Cytotoxic T lymphocyte antigen‐4 regulates development of xenogenic graft versus host disease in mice via modulation of host immune responses induced by changes in human T cell engraftment and gene expression

Abstract

Graft versus host disease (GvHD) is a major clinical problem with a significant unmet medical need. We examined the role of cytotoxic T lymphocyte antigen‐4 (CTLA‐4) in a xenogenic GvHD (xeno‐GvHD) model induced by injection of human peripheral mononuclear cells (hPBMC) into irradiated non‐obese diabetic (NOD) SCID gamma (NSG) mice. Targeting the CTLA‐4 pathway by treatment with CTLA‐4 immunoglobulin (Ig) prevented xeno‐GvHD, while anti‐CTLA‐4 antibody treatment exacerbated the lethality and morbidity associated with GvHD. Xeno‐GvHD is associated with infiltration of hPBMCs into the lungs, spleen, stomach, liver and colon and an increase in human proinflammatory cytokines, including interferon (IFN)‐γ, tumor necrosis factor (TNF)‐α and interleukin (IL)‐5. Infiltration of donor cells and increases in cytokines were attenuated by treatment with CTLA‐4 Ig, but remained either unaffected or enhanced by anti‐CTLA‐4 antibody. Further, splenic human T cell phenotyping showed that CTLA‐4 Ig treatment prevented the engraftment of human CD45⁺ cells, while anti‐CTLA‐4 antibody enhanced donor T cell expansion, particularly CD4⁺ (CD45RO⁺) subsets, including T box transcription factor TBX21 (Tbet)⁺ CXCR3⁺ and CD25⁺ forkhead box protein 3 (FoxP3) cells. Comprehensive analysis of transcriptional profiling of human cells isolated from mouse spleen identified a set of 417 differentially expressed genes (DEGs) by CTLA‐4 Ig treatment and 13 DEGs by anti‐CTLA‐4 antibody treatment. The CTLA‐4 Ig regulated DEGs mapped to down‐regulated apoptosis, inflammasome, T helper type 17 (Th17) and regulatory T cell (T_reg) pathways and enhanced Toll‐like receptor (TLR) receptor signaling, TNF family signaling, complement system and epigenetic and transcriptional regulation, whereas anti‐CTLA‐4 antibody produced minimal to no impact on these gene pathways. Our results show an important role of co‐inhibitory CTLA‐4 signaling in xeno‐GvHD and suggest the therapeutic utility of other immune checkpoint co‐inhibitory pathways in the treatment of immune‐mediated diseases driven by hyperactive T cells.

We examined the role of Cytotoxic T‐Lymphocyte Antigen‐4 (CTLA‐4) in a mouse model of xenogenic GvHD (xeno‐GvHD) using CTLA‐4 Ig and anti‐CTLA‐4 antibody. We show that CTLA‐4 signaling plays a key role in regulating autoimmune responses in GvHD.

INTRODUCTION

Allogeneic hematopoietic stem cell transplantation (HSCT) can be a potential curative therapy for a variety of malignant and non‐malignant hematological diseases, including autoimmunity [1, 2, 3]. In the tumor setting, the development of graft‐versus‐tumor immune responses elicited by the transferred donor‐derived T cells are beneficial. However, the alloreactivity of these cells against human leukocyte antigen (HLA) disparities and minor antigens can result in the initiation of graft versus host disease (GvHD) [4]. GvHD is a potentially life‐threatening complication that often commences with an acute phase followed by a chronic form of the disease. A clearer understanding of the biological pathways underpinning GvHD can potentially lead to novel therapeutics to many immune‐mediated diseases driven by hyperactive T cells.

T cells play a critical role in the regulation of the immune response against self‐antigens in autoimmune conditions such as rheumatoid arthritis (RA), multiple sclerosis (MS) and GvHD [5, 6, 7]. The activation of naive T cells requires two critical signals: an antigen‐specific signal through the T cell receptor (TCR) complex and a co‐stimulatory signal through CD28 [8, 9].

Cytotoxic T lymphocyte antigen‐4 (CTLA‐4; also known as CD152) is a negative regulator of T cell activation [10] with a key role in regulating peripheral tolerance [11, 12, 13, 14]. CTLA‐4 is a transmembrane glycoprotein and belongs to the CD28/B7 immunoglobulin superfamily [15]. It binds to CD80 and CD86 with a greater affinity and avidity than CD28, thus enabling it to outcompete CD28 for its ligands [16]. CTLA‐4 is expressed transiently on the T cell surface upon activation [17] and delivers an inhibitory signal to T cell activation through a reduction of kinases and transcription factors downstream of TCR activation [e.g. mitogen‐activated protein (MAP) kinases, extracellular signal‐regulated kinase (ERK) and Jun N‐terminal kinase (JNK) and transcription factors such as nuclear factor‐kappa B (NF‐κB), activator protein 1 (AP‐1) and nuclear factor of activated T cells (NF‐AT)]. The net result is the inhibition of cell cycle progression, T cell proliferation and cytokine production [11, 18].

The physiological role of CTLA‐4 signaling in immune homeostasis is evidenced by an autoimmune phenotype in CTLA‐4‐deficient mice, i.e. polyclonal T cell activation, lymphoproliferation and cytokine storm, suggesting that CTLA‐4 functions as a brake to restrain the expansion and activation of self‐reactive T cells and to enforce tolerance [19, 20, 21, 22]. In humans, single nucleotide polymorphisms in CTLA‐4 are associated with a number of autoimmune conditions including RA and systemic lupus erythematosus (SLE) [23]. Thus, targeting the CD28/CTLA‐4 pathway by CTLA‐4 neutralizing antibodies and CTLA‐4 fusion proteins has been explored for the treatment of various forms of cancer and autoimmune diseases, respectively [24, 25, 26]. Blocking of CTLA‐4 causes tumor regression in murine models of cancer [27, 28], and two CTLA‐4 antibodies, ipilimumab (Yevroy™) and tremelimumab [a monoclonal antibody (mAb) in development] showed significant efficacy in human cancers [26, 29, 30]. Although immune checkpoint blockade therapy ushered in the era of cancer immunotherapy, these therapies can trigger autoimmune adverse reactions in susceptible individuals [26].

Fusion proteins of CTLA‐4 (CTLA‐4 Ig) attenuate T cell activation in vivo leading to the maintenance of immune tolerance and inhibition of T cell proliferation and production of interleukin (IL)‐2 [10, 13, 31]. In preclinical studies, CTLA‐4 Ig enhanced long‐term survival of cardiac and pancreatic islet cell allografts [32, 33] and attenuated arthritic inflammation [34, 35]. CTLA‐4 Ig fusion protein, abatacept (Orencia™), is approved to treat RA, psoriatic arthritis and juvenile idiopathic arthritis [36]. Another CTLA‐4 Ig fusion protein, belatacept (Nujolix™), is approved for kidney transplantation [37]. Taken together, these data suggest that the CD28/CTLA‐4 signaling axis plays a critical role in modulating T cell activation and function.

Despite the importance of CTLA‐4 to autoimmunity and anti‐tumor immunotherapy, the precise mechanisms responsible for its function are unknown. Considerable debate exists regarding whether CTLA‐4 inhibits T cell responses by cell‐intrinsic and/or ‐extrinsic mechanisms [16, 17, 18, 19, 20, 21, 22]. Cell‐intrinsic mechanisms would reflect direct effects of the co‐receptor on the expressing cell (i.e. signal transduction), while cell‐extrinsic effects relate to the regulation of function via a distal cell such as a dendritic cell or via a mediator.

GvHD is a clinically important autoimmune condition where donor bone marrow or peripheral blood stem cells sense the recipient’s body as foreign and attack the host tissues [6]. Allogenic‐ GvHD (allo‐GvHD) can be induced in inbred F1 mice by the injection of T cells of parental origin [38]. CTLA‐4 Ig efficiently prevented the development of clinical signs of allogenic GvHD, expansion and activation of donor CD4⁺ and CD8⁺ T cells and inflammatory Th1/Th2 cytokine production [34, 35]. These data, combined with the exacerbation of allo‐GvHD by an anti‐CTLA‐4 antibody, firmly establishes a homeostatic role of CTLA‐4 signaling in GvHD pathogenesis [39].

In recent years, humanized mouse models of GvHD [referred to as xenogeneic‐GvHD (xeno‐GvHD)] have been developed to study new approaches of GVHD prevention. The pathogenesis of xeno‐GvHD remains poorly understood, and the precise role of CTLA‐4 in the context of xeno‐GvHD has not been explored.

Therefore, in the present study, we utilized the xeno‐GvHD model to understand the pathology of GvHD in target organs such as skin, colon, lungs and liver, infiltration and activation of human immune cells and explored the host response. In addition, we evaluated the effects of CTLA‐4 Ig and CTLA‐4 neutralizing mAb on pathophysiology of xeno‐GvHD and conducted transcriptomic analysis to identify novel molecular pathways associated with CTLA‐4 signaling.

MATERIALS AND METHODS

Materials

Antibody–fluorochrome conjugates [CD45RO fluorescein isothiocyanate (FITC), CD3 BV510, CD4 peridinin chlorophyll (PerCp), CD8 allophycocyanin‐cyanin 7 (APC‐Cy7), T‐box gene expressed in T cells (Tbet) AF647, forkhead box protein 3 phycoerythrin (FoxP3 PE), CD25, BV650, CD127 PE‐Cy7, C‐X‐C motif chemokine receptor 3 (CXCR3) BV711, granzyme B PE‐CF954, perforin BV421] were obtained from BioLegend (San Diego, California, USA) or from BD Biosciences (Franklin, New Jersey, USA). CTLA‐4 Ig and anti‐CTLA‐4 mAb were obtained from Bristol‐Myers Squibb Company (Princeton, New Jersey, USA).

Mice

Female non‐obese diabetic (NOD) SCID gamma (NSG) mice (6–8 weeks old; Jackson Laboratories, Bar Harbor, Maine, USA) were maintained under specific‐pathogen‐free (SPF) conditions in filter‐top cages (five mice per cage) with free access to water and diet. All mouse experiments were performed according to the Institutional Animal Care and Use Committee (IACUC)‐approved mouse protocol at Janssen Pharmaceutical Companies of Johnson & Johnson, Spring House, Pennsylvania, and studies were conducted in an American Association for Accreditation of Laboratory Animal Care (AAALAC)‐accredited vivarium.

Irradiation of mice

Animals were randomized by body weight and divided into various treatment groups prior to irradiation. Conscious and freely moving mice received total body irradiation of 100 Rad (1 Gy) using Gammacell^® 3000 Elan Irradiator (9.72 Gy/min). Following irradiation, animals were returned to their home cages with access to food and water.

Preparation of human peripheral blood mononuclear cells (hPBMCs)

Large batches of frozen, healthy donor, hPBMCs isolated from buffy coats were obtained from a commercial source (All Cells, Alameda, California, USA) and cells were stored in 1‐ml aliquots (~1.0–1.5 × 10⁷ cells/tube) in liquid nitrogen. Before injection, frozen cells were quickly thawed at 37°C in a water‐bath and washed three times with sterile phosphate‐buffered saline (PBS) followed by centrifugation at 500  g for 5 min at room temperature. Cells were counted and finally resuspended in cold PBS to obtain a final cell concentration of 4–5 × 10⁷/ml. Patient consent was not applicable to this study.

Induction of GvHD

Twenty‐four hours after irradiation (day 0), mice received 2.0 × 10⁷ hPBMCs/mouse in a volume of 500 µl via intraperitoneal (i.p.) injection. In initial experiments, we conducted a dose–response study with varying numbers of injected hPBMC (1–3 × 10⁷ cells/mouse) to identify the optimum dose of the cells. Irradiated mice receiving PBS alone served as controls. Animals were observed for the emergence of clinical signs of GvHD and weighed daily. The following rating scale was used to characterize GvHD: score 0, normal alert and reactive; score 1, ruffed hair coat, decreased activity, ocular discharge; score 2, hunched posture, moderate hypothermia or hyperthermia (sensed by touch), labored breathing during prodding; score 3, labored breathing during rest, ataxia, tremor, hypothermia or hyperthermia (sensed by touch); and score 4, loss of ability to ambulate with gentle prodding and unconsciousness. Mice that lost ≥ 20% of initial bodyweight or reached to score 4 were humanely euthanized in accordance with institutional IACUC guidelines. Mice terminated prior to the planned termination were given a GvHD score of 5, and this score was carried forward until the end of planned termination. At the termination of the study tissues were harvested for histopathological analyses, and blood samples were drawn for flow cytometry and inflammatory cytokines/mediator analysis.

Treatment with CTLA‐4 Ig and anti‐CTLA‐4 antibody

Mice received CTLA‐4 Ig or anti‐CTLA‐4 mAb at a dose of 10 mg/kg i.p. given once in 3 days (Q3D), with the first dose given 1 h before the hPBMC injection. Control mice received isotype control in a similar fashion.

Serum biochemical analysis

Serum biochemistry determinations were carried out following standard procedures. Specifically, liver function was determined by measuring aspartate aminotransferase (AST) and alanine aminotransferase (ALT) according to a modified method of the International Federation of Clinical Chemistry (IFCC) [40]. The analytical range was 0–1100 U/ml for ALT and 0–1000 U/l for AST. Serum glucose was measured by the hexokinase/glucose‐6‐phosphate dehydrogenase (G6PD) method with the analytical range from 0 to 700 mg/dl [41]. Serum triglyceride was measured by the glycerol‐3‐phosphate oxidase (GPO) method with an analytical range of 0–550 mg/dl [42].

Histopathological analysis

After euthanasia, esophagus, stomach, jejunum, duodenum, colon, rectum, heart, spleen, liver, lung, skin/subcutis, pancreas, kidney, thyroid, adrenal gland and skeletal muscle were harvested and fixed in 10% neutral‐buffered formalin. After 48 h, tissues were transferred to 70% ethanol and subsequently paraffin‐embedded. Approximately 4–5 µm sections from the formalin‐fixed paraffin‐embedded tissue specimens were stained with hematoxylin and eosin (H&E). Histopathological changes were assessed blindly by a board‐certified veterinary pathologist using light microscopy. Tissue sections were graded for mononuclear cell infiltration on semiquantitative scores (0–4), with grade 0 indicating no changes; grade 1, minimal or small changes; grade 2, medium changes; grade 3, moderate changes; and grade 4, extensive changes, relative to controls. Images were acquired using an Olympus VS‐120 scanner or an Olympus light microscope using DP controller software version 3.3.1.292 (Olympus Corporation, Center Valley, Philadelphia, USA).

Colon explant culture

Fresh colon tissues were collected, cut longitudinally and 3‐mm² explants were prepared using a scalpel. Explants were briefly washed in RPMI‐1640 medium [supplemented with 25 mM HEPES, 10% fetal bovine serum (FBS), 1% penicillin/streptomycin and 50 μg/ml gentamycin] and transferred to a 48‐well plate containing 300 µl media/well for incubation at 37°C in a humidified incubator supplemented with 5% CO₂. Supernatants of explant cultures were collected after 24 h for cytokine analysis.

Analysis of human cytokine in colon explants

Human cytokines were measured using the human cytokine/chemokine magnetic bead panel – premixed 41 Plex (Millipore, Billerica, Massachusetts, USA), according to the manufacturer’s instructions. Briefly, 25 µl of supernatants were measured in duplicate using the Bio‐Plex^® 200 systems. The reports automatically generated by Bio‐Plex Manager version 6.1 were reviewed, and only cytokines that were greater than the lower limit of detection and below the saturation value were considered valid. To avoid detection of non‐species‐specific cytokines, the cytokines that have cross‐reactivity when mouse and human cytokine detection panels are applied were excluded.

Analysis of human immune cells by flow cytometry

To phenotype human immune cells by flow cytometry, mouse spleens were harvested from engrafted NSG mice into ice‐cold RPMI‐1640 medium. Spleens were disrupted to make a single cell suspension, followed by lysis of red blood cells (RBCs) in 2 ml of ACK lysing buffer (Thermo Fisher Scientific, Fremont, California, USA) on ice. Immediately after RBC lysis, cells were centrifuged at 300  g for 5 min at 4°C. Supernatants were decanted and splenocytes were resuspended in 2 ml of fluorescence activated cell sorter (FACS) buffer. The cell suspension was filtered through a second 40‐µm filter and live cells were counted after staining with acridine orange/propidium iodide (AO/PI) stain. Following incubation, cells were washed with FACS buffer and centrifuged at 300  g for 5 min. Resulting cells were resuspended in 100 µl/well of FACS buffer containing human Fc block (1:100 dilution of stock) and cell surface staining cocktails (CD45RO FITC, CD3 BV510, CD4 PerCp, CD8 APC‐Cy7, CD25 BV650, CXCR3 BV711, CD127 PE‐Cy7) and incubated for 20 min at room temperature. After incubation, cells were washed and centrifuged at 300  g for 5 min at 4°C. Supernatants were decanted and samples were resuspended in 200 µl FACS buffer before being analyzed on a BD LSR II flow cytometer.

For intracellular staining, cells were resuspended in 100 µl BD Cytofix/fix buffer and incubated for 15 min in the dark at room temperature before being washed twice with ×1 BD perm/wash buffer. Fixed cells were then incubated with 100 µl FACS intracellular stain cocktails (FoxP3 PE, granzyme B PE‐CF594, perforin BV421) for 20 min at room temperature, after which they were washed twice with ×1 perm/wash buffer. Fluorescence minus one controls were included for each stain panel tested. Finally, the cells were resuspended in 150 µl stain buffer before being acquired on the BD LSR II flow cytometer.

Human gene expression in the spleens of GvHD mice

RNA from mouse spleen was isolated using RNeasy Mini Kit (Qiagen, Valencia, California, USA), following the manufacturer’s instructions, and quantitated using Nanodrop (Thermo Fisher Scientific). Gene expression analysis was conducted on the nCounter™ analysis system using the human autoimmune panel from NanoString™ Technologies (Seattle, Washington, USA) on 100 ng RNA per sample. Expression levels were assessed using nSolver™ analysis software to generate an advanced analysis report, following normalization to the geometric mean of positive control spike‐ins and the gene expression of housekeeping genes. This technology offers high levels of precision and sensitivity. Each color‐coded barcode is attached to a single target‐specific probe corresponding to a single gene which can be individually counted without amplification. Any cross‐reactivity to mouse genes was excluded and only the genes that showed statistical significance of p ≤ 0.05 with expression fold changes of ≥ 1.5‐fold over isotype control were counted.

Data analysis

The area under the curve (AUC) of clinical scores and body weight were calculated using trapezoidal rule and with last observation carried over (LOCO) for animals that did not survive to the end of the study. The score AUC values were log₂‐transformed, and their group averages compared using linear models/analysis of variance (ANOVA). Body weight AUCs were analyzed without log transformation. Additionally, the last time‐point analysis of body weight in the time‐course analysis was performed using ANOVA.

Principal components analysis (PCA) was performed on gene expression data for the following six genes: CD40, CPI1, CXCL13, IRF4, CD70 and SOCS3. This multi‐dimensional analysis allowed inference of the combined gene expression data in addition to gene‐by‐gene analysis. PCA used singular value decomposition (SVD) of centered data. A linear transformation was applied to the original data projecting it onto a new, orthogonal space of principal components (PC1–6), such that the first principal component (PC1) was in the direction of most variability in the data, PC2 is in the second most, and so on. The results were then presented graphically as a biplot where the x‐axis represented PC1, the y‐axis, PC2, points, the samples (color‐coded for the treatment groups) and arrows and direction of increased expression for each gene with the length of the arrows being proportional to their contribution. Multinomial regression models with the treatment groups as the multi‐class response variable and the principal components as independent predictors of the classes were then used to quantitatively assess the class separation.

RESULTS

CTLA‐4 Ig treatment dampens whereas anti‐CTLA‐4 antibody exacerbates the development of xeno‐GvHD

In the xeno‐GvHD model, human cells efficiently engraft into various organ and drive tissue‐specific immune‐mediated pathology. These attributes allow for evaluation of the effects of therapeutics on human immune cells/signaling both at the macroscopic (whole organism) and microscopic (tissue/cellular) level. Incidence of mortality and severity of GvHD depends upon the number of hPBMCs injected and the engraftment efficiency. Typically, engraftment of 2.5–3.0 × 10⁷ hPBMCs resulted in 50–60% mortality with 15–20% weight loss by day 21 (data not shown). As this paradigm results in the mortality of many diseased animals by the termination of the study, we decided to engraft NSG mice with lower number of cells (2.0 × 10⁷/mouse) to reduce the incidence of mortality (≤ 20%).

GvHD mice treated with an isotype control exhibited gradual weight loss (Figure 1a,d), approximately 10% mortality (Figure 1c) and progression of GvHD symptoms (Figure 1b,e). CTLA‐4 Ig was effective in preventing the mortality (100% survival), weight loss and development of GvHD (Figure 1). In contrast, compared to isotype control, anti‐CTLA‐4 antibody worsened both GvHD and survival (10 versus 60%). These effects were also mirrored in body weight changes (Figure 1a,d). CTLA Ig‐treated mice (Q3D, last dose on day 18) survived and remained disease‐free until day 65 of the observation period (data not shown).

FIGURE 1.

1 Cytotoxic T lymphocyte antigen‐4 (CTLA‐4) immunoglobulin (Ig) prevents, whereas anti‐CTLA‐4 exacerbates, clinical graft versus host disease (GvHD) in mice. Sublethally irradiated female non‐obese diabetic (NOD) SCID gamma (NSG) mice were transplanted human peripheral blood mononuclear cells (hPBMCs). The body weight (a) and (d), survival (c) and disease score (b) and (e) were recorded daily. Surviving mice were euthanized on day 21. To examine the effect of test molecules, groups of mice were treated with isotype [10 mg/kg intraperitoneally (i.p.)], CTLA‐4 Ig (10 mg/kg i.p.) or anti‐CTLA‐4 (10 mg/kg i.p.) once per 3 days (Q3D). Body weight and clinical score of the mice are presented as mean ± standard error of the mean (SEM); n = 20 mice/group; **p < 0.01; ***p < 0.001; ****p < 0.0001

CTLA‐4 Ig prevents, whereas anti‐CTLA‐4 antibody increases, the infiltration of donor mononuclear cells in peripheral tissues in xeno‐GvHD

Donor cells infiltrate into various tissues during the development of GvHD [38]. We thus evaluated various tissues, including esophagus, stomach, jejunum, duodenum, colon, rectum, heart, spleen, liver, lung, skin/subcutis, pancreas, kidney, thyroid, adrenal gland and skeletal muscles of GvHD mice, for cellular infiltration using H&E staining. Although infiltration of mononuclear cells was detected in all the tissues evaluated, the most affected tissues were lungs (Figure 2b), stomach (Figure 2f), liver (Figure 2j) and colon (Figure 2n). Pulmonary interstitial mononuclear cell infiltrates were seen in animals treated with isotype control, and the level of infiltrating cells in anti‐CTLA‐4 antibody‐treated groups was readily distinguishable from naive controls (Figure 2a,b,d and Figure 3a). In contrast, mice treated with CTLA‐4 Ig prevented the infiltration of mononuclear cells into the lungs (Figures 2, 3). Similarly, in the animals treated with isotype control, mononuclear cell infiltrates were observed in the serosal layer and/or tunica muscularis of stomach (Figure 2f), in the parenchyma of the liver (Figure 2j) and in the lamina propria of the colon (Figure 2n). Anti‐CTLA‐4‐treated mice showed an equal or higher level of mononuclear cell infiltration in stomach liver and the colon (Figures 2 and 3a). In CTLA‐4 Ig‐treated mice, the level of infiltration (Figures 2g,k,o and 3a) was significantly reduced almost to the level seen in naive mice (Figures 2e,i,m and 3a). Overall, the treatment of GvHD mice with CTLA‐4 Ig mitigated donor cell infiltration, whereas anti‐CTLA‐4 antibody either had no effect or enhanced the human immune cell infiltration. A similar observation was made with other tissues examined, such as thyroid, pancreas and subcutis (data not shown). Correlation analysis between the grade of mononuclear cell infiltration in different tissues and clinical score showed a positive correlation, indicating that all these tissue damages were associated with each other and with overall health status of GvHD mice (Figure 3). The highest correlation between mononuclear infiltration and GvHD score was observed for lung and liver (coefficient = 0.92).

FIGURE 2.

Human peripheral blood mononuclear cells (hPBMCs) transplanted non‐obese diabetic (NOD) SCID gamma (NSG) mice were treated with isotype [10 mg/kg intraperitoneally (i.p.)], cytotoxic T lymphocyte antigen‐4 (CTLA‐4) immunoglobulin (Ig) (10 mg/kg i.p.), anti‐CTLA‐4 (10 mg/kg i.p.) once per 3 days (Q3D). Surviving mice were euthanized on day 21 and their lungs, liver, stomach and colon tissues were harvested and stained with hematoxylin and eosin (H&E); n = 5 mice/group. Representative images are shown at ×20 magnification.

FIGURE 3.

Effect of cytotoxic T lymphocyte antigen‐4 (CTLA‐4) immunoglobulin (Ig) and anti‐CTLA‐4 on mononuclear cell infiltration in various tissues and correlation of infiltration with the clinical score in graft versus host disease (GvHD). Mononuclear cell infiltration was evaluated in the hematoxylin and eosin (H&E)‐stained lungs, stomach and liver tissues of naive and isotype control, CTLA‐4 Ig and anti‐CTLA‐4‐treated xeno‐GvHD mice on a scale of 1–4 (a). Correlation matrix between clinical score and mononuclear cell infiltration was plotted using R script with corrplot function. The correlation coefficients were either shown by color intensity and the size of the circle at upright or by numbers listed in the left; n = 5 mice/group.

CTLA‐4 Ig but not anti‐CTLA‐4 inhibits serum liver enzyme levels in xeno‐GvHD

In addition to histopathological changes observed in the liver, serum AST and ALT were also measured to assess GvHD‐associated liver damage. Significant increases in both enzymes were observed in mice that received isotype control (Figure 4a,b). CTLA‐4 Ig attenuated amounts of both enzyme levels to near baseline, while anti‐CTLA‐4 antibody did not affect these parameters versus isotype (Figure 4a,b). Serum glucose and triglyceride levels were reduced in isotype‐treated mice relative to naive mice; these were normalized by the CTLA‐4 Ig treatment and unaffected by anti‐CTLA‐4 (Figure 4c,d). Changes in alkaline phosphatase (ALP) and gamma glutamyl transferase (GGT), potential markers of hepatobiliary injury, were inconsistently altered (data not shown). Bile acid levels were not measured in the sera. In addition, definitive histopathological evidence for biliary injury was lacking. As such, the changes in this acute setting are suggestive of hepatic injury. It is possible that in a chronic setting the hepatic injury seen in this study could evolve into hepato‐biliary injury.

FIGURE 4.

Serum levels of alanine aminotransferase (ALT) and aspartate aminotransferase (AST) in graft versus host disease (GvHD) mice treated with cytotoxic T lymphocyte antigen‐4 (CTLA‐4) immunoglobulin (Ig) and anti‐CTLA‐4. Serum levels of AST (a), ALT (b), glucose (c) and tryglyceride (d) were measured on day 21 in GvHD mice treated with isotype [10 mg/kg intraperitoneally (i.p.)], CTLA‐4 Ig (10 mg/kg i.p.) and anti‐CTLA‐4 (10 mg/kg i.p.) once per 3 days (Q3D). Data are presented as mean ± standard error of the mean (SEM); n = 8–10 mice/group; *p < 0.05; **p < 0.01; ***p < 0.001

CTLA‐4 Ig treatment inhibits inflammatory cytokines in sera from xeno‐GvHD mice

Cytokines produced during immune activation play a critical role in tissue damage and perpetuating T cell activation/migration in GvHD. T helper type 1 (Th1) cells secrete interferon (IFN)‐γ and tumor necrosis factor (TNF)‐α), while Th2 cells secrete IL‐4, IL‐5 and IL‐13. IFN‐γ is known to stimulate the proliferation and activation of CD8 T cells. As shown in Figure 5a–c, human IFN‐γ, TNF‐α and IL‐5 were elevated in the serum of GvHD mice. CTLA‐4 Ig treatment completely inhibited these changes, whereas anti CTLA‐4 treatment had no effect. Granulocyte–macrophage colony‐stimulating factor (GM‐CSF), sCD40L and macrophage inflammatory protein (MIP)‐1β also showed a similar pattern to TNF‐α, IL‐5 and IFN‐γ (data not shown). Levels of TGF‐α, IL‐12, IL‐1α, IL‐1β, IL‐2, IL‐4, IL‐8 and vascular endothelial growth factor (VEGF) remained below the detection limit in all groups (data not shown). IL‐6, IL‐7, VEGF and eotaxin, however, were detected at very low concentrations (below 10 pg/ml) with no effect of any of two treatments (data not shown).

FIGURE 5.

Cytokine levels in graft versus host disease (GvHD) mice treated with cytotoxic T lymphocyte antigen‐4 (CTLA‐4) immunoglobulin (Ig) and anti‐CTLA‐4. Serum levels of human tumor necrosis factor (hTNF)‐α (a), human interleukin (hIL)‐5 (b), human interferon (hIFN)‐γ (c), human regulated on activation, normal T cell expressed and secreted (hRANTES) (d) and colon explant RANTES (e) were measured on day 21 in GvHD mice treated with isotype [10 mg/kg intraperitoneally (i.p.)], CTLA‐4 Ig (10 mg/kg i.p.) and anti‐CTLA‐4 (10 mg/kg i.p.). Data are presented as mean ± standard error of the mean (SEM); n = 8–10 mice/group; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

Previous studies have shown that expression of regulated on activation, normal T cell expressed and secreted (RANTES) or C‐C chemokine receptor type 5 (CCR5) in donor T cells plays a critical role in their accumulation into lymphoid tissues after allogeneic transplantation [43, 44]. We thus examined the effect of CTLA‐4 targeting on serum RANTES levels in GvHD. CTLA‐4 Ig completely abrogated the induction of serum RANTES levels. In contrast, anti‐CTLA‐4 antibody treatment caused a significant increase over the isotype control response (Figure 5d).

Activated donor T cells traffic and cause organ‐specific cytotoxicity in various host tissues, and one such organ is the colon. To explore if the CTLA‐4 axis affects colonic inflammation, colon explants from GvHD mice treated with isotype, CTLA‐4 Ig or anti‐CTLA‐4 antibody were cultured for 24 h. Like serum, a significant increase of RANTES was observed in media from the colon explants (Figure 5e). CTLA‐4 Ig inhibited this chemokine response, whereas anti‐CTLA‐4 antibody caused a further increase over that seen with the isotype.

CTLA‐4 Ig prevents where anti‐CTLA‐4 antibody increases T cell engraftment in xeno‐GvHD

We evaluated the effects of CTLA‐4 Ig and anti‐CTLA‐4 antibody treatments on donor T cell engraftment and activation status of the T cells in the spleens of surviving GvHD mice on day 21. The gating strategy for the enumeration of engrafted human cells is shown in Figure 6a.

FIGURE 6.

Effect of cytotoxic T lymphocyte antigen‐4 (CTLA‐4) immunoglobulin (Ig) and anti‐CTLA‐4 on human T cell engraftment in the spleen of graft versus host disease (GvHD) mice. Human PBMC‐transplanted non‐obese diabetic (NOD) SCID gamma (NSG) mice were treated with isotype [10 mg/kg intraperitoneally (i.p.)], CTLA‐4 Ig (10 mg/kg i.p.) and anti‐CTLA‐4 (10 mg/kg i.p.) once per 3 days (Q3D). Human CD45RO (hCD45RO) + cell engraftment and T cell phenotyping was examined on day 21 by flow cytometry. Splenocytes were stained for surface expression of human CD3, CD4, C‐X‐C motif chemokine receptor 3 (CXCR3) and CD25. Cells were also stained to measure intracellular expression of human granzyme B, perforin and forkhead box protein 3 (FoxP3). Gating strategy (a) and frequencies of T cell phenotypes (b–h) are described. Data are presented as mean ± standard error of the mean (SEM); n = 4–10 mice/group; *p < 0.05; **p < 0.01; ***p < 0.001. CTLA‐4 Ig and anti‐CTLA‐4 antibody differentially affect T cell engraftment resulting in lower CD45RO⁺ (or hCD3⁺) cells/spleen in CTLA‐4 Ig versus isotype and higher CD45RO⁺ (or hCD3⁺) cells/spleen versus isotype (b,c), respectively. Two types of normalization were used. Fluorescence activated cell sorter (FACS) plots (a) are shown in which subpopulations are represented as a fraction of CD45RO⁺ cells in each treatment group. (d–f) The subpopulations are shown as a fraction of live single‐positive (hCD4) cells.

Flow cytometric analysis of splenic cells from isotype control mice showed engraftment of human T cells (CD45RO ⁺/CD3⁺) (Figure 6b,c). Administration of CTLA‐4 Ig reduced (although not statistically significantly) donor T cell [human CD45RO⁺ (hCD45RO⁺)/CD3⁺] expansion compared with the isotype control group (Figure 6b,c). In contrast, blockade of CTLA‐4 by anti‐CLTA‐4 antibody greatly enhanced donor CD45RO⁺/CD3⁺ T cell expansion (Figure 6b,c).

CTLA‐4 Ig and anti CTLA‐4 antibody differentially affect the T cell engraftment, resulting in lower CD45RO⁺ [or human CD3⁺ (hCD3⁺] cells/spleen in CTLA‐4 Ig versus isotype and higher CD45RO⁺ (or hCD3⁺) cells/spleen versus isotype (Figure 6b,c), respectively. Therefore, the subset analysis differs depending on the normalization factors used. Two types of normalization were used. FACS plots (Figure 6a) are shown in which subpopulations are represented as a fraction of CD45RO⁺ cells in each treatment group. For Figure 6d–f, the subpopulations are shown as a fraction of live single‐positive (hCD4) cells.

Further analysis of T cell activation in various subsets showed a significant reduction in memory CD8⁺ T cells (CD45RO⁺/CD8⁺) by CTLA‐4 Ig compared with isotype control (Figure 6d). In contrast, anti‐CTLA‐4 antibody treatment increased the engraftment of both CD4⁺ and CD8⁺ memory T cells, as shown in Figure 6d,e. Although not statistically significant, a marginal increase in regulatory memory CD4 T cells (CD45RO⁺/CD25⁺/FoxP3⁺) was also observed with both CTLA‐4 Ig and anti‐CTLA‐4 compared with the isotype control (Figure 6f). CTLA‐4 Ig did not significantly impact memory CD4 T cells compared with the isotype control (Figure 6e). Cytotoxic lymphocytes are critical for eliminating virus‐infected or transformed cells through inducing apoptosis by activating the perforin/granzyme pathway [45]. Thus, donor CD8⁺ cells were enumerated for the presence of cytotoxic molecules, granzyme and perforin. Granzyme/perforin double‐positive CD8 memory cells were detected in isotype‐treated mice and these were significantly decreased by CTLA‐4 Ig, but slightly increased by anti‐CTLA‐4 antibody (Figure 6g).

CD4 T cells perform multiple cellular functions through expression of transcription factors that initiate distinct programs of gene expression, including chemokine receptor CXCR3, expression of which is increased upon activation. We therefore examined the effect of CTLA‐4 pathway on the expression of Tbet and CXCR3 on CD4⁺ T cells. Although no statistical significance was observed by one‐way ANOVA compared to the isotype control, CTLA‐4 Ig treatment showed a trend towards a decreased (Figure 6h), whereas the anti‐CTLA‐4 Ig increased, the Tbet⁺/CXCR3⁺ CD4 cell population (Figure 6h).

Differential effects of CTLA‐4 Ig and anti‐CTLA‐4 antibody on human gene expression in xeno‐GvHD

To gain further mechanistic insights into immune activation and impact of CTLA‐4 pathway modulation, we compared gene expression profiles in human cells isolated from spleens of GvHD mice treated with isotype, CTLA‐4 Ig or anti‐CTLA‐4 (Supporting information, S1–S2). For these studies we utilized the NanoString™ platform (human immune panel; 770 genes), a novel digital color‐coded barcode technology based on direct multiplexed measurement of gene expression [46]. This technology offers a high level of precision and sensitivity for mRNA quantitation without amplification. Human gene‐specific probes with unique color‐coded barcode are used for mRNA quantitation and can be individually counted without amplification [47]. Probes with any cross‐reactivity to mouse genes were excluded from the analysis. Heat‐map depiction showed an overlap of gene expression between isotype control and anti CTLA‐4 and a profile for CTLA Ig that is distinct both from the isotype control and anti‐CTLA‐4 treatments (Figure 7a). We next plotted the most differentially expressed genes after treatment with CTLA‐4 Ig or anti‐CTLA‐4 antibody versus isotype (Figure 7b–g). Compared to isotype control, expression of IRF4, SOCS3 and CD70 genes was significantly down‐regulated, while CD24 and SPl1 genes were up‐regulated by CTLA‐4 Ig. The anti‐CTLA‐4 antibody had an opposite effect on these selected genes.

FIGURE 7.

Clustering and expression of differentially expressed genes in the spleen of cytotoxic T lymphocyte antigen‐4 (CTLA‐4) immunoglobulin (Ig) and anti‐CTLA‐4‐treated mice in graft versus host disease (GvHD). Human PBMC transplanted non‐obese diabetic (NOD) SCID gamma (NSG) mice were treated with isotype [10 mg/kg intraperitoneally (i.p.)], CTLA‐4 Ig (10 mg/kg i.p.) and anti‐CTLA‐4 (10 mg/g i.p.) once per 3 days (Q3D). Surviving mice were euthanized on day 21 and spleens were harvested. Differential expression (a–g), Venn diagram (f), principal component (PC) plot (i) of human genes in the engrafted human cells in mouse spleens of isotype control, CTLA‐4 Ig and anti‐CTLA‐4 was analyzed using nSolver advanced analysis; n = 6–10 mice/group; *p < 0.05; **p < 0.01; ****p < 0.0001

The gene expression similarities and differences between CTLA‐4 Ig versus isotype and anti‐CTLA‐4 antibody versus isotype was then compared using a Venn diagram analysis. CTLA‐4 Ig versus isotype treatments showed that 417 genes were differentially expressed (DEGs, p < 0.05, > 1.5‐fold increase or decrease over isotype expression). In contrast, only 13 genes were differentially expressed (p < 0.05, > 1.5‐fold increase or decrease over isotype expression) when comparing anti‐CTLA‐4 treatment to isotype control. Among the DEGs regulated by CTLA‐4 Ig and anti‐ CTLA‐4 treatment, six genes: IRF4, SOC3, CXCL13, CD70, CD24 and SPI1, were shared (Figure 7h). Among these six genes, CD24 appeared to be up‐regulated by both CTLA‐4 Ig and anti‐CTLA‐4 treatments.

The PCA analysis showed a clear separation between the three treatment groups: the isotype, CTLA‐4 Ig and anti‐CTLA‐4 (Figure 7i). The separation was driven by higher expression levels of SPI1 and CD24 and lower expression of CXCL13, IRF4, CD70 and SOCS3 in anti‐CTLA‐4‐treated samples compared to the isotype and CTLA‐4 Ig‐treated samples. Multinomial regression with PC1 as the only predictor correctly classified all eight isotype samples in nine of 10 CTLA‐4 Ig and five of six anti‐CTLA‐4 samples. Adding PC2 as a predictor improved the model sensitivity by classifying all samples correctly.

To further explore the biology of CTLA‐4 Ig treatment, a pathway enrichment analysis for the 417 DEGs regulated by CTLA‐4 Ig and the 13 DEGs impacted by anti‐CTLA‐4 antibody (Figure 8a) was performed. For each pathway, an expression score was calculated for each sample to illustrate the impact of treatment on the expression of a group of genes associated with a biological pathway. The clustering analysis of the pathways (Figure 8a) is consistent with the heat‐map analysis of single genes: isotype control and anti CTLA‐4 induced gene expression that mapped to similar signaling pathways, while the CTLA‐4 Ig‐induced gene expression and associated pathways showed a distinct, non‐overlapping pattern (Figure 8b). As the CTLA‐4 Ig treatment yielded the most gene expression shift compared to isotype controls, further analysis was performed on the impacted genes. This analysis revealed that apoptosis, inflammasome, Th17 and regulatory T cell (T_reg) pathways were down‐regulated, whereas TLR receptor signaling, TNF family signaling, complement system and epigenetic and transcriptional regulation were up‐regulated by CTLA Ig. In contrast, anti‐CTLA‐4 antibody treatment produced minimal to no impact on gene pathways relative to isotype (Figure 8b).

FIGURE 8.

Differentially expressed gene pathways in the spleen of cytotoxic T lymphocyte antigen‐4 (CTLA‐4) immunoglobulin (Ig) and anti‐CTLA‐4‐treated mice in graft versus host disease (GvHD). Human PBMC transplanted non‐obese diabetic (NOD) SCID gamma (NSG) mice were treated with isotype [10 mg/kg intraperitoneally (i.p.)], CTLA‐4 Ig (10 mg/kg i.p.) and anti‐CTLA‐4 (10 mg/kg i.p.) once per 3 days (Q3D). Surviving mice were euthanized on day 21 and their spleens were harvested. Total RNA was isolated from mouse spleens that had human cells engrafted into the tissue. The cells were not sorted to isolate human populations. Differential expression of human genes pathways between isotype control splenocytes and CTLA‐4 Ig and anti‐CTLA‐4 was analyzed; n = 6–10 mice/group.

DISCUSSION

Effective T cell activation is mediated by a productive interaction between TCR and MHC and obligatory support from a co‐stimulatory signal via the B7/CD28 axis. CTLA‐4 is constitutively expressed on T_regs, but only up‐regulated in conventional T cells 24–48 h after cellular activation. CD80 and CD86, expressed on several antigen‐presenting cells (e.g. Langerhan’s cells, macrophages, dendritic cells and B cells; are the cognate ligands for both CD28 and CTLA‐4 [18, 48]. However, CTLA‐4 binds to CD80/86 with a greater affinity and avidity than CD28, thus enabling it to outcompete CD28 for its ligands [16]. As such, CTLA‐4 engagement is an effective immune inhibitory checkpoint mechanism for self‐tolerance and immunological homeostasis.

Aberrant T cell activation plays a central role in the pathogenesis of several autoimmune diseases, including RA, SLE and GvHD. Further, a lack of T cell infiltration into tumors coupled with a compromised immune response against tumor antigens can lead to neoplasms. The xeno‐GvHD model induced by the transfer of hPBMCs into NSG mice, resulting in engraftment of donor cells into the host, activation and expansion of human T cells triggering a multi‐organ pathology [48, 49, 50], is one of the few tools currently available to study human T cell functions in an in‐vivo setting. In the present study, we used a xeno‐GvHD model to examine the impact of inhibition or activation of the CTLA‐4 pathway on human immune cell engraftment, organ‐specific pathology and T cell activation/signaling.

Our results demonstrate differential effects of administration of CTLA‐4 Ig and anti‐CTLA‐4 antibody on the pathophysiology of xeno‐GvHD and associated gene expression profiles that map to distinct biochemical pathways. To the best of our knowledge, no such attempts have been made in the literature to compare the effects of CTLA‐4 Ig and anti‐CTLA‐4 antibodies at pathological and molecular levels, side by side, in the xeno‐GvHD setting.

CTLA‐4 Ig [Orencia™ (abatacept)] binds to B7 with high affinity/avidity and blocks T cell‐dependent immune responses, including graft rejection [34, 35, 38, 44, 46]. Conversely, anti‐CTLA‐4 [Yervoy™ (ipilimumab)] augments T cell activation and enhances T cell responses against tumor antigens due to the abrogation of CTLA‐4‐induced inhibitory signals [45]. Administration of CTLA‐4 Ig at the time of hPBMC transfer inhibited the development of GvHD parameters such as disease score, weight loss and moribundity/mortality, while administration of anti‐CTLA‐4 antibody exacerbated the development of GvHD. The salutary effects of CTLA‐4 Ig in the model are consistent with the established role of CTLA‐4 as a critical regulator of T cell activation. While cell‐intrinsic mechanisms of CTLA‐4 have been proposed to explain its immunological effects [16, 17, 18, 19, 20, 21, 22], Bachman et al. [51] reported a cell‐extrinsic pathway for CTLA‐4 via the demonstration that recombinase activating gene 2 (Rag2)‐deficient mice reconstituted with a mixture of wild‐type and CTLA‐4‐deficient bone marrow cells failed to develop autoimmune disease, while the transfer of CTLA‐4^−/− bone marrow cells alone transferred disease. Additional studies by Qureshi et al. [52] and Wang et al. [53] further support cell extrinsic mechanisms for in‐vivo actions of CTLA‐4. The effects of CTLA‐4 Ig observed in this investigation are likely to involve cell extrinsic mechanisms, potentially at the level of the host DC:human T cell interface, leading to attenuated T cell activation.

Xeno‐GvHD‐induced organ‐specific injury requires engraftment of injected human cells [37]. We characterized human immune cells in the spleen on day 21 of the study as a representation of tissue engraftment. Immunophenotyping of splenic cells on day 21 showed an exclusive presence of donor T cells. The relative lack of human B cells at this time‐point is probably the result of poor survival of B cells in this setting [54, 55]. In preliminary time‐course studies involving the transfer of hPBMCs into NSG recipient mice, we detected up to ~15% of human CD45⁺CD19⁺ B cells engrafted in the spleen after 14 days post‐transfer followed by a rapid decline. Engrafted B cells around day 14 were mainly of activated phenotype and included CD27⁺IgD⁻ memory B cells and CD27⁻IgD⁻IgM⁻CD138⁺ plasma cells (data not shown).

Immunophenotyping of the splenic human leukocyte population further revealed differential effects of CTLA‐4 Ig versus anti‐CTLA‐4 antibody. The CTLA‐4 Ig treatment showed a trend towards decreased leukocyte engraftment (reductions in hCD45⁺ or hCD3⁺ cells), markedly reduced effector memory CD8 T cells (CD45RO⁺CD8⁺), cytotoxic (granzyme⁺) CD8⁺ T cells and T box transcription factor TBX21 (Tbet)⁺CXCR3⁺CD4 T cells. The net effect is a reduction in leukocyte engraftment and attenuated T cell activation/survival leading to reduced effector and cytotoxic T cells. In contrast, anti‐CTLA‐4 antibody increased leukocyte engraftment and increased the frequency of effector memory CD4⁺ T cells and Tbet⁺CXCR3⁺CD4 T cells. Inflammatory cytokines induce Tbet expression in a graded manner which, in turn, regulates a distinct differentiation program. For example, Tbet can act as a fulcrum between Th1 and Tfh cell differentiation, pathogenic and non‐pathogenic Th17 cells and CD8 effector and memory T cells. A marked increase in Tbet⁺CXCR3⁺CD4 T cells in CTLA‐4 Ig‐treated mice supports exaggerated effector/memory T cell functions following CTLA‐4 blockade.

In preliminary studies, we observed that a significant proportion of CD3⁺ T cells in the spleens of GvHD mice to be CD4–CD8 double‐positive. The presence of double‐positive T cell subsets in the peripheral blood and tissues have been reported in a variety of pathological conditions, including RA, atopic dermatitis and GvHD [56, 57, 58]. Hussen et al. [56] reported the presence of double‐positive tissue‐infiltrating T lymphocytes in the liver and lungs of GvHD mice, suggesting that these cells might be selectively recruited, differentiated and amplified in response to antigenic exposure in the tissues. As the focus of the current study was on CD4 and CD8 single‐positive T cells, the effects of CTLA‐4 Ig or CTLA‐4 antibody have not been explored in the double‐positive population, and additional experimentation is needed to fully elucidate the role of the double‐positive cells in the pathobiology of GvHD.

CTLA‐4 Ig inhibited both acute and chronic allo‐GvHD by preventing the activation and expansion of T cells [59] and IL‐2 production by donor CD4⁺ T cells [60]. Conversely, anti‐CTLA‐4 mAb enhanced the activation and expansion of superantigen‐specific T cells in response to priming [61] and worsened GvHD. Our xeno‐GvHD results are consistent with previous results using allogenic GvHD models. Sondergaard et al. [55] have demonstrated that CTLA‐4 Ig treatment abrogated the development of disease and human cell expansion in GvHD mice. Thus, increased expansion, activation of donor T cells and their infiltration into target tissues might be a potential mechanism for the worsening of GvHD caused by anti‐CTLA‐4 treatment in this study. Anti‐CTLA‐4 is known to enhance T cell‐mediated autoimmune diseases such as murine models of encephalitis [62], diabetes [63], myasthenia gravis [64] and allogenic GvHD [39]. In allogenic GvHD, exaggerated immune activation by anti‐CTLA‐4 appears be mediated by CD8⁺ T cells [39, 65]. These phenotypical differences in T cells could contribute to strong protection against GvHD by CTLA‐4 Ig and its worsening by CTLA‐4 antibody.

GvHD is accompanied by organ‐specific pathologies driven by engraftment of donor cells into these tissues and their unrestrained activation. The multi‐organ pathology most seen is in the lung, gastrointestinal tract, skin and the liver [49, 50, 66]. Inflammatory cytokines and chemokine milieu can drive activation and migration of immune cells resulting in tissue damage. In our study, a profound influx of mononuclear cells into various tissues was seen, with the highest levels of influx into the lungs, stomach, liver and the colon. CTLA‐4 Ig significantly reduced mononuclear cells in several host tissues, whereas CTLA‐4 antibody had no effect and, in some instances, showed a qualitative trend for an increase. While additional studies are needed for definite immunophenotyping of mononuclear cells, earlier work by Burlion et al. [67] demonstrated numerous infiltrates of mononuclear cells (largely T cells) in lung and liver tissue in the xeno‐GvHD model. The cellular infiltration into the liver was associated with elevations in liver enzymes, AST and ALT, suggestive of hepatic injury. Clear evidence for hepatobiliary injury is lacking.

Sera from GvHD sera contained detectable levels of several human proinflammatory cytokines, and the CTLA‐4 Ig treatment attenuated T cell‐derived cytokines such as IL‐5, IFN‐γ and TNF‐α. These data are consistent with elevations in murine Th1/Th2 cytokines in an allogeneic GvHD model and their reductions following CTLA‐4 Ig treatment [59, 67]. Another key finding of our report is the detection of significant levels of RANTES in the serum of GvHD mice and in the incubation media from colon explants of GvHD mice. Previous studies have shown that expression of RANTES or CCR5 in donor T cells plays a critical role for their accumulation into lymphoid tissues after allogeneic transplantation [43, 44]. In an allo‐GvHD model in mice, eliminating the expression of a CCR5 ligand, CCL3, from the donor T cells resulted in reduced CD8⁺ T cell accumulation in the spleen [44]. Furthermore, the CCL5:CCR1 interaction also contributes to target organ injury, as the blockade of this interaction with a CCR5 antibody resulted in suppression of alloreactive T cell activation, leading to decreased liver and intestinal injury [68].

Our study demonstrated that CD28–CD80/CD86 interactions play an important role in GvHD; however, molecular mechanisms are not fully understood. To this end, we determined unique transcripts affected by treatment with CTLA‐4 Ig or anti‐CTLA‐4 antibody in xeno‐GvHD using the NanoString™ platform. CTLA‐4 Ig and anti‐CTLA‐4 antibody treatments resulted in distinct gene expression profiles, and 417 DEGs in CTLA‐4 Ig (versus control) and 13 DEGs in anti‐CTLA‐4 antibody (versus control) and six overlapping genes in both groups were identified. Among these six genes, interferon regulatory factor 4 (IRF4) [69], suppressor of cytokine signaling (SOCS) 3 [70], CD70 [71], CXCL13 [72] and CD24 [73] are associated with T cell signaling and differentiation, whereas SPl1 is responsible for epigenetic and transcriptional regulation [69]. We found that IRF4, SOCS3 and CD70 transcripts were significantly decreased, whereas transcripts of CD24 and SPl1 were enhanced by CTLA‐4 Ig. Anti‐CTLA‐4 treatment, however, increased expression of IRF4, SOCS3 and CD70. CTLA‐4 Ig treatment was associated with decreased expression of genes that regulate apoptosis, T_reg and Th17 differentiation, TLR and TNF family signaling, and thereby complement linking changes in co‐stimulatory signaling to impaired activation, proliferation and decreased abundance.

When we explored the relationships between clinical score versus gene expression and serum cytokines, a relatively lower shift between the anti‐CTLA‐4 group versus the isotype control group was observed compared with the overall clinical outcome. This may be due to the data inclusion criteria of animals at different end‐points. For the clinical GvHD score, mice that died before day 21 or were euthanized due to weight loss of > 20% were included (i.e. last observation carried over) and given the highest clinical score. However, only animals surviving on day 21 were included in histology, serum cytokine and gene expression analysis and the mice that died or were euthanized before the end‐point (day 21) were not included in the analysis. This means that the mice with most severe disease (based on the survival curve), almost exclusively in the anti‐CTLA‐4 group, were not included in the sample analysis. A detailed time‐course of xeno‐GvHD would provide clearer insight into the clinical outcomes as well as differential gene expression.

In summary, CTLA‐4 Ig treatment prevents the development of xeno‐GvHD, whereas anti‐CTLA‐4 antibody exacerbates the disease, and these differential profiles are reflected in contrasting effects on leukocyte engraftment, effector T cell phenotype, inflammatory cytokine milieu and finally distinct gene expression profiles. We believe that our results contribute to a clearer understanding of CD28/CD80/86 signaling in the pathophysiology of GvHD. Furthermore, results obtained from these studies may facilitate the identification of the novel and previously unrecognized functional role of genes and biological pathways to develop better clinically translatable strategies for the treating of T cell‐mediated diseases by targeting the co‐stimulatory pathway.

CONFLICTS OF INTEREST

The authors are/were employees of Janssen Research and Development. There are no other conflicts of interest.

CLINICAL TRIAL REGISTRATION

Not applicable.

AUTHOR CONTRIBUTIONS

Study conceptualization, data analysis, interpretation, manuscript preparation: C.G., M.‐C.T., T.R., R.M. Experimental conduct, methods, initial data analysis: C.G., D.G., D.S., M.C., M.‐C.T., S.H., H.D. Histopathology: A.C. Manuscript writing, reviewing and editing: M.C.‐T., T.R., R.M. All authors had access to the data and reviewed the manuscript.

Supporting information

Table S1

Table S2

Gao C, Gardner D, Theobalds M‐C, Hitchcock S, Deutsch H, Amuzie C, et al. Cytotoxic T lymphocyte antigen‐4 regulates development of xenogenic graft versus host disease in mice via modulation of host immune responses induced by changes in human T cell engraftment and gene expression. Clin Exp Immunol. 2021;206:422–438. 10.1111/cei.13659

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.