studies using this model have generated important lessons regarding, e.g. IL-21, inducible regulatory T cell protocols and intravenous (i.v.) immunoglobulins during human T cell expansion and activation [11-13]. However, the mode of action (MoA) driving the T cell response in this model and the role of individual T cell subsets still remain unclear, thus limiting the use and interpretation of data from such a model. In this study, we wanted to explore the MoA involved in the human T cell response that develops in PBMC-injected mice and eventually leads to xenogeneic GVHD; to specifically dissect the interplay of CD4+ and CD8+ T cell subsets in the model and investigate its feasibility for studies of T cell-targeting therapies.

Using NOD.RAG1−/− IL-2Rγc−/− and NOD.SCID IL-2Rγc−/− mice transferred with human PBMCs, we show that a polyclonal repertoire of activated CD4+ and CD8+ T cells engraft and expand in the model; CD8+ T cells are the main drivers of disease, whereas CD4+ T cells play a role in part by regulating the response through CD4+CD25+ T cells. Furthermore, we demonstrate that several anti-inflammatory compounds blocking distinct T cell signalling pathways modulate T cell expansion and disease development. Taken together, our data suggest that the human T cell response that develops in mice is an attractive model for in-vivo studies of intrinsic human T cell pathology and could be relevant for the testing of novel T cell-targeting therapies.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

Mice

NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) and NOD.Cg-Rag1tm1Mom Il2rgtm1Wjl/SzJ (NRG) mice were purchased from Jackson Laboratories (Bar Harbor, ME, USA) and NOD.Cg-Prkdcscid Il2rgtm1Sug/JicTac (NOG) mice from Taconic (Germantown, NY, USA). Mice were housed in a pathogen-free environment in a dedicated immunodeficient mouse facility with unlimited access to sterile food and water. Mice aged 8–10 weeks were used in all experiments, which were carried out in accordance with governmental and local animal ethics guidelines.

Drugs

Cyclosporin A (CsA) (Sandimmune®), tacrolimus (Tac) (Prograf®), tumour necrosis factor receptor (TNFR)II-hIg-Fc (Enbrel®) and cytotoxic T lymphocyte-4 (CTLA-4)-hIg-Fc (Orencia®) were purchased through Nomeco, Denmark. Human immunoglobulin (Ig)G-Fc control protein was purchased from BioXCell (West Lebanon, NH, USA).

Flow cytometric analysis

Anti-human antibodies used were as follows: fluorescein isothiocyanate (FITC)- and allophycocyanin (APC)-conjugated CD45, peridinin-chlorophyll (PerCP)-conjugated CD4, Pacific Blue-conjugated CD8, phycoerythrin-cyanin 7 (PE-Cy7)-conjugated CD19, PE-conjugated CD45RO (all from BD Biosciences, San Jose, CA, USA), APC-conjugated CD62L (eBiosciences, San Diego, CA, USA) and Qdot 655-conjugated CD3 (Invitrogen, Carlsbad, CA, USA). The anti-mouse antibodies used were Pacific Orange-conjugated CD45. Non-viable lymphocytes were excluded using LIVE/DEAD® Fixable Near-IR Dead Cell Stain Kit (Invitrogen). Human Fc-receptor blocking reagent (Miltenyi Biotec, Bergisch Gladbach, Germany) was used in all experiments. The IO Test Beta Mark Kit (Beckmann-Coulter, Indianapolis, IN, USA) was used for T cell receptor variable region beta (TCRVβ) analysis.

Cell suspensions were stained in tubes or 96-well plates for 30 min at 4°C in the dark and washed between incubations. Acquisition and analysis was performed on an LSR-II flow cytometer using FACSDiva software (BD Biosciences).

Human PBMC-injected mouse model

Human PBMCs were purified using Ficoll-Paque PLUS (GE Healthcare, Brøndby, Denmark) in Leucosep-tubes (Greiner Bio-One, Frickenhausen, Germany) from buffy-coats donated by healthy donors at the blood bank, Rigshospitalet, Denmark, under informed consent.

Female NRG, NSG or NOG mice were injected i.v. via the tail vein with 20 × 106 human PBMCs in phosphate-buffered saline (PBS). CD4+, CD8+ and CD25+ cell depletion or CD4+ T cell purification from PBMCs was performed using an AutoMACS® magnetic cell separator through depletion of magnetically labelled cells or negative selection of non-labelled cells according to the manufacturer's protocol (Miltenyi Biotec). CD8+ T cell purification was performed using negative selection of non-magnetically labelled cells on a RoboSep® cell separator according to the manufacturer's protocol (Stemcell Technologies, Grenoble, France). Following injection of cells, animals were randomized and monitored three times weekly for weight loss and other signs of reduced health, e.g. hunched back, ruffled fur and reduced mobility. Blood samples were taken weekly to monitor human cell engraftment. Preventive treatments were started on day 0 following PBMC injections and therapeutic treatments were started post-day 15, when individual mice experienced either >10% weight loss or showed weight loss on three consecutive measurements.

Mice were euthanized if they experienced >20% weight loss or showed signs of compromised health and were considered to have human graft-related disease if an expanded population of human cells was detected simultaneously in blood.

Plasma cytokine and chemistry analysis

Plasma samples were analysed for the liver enzymes alanine transaminase (ALT) and aspartate transaminase (AST) on a Hitachi 912 Chemistry Analyser (Roche, Hvidovre, Denmark). Plasma cytokines were analysed using multiplex suspension analysis on a Bio-Plex® system (Bio-Rad Laboratories, Copenhagen, Denmark) and a custom-made 8-plex human cytokine-detection kit for IL-2, IL-4, IL-6, IL-10, IL-12p70, IL-17, IFNγ and TNFα (Bio-Rad Laboratories).

Immunohistopathology

Liver, lung, skin, ileum, colon and kidney were harvested at time of 20% weight loss, fixed in neutral-buffered formalin, embedded in paraffin and sectioned (6 μm). Sections were stained with haematoxylin and eosin (H&E) or with anti-human CD3 (SP7) (Thermo Scientific, Erembodegem, Belgium) by immunohistochemistry. A histopathological inflammation score was evaluated on H&E-stained sections in a blinded fashion by a histopathologist and given a mean score based on infiltration of mononuclear cells, extent of changes from normal histology and extent of necrosis (score range for each parameter: 0–4; 0 = none, 1 = mild, 2 = moderate, 3 = pronounced, 4 = severe).

Statistics

The Mantel–Cox log-rank test was used to evaluate statistical differences in Kaplan–Meier analyses. Student's t-test (two-tailed, assuming equal variance), two-tailed Mann–Whitney U-test or Kruskal–Wallis test with Dunn's post-test were used for statistical evaluations as indicated. Data are shown as individual observations or as mean ± standard error of the mean (s.e.m.) and P-values less than 0·05 were considered significant.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

Polyclonal expansion of activated T cells in human PBMC-injected mice

Initially NOG, NSG and NRG mice were evaluated and showed similar T cell engraftment and disease development (data not shown), thus NOG mice were used hereafter for most studies as indicated. PBMC injection into mice resulted in a significant expansion of human T cells in blood, followed by the development of severe weight loss necessitating euthanasia (Fig. 1a,b). No detection of hCD14+ monocytes, hCD11c+ dendritic cells (DC), hCD56+ natural killer (NK) cells or hCD123+ plasmacytoid DCs was found in blood samples (data not shown). A variable frequency of CD19+ B cells was detected ∼0·5–2% of hCD45+ in blood and spleen (data not shown and Fig. S1). However, large quantities of human IgG were detected in plasma, suggesting the presence of IgG-producing B cells and/or plasma cells (Fig. S2).

figure

Figure 1. Human T cells show polyclonal expansion of activation in human peripheral blood mononuclear cell (PBMC)-injected mice resulting in multi-organ pathology. Groups of NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG), NOD.Cg-Rag1tm1Mom Il2rgtm1Wjl/SzJ (NRG) (a–c) and NOD.Cg-Prkdcscid Il2rgtm1Sug/JicTac (NOG) (d–g) mice were injected with 20 × 106 human PBMCs from three different donors (A, B and C) and monitored for disease development by weight loss and human cell expansion in blood. (a) Graph depicts the percentage of starting weight and Kaplan–Meier survival curves depict the percentage of mice with <20% weight loss. (b) Blood samples were drawn once weekly and analysed by fluorescence activated cell sorter (FACS); the percentage of hCD45+ cells of all live cells and the percentage of hCD3+ of hCD45+ cells are plotted. (c) Activation of human T cells in blood was evaluated by expression of CD62L and CD45RO; representative FACS plots show CD62L and CD45RO expression by CD4+ (top panel) and CD8+ (bottom panel) and bar plots summarize the expression of CD62L and CD45RO on T cells over time. (d) The clonality of the T cell response during the xenogeneic reaction was tested by evaluation of the T cell receptor variable region beta (TCRVβ) distribution in T cells from PBMCs before injection and at the indicated time post-injection. (e) Liver, lung, kidney, colon, ileum and skin were harvested at the time of graft-versus-host disease (GVHD) and evaluated by histopathology. Representative pictures of microsections of livers from naive and PBMC-injected NOG mice stained with haematoxylin and eosin or anti-CD3 are shown. Scale bars = 180 μm. (f) Bar plots show the histopathology score of indicated organs evaluated blindly, as outlined in Materials and methods. (g) Plasma samples from naive and human PBMC-injected mice at time of 20% weight loss were analysed for the liver enzymes alanine transaminase (ALT) and aspartate transaminase (AST). Data are shown as mean ± standard error of the mean, n = 6–8 (a–c), n = 11 (f), or as individual data points (d,g) of n = 2 experiments (d); **P < 0·01 by Mantel–Cox log-rank test of PBMC-injected versus naive mice.

Download figure to PowerPoint

Engrafted T cells acquired an activated/effector phenotype in vivo, illustrated by increased CD45RO and reduced CD62L expression (Fig. 1c). The TCRVβ repertoire present in PBMCs prior to injection was largely maintained during expansion and disease development, suggesting a polyclonal T cell expansion, although some perturbation was observed particularly in the CD8+ T cell repertoire where, e.g. Vb11·0- and Vb12·0-expressing clones show a relatively restricted expansion during disease progression (Fig. 1d).

While substantial T cell expansion and activation could lead to T cell anergy, we found that organ-infiltrating T cells at the time of euthanasia had maintained TCR-responsiveness and enhanced proliferation to IL-2 and IL-15, but not IL-12, IL-18 or IL-21 (Fig. S3). Thus, polyclonal, activated and functional T cells were the predominant human cell type in PBMC-injected mice.

Multi-organ immune infiltration and liver pathology by human T cells in mice

At time of euthanasia, mice presented mainly with significant weight loss and general malaise, but to evaluate the dissemination of disease further, a histopathological evaluation of several major organs at the time of euthanasia was performed, as described in Materials and methods. Representative images of liver microsections from mice injected with human PBMCs showed substantial mononuclear CD3+ cell infiltration compared to naive mice (Fig. 1e). Histopathological evaluation showed that the main organs presenting pathological changes were the liver and lung, with lower scores found in kidney and skin and near-normal gut histology (Fig. 1f). Liver leucocytes were highly increased in diseased mice and were predominantly human T cells (Fig. S1). Furthermore, the liver enzymes AST and ALT were elevated significantly at the time of euthanasia compared to naive mice (Fig. 1g), suggesting pronounced liver damage by the infiltrating human T cells.

Human CD8+ T cells are required and sufficient for xenogeneic disease development in mice, whereas CD4+ T cells show regulatory activity

To study the role of human T cell subsets during expansion and disease development, we depleted CD4+ or CD8+ T cells from PBMCs before injection. Depletion of CD8+ T cells significantly postponed the time to euthanasia and reduced disease incidence, suggesting that CD8+ T cells were the main drivers of the observed pathology and weight loss (Fig. 2a). A few animals receiving CD8-depleted PBMCs were euthanized due to weight loss, but this was due most probably to a delayed expansion of few remaining CD8+ T cells (Fig. 2a). In contrast, depletion of bulk CD4+ T cells from PBMCs did not change the rate or frequency of disease development (Fig. 2a). Injection of purified CD4+ or CD8+ T cells corresponding to numbers in total PBMCs (8 × 106 cells or 1 × 106 cells, respectively), demonstrated that CD4+, but not CD8+ T cells, were able to expand but without causing significant weight loss (Fig. 2b). However, by increasing the amount of injected CD8+ T cells, purified CD8+ T cells could expand and were sufficient to cause significant weight loss with accelerated manifestations (Fig. 2c). Finally, CD25+forkhead box protein 3 (FoxP3)+CD4+ regulatory T cells (1–2% of all CD4+ T cells) were detectable in the blood of PBMC-injected animals until day 20 (data not shown). By depleting CD25+ cells before injection of PBMCs, T cell expansion and weight loss were accelerated significantly (Fig. 2d), suggesting that regulatory CD25+CD4+ T cells were involved in controlling human T cell-induced pathology in mice.

figure

Figure 2. Human CD8+ T cells are required and sufficient for xenogeneic disease development, whereas CD4+ T cells show regulatory activity. Groups of NOD.Cg-Prkdcscid Il2rgtm1Sug/JicTac (NOG) mice were injected with 20 × 106 peripheral blood mononuclear cells (PBMCs) and weight loss and cell expansion were compared to mice injected with 20 × 106 CD4-depleted or CD8-depleted PBMCs (a), indicated number of purified CD4+ or CD8+ T cells (b), purified CD8+ T cells (c) and 20 × 106 CD25-depleted PBMCs (d). Kaplan–Meier survival curves depict the percentage of mice with <20% weight loss and bar plots show the mean ± standard error of the mean of the absolute number of human CD4+ and CD8+ T cells in blood on the indicated days post-injection. **P < 0·01; ***P < 0·001 by Mantel–Cox log rank test or Mann–Whitney U-test compared to PBMCs ΔCD8 in (a) and to PBMCs in (b–d).

Download figure to PowerPoint

Calcineurin-inhibitors attenuate human T cell-mediated disease in mice

Given the key role of human T cells in this xenogeneic disease development, we tested the efficacy of the calcineurin inhibitors CsA and Tac [14], which are the standard of care for patients with, e.g. bone marrow transplant-associated allogeneic GVHD. Doses of 3 mg/kg of Tac and 30 mg/kg of CsA were chosen based on previous work [15]. Mice were treated from day 0 and 5×/week until day 15 and 3×/week for the remainder of the study by subcutaneous (s.c.) administration. Results showed that both Tac and CsA delayed the time to significant weight loss and human CD45+ cell expansion in blood (Fig. 3a). CsA showed a near-complete inhibition of human CD45+ cell expansion; Tac only showed a suboptimal effect on CD45+ cell expansion, and following treatment withdrawal substantial CD45+ cell expansion was observed in Tac-treated mice resulting in rapid weight loss (Fig. 3a). A dose-titration of CsA (5 and 30 mg/kg, s.c., 3×/week) showed significantly delayed disease development at both doses; however, with 5 mg/kg CsA fewer than half the mice were resistant to weight loss at study termination, compared to 100% at 30 mg/kg (Fig. 3b). Interestingly, human cell expansion in blood following 5 mg/kg CsA was largely comparable to NaCl-treated mice, suggesting that CsA does not work simply by inhibiting cell expansion (Fig. 3b).

figure

Figure 3. Cyclosporin A and tacrolimus attenuate human T cell-mediated disease in mice. Groups of NOD.Cg-Prkdcscid Il2rgtm1Sug/JicTac (NOG) mice were injected with 20 × 106 human peripheral blood mononuclear cells (PBMCs) and monitored for weight loss and cell expansion as in Fig. 1. (a) Mice were treated by subcutaneous administration of 30 mg/kg cyclosporin A (CsA), 3 mg/kg tacrolimus (Tac) or NaCl 5×/week until day 15 and thereafter 3×/week until withdrawal of dosing on day 64 (indicated by the vertical dotted line). (b) Mice were treated with 30 mg/kg or 5 mg/kg of CsA or NaCl 3×/week throughout the study. Kaplan–Meier survival curves depict percentage of mice with <20% weight loss. Bars plots show the number of hCD45+ cells in blood drawn on the indicated days post-injection of PBMCs as mean ± standard error of the mean. *P < 0·05; **P < 0·01; ***P < 0·001 by Mantel–Cox log-rank test in survival analysis or by Kruskal–Wallis test with Dunn's post-test in blood cell analysis, all compared to NaCl-treated mice.

Download figure to PowerPoint

TNF-α blockade attenuates human T cell-mediated disease in mice

Next, we sought to determine whether or not the role of human proinflammatory cytokines could be studied in the human PBMC-injected mice and whether T cell function would be affected by neutralization of TNF-α. In this study, TNF-α receptor II-Ig (TNFRII-Ig) fusion protein showed a significant deceleration of weight loss as well as human CD45+ cell expansion using 10 mg/kg, 3×/week from day 0 (Fig. 4a). Interestingly, plasma cytokine analysis revealed that TNF-α detection in NaCl-treated mice was low, but increased significantly in TNFRII-Ig-treated mice (Fig. 4b) suggesting significant production of TNF-α in the model, and that TNFRII-Ig works by sequestering TNF-α in the circulation. Furthermore, a significant reduction in plasma IFN-γ and IL-10 was observed following TNFRII-Ig, whereas IL-2, IL-4, IL-6, IL-12p70 and IL-17 were all below the detection limit (data not shown).

figure

Figure 4. Blockade of tumour necrosis factor (TNF)-α attenuates human T cell-mediated disease in mice. Groups of NOD.Cg-Prkdcscid Il2rgtm1Sug/JicTac (NOG) mice were injected with 20 × 106 human peripheral blood mononuclear cells (PBMCs) and monitored for weight loss and cell expansion as in Fig. 1. Mice were treated from day 0 with 10 mg/kg TNF-αRII-hIg fusion protein or NaCl 3×/week throughout the study. (a) Kaplan–Meier survival curves depict percentage of mice with <20% weight loss and bars indicate mean ± standard error of the mean of the number of hCD45+ cells in blood drawn on the indicated days post-injection of PBMCs. (b) Plasma samples taken on indicated time-points from mice treated with TNF-αRII-hIg or NaCl were analysed for the presence of human cytokines by multiplex analysis. Horizontal dotted lines indicate the lower limit of detection in the assay for each cytokine. *P < 0·05; **P < 0·01; ***P < 0·001 by Mantel–Cox log-rank test in survival analysis or by Mann–Whitney U-test in blood cell analysis, all compared to NaCl-treated mice.

Download figure to PowerPoint

CTLA-4-Ig inhibits xenogeneic T cell expansion and treats a subset of mice with active disease

Finally, we sought to determine whether or not T cell co-stimulation was required for the xenogeneic T cell reaction. To address this, we used human CTLA-4-Ig fusion protein, which is reported to be active in mice [16]. Treatment with CTLA-4-Ig at 10 mg/kg, 3×/week, from day 0, completely abrogated the development of disease and human cell expansion in blood compared to hIgG1-Fc treatment (Fig. 5a). To evaluate whether continuous co-stimulation was required, we treated mice therapeutically by individual inclusion criteria (see Materials and methods). Interestingly, a significant fraction of mice (six of 19) could be rescued from active disease associated with a contraction of human cells in their blood (Fig. 5b). Furthermore, histopathological analysis showed that CTLA-4-Ig responders had lower histopathological scores in all evaluated organs compared to non-responders (Fig. 5c). These data suggest critical and continued involvement of co-stimulation in the xenogeneic T cell reaction in mice and that CTLA-4-Ig can limit T cell expansion in both blood and target tissue.

figure

Figure 5. Cytotoxic T lymphocyte (CTLA)-4–immunoglobulin (Ig) inhibits human T cell expansion in mice and treats a subset of mice with active disease. Groups of NOD.Cg-Prkdcscid Il2rgtm1Sug/JicTac (NOG) mice were injected with 20 × 106 human peripheral blood mononuclear cells (PBMCs) and monitored for weight loss and cell expansion as in Fig. 1. (a) Mice were treated from day 0 with 10 mg/kg of the CTLA-4-hIg fusion protein or human IgG1-Fc protein 3×/week throughout the study. (b) Mice were treated therapeutically with individual onset of treatment based on the inclusion criteria: post-day 15 and >10% weight loss or with progressive weight loss on three consecutive measurements. Kaplan–Meier survival curves depict the percentage of mice with <20% weight loss and bars indicate mean ± standard error of the mean (s.e.m.) of the number of hCD45+ cells in blood drawn on the indicated days post-injection of PBMCs. (c) A histopathological evaluation as outlined in Materials and methods was performed on haematoxylin and eosin-stained microsections from the indicated organs and groups from the therapeutic study, CTLA-4-Ig Rs and CTLA-4-Ig non-Rs refer to responders and non-responders to the therapy. Data are shown as an accumulated score indicating the contribution from each organ (left) and as an average score per organ where bars indicate mean ± s.e.m. (right). *P < 0·05 by Kruskal–Wallis test with Dunn's post-test in organ histopathology evaluation compared to all other groups. **P < 0·01 by Mantel–Cox log-rank test in survival analysis compared to hIgG-Fc and PBMCs.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

Dysregulation of T cells is a mainstay in several autoimmune and inflammatory disorders [17]. Thus, it could be argued that mouse models to study human T cell expansion and effector function could provide a novel insight into the function of pathological human T cells in vivo and be useful for the testing of novel T cell-targeting therapeutics. In this regard, injection of human PBMCs into immunodeficient mice represents one approach where human T cells can engraft in mice and give rise to pathology [9]. However, the human T cell response that develops in mice remains poorly understood, limiting the use of such a model for mechanistic as well as pharmacological studies. In this study, we present data from PBMC-injected mice where we have dissected the human T cell response leading to xenogeneic disease. We describe the human T cell phenotype that arises in mice and delineate the role of CD8+ and CD4+ T cell subsets in disease development. Furthermore, we show that blockade of several clinically relevant inflammatory signals delays or prevents human T cell expansion and disease development, further delineating the active molecular pathways in xenogeneic human T cell expansion and supporting the relevance of mechanistic and pharmacological studies in this model.

Initially, we found that injection of human PBMCs resulted predominantly in T cell engraftment with evidence of some B cell or plasma cell engraftment. Thus, this model would be relevant mainly for studies of T cell function, although human B cell maturation, Ig production or autoantibody production could be explored further as readouts, as suggested previously in PBMC-injected mice [18-20]. The initial engraftment of T cells resulted in a rapid and substantial expansion of T cells with a shift in surface marker expression from predominantly naive CD62L+CD45RO T cells towards an activated effector/memory phenotype CD62LCD45RO+, suggesting that processes governing both T cell expansion and activation could be studied in this model. It has been suggested previously that a restricted T cell repertoire emerges in reconstitution of lymphopenic hosts [21, 22]. However, we have established here that the human T cell expansion in this model is polyclonal and largely reflects the donor clonal diversity with minor perturbation in the CD8+ T cell repertoire; this observation is also in accord with reports from allogeneic reactions in humans [23]. Thus, rather than studying a few xenogeneic clones, this model allows studies of polyclonal human T cell expansion and activation.

In early studies of PBMC-injected SCID mice, it was reported that human T cells were anergic to TCR stimulation ex vivo, but responsive to IL-2 [24]; however, we found both TCR- and cytokine- responsive T cells at the time of disease, suggesting the presence of non-anergic effector T cells.

Both CD4+ and CD8+ T cells can contribute to the pathology of allogeneic reactions in mice and humans [25], but the capacity of individual human T cell subsets to expand and induce disease in mice has been unclear. Interestingly, we established that human CD8+ T cells were sufficient for expansion and required for disease induction in mice, which has also been reported in allogeneic mouse GVHD studies [26-28]. In contrast, while purified human CD4+ T cells could engraft and expand in mice, they did not induce acute weight loss, as found recently using irradiated mice [29]. It is likely that irradiation-induced tissue damage could enhance CD4+ T cell activation and pathology, explaining these different observations.

Host APCs are thought to be the core of T cell activation during allogeneic responses [30, 31]. In this study we show that human T cell expansion and activation were maintained using highly purified human T cells, suggesting that human APCs were dispensable for xenogeneic responses and that interaction with host stimulatory molecules/APCs/tissues and T cell intrinsic factors were sufficient for human T cell activation and expansion in mice. This is also supported by studies using mouse MHC class I and II knock-out mice [10].

Defects in regulatory CD4+ T cells (Tregs) have been associated with a range of autoimmune and inflammatory diseases [32-35], and the possibility to study human Treg function and modulations thereof in vivo represents an attractive opportunity. Our results show that Tregs exert moderate control of the xenogeneic T cell reaction, highlighting the possibility for direct studies of human Treg function and modulations of Treg numbers and function in this model. Whether the Tregs present in this model represent natural Tregs or adaptive xeno-specific Tregs remains to be determined. Furthermore, our data on bulk CD4+ T cell depletion suggests that the CD4+ T cell compartment in this model contains not only regulatory, but also helper/effector T cells, as bulk CD4+ T cell depletion neither reduced nor accelerated disease.

While the resulting disease phenotype in human PBMC-injected mice resembles mainly acute GVHD in humans, there are some discrepancies. GVHD in humans is characterized by multi-organ inflammation, predominantly liver damage, skin rash and gut pathology resulting in diarrhoea. We found that immune infiltration in the liver and liver damage were part of the xenogeneic pathology, in accord with human disease. In contrast, we found substantial lung pathology, minor histological changes in skin and kidney and no major changes in gut histology or observations of diarrhoea. The somewhat distinct histopathology of xenogeneic GVHD suggests a preferential homing or activation of human T cells to certain mouse tissues. The particular lack of gut pathology, which is reported in PBMC-injected NOD-SCID mice [36], could be due to the poor development of Peyer's patches (PPs) in IL-2Rγ−/− mice [37] and their importance in gut repopulation and GVHD [38]. Another possibility could be that irradiation, which is inherent to human bone marrow transplantation protocols but not used in this model, could cause the accelerated gut pathology, but this remains to be determined.

In order to understand further the molecular mechanisms behind the human T cell expansion and activation in mice we used several anti-inflammatory compounds with distinct modes of action. CsA and Tac inhibited T cell expansion and disease development significantly, suggesting that calcium signalling via calmodulin–calcineurin was important for human T cell expansion and activation in mice. TNFR-Ig treatment showed attenuation of both T cell expansion and development of disease by sequestering TNF-α in circulation, indicating that TNF-α signalling was partly responsible for driving the human T cell expansion in mice and also suggesting the relevance of this model in mechanistic studies of cytokine-activated T cells. This notion is supported further by a recent study showing that blockade of IL-21 modulated the T cell response and disease development significantly in a similar model [12]. Using CTLA-4-Ig, we found pronounced inhibition of T cell expansion and disease development as preventive treatment, suggesting that co-stimulation, probably via CD28–CD80/CD86 interaction, was essential for initiation of the xenogeneic T cell reaction. Interestingly, we found that CTLA-4-Ig as a therapeutic intervention could contract activated T cells and reverse active disease, indicating the continued requirement for co-stimulation via CD80/CD86 to progress the xenogeneic T cell reaction. In allogeneic mouse GVHD models the use of CTLA-4-Ig has produced mixed responses [39-41], suggesting that a different balance between negative CTLA-4- and positive CD28-signalling exists across GVHD models. The use of antagonistic anti-human CTLA-4 antibodies has shown previously that immunity and T cell expansion were enhanced in human PBMC-engrafted SCID mice [42], suggesting that negative CTLA-4-signalling might also be involved in xenogeneic T cell activation. However, our results suggest that CD28–CD80/CD86 co-stimulation is the dominant signal. Whether CD80/CD86 molecules expressed on engrafting human cells, mouse tissues or both are targeted by the CTLA-4-Ig remains to be shown, and whether the effect of CTLA-4-Ig results in T cell intrinsic or extrinsic inhibitory signals, as discussed recently [43], also remains unclear.

In addition to the mechanistic insights provided by calcineurin inhibitors TNFR-Ig and CTLA-4-Ig treatment in PBMC-injected mice, their effects also reveal that the possible predictive validity of the model for T cell-targeting compounds is not only limited to GVHD-relevant therapeutics. Calcineurin inhibitors are used commonly for the treatment of GvHD, whereas TNFR-Ig is approved for rheumatoid arthritis, although some clinical effects have been reported in GVHD [37, 38]. Early clinical trials in acute GVHD prevention are being conducted currently with CTLA-4-Ig (ClinicalTrials.gov identifier: NCT01012492), but so far CTLA-4-Ig is approved only for rheumatoid arthritis patients with inadequate responses to anti-TNF treatment. Thus, in addition to studies of the mechanisms related to human T cell expansion, activation and regulation and aspects of GVHD pathology, it could be argued that this model might also be useful for evaluation of efficacy and mode-of-action of novel T cell-targeting compounds.

Taken together, we demonstrate several new mechanisms underlying human T cell expansion and activation in humanized mice, including the role of different T cell subsets and specific T cell-signalling pathways. We show the resemblance of this model with acute GVHD in humans, but speculate that the model may have a broader application for studies of T cell activation and effector function and could help to advance novel T cell-targeting therapies.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

We thank Trine Larsen, Rasmus Mark Mortensen and Ella Louise Hjort Kristensen for technical assistance with all animal experiments, Jeanette Juhl for technical assistance with histology and Jesper Damgaard for assistance with liver enzyme measurements. Furthermore, we thank the Animal Unit at Novo Nordisk A/S for the daily maintenance of animals.

Disclosure

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

All authors are employed by Novo Nordisk A/S, but otherwise have no financial interests to disclose.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
cei12051-sup-0001-fs1.tif325K

Fig. S1. Livers and spleens were harvested from human carboxyfluorescein succinimidyl ester (CFSE)-injected NOD.Cg-Prkdcscid Il2rgtm1Sug/JicTac (NOG) mice at the time of 20% weight loss. (a) Liver leucocytes from naive or humanized NOG were isolated using 33·75% isotonic Percoll gradient (Amersham Biosciences) and counted. (b) Liver leucocytes and splenocytes were analysed by flow cytometry for the indicated surface markers. hCD45+ cells are shown as the percentage of live and single cells based on forward/side-scatter (FSC/SSC) properties and exclusion of dead cells by LIVE/DEAD® Fixable Near-IR Dead Cell Staining (Invitrogen). CD3+ and CD19+ cells are shown as the percentage of hCD45+ and CD4+ and CD8+ cells are shown as te percentage of CD3+ cells.

cei12051-sup-0002-fs2.tif138K

Fig. S2. Plasma was collected from groups of human peripheral blood mononuclear cell (PBMC)-injected NOD.Cg-Prkdcscid Il2rgtm1Sug/JicTac (NOG) mice on the indicated days post-injection of PBMCs. Plasma samples were analysed for hIgG using Easy-Titer® human immunoglobulin (Ig)G (H + L) assay kits (Thermo Scientific), according to the manufacturer's instructions. Plasma samples from naive NOG mice were all below detection in the assay (data not shown).

cei12051-sup-0003-fs3.tif3205K

Fig. S3. Splenocytes isolated from human blood mononuclear cell (PBMC)-injected NOD.Cg-Prkdcscid Il2rgtm1Sug/JicTac (NOG) mice at the time of 20% weight loss or untouched human PBMCs (frozen) were labelled with carboxyfluorescein succinimidyl ester (CFSE). Cells were stimulated in vitro for 72 h with plate-bound anti-CD3 and soluble anti-CD28, or the indicated cytokines and proliferation were measured as percentage CFSElow per number of cell divisions. (a) Histograms show the gating of each cell division in unstimulated (left) and stimulated (right) CD4+ and CD8+ T cells from PBMCs. (b) Bar plots show the mean ± standard deviation (s.d.) of the percentage of CFSElow cells per number of cell divisions of cells from PBMCs or humanized spleens in duplicate wells and is representative of two individual experiments. The horizontal bar in (b) indicates the level of CFSE dilution in cells from unstimulated wells. (c) Magnetically purified CD8+ and CD4+ T cells isolated from spleens of humanized mice at the time of graft-versus-host disease (GVHD) were stimulated with the indicated cytokines. Cells were incubated for 72 h and proliferation was assayed by [3H]-thymidine incorporation for the last 18 h. Data were measured on a top-counter (Perkin-Elmer) and the graph shows mean ± standard error of the mean (s.e.m.) of triplicate wells. Representative of four independent experiments.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.