Research ArticleType I Diabetes

Teplizumab Induces Human Gut-Tropic Regulatory Cells in Humanized Mice and Patients

See allHide authors and affiliations

Science Translational Medicine  25 Jan 2012:
Vol. 4, Issue 118, pp. 118ra12
DOI: 10.1126/scitranslmed.3003401


The development and optimization of immune therapies in patients has been hampered by the lack of preclinical models in which their effects on human immune cells can be studied. As a result, observations that have been made in preclinical studies have suggested mechanisms of drug action in murine models that have not been confirmed in clinical studies. Here, we used a humanized mouse reconstituted with human hematopoietic stem cells to study the mechanism of action of teplizumab, an Fc receptor nonbinding humanized monoclonal antibody to CD3 being tested in clinical trials for the treatment of patients with type 1 diabetes mellitus. In this model, human gut-tropic CCR6+ T cells exited the circulation and secondary lymph organs and migrated to the small intestine. These cells then produced interleukin-10 (IL-10), a regulatory cytokine, in quantities that could be detected in the peripheral circulation. Blocking T cell migration to the small intestine with natalizumab, which prevents cellular adhesion by inhibiting α4 integrin binding, abolished the treatment effects of teplizumab. Moreover, IL-10 expression by CD4+CD25highCCR6+FoxP3 cells returning to the peripheral circulation was increased in patients with type 1 diabetes treated with teplizumab. These findings demonstrate that humanized mice may be used to identify novel immunologic mechanisms that occur in patients treated with immunomodulators.


In recent years, the number of immunomodulatory agents available to treat human disease has grown exponentially (13). These drugs typically target immunological mechanisms that have been studied in rodent preclinical models of human disease combined with studies of human cells in vitro. Based on these studies, early-stage clinical trials have been undertaken to test the safety and efficacy of treatment in patients. This approach requires many years to optimize a successful therapy, is costly, and potentially has risks for patients (47). Studies of the mechanism of action of these reagents using samples of peripheral blood often do not confirm observations in the preclinical studies that led to the development of the biologic. The reasons for this disparity include differences in murine and human immune responses as well as lack of access to tissue compartments where important immune response may be occurring in patients (8).

Preclinical studies in the nonobese diabetic (NOD) mouse model originally demonstrated that treatment with a brief course of anti-CD3 monoclonal antibody (mAb) at onset of diabetes induced sustained remission by induction of tolerance to pancreatic β cells (9, 10). These studies identified a reduction in T cells followed by the restoration of tolerance by the induction and activation of a T regulatory population (11). In mice, the remission from diabetes is complete and sustained, precluding the need for repeat administration (10). Similarly, in humans treated with Fc receptor (FcR) nonbinding mAb to CD3, there was a rapid reduction in T cells at initiation of anti-CD3 treatment followed by a rapid recovery of T cells after treatment (12, 13). This reduction and reappearance in T cells in humans may be due to not only depletion but also migration of T cells from the peripheral circulation (14). In addition to the trafficking of T cells, non-FcR nonbinding anti-CD3 mAbs have been found in mice to induce adaptive regulatory T cells (Tregs) that express CD25 and function through a transforming growth factor–β (TGF-β)–dependent mechanism (9).

Humanized monoclonal anti-CD3 antibodies teplizumab [hOKT3γ1(Ala-Ala)] and otelixizumab (ChAglyCD3) have shown efficacy in early human clinical trials for the treatment of new-onset type 1 diabetes (T1D) (13, 15, 16). However, further phase 2 clinical trials of teplizumab and otelixizumab have had mixed results. Teplizumab has been shown to be effective in the preservation of endogenous insulin production at a high dose, whereas otelixizumab was ineffective at a low dose (17, 18). Furthermore, analysis of peripheral blood from study subjects has failed to identify human Treg populations similar to those found in mice. IL-10+CD4+ T cells, which were not described in the preclinical rodent models, have been identified transiently in the peripheral circulation of treated patients, but their location and the way in which these cells are generated are unclear (19). Understanding the mechanisms of action of these mAbs in humans is essential for the effective design and implementation of future clinical trials.

Therefore, to clarify the mechanism of action of teplizumab on human cells in vivo, we used a tolerogenic humanized mouse model to study the effects of the drug (20). The use of the humanized mouse engrafted with human hematopoietic stem cells allows for the reconstitution of human immune systems without the development of graft-versus-host responses that would preclude the evaluation of immunomodulators (21, 22). Moreover, the humanized mouse permits access to tissue compartments and secondary lymph organs that are unavailable in humans for study. The model facilitates the direct in vivo study of human biologics and monoclonal therapies.

We found that teplizumab treatment induced human T cells to migrate to the small intestine where they produced interleukin-10 (IL-10) and acquired regulatory function. These studies in mice identified a specific population of CD4+CCR6+IL-10+ T cells that we also found in patients treated with the drug. In patients, we found that the gut-tropic Tregs transiently decline during therapy but express higher levels of IL-10 and FoxP3 after completion of treatment.


Humanized mouse reconstitution and treatment with anti-CD3 mAb

Figure 1A shows staining for human lymphocytes (hCD45), T cells (hCD4/hCD8), and B lymphocytes (hCD19) in a representative mouse, and Fig. 1B shows the cell populations in all of the mice used for these studies. The mean percentages of hCD45+ cells in the group treated with teplizumab (n = 11) were 37.44 ± 7.48% (mean ± SEM) compared with the hIg (human immunoglobulin) group (30.01 ± 7.6%) (n = 11); CD4, 2.56 ± 0.74% compared with 3.84 ± 1.5%; CD8, 6.3 ± 3.29% compared with 6.34 ± 2.76%; and hCD19, 27.43 ± 6.97% compared with 18.07 ± 7.79% (all P values not statistically significant).

Fig. 1

Humanized mouse reconstitution and treatment with teplizumab. (A) FACS dot plot from peripheral blood of a representative humanized mouse before treatment showing the degree of human reconstitution. Staining for mouse CD45 (mCD45), human CD45 (hCD45), human CD8 (hCD8), human CD4 (hCD4), and human CD19 (hCD19) after gating on total lymphocyte populations. (B) Humanized mouse reconstitution before treatment with teplizumab or control hIg (n = 22). The percentages of hCD45+, CD4+, CD8+, and CD19+ human cells are shown for individual mice (the line designates median value). (C) CD4/CD8 ratio in mice treated with teplizumab or hIg showing decrease in the ratio after teplizumab treatment (*P < 0.05). (D) Coating of human CD3-TCR complex after a single injection of 5 μg of teplizumab (line histogram) or control hIg (gray histogram). Binding of OKT3 to T cells was significantly reduced in the teplizumab-treated group (***P < 0.001, n = 11). The percent coating was 39 ± 8.39% (mean ± SEM).

To replicate the effects of drug treatment in patients, we treated mice with a single dose (5 μg) of drug, which is the equivalent to the total dose administered to clinical trial subjects adjusted for the weight of the mice. Control animals received 5 μg of hIg. After teplizumab treatment, the CD4/CD8 ratio decreased in the peripheral blood from 2.03 ± 0.53 to 1.01 ± 0.50 (P < 0.05), analogous to the decline seen in patients who were treated with the drug (Fig. 1C) (15). We measured the coating of CD3 by staining peripheral blood mononuclear cells (PBMCs) with OKT3, which competes with teplizumab for binding to CD3ε. The mean fluorescence intensity (MFI) of CD3 staining decreased from 2405 ± 414 to 1031 ± 226 MFI units (P < 0.001), which corresponds to 39 ± 8.39% coating of CD3 molecules (Fig. 1D). These data are similar to those reported in patients after the first full dose of drug (15).

Decrease in human T cells after treatment with anti-CD3 mAb

The effects of teplizumab treatment on the populations of total human lymphocytes (hCD45), total human T cells (hCD2), hCD4 cells, hCD8 cells, and hCD19 cells were examined in five tissue compartments (peripheral blood, spleen, lung, bone marrow, and the small intestine). In the peripheral blood, there was a decline in circulating hCD4 cells after treatment with teplizumab (11.03 ± 5.65%) compared to control hIg (26.2 ± 8.47%; P < 0.05) (Fig. 2A). In the spleen, the total number of hCD45 cells did not show a significant change (4.5 × 106 ± 1 × 106 compared with 2.1 × 106 ± 7.2 × 105; P = 0.1). However, there was an increase in the proportion and absolute number of CD19 cells (3.3 × 106 ± 9 × 104 compared with 1.2 × 106 ± 5 × 104; P < 0.05) (Fig. 2B), a significant decrease in the proportion of CD4+ cells (8.38 ± 4.613% compared with control hIg, 24.23 ± 7.9%; P < 0.05), and a trend for a similar decrease in the CD8+ T cells (4.85 ± 2% compared with control hIg, 14.37 ± 5.6%; P = not significant) (Fig. 2B). There was also a small reduction in the proportion of T cells in the bone marrow and lung in the teplizumab-treated group (P = not significant) (Fig. 2C).

Fig. 2

Analysis of T cells in tissue compartments of humanized mice after treatment with teplizumab. (A) Analysis of peripheral blood. (Left) Dot plot of peripheral blood from a representative mouse in the teplizumab-treated group showing decline in the population of human T cells 18 hours after anti-CD3 treatment. (Right) Analysis of T and B cell populations comparing the teplizumab- and hIg-treated groups with gating on hCD45+ cells. There was a significant decline of the hCD4 population in the anti-CD3–treated group (n = 11, *P < 0.05) (mean ± SEM). (B) Analysis of spleen from the anti-CD3– and hIg-treated mice. (Left) There was a decline in the proportion of hCD4 cells in the anti-CD3 group spleens (*P < 0.05). (Right) This is accompanied by an increase in the number of CD19 cells (*P < 0.05) and a modest decline in the absolute number of CD4+ and CD8+ T cells (n = 11). (C) Analysis of the lung and bone marrow showed no evidence of accumulation of human lymphocytes with anti-CD3 treatment. Dot plots and bar graphs were gated on total live hCD45 cells, and percentages of human T and B cells were of total live human lymphocytes.

Migration of T cells to the lamina propria of the small intestine after treatment with teplizumab

Eighteen hours after treatment with anti-CD3 mAb, there was an increase in the total number of human CD45+ cells infiltrating the lamina propria (LP) of the small intestine (359 ± 92 compared with 26 ± 5; P < 0.002) (Fig. 3, A and B). Both CD4+ and CD8+ T cells were found, but there was a slightly greater proportion of CD4 cells (22.75 ± 6.15%) compared to CD8 cells (18.55 ± 3.35%). Immunohistochemistry studies showed scattered hCD45+ cells but no evidence of an organized infiltrate in the duodenum. The architecture of the villi, crypts, and luminal epithelial surfaces was intact (Fig. 3C). We did not identify any human lymphocytes in the IEL compartment of the small intestine from the anti-CD3 mAb–treated mice by fluorescence-activated cell sorting (FACS) analysis.

Fig. 3

Migration of human T cells to the LP of the small intestine. (A) Analysis of human lymphocytes extracted from the LP of the small intestine from representative animals. Dot plots on the left show live hCD45 cells as a percentage of total cells gated on scatter in a teplizumab-treated mouse compared to an hIg-treated control. The right plots show the phenotypes of these cells. Data from a single mouse representative of 11 mice. FSA, forward scatter area. (B) The number and phenotype of hCD45+ cells were determined by FACS analysis of cells isolated from the gut. In the teplizumab group, there was an increase in the total count of human lymphocytes. Each dot represents a single mouse (***P < 0.0005). The phenotypes included CD4+ and CD8+ T cells. (C) Immunohistochemistry of duodenum showing human lymphocytes in a teplizumab- but not control Ig–treated mice. The arrows show hCD45+ (brown) T cells located in the LP of the teplizumab-treated mouse (red arrow). There is no evidence of inflammation and epithelial surfaces are intact. a, teplizumab-treated (×40). b, hIg-treated (×40). c, teplizumab-treated (×20). d, hIg-treated (×20). Individual animals representative of six in each group are shown.

Microarray analysis of migrating T cells

These studies suggested that cells from the periphery migrated to the LP in response to anti-CD3 mAb treatment. To identify markers that distinguished the migrating lymphocytes, we compared, by microarray, gene expression in sorted hCD4 splenocytes after anti-CD3 mAb treatment in vivo (n = 4) to CD4+ cells harvested from splenocytes treated with anti-CD3 mAb in vitro (n = 4). Comparing the cells that were trapped in the culture wells to splenocytes allowed us to identify the differential expression of genes by the migrating cells. We confirmed the expression of genes among the gut-infiltrating cells by quantitative polymerase chain reaction (qPCR). We identified increased expression of IL-10 (sixfold), IL-8 (sevenfold), and CCR6 (threefold), as well as seven other genes (Fig. 4A).

Fig. 4

Teplizumab-induced migration of human CD4 T cells to the small intestine by the CCR6/CCL20 axis. (A) Microarray analysis of the migrating CD4+ T cell population to the gut from the spleen of humanized mice (n = 4). The genes shown were increased twofold with P < 0.05 in comparison with CD4+ T cells from the spleens of mice treated with teplizumab versus CD4+ splenocytes that were cultured with teplizumab in vitro. The migrating hCD4 cells had increased expression of CCR6, IL-10, and IL-8. (B) Expression of CCR6 on splenocytes from humanized mice cultured with teplizumab (red) and control Ig (blue) for 18 hours. Isotype staining is shown in green. There was a significant increase in the expression of CCR6 (n = 6) (*P < 0.05) but not CCR9 on CD4+ T cells during culture with teplizumab. (C) hCD4 cells have increased expression of CCR6 and reduced expression of CD62L within the small intestine (*P < 0.05; **P < 0.01; ***P < 0.001). (D) Expression of mouse CCL20 (mCCL20) was measured by qPCR as described in cellular isolates of the duodenum and ileum. mCCL20 was increased in the duodenum, of the anti-CD3 mAb–treated mice compared to ileum or to control hIg–treated mice (**P < 0.01). (E) Expression of human CCL20 was measured by qPCR in sorted human CD4 cells and cellular isolates from teplizumab- and hIg-treated mice. Expression was increased in hCD4 cells isolated from the small intestine (***P < 0.001).

Changes in expression of CCR6 and CD62 ligand on T cells after teplizumab treatment

Our array screen suggested that CCR6, induced by anti-CD3 mAb, might be a homing receptor used by cells that migrate to the gastrointestinal tract. We therefore studied the expression of CCR6 on splenocytes from the humanized mice cultured with teplizumab. There was a modest but significant increase in expression of CCR6 on CD4+ T cells (36.12 ± 5.17% compared with 31.28 ± 4.25%; P < 0.05, n = 5) (Fig. 4B), whereas the expression of another chemokine receptor, CCR9, was unchanged. In the peripheral circulation, we found a trend toward a reduced proportion of hCD4+CCR6+ cells (34.82 ± 1.19% compared with 42.68 ± 3.27%; P = not significant, n = 6). We also found increased expression of CCR6 mRNA in the CD4+ T cells in the gut compared to cells in the spleen (P < 0.001, n = 4; Fig. 4C). In addition, we found by qPCR lower levels of CD62L on the gut-infiltrating T cells (P < 0.05) (Fig. 4C).

CCL20 is the known homing ligand for CCR6. We extracted total RNA from the duodenum and ileum from mice that had been treated with anti-CD3 mAb and analyzed expression of both human and mouse CCL20 by qPCR using species-specific primers. We found increased expression of both human and mouse CCL20 in the duodenum but not the ileum in teplizumab-treated mice (Fig. 4, D and E). The increased expression of hCCL20 was on CD4+ T cells because we found increased gene expression of hCCL20 in these cells when compared to the total duodenal and ileum cell isolates (P < 0.001) (Fig. 4E). In other studies, we found increased expression of IL-8 mRNA in the gut-infiltrating cells compared to splenocytes from these mice (fig. S1; P < 0.001).

Expression of IL-10 and FoxP3 and regulatory function of gut migrating cells

We compared the expression of IL-10, a regulatory cytokine, and other associated genes in the T cells from the spleen and gut of the drug- and control Ig–treated mice by qPCR. Because there were few human cells in the LP of control mice, we compared gene expression by the gut-infiltrating cells to splenocytes from hIg-treated mice (n = 6) (Fig. 5A). The qPCR analysis confirmed increased expression of IL-10 among the CD4+ T cells that had migrated to the gut (n = 4, P < 0.01; Fig. 5A). In addition, the relative expression of FoxP3 in hCD4 T cells from the small intestine was significantly higher compared to resident spleen T cells from teplizumab- or hIg-treated controls (P < 0.01) (Fig. 5B). Surprisingly, TGF-β expression was lower in the gut T cells than in the spleen of anti-CD3 mAb–treated mice (fig. S2; P < 0.05). IFN-γ (interferon-γ) and IL-17 mRNA were undetectable in the CD4+ T cells that infiltrated the LP. To determine the function ex vivo of human CD4+ T cells that had migrated to the gut, we isolated human CD4 cells from the LP and spleens from four teplizumab-treated animals and tested the ability of these cells to inhibit proliferation of CD25-depleted splenocytes activated with anti-CD3/28 beads. We found that the CD4+ cells from the LP of teplizumab-treated animals were 10 times more effective at inhibiting the proliferation of CD4+CD25 cells from untreated animals than CD4+ splenocytes at 1:2 and even 1:8 Treg/responder ratios (P < 0.001), although the overall level of inhibition was modest (8.46 ± 0.39% compared with 0.81 ± 0.33%, n = 4) (fig. S3).

Fig. 5

Increased expression of IL-10 and FoxP3 by human CD4+ cells in the small intestine LP. (A and B) IL-10 (A) and FoxP3 (B) expression were determined by qPCR analysis in CD4+ cells from the small intestine and spleens of the treatment groups. IL-10 (***P < 0.001) and FoxP3 (***P < 0.001) are increased in the small intestine in anti-CD3–treated mice compared to anti-CD3–treated, hIg-treated, anti-CD3 + natalizumab–treated, and hIg + natalizumab–treated CD4+ splenocytes.

Blockade of migration of T cells to the small intestine and induction of Tregs after teplizumab treatment by natalizumab

The earliest events in lymphocyte migration to the gastrointestinal tract involve the direct binding of α4β7 to MAdCAM-1 that is expressed on the endothelial cells. Therefore, to determine whether T cell migration to the small intestine was required for induction of IL-10 and FoxP3, we cotreated mice with natalizumab (anti-human α4 mAb) and teplizumab. With coadministration of the mAbs, there was no longer a decrease in the number of CD4+CCR6+ cells in the spleen (n = 8) when compared to the teplizumab treatment group (n = 6) (4.2 × 105 ± 1.6 × 105 compared with 2.3 × 106 ± 4.5 × 105; P < 0.05) (Fig. 6A).

Fig. 6

Blockade of teplizumab-induced migration of T cells to the small intestine by natalizumab. (A) The total number of CCR6+CD4+ splenocytes was determined by flow cytometry (n = 8 mice per group). There was a significant decrease in the number of CCR6+CD4+ T cells after teplizumab treatment (*P < 0.05) that was reversed with natalizumab treatment. Changes in CD4+CCR6+ cells were not significantly different when compared to teplizumab treatment (n = 6) alone (P < 0.005). (B) Levels of IL-10 were measured in the serum of mice by Luminex before and after treatment with teplizumab in the four treatment groups. There was a significant increase in the concentration of IL-10 only in the teplizumab group (***P < 0.001 by paired t test, n = 6 to 9 per group, plots show median and 10th and 90th percentiles). Cotreatment with natalizumab ablated this rise in IL-10. (C) Murine allogeneic skin graft from B6 donor to reconstituted NSG mouse at day 15 after treatment with teplizumab (top) and hIg (below). The hIg-treated mouse shows evidence of rejection. Cotreatment with natalizumab and teplizumab abated the treatment effect of teplizumab on graft survival. Each symbol represents an individual mouse (P = 0.003, hIg versus teplizumab; P = 0.004, teplizumab versus teplizumab + natalizumab).

We also extracted lymphocytes from the LP of the mouse groups treated with natalizumab and found a 73% reduction in the number of hCD45+ cells infiltrating the LP of the small intestine in the teplizumab + natalizumab group (97 ± 14) compared to the teplizumab alone group (359 ± 92; P < 0.05). There was not a significant difference in the number of human cells in the LP in the hIg-treated, hIg + natalizumab (56 ± 13) (n = 8)–treated, and teplizumab + natalizumab–treated groups (P = not significant) (fig. S4). In addition to the reduced number of cells, expression of IL-10 and FoxP3 was reduced in the gut as well as in the spleen cells (Fig. 5, A and B).

To determine whether anti-CD3 mAb–induced IL-10 expression in the gut-migrating CD4 cells was an early marker of cell migration and differentiation, we measured the concentrations of IL-10 in the serum on day 1 after anti-CD3 mAb treatment and tested the effects of natalizumab treatment. There was an increase in the concentration of IL-10 after teplizumab treatment when compared to levels before treatment (day 0, 8.45 ± 2.1 pg/ml, compared with day 1, 18.46 ± 2.66 pg/ml; P < 0.005) but not in the hIg-treated group (n = 6) (Fig. 6B). Furthermore, we found that natalizumab treatment ablated the increase in systemic IL-10 mediated by teplizumab treatment (n = 6) (Fig. 6B). There was no significant change in serum levels of IFN-γ or TNF-α (tumor necrosis factor–α) in treatment groups, and IL-6 was not detectable in the serum (fig. S5, A and B).

Finally, to test the functional consequences of blockade of gut migration, we tested the effects of teplizumab with and without natalizumab cotreatment on rejection of mouse skin allografts (C57BL/6). Teplizumab treatment alone prolonged skin graft survival compared to hIg (73 ± 10 days compared with 25 ± 1 days; P < 0.003) (Fig. 6C). However, when the anti-CD3 mAb–treated mice also received natalizumab, there was rapid rejection of allogeneic skin (22 ± 1 days; P < 0.004). Natalizumab treatment + hIg had no effect on graft survival compared to hIg alone (24 ± 0.5 days compared with 25 ± 1 days) (Fig. 6C).

Teplizumab-induced changes in CD4+ and CCR6+ cells in treated subjects

To determine whether the mechanisms we had identified in the humanized mice occurred in patients, we first measured the concentrations of IL-10 in the serum of patients with T1D who received a 14-day course of teplizumab (n = 11) as part of a clinical trial. There was a rise in the concentration of IL-10 on the day after the first full dose of the drug (day 6, P < 0.05; Fig. 7A). There was also a decline in the percentage of CD4+CCR6+ T cells in the peripheral blood when study day 0 (untreated) was compared to study day 14 (before last treatment) (8.64 ± 0.96% compared with 3.32 ± 0.38%; P < 0.001; Fig. 7, B and C). About 2 months after treatment, the population of CD4+CCR6+ cells had increased to 6.54 ± 0.9% but remained lower than before teplizumab treatment (P < 0.05) (Fig. 7C). We did not identify a significant change in the CD4+CCR9+ or CD4+CD49d+ population of circulating T cells in the teplizumab-treated subjects or changes in CCR6+ populations in placebo-treated control subjects (n = 3) (fig. S6). We then stained the CD4+CD25+CCR6+ cells for intracellular IL-10 and FoxP3, without activation ex vivo, in two cohorts of treated patients (n = 10 and 7) and found that expression of both was increased compared to before treatment (Fig. 7, D to G). On day 14, at the conclusion of teplizumab treatment, the percentage of CD4+CD25+CCR6+FoxP3+ cells increased from 3.9 ± 0.98% to 18.2 ± 3.45% (P < 0.01) and the percentage of CD4+CD25+CCR6+IL-10+ cells increased from 6.6 ± 0.7% to 15.13 ± 2.1% (P < 0.05) (Fig. 7F). The frequency of the FoxP3+ and IL-10+ cells remained increased even 2 months after anti-CD3 mAb treatment compared to baseline, but the difference did not reach statistical significance. Finally, we found that the absolute number of CD4+CD25+CCR6+FoxP3+ cells expressing IL-10 increased in teplizumab-treated subjects (n = 4) from day 0 to day 14 (0.99 × 10−6 ± 0.17 compared with 1.5 × 10−6; P < 0.05) without a significant change in absolute lymphocyte count (P = not significant).

Fig. 7

Expression of IL-10 by CCR6+CD4+CD25+FoxP3+ T cells in patients treated with teplizumab. (A) Serum cytokine analysis from the treated patients showing increase in systemic IL-10 on the day after full-dose teplizumab treatment (*P < 0.05 by paired t test). (B) Representative FACS plot showing decline of CD4+CCR6+ T cells in the peripheral blood during treatment. Day 0, before treatment; day 14, immediately after treatment; day 60, after treatment (2 months). Data are from a single patient; representative of 10 is shown. (C) In teplizumab-treated subjects, CD4+CCR6+ cells decline in the peripheral blood with treatment (***P < 0.001) and remain lower after treatment (**P < 0.01 by paired t test). (D) During treatment, there is an increase in the expression of FoxP3 on CD4+CD25hiCCR6+ (**P < 0.01). (E) Representative FACS plot showing changes in IL-10 expression among CCR6+ and CCR6 cells with gates on CD4+CD25+ cells. The increase in IL-10 production is primarily on the CCR6+ cells. A single subject representative of seven is shown. Crosshairs were drawn at <1.5% staining with control antibodies. (F) Increased expression of IL-10 by CD4+CD25hiCCR6+ cells from days 0 to 14 (*P < 0.01). (G) Increase in the numbers of CD4+CD25hiCCR6+FoxP3 cells expressing IL-10 in the peripheral circulation during teplizumab treatment. *P < 0.05 by paired t test of Ln[counts (×10−6) + 1]. The cell counts were calculated from the blood counts obtained during the clinical trial.


We used humanized NOD/SCID (severe combined immunodeficient) IL-2γc−/− mice to study the initial events that occur with human cells after treatment with FcR nonbinding anti-CD3 mAb. Immunodeficient mice reconstituted with human hematopoietic stem cells develop lymphoid and myeloid lineage cells without graft-versus-host disease (21, 22). In previous studies, these mice have been shown to develop productive adaptive immune responses to Salmonella typhi (23, 24) as well as a functional Treg compartment (25). However, the mice do not fully develop lymphoid tissue. For example, the development of the gut-associated lymphoid tissue is incomplete, and in the mice we studied, we did not find gut-infiltrating human lymphocytes in the absence of anti-CD3 mAb therapy even though murine lymphocytes are found at that location in wild-type mice. Moreover, secondary immune responses to antigen have been difficult to achieve (26). However, the model provides a human immune system that can be used to investigate the mechanism of action of human mAbs in vivo without the use of human subjects.

We found that non-FcR binding mAb to CD3 causes migration of T lymphocytes to the proximal small bowel where they acquire regulatory function. This occurs after induction of CCR6 on T cells and expression of murine and human CCL20 in the cells of the gastrointestinal tract. The migrating cells produce IL-10 and then appear in the peripheral circulation. The cells that we extracted from the gastrointestinal tract had modest effects in a standard Treg assay compared to conventional natural Tregs, but this may reflect the fact that, in our assay, we used the entire gut-infiltrating CD4+ cell population, rather than a subpopulation selected on phenotypic or functional markers.

Our studies with the major histocompatibility complex (MHC) mismatch mouse skin graft model were able to show the requirement for gut migration for the immunomodulating effects of the anti-CD3 mAb and for their regulatory function. We used the anti-human α4 antibody natalizumab to test the postulated mechanism of action (2729). Blockade of α4 integrin inhibits one of the earliest steps in T cell migration across the gastrointestinal endothelium, and therefore, entry into the small intestine was inhibited even though homing molecules, such as CCR6, were still expressed by the cells. When migration of CD4+ T cells was blocked, the teplizumab treatment effect on prolongation of graft survival was abolished.

Data from humans treated with teplizumab have suggested that lymphocytes may migrate out of the systemic circulation in response to therapy. This is based on the kinetics of change in the T cell compartment and the absence of T cell receptor (TCR) excision circles when the cell counts recover after treatment, suggesting that the recovering cells were not new thymic emigrants (12, 13). A recent study of FcR nonbinding anti-CD3 mAb in mice suggested that there is at least some depletion of T cells, a possibility that we cannot exclude in our studies because we do not know the total number of cells in every compartment (30). Nonetheless, our finding that the decrease in CD4+ T cells in the spleen was blocked when natalizumab was added to teplizumab suggests that cell migration was the most important basis for changes in cell counts. Curiously, we found an increase in the absolute number of B lymphocytes in the spleen together with a decrease in the proportion of T cells. We cannot be certain whether the increase in B cells reflects homeostatic proliferation, perhaps in response to “space” created by migration or even depletion of T cells in the spleen and peripheral blood.

Our array studies initially identified IL-10 as one of the factors produced by the migrating cells, but the array studies could not identify the requirements for tissue migration. Treatment with teplizumab leads to T cell activation (19) and up-regulation of CCR6 on CD4+ T cells. This was a selective effect of anti-CD3 mAb on T cells because we did not see increased expression of CCR9 or CD49d (α4 integrin). In addition, the CD3 mAb treatment increased expression of the ligand CCL20 in the gastrointestinal tract. Previous studies by gastrointestinal biopsy have demonstrated that the CCR6/CCL20 interaction mediates migration of T cells to the upper gastrointestinal tract in humans (31, 32). The reasons why both human and murine CCL20 were increased are not clear, but these data suggest activity of a soluble factor. We did not find histologic evidence of inflammation, but we did find increased expression of IL-8 on the gut-infiltrating cells, which may induce expression of human and murine CCL20. Recent evidence suggests that IL-8 could be involved in the regulation of CCL20 expression by the intestinal epithelial cells (33). Once in the LP of the small intestine, the CCR6+ T cells increased their expression of FoxP3 and IL-10 compared to the spleen. The changes were site-specific to the small intestine and required migration because it was not seen when the migration was blocked with natalizumab. This suggests that the CD4+CCR6+ cells acquire a regulatory phenotype in the small intestine. There was no increase in systemic IL-10 and no decrease in the number of CD4+CCR6+ T cells in the spleen when migration to the small intestine was blocked.

Our findings in this humanized mouse model suggest an alternative mechanism from that recently reported by Esplugues et al. (34). They reported that anti-CD3 mAb induced expression of IL-17 in gut-infiltrating T cells and that these IL-17+ cells were controlled by elimination in the intestinal lumen with colitis or acquired a regulatory phenotype associated with production of IL-10 and TGF-β (34). However, we were unable to detect IL-17 expression by the human CD4 T cells, and the expression of TGF-β was lower in the gut-infiltrating cells compared to other locations. The reasons for the differences in mechanisms identified by the two studies may reflect the nature of the cells reconstituting the mice because the previous study used PBMCs, which may have contained IL-17 precursor cells, as well as dose of the anti-CD3 mAb and timing of the analysis.

Our studies were able to recapitulate many of the features that occur with treatment of patients with anti-CD3 mAb such as coating of human CD3, reversal of the CD4/CD8 ratio, and induction of IL-10. In patients treated with teplizumab, we found induction of an IL-10+CCR6+ population of CD4+ T cells that we postulate also migrate to the gut. In previous studies, we had detected CD4+IL-10+ cells transiently in the systemic circulation after treatment. These findings have identified the subpopulations of CD4+CD25hiCCR6+FoxP3+ T regulatory cells that can produce IL-10 and the basis for their induction. The IL-10 production was found on T cells on day 14 after the last dose of mAb without stimulation ex vivo. Although the frequency of the cells that spontaneously produce IL-10 appears to decline after that time (at month 3 after treatment), these cells would be expected to again produce IL-10 in response to antigen, which may account for lasting effects of the drug even beyond the period of drug administration. This information may be useful for monitoring the effects of drug treatment in clinical trials with teplizumab.

The ability of the reconstituted mice to accurately replicate immunological mechanisms with human immune cells is an advance over current murine models because there are inherent differences in human and murine immune cells. Furthermore, our findings highlight the limitations of mechanistic studies that restrict themselves to analysis of peripheral blood in humans. In the case of anti-CD3 mAb treatment, analysis of the peripheral blood in patients was previously able to identify a decline in circulating cells but neither the basis for the change in cell number nor the phenotype of the migrating cells. This use of the humanized mouse may now be extended to investigate the mechanism of action of other human immunotherapies.

Materials and Methods


NOD/SCID IL-2γc−/− (NSG) and C57BL/6 mice were purchased from the Jackson Laboratory and bred at our animal facility. Mice were kept under specific pathogen-free conditions. Experiments were approved by the Institutional Animal Care and Use Committee of Yale University. Samples were obtained from participants in clinical trial NCT00378508 (“Delay”). The trial and use of human umbilical vein blood were approved by the Yale University Institutional Review Board. Written informed consent was given for the use of the samples for mechanistic studies.

mAbs for staining

Before staining, all samples were preincubated with Mouse BD Fc Block (BD Biosciences). The following conjugated mAbs were used for flow cytometry analysis: murine CD45 (30-F11) [peridinin chlorophyll protein (PerCP)], CD45 [NS-1 fluorescein isothiocyanate (FITC) or PerCP], CD19 (HIB19) (FITC), CD4 (RPA-T4) [allophycocyanin (APC) or FITC], human CD45 (HI30) [phycoerythrin (PE) or Pacific Blue (PB)],CD8 (HITa) (PerCP), and CD2 (RPA-2.10) (APC). The following antibodies were also used: CD25 (M-A251) (PE), CD127 (HCD127) (PerCP), CD8 (RPA-T8) (APC), CCR6 (TG7) (PE), CD8 (RPA-T8) (PerCP-Cy5.5), CCR9 (BL) (Alexa Fluor 647), CD49d (9F10) (PE), CD62L (DREG-56) (PE-Cy5), APC CD8 (RPA-T8), FoxP3 (206D) (FITC), or IL-10 (JES3-9D7) (PECy7). Viability of cells was confirmed by use of a Live/Dead fixable dead stain kit (Invitrogen).

Isolation of human CD34+ stem cells and reconstitution of humanized mice

Human umbilical cord blood was obtained from human placentas within 4 hours of delivery. CD34+ cells were isolated on a discontinuous density gradient of Ficoll-Paque Plus (GE Healthcare) and purified with magnetically labeled microbeads conjugated to anti-human CD34+ (Miltenyi Biotec). Fresh CD34+ cells (2 × 105) were then injected intrahepatically into 1- to 2-day-old NSG pups that had been irradiated (1 Gy). The reconstitution of these mice with human cells was determined in peripheral blood at 12 weeks by FACS analysis on a FACSCalibur (BD). All FACS staining was performed in phosphate-buffered saline (PBS) supplemented with 3% heat-inactivated fetal calf serum (FCS) (Atlanta Biologicals) and 0.01% sodium azide (Sigma-Aldrich). Results from FACS were analyzed with FlowJo software (version 7.6.3).

Teplizumab dose, natalizumab dose, T cell ratio, and CD3-coating studies

The total dose of teplizumab [hOKT3γ1(Ala-Ala)] administered to a 70-kg human subject in NCT00378508 is 17 mg (0.24 mg/kg). We administered 5 μg of teplizumab intraperitoneally per mouse (0.025 kg), which is equivalent to the human dose at 0.2 mg/kg. Control mice were administered 5 μg of hIg (Sigma) intraperitoneally. The same dose of teplizumab/hIg and duration of treatment (18 hours) were used in all in vivo and ex vivo experiments. For coating studies and measurement of CD4/CD8 ratio, mice were bled immediately before treatment and at sacrifice. Coating of the CD3-TCR complex was assessed by competitive staining with mAb human CD3 (OKT3) (FITC). Coating of CD3 on circulating T cells and the CD4/CD8 ratio were calculated as previously described (16, 35). In some experiments, we also administered 600 μg of natalizumab (IgG4κ humanized mAb to human α4 integrin) intravenously 2 hours before treatment with 5 μg of teplizumab or hIg.

Extraction, purification, and analysis of human lymphocytes

Cells were isolated from the lung, bone marrow, and spleen by tissue homogenization. Peripheral blood was collected in lithium heparin–coated tubes (BD Microtainer). Harvested lymphocytes were treated with ACL lysing buffer (Lonza) to remove red blood cells. Lamina propria (LP) lymphocytes and intraepithelial lymphocytes from the small intestine were isolated as described with the following modifications. Extracted and homogenized small intestines were incubated with 10% FCS RPMI (Cellgro), collagenase (2 U/ml) (Sigma), and deoxyribonuclease I (DNase I) (10 μg/ml) (Roche) for 45 min in a shaker at 37°C. After washing with PBS, cells were loaded onto a Percoll Plus (GE Healthcare) gradient and centrifuged. Lymphocytes were harvested between the 44% and 67% interfaces with debris discarded. FACS analysis was performed on an LSRII (BD Biosciences). In certain experiments, PBMCs from normal donors were isolated by Ficoll discontinuous density gradient. These were cultured in AIM-V medium with teplizumab (5 μg/ml) or control hIg. Isolated mouse splenocytes were cultured in a similar manner.

Unless indicated, an electronic gate was placed on live hCD45 cells. Human T and B cell lymphocyte quantities were then expressed either as percent of hCD45+ cells or as absolute counts (by multiplying by the cell counts). Human CD4 cells from spleen and small intestine were purified by positive selection on magnetic-activated Dynabeads (Invitrogen). We were unable to amplify murine actin from the hCD4+ cells isolated from the small intestine in this manner.

Immunohistochemistry of humanized mouse small intestine

After treatment of humanized mice with teplizumab or hIg, the small intestine was harvested and fixed in 1% paraformaldehyde overnight at 4°C. Tissue was then placed in a PBS dextrose gradient of 5% for 30 min, 10% for 1 hour, 20% for 1 hour, and 30% overnight at 4°C. They were then washed and mounted in OCT (optimal cutting temperature) compound followed by snap-freezing. Frozen sections were cut at 5 μm on a Leica CM1850 microtome at −20°C. Antigen retrieval was performed by heating the section to 85°C for 2 hours in a Coplin jar in target retrieval solution (pH 9) (Dako). After washing, the sections were incubated for 15 min with 1% goat serum in PBS followed by avidin/biotin blocking as per the manufacturer’s protocol (Vector Labs). Anti-human CD45 (2B11 + PD7/26) was applied overnight at 4°C followed by biotinylated goat anti-mouse (Dako) at room temperature and then with streptavidin–horseradish peroxidase (BD) and staining with peroxidase substrate (Vector Labs). Hematoxylin counterstain was applied and sections were then mounted in Crystal Mount (BioMeda).

Microarray analysis of migrating T cell population

To identify genes expressed by cells that leave the spleen, we harvested the spleens from untreated reconstituted mice and prepared a single-cell suspension in AIM-V serum-free medium (Gibco). The cultures were treated with either teplizumab or hIg for 18 hours. Splenocytes were also harvested 18 hours after reconstituted mice were treated with teplizumab. RNA was prepared from cells with the RNeasy purification system (Qiagen). RNA (100 ng) from each sample was amplified as per the manufacturer’s protocol with an Illumina TotalPrep RNA amplification kit (Ambion). Biotinylated complementary RNA (cRNA) (1000 ng) was provided for expression analysis with HumanHT-12 v3 Expression BeadChip (Illumina). Twelve samples (four in vivo anti-CD3–treated group, four ex vivo anti-CD3–treated group, and four ex vivo hIg–treated group) for hCD4 were analyzed with GeneSpring GX 10 (Agilent Technologies). Genes of interest were defined as having at least twofold change in expression in at least three samples from each group (P < 0.05).

qPCR analysis

RNA from cells was converted to complementary DNA (cDNA) with reverse transcriptase (Superscript II, Invitrogen) and oligo(dT) primers following the manufacturer’s protocol. qPCR was performed with a QuantiTect SYBR Green PCR kit (Qiagen) on an IQ5 multicolor real-time PCR (Bio-Rad). Primers were designed for genes of interest (sequences in table S1). All results were normalized to the human or mouse actin genes that were quantified in parallel PCRs. Individual assays were performed as duplicates or triplicates. Results were analyzed with QGene software (36).

Cytokine analysis

Humanized mice were bled and serum was isolated immediately before treatment and at 18 hours after treatment. Samples were obtained from participants in clinical trials at predetermined time points. All serum samples were stored at −80°C. Human cytokines were measured with an ultrasensitive Milliplex xMAP assay (Millipore). The reported sensitivity of this assay for the cytokines tested is 1 pg/ml (37). Samples were analyzed on a Bioplex analyzer (Bio-Rad). To ensure the sensitivity of this assay, we validated our results with a second assay (Human Multi-spot, Meso Scale).

Suppression assays

Lymphocytes were isolated and pooled from the spleens and LP of four reconstituted mice treated with teplizumab. Responder lymphocytes were harvested from the spleens of two untreated reconstituted mice engrafted with hCD34+ cells from the same umbilical cord. Contaminating mouse lymphocytes were removed by positive selection with anti-mouse CD45 magnetic beads on an LS column (Miltenyi Biotec). Human CD4+ cells were purified by negative selection with magnetic beads (Stemcell), and the CD4+CD25+ cells were removed by positive selection. CFSE (carboxyfluorescein diacetate succinimidyl ester) (1 μM)–labeled T responder cells (1 × 104 cells per well) were cultured in 96-well round-bottom plates in the presence of dilutions of hCD4+ cells isolated from the LP and spleen. Cell cultures were stimulated with CD2/CD3/CD28 beads (Miltenyi Biotech). After 4 days, cells were harvested and analyzed by flow cytometry. Inhibition of proliferation was calculated as [1 − (% divided with added cells/% divided without added cells)] × 100.

Murine skin allografts

The mouse skin allografts were performed as previously described with the following modifications (38). Donor skin was harvested from adult C57BL/6 mice (H2b) and engrafted onto adult reconstituted NSG mice. Postoperative mice received meloxicam (1.64 mg/ml) in drinking water for 5 days. Dressings were removed at day 7 after graft. Skin grafts were monitored daily, and the day of rejection was defined as complete loss of the graft. Syngeneic skin grafts were not rejected in the reconstituted mice.

Analysis of lymphocytes from patients treated with teplizumab

Patient samples were obtained from subjects enrolled in trial NCT00378508 and had been treated with teplizumab. For each individual subject, serum cytokines and lymphocytes were analyzed before treatment (day 0), immediately after the last dose (day 14), and 2 months after treatment (day 60). Absolute lymphocyte counts were determined in a local clinical laboratory. PBMCs for FACS were extracted on the day of the patient visit by a Ficoll discontinuous density gradient and cryopreserved in liquid nitrogen until further processing. Intracellular staining was performed for FoxP3 or IL-10 as described above.

Statistical analysis

A Student’s t test for paired data or unpaired data was used where indicated. A Mann-Whitney test for unpaired data was used for data that were not normally distributed. Cytokine levels were log-transformed and compared by a t test. For multiple groups, analysis of variance (ANOVA) test with Bonferroni’s multiple comparison test was applied. A P value of <0.05 was considered statistically significant.

Supplementary Material

Fig. S1. qPCR analysis of IL-8.

Fig. S2. qPCR analysis of TGF-β.

Fig. S3. T cell suppression assay.

Fig. S4. Effect of natalizumab treatment on lymphocyte migration to the small intestine.

Fig. S5. Serum measurement of IFN-γ and TNF-α.

Fig. S6. Expression of CCR9 and CD49d after treatment with teplizumab in humans.

Table S1. qPCR primer sequences.

References and Notes

  1. Acknowledgments: We thank J. Alderman, E. Esplugues, and S. Humber for their advice with the humanized mice and helpful discussions. We also thank G. Lyon and L. Devine for expert help with regard to FACS staining. Teplizumab was supplied by Macrogenics. Funding: F.W.-L. is supported by the Dr. Richard Steevens’ Scholarship from the Health Service Executive (Ireland). Supported by grants JDRF 2007-1059, NIH DK057846, UL1 RR024139, JDRF 2006-351, and 2006-502. Author contributions: F.W.-L. and K.C.H. designed the study and wrote the manuscript. R.F. assisted with the design of the study, provided umbilical cord samples, and contributed to the critical reading of the manuscript. O.H. assisted with the microarray and quantitative PCR analysis. P.P.-H., S.D., and J.T. provided experimental assistance and discussion. F.W.-L. conducted all other experiments and analyzed the data. Competing interests: The authors declare that they have no competing interests. Data availability: The gene expression omnibus number for gene array data deposited in the National Center for Biotechnology Information database is GSE34163.

Stay Connected to Science Translational Medicine

Navigate This Article